You are on page 1of 698

3D/2D modelling suite for integral water solutions

Delft3D

Hydro-Morphodynamics

User Manual

Delft3D-FLOW
Simulation of multi-dimensional hydrodynamic flows and transport phenomena, including sediments

User Manual

Hydro-Morphodynamics

Version: 3.15.25157 5 December 2012

Delft3D-FLOW, User Manual

Published and printed by: Deltares Rotterdamseweg 185 p.o. box 177 2600 MH Delft The Netherlands

telephone: fax: e-mail: www:

+31 88 335 82 73 +31 88 335 85 82 info@deltares.nl http://www.deltares.nl

For sales contact: telephone: +31 88 335 81 88 fax: +31 88 335 81 11 e-mail: sales@deltaressystems.nl www: http://www.deltaressystems.nl

For support contact: telephone: +31 88 335 81 00 fax: +31 88 335 81 11 e-mail: support@deltaressystems.nl www: http://www.deltaressystems.nl

Copyright © 2012 Deltares All rights reserved. No part of this document may be reproduced in any form by print, photo print, photo copy, microfilm or any other means, without written permission from the publisher: Deltares.

Contents

Contents
1 A guide to this manual 1.1 Introduction . . . . . . . . . . . . . . . . 1.2 Manual version and revisions . . . . . . . 1.3 Typographical conventions . . . . . . . . 1.4 Changes with respect to previous versions 2 Introduction to Delft3D-FLOW 2.1 Areas of application . . . . . . . . . . . 2.2 Standard features . . . . . . . . . . . . 2.3 Special features . . . . . . . . . . . . . 2.4 Coupling to other modules . . . . . . . 2.5 Utilities . . . . . . . . . . . . . . . . . 2.6 Installation and computer configuration 3 Getting started 3.1 Overview of Delft3D . . . . . . 3.2 Starting Delft3D . . . . . . . . 3.3 Getting into Delft3D-FLOW . . 3.4 Exploring some menu options 3.5 Exiting the FLOW-GUI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 2 2 3 7 7 7 8 8 9 9 11 11 11 12 16 18 19 19 19 20 21 21 24 26 27 27 27 33 35 37 39 41 44 47 54 63 65 66 71 75 79 83 87 90

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

4 Graphical User Interface 4.1 Introduction . . . . . . . . . . . . . . . . . . . . 4.2 MDF-file and attribute files . . . . . . . . . . . . 4.3 Filenames and conventions . . . . . . . . . . . . 4.4 Working with the FLOW-GUI . . . . . . . . . . . 4.4.1 Starting the FLOW-GUI . . . . . . . . . 4.4.2 Visualisation Area window . . . . . . . . 4.5 Input parameters of MDF-file . . . . . . . . . . . 4.5.1 Description . . . . . . . . . . . . . . . . 4.5.2 Domain . . . . . . . . . . . . . . . . . . 4.5.2.1 Grid parameters . . . . . . . . 4.5.2.2 Bathymetry . . . . . . . . . . 4.5.2.3 Dry points . . . . . . . . . . . 4.5.2.4 Thin dams . . . . . . . . . . . 4.5.3 Time frame . . . . . . . . . . . . . . . . 4.5.4 Processes . . . . . . . . . . . . . . . . 4.5.5 Initial conditions . . . . . . . . . . . . . 4.5.6 Boundaries . . . . . . . . . . . . . . . . 4.5.6.1 Flow boundary conditions . . . 4.5.6.2 Transport boundary conditions 4.5.7 Physical parameters . . . . . . . . . . . 4.5.7.1 Constants . . . . . . . . . . . 4.5.7.2 Viscosity . . . . . . . . . . . . 4.5.7.3 Heat flux model . . . . . . . . 4.5.7.4 Sediment . . . . . . . . . . . 4.5.7.5 Morphology . . . . . . . . . . 4.5.7.6 Wind . . . . . . . . . . . . . . 4.5.7.7 Tidal forces . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

Deltares

iii

Delft3D-FLOW, User Manual Numerical parameters . . . . . . Operations . . . . . . . . . . . . 4.5.9.1 Dredging and dumping 4.5.10 Monitoring . . . . . . . . . . . . 4.5.10.1 Observations . . . . . 4.5.10.2 Drogues . . . . . . . . 4.5.10.3 Cross-sections . . . . 4.5.11 Additional parameters . . . . . . 4.5.12 Output . . . . . . . . . . . . . . 4.5.12.1 Storage . . . . . . . . 4.5.12.2 Print . . . . . . . . . . 4.5.12.3 Details . . . . . . . . . Save the MDF and attribute files and exit Importing, removing and exporting of data 4.5.8 4.5.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 95 100 100 100 102 103 105 106 107 109 110 111 113 115 115 116 117 117 120 120 121 121 124 125 127 129 130 132 132 138 138 139 140 141 142 143 146 146 148 149 150 151 153 154 155 155

4.6 4.7

5 Tutorial 5.1 Introduction – MDF-file and attribute files 5.2 Filenames and conventions . . . . . . . . 5.3 FLOW Graphical User Interface . . . . . 5.3.1 Introduction . . . . . . . . . . . . 5.3.2 Saving the input data . . . . . . . 5.4 Description . . . . . . . . . . . . . . . . 5.5 Domain . . . . . . . . . . . . . . . . . . 5.5.1 Grid parameters . . . . . . . . . 5.5.2 Bathymetry . . . . . . . . . . . . 5.5.3 Dry points . . . . . . . . . . . . 5.5.4 Thin dams . . . . . . . . . . . . 5.6 Time frame . . . . . . . . . . . . . . . . 5.7 Processes . . . . . . . . . . . . . . . . . 5.8 Initial conditions . . . . . . . . . . . . . . 5.9 Boundaries . . . . . . . . . . . . . . . . 5.10 Physical parameters . . . . . . . . . . . 5.10.1 Constants . . . . . . . . . . . . . 5.10.2 Roughness . . . . . . . . . . . . 5.10.3 Viscosity . . . . . . . . . . . . . 5.10.4 Wind . . . . . . . . . . . . . . . 5.11 Numerical parameters . . . . . . . . . . 5.12 Operations . . . . . . . . . . . . . . . . 5.13 Monitoring . . . . . . . . . . . . . . . . . 5.13.1 Observation points . . . . . . . . 5.13.2 Drogues . . . . . . . . . . . . . 5.13.3 Cross-sections . . . . . . . . . . 5.14 Additional parameters . . . . . . . . . . . 5.15 Output . . . . . . . . . . . . . . . . . . . 5.16 Save MDF-file . . . . . . . . . . . . . . . 5.17 Additional exercises . . . . . . . . . . . . 5.18 Execute the scenario . . . . . . . . . . . 5.19 Inspect the results . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6 Execute a scenario 161 6.1 Running a scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 iv Deltares

Contents 6.2 6.3 Run time . . . . . . . . . . . . . . . . . . . . . . Files and file sizes . . . . . . . . . . . . . . . . 6.3.1 History file . . . . . . . . . . . . . . . . 6.3.2 Map file . . . . . . . . . . . . . . . . . . 6.3.3 Print file . . . . . . . . . . . . . . . . . . 6.3.4 Communication file . . . . . . . . . . . . Command-line arguments . . . . . . . . . . . . 6.4.1 Command-line arguments deltares hydro 6.4.2 Command-line arguments delftflow . . . Frequently asked questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 164 165 165 166 166 167 167 169 170 173 173 173 173 175 179 179

6.4

6.5

7 Visualise results 7.1 Introduction . . . . . . . . . . . . 7.2 Working with Delft3D-GPP . . . . 7.2.1 Overview . . . . . . . . . 7.2.2 Launching Delft3D-GPP . 7.3 Working with Delft3D-QUICKPLOT 7.4 Delft3D-GISVIEW interface . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

8 Manage projects and files 181 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 8.1.1 Managing projects . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 8.1.2 Managing files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 9 Conceptual description 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 General background . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.1 Range of applications of Delft3D-FLOW . . . . . . . . . . . . . 9.2.2 Physical processes . . . . . . . . . . . . . . . . . . . . . . . . 9.2.3 Assumptions underlying Delft3D-FLOW . . . . . . . . . . . . . 9.3 Governing equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.1 Hydrodynamic equations . . . . . . . . . . . . . . . . . . . . . 9.3.2 Transport equation (for σ -grid) . . . . . . . . . . . . . . . . . . 9.3.3 Coupling between intake and outfall . . . . . . . . . . . . . . . 9.3.4 Equation of state . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.1 Flow boundary conditions . . . . . . . . . . . . . . . . . . . . 9.4.1.1 Vertical boundary conditions . . . . . . . . . . . . . 9.4.1.2 Open boundary conditions . . . . . . . . . . . . . . 9.4.1.3 Shear-stresses at closed boundaries . . . . . . . . . 9.4.2 Transport boundary conditions . . . . . . . . . . . . . . . . . . 9.4.2.1 Open boundary conditions for the transport equation 9.4.2.2 Thatcher-Harleman boundary conditions . . . . . . . 9.4.2.3 Vertical boundary conditions transport equation . . . 9.5 Turbulence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5.1 Algebraic turbulence model (AEM) . . . . . . . . . . . . . . . . 9.5.1.1 Algebraic closure model (ALG) . . . . . . . . . . . . 9.5.1.2 Prandtl’s Mixing Length model (PML) . . . . . . . . . 9.5.2 k -L turbulence model . . . . . . . . . . . . . . . . . . . . . . 9.5.3 k -ε turbulence model . . . . . . . . . . . . . . . . . . . . . . . 9.5.4 Low Reynolds effect . . . . . . . . . . . . . . . . . . . . . . . 183 183 183 183 184 185 186 192 200 203 204 206 207 207 211 218 218 218 219 221 221 225 225 226 227 229 231

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

Deltares

v

Delft3D-FLOW, User Manual 9.6 9.7 Secondary flow; sigma-model only . . . . . . . . . . . . . . . . . . . Wave-current interaction; σ -model only . . . . . . . . . . . . . . . . 9.7.1 Forcing by radiation stress gradients . . . . . . . . . . . . . 9.7.2 Stokes drift and mass flux . . . . . . . . . . . . . . . . . . . 9.7.3 Streaming . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.7.4 Wave induced turbulence . . . . . . . . . . . . . . . . . . . 9.7.5 Enhancement of the bed shear-stress by waves . . . . . . . 9.8 Heat flux models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.8.1 Heat balance . . . . . . . . . . . . . . . . . . . . . . . . . . 9.8.2 Solar radiation . . . . . . . . . . . . . . . . . . . . . . . . . 9.8.3 Atmospheric radiation (long wave radiation) . . . . . . . . . . 9.8.4 Back radiation (long wave radiation) . . . . . . . . . . . . . . 9.8.5 Effective back radiation . . . . . . . . . . . . . . . . . . . . . 9.8.6 Evaporative heat flux . . . . . . . . . . . . . . . . . . . . . . 9.8.7 Convective heat flux . . . . . . . . . . . . . . . . . . . . . . 9.8.8 Overview of heat flux models . . . . . . . . . . . . . . . . . 9.9 Tide generating forces . . . . . . . . . . . . . . . . . . . . . . . . . 9.9.1 Tidal potential of Equilibrium tide . . . . . . . . . . . . . . . 9.9.2 Tidal potential of Earth tide . . . . . . . . . . . . . . . . . . 9.10 Hydraulic structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.10.1 3D gate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.10.2 Quadratic friction . . . . . . . . . . . . . . . . . . . . . . . . 9.10.3 Linear friction . . . . . . . . . . . . . . . . . . . . . . . . . . 9.11 Flow resistance: bedforms and vegetation . . . . . . . . . . . . . . . 9.11.1 Bedform heights . . . . . . . . . . . . . . . . . . . . . . . . 9.11.1.1 Dune height predictor . . . . . . . . . . . . . . . . 9.11.1.2 Van Rijn (2007) bedform roughness height predictor 9.11.2 Trachytopes . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.11.2.1 Trachytope classes . . . . . . . . . . . . . . . . . 9.11.2.2 Averaging and accumulation of trachytopes . . . . 10 Numerical aspects of Delft3D-FLOW 10.1 Staggered grid . . . . . . . . . . . . . . . . . . . . 10.2 sigma-grid and Z -grid . . . . . . . . . . . . . . . . . 10.3 Definition of model boundaries . . . . . . . . . . . . 10.4 Time integration of the 3D shallow water equations . 10.4.1 ADI time integration method . . . . . . . . . 10.4.2 Accuracy of wave propagation . . . . . . . . 10.4.3 Iterative procedure continuity equation . . . 10.4.4 Horizontal viscosity terms . . . . . . . . . . 10.4.5 Overview time step limitations . . . . . . . . 10.5 Spatial discretizations of 3D shallow water equations 10.5.1 Horizontal advection terms . . . . . . . . . . 10.5.2 Vertical advection term . . . . . . . . . . . . 10.5.3 Viscosity terms . . . . . . . . . . . . . . . . 10.6 Solution method for the transport equation . . . . . . 10.6.1 Cyclic method . . . . . . . . . . . . . . . . 10.6.2 Van Leer-2 scheme . . . . . . . . . . . . . 10.6.3 Vertical advection . . . . . . . . . . . . . . 10.6.4 Forester filter . . . . . . . . . . . . . . . . . 10.7 Numerical implementation of the turbulence models . vi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 236 239 240 241 242 243 248 250 251 253 254 254 255 258 258 261 262 264 265 265 265 266 266 266 267 269 270 271 277 281 281 283 283 285 286 288 289 289 290 290 291 295 295 296 297 298 298 299 301

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

Deltares

Contents 10.8 Drying and flooding . . . . . . . . . . . . . . 10.8.1 Bottom depth at water level points . . 10.8.2 Total water depth at velocity points . 10.8.3 Drying and flooding criteria . . . . . 10.9 Hydraulic structures . . . . . . . . . . . . . . 10.9.1 3D Gate . . . . . . . . . . . . . . . 10.9.2 Quadratic friction . . . . . . . . . . . 10.9.2.1 Barrier . . . . . . . . . . . 10.9.2.2 Bridge . . . . . . . . . . . 10.9.2.3 Current Deflection Wall . . 10.9.2.4 Weir . . . . . . . . . . . . 10.9.2.5 Porous plate . . . . . . . . 10.9.2.6 Culvert . . . . . . . . . . . 10.9.3 Linear friction . . . . . . . . . . . . . 10.9.4 Floating structure . . . . . . . . . . . 10.10 Artificial vertical mixing due to σ co-ordinates 10.11 Smoothing parameter boundary conditions . 10.12 Assumptions and restrictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 302 304 308 310 311 311 312 313 314 314 316 316 320 320 320 324 325 327 327 327 327 328 329 329 330 331 331 331 332 333 333 333 334 335 335 335 335 336 336 336 338 339 340 343 344 344 344 345 345 vii

11 Sediment transport and morphology 11.1 General formulations . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 11.1.2 Suspended transport . . . . . . . . . . . . . . . . . . 11.1.3 Effect of sediment on fluid density . . . . . . . . . . . 11.1.4 Sediment settling velocity . . . . . . . . . . . . . . . 11.1.5 Dispersive transport . . . . . . . . . . . . . . . . . . 11.1.6 Three-dimensional wave effects . . . . . . . . . . . . 11.1.7 Initial and boundary conditions . . . . . . . . . . . . . 11.1.7.1 Initial condition . . . . . . . . . . . . . . . 11.1.7.2 Boundary conditions . . . . . . . . . . . . 11.2 Cohesive sediment . . . . . . . . . . . . . . . . . . . . . . . 11.2.1 Cohesive sediment settling velocity . . . . . . . . . . 11.2.2 Cohesive sediment dispersion . . . . . . . . . . . . . 11.2.3 Cohesive sediment erosion and deposition . . . . . . 11.2.4 Interaction of sediment fractions . . . . . . . . . . . . 11.2.5 Influence of waves on cohesive sediment transport . . 11.2.6 Inclusion of a fixed layer . . . . . . . . . . . . . . . . 11.2.7 Inflow boundary conditions cohesive sediment . . . . 11.3 Non-cohesive sediment . . . . . . . . . . . . . . . . . . . . . 11.3.1 Non-cohesive sediment settling velocity . . . . . . . . 11.3.2 Non-cohesive sediment dispersion . . . . . . . . . . 11.3.2.1 Using the algebraic or k -L turbulence model 11.3.2.2 Using the k -ε turbulence model . . . . . . . 11.3.3 Reference concentration . . . . . . . . . . . . . . . . 11.3.4 Non-cohesive sediment erosion and deposition . . . . 11.3.5 Inclusion of a fixed layer . . . . . . . . . . . . . . . . 11.3.6 Inflow boundary conditions non-cohesive sediment . . 11.4 Bedload sediment transport of non-cohesive sediment . . . . 11.4.1 Basic formulation . . . . . . . . . . . . . . . . . . . . 11.4.2 Suspended sediment correction vector . . . . . . . . 11.4.3 Interaction of sediment fractions . . . . . . . . . . . . Deltares

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Delft3D-FLOW, User Manual 11.4.4 Inclusion of a fixed layer . . . . . . . . . . . . . . . . . . . . . . . . 11.4.5 Calculation of bedload transport at open boundaries . . . . . . . . . 11.4.6 Bedload transport at U and V velocity points . . . . . . . . . . . . . 11.4.7 Adjustment of bedload transport for bed-slope effects . . . . . . . . Transport formulations for non-cohesive sediment . . . . . . . . . . . . . . . 11.5.1 Van Rijn (1993) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.2 Engelund-Hansen (1967) . . . . . . . . . . . . . . . . . . . . . . . 11.5.3 Meyer-Peter-Muller (1948) . . . . . . . . . . . . . . . . . . . . . . . 11.5.4 General formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.5 Bijker (1971) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.5.1 Basic formulation . . . . . . . . . . . . . . . . . . . . . . 11.5.5.2 Transport in wave propagation direction (Bailard-approach) 11.5.6 Van Rijn (1984) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.7 Soulsby/Van Rijn . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.8 Soulsby . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.9 Ashida–Michiue (1974) . . . . . . . . . . . . . . . . . . . . . . . . . Morphological updating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.6.1 Bathymetry updating including bedload transport . . . . . . . . . . . 11.6.2 Erosion of (temporarily) dry points . . . . . . . . . . . . . . . . . . . 11.6.3 Dredging and dumping . . . . . . . . . . . . . . . . . . . . . . . . . 11.6.4 Bed composition models and sediment availability . . . . . . . . . . Specific implementation aspects . . . . . . . . . . . . . . . . . . . . . . . . Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 347 347 348 351 351 356 356 357 357 358 359 361 363 365 367 367 370 371 372 373 374 375 377 377 380 380 381 381 381 382 382 382 383 384 384 386 386 387 388 388 389 389 390 390 391 393 393 393 394

11.5

11.6

11.7 11.8

12 Fixed layers in Z -model 12.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Time integration of the 3D shallow water equations . . . . . . . . . 12.2.1 ADI time integration method . . . . . . . . . . . . . . . . . 12.2.2 Linearisation of the continuity equation . . . . . . . . . . . 12.3 Bed stress term . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4 Horizontal viscosity terms . . . . . . . . . . . . . . . . . . . . . . . 12.4.1 Overview time step limitations . . . . . . . . . . . . . . . . 12.5 Spatial discretisations of 3D shallow water equations . . . . . . . . 12.5.1 Horizontal advection terms . . . . . . . . . . . . . . . . . . 12.5.2 Vertical advection term . . . . . . . . . . . . . . . . . . . . 12.5.3 Viscosity terms . . . . . . . . . . . . . . . . . . . . . . . . 12.6 Solution method for the transport equation . . . . . . . . . . . . . . 12.6.1 Van Leer-2 scheme . . . . . . . . . . . . . . . . . . . . . 12.6.2 Vertical advection . . . . . . . . . . . . . . . . . . . . . . 12.6.3 Forester filter . . . . . . . . . . . . . . . . . . . . . . . . . 12.7 Baroclinic pressure term . . . . . . . . . . . . . . . . . . . . . . . 12.8 Numerical implementation of the turbulence models . . . . . . . . . 12.9 Drying and flooding . . . . . . . . . . . . . . . . . . . . . . . . . . 12.9.1 Bottom depth at water level points . . . . . . . . . . . . . . 12.9.2 Total water depth at velocity points . . . . . . . . . . . . . 12.9.3 Upwinding of the water level in defining the total water depth 12.9.4 Drying and flooding criteria . . . . . . . . . . . . . . . . . 12.10 Cut-cell and 45 degrees closed boundaries . . . . . . . . . . . . . 12.10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 12.10.2 Cut Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.10.3 45 degrees closed boundary . . . . . . . . . . . . . . . . . viii

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . .

Deltares

Contents 12.11 Hydraulic structures . . . . . 12.11.1 3D Gate . . . . . . 12.11.2 Quadratic friction . . 12.11.3 Linear friction . . . . 12.11.4 Floating structure . . 12.12 Assumptions and restrictions References Glossary of terms A Files of Delft3D-FLOW A.1 MDF-file . . . . . . . . . . . . . . . . . . . . . . . . A.1.1 Introduction . . . . . . . . . . . . . . . . . . A.1.2 Example . . . . . . . . . . . . . . . . . . . A.1.3 Physical parameters . . . . . . . . . . . . . A.1.3.1 Tide Generating Forces . . . . . . A.1.3.2 Thatcher-Harleman Conditions . . A.1.4 Output options . . . . . . . . . . . . . . . . A.2 Attribute files . . . . . . . . . . . . . . . . . . . . . A.2.1 Introduction . . . . . . . . . . . . . . . . . . A.2.2 Orthogonal curvilinear grid . . . . . . . . . . A.2.3 Computational grid enclosure . . . . . . . . A.2.4 Bathymetry . . . . . . . . . . . . . . . . . . A.2.5 Thin dams . . . . . . . . . . . . . . . . . . A.2.6 Dry points . . . . . . . . . . . . . . . . . . A.2.7 Time-series uniform wind . . . . . . . . . . A.2.8 Space varying wind and pressure . . . . . . A.2.8.1 Defined on the computational grid A.2.8.2 Defined on an equidistant grid . . A.2.8.3 Defined on a curvilinear grid . . . A.2.8.4 Defined on a Spiderweb grid . . . A.2.9 Initial conditions . . . . . . . . . . . . . . . A.2.10 Open boundaries . . . . . . . . . . . . . . . A.2.11 Astronomic flow boundary conditions . . . . A.2.12 Astronomic correction factors . . . . . . . . A.2.13 Harmonic flow boundary conditions . . . . . A.2.14 QH-relation flow boundary conditions . . . . A.2.15 Time-series flow boundary conditions . . . . A.2.16 Time-series transport boundary conditions . A.2.17 Time-series for the heat model parameters . A.2.18 Bottom roughness coefficients . . . . . . . . A.2.19 Horizontal eddy viscosity and diffusivity . . . A.2.20 Discharge locations . . . . . . . . . . . . . A.2.21 Flow rate and concentrations at discharges . A.2.22 Dredge and dump characteristics . . . . . . A.2.23 Dredge and nourishment time-series . . . . A.2.24 Dredge and dump areas . . . . . . . . . . . A.2.25 Observation points . . . . . . . . . . . . . . A.2.26 Moving observation points . . . . . . . . . . A.2.27 Drogues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394 394 395 396 396 396 397 407 413 413 413 413 416 417 417 418 422 422 423 424 425 426 427 429 429 430 434 434 435 435 437 438 440 441 442 443 448 452 455 457 458 459 461 469 471 472 473 475

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Deltares

ix

Delft3D-FLOW, User Manual A.2.28 Cross-sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.2.29 Fourier analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . B Special features of Delft3D-FLOW B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.2 Decay rate constituents . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3 Hydraulic structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.1 3D gates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.2 Quadratic friction . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.2.1 Barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.2.2 Real-time control . . . . . . . . . . . . . . . . . . . . . B.3.2.3 Bridge . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.2.4 Current deflection wall (CDW) . . . . . . . . . . . . . . B.3.2.5 Weirs (2D model) . . . . . . . . . . . . . . . . . . . . . B.3.2.6 Local weir . . . . . . . . . . . . . . . . . . . . . . . . . B.3.3 Porous plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.4 Culvert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.4.1 Definition of culvert in the discharge input file (<∗.src>) . B.3.4.2 Properties for culverts defined in INI file (<name.cul>) . B.3.4.3 Additional key-value pairs for culvert of type ‘c’ . . . . . . B.3.4.4 Additional key-value pairs for culvert of type ‘d’ or ‘e’ . . . B.3.4.5 Additional key-value pairs for culvert of type ‘u’ . . . . . . B.3.4.6 More culverts . . . . . . . . . . . . . . . . . . . . . . . B.3.5 Linear friction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.5.1 Rigid sheet . . . . . . . . . . . . . . . . . . . . . . . . B.3.6 Floating structure . . . . . . . . . . . . . . . . . . . . . . . . . . . B.3.7 Upwind at Discharges . . . . . . . . . . . . . . . . . . . . . . . . B.3.8 User defined discharge through a structure . . . . . . . . . . . . . B.4 Space varying Coriolis coefficients . . . . . . . . . . . . . . . . . . . . . . B.5 Temperature modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.5.1 Direct specification of net solar radiation . . . . . . . . . . . . . . B.5.2 Specification of the coefficient of free convection . . . . . . . . . . B.5.3 Output of computed heat fluxes . . . . . . . . . . . . . . . . . . . B.6 Evaporation and precipitation . . . . . . . . . . . . . . . . . . . . . . . . . B.7 Space varying wind and pressure . . . . . . . . . . . . . . . . . . . . . . B.7.1 Space varying wind and pressure on an equidistant grid . . . . . . B.7.2 Space varying wind and pressure on a separate curvilinear grid . . B.7.3 Space varying wind and pressure on a Spiderweb grid . . . . . . . B.8 Horizontal large eddy simulation . . . . . . . . . . . . . . . . . . . . . . . B.9 Sediment transport and morphology . . . . . . . . . . . . . . . . . . . . . B.9.1 Sediment input file . . . . . . . . . . . . . . . . . . . . . . . . . . B.9.2 Morphology input file . . . . . . . . . . . . . . . . . . . . . . . . . B.9.3 Sediment transport input file . . . . . . . . . . . . . . . . . . . . . B.9.4 User defined transport routine for sand or bedload fractions . . . . B.9.5 User defined transport routine for mud fractions . . . . . . . . . . . B.9.6 User defined routine for the settling velocity . . . . . . . . . . . . . B.9.7 Sediment transport and morphology boundary condition file . . . . B.9.8 Morphological factor file . . . . . . . . . . . . . . . . . . . . . . . B.9.9 Initial bed composition file . . . . . . . . . . . . . . . . . . . . . . B.10 Fluid mud (2-layer approach) . . . . . . . . . . . . . . . . . . . . . . . . . B.10.1 Two layer system . . . . . . . . . . . . . . . . . . . . . . . . . . . x 476 477 481 481 482 483 483 484 484 485 486 486 487 489 491 493 493 494 495 495 497 497 498 498 500 502 502 512 513 514 514 514 515 517 517 522 526 531 536 537 543 553 557 561 565 568 571 573 574 576

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Deltares

Contents B.10.1.1 Suspension layer . . . . . . . . . . . . . . . . . . . . . . B.10.1.2 Fluid mud layer . . . . . . . . . . . . . . . . . . . . . . . B.10.1.3 Mathematical modelling of fluid mud layer . . . . . . . . . B.10.2 Applying fluid mud . . . . . . . . . . . . . . . . . . . . . . . . . . . B.10.2.1 DelftIO library . . . . . . . . . . . . . . . . . . . . . . . . B.10.2.2 Running a simulation in foreground . . . . . . . . . . . . . B.10.2.3 Running a simulation in background . . . . . . . . . . . . B.10.2.4 Pitt falls . . . . . . . . . . . . . . . . . . . . . . . . . . . B.10.3 Post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . Z-model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.11.1 Grid definition and construction . . . . . . . . . . . . . . . . . . . . B.11.2 Defining the keywords in the FLOW-GUI . . . . . . . . . . . . . . . B.11.3 Restrictions and limitations . . . . . . . . . . . . . . . . . . . . . . . B.11.3.1 Defining Cut-cells and 45 degrees closed boundaries . . . B.11.4 45 degrees staircase closed boundary points (Z -model only) . . . . B.11.5 Cut-cell closed boundary points (Z -model only) . . . . . . . . . . . . Non-hydrostatic solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.12.1 The use of hydrostatic and non-hydrostatic models . . . . . . . . . . B.12.2 Governing equations . . . . . . . . . . . . . . . . . . . . . . . . . . B.12.3 A pressure correction technique for computing the non-hydrostatic pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.12.4 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . B.12.5 Conjugate gradient method (CG) . . . . . . . . . . . . . . . . . . . B.12.6 Practical aspects of using the non-hydrostatic flow module . . . . . . B.12.6.1 Switches in MDF-file . . . . . . . . . . . . . . . . . . . . B.12.6.2 Grid spacing . . . . . . . . . . . . . . . . . . . . . . . . . B.12.6.3 Vertical mixing . . . . . . . . . . . . . . . . . . . . . . . . B.12.6.4 Convergence criterion CG solver . . . . . . . . . . . . . . B.12.6.5 Defining the input (keywords) for the non-hydrostatic pressure approach . . . . . . . . . . . . . . . . . . . . . . . . User defined functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.13.1 Boundary conditions for turbulence models . . . . . . . . . . . . . . B.13.2 Diagnostic mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.13.3 Particle wind factor . . . . . . . . . . . . . . . . . . . . . . . . . . . Domain decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.14.2 Motivations for domain decomposition . . . . . . . . . . . . . . . . . B.14.3 Local refinement horizontal and vertical . . . . . . . . . . . . . . . . B.14.4 Pre-processing, processing and post-processing . . . . . . . . . . . B.14.5 Restrictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.14.6 How to set-up a domain decomposition model . . . . . . . . . . . . Surfbeat/roller model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.15.1 Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.15.2 Formulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B.15.3 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . B.15.4 Coupling with other modules . . . . . . . . . . . . . . . . . . . . . . B.15.5 Modes of operation . . . . . . . . . . . . . . . . . . . . . . . . . . . B.15.6 Input description . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bedform heights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Trachytopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 577 577 577 579 580 581 581 582 582 582 582 584 584 585 585 587 588 588 589 590 594 594 594 595 595 597 597 598 599 599 601 601 602 602 602 606 608 609 610 612 612 613 614 616 616 616 620 624

B.11

B.12

B.13

B.14

B.15

B.16 B.17

Deltares

xi

. . . . . . . . . . . . . . . .19 B.4 Example . . . . .1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . B. . . . . . . . . .6 Model results .4 Input description . 651 C. . . . . . . . . . . . . . . . . . . . . . . . . .23. . . . .22. . . . . . . . . .23.22. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. . . . . . . . . . . . . . . . . . . B. . . . . . . . . B. . .3. . . . . . . . . .1. . . . Reuse temporary files . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . . .1. B.23. . .18 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . B. . . . . . User Manual B.3. . .22. . . . . . . . .23.23. . . . . . . . . .3. . . 1D–3D Coupling . . . . . .dis>) with amount of entrained water . . . . . .4. B. . . . . . . . .1 Introduction . . . . D.3 Extending the time-series file (<∗. .1 Trachytope definition file . . .2 Implementation of Delft3D-SOBEK coupling . . . . . . . . . . . . . .3. . . .3. . . . . . .2 Bubble screen or line bubble plume . . . . . E Computational grid F Delft3D-NESTHD F. . . B. B. . . . . . . . . . . . . . .22. . . B. . . . . . . . . . . . . . . F. . . . .3 Model setup and input (including best practise) . . .22. . . . . Dry run . . . . . . . . . . . . . . .2 List of astronomical constituents . . . .3 Running of the coupled model system . . . . . . . . . . .22. . D. . . . . . . . . . . . . . . . . . . . . . . . . . . .22. . . . . . D. . . . . . . . . . . . . .2 Setup of the communication file used by coupling . .1 Motivation for online Delft3D-SOBEK coupling . . 657 657 657 658 659 659 659 661 667 667 669 670 671 . .22. . . . . . B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 Selection of the vertical turbulence model D. . . . . . . B. . . B. . . . . . . . . . . . . . .23. . . . .2 Theory .22. . . .22. .Delft3D-FLOW. . . . . . . . . .20 B. . . . . . . . . Bubble screen . . B. . .2 Extending the discharge locations file with bubble screens B. . . . . . . B. . .1 Preparation of the Delft3D-FLOW and SOBEK models . . .2 Area files . . .3 Strongly stratified . . .1 Single nozzle bubble plume . . . . .3 Bubble plume in stagnant stratified water . . . . . . . . B. . . . . Change the update frequency of the nodal factors . . . . . . . . . . . . . . . . .3 How to use NESTHD2 F. . . . . .3 Numerical implementation . . .17. . .1 Introduction . . . . . . . . . . . . . . . . . . .22. . . . .4 Best practice with regard to running coupled Delft3D-SOBEK simulations . . . . . . . . . . . . .2 How to use NESTHD1 F. . . . . . . . . .5 Coupling with other models . . . . . . . . . . . . . B. . . . . . . . .1 Well-mixed . . . . . . . .22. . .1 Entrained water as function of the air injection . . . . . . . . .21 B. . . . . .4 Versions and limitations . . . . . . . . . . . . . . . . . . . . . . . . 651 D Some modelling guidelines D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 625 627 628 629 629 629 630 631 632 633 633 633 637 638 639 640 640 641 641 642 642 643 644 645 646 648 648 649 B. . . . . . . B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Deltares .22 B. .1 Generating a file with bubble screen locations . . . . . . . . . Creating D-Water Quality input files . . . . . . . . . . . . . . . . . .4. .3. . . . . . . . . . . .23 C Astronomical constituents 651 C. . . . . . . . . B. . . . . . . . . .17. . . . .2 Partly mixed . . . .23. . . .2 Depth-averaged or 3D model . . . . . . . . . . . . . . . . . . . .3. .23. . . . . . . . B. . . . . . . . .1. D. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

17 Main window of the FLOW-GUI . . . . . Visualisation Area Window . . . Specifying the physical constants . . . . . . . . . . . . . . . . . . . . . . . . . .34 System architecture of Delft3D . Sub-data group Domain → Bathymetry . . . . Contents of a Component set with two tidal constituents having corrections . . . . . Dry point at grid location (m. . . Transport conditions. Straight channel. . .18 4. . . . . . . . . . . 8 Title window of Delft3D . . . . . . . . . . . .13 4.2 4. .33 4. . . . . . . .3 3.5 4. . . . . . . .14 4. Window with HLES parameters . . . . . . 13 Select working directory window . . . . . . . . . . . . . . . Sediment definition window . . . . . . . . . . . . . . . .19 4. . . . . . 15 Menu bar of the FLOW-GUI . . . . . . . . . . . . . . . . . . 12 Selection window for Hydrodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 4. . . . . . .List of Figures List of Figures 2. . . . Sub-data group Roughness . .4 3. . . . Defining the eddy viscosity and eddy diffusivity . . . . . . . .10 3. . .30 4. . .2 3. . . (M−1 to M+1. . . . . Sub-data group Thin dams . . . . . . . Specifying harmonic boundary conditions . . . . . .1 3. . . . . . . . n) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 24 25 26 27 28 30 31 32 33 35 36 37 38 39 42 43 45 49 50 51 52 56 57 58 61 62 63 64 66 67 69 71 72 Deltares xiii . . Specifying time-series boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . .28 4. Equivalence of v -type thin dams (left) and u-type thin dams (right) with the same grid indices. . . . . . . Sub-data group Grid parameters . . . . . . . . . . . . . . . . . . . . Data Group Processes . . . . . . . . . . . . . . . . . . . . . . . . . .26 4. . . . . . . . . . . . . . . . . . . . . Sketch of cross-section with 8 grid cells . . . . . . . . . . . . . . . . . . . . . location of open and closed boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . .22 4. . . .20 4. . . . . . . . . . . . . . . Specifying QH-relation boundary conditions . . . .7 4. . . . . . . . . . . . . .21 4. . . . . . . . . . . . .11 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 4. .11 4. . . . 17 Save changes window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 4. . . . . Data Group Time frame . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 4. . 15 File drop down menu . . . . . . . . . . . . . . 13 Select working directory window to set the working directory to <flow\friesian tidal inlet> 14 The current working directory is not shown in the title bar due to its length . . . . . . 14 Main window of the FLOW Graphical User Interface .8 3. . . . . .31 4. . . . . . .6 4. . . . . . . . . . . . . . . . . . . . . . . Main window for defining open boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Data Group Domain selection and input fields . . . . . Definition sketch grid system to North orientation . . . .5 3. . . . . . . . . Thatcher Harleman time lags . . .6 3. 11 Main window Delft3D-MENU . .12 4. . . . . . .24 4. . . . . .17 4. . . . . . . . . . . . . . . .25 4. . . . . . . . . . . . . . . . . . . Open and save window for boundary locations and conditions . . . . Specifying the layers thickness . . .23 4. . . . Data Group Description . . .29 4. . . . . . . . . .15 4. . . . . . . . . . . . . . . Possible selections of View → Attributes . . . . . . . . .4 4. . . . . . . . Specifying transport boundary conditions . . . Display symbols of all grid related quantities . . . . . . . . . . . . . . . . . . . . . . . . . Sub-data group Dry points . .1 3. . . . . .8 4. . . . . . . .32 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data group Initial conditions . . Examples of the wind drag coefficient . . . . .9 3.7 3. . . . . . . . . . . . . . . . . Staggered grid of Delft3D-FLOW . . . . .10 4. . . . . . . . . . . . . . . . . . . . . . . . . N) . . . . . . . . . . . . . . .27 4. . . . . . . . . . Specifying astronomical boundary conditions . . . . . . . .

. . . . Data Group Numerical parameters . . . . . . . .Delft3D-FLOW.17 5. . cohesive sediment parameters (continued) .56 4. . . . . . . .43 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Processes: Pollutants and tracers sub-window . . . . . . . . . . . . . Sub-data groups of Domain . . . . . . . . . . .1 5. . . . . . . . . . . . . . . . . . . .35 4. . . . . . . . . . . . . . . . . . . . . User Manual 4. . . . . . . non-cohesive sediment parameters . . . . . . . . . . . overall and cohesive sediment parameters . . . . . . . Sub-data group Bathymetry . . . . .54 4. . . . . . Sub-window to define the discharge rate and substance concentrations Decomposition of momentum released by a discharge station in (m. . . . .8 5. . . . . . Data Group Description sub-window . . . . . . . Boundaries sub-window . . . . . . . . . . . . . . . . . . . . . .Constants sub-window . . . . . . . Initial conditions sub-window . . . . . . . .53 4. . .3 5. . . . . . . . . . . . Sub-data group Grid parameters . . . . .16 5. . . . . . . . . . . . . . . . . . . . . . Boundaries: Transport Conditions window . . . . . . . . . . Sub-data group Dry Points . . . . . . . Sub-data group Sediment. . . . . . . . . . . . Sub-data group Sediment. . . . . . . . . . . . . . . . . . . . . . .23 xiv Sub-data group Heat flux model . . . . . . . . . . . . . . . . . . . . . . . . .47 4.44 4. . Output Specifications window . . . . . . . . . . . .39 4. . . Data Group Discharges . Equivalence of v -type thin dams (left) and u-type thin dams (right) with the same grid indices. Sub-data group Monitoring → Cross-Sections . . . Sets of thin dams blocking v -velocities (left) and blocking u-velocities (right) . . . . Sub-data group Morphology . . . . . . . 76 80 80 81 82 84 88 89 90 91 96 98 98 100 100 101 102 103 105 106 107 110 111 111 112 119 120 121 121 122 123 123 124 125 125 127 128 129 129 131 131 132 133 134 135 136 137 139 Starting window of the FLOW Graphical User Interface . Deltares . . . . . . . . . . . . . . . . .51 4. . . . . . . . . . . . Data group Output . . . . . . . . . .36 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 5. . Window Time frame . . . . . . . . . .21 5. . . .49 4. . . Sub-data group Thin dams .7 5. . . . . . . . . . . . . . . . . . . . . .40 4. . . . . . . . . . .5 5. . . . . . . . . Sub-data group Output → Print . .41 4. . . . . . . . . . . . . . . . . . . . . Sub-window Tidal forces . . . . . Sub-window for Monitoring locations . Location of a dry point at grid indices (m. . . . . . . . . . . . . . . . . . . . Visualisation Area window . . . . . . .59 5.15 5. Data Group Additional parameters . . . . . . . . Physical parameters sub-data groups . . . . . . . . . . . . . . . . Wind definition window . . . .57 4. . . . . Sub-data group Output storage . . . . . . . . . .48 4. . . . . . . . . . . .52 4. . . . . . . . . . . . . . . . . . . .38 4. Processes window . . . . . . . . . . . . . . . . . . . . Sub-data group Grid parameters. Nautical definition wind direction . . . . . . . . . . . .46 4. . . . . . . . . . . . . . . . . . . .22 5. . . . . .12 5. . . . .14 5. . .19 5. .55 4.10 5. . . . . . . .11 5. n) . . . . . . . .45 4. . . . . . . . .18 5. . . . . . . . . Sub-data group Monitoring → Drogues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Staggered grid used in Delft3D-FLOW . . . . . cohesive sediment parameters (continued) . . . .6 5. . . . . . . . . . . . . Sub-window for Observation points . . . .9 5. . . . . . . . . . . . . . . . Physical parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42 4. . . . . . Harmonic boundary conditions . . . . . . . . . . . . . . . . . . . . . . . .50 4. . . . . . . . . . . . . . . Open/Save Boundaries sub-window . . . . . . . .37 4. . . . . . . . . . . . . . . . . . . . . . . . . n) Sub-data group Dredging and dumping . . . . . . . . . . . . . filenames. . . . . . . . . . . . . . . . . . Sub-data group Sediment. . . . . . . . . .20 5. . . . . n) . . . . . . File drop down menu . . . . Save changes window . . .13 5. . . . . . . . . . . .4 5. . . . . . . . . . . . . type of co-ordinate system and grid dimensions . . . . . . .58 4. . . . . . Sub-data group Sediment. . . . . . . . (m − 1 to m + 1. . . . . . . . . . . . . . . . .

. 139 140 141 142 142 144 145 145 146 147 147 148 149 149 150 150 152 153 154 155 157 158 159 161 162 162 163 174 175 177 178 179 Hierarchy of Delft3D-GPP . . .29 5. . . . . . . . . .2 9. . . . . . . . . . . . . 90)) . . . . . . . . depth (h) and total depth (H . . . . . .2 6. . . . . . . . . . . . . Parameters and locations in the <trih-f33. . . . . . . Delft3D-QUICKPLOT interface to Delft3D result . . . . . . . . . . . .1 7. . . . Observation points sub-window . . . . . .31 5. . . . . . .39 5. . . . . . . . . . . . p. . . . . . . . . . . . . . . . . . . . .6 Spiral motion in river bend (from Van Rijn (1990)) .5 9. . . . . . . . . . Data Group Operations. . . Representation of a v-cross-section in the Visualisation Area window . Discharges sub-window . . . . . . . . . . . . . . . . . . . . . . . . . . . . Definition of water level (ζ ). . . . . . Output details sub-window . . . . . 5. . . . Representation of a drogue in the Visualisation Area window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41 5. Select the MDF-file to be verified . . . . . . . 194 Examples wind drag coefficient depending on wind speed. . . . . Drogues sub-window . . .1 6. .2 7. . . . . . . . . . . . Filled time table in the Wind sub-window . . . . . . . . . . Part of the report to the output window Select a report file for inspection . . . . . . . . . . Output sub-window .37 5. . and B–B’)216 9. . . . . . . . Output storage sub-window . .44 Roughness sub-window .27 5. . . . . . . . Representation of an observation point in the Visualisation area window . . . . . . .3 9. . . . p. . . . .5 Illustration of memory effect for open boundary . . . . . . . . . . . . . . . . . Computed time-series of the water level.4 MENU-window for Hydrodynamics . . . . . . . . . . . . . .dat> file Some options to change the plot attributes . . . . . . . 89)) . . . . . . . . (1993b. . . 5. . . . . . current and salinity in observation point Obs4 . . . . . . . . . . . . . . .32 5. . . . Part of the <tri-diag> report file . . . . . . . . . . and contours of water level on 6 August 1990 01:00 hr . . . . . .35 5. . . . . . . . . . . . . . . . . . . . . . . . 242 9. . . . . . . .7 Vertical profile secondary flow (V ) in river bend and direction bed stress . .28 5. . .4 7. . . . . . . . .34 5. . . .List of Figures 5. .9 Schematic view of non-linear interaction of wave and current bed shear-stresses (from Soulsby et al. . . . . . Figure 16. . .3 6. . . . . . . . . . . . . . . . . . . Select scenario to be executed . . . . . . . . . . . . . . . . . . Discharge Data sub-window . . . . . . . . .45 Computational grid with drogue Dr4. . . . . . . . . . . . . . . . . . . . . . Figure 17. . . . . .25 5. 232 9. . .40 5. . . . . 244 9. . . . . . .8 Vertical distribution of turbulent kinetic energy production . . . . . . .3 7. . . . . 211 Hydrodynamic model of coastal area with three open boundaries with offshore boundary (A–B at deep water) and two cross shore boundaries (A–A’. . . . . . . . . . . . . . . . . . . .10 Inter-comparison of eight models for prediction of mean and maximum bed shear-stress due to waves and currents (from Soulsby et al. . . . . . . . . 220 9. . . . . . . . . . . . . . . . 193 Example of σ . . . (1993b. . . . . . . . Viscosity sub-window . . . . . . . . . . . . . . . . . . . Wind sub-window . . . . . . . . .46 Vector velocities and contours of salinity on 6 August 1990 01:00 hr . . . . . .43 5. . . . Main window of Delft3D-GPP . . . . . . . . . . . . . . . . . . . . . . . .4 7. . . . . . . . . . . . .30 5. . . . . . . 6. . . . . 233 9. . . . . . . . . . . . . . . Monitoring sub-window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42 5. . . . .1 9. . . . . . . . . . . . . . . . . . . .38 5. . . . . .and Z -grid . Numerical parameters sub-window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26 5. . . . . . . . .33 5. . . . . . . . . . . . . . . . 246 Deltares xv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-sections sub-window . . . . .24 5. . . . . Representation of a discharge in the Visualisation Area window . . . . .36 5. . . . . . .

. . . see Eq. . . . . 10. . . . . . . . . . . . (a) Control Volume for mass for the Flooding scheme. . . . 378 379 383 385 xvi Deltares . . . . . . . . . . . .2 Vertical computational grid Z -model (left) and σ -model (right) . .12 9. . . . . .1 11. . . . . . . . . . . 10. . . . . . . . . . . . . . . . . the velocity point is set dry. 10. .2 11. . . . . . . . MEAN-option . 10. . . upwind approach. .1 10. . . . . . . . .20 Finite Volume for diffusive fluxes and pressure gradients . . 10. . . . .1 Irregular representation of bottom boundary layer in the Z -model . . . .18 Computational layer partially blocked at bottom of gate . . see Eq.17 Example of a 3D Gate (vertical cross-section) . Numerical region of influence for one time step. . .13 9. . . . . .4 Aggregation of Control volumes in the vertical due to variation position free surface . . . . . . .64. . . . the velocity point is set dry. . . . . . . . . . Mapping of physical space to computational space . . . . . . . . . . . . . . (b) Control Volume for momentum in horizontal and vertical direction for the Flooding scheme . . .Delft3D-FLOW. Grid points in difference stencil dependent on flow direction . . . . . . . .64. . . . . . . . . . . . . . . . . . . . . . 10. . . .10 Definition bottom depth on FLOW grid . 10. . The flow-through height is based on the maximum water level. . . . 12. .63. Effect of tide generating force on the computed water elevation at Venice Earth ocean tidal interaction (after Schwiderski (1980)) . . 10. 10. . . . . . 10. . . . . . . . . . where a is Van Rijn’s reference height . . .15 Overtopping of a river bank. . . .6 Morphological control volume and bedload transport components . . . . . .8 Layer numbering in σ -model . . . . . . . . . The flow-through height is determined by flow direction. . . . . . . . . . . . .21 Left and right approximation of a strict horizontal gradient . . . . . 348 11. . . .12 Drying of a tidal flat. . “Zig-zag channel” . .16 Special limiter for critical flow due to a sudden contraction (Flooding scheme and increase in bottom larger than DGCUNI ) . . . . 340 Approximation of concentration and concentration gradient at bottom of kmx layer . . . . . 340 Schematic arrangement of flux bottom boundary condition . . . . Example of Delft3D-FLOW model area . 10. . . . . . . . . (b) Hδσ < σ ∂x δx . . . . . . . . . 3D view (left) and top view (right) . . . .3 10. .14 10. . . . . the velocity point remains wet. . . . . . . . . . .14 Drying of a tidal flat. . . . . .4 305 306 307 307 311 312 321 322 323 325 Sediment mixing coefficient in non-breaking waves (Source: Van Rijn (1993)) 337 Selection of the kmx layer. . . . . . averaging approach. . 10. . . . The flow-through height is based on the average water level. . . . upwind approach. . . . . . . . . . . 10. . 12. . .63. . . . . . . . see Eq. . . . . . . . . see Equation 10. The flow-through height is based on the average water level. . . . . . User Manual 9. . . . . . . . 10. .13 Overtopping of a river bank (weir). . . . . .11 Negative control volume with two positive flow-through heights. . . . . . . .22 Cold start with damping of eigen oscillations due to bottom friction . . . . . . . . . . . . . . . . . 12. . . . . . . . . . . . . . . . . . . . . . . . . 10. 371 12. . . . . . . . . . . . . . . 10. . . . . .5 Setting of bedload transport components at velocity points .2 10. . . . . . . . . . . . . . . . . .7 Overview of the heat exchange mechanisms at the surface . . . . . . . . . Example of Delft3D-FLOW grid . 341 11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Co-ordinate system position Sun . . . . . . . averaging approach. . the velocity point remains wet. . . . 10. . . . . .3 discretisation along streamlines. . . . . . . . .19 Example of a hydrostatical consistent and inconsistent grid. . 10. . .5 10. 249 252 262 264 282 282 282 284 285 288 293 295 300 303 304 305 Example of a grid in Delft3D-FLOW . . . . . . . . . . . . . . . . .11 9. . . . . . . . .4 10. . . . . .3 11. . 11. . . . . . . ∂H (a) Hδσ > σ ∂H ∂x δx.9 Illustration of wiggles in vertical direction . . . . . . . . . . . Grid staggering. .6 10.

. . 488 Top view of 2D weirs . .2 B. 604 Schematised island with domain decomposition . . . . .19 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . 603 Example of coupling of models with a different dimension . . . 604 Schematised island without domain decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607 Problem layout sketch .17 B. . . . . . .26 B. . . . . . . . . . . . 528 Spiderweb grid definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 A. . . . . . . . . .1 B. . . . . . . . 584 Inserting appropriate keywords to switch on the Z -grid-model . . . . . . . . . . . . . . . . . . (1999). . . . . . . . . .3 A. . . . . . . . . . . . . . . . . . 12. . . .8 left: Cut Cell (definition) and right: defined by shifting (exaggerated) the corner point to boundary. . . . . . . . Cross-sections in model area . . . . . . . . . . .2 A. . . . . . . . . . .4 A. . 491 Top view of rigid sheets . . . . . . . . . 486 Example of CDW in perspective view (left) and top view (right) . . . . . . . . . .4 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16 B. . . . . . . . . . . . . . . . . . . . . . . . . . 12. . . . . . 580 Vertical grid construction. . . . . . . . . . . . . . . . 12. . . . . . . .22 B. . . . . . . . . . . . . . . . . . . . . 12. . . .18 B. . . . . . . . . . . . . . .27 Example of computational grid enclosures . . . .7 B. . . .8 B.10 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14 B. . . . . . . .25 B. . . . . . . . . . . . . .5 Horizontal fluxes between neighbouring cells with variation in position free surface . . . . . . . . . . .1 A. . . . . . . . . . . . . . .9 Flow along staircase boundary. . . . . . . .6 Definition bottom depth on FLOW grid . . . . . . . . . .11 B. 593 Area where a non-hydrostatic pressure is taken into account . . . 490 Local weir . . . . . . . . . . . . . 587 Cut-cell closed boundary (not related to the data specified in the example above)588 Schematic representation of the free surface boundary condition for the pressure correction. . . . . . . . . . . . . . . . . . .6 B. . . . . . . . . . . . . . 528 Definition of truncation wave numbers due to resolution and numerical damping 534 Schematic representation of the governing processes between suspension layer and fluid mud layer Winterwerp et al. . . . . .12 B. . . .5 B. . . . . . . . . 500 Cross-sectional view floating structure The vertical lines are drawn through the velocity points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The bottom is represented as a staircase around the depth in water level points. . . . . . . . . . . . . . . . . . .13 B. . 502 Illustration of the data to grid conversion for meteo input on a separate curvilinear grid .15 B. . . . . . . . . . . . . . . . 12. . . . . . . . . . . . . . . . . . Dry points in model area . . .9 B. . .21 B. . . . . . . . . . 586 45 degrees staircase closed boundary . . . .20 B. . . Z -model . . . .6 B. 12. . . . 386 389 391 393 394 394 395 395 425 427 428 430 433 477 Example of 3D gates in perspective view (left) and top view (right) . . . . 12. . . . . . . . 525 Wind definition according to Nautical convention . . . . . . . . . 585 Defining cut-cell and 45 degree closed boundaries . . . . . . . . . . . .24 B. . . . . . . 596 Defining the Non-hydrostatic solver using the Z -model in the FLOW-GUI . . . 484 Barriers in model area . . .3 B. . . 613 Deltares xvii .10 Reflection of velocities . . . . .11 Example of a 3D Gate (vertical cross-section) . . . . . . . 598 Example of grid refinement . A. . . . . . . . . . . . . . . . . . . . . . . . . . . . 576 A schematic representation of two Delft3D-FLOW modules running simultaneously simulating a fluid mud problem . . . . . . . . . . . . .7 The flow-through height is determined by the flow direction. . . . . . . . . . . . . . . . .12 Computational layer partially blocked at the bottom of the 3D gate . Example of thin dams in a model area . . .23 B. . . . . . Definition sketch of wind direction according to Nautical convention Definition wind components for space varying wind . . . . . .List of Figures 12. . . 605 Example of grid refinement in the horizontal direction . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Additional tools window with the NESTHD1 and NESTHD2 options Specification of input and output files for NESTHD1 . . . . . . 630 B. . . . . Above the first plunge point the second plunge point creates an intrusion and recirculation inside the epilimnion. . . . .3 E. . . 666 Hydrodynamics selection window with the Tools option . . . . . . . . . . . . . . . . . . . . . . .36 Enable Delft3D-FLOW in SOBEK settings and select MDF-file and communicationfile. . . . . . . . . . . . . .33 Initial temperature profile (blue) and after application of a bubble screen at z = −16 m (red line). mid. . . . . . . . .1 E. . . . . . . . . . . . . . . . . . 641 B. . . . . . . . . . . . . . . . . . . . .4 E. 658 Left: items with the same (array) number. . . . . . . . . . . . . 662 Definition sketch of a (12 ∗ 7) staggered grid with grid enclosure (thick line) and numerical grid . . 643 B. . . . . .2 F. . . . . . . . . .29 Definition of near-. . . . . . . .35 Coupling of the Delft3D-FLOW and SOBEK grids.and far-field of the circulation induced by a bubble screen and the vertical profile of the vertical (downward) velocity in the midfield circulation cell . . . .34 Explicit exchange of water levels and discharges between Delft3D-FLOW and SOBEK. . . .2 E.30 Flux of entrained water as function or air flow under atmospheric conditions and height above the nozzle in stagnant non-stratified water based on experiments in (Milgram. . . . . . . 631 B. . 643 B. . . . . . . . . . 646 D. . . . . .1 E. . . . . . . User Manual B. . Right: a computational control volume662 Lower-left (left) and lower right (right) computational grid cell . . . . . 634 B. . . . . . . . . . . . . . . . . . .1 F. . . . . . . Overview grids overall and nested models .28 Schematic axisymmetric bubble plume with entrainment of water by the rising bubbles . . .32 The mid field averaged heat equation with vertical distribution of sources/sinks and vertical (downward) velocity profile and the model equation applied to a single grid box. . . . . . . . . . . . . . . . . . . . . . . . . . . 663 Location and combination of water level and velocity controlled open boundaries664 Straight channel with 10 ∗ 5 computational grid cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 F. . . . . 668 668 669 670 671 xviii Deltares . . . . . . . . . . . . 639 B. . Specification of input and output files for NESTHD2 . . . . . . . .31 Schematic overview of the introduction of cold hypolimnion water into the lower part of the warmer epilimnion by a bubble plume. . . . . . . . . .3 F. . . . . . . .4 F. . . . . 632 B. . . . . . . . .Delft3D-FLOW. . . . . . . . . . . . . . On horizontal axis the temperature (in ◦ C) and on the vertical axis the vertical position in the water column (in m). . . 1983) The injected air flux is given in [Nm3 /s] where N stands for normal atmospheric conditions. . . .5 Steps to determine if a 3D model is required . .

. . . . . . . . . phases and amplitude on alongshore waterlevel boundary and corresponding frequenties. . . . . . . . . . . . . . 1993a) . . . . . . . . . .11 Summary of time dependent input data of the heat flux models . . . . Description of the configuration file for deltares hydro . . . . . . . . . . . . Print flags for history-data . . phases and amplitudes for the cross-shore Neumann boundaries . . . . . . . . . 253 9. . . .1 4. . . . . . . . . .6 B. . . . . . . . . . . . .7 B. . . .7 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphological input file without keywords . 366 11. Definition of open and closed boundaries. . . . . . . 382 418 418 419 419 419 420 420 462 535 537 539 541 543 550 551 554 554 xix . . . . . . . . . . . . . . . . . . . . . . . . . . continued (Soulsby et al. . . . . . . . . . . . . . . . . . 203 Frequencies. . . . . . . . . . . . . . . . . . . . 351 11. Transport formula parameters . . . . . . . . . . . . . . .3 Overview of the coefficients used in the various regression models. . . .2 B. . . . . . . Time step limitations shallow water solver Delft3D-FLOW . . . . . . . . . . .2 9. . . . . . .. . . .4 A. . . . . . . . . . . . . . . . . . . . . . . .3 B. . . . . . . . . . . . . 263 10. . . . . . . .7 Terms of the heat balance used in heat model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 11.2 A. . . . . .1 Additional transport relations . . . . . . . . . . . . . Overview of eddy viscosity options in Delft3D-FLOW . . . . . . . . . . . . . . .9 Deltares Print flags for map-data . . . . . . . . . . . . . . .5 B. . . .6 Albedo coefficient and cloud function . . . . . . . . . . . . . . . 366 12. . . . . . . . . . . . . Default parameter settings for the SGS model Sediment input file with keywords . .1 6.10 Terms of the heat balance used in heat model 5 . . . . . . . . . . . . . 259 9. . . . . . . . . . . . . . . . . . . . . . . . Optional output flags under Additional parameters Dredge and dump input file with keywords . . . .3 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Storage flags for history-data . . . . . . . . . 199 Overview of eddy diffusivity options in Delft3D-FLOW .List of Tables List of Tables 4. . . Additional transport relations .9 Terms of the heat balance used in heat model 4 . . . . . . . Storage flags for map-data . . . . . . .6 A. . . . . . 245 9. . . . . . . 260 9. . . . . . . . . .1 Time step limitations shallow water solver Delft3D-FLOW . . . . . . . . . . . . . . . . . . . . . . .2 4. Morphological input file without keywords . . . . . . . . . . . . . . 261 9. . . . . . . . 22 40 52 118 164 167 Simulation performance on different operating systems . . . . . . . . . . . . .4 Overview of the main attribute files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Fitting coefficients for wave/current boundary layer model . . .3 A. .5 A. . . . .. . . . . . . . . . . . . . . . . . . . . . . . .5 A. . . . . . . . . Overview of attribute files. . . .1 Time step limitations shallow water solver Delft3D-FLOW A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 B.8 Terms of the heat balance used in heat model 2 . . . . . . . .2 9. . . . . 260 9. . . . . Storage flags for history-data . . . . . . . . . . . . . . . .4 B. . . . . . .3 A. . . . . . . . . . . . .12 Constants of major tidal modes . . . .1 6. . . . Print flags for history-data . . . Morphological input file with keywords . . . 217 9. . . . . . . . . . . . 1993a) . . . . . . . . . .7 B. . . . . . . . . . . . . .3 5. . Options for sediment diameter characteristics Sediment input file without keywords . . . . 259 9. .2 Overview of the coefficients used in the various regression models (Soulsby et al.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .10 B. . . .Delft3D-FLOW. Bedform keywords in mdf file . . Example of a <name. . . . . . . . . .1 Sediment transport formula input file with keywords Initial bed composition file keywords . . . . xx Deltares . . . . . . 556 573 607 611 621 651 Astronomical constituents . Example of vertical grid refinement . . . . . . . . . . .13 B.12 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . .14 C. . User Manual B. . . . . . .ddb> file . . . . . . . . . . .11 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 11: Sediment transport and morphology.and post-processing tools. bedload transport and morphological updating of the bottom. To make this manual more accessible we will briefly describe the contents of each chapter and appendix. Chapter 12: Fixed layers in Z -model. in verifying this input. discusses the various grids. Delft3D-FLOW. provides practical information on the selection of all parameters and the tuning of the model. the required computer configuration and how to install the software. Chapter 8: Manage projects and files.1 A guide to this manual 1. If this is your first time to start working with Delft3D-FLOW we suggest you to read and practice the getting started of Chapter 3 and the tutorial of Chapter 5. Chapter 9: Conceptual description. provides a detailed insight into the managing of projects and scenarios. provides a list of publications and related material on the Delft3D-FLOW module. Chapter 6: Execute a scenario. describes the three-dimensional transport of suspended sediment. as well as all practical matters about the implications of parameter selections. explains in short the visualisation of results.. which gives access to all Delft3D modules and to the pre. These chapters explain the user interface options and guide you through the definition of your first simulation. provides specifications of Delft3D-FLOW. Chapter 4: Graphical User Interface. used to define the input required for a flow simulation. It introduces the post processing program Delft3D-GPP to visualise or animate the simulation results. Last but not least you will get a first introduction into the FLOW Graphical User Interface (GUI). Chapter 5: Tutorial. explains the use of the overall menu program. describes the theoretical physics modelled in Delft3DFLOW. References. Chapter 7: Visualise results. in executing the simulation and in inspecting the results. emphasis at giving you some first hands-on experience in using the FLOW-GUI to define the input of a simple problem. grid-numbering etc. Deltares 1 . such as the areas of applications. the concept of fixed. the standard and specific features provided. Chapter 10: Numerical aspects of Delft3D-FLOW .1 Introduction This User Manual concerns the hydrodynamic module. Chapter 2: Introduction to Delft3D-FLOW . of the Delft3D software suite. discusses how to verify and execute a scenario and provides information on run times and file sizes. horizontal layers in the vertical grid are given. Chapter 3: Getting started.

a non-hydrostatic module. This manual applies to Delft3D-FLOW version 3. user-defined functions. Appendix A: Files of Delft3D-FLOW . surfbeat or roller model.60. angular frequency. discusses the location of open and closed boundaries on the staggered grid used in Delft3D-FLOW. However. Generally. gives a description of the files that can be used in the input of Delft3D-FLOW. User Manual Glossary of terms. space varying Coriolis coefficients. amplitude in the equilibrium tide and the relation if an amplitude coupling exists with other components. 1. domain decomposition. these files are generated by the FLOW-GUI and you need not to be concerned about their internal details. the following conventions help you to distinguish between different elements of text to help you learn about the FLOW-GUI. This value can be a string of characters referring to a file that contains additional input parameters for this function.3 Typographical conventions Throughout this manual. for instance to generate them by means of other utility programs. and FLOW-GUI version 3. Nesting in Delft3D-FLOW is executed in three steps. Appendix E: Computational grid. Appendix D: Some modelling guidelines. fluid mud.Delft3D-FLOW. Z -model. this appendix discusses some guidelines to determine when you need a 3D computation and which vertical turbulence model you need. 2 Deltares . Appendix F: Delft3D-NESTHD . trachytopes. discusses the steps to generate boundary conditions for a nested Delft3D-FLOW model. For normal use of Delft3D-FLOW you can skip this appendix.39. this appendix gives a complete overview of the astronomical components supported by Delft3D-FLOW. using two separate utilities and the Delft3D-FLOW program. gives an overview and description of the additional functions provided by Delft3D-FLOW.2 Manual version and revisions A manual applies to a certain release of the related numerical program. Reading this appendix is suggested when you want to know all the details of the staggered grid and specific implementation aspects and consequences. The definition and use of the grid enclosure is discussed in detail. 3D sediment and morphology. in certain cases it can be useful to know these details. contains a list and explanations of the terms and abbreviations used in this manual. An additional function provides specific functionalities which are not yet supported by the user interface. but which are recognised by a keyword in the MDF-file with one or more values. Appendix C: Astronomical constituents. horizontal large eddy simulation. hydraulic structures. evaporation and precipitation. These features are: decay rate for constituents.28. In case the hydrodynamic and transport boundary conditions of a Delft3D-FLOW model are generated by a larger (overall) model we speak of a nested model. For each component is given: its name. space varying wind and pressure. given the type of modelling application. 1. Appendix B: Special features of Delft3D-FLOW .00.

Sub-windows are displayed in the Module window and cannot be moved.15 Description Meteo input (wind. <>. such as the Visualisation Area window.01. In case of an input field you are supposed to enter input data of the required format and in the required domain. and path names are expressed between angle brackets.8 and B. Upon selecting this item (click or in some cases double click with the left mouse button on it) a related action will be executed. see Appendices A. Deltares 3 . Leaving them out might result in misinterpretation. pressure. For the Linux and UNIX environment a forward slash (/) is used instead of the backward slash (\) for PCs. Windows can be moved independently from the Module window. Bubble screen added. etc.A guide to this manual Example Waves Boundaries Description Title of a window or sub-window. in most cases it will result in displaying some other (sub-)window. 10 points.02 and higher). 1D–3D Coupling added.7. Item from a menu. Selections of menu items. are described as such: for instance ‘select Save and go to the next window’.2. Units are given between square brackets when used next to the formulae. title of a push button or the name of a user interface input field. Data to be typed by you into the input fields are displayed between double quotes.mdw> “27 08 1999” delft3d-menu [m/s] [-] 1. Save <\tutorial\wave\swan-curvi> <siu. filenames. Commands to be typed by you are given in the font Courier New.60. Directory names.) modified (FLOW v3. option boxes etc.4 Changes with respect to previous versions Version 3.

example <∗.e. Chapter 12. and wind is active in the FLOW simulation.runid> changed in <runid.4. New functionality: Maximum number of constituents (pollutants and/or sediments) increased to 99. References: References on UNESCO formulation added.cor> file. Astronomical component A0 cannot be corrected by a <∗.2.bcc> file: time function should be time-function. Appendix B. Section 9. Chapter 11 updated.3. New keyword (IniSedThick) for Initial sediment layer thickness at the bed. GUI reads old keyword and converts data to new keyword.exe. MENU screens updated. Section A.ddb> you have to replace the grid filenames by the MDF-filenames. Online Delft3D-WAVE and Online coupling is now possible.1: description horizontal viscosity extended taking into account Subgrid scale viscosity and HLES viscosity. description of η and η added.3.3: UNESCO formulation added.3. New functionality: Initial conditions from a map-file described. Section 9.mft> file: time function should be time-function. Maximum number of discharges increased to 500. Equation (11. Section A. Units for reflection parameter at open boundaries added.13 4 Deltares .97): V replaced by q .bct> file: time function should be time-function.21.5. example <∗. Equation (9.fil> with runid not truncated. GUI output file <attribute files.94): − before max changed to +.8. example <∗.8. New functionality: Time-varying morphological scale factor.5. User Manual Version 3. New functionality: reflection parameter for discharge boundaries implemented.5 (Command-line arguments) added. New functionality: The name of the MDF-file may be up to 256 characters. Appendix B. Appendix F. Limitation of 4 Gb for NEFIS files added.1.dis> file: time function should be time-function.12.3. example <∗. New functionality: Section 6.8 updated. Momentum equation adapted.exe and trisim.8.15. Section B. the runid is not limited anymore to 3 characters.2. 3. Unit is metre. this wind can be used in the WAVE simulation.2: description horizontal diffusivity extended taking into account Subgrid scale viscosity and HLES diffusivity. In Data Group Numerical parameters. New functionality: if FLOW runs online with WAVE. Section B. Transport equation adapted. Section B13. Depth specified at cell corners or cell centres introduced.173) corrected.1: Restriction added: The boundaries of the nested model may not consist of a single grid cell. New functionality: Neumann boundaries described. runprocedure adjusted to use arguments for tdatom. see B.9. Text adjusted accordingly. Equation (11.13: Before using <model id.4. Section B.14 Description Section 9.Delft3D-FLOW. i.

sediment and morphology. In Chapter 10: description of the Flooding Scheme.11 3. Special treatment for ‘staircase’ closed boundary added for Z -model. Operation of Change working directory updated in Chapter 3.12 respectively. Section 10. Reference and Glossary updated and renamed to Chapter 12. is dedicated for specifics of the Z -model.6. Chapter 4 updated with latest GUI developments (Cartesian or spherical grids. C(urrent)D(eflection)W(all)-features added (for σ . Also in Chapter 4 the description of In-out discharges improved. specification of tide generating forces moved to Physical parameters. Reference version for these change notes.50. Inclusion of Z -model (horizontal layers) and Non-hydrostatic approach.8: Update of Sediment and Morphology latest developments. New Chapter.10: Input description for Z -model.14 added: Surfbeat/roller model.09. In Chapter 7 Delft3D-GIS updated to Delft3D-GISVIEW. Chapters 9 and 10 are amended.12 and 12. heat flux model parameters added. additional drying and flooding options.02 and higher).2: Equation (10. 3. Section 10. Theoretical background of Sediment and morphology updated and moved from Appendix B. Update of MENU windows with second postprocessing tool Delft3DQUICKPLOT.12 Description Chapter 4: the contribution from the HLES sub-grid model will be added to the background values (FLOW v3. Horizontal Large Eddy Simulations. specification of wind moved to Physical parameters. online Delft3D-WAVE. Appendix B. specification of astronomical flow boundary conditions changed. In Chapter 4 emphasized that the stop time may not be too far ahead of the Reference date.4.and Z -model). no slip condition added. 12.A guide to this manual Version 3. Appendix B. Appendix B.10 Deltares 5 .8 to a new Chapter 11. dredging and dumping.3: Description of AOI method removed. momentum solver options).35) 0 and 1 interchanged. Chapter 4 updated with online coupling.

User Manual 6 Deltares .Delft3D-FLOW.

Density driven flows (pressure gradients terms in the momentum equations). It has been designed for experts and non-experts alike. Simulations in deep lakes and reservoirs. Robust simulation of drying and flooding of inter-tidal flats. river discharges). fully integrated computer software suite for a multi-disciplinary approach and 3D computations for coastal. 2. Drogue tracks. water quality.1 Areas of application Tide and wind-driven flows (i.2 Introduction to Delft3D-FLOW Deltares has developed a unique. Cooling water intakes and waste water outlets. hydraulic jumps. Advanced turbulence models to account for the vertical turbulent viscosity and diffusivity based on the eddy viscosity concept. bores and flood waves. Fresh-water river discharges in bays. Advection-diffusion solver included to compute density gradients with an optional facility to treat very sharp gradients in the vertical. 2. storm surges). The Delft3D suite is composed of several modules. Online sediment transport and morphology. Deltares 7 . grouped around a mutual interface. waves. algebraic and constant model. river and estuarine areas. Space and time varying wind and atmospheric pressure. Thermal stratification in lakes. sediment transports. Transport of dissolved material and pollutants. Simulation of Tsunamis. the vertical grid is defined following the σ co-ordinate approach. Salt intrusion. Stratified and density driven flows.2 Standard features Tidal forcing. morphological developments and ecology. In 3D simulations. Non-hydrostatic flows. Time varying sources and sinks (e. Simulation of the thermal discharge.g. which this manual is about. is one of these modules. It can carry out simulations of flows. effluent discharge and the intake of cooling water at any location and any depth. Wave-driven currents. Four options are provided: k-ε.e. boundary fitted grid. Delft3D-FLOW is a multi-dimensional (2D or 3D) hydrodynamic (and transport) simulation program which calculates non-steady flow and transport phenomena that result from tidal and meteorological forcing on a rectilinear or a curvilinear. k-L. seas and reservoirs. while being capable to interact with one another. Delft3D-FLOW. River flow simulations. The effect of the Earth’s rotation (Coriolis force).

water elevations. in river bends) which is especially important in sedimentation and erosion studies (for depth averaged — 2DH — computations only). Optional facility for user-defined functions. Wave induced stresses and mass fluxes. Various options to model the heat exchange through the free water surface. Optional facility to switch between a number of advection solvers. 2. Current Deflecting Wall (CDW) floating structures.Delft3D-FLOW. Influence of waves on the bed shear stress. which are (see Figure 2. salinity. Built-in automatic switch converting 2D bottom-stress coefficient to 3D coefficient. User Manual Figure 2.g.1): module Delft3D-WAVE D-Water Quality Delft3D-PART Delft3D-ECO Delft3D-SED description short wave propagation far-field water quality mid-field water quality and particle tracking ecological modelling cohesive and non-cohesive sediment transport 8 Deltares . Optional facility for special points such as 3D gates. vertical eddy viscosity and vertical eddy diffusivity) calculated in the Delft3D-FLOW module are used as input to the other modules of Delft3D.4 Coupling to other modules The hydrodynamic conditions (velocities. culverts. bridges. Optional facility for tidal analysis of output parameters. Domain decomposition. curvilinear or spherical). density. Optional facility to calculate the intensity of the spiral motion phenomenon in the flow (e.3 Special features Various options for the co-ordinate system (rectilinear. porous plates and weirs. Built-in switch to run the model in either σ -model or in Z -model. Built-in anti-creep correction to suppress artificial vertical diffusion and artificial flow due to σ -grids.1: System architecture of Delft3D 2.

salinity or concentrations of constituents.6 Installation and computer configuration See the Delft3D Installation Manual. such as bathymetry or initial conditions for water levels.5 Utilities For using Delft3D-FLOW the following utilities are important: module Delft3D-RGFGRID Delft3D-QUICKIN description for generating curvilinear grids for preparing and manipulating grid oriented data. for performing off-line tidal analysis of time series generated by Delft3D-FLOW for performing tidal analysis on time-series of measured water levels or velocities for generating (offline) boundary conditions from an overall model for a nested model for visualisation and animation of simulation results a second tool for visualisation and animation of simulation results Delft3D-TRIANA Delft3D-TIDE Delft3D-NESTHD Delft3D-GPP Delft3D-QUICKPLOT For details on using these utility programs you are referred to the respective User Manual. 2.Introduction to Delft3D-FLOW 2. Deltares 9 .

Delft3D-FLOW. User Manual 10 Deltares .

In Figure 3. Tutorial. Figure 3.1: After a short while the main window of the Delft3D-MENU appears. In this chapter we will guide you through some of the input screens to get the look-and-feel of the program. 3. Each module can be executed independently or in combination with one or more other modules.2. In Chapter 5. Several menu options are shown. you will learn to define and run a simple scenario.3 Getting started 3.2 Starting Delft3D To start Delft3D: On an MS Windows platform: select Delft3D in the Programs menu or click on the Delft3D icon on the desktop. Figure 3. Other. Figure 3. For now. On Linux machines: type delft3d-menu on the command line. each module writes results required by another module to this communication file and reads from the file the information required from other modules. module-specific.2 all options are sensitive. only concentrate on exiting Delft3D-MENU. files contain results of a computation and are used for visualisation and animation of results. Delft3D is provided with a menu shell through which you can access the various modules. Next the title window of Delft3D is displayed. The information exchange between modules is provided automatically by means of a so-called communication file.1 Overview of Delft3D The Delft3D program suite is composed of a set of modules (components) each of which covers a certain range of aspects of a research or engineering problem.1: Title window of Delft3D Deltares 11 . hence: Click on the Exit push button.

Next the Select working directory window. Open the <flow> directory. Remark: In this and the following chapters several windows are shown to illustrate the presentation of Delft3D-MENU and Delft3D-FLOW.2: Main window Delft3D-MENU The window will be closed and you are back in the Windows Desktop screen for PCs or on the command line for Linux workstations. you must select the directory in which you are going to prepare scenarios and execute computations: Click the Select working directory button. 3.4. Browse to and open the <Tutorial> sub-directory of your Delft3D Home-directory. 12 Deltares . User Manual Figure 3.Delft3D-FLOW. to execute a computation in foreground (including online WAVE or online coupling).3. Next the selection window for Hydrodynamics is displayed for preparing a flow input (MDF-)file or wave input (MDW-)file. Figure 3. For Linux workstations the content of the windows is the same.3 Getting into Delft3D-FLOW To continue restart the menu program as indicated above. is displayed (your current directory may differ. On the PC-platform you can set your preferred colours by using the Display Properties. These windows are grabbed from the PCplatform. Before continuing with any of the selections of this Hydrodynamics (including morphology) window. depending on the location of your Delft3D installation). to inspect the report files with information on the execution and to visualise the results: Figure 3. Click the Flow button. but the colours may be different.

Getting started Figure 3.4: Select working directory window Deltares 13 .3: Selection window for Hydrodynamics Figure 3.

User Manual Figure 3.Delft3D-FLOW. In the main Hydrodynamics (including morphology) menu.5: Select working directory window to set the working directory to <flow\friesian tidal inlet> Figure 3. 14 Deltares . also called the “Master Definition Flow” file (MDF-file) which contains all information to execute a flow simulation. Figure 3. In this guided tour through Delft3D-FLOW we limit ourselves to inspecting some windows of the FLOW Graphical User Interface (GUI).6: The current working directory is not shown in the title bar due to its length Enter the <friesian tidal inlet> sub-directory and close the Select working directory window by clicking OK. but now the changed current working directory is displayed in the title bar (if the name is not too long). see Figure 3. execute and visualise a scenario.6. The FLOW-GUI is loaded and the primary input screen is opened. The purpose of this FLOW-GUI is to create the input file of Delft3D-FLOW. you can define. Hence: Click on Flow input.3. Figure 3.7. Remark: In case you want to start a new project for which no directory exists yet. you can select in the Select working directory window to create a new folder.5. see Figure 3. Next the Hydrodynamics (including morphology) window is re-displayed.

8: Menu bar of the FLOW-GUI Deltares 15 .7: Main window of the FLOW Graphical User Interface Figure 3.Getting started Figure 3.

7 the Description text box is displayed in the canvas area.7 is displayed with the Data Group Description selected and displayed. tool to change table oriented data by adding or deleting rows or values. or listboxes to define the various kinds of input data required for a simulation. For example.4 Exploring some menu options The menu bar of the FLOW-GUI displays four options: File Table View About select and open an MDF-file. to save the MDF-data under a new name. Each option provides one or more selections.9: File drop down menu 3. Figure 3. to open an existing MDF-file. save an MDF-file under a different name. will result in the window shown in Figure 3. Though several input items are 16 Deltares . These Data Groups are represented by the large grey buttons at the left of the main window. The input parameters that define a hydrodynamic scenario are grouped into Data Groups. You are encouraged to explore the various Data Groups and sub-windows to get a first impression on the items the Data Groups are composed of. to save all attribute data under a new name + MDF-file. for instance. Upon starting the FLOW-GUI. Click on a Data Group and see what happens. The area to the right of the Data Groups is called the canvas area. to exit the FLOW-GUI and return to the Hydrodynamics (including morphology) window. tables. clicking on File enables the selections: New Open Save MDF Save MDF As Save All Save All As Exit to clean-up the internal data structure and start with a new scenario. to save the MDF-data under its current name. save attribute files or ‘exit’ the FLOW-GUI. In Figure 3. clicking the Domain button and next the sub Data Group Grid parameters. About information. This canvas area will be dynamically filled with input fields. visualisation area or list of attribute files. save an MDF-file. The Tutorial in Chapter 5 will make you become fully acquainted with the various input windows that result from this main window. to save all attribute data in the current attribute files + MDF-file.10. User Manual Figure 3.Delft3D-FLOW.

Getting started Figure 3. during this introduction you can neglect these messages and press the Ignore button if requested. Deltares 17 . Occasionally you might get a warning or error message that some data is not saved or not consistent with earlier defined data.10: Data Group Domain selection and input fields Figure 3.11: Save changes window related there is no fixed or prescribed order in defining the input data. No harm will be done on existing input files as you are not going to save the input data of this exercise.

don’t save the first unsaved data item. abort this Exit action and return to the FLOW-GUI. Ignore all other options and just: Click Return to return to the main window of Delft3D-MENU.5 Exiting the FLOW-GUI To exit the FLOW-GUI: From the File menu. proceed to the next unsaved data. the unsaved attribute files will be listed. User Manual 3. save all unsaved data items and request a file name if not yet defined. if you have changed data that must be saved into a so-called attribute file. Figure 3. 18 Deltares . In this case only the MDF-data was not saved. We encourage new users next to run the tutorial described in Chapter 5. Select one of the options displayed: Yes YesAll No NoAll Cancel save the first unsaved data item. select Exit.Delft3D-FLOW. If you have made any change to any input field and have not explicitly saved both the attribute data and/or the MDF-data Figure 3.3. request for a file name if not yet defined and proceed to the next unsaved data item.2. Figure 3. Neglect any unsaved data and exit: Click NoAll. exit without saving any unsaved data item. In this Getting Started session you have learned to access the FLOW-GUI and to load and inspect an existing input (MDF-)file. Click Exit.11 is displayed. The window is closed and the control is returned to the desktop or the command line. You will be back in the Hydrodynamics (including morphology) window of the Delft3DMENU program.

it is likely that the solution is sensitive to the selected parameter values.1 Introduction In order to set up a hydrodynamic model you must prepare an input file. Although you are not supposed to work directly on the MDF-file it is useful to have some ideas on it as it reflects the idea of the designer on how to handle large amounts of input data and it might help you to gain a better idea on how to work with this file. In Section 4. is called the Master Definition Flow file or MDF-file. time-dependent or space varying data). In Section 4. Being an ASCII-file. Currently. including their restrictions and their valid ranges or domain.2 MDF-file and attribute files The Master Definition Flow file (MDF-file) is the input file for the hydrodynamic simulation program. it can be transported to an arbitrary hardware platform. Each line contains a maximum of 300 characters. If you have not used Delft3D-FLOW before. Within the range of realistic values. or if you are not familiar with the FLOW Graphical User Interface (GUI) we suggest you to execute the tutorial given in Chapter 5 first and then return to this chapter. removing and exporting of data and their references in the MDF-file. The input data defined is stored into an input file which. This will be particularly the case when parameters contain a large number of data (e. In Section 4.4 we discuss working with the FLOW-GUI and the Visualisation Area window. so a concise description of all parameters is required. The MDF-file and all possible user-definable attribute files are listed and described in Appendix A. In Section 4. Each (set of) input parameter(s) is preceded by a (set of) keyword (s). but you can easily write a couple of scripts to run your pre. In many cases we give a short discussion on the criteria to determine a parameter or to select a certain formulation. Delft3D does not support remote or distributed processing. The MDF-file is an intermediate file between the FLOW-GUI and the hydrodynamic simulation program. It contains all the necessary data required for defining a model and running the simulation program. All parameters to be used originate from the physical phenomena being modelled.g. Consequently.7 the aspect of importing. we discuss in Section 4. In Section 4. and finally.3 we discuss shortly the filenames and their extension. Finally. such as the turbulence closure model for the vertical turbulent eddy viscosity and turbulent eddy diffusivity.2 we discuss some general aspects of the MDF-file and its attribute files.6 we discuss saving the MDF-file and exiting the FLOW-GUI.and Deltares 19 . In the MDF-file you can define attribute files in which relevant data (for some parameters) is stored. the hydrodynamic simulation program and the FLOW-GUI do not necessarily have to reside on the same hardware platform. from decisions being made to control the simulation and to store its results. Also from the numerical techniques being used to solve the equations that describe these phenomena.4 Graphical User Interface 4. as you may recall.5 we discuss all input parameters. 4. The basic characteristics of an MDF-file are: It is an ASCII file.

Result files The results of a Delft3D-FLOW computation are stored in several types of files: communication file: <com-runid. The MDF-file only contains permanent input parameters and references to these attribute files.dat>.Delft3D-FLOW. its length is restricted to 300 columns.def> and <trim-runid. MDF-file The name of an MDF-file must have the following structure: <runid.mdf> may consist of up to 256 alpha-numeric characters and may not contain blanks.def> and <trid-runid.def> and <com-runid.dat>. restart files <tri-rst.time>. you must make sure that the correct input formats are used. Restrictions: 20 Deltares . This file could indicate the flow-input file of the first calibration run of a project. The results of all modules are written to platform independent binary files. When you have many computations we suggest to use a combination of one alpha.date. 4.mdf>.3 Filenames and conventions The names of the MDF-file and its attribute files have a specific structure. since supplying long time-series manually is not very practical.dat>.def> and <trih-runid. some aspects are mandatory while others are only advised or preferred. To maintain a good overview of the file. It can therefore be used as model archive.dat>. drogue file: <trid-runid.and two numeric-characters followed by a useful name of your project. As you will see in Chapter 5. so also these result files you can transfer across hardware platforms without any conversion. User Manual postprocessing on one hardware platform and run the computational intensive simulation at an other hardware platform.mdf>. The result files are stored in the working or project directory. history file: <trih-runid. after having specified certain types of input parameters you can store them in attribute files. Example: <r01-calibration.e. The formats of all attribute files (and of the MDF-file itself) are described in detail in Appendix A.runid. If you wish to create attribute files in advance. i. The <runid. An overview of the attribute files is given in Section 4. The runid part of the filename is used as a run-id in the names of the result files to safeguard the link between an MDF-file and the result files. Contact our support manager if you need remote or distributed computational functionalities.3. The MDF-file is self contained. map file: <trim-runid. it contains all the necessary information about the model concerned.

With respect to the parameters in the MDF-file. The history file contains results of all computed quantities in a number of user-defined grid points at a user-defined time interval. i. in this section we briefly recapitulate how to work with the FLOW-GUI (alternatively. The name of an attribute file must have the following structure: <name. <name-number. Click the menu item FLOW. Change to your project or working directory. use an underscore instead.mdf> instead of <r01 test.e. when you have many computations we suggest to use a character followed by a two digit number. To start the FLOW-GUI you must in short execute the following commands. i. The extension is mandatory as indicated in Table 4.e. There is no limitation other than the platform dependent limitations. Click the menu item Flow input .extension>. but their extension is mandatory as indicated in Table 4. you are referred to your hardware platform manual for details. The map file contains results of all quantities in all grid points at a user-specified time interval. 4. the full path. <r01 test.4.1. Attribute files Attribute files contain certain input quantities. Where: <name.4 4. such as the water quality module. The communication file contains results that are required by other modules. Avoid spaces in a filename.Graphical User Interface Each scenario must have a unique run-id.extension>. An attribute file can be created by the FLOW-GUI if the required data was specified. such as monitoring points or time dependent input data such as wind. the FLOW-GUI follows either one of the following options: A single parameter is updated and included in the MDF-file. the FLOW-GUI will be started and the main window will Deltares 21 . For your convenience.extension> may consist of up to 256 alpha-numeric characters and may contain (sub-) directories. We suggest to add some continuation character. i.1. for instance -number to the name to distinguish between various updates or modifications of the file.e. The names of the main attribute files are basically free. The extension mdf is mandatory. consult Chapters 3 and 5). A reference to an attribute file is updated and included in the MDF-file.mdf>. see Chapter 3 for details: Click the Delft3D-MENU icon on the desktop (PC) or execute the command delft3d-menu on the command line (Linux and UNIX).1 Working with the FLOW-GUI Starting the FLOW-GUI The purpose of the FLOW-GUI is to provide a graphical user interface which simplifies the preparation of an MDF-file.

tem> <name.bcq> <name.rgh> <name.enc> <name.bct> <name.sed> <name.par> <name.fou> <name.src> <name.Delft3D-FLOW.bch> <name.dad> <name.bca> <name.thd> <name.crs> <name. User Manual Table 4.edy> <name.cor> <name.ini> <name.dry> <name.obs> <name.bnd> <name.1: Overview of the main attribute files Quantity Astronomic correction factors Bathymetry or depth Bottom roughness Constituents boundary conditions Cross-sections Curvilinear grid Discharge locations Discharges rates Dredge and dump characteristics Drogues or floating particles Dry points Flow boundary conditions (astronomic) Flow boundary conditions (harmonic) Flow boundary conditions (QH-relation) Flow boundary conditions (time-series) Fourier analysis input file Grid enclosure Horizontal eddy viscosity and diffusivity Initial conditions Morphology characteristics Observation points Open boundaries Sediment characteristics Temperature model parameters Thin dams Wind Filename and mandatory extension <name.grd> <name.dep> <name.dis> <name.bcc> <name.mor> <name.wnd> 22 Deltares .

For viewing the grid related parameters or for listing the attribute files used (only their referenced name. Save All.1. Save MDF As. Save MDF. Open. or saving an MDF-file with another name. In the menu bar you can choose from the following options: File For opening and saving an MDF-file. You are now ready to start defining or modifying all input parameters grouped into so-called data groups. Sub-menus are: Contents and About. The first is not implemented yet. for cleaning up the internal data structure and for exiting the FLOW-GUI. For getting online and context sensitive help. Copy row above. for saving attribute files under the same name or under a new name. not their contents). Save All As and Exit. Delete row and Copy value to all rows. Table View Help Deltares 23 . see Figure 4. Sub-menus are: Insert row above.Graphical User Interface Figure 4.1: Main window of the FLOW-GUI be opened. Sub-menus are: New. To change attribute files by adding or deleting rows or values in table quantities. Sub-menus are: Visualisation Area and Attribute files.

To set the font. The main and sub-menus of the Visualisation Area window are: File To Open some of the grid defining files. To Zoom In and Zoom Out the whole visualisation area. size. to print (Print area) the Visualisation Area window or Exit (close) the Visualisation Area window.3. Zoom Box in a user-defined area and Zoom Reset to return to the initial situation. observation points. cross-sections and open boundaries. These grid related data are: dry points. You can activate (display) or de-activate (hide) the various attributes. discharges.4.2 Visualisation Area window Most grid related data specified in the MDF-file can be visualised and defined in the Visualisation Area window. Modify (move) and View the quantities selected in Edit. To Add. To select one of the grid related quantities that can be visualised. To select which quantities will be displayed in the visualisation area and to refresh the display.2: Visualisation Area Window 4. Delete. User Manual Figure 4. To set the colours for visualising the bathymetry. but after you have defined or selected a grid it will display the grid and several grid related quantities. The visualisation area is still blank. thin dams. For online information of using the Visualisation Area window.2. drogues. Deltares Edit Edit Mode Zoom View Fonts Colors Options Help 24 . The various selections of View → Attributes are displayed in Figure 4. of the attribute names. Upon selecting View → Visualisation Area the Visualisation Area window will pop up. To switch on or off viewing attributes and/or attribute names.Delft3D-FLOW. etc. Figure 4.

without having to type locations manually. Make sure the Edit Mode is selected properly. You can save the data in an attribute file before you change the quantity to be worked on.3: Possible selections of View → Attributes Remarks: The Edit and Edit Mode options make it possible to graphically define.Graphical User Interface Figure 4.4 is displayed. Deltares 25 . modify (move) or just view quantities of a certain type. Remark: Symbols are grouped in colours. but you can also postpone this until the end of your input session. delete. if not you might accidentally move a quantity without notice! To see how each quantity is represented in the Visualisation Area display the legend: Select View → Legend and next Figure 4. This allows you to define (Add) or delete (Delete) all kind of quantities without having to set and reset the Edit Mode. The Edit Mode shifts back to View mode as soon as you leave the Visualisation Area window. to prevent unintended modifications of your grid related quantities. If a quantity can have a user-defined name you can fill that in after having defined its location graphically. but with different representation to support easy recognition. The Edit Mode remains in its selected mode as long as you are working in the Visualisation Area window.

5 Input parameters of MDF-file In this section all input parameters of the MDF-file will be described in the order as they appear in the FLOW-GUI.4: Display symbols of all grid related quantities 4. Dry points and Thin dams. In many cases we add a more comprehensive discussion to put the quantity and its use in perspective. For instance. We will now describe all data groups in consecutive order. in the Data Group Operations you can define all aspects related to a discharge. its location. For each input quantity we give: A short description of its meaning. A data group is a coherent set of input parameters that together define a certain type of input data. such as its name. 26 Deltares .Delft3D-FLOW. that consists of four sub-data groups: Grid parameters. called its domain. if the momentum of the discharge is to be taken into account and if so in which direction and last but not least the concentration of all substances released. Remark: Before you can define grid related quantities. where the time-series can be specified manually or read from a file. When editing an existing input file there is no preferred or mandatory order in which to address the various data groups. such as the Data Group Domain. its discharge rate. Several of these items can be specified as a function of time. User Manual Figure 4. The restrictions on its use. and its default value (if applicable). The range of allowed values. Bathymetry. After starting the FLOW-GUI data groups become available for defining or changing the input parameters in the MDF-file. you must define or select the grid structure. Some data groups are organised in sub-data groups.

the velocity components or concentration of substances. Domain: Parameter Descriptive text Lower limit Upper limit Default Empty lines Unit none Any printable character Restriction: 10 lines of text each containing a maximum of 30 characters. The staggered grid applied in Delft3D is given in Figure 4.5 is displayed. are defined at the same location in the numerical grid (and thus in the physical space).2. If you started from an existing MDF-file its name is displayed above in the title bar.6.1 Description The Data Group Description is a text box of up to 10 lines of text.5. Closed boundaries are defined through u.2 Domain The Data Group Domain contains the following sub-data groups: Grid parameters. Deltares 27 .Graphical User Interface Figure 4. Upon selecting the data group Figure 4. Dry points and Thin dams. such as the water level.5. In a staggered grid not all quantities.5. which you can use to describe the purpose of the present model and for discriminating the present run from the (possibly) other runs with the same model. 4. the depth. Before continuing with discussing the Grid parameters you should familiarise yourself with the concept of the staggered grid applied in Delft3D-FLOW. The description is only used for reference. open boundaries through either u-. 4. the latitude of the model area and the number of (vertical) layers.or v -points. Bathymetry.1 Grid parameters In the sub-data group Grid parameters you specify the grid used.5: Data Group Description 4.

and the co-ordinates of the grid points. The type of co-ordinate system is stored in the grid file. concentration of constituents.and n-direction) • depth below mean (still) water level (reference level) Figure 4. 28 Deltares . Typically. because the grid must fulfill the following criteria: It must fit as closely as possible to the land-water boundaries (in short land boundaries) of the area to be modelled. User Manual Legend: full lines the numerical grid grey area items with the same grid indices (m. The grid spacing must vary smoothly over the computational region to minimise inaccuracy errors in the finite difference operators. The construction of a suitable curvilinear grid is not a simple task. this will be used to calculate a fixed Coriolis force for the entire area. temperature − horizontal velocity component in ξ -direction (also called u. such as a regional model. together with the number of grid points in both directions. grid lines must intersect perpendicularly. For a spherical grid the Coriolis force is calculated from the latitude co-ordinates in the grid file and thus varies in the latitude direction. It must be orthogonal.6: Staggered grid of Delft3D-FLOW v . The location of other grid related quantities.or water level (ζ -) points depending on the type of boundary condition such as velocity or water level.and m-direction) | horizontal velocity component in η -direction (also called v . such as discharges and observation points are given when appropriate.Delft3D-FLOW. In Delft3D-FLOW we support two types of co-ordinate systems in the horizontal: Cartesian: the co-ordinates are in metres Spherical: the co-ordinates are in decimal degrees For a Cartesian grid you have to specify the latitude of the model area. i. n) + water level. salinity.e. you use spherical co-ordinates for large areas.

A measure for the orthogonality is the angle. Remark: Flooding and drying is less accurate if the grid size increases. A grid enclosure is a closed polygon specified on a grid through the water level points. To resolve an important geometrical or hydrodynamic phenomenon you will need at least 5 grid cells.7. wet) computational cells in the com- Deltares 29 . see Figure 4.enc> file that belongs to your grid file. and open a <∗. as well as the number of grid points in both directions. see Figure 4. delete or add locally individual grid cells. A file window opens in which you can browse to the required directory. the co-ordinate system used is displayed.7. working from a coarse version of the grid to finer versions until the required resolution is achieved.02. Opening a grid and enclosure file Select Grid parameters → Open grid. such as refine or de-refine the grid globally or locally. Its purpose is to define the active or potentially active (i. define separately a grid in a sub-area and glue it to the overall grid and orthogonalise the grid. A measure for the grid smoothness is the aspect ratio of grid cells (ratio of the grid cell dimension in ξ . Delft3D-RGFGRID provides all kind of features to develop a grid. where ϕ is the angle between the grid lines.4 far away. to resolve a horizontal circulation the grid size should be 1/10th or less of the size of the circulation. or the cosine of the angle. You can use Delft3D-RGFGRID to inspect these and other properties graphically. For details see the Delft3D-RGFGRID User Manual. unless the flow is predominantly along one of the grid lines. The quality of a grid is to a large extent determined by its orthogonality and the rate with which certain properties change over the area to be modelled (smoothness). After the grid file has been opened. The actual construction of a grid is realised in an iterative procedure allowing for a stepwise generation of the model grid. and open the <∗.and η -direction) and the ratio of neighbouring grid cell dimensions. Delft3D-RGFGRID provides features to inspect the quality of the grid.2 in the area of interest up to 1.and η -direction. Next Select Grid parameters → Open grid enclosure. Aspect ratio: must be in the range [1 to 2].Graphical User Interface The Delft3D modelling suite contains the grid generator program Delft3D-RGFGRID that allows you to generate a curvilinear grid (in Cartesian or spherical co-ordinates) with the required resolution and properties. As a guideline we suggest the following overall quality criteria: Orthogonality: cos(ϕ) < 0. between the grid lines in ξ . Ratio of neighbouring grid cells: should be less than 1. A file window opens in which you can browse to the required directory.e.grd> file. The horizontal resolution of the grid depends on the characteristic length scale of the bathymetry and the land-water boundary and of flow patterns you want to resolve.

Delft3D-FLOW. Remarks: We strongly suggest generating the grid and the grid enclosure in all cases with the grid generator program Delft3D-RGFGRID. If no grid enclosure is specified a default polygon is generated through the four corner points of the numerical grid. However. you must be aware of certain details related to definition and implementation.2.3 related to the file structure and content and in Appendix E for a detailed discussion of the grid enclosure and the location of open and closed boundaries.7: Sub-data group Grid parameters putational domain and the location of the open and closed boundaries. its use is under all practical conditions completely transparent to you and you do not need to be aware of the definition and implementation details. 30 Deltares . such as the grid file or the bathymetry file and want to check all details in relation to values and locations. if you want to inspect certain files. in Appendix A. User Manual Figure 4. The grid enclosure is generated by the grid generator Delft3D-RGFGRID. You can find these details in Chapter 10 as far as related to the numerical scheme.

The layer thickness at the top is however determined by the actual water level and at the bottom by the local topography. The angle is positive if the rotation is clockwise. an additional input field Layer thickness is displayed.e. you must specify the thickness of the layers in the vertical. For a 3D computation. Remark: The Coriolis force in a spherical model varies in the North-South direction and is determined by the actual latitude. the angle of latitude (in degrees North). two types of vertical grid with distinctive layer thickness characteristics are supported: With the σ -grid in the vertical the layer thickness varies with the depth. In the vertical direction. i. The Coriolis force is determined by the location of the model area on the Earth’s globe.9. the Number of layers is larger than 1. and the number of active layers is constant (denoted in this manual as the σ -model) and Z -grid. see Figure 4. see Figure 4.8. Number of layers For a depth-averaged computation you must set the Number of layers to 1.Graphical User Interface Figure 4. Orientation The model orientation is defined as the angle between the true North and the y-axis of the Cartesian co-ordinate system. Upon setting the Number of layers larger than 1.8: Definition sketch grid system to North orientation Latitude and orientation of the model area If the grid is defined in Cartesian co-ordinates you have to specify the latitude and orientation of the model: Latitude Latitude location of the model on Earth. here the layer thickness is fixed and the number of active layers varies with the depth. The model using this grid is referred to as the Z -model. Furthermore. In the northern hemisphere you must enter a positive value. in the southern hemisphere you must enter a negative value. Deltares 31 .

e. To resolve the logarithmic profile of the horizontal velocity components in the vertical the thickness of the bottom layer should be small.5 %. For a ten σ -layers example the suggested layer thickness is {2.4. of the water depth. 10 and 12 for more details of the vertical co-ordinate system. Remark: In the σ -model layer 1 corresponds to the surface layer while in the Z -model layer 1 refers to the bottom layer. 6. You can specify an arbitrary number or distribution as long as the total sum of the layers is 100 %.7 to 1. water depth for the σ -model. i. 20} %. You can specify an arbitrary number or distribution as long as the total sum of the layers is 100 %. The thickness of a layer is defined as: a percentage of the initial water depth for the Z -model and a percentage of the. An indicative value for the variation-factor for each layer is 0. It is recommended to choose the bed layer thickness to be about 2 % of the water depth.75 %. 4. 6. 15. Thus.75 % etc. 10. going down from the surface the layer thickness should not exceed 3 %. 3. The variation in the layer thickness should not be large. 6. 20. 4.9: Specifying the layers thickness To achieve a constant number of layers in the σ -model. of the water depth. Domain: 32 Deltares . If wind is one of the important driving forces also the layer thickness near the surface should not exceed 2 % of the water depth. 4. etc. 8. User Manual Figure 4.Delft3D-FLOW.5 %. a σ co-ordinate transformation in the vertical is used. A similar argument holds when you want to resolve sharp density gradients in the vertical: you must have a sufficiently fine grid to resolve the vertical profile. See Chapters 9. Going from bottom to surface the suggested layer thickness should not exceed 3 %. the layer thickness must have a smooth distribution. 12. time varying.

Upon selecting File an Open button is displayed. 4. A nonuniform (space-varying) bathymetry is given in an attribute file with the extension <dep>. The maximum number of layers is 100.0 360.10.0 100 100. To enter a uniform depth. The grid enclosure (polygon) must be closed. For a uniform bathymetry the sub-window is given in Figure 4.5.0 Default 0. Domain: Deltares 33 .10: Sub-data group Domain → Bathymetry Parameter Latitude Orientation Number of layers Thickness Lower limit -90. To enter a uniform depth value for the whole model area. Remark: The reference level of the depth is a horizontal plane. A line segment may not intersect or touch another line segment.0 0.Graphical User Interface Figure 4. To open and read the bathymetry file with extension <dep>.0 1 0.2. Upon selecting Uniform an input field Depth is displayed.0 Unit degrees N degrees % total depth Restrictions: The direction of the line segments in the polygon of the grid enclosure must form a multiple of a 45 degree angle with the numerical grid axis.01 Upper limit 90. see Appendix A for its file format.2 Bathymetry The depth schematisation may be uniform or non-uniform across the model area. A negative value defines a depth above the reference plane.0 1 100. You can either select: File Open Uniform Depth To specify a file with bathymetry data.0 0. The sum of the layer thickness must be equal to 100 %.

However. this does not mean that the best bathymetry is obtained by always assigning the actual depths to all grid points. The interpolation of these data to the depth points of the grid should result in a bathymetry that resembles the natural bathymetry as closely as possible. Thus. you must be careful when combining depth data originating from different sources (different chart datums).000 Default 10. User Manual Parameter Depth Lower limit -1. before deciding to either include or discard it.0 Unit m The task of assigning depth values to grid points can be split in two main components. The triangulation network is designed in such a way. In the opposite situation. nor may they cover the complete area of the grid. In this way. the integral bathymetric features are best accounted for. Using available measurements (echo-soundings). The resulting bathymetry on the numerical grid is shown by way of iso-lines. Extracting the bottom schematisation of the area to be modelled from the bottom schematisation of an overall coarser hydrodynamic model. Corrections might be required to ensure that all depth values refer to only one reference level. thus spoiling interpolation results that might have been good if properly dealt with. Since the FLOW module calculates averaged flow velocities and water levels. you must carefully evaluate the quality of the various bathymetric data sets. a maximum correlation between the numerical bathymetry and known bathymetric data points is then obtained. The first one is the gathering of the raw bathymetric data. that minimum triangle side lengths are achieved. You can obtain the bathymetric data by: Digitising bathymetric charts (Admiralty Charts. 34 Deltares . when there are less data points than grid points in a given area. The combined bathymetric data may not all be of the same resolution. you will apply some kind of interpolation method on a triangulation network. Therefore.000 Upper limit 20. These options can be combined to obtain the most elaborate depth data. Remark: Do pay special attention to the reference levels of different raw bathymetric data sources. Fair Sheets). However. the second one is the actual interpolation of these raw data on the structured grid. If all data are simply stacked into one file. equality of averaged bathymetric features is more important than equality of bathymetric features at discrete grid points. neither of the same quality with respect to accuracy. You can correct interactively depth values of individual samples or grid points.Delft3D-FLOW. there will be the problem that high quality data becomes contaminated with low quality data. You can apply the utility program Delft3D-QUICKIN that enables you to select a sequence of data files and to control the interpolation areas and the interpolation method. Hence. you should adopt a volume-preserving interpolation method that uses all data points if there is redundancy of data in a given grid cell.

This can produce many (short wave length or wave period) disturbances (noise) in the simulation results. n − 1) and (m − 1. In Figure 4. If the bottom has steep gradients across a large area on a tidal flat. It may generate high frequency disturbances in the results. Dry points are specified as a line of dry points. n − 1). You can specify dry points either manually. (m − 1. n) Bathymetry in relation to drying and flooding Drying and flooding is a discontinuous process. a large area may be taken out of the flow domain in just one half integration time step. 4. n). (m. irrespective of the local water depth and without changing the water depth as seen from the wet points.11 a single dry point is defined at location (m. In general. n) grid indices. Finally.11: Dry point at grid location (m. You can apply one or more of the following options: Deltares 35 . and for discharge points near tidal flats in particular. n) remain unchanged. You should also pay attention to the fact that depth values at points near closed boundaries are used in the drying and flooding procedure.5.e. at (m. Upon selecting the Data Group Dry points the sub-window of Figure 4. n). i. you must ensure that the topography for points near tidal flats in general. As a result the water depth in the surrounding water level points (+) are not influenced by the presence of the dry point. For details of the numerical aspects of drying and flooding see Chapter 10.Graphical User Interface Figure 4. Finally.12 is displayed. You can avoid this by smoothing the bottom gradients. you must avoid drying and flooding at open boundaries.2.e. a single dry point is specified as a line of unit length. Dry points are characterised by their (m.3 Dry points Dry points are grid cells centred around a water level point that are permanently dry during a computation. The disturbances are small if the grid size is small and the bottom has smooth gradients. as defined by the bathymetry. the magnitude of the disturbance will depend on the grid size and the bottom topography. graphically in the Visualisation Area window or by reading from an attribute file (file extension <dry>). are appropriately schematised to prevent the cells to be set wet and dry at each integration time step. i. The depth at the corner points.

User Manual Figure 4. In reverse order the manually defined dry points are overwritten by those of the file.e. To save all dry points in the same or a new attribute file. n2). or Use the Visualisation Area window. i. Specify the grid indices of the begin and end points. n1) and (m2. n) Lower limit Upper limit Default none Unit none Anywhere in the computational domain 36 Deltares .4. Open Save To read dry points from an attribute file with extension <dry>. Remark: You can save the dry points here or postpone it to the end of the input definition when you save the MDF-file. (m1.Delft3D-FLOW.12: Sub-data group Dry points Add To add one or a line of dry points either: Click Add. Click Delete. Remark: If you want to combine dry points read from a file and inserted manually you must read the file first and then add the manually defined dry points. or Use the Visualisation Area window (see Section 4.2 on how to use the Visualisation Area window). Domain: Parameter Indices (m. Delete To delete a single dry point or a line of dry points either: Select the dry points to be deleted in the list box.

Thin dams are specified as a line of thin dams. n2).Graphical User Interface Figure 4. they only differ in the direction. (m1. with the computational grid axis.4 Thin dams Thin dams are infinitely thin objects defined at the velocity points which prohibit flow exchange between the two adjacent computational cells without reducing the total wet surface and the volume of the model.or v -direction). N) Restrictions: Dry points may only be specified along line segments which form a 45 degrees angle. The purpose of a thin dam is to represent small obstacles (e. If you need to apply a different depth on both sides you cannot apply a thin dam. Thin dams can be specified either manually or via an imported file with mask <thd>.5. The line of thin dams is defined by its indices of begin and end point. You can apply one or more of the following selections: Add To add a single or a line of thin dams either: Click Add. 4. If dry points are read from file and defined manually.14 is displayed. Upon selecting Thin dams the sub-window of Figure 4. (m1. Note that these thin dams have the same grid indices. and the direction of thin dam (u. n1) and (m2. breakwaters. i. dams) in the model which have sub-grid dimensions. Deltares 37 . but large enough to influence the local flow pattern.13 the location is shown for three single u-thin dams (left) and a line of three v -thin dams (right).13: Equivalence of v -type thin dams (left) and u-type thin dams (right) with the same grid indices. but you should use a line of dry points instead. Remark: Thin dams separate the flow on both side. but they do not separate the bathymetry on both sides.2. Specify the grid indices of the begin and end point. the file must be read first. (M−1 to M+1.e. n1) and (m2. Depth points are located at the thin dam and so this depth is used on both sides of the thin dam. In Figure 4. n2). or the multiple of it.g. respectively. a single thin dam is specified as a line of unit length. Upon saving all dry points are stored in the (new) attribute file.

n) Lower limit Upper limit Default none Unit none Anywhere in the computational domain Restriction: Thin dams can only be specified along lines parallel to one of the numerical grid axes 38 Deltares . Open To read thin dams from an attribute file with extension <thd>.4. see Section 4.or v direction. In reverse order the manually defined thin dams are overwritten by those of the file. or Use the Visualisation Area window. A thin dam can either be defined as blocking the flow in u.Delft3D-FLOW. select either direction. Click Delete. Save Direction of Thin Dam Domain: Parameter Indices (m.4. Delete To delete a single or a line of thin dams either: Select the thin dam to be deleted in the list box. Remark: You can save the thin dams here or postpone it to the end of the input definition when you save the MDF-file. User Manual Figure 4. Remark: If you want to combine thin dams read from a file and inserted manually you must read the file first and then add the manually defined thin dams. see Section 4.2.14: Sub-data group Thin dams or Use the Visualisation Area window.2. To save the thin dams to the same or a new attribute file.

15.15: Data Group Time frame or along lines which form a 45 degrees angle with the numerical grid axis. The start date and time of the simulation. but in the simulation a time is determined by its number of time steps after the simulation reference date at time [00 00 00]. The reference date defines the (arbitrary) t = 0 point for all timeseries as used in the simulation. This reference date is also given in the header of files containing time-series.21. Upon selecting the Data Group Time frame the sub-window given in Figure 4. Thin dams aligned along an open boundary section are strongly discouraged. in most cases stability is not an issue. If done so these thin dams do not make sense anymore.2. The accuracy is.Graphical User Interface Figure 4. Generally. such as boundary conditions of type Time-Series. 4. or flow rate and concentrations of discharges. By default the reference date is set equal to the current date. Thin dams perpendicular to open boundaries are allowed. All time-series are specified in minutes after this t = 0 point. such as the reproduction of the Simulation start time Simulation stop time Time step Deltares 39 . All time dependent input is defined by the date and time as [dd mm yyyy hh mm ss]. The time step used in the simulation in minutes. Appendix A. Remarks: Defining thin dams at the grid boundaries (other then open boundaries) does not make sense since these boundaries are already closed per definition. see the first remark.2. You can better subdivide such a section in sub-sections excluding the thin dams.3 Time frame In the Data Group Time frame you define the relation between the time axis of the real world and that of the simulation. among several other parameters.15 is displayed. You must specify the following input data: Reference date The reference date of the simulation. The stop date and time of the simulation. Appendix A.5. you can choose the time step based on accuracy arguments only.

|v |) (4.2) ∆t ≤ min (∆x. Which of the limitations is most restrictive is dependent on the kind of application: length scale. with or without densitycoupling etc. The local Time Zone is used for for two processes: Deltares 40 . dependent on the Courant-Friedrichs-Lewy number (CFL). Generally.3) Local time zone Remark: You should check the influence of the time step on your results at least once. and {∆x.4 for a discussion on the numerical scheme used in Delft3D-FLOW in relation to stability and accuracy. For your convenience we repeat the summary of that section.Delft3D-FLOW. ∆y } (4.1) where ∆t is the time step (in seconds). ∆y ) max (|u|. the Courant number should not exceed a value of ten. velocity scale.2: Time step limitations shallow water solver Delft3D-FLOW Points per wave period T Accuracy ADI for barotropic mode for complex geometries Stability baroclinic mode Explicit algorithm flooding Stability horizontal viscosity term ∆t ≤ 1 40 T 1 ∆ x2 Ct = 2∆t gH 2∆t u∆t ∆x ∆ρ ρ gH + 1 ∆y 2 √ <4 2 1 ∆ x2 + 1 ∆y 2 <1 <2 1 ∆ x2 ∆tνH + 1 ∆y 2 <1 important spatial length scales by the numerical grid. ∆y } is a characteristic value (in many cases the minimal value) of the grid spacing in either direction. but for problems with rather small variations in both space and time the Courant number can be taken substantially larger. User Manual Table 4. For the Z -model the following additional limitations apply: ∆t ≤ and min (∆x. The time difference between local time and UTC. ∆y ) ∆ρ ρ gH (4. The time zone is defined as the time difference (in hours) between the local time (normally used as the time frame for Delft3D-FLOW) and Coordinated universal time (UTC). g is the acceleration of gravity. Let ∆x and ∆y be horizontal grid sizes. defined by: √ ∆t gH CF L = {∆x. H is the (total) water depth. See Section 10. In the time step limitations are given for the shallow water code Delft3D-FLOW. For a curvilinear grid ∆x = Gξξ and ∆y = Gηη .

0 then the simulation time frame will refer to UTC. this enables the process Dredging and dumping.0 12. wind velocities and atmospheric pressure. To run FLOW Online with WAVE you have to write the communication file. upon which Figure 4. These can be specified in a different time zone.5. The simulation stop time must be equal to or larger than the simulation start time. such as Initial conditions or Boundaries. and you have to prepare an MDW-file prior to execute these simulations Secondary flow is only displayed for depth averaged computations using the σ -model. If the Local Time Zone = 0. Remarks: To include Waves (offline) you need a communication file with wave data. In Figure 4. To compare the local time of the simulation with the times at which meteo input is specified. the parameters required for these processes are defined in other Data Groups. Here you only define which processes you are going to apply.0 0.Graphical User Interface To determine the phases in local time of the tidal components when tide generating forces are included in the simulation.0 1.16 the process Sediments has been selected. e. Domain: Parameter Date Time 00 00 00 23 59 59 Lower limit Upper limit Default Current date 00 00 00 Unit none hours minutes seconds minutes hours Time step Local Time Zone 0.0 Restrictions: If open boundaries are used with forcing type Astronomic then dates may not be before 1 January 1900. Click the Data Group Processes.16 is displayed. see Data Group Processes. The start and stop date must be equal to or larger than the reference date.g. The time step must be positive. but not for the Z -model Tidal forces is only displayed if the grid is in spherical co-ordinates. Deltares 41 .0 999. 4.0 -12. The start and stop time must be integer multiples of the time-step.4 Processes In the Data Group Processes you specify which processes or quantities that might influence the hydrodynamic simulation are taken into account.

This functionality includes the transport of suspended sediments (cohesive and non-cohesive). a horizontal and vertical density gradient may occur which induces density driven flows. bedloads and optionally updating the bathymetry.Delft3D-FLOW.17.2 for a description on how to use this option. Salinity Temperature The salinity concentration will be taken into account including its influence on the water density and other processes. see Figure 4. see the sub-data group Physical parameters → Heat flux model. As a consequence. Furthermore. If Pollutants are selected a window appears to let you specify the names of the constituents applied. The density is calculated at cell centres. the tabs Sediment and Morphology are available in the Data Group Physical parameters where you can define the sediment Deltares Sediments 42 . Several heat flux models are available. Pollutants and tracers The spreading of up to 99 constituents (pollutants and/or sediments) can be simulated simultaneously. The names of sediment substances have to start with “Sediment”. The names of the constituents can be up to 20 characters long. The temperature is computed and the influence of the temperature on the water density is taken into account. If Sediments are selected a window appears to let you specify the names of the cohesive and/or non-cohesive sediments applied. Salinity and temperature affect the density of the water.16: Data Group Processes You can select any combination of the processes by: Ticking off its check box and specifying names if required. User Manual Figure 4. Remark: You can specify a first order decay rate for each of the constituents. See Appendix B.

To take into account the influence of short waves on the bed-stress and the radiation stress on the overall momentum transport. using the latest WAVE results.17: Sediment definition window Wind Wave Online Delft3D-WAVE Secondary flow Deltares and morphology characteristics. There you can specify a global time varying wind field.9 for details on how to use the Sediment and morphology functionality. An extension to the previous process is Online Delft3D-WAVE. Remark: If FLOW runs online with WAVE. subsequently a WAVE simulation is performed. such 43 . If wind is selected a Wind tab is available in the Data Group Physical parameters. Every time the communication file is written. Prior to starting the FLOW simulation with online Delft3D-WAVE you also have to prepare a WAVE input file. The latter is useful for the simulation of storm surges or typhoons in a regional model. the wind specified in FLOW can be used in WAVE. or read a time and space varying wind field and pressure from a file. In this case there is a direct coupling with Delft3D-WAVE. The process Secondary flow adds the influence of helical flow. Restriction: The Wave process is only available for the σ -model. Remark: You first have to run the WAVE module to create a communication file with the required wave information. To include a wind field and its influence on the flow and spreading of substances. See Section 9. Remark: See Chapter 11 and Appendix B. You select the formulation for the influence of short waves on the bottom roughness in the Data Group Physical parameters → Roughness.Graphical User Interface Figure 4. For details of modelling waves you are referred to the User Manual of Delft3D-WAVE.7 for details on how to take the wave-current interaction into account. For details of modelling waves you are referred to the User Manual of Delft3D-WAVE. Then FLOW resumes.

You can select the initial conditions arbitrarily within the domain specified below. to the momentum transport. A large discrepancy between the initial condition and the boundary conditions at the simulation start time can result in short wave disturbances that propagate into the model area. temperature. salinity and/or temperature in the case of an inhomogeneous computation. Tidal forces can only be specified for spherical co-ordinates. see Section 9. secondary flow and constituents are selected in the Data Group Processes.5. i. User Manual as in river bends.1. The processes salinity. the values of the variables in the (dynamic) equilibrium of the model and to some extent the topography. in tidal situations one or even several tidal cycles. Tidal forces can only be applied for spherical grids.9. Secondary flow is not available for the Z -model.6 for details of the formulations. For details see Section 9. The optimum values depend on a combination of the values of the boundary conditions at the start time of the simulation. flow velocity components. The time to reduce these short wave disturbances by internal dissipation such as bottom friction might be quite large. such as a uniform value in the whole area. 4. you can use the functionality to dredge and dump sediments.5. Initial conditions can be very simple.1 Initial conditions are required for all dependent variables. Dredging and dumping When the process Sediments is switched on.9. such as water level.5 Initial conditions In the Data Group Initial conditions you can specify the initial values the computation will start with. and for the secondary flow and constituents if included in the simulation. The file with dredging and dumping sites and dredge characteristics must be opened in the Data Group Operations → Dredging and dumping.e. cannot (and need not) be selected. see Section 4. The influence of spiralling flow can either be taken into account through an equilibrium formulation or as an evolutionary process by solving an advection-diffusion equation. Restriction: Secondary flow is only available for the σ -model. Restrictions: The process Wave is not available for the Z -model. Tidal forces To specify the parameters for tide generating forces. The influence of helical flow is resolved in the 3D formulation itself. If the discrepancy is so large that the flow velocity becomes close to or above the critical flow velocity the simulation might even become unstable. or more complex as a space varying value obtained from ‘somewhere’ or taken from a previous run. If tidal forces is selected a tab Tidal forces is available in Data Group Physical parameters.Delft3D-FLOW. The effects of a discrepancy between the initial condition and the boundary conditions at the 1 Initial condition file is not yet available for the Z -model 44 Deltares . Remark: In 3D simulations the check box of Secondary flow is invisible.

Figure 4.18 is displayed. 45 Deltares . but the transient period to reach the (dynamic) equilibrium will increase. To display the Initial conditions window: Click the Data Group Initial conditions. two sediments and set the option in Figure 4. the final results will be the same. For uniform initial conditions you must specify one or more of the following quantities. selected all processes in the Data Group Processes.Graphical User Interface Figure 4. Mismatches between the initial condition and the boundary condition are washed out of the system instead of propagating inward.18 to Uniform. If you specify less processes less data fields will be displayed. Next. see the Data Group Numerical parameters. depending on the processes you have selected in the Data Group Processes: Water level Specify the initial condition above the reference plane (i. For complicated topographies with large drying and flooding areas we suggest to start at high water and define an initial condition to this situation. Remark: For the layout to appear as displayed in Figure 4. defined one pollutant. the still water level) in [m].18 we have selected a depth averaged computation (Number of layers = 1 in the Data Group Domain → Grid parameters).18: Data group Initial conditions start time of the simulation can be substantially reduced by applying a transition period from the initial condition to the actual boundary conditions and determine the intermediate values by a smooth interpolation. If you do not use optimum initial conditions.e. Above the reference plane is positive. You can select one of the following options: Uniform To specify a uniform initial condition for the water level and each of the constituents defined in the Data Group Processes. Remark: The uniform initial condition for the velocity components is by default set to zero.

0 Unit m ppt ◦ C m/s kg/m3 Restrictions: The uniform initial conditions for the flow velocity components in both horizontal directions are zero by default and cannot be changed. User Manual Salinity Temperature Secondary flow Specify the initial (sea)water salinity in [ppt].000 Default 0. displayed in the lower half of Figure 4.0 -10.date. The initial conditions are read from a map file from a previous run. the same processes must be used in both computations.Delft3D-FLOW.∗> or map file <trim-∗> must reside in the current working directory.0 31.0 0. Upon selecting this option you must specify the file.0 10.time> file. for instance from a previous run. This option can only be applied when you have selected a depth averaged simulation (Number of layers = 1 in Data Group Domain → Grid parameters) and the σ -model. created by a separate program or constructed from measurements. when using a map-file: the start time of the current simulation must occur on the map file. To identify this file you must select a <trim-runid > file. Specify the initial water temperature in [◦ C]. For the Restart file and Map file options to work properly: the restart-runid must differ from the current runid the restart file <tri-rst.0 60.0 0.0 Upper limit 1. Specify the initial secondary flow in [m/s]. secondary flow and constituents need only be specified if the corresponding process is selected in the Data Group Processes.000 100. or. The initial conditions are read from a restart file from a previous run. Initial conditions file The initial conditions are read from an attribute file that you some how obtained. The initial conditions for salinity.000 0.0 0. in case of a 3D-application both computations must have the same number of layers 46 Deltares .runid.runid > or a <trirst. sediment) concentrations in [kg/m3 ].18 are not displayed. see Data Group Output → Storage.0 15. To identify this file you must select either a <tri-rst. the simulation start time of the current scenario must coincide with the time instance at which the restart file is written in the selected scenario.0 0. temperature. Specify the initial constituent (pollutant. Restart file Map file Domain: Parameter Water level Salinity Temperature Secondary flow Constituent Lower limit -1.0 1. Constituents When you select either of the following options the input fields of the uniform values.

the area beyond the model area which is not modelled. Neumann boundaries can only be applied on cross-shore boundaries in combination with a water level boundary at the seaward boundary. These conditions represent the influence of the outer world. the type of boundary forcing and last but not least the data itself. When modelling cross flow in front of a harbour. the latter three yield a much weaker control over the water level. discharges (total or per grid cell) and a combination of water levels and currents. 2 This feature has only been tested for the σ -model Deltares 47 . The Neumann type of boundary is used to impose the alongshore water level gradient. you can divide an open boundary into sections. Remark: Delft3D-FLOW does not check for all these requirements. water level gradients. However. 4. Other examples are: When modelling river flow. the upstream flux is prescribed in combination with the downstream water level. you will prescribe the known water levels at the entrance of the estuary. when you are modelling water levels at the inland side of an estuary. A boundary section is characterised by a name. But. temperature and/or constituents is prescribed by specifying the inflow concentrations. The following types of boundary conditions are available: Water level Velocity Neumann (water level gradient) Discharge or flux (total or per grid cell) Riemann or weakly reflective boundaries2 The choice of the type of boundary condition used depends on the phenomena to be studied. You can specify the boundary conditions for these parameters as time-series only.6 Boundaries In the Data Group Boundaries you can define the open boundaries. their location. i. type and all input data related to driving the simulation. Depending upon the variability of the parameters along the open boundary. velocity components are prescribed. fluxes or weakly reflective conditions.5.Graphical User Interface and layer thickness. The hydrodynamic forcing can be prescribed using harmonic or astronomical components or as time-series. which is needed to make the solution of the mathematical boundary value problem well-posed. For water level forcing the boundary conditions can also be specified in terms of QH-relations.e. For instance. The flow may be forced using water levels. The transport of salinity. since velocities are only weakly coupled to water levels. currents. especially for the more complex flow situations. indices of the begin and end point of the boundary section. the same internal solution may be achieved by prescribing flow velocities. At an open boundary the flow and transport boundary conditions are required.

Often. two velocity boundaries at both ends of a straight channel may lead to continuity problems. There is a substantial correlation between water levels in locations being not too far apart. and End or B).Delft3D-FLOW. i.e. The actual signal at the support points may either be presented to the program as a harmonic type of signal (frequencies. In this case it is better to prescribe the normal velocity component at one end of the channel and the water level at the other. forcing by prescribing water levels only is generally a sound procedure. User Manual The Riemann type of boundary is used to simulate a weakly reflective boundary. meaning that a small error in the prescription of water levels can only be compensated by a large response in the velocity components. For these models. The main characteristic of a weakly reflective boundary condition is that the boundary up to a certain level is transparent for outgoing waves. the signal itself will be a tidal signal composed of amplitudes and phases at astronomic frequencies. When prescribing water level boundaries. For example. you should avoid applying the same type of boundary condition at all boundaries. The boundary condition is prescribed at two so-called boundary support points (Begin or A. most of the larger sea models are likely driven by water level boundaries only since these are the only quantities known with some accuracy. A physically stable result is then obtained. we locate boundaries as far away from the areas of interest as possible. but rather to the entire physical area. Outgoing waves can cross the open boundary without being reflected back into the computational domain as happens for the other types of boundaries. In practice. Since we do not want small errors in the boundary conditions to significantly influence the model results. the channel will eventually dry up or will overflow. Verboom and Segal (1986) and Chapter 9. For more details about the weakly reflective boundary condition see Verboom and Slob (1984). The area of influence of this phenomenon is not limited to a certain number of grid points near the boundary. Points that lie in between these two support points are calculated by linear interpolation of the forcing at both ends. if the fluxes that are a result of these velocity components and their respective water levels are not compatible. If there is more than one open boundary in the model area. amplitudes and phases) or directly as a time-series. For large area models (horizontal dimensions of the order of the tidal wave length or larger).g. the small relative error introduced by the interpolation procedure becomes more and more significant and important when the method is applied for smaller area models (e. However. Remark: The interpolation between the begin and end point of a boundary section are obtained by linear interpolation of the values at these (support) points. For instance. which divide a complete boundary into several segments. This coefficient Alfa must be ≥ 0 (recommended value for tidal computations is 50 or 100). A weakly reflective form of the other boundary types is obtained by specifying a reflection coefficient. the boundary condition should be obtained by nesting the model into a calibrated larger area model. When modelling tidal flow in a large basin. coastal models). For boundary conditions expressed in amplitudes and phases the interpolation is executed on these quantities 48 Deltares . you should keep in mind that water level is a globally varying quantity. the type of boundary condition applied often depends mainly on the available data. the amplitudes and phases can be interpolated from nearby ports. such as short wave disturbances.

This is a rather difficult task if the bathymetry changes appreciably over the cross-section. If you want to prescribe a certain discharge through the total cross-section you must translate this total discharge into discharges at the end points.+.+. Upon selecting the Data Group Boundaries the following sub-window is displayed. The section is given the default name “-Unnamed-”. the total discharge through the cross-section and the total number of sections in the crosssection. so you can prescribe the discharge through a cross-section using the type of boundary Total discharge.+. Hi . or discharge points.+. the discharge prescribed in an end point is the discharge through the grid cell related to that specific end point. A better approach is to define boundary sections of just one grid cell and to determine the boundary conditions per grid cell (still given at its end points A and B.e. i. In the present release of Delft3D-FLOW the procedure mentioned above is incorporated in the FLOW-GUI.+.19: Sketch of cross-section with 8 grid cells 3/2 N j =1 Hj -+. the discharge is prescribed at both ends of a boundary section and intermediate values are determined by linear interpolation. respectively. displayed in the Section name field. You can select one or more of the following options: Add Deltares .+ (and not on the computed time dependent boundary values).+. the average depth in section i. Q and N are the discharge through section i.+.20.+.+.Graphical User Interface - 1 2 3 4 5 6 7 8 C ross-section V irtual line of grid points Discharge boundary -+-+-+.+-+. see Figure 4. However. 49 where Qi .+ Q (4.+. Discharge per grid cell and total discharge To specify boundary conditions of type Discharge per cell you must apply a similar procedure as for the other boundary types.+. which now are the same) using the following approach: Qi = Hi 3/2 -+. Remark: The depth is defined at the grid lines between the velocity.+ -+.+ -+.+.19 details.4) To Add an open boundary section either: Click Add.+ Top view Figure 4. see Figure 4.+.

type of open boundary. User Manual Figure 4. Click Delete. or Use the Visualisation Area window.21 is displayed.Delft3D-FLOW. Specify the grid indices of the begin and end point given by (M1. Upon selecting Open / Save Figure 4. forcing type) of the open boundary sections in an attribute file with extension bnd in the working directory. or Use the Visualisation Area window. reflection parameter. To Delete an open boundary section either: Select the boundary section to be deleted in the list box. Open/Save Open or Save the definitions (location. N2). Delete 50 Deltares .20: Main window for defining open boundaries Replace the name “-Unnamed-” by a useful name. N1) and (M2. There is no preferred direction in the definition of a boundary section. Only an open boundary of the type Water level may be defined along a diagonal grid line. Remarks: Open boundary sections typically extend over more than just one grid cell. so you have to specify their begin and end point.

21: Open and save window for boundary locations and conditions Deltares 51 .Graphical User Interface Figure 4.

3. Boundary Indices of begin and end point Displayed in FLOW-GUI Used in computation (1. 2) and (12. 1) (12.Delft3D-FLOW. 6) North boundary (2. To illustrate this we take the example given in Appendix E for a straight rectilinear channel of 10 ∗ 5 grid cells or 11 ∗ 6 grid points. 7) Discussion: The location and numbering of open boundaries need some clarification and illustration. 6) u -boundary (2. Water level boundaries are located just outside the first or last computational grid cell and velocity boundaries are located at the edge of the first or last computation cell. 2) and (12. are given in Table 4. 6) (2. 1) and (11. 6) v -boundary West boundary South boundary East boundary (1. see Figure 4. 1) (12. 1) and (11. location of open and closed boundaries Table 4. The definition of the boundary sections. Remark: Much confusion on the location and numbering of open and closed boundaries can be avoided by thinking in terms of computational grid cells.22. 2) and (11. Both open and closed boundaries are defined at the grid enclosure (through the water level points). 52 Deltares . See Appendix E for a discussion and illustration. 6) ζ -boundary (11. 7) and (11. both as displayed in the FLOW-GUI and as used in the computation. However.3: Definition of open and closed boundaries. 6) (2. 7) and (11. 7) ζ -boundary (2. physically closed and open velocity controlled open boundaries are located at the nearest velocity point inside the grid enclosure. 6) and (12. User Manual Figure 4.22: Straight channel. 2) and (1. 2) and (1.

A boundary section must be located on the grid enclosure.0 10. Type of open boundary Select either one of the options: Water level Current Neumann Discharge per cell Total discharge Riemann Remarks: Riemann type of boundary has only been tested for the σ model. neither for a 2D simulation.Graphical User Interface Domain: Parameter Name boundary Lower limit Upper limit Default -Unnamed0 0.0 Unit none none none Any printable character At the grid enclosure 0. Select either one of the options: Astronomic. n) Alfa Restrictions: A boundary section name must be non-blank. Only an open boundary of the type Water level may be defined along a diagonal grid line. Remark: The reflection parameter does not apply for Neumann or Riemann controlled boundary sections. and can contain up to 20 characters. Vertical profile for hydrodynamics The vertical profile does not apply for Water level or Neumann controlled boundary sections. Harmonic. Specify the amount by which you to want to make the open boundary less reflective for short wave disturbances that propagate towards the boundary from inside the model.000. the flow conditions are specified using user-defined frequencies. Time-series. Next you must specify for each boundary section the type of forcing and the forcing itself. the flow conditions are specified using tidal constituents. The profile can be: Reflection parameter Alfa Forcing type Deltares 53 . The nodal amplitude and phase factors are determined at the start of the simulation and updated after each NodalT time interval. The number of boundary sections is limited to 300. QH-relation. the flow conditions are specified as time-series.0 (m. amplitudes and phases. (only in combination with water level boundary type) the water level is derived from the computed discharge leaving the domain through the boundary. amplitudes and phases.

Delft3D-FLOW. compound and higher harmonic constituents may have to be added. Edit flow conditions Edit transport conditions To define the boundary condition data press the Edit flow conditions button. User Manual Uniform The velocity profile is uniform over the water depth.6. 4.6. In the current implementation you must add this type manually in the <bnd>-file.2.1 Flow boundary conditions For flow boundary conditions you can either select: Astronomic Harmonic QH-relation (for water levels only) Time-series The window to specify the boundary data depends on the forcing type selected.2.2. In this representation by means of the primary constituents.5. Details are specified in a separate window. Remark: This option is activated if you open a boundary location <bnd>file with boundary type ‘3-d profile’.10 and A. and bottom friction give rise to non-linear 54 Deltares . such as Delft3D-TIDE or Delft3D-TRIANA.5. each with its own characteristic frequency ω (angular velocity). To specify the boundary data for the hydrodynamic simulation. Per layer The velocity profile is embedded in the boundary condition file taken from a larger area model by nesting.5. see Section 4. This is the case in shallow water areas for example. where advection. large amplitude to depth ratio.6. The observed tidal motion can be described in terms of a series of simple harmonic constituent motions. To specify the boundary data for the constituents transported with the flow and only for those constituents specified in the Data Group Processes. Logarithmic The velocity profile is a logarithmic function over the water depth. The boundary conditions are stored in an attribute file with extension <bca>. Astronomic boundary conditions You can specify the boundary conditions in terms of astronomical components. These astronomical components are mostly obtained from some type of tidal analysis program.15 for details. The sub-window displayed depends on the Type of open boundary selected. as tens of components can be prescribed. Details are specified in a separate window.1. see Section 4. Specifying astronomical components in a manner similar as harmonic components is a laborious activity. The amplitudes A and phases G of the constituents vary with the positions where the tide is observed. Remark: The Vertical profile for hydrodynamics forms also part of the definition and it will be saved in the <bnd> file as well. see Appendix A.

V0 is frequency dependent.5) in which: H (t) A0 k i Ai Fi ωi (V0 + u)i Gi water level at time t mean water level over a certain period number of relevant constituents index of a constituent local tidal amplitude of a constituent nodal amplitude factor angular velocity astronomical argument improved kappa number (= local phase lag) Fi and (V0 + u)i are time-dependent factors which. The amplitudes will be 1 and the phases 0. Remark: When adding a Component set. Amplitude and Phase.6 years. see Appendix C. Upon selecting Astronomic and next Edit flow conditions Figure 4. You only have to provide the amplitudes. reflecting the boundary and the end point (A or B) for which this set will be used. Ai and Gi are position-dependent : they represent the local character of the tide. can easily be calculated and are generally tabulated in the various tidal year books. or to copy a value. V0 is the phase correction factor which relates the local time frame of the observations to an internationally agreed celestial time frame. together with ω . See section B. At each end you have to assign a Component set. For most frequencies they have a cyclic period of 18. Delft3D-FLOW re-calculates the nodal amplitude factors and astronomical arguments every 6 hours. In the Name field: select from the dropdown list the required tidal constituent.23 is displayed.Graphical User Interface interactions. Deltares 55 . this set will have the same tidal constituents as the first set. phases and frequencies (in terms of constituent names).21 for changing this default. Each open boundary section has a begin (A) and an end (B) point. Replace the name ‘-Unnamed-’ in the Selected set field by a useful name. A Component set consists of a number of tidal constituents with their Name. A0 . The set is given the default name ‘-Unnamed-’. To add a Component set : Click Add. To add a tidal constituent to a Component set : First select the required set from the list of available Component sets. For a list of primary and compound constituents. The general formula for the astronomical tide is: k H (t) = A0 + i=1 Ai Fi cos (ωi t + (V0 + u)i − Gi ) (4. Use the menu item Table in the menu bar to insert. Click in the Amplitude or Phase field. copy or delete a row. By default. Fi and ui are slowly varying amplitude and phase corrections and are also frequency dependent.

Under Selected component sets. Remark: The mean value (A0) has no phase. see Figure 4. If you do not want this. Under Selected component sets.Delft3D-FLOW.24. an intermediate point will have a phase of 180 degrees. select from the dropdown list the required set for End B. These corrections can be applied to a sub-set of the boundary sections and to a sub-set of the astronomic components applied. Specify the same set for End A and End B. Click Edit flow conditions. To assign a Component set to a boundary: First select the required open boundary. You can apply corrections to the defined astronomic components. you can specify 361 degrees at 56 Deltares . User Manual Figure 4. only one Component set is needed for this open boundary. For instance. This attribute file has extension <cor>. If the type of open boundary is Total discharge. if you have a phase of 1 degree at End A. Remarks: The correction does not apply for the A0 astronomic component. These corrections are typically applied during the calibration phase of a model.23: Specifying astronomical boundary conditions Specify the Amplitude and Phase. Both the amplitude (multiplicative) and the phase (additive) can be corrected. select from the dropdown list the required set for End A. and a phase of 359 degrees at End B. Be aware that amplitudes and phases for intermediate points at the open boundary are linearly interpolated from the values at the end points.

23 the component sets from the imported file are listed under Component sets.Graphical User Interface Figure 4. The information about which set is assigned to which boundary end point is stored in the <bnd> file. delete or modify them at will.21. the corrections and the specification of the sets to boundary end points: Use Open / Save in Figure 4. To save the component sets. Domain: Deltares 57 . To assign a Component set to an end point of an open boundary section. Remarks: The components sets are stored in a file with extension <bca>.20 and Open Astronomical flow conditions in Figure 4.21. You can extend. follow the procedure above. In the window of Figure 4.20 to enter the Open/Save Boundaries window of Figure 4.24: Contents of a Component set with two tidal constituents having corrections End A. You can import astronomic boundary conditions by selecting Open/Save in Figure 4.

000 Default 0.0 10. Boundary conditions of type Astronomic must precede boundary conditions of type QHtable and Time-series.0 0.0 0.0 0.Delft3D-FLOW. 58 Deltares .25: Specifying harmonic boundary conditions Parameter Water level Neumann Current Discharge Riemann Phase Alpha Lower limit 0.0 0.0E+06 100.0 0.0 10.0 0.0 0. The maximum number of frequencies is 234. User Manual Figure 4.0 1.0 Unit m m/s m3 /s m/s degrees none Restrictions: Astronomical forcing can not be mixed with harmonic forcing.0 Upper limit 100.0 100.0 -360.0 0.0 0.0 0.0 0.0 360.

25 is displayed. if you have a phase of 1 degree at End A. Restrictions: Deltares 59 .6) where N is the number of frequency components. the amplitudes and phases for the whole section are required. Be aware that amplitudes and phases for intermediate points at the open boundary are linearly interpolated from the values at the end points. The unit of the amplitudes depends on the quantity prescribed. N1) and (M2. and [m/s] for Riemann boundaries. respectively. N2). For each of the boundary sections you can either import the boundary conditions by selecting Open/Save. copy or delete a row. you can specify 361 degrees at End A. an intermediate point will have a phase of 180 degrees. The boundary signal F (t) is constructed by super-imposing the following individual components: N F (t) = i=1 Ai cos (ωi t − ϕ) (4. Remark: If the type of open boundary is Total discharge. see Figure 4. The phase of the harmonic component in [degrees] at the Begin point A and at the End point B defined at grid indices (M1. Remark: The data you import here will be included in the <bch> attribute file when you save the boundary locations and conditions. or specify the harmonic conditions manually in a table. Remarks: The mean value is specified by zero frequency and zero phase. [m3 /s] or [m/s] at the Begin point A and at the End point B defined at grid indices (M1.Graphical User Interface Harmonic boundary conditions Upon selecting Harmonic and next Edit flow conditions Figure 4. [-]. they are respectively [m] for water elevations. [m/s] for velocities. Phase Use the menu item Table in the menu bar to insert. and a phase of 359 degrees at End B. If you do not want this. [-] for Neumann. The mean value must always be specified. The amplitude of the harmonic component in [m]. If not taken care of by you. All boundaries with Harmonic forcing must have the same frequencies. or to copy a value. [m3 /s] for fluxes. N2). For instance. Open Import the harmonic boundary conditions from an attribute file with extension <bch>. the GUI will save the frequencies of the last Harmonic boundary. N1) and (M2. In the table you must specify: Frequency Amplitude The frequency of the harmonic component in [degrees/hour].21. respectively.

The sub-window displayed has just one column for the discharge and one column for the water level.0E+06 100.26 is displayed. or to copy a value.0 0. If you have defined all data for the currently selected boundary section: Close the sub-window and select another boundary section to define its boundary conditions.0 10.0 0.0 0. User Manual Harmonic forcing may not be mixed with astronomical forcing and must precede QH and time-series forcing.0 0. First click in either the Discharge or Water level input field. A relaxation parameter can be specified in the data group Numerical parameters. Upon selecting QH-relation and next Edit flow conditions Figure 4. Use the menu item Table in the menu bar to insert. If you have defined all data for the currently selected boundary section: Close the sub-window and select another boundary section to define its boundary conditions.0 Default 0.0 Unit m m/s m3 /s m/s degrees QH-relation boundary conditions (for water level boundaries only) For QH-relations you must specify the relation between the outflowing discharge and the water level at the boundary.0 1. Domain: Parameter Discharge Water level Lower limit -1.0E+6 -100. copy or delete a row.0 Unit m3 /s m 60 Deltares .0 0. Domain: Parameter Water level Neumann Current Discharge Riemann Phase Lower limit 0. Above the highest specified discharge and below the lowest specified discharge the water level is kept constant.0 0.Delft3D-FLOW.0 100.0 0. The maximum number of frequencies is 234.0 360.0 Upper limit 1. Intermediate values are determined by linear interpolation.0 Upper limit 100.0 0.0 0.0E+06 100.0 -360.0 Default 0. You cannot change the type of interpolation.0 0.

for positive discharges from small to large and for negative discharges from large to small. respectively. That is. current. copy or delete a row. Intermediate values are determined by linear interpolation.Graphical User Interface Figure 4.e. Intermediate values are determined by linear interpolation. discharge or Riemann. Above the highest specified discharge and below the lowest specified discharge the water level is kept constant. For time-series conditions you must specify the boundary conditions in terms of values of the selected kind. Time-series boundary conditions Upon selecting Time-series and next Edit flow conditions Figure 4. Use the menu item Table in the menu bar to insert.26: Specifying QH-relation boundary conditions Restrictions: The discharges should be specified in increasing order. water level. i. Boundaries with forcing by a QH relation must precede time-series boundaries and they must follow boundaries with harmonic and astronomical forcing. Deltares 61 . only one (1) discharge is required for the whole section. Neumann. or to copy a value. If you have defined all data for the currently selected boundary section: Close the sub-window and select another boundary section to define its boundary conditions.27 is displayed. at time breakpoints for both the Begin point A and the End point B. Remark: If the type of open boundary is Total discharge. You cannot change the type of interpolation.

0 Upper limit 100. The first time breakpoint must be before or equal to the simulation start time.000 Default 0. and the last time breakpoint must be equal or later than the simulation stop time (see Data Group Time frame).0 Unit m m/s m3 /s m/s none Restrictions: The time breakpoints must differ from the simulation start time by an integer multiple of the time-step (Data Group Time frame). Time-series forcing must follow astronomic.0 100.0E+06 100.0 0.0 0. Intermediate values are determined by linear interpolation.0 10.0 0. Time breakpoints must be given in ascending order.0 10.0 -1.0 -100.Delft3D-FLOW.0E+6 -100.0 0. User Manual Figure 4.0 1.0 -10.0 0.0 0.27: Specifying time-series boundary conditions Domain: Parameter Water level Neumann Current Discharge Riemann Alfa Lower limit -100. 62 Deltares . harmonic or QH forcing when these type of boundary conditions are applied in the same model area.

2 Transport boundary conditions To specify the boundary conditions for all quantities defined in the Data Group Processes you must for each boundary section: Select the boundary section in the list box.5. The return time must be specified for each open boundary section. Remark: When you specify a negative value for the time lag the boundary treatment for constituents. In a 3D simulation you can specify a surface and a bottom value.Graphical User Interface Figure 4. but now for all times. Deltares 63 . see Figure 4. At the sea-side boundary a common problem for numerical models of estuarine areas is encountered when the boundary conditions for a constituent are to be prescribed.20. Select Edit transport conditions. Specify if required Thatcher Harleman time lags. the value of the upper layer. For 3D simulations you can specify different values at the surface and at the bottom to represent the influence of stratification and the different flow conditions above and below the interface. the inflowing water mass immediately after low water slack originates from the outflowing water mass a moment earlier. see Figure 4. Next the window Boundaries: Transport Conditions is displayed. In a physical (unbounded) world. the solution for these parameters will be fully determined by the processes inside the model area. The return time depends on the flow conditions outside the estuary. in a 2D simulation you need to specify only the surface value (applied over the whole depth). If there is a strong circulation the return time is short. one for the top layer and the other for the bottom layer.28: Transport conditions. In a 2D model only the first value is used. Thus. The window displayed is for a 3D simulation. this time lag (return time) is often modelled by means of a “Thatcher-Harleman” boundary condition (Thatcher and Harleman. the concentration of the inflowing water is commonly not equal to the concentration Cmax which has been prescribed along this open boundary.6. In numerical models.29.28. It will take some time before the concentration along this open boundary reaches the Cmax value. This negative value has no physical meaning or background. salinity and temperature will be the same as at outflow. 1972). Thatcher Harleman time lags 4. Thatcher-Harleman time lag The return time for concentrations from their value at outflow to their value specified by the boundary condition at inflow. Consequently. it only offers the option to fully exclude the influence of the outer world on the inner solution. for a 2D simulation all input fields related to 3D are either absent or insensitive. The return times for the layers in between will be determined through a linear interpolation of these two values. see Figure 4.

depending whether you have a 2D or a 3D simulation. In the last case the location of the profile jump. n1) and (m2. respectively. n2). salinity or temperature. or to copy a value. In the table you must specify time breakpoints and concentrations at the Surface and Bottom layer for the Begin point A and the End point B of the current boundary section.Delft3D-FLOW. defined at grid indices (m1.e. The options for the vertical profile for these quantities are respectively: Uniform. Linear Surface and bottom values are linearly interpolated for intermediate depth. for each of the quantities selected in the Data Group Processes you must: Constituent Select the quantity to be defined next in the table. Per layer The vertical profile is embedded in the boundary condition file taken from a larger area model by nesting. You can either select: Uniform The value is applied over the whole water depth. etc) is included in a 3D model then the vertical profiles for these quantities must be prescribed. Linear and Step (discontinuous profile).29: Specifying transport boundary conditions Transport boundary conditions always are specified as time-series. The surface value is maintained down to this depth and the bottom value is used from the jump depth down to the bottom. If a constituent (i. Vertical profile To specify the vertical profile of the concentration at the current boundary section. measured in me- 64 Deltares . You can specify one or more of the following options. User Manual Figure 4. Step You can define a jump in the vertical profile. Upon selecting this option you can specify the depth (Profile jump) below the free surface at which the profile jump occurs. Use the menu item Table in the menu bar to insert. copy or delete a row. only the surface values need to be specified. besides the usual boundary condition values.

Consequently.0 Unit ppt ◦ C kg/m3 Restrictions: The time breakpoints must differ from the simulation start time by an integer multiple of the time-step (Data Group Time frame).0 0.Graphical User Interface tres from the water surface. Time breakpoints must be given in ascending order. The other tabs are only available if the associated processes are switched on: Heat flux model (requires Temperature process activated) Sediment and Morphology (requires Sediments process activated) Wind (requires Wind process activated) Tidal forces (requires grid in spherical co-ordinates and Tidal forces process activated) Deltares 65 . 4.0 1. a mismatch between your boundary condition and the concentration distribution in the vertical. This mismatch can produce spurious oscillations in your result. the position of the jump will be determined using the water elevation read from this file. some are always needed and some are related to the processes switched on or off. may occur.5.0 0. Domain: Parameter Salinity Temperature Pollutant or sediment Lower limit 0. Once determined. just inside the open boundary. The first time breakpoint must be before or equal to the simulation start time. The tabs for Constants.000 Default 0. If you have defined all data for the currently selected boundary section: Close the sub-window and select another boundary section to define its transport boundary conditions. so this layer number depends on the initial water level. Remark: When the initial conditions are read from a restart file (hot start).0 0.0 Upper limit 100.0 60. this layer number will be fixed throughout the simulation. The physical parameters can be split into several classes. The Data Group Physical parameters contains several tabs. During initialisation the jump in the profile will be translated into a layer number.7 Physical parameters In the Data Group Physical parameters you can select or specify a number of parameters related to the physical condition of the model area. Roughness and Viscosity are always visible.0 0. and the last time breakpoint must be equal or later than the simulation stop time (see Data Group Time frame). must be prescribed.

The background water salinity in [ppt]. The background water density in [kg/m3 ]. You can specify one or more of the following constants: Gravity Water density The gravitational acceleration in [m/s2 ]. This so-called ‘secondary flow’ (spiral motion) is of importance for the calculation of changes of the river bed in morphological models. The density of air in [kg/m3 ].e. Fraction of the shear stress due to secondary flow taken into account in the momentum equation. Air density Temperature Salinity Beta c Equilibrium state 66 Deltares . This spiral motion is automatically taken into account in a 3D simulation.Delft3D-FLOW. No Compute the secondary flow as a dynamical process.30 is displayed. This component is directed to the inner bend along the river bed and directed to the outer bend near the water surface. and the process Salinity is. The flow in a river bend is basically three-dimensional. The velocity has a component in the plane perpendicular to the river axis. and the process Temperature is. User Manual Figure 4. i. This value is only required for a homogeneous simulation.30: Specifying the physical constants 4. The background water temperature in [◦ C]. This value is only required if the processes Salinity is not selected.1 Constants Upon selecting Physical parameters → Constants the sub-window given by Figure 4.7.5. used in the equation of state. used in the equation of state. the processes temperature and salinity are not selected. Tick off if you want to apply the equilibrium state for the secondary flow process (only for depth-averaged flow): Yes Apply the equilibrium state. used in the wind stress formulation. This value is only required if the process Temperature is not selected.

The actual local velocity distribution originates from a multiplication of this function with the spiral motion intensity. As a result. The βc parameter described here influences the extent to which the effect of spiral motion in a depth-averaged model is accounted for. This assumption might be violated in models that cover only a very small area. Breakpoint (C). Inbetween the breakpoints there is a linear interpolation applied. reflecting increasing roughness of the water surface with increasing wind speed. The parameter Beta c specifies to which extent the shear stresses due to the secondary flow are included in the momentum equation (Beta c = 0 means no inclusion). you can specify a constant wind drag coefficient. Wind drag You can specify three Wind drag coefficient values and three Wind speed values at which these coefficients apply. Domain: Deltares 67 . The last two coefficients. The first two coefficients. The secondary flow may be computed from a local equilibrium approach or from an advection/diffusion equation.31.31.6 for full details. The wind drag coefficient may be linearly dependent on the wind speed. a linearly increasing wind drag coefficient or a piece-wise linear function of wind speed. The vertical distribution of this horizontal velocity component is assumed to be a universal function of the vertical co-ordinate. can be used to determine the deviation of the direction of the bed shear stress from the direction of the depth-averaged flow. The intensity of the secondary flow is a measure for the magnitude of the component of the secondary flow normal to the depth averaged flow. as shown in Figure 4. The three wind drag coefficients determine three breakpoints in the piece-wise linear function of wind drag coefficient and wind speed. The underlying assumption is that the secondary flow is locally determined inside the model area and not transported into the model area through the open boundaries. The spiral motion intensity. Breakpoint (A).31: Examples of the wind drag coefficient but in 2D simulations the flow equations must be extended to take this effect into account.Graphical User Interface 0 Figure 4. this is done in the morphological module if the secondary flow is included as a process. see Figure 4. Remark: You do not need (can not) prescribe boundary conditions for the transport equations that describe the secondary flow. specify the constant wind drag coefficient from the specified wind speed and higher. I . See Section 9. determine the constant wind drag coefficient from zero wind speed up to the wind speed specified.

0 0. Upon selecting Physical parameters → Roughness the window given by Figure 4.0 0. either as a uniform value in each direction.0 Unit m/s2 kg/m3 kg/m3 ◦ C ppt none none none none m/s m/s m/s Roughness In the sub-data group Roughness you can specify the bottom roughness and the roughness of the side walls.0 31. White-Colebrook. Z0 (for 3D simulations only). User Manual Parameter Gravity Water density Air density Temperature Salinity Beta c Coefficient A Coefficient B Coefficient C Wind speed A Wind speed B Wind speed C Lower limit 9. or space-varying imported from an attribute file.5 900.0 0.(or ξ -) and y .0 1.0 0. Bottom roughness You can select and define the following options and parameters: Roughness formula The bottom roughness can be computed according to: Manning. The bottom roughness can be computed with several formulae and you can specify different coefficients for x. The Chezy C= H 1/6 n (4.32 is displayed.5 0.0 100.1 100.81 1.0 0. In the Manning formulation the Manning coefficient.00063 0. ´ Chezy.7) 68 Deltares .Delft3D-FLOW.0 100.0 0.0 100.0 100.0 1.0 0.00723 0.000 1.00723 0. n.0 Upper limit 12.5 60.0 Default 9.0 0.0 0.0 100. must be spec´ friction coefficient is calculated from: ified.1 0.0 0.(or η -) direction.1 0.500 1.0 0.5 0.0 15.

Graphical User Interface

Figure 4.32: Sub-data group Roughness

where H is the water depth. A typical Manning value is 0.02 [s/m1/3 ]. In the White-Colebrook formulation you must specify the equivalent ´ friction cogeometrical roughness of Nikuradse, ks [m]. The Chezy efficient is calculated from:

C = 18 10 log

12H ks

(4.8)

A first estimate of the bed roughness length z0 [m] in 3D computations can be derived from the equivalent geometrical roughness of Nikuradse, ks :

z0 =

ks 30

(4.9)

Typical values of ks range from 0.15 [m] for (river) beds with sediment transport down to 0.01 [m] or less for (very) smooth surfaces; see Section 9.4.1 for more details. ´ friction coefficient is given by: A first estimate of the Chezy

C = 25 + H

(4.10)

Uniform File

Stress formulation due to wave forces

Remark: ´ The Chezy friction coefficient discussed above applies to a depth averaged computation. Specify the values in u- and v -direction in the input fields. Select the space varying coefficients in u- and v -direction from an attribute file (extension <rgh>). See Appendix A.2.18 for a format description of the roughness file. The file is not read. The bottom stress due to wave forces can be computed with several formulae. You can select one of the following formulae, see also Section 9.7.5: 69

Deltares

Delft3D-FLOW, User Manual Fredsøe (1984) Myrhaug and Slaattelid (1990) Grant and Madsen (1979) Huynh-Thanh and Temperville (1991) Davies et al. (1988) Bijker (1967) Christoffersen and Jonsson (1985) O’ Connor and Yoo (1988) Van Rijn (2004)

Wall roughness The friction of side walls is computed from a ‘law of the wall’ formulation if selected: Slip condition You can select one of the following: Free; zero tangential shear stress Partial ; specify the roughness length (m). No; zero velocity at the wall assumed. In large-scale hydrodynamic simulations, the tangential shear stress for all lateral boundaries or vertical walls can be safely neglected (free slip). For simulations of small-scale flow (e.g. laboratory scale) the influence of the side walls on the flow can be significant and this flag can be activated. The partial slip only applies to the side walls of permanently dry areas, including dry points. The tangential shear stress is calculated based on a logarithmic law of the wall where the roughness length needs to be specified. See Chapter 9 for details. Domain:
Parameter Manning White-Colebrook ´ Chezy Z0 Lower limit 0.0 0.0 0.0 0.0 Upper limit 0.04 10.0 1,000 1.0 Default 65.0 65.0 65.0 65.0 Unit m−1/3 s m m1/2 /s m

Remark: ´ The default value is set for the Chezy-formulation.

70

Deltares

Graphical User Interface

Figure 4.33: Defining the eddy viscosity and eddy diffusivity

4.5.7.2

Viscosity The background horizontal eddy viscosity and horizontal eddy diffusivity are user-defined. The viscosity and diffusivity calculated with the Horizontal Large Eddy Simulation (HLES) sub-grid model will be added to the background values (for FLOW v3.50.09.02 and higher). Remark: If you use an older version of FLOW, or if you re-run an old MDF-file with FLOW v3.50.09.02 or higher, be aware of this adding operation. The user-defined parameters can be entered as one uniform value for each quantity (uniform for both directions) or as non-uniform values read from a user-specified file with file extension <edy>. If the model is 3D, you may specify constant background values for the turbulent vertical eddy viscosity. If constituents are modelled you may also specify a constant background value for the vertical turbulent eddy diffusivity and the Ozmidov length scale. Finally the model for 3D turbulence can be specified. For details see Section 9.5. Upon selecting Physical parameters → Viscosity Figure 4.33 will be displayed.

Deltares

71

Delft3D-FLOW, User Manual

Figure 4.34: Window with HLES parameters

Horizontal eddy viscosity and diffusivity You can select or specify the following parameters:

Background values Uniform Horizontal eddy viscosity Horizontal eddy diffusivity File

A uniform horizontal eddy viscosity and eddy diffusivity. The values are defined in the separate input fields. Specify a uniform horizontal eddy viscosity in [m2 /s]. Specify a uniform horizontal eddy diffusivity in [m2 /s]. Select the horizontal eddy viscosity and eddy diffusivity from an attribute file with extension <edy>. The data is not read.

Model for 2D turbulence Subgrid scale HLES Upon selecting Figure 4.34 will be displayed. Details about the input parameters and theory can be found in Appendix B.8. Remarks: The contribution from the HLES model will be added to the background values. If the relaxation time is less than 0.0, the high-pass filtering operation will be cancelled.

72

Deltares

Graphical User Interface Background vertical eddy viscosity and diffusivity You can select or specify the following parameters: Vertical eddy viscosity Specify a uniform vertical turbulent eddy viscosity in [m2 /s]. Vertical eddy diffusivity Specify a uniform vertical turbulent eddy diffusivity in [m2 /s]. The resulting viscosity is the maximum of the user-defined background value and the calculated value from the turbulence model. Similar for the diffusivity. Ozmidov length scale The Ozmidov length scale determines the minimal value for the vertical diffusivity from the calculated vertical diffusivity and the equation ¨ al ¨a ¨ frequency and Ozmidov length scale; see Chapusing Brunt-Vais ter 9 for details.

Model for 3D turbulence Turbulence Model For a 3D simulation you can select the type of turbulence formulation used to determine the vertical turbulent eddy viscosity and the vertical turbulent eddy diffusivity additional to the background values. You can select from: Constant Only the background values will be applied. Algebraic The coefficients are determined by algebraic equations for the turbulent energy and the mixing length. k-L The coefficients are determined by a transport equation for the turbulent kinetic energy k and mixing length L. k-epsilon The coefficients are determined by transport equations for both the turbulent kinetic energy and the turbulent kinetic energy dissipation. Remark: You do not need (can not) prescribe boundary conditions for the transport equations used in the selected turbulence model. The underlying assumption is that the turbulence is mainly generated inside the model area and not transported into the model area through the open boundaries. For very small area models this assumption might be violated and a special add-on to the flow module can be provided to prescribe the required boundary conditions. In 3D-simulations the vertical turbulent eddy viscosity and turbulent diffusivity are computed using a turbulence closure model. The four models differ in their prescription of the turbulent kinetic energy, the dissipation of energy and/or the mixing length. Constant eddy viscosity and diffusivity: you must specify background values that are used in both (horizontal) directions. Algebraic model: the first closure model uses an algebraic formula for k and L.

k depends on the (friction) velocities, whereas for the mixing length L a function following
Bakhmetev (1932) is used. In the case of vertical density gradients, the turbulent exchanges are limited by buoyancy forces and the mixing length L must be corrected. This correction involves a so-called damping function. The approach is sufficiently accurate to be applied in

Deltares

73

Delft3D-FLOW, User Manual cases where weakly stratified flow conditions exist. See Section 9.5.1 for full details. k -L model: the second closure model is a first-order turbulence closure model. In this model an identical approach is applied for the mixing length L. To find the turbulent kinetic energy k a transport equation is solved. See Section 9.5.2 for full details. k -ε model: the third model is a k -ε model which is a second-order turbulence closure model. Here, both the turbulent kinetic energy k and the turbulent kinetic dissipation ε are prescribed by a transport equation. From k and ε the mixing length and the vertical turbulent eddy viscosity value are determined. The mixing length is now a property of the flow, and in the case of stratification no damping functions are needed. The eddy diffusivity is derived from the eddy viscosity. See Section 9.5.3 for full details. Remark: Both parameters, the vertical eddy viscosity and eddy diffusivity, are defined at the layer interfaces. Therefore, when you plot or print these values you will obtain one value more than the number of layers. The value for both the horizontal eddy viscosity and the horizontal eddy diffusivity depends on the flow and the grid size used in the simulation. For detailed models where much of the details of the flow are resolved by the grid, grid sizes typically tens of metres or less, the values for the eddy viscosity and the eddy diffusivity are typically in the range of 1 to 10 [m2 /s]. For large (tidal) areas with a coarse grid, grid sizes of hundreds of metres or more, the coefficients typically range from 10 to 100 [m2 /s]. Both coefficients are so-called calibration parameters; their value must be determined in the calibration process. The use of the uniform vertical eddy viscosity and eddy diffusivity coefficient depends on the selected turbulence closure model: If a constant turbulence closure model is selected the prescribed (constant) background values are applied. In all other cases the uniform values are used as a background (minimum) value for the turbulent contribution. The molecular viscosity and diffusivity are added to the turbulent part. In case of strongly stratified flow, a background vertical eddy viscosity may be needed to damp out short oscillations generated by e.g. drying and flooding, boundary conditions, wind forcing etc. Without this background eddy viscosity these short oscillations will only be damped in the bottom layer. The layers behave as if they were de-coupled. The value suggested for the background eddy viscosity depends on the application. The value must be small compared to the vertical eddy viscosity calculated by the turbulence closure model. Our experience is that for stratified estuaries and lakes a value of 10−4 to 10−3 [m2 /s] is suitable. For the eddy diffusivity similar arguments hold as for the eddy viscosity with one exception. Actually, a background eddy diffusivity is not necessary and will generate additional vertical mixing of matter through the pycnocline. However, the access to this value is retained for reason of compatibility with the eddy viscosity and for experimental purposes. Domain: 74 Deltares

Graphical User Interface

Parameter Horizontal eddy viscosity Horizontal eddy diffusivity Vertical eddy viscosity Vertical eddy diffusivity Ozmidov length scale

Lower limit 0.0 0.0 0.0 0.0 0.0

Upper limit 100.0 1,000 100.0 1,000 Max. number

Default 10..0 10.0 1.0E-6 1.0E-6 0.0

Unit m2 /s m2 /s m2 /s m2 /s m

Restriction: Due to the manner in which the horizontal shear stresses containing cross derivatives are numerically implemented (explicit integration) the, otherwise unconditionally stable, numerical scheme for the viscosity term will become conditionally stable. Please check the stability criterion in Chapter 10.

4.5.7.3

Heat flux model If you have switched on the Temperature process, you can specify how the exchange of heat through the free surface is modelled. Upon selecting the tab Heat flux model in the Data Group Physical parameters the window given by Figure 4.35 is displayed. In Figure 4.35 the Ocean heat flux model is selected. Depending on the selected heat flux model parameters are displayed or not.

Heat flux model You can select one of the following options: No flux Absolute flux, total solar radiation The background temperature is used throughout the model area. The absolute temperature is computed. The relative humidity, air temperature and the (short wave) solar radiation for a clear sky have to be prescribed. The net atmospheric (long wave) radiation and the heat losses due to evaporation and convection are computed by the model. Absolute flux, net solar The absolute temperature is computed. The relative humidity, air radiation temperature and the combined net (short wave) solar and net (long wave) atmospheric radiation have to be prescribed. The terms related to heat losses due to evaporation and convection are computed by the model. Excess temperature The heat exchange flux at the air-water interface is computed; only the background temperature is required. Murakami The relative humidity, air temperature and the net (short wave) solar radiation have to be prescribed. The effective back radiation, and the heat losses due to evaporation and convection are computed by the model. The incoming radiation is absorbed as a function of depth.

Deltares

75

Delft3D-FLOW, User Manual

Figure 4.35: Sub-data group Heat flux model

76

Deltares

Graphical User Interface The evaporative heat flux is calibrated for Japanese waters. The relative humidity, air temperature and the fraction of the sky covered by clouds is prescribed (in %). The effective back radiation and the heat losses due to evaporation and convection are computed by the model. Additionally, when air and water conditions are such that free convection occurs, free convection of latent and sensible heat is computed by the model and added to the forced convection. This model formulation typically applies for large water bodies.

Ocean

The absolute temperature models compute the heat fluxes through the water surface by considering incoming radiation, back radiation, evaporation and convection. Evaporation and convection depend on air temperature, water temperature, relative humidity and wind speed. When modelling the effect of a thermal discharge, it is often enough to compute the excess temperature distribution and decay. Excess heat is transferred from the water surface to the atmosphere by evaporation, long-wave radiation and convection. This total heat flux is proportional to the excess temperature at the surface. The heat transfer coefficient depends mainly on the water temperature and the wind speed. Remarks: If you have selected a heat flux model which includes evaporation, you can specify if the evaporated water should be taken into account. Specify Maseva = #Y# in Data Group Additional parameters. Precipitation and evaporation is an add-on of FLOW independent if you model temperature or not. See Appendix B.6 for how you can use this add-on. If you use this add-on together with the process Temperature, then the user-defined evaporation overrules the computed evaporation from the heat flux model. The user-defined evaporated water is always included in the continuity equation.

Global parameters for the heat flux models You can define: Water surface area The water area exposed to wind in [m2 ]. Required for heat flux models 1, 2 and 3.

A secondary effect is included in the heat models by taking into account the surface area of the water body exposed to the wind. Wind strongly affects the evaporation and convection process. The value for the surface area can differ from the effective water surface in the model. Its value depends on the model geometry and on the direction of the wind (Sweers, 1976). Effectively, the surface area can be regarded as a calibration parameter. Sky cloudiness The fractional blocking of the solar radiation by clouds in [%]. Required by heat flux models 1, 4 and 5.

The incoming solar and atmospheric radiations are affected by clouds. The Sky cloudiness parameter describes the percentage of the solar radiation blocked by the clouds; 0 % refers to clear sky condition whereas a value equal to 100 % implies that most of the solar radiation will be absorbed by the atmosphere. Only 35 % (excluding the reflection) of the specified solar radiation will reach the water surface. Consequently, due to increased absorption the atmosphere will ultimately radiate more energy.

Deltares

77

Delft3D-FLOW, User Manual Secchi depth Dalton number Stanton number Required by heat flux models 4 and 5. for evaporative heat flux. Required by heat flux models 4 and 5. for heat convection. Required by heat flux model 5.

Time dependent heat flux model input Interpolation The type of interpolation used for the time-series at intermediate time instances. You can select either Linear or Block interpolation. Open / Save To open and read or save the time-series from or to an attribute file with extension <tem>. In the table you must specify time-series for the following parameters: Incoming solar radiation for a clear sky in J/(m2 s) (only for heat flux model 1) Net (short wave) solar radiation in J/(m2 s) (heat flux model 4) Combined net (short wave) solar and net (long wave) atmospheric radiation in J/(m2 s) (heat flux model 2) Background temperature in ◦ C (Excess heat flux model) Air temperature in ◦ C (all heat flux models except the Excess model) Relative humidity in % (all heat flux models except the Excess model) Fraction of the sky covered by clouds (in %) (Ocean heat flux model) The background temperature used in the excess temperature heat model is already prescribed in Data Group Physical parameters → Constants. Domain:
Parameter Water surface Sky cloudiness Secchi depth Dalton number Stanton number Temperature Humidity Solar radiation Lower limit 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 60.0 100.0 Max. Number Upper limit Max. Number 100.0 Default 0.0 0.0 2.00 0.00130 0.00130 0.0 0.0 0.0 Unit m2 % m ◦

C

% W/m2

Restrictions: The time breakpoints must differ from the simulation start time by an integer multiple of the time-step (Data Group Time frame). The first time breakpoint must be before or equal to the simulation start time, and the last time breakpoint must be equal or later than the simulation stop time (see Data Group Time frame). Time breakpoints must be given in ascending order.

78

Deltares

Graphical User Interface 4.5.7.4 Sediment Sediment transport offers the following functionalities (for details see Chapter 11): Cohesive sediment transport, including the effect of salinity on flocculation. Non-cohesive suspended sediment (sand) transport. Bedload transport, including effect of wave asymmetry. Influence of waves and hindered settling. Updating the bed-level and feedback to the hydrodynamics. Up to 99 fractions of sand or combination of sand and mud. Numerous sediment transport formulations If you have switched on the Sediments process, you can specify the characteristics of the (non-) cohesive sediments, the critical shear stresses for sedimentation and erosion (for cohesive sediments) and the initial sediment at the bed. Upon selecting the tab Sediment in the Data Group Physical parameters the window given by Figure 4.36 is displayed. Open / Save The sediment characteristics can be read from or saved to the file <∗.sed>. See Appendix B.9 for a detailed description.

Overall sediment data Reference density for In high concentration mixtures, the settling velocity of a single parhindered settling ticle is reduced due to the presence of other particles. In order to account for this hindered settling effect Richardson and Zaki (1954) are followed. The reference density is a parameter in their formulation. See Section 11.1.4.

Cohesive sediment Specific density

Dry bed density

Settling velocity

For sediment transport the Eckart relation is extended to include the density effect of sediment fractions in the fluid mixture, see Section 11.1.3. The specific density of each sediment fraction is part of this formulation. The thickness of the sediment above the fixed layer is calculated by dividing the mass of sediment available at the bed by the user specified dry bed density. The settling velocity of the cohesive sediment. Remarks: The settling velocity has to be specified in mm/s in the GUI but is written in [m/s] to the <∗.sed> file In salt water cohesive sediment tends to flocculate to form sediment “flocs”, with the degree of flocculation depending on the salinity of the water. This has been implemented, see Section 11.2.1. Though the GUI does not support this formulation, you can still edit the <∗.sed> file to use this functionality.

Press Next to browse to the next cohesive input parameters, see Figure 4.37.

Deltares

79

Delft3D-FLOW, User Manual

Figure 4.36: Sub-data group Sediment, overall and cohesive sediment parameters

Figure 4.37: Sub-data group Sediment, cohesive sediment parameters (continued)

80

Deltares

Graphical User Interface

Figure 4.38: Sub-data group Sediment, cohesive sediment parameters (continued)

For cohesive sediment fractions the fluxes between the water phase and the bed are calculated with the well-known Partheniades-Krone formulations (Partheniades, 1965): Critical bed shear stress for sedimentation Critical bed shear stress for erosion If the bed shear stress is larger than the critical stress, no sedimentation takes place. If the bed shear stress is smaller than the flux is calculated following Partheniades-Krone, see Section 11.2.3. If the bed shear stress is smaller than the critical stress, no erosion takes place. If the bed shear stress is larger than the flux is calculated following Partheniades-Krone, see Section 11.2.3.

Space-varying shear stress files can be made with Delft3D-QUICKIN. Press Next to browse to the next cohesive input parameters, see Figure 4-38. Sediment erosion rate Initial sediment layer thickness at bed Erosion parameter in the formulation of Partheniades-Krone, see Section 11.2.3. The initial sediment layer thickness at the bed in metre. Remarks: The initial bed composition should be specified at the cell centres. You can use Delft3D-QUICKIN with the Operations → Data in Cell Centre option on to prepare the file with spacevarying initial sediment layer thickness at the bed. GUI v3.39.25 and higher writes a new keyword (IniSedThick) to the <∗.sed> file. The sediment now is in metre, before it was in kg/m2 . Old <∗.sed> files will be converted by the GUI.

Deltares

81

Dry bed density Median sediment diameter Initial sediment layer thickness at bed Domain: 82 Deltares .39.3.sed> file. User Manual Figure 4. You can use Delft3D-QUICKIN with the Operations → Data in Cell Centre option on to prepare the file with spacevarying initial sediment layer thickness at the bed. The formulation used depends on the diameter of the sediment in suspension: See Section 11. see Section 11. Old <∗. The thickness of the sediment above the fixed layer is calculated by dividing the mass of sediment available at the bed by the user specified dry bed density. Remarks: The initial bed composition should be specified at the cell centres. The initial available sediment at the bed. The sediment now is in metre.25 and higher writes a new keyword (IniSedThick) to the <∗.1. before it was in kg/m2 .sed> files will be converted by the GUI.39. GUI v3.3. The specific density of each sediment fraction is part of this formulation. you have to specify the following parameters. Remark: The sediment diameter has to be specified in µm in the GUI but is written in m to the <∗.sed> file. The settling velocity of a non-cohesive (“sand”) sediment fraction is computed following the method of Van Rijn (1993).39: Sub-data group Sediment.Delft3D-FLOW. Specific density For sediment transport the Eckart relation is extended to include the density effect of sediment fractions in the fluid mixture. see Figure 4. non-cohesive sediment parameters Non-cohesive sediment If you have defined a non-cohesive sediment fraction.1.

0 3000.0 1000.0 0. This means that the sediment load entering through the boundaries will be near-perfectly Deltares 83 .sed>.0 100.001 0.0 2000. Number 4000. See Section 11. The mean sediment diameter value is written in [m] to the file <∗.0 0. density For coarser non-cohesive material you can specify that.05 sand: 5.0 mud: 100. You can include the influence of waves by running this version of Delft3D-FLOW in combination with the Delft3D-WAVE module.0 Default 1600. Include effect of You can include or neglect the effect of sediment on the fluid density sediment on fluid by setting this option accordingly.0 1.0 sand: 50.0 3000. General Update bathymetry If you want to take into account the feedback of bottom changes on during FLOW the hydrodynamics.0 Unit kg/m3 kg/m3 kg/m3 kg/m3 mm/s Settling velocity Median sediment diameter Critical shear stress for sedimentation Critical shear stress for erosion Sediment erosion rate Initial sediment layer at bed > 0.0 sand: 1600.0 µm N/m2 N/m2 kg/m2 /s m Remarks: The settling velocity value is written in [m/s] to the file <∗. 4. the quantity of sediment available at the bed will still be updated. See Appendix B.6 for details.sed>.0 0.0 64.7.5.40.0 30.25 200 1000. tick off this option.0 100.5 Morphology With the feedback of bottom changes to the hydrodynamic computation you can execute a full morphodynamic computation.0 mud: 500.0 Upper limit Max. the flow should enter carrying all “sand” sediment fractions at their equilibrium concentration profiles.5 0.0 2650. at all open inflow boundaries.0 mud: 10. regardless of the state of this flag. Open / Save The morphology characteristics can be read from or saved to the file <∗.9 for a detailed description. Selecting the tab Morphology in the Data Group Physical parameters opens Figure 4.0 0.mor>. simulation Remark: The use of this update option only affects the updating of the depth values (at ζ and velocity points) used by flow calculations at subsequent time-steps.0 sand: 500.Graphical User Interface Parameter Reference density Specific density Dry bed density Lower limit 100.0 0.0001 mud: 0.

This 84 Deltares .40: Sub-data group Morphology adapted to the local flow conditions. or years to change significantly).Delft3D-FLOW. When not activated the inflow concenconcentration profile at trations specified in Data Group Boundaries → Transport conditions inflow boundary will be used. and very little accretion or erosion should be experienced near the model boundaries. tidal flows change significantly in a period of hours.8. months. One of the complications inherent in carrying out morphological projections on the basis of hydrodynamic flows is that morphological developments take place on a time scale several times longer than typical flow changes (for example. factor Remark: The Morphological scale factor can also be time-varying. whereas the morphology of a coastline will usually take weeks.9. One technique for approaching this problem is to use a Morphological time scale factor whereby the speed of the changes in the morphology is scaled up to a rate that it begins to have a significant impact on the hydrodynamic flows. User Manual Figure 4. Morphological scale The above can be achieved by specifying a non-unity value. see Appendix B. Equilibrium sand Tick off to use this option.

Minimum depth for In the case of erosion near dry points a threshold depth for computsediment calculation ing sediment transport can be used. see Section 11. During this time interval all other calculations will proceed as normal (sediment will be available for suspension for example) however the effect of the sediment fluxes on the available bottom sediments will not be taken into account. Because of the more complex.6. Sediment transport parameters The following parameters are only relevant for non-cohesive sediments. This allows accelerated bed-level changes to be incorporated dynamically into the hydrodynamic flow calculations. van Rijn’s reference height factor For non-cohesive sediment (e. If the remaining sediment thickness is less than the userspecified threshold and erosive conditions are expected then the source and sink sediment flux terms. erosive flux terms used for “sand” type sediments it is not possible to use the simple source limitation technique used for cohesive sediments. we follow the method of Van Rijn (1993) for the combined effect of waves and currents.mor> file manually. The reference height formulation contains a proportionality factor called van Rijn’s reference height factor. Frequently. Instead. It is likely that during this stabilisation period the patterns of erosion and accretion that take place do not accurately reflect the true morphological development and should be ignored. see Eq. sand). at each computational time-step. the wave related roughness is related to the estifactor mated ripple height. (11.Graphical User Interface feature is not yet supported by the GUI. At each time-step the thickness of the bottom sediments is calculated. Spin-up interval before Specify a time interval (in minutes after the start time) after which the morphological changes morphological bottom updating will begin. see Section 11. are reduced in the following manner: Estimated ripple height In case of waves. You have to edit the <∗.3. see Section 11.g.3. a hydrodynamic simulation will take some time to stabilise after transitioning from the initial conditions to the (dynamic) boundary conditions.2. partly explicit — partly implicit.3. Threshold sediment thickness Deltares 85 . you must specify a Threshold sediment thickness.32). The implementation of the Morphological time scale factor is achieved by simply multiplying the erosion and deposition fluxes from the bed to the flow and vice-versa by this scale factor.

1. where for each wet cell. Sbedload waves .12) where Sbedload current . the standard scheme will not allow erosion of the adjacent cells. if this factor equals zero. Stotal = BED · Sbedload current + BEDW · Sbedload waves + SUSW · Ssuspended waves (4. Current-related reference concentration factor Only Van Rijn (1993) and Van Rijn (2004) compute explicitly a wave-related transport component. see Section 11. The reference concentration is calculated in accordance with Van Rijn (2000). the erosion volume is distributed over the wet cell and the adjacent dry cells. Entrainment. In the case of erosion near a dry beach or bank. Therefore a scheme has been implemented. even when a steep scour hole would develop right next to the beach. Ssuspended waves .Delft3D-FLOW. Wave-related bedload transport factor : — Domain: 86 Deltares . the standard scheme is used. and Deposition depend on the sediment transport formula used.3. Current-related transport vector magnitude factor : — Wave-related suspended transport factor : The wave-related suspended sediment transport is modelled using an approximation method proposed by Van Rijn (2001).11) dDP S = Stotal out − Stotal in + SUS · (Entrainment − Deposition) dt (4. User Manual Multiplication (calibration) factors The following parameters are only relevant for non-cohesive sediments. if this factor equals 1. if there are dry points adjacent to it.3. Factor for erosion of adjacent dry cells The distribution is governed by a user-specified factor for erosion of adjacent dry cells.5. which determines the fraction of the erosion to assign (evenly) to the adjacent cells. see Section 11. In most cases it holds Sbedload waves = 0 and Ssuspended waves = 0. all erosion that would occur in the wet cell is assigned to the adjacent dry cells.

0 10.0 Unit none none none min m m - 4.0 0. see Figure 4.0 0. Deltares 87 .0 0.0 1.0 100. In the Wind tab you can select: Space varying wind and pressure Uniform To define a space and time varying wind and pressure field.0 2. To define a wind field that is time dependent but uniform in space.1 0.0 1.0 1.0 0. Wind fields can be specified as uniform but time dependent or as a space and time varying wind (and pressure) field.0 0.7.0 1.5. Open the file with wind and pressure data.0 1.0 0.0 0.0 0.4 0. Selecting this tab displays Figure 4.1 1.0 Upper limit Default yes no yes 1.0 0.6 Wind When the Wind process is activated in Data Group Processes.0 100.005 1.0 720.0 100.16.0 5.0 0.0 10000. Upon selecting Uniform you can either specify the wind data or read an attribute file.0 10.41.05 2. the tab Wind is available in the Data Group Physical parameters.0 100.Graphical User Interface Parameter Update bathymetry during FLOW simulation Include effect of sediment on fluid density Equilibrium sand concentration profile at inflow boundary Morphological scale factor Spin-up interval Minimum depth for sediment calculation Van Rijn’s reference height Threshold sediment thickness Estimated ripple height factor Factor for erosion of adjacent dry cells Current-related reference concentration factor Current-related transport vector magnitude factor Wave-related suspended transport factor Wave-related bedload transport factor Lower limit yes or no yes or no yes or no 0.

Domain: 88 Deltares . Specify the time breakpoints in [dd mm yyyy hh mm ss].Delft3D-FLOW. i. Interpolation Type Time-Series The wind direction is defined according to the nautical definition.41: Wind definition window Uniform wind For a uniform wind you can specify: Open Save To read an attribute file with a uniform but (potentially) time dependent wind field. file extension <wnd>. If you want to combine wind data read from a file and defined manually you must first import the file and then add. Remark: The wind direction is measured (clockwise) starting from true North. To save the manually defined or modified wind field to a file. User Manual Figure 4. East-North-East wind thus has a direction of +67. to determine the value at intermediate times you can apply a linear or a block type of interpolation. delete or modify the imported data. an East-North-East wind.42 the wind direction is about +60 degrees.e. Time dependent data is specified at time instances called breakpoints.5 degrees. reversing this order will overwrite your manually defined data. the wind speed in [m/s] and the wind direction in degrees (nautical definition). In Figure 4. relative to true North and positive measured clockwise. i.e.

7 for details.7. which in turn causes a water level gradient. and the last time breakpoint must be equal or later than the simulation stop time (see Data Group Time frame).0 360. To reduce this effect.0 0. When using space varying wind and pressure. It is also possible to use space varying wind and pressure defined on an independent wind grid.1. Time dependent data is specified at time instances called breakpoints.2.0 Default 0.Graphical User Interface N orth W ind direction W est E ast South Figure 4.8. Space varying wind and pressure For a space varying wind and pressure you can specify: Select file Interpolation Type To select an attribute file with space and time varying wind and pressure data. Deltares 89 .0 0. This functionality is described in Section B. Time breakpoints must be given in ascending order. Remarks: The space varying wind and pressure selected in this sub-data group must be defined on the hydrodynamic grid. the input is applied at the interior of the domain. See Appendices A.42: Nautical definition wind direction Parameter Wind speed Wind direction Lower limit 0.0 Unit m/s degrees North Restrictions: The time breakpoints must differ from the simulation start time by an integer multiple of the time-step (see Data Group Time frame).2 and B. to determine the value at intermediate times you can apply a linear or a block type of interpolation. it is possible to apply a pressure correction on open boundaries. The first time breakpoint must be before or equal to the simulation start time. On open boundaries this can cause a pressure gradient.0 Upper limit 100. The file is not read.

Several drying/flooding procedures are available in Delft3D-FLOW.8 for more details on drying and flooding. see Section 10. If the flow at the four faces of a computational grid cell is blocked this grid cell is set dry.44. User Manual Figure 4. then you will see less input parameters. Note: if your model is not 3D. You must select or define one or more of the following parameters: Drying and flooding check 90 Determine if you want to apply an additional check on the water Deltares . See Section 9.9 for details of tide generating forces.7 Tidal forces When the process Tidal forces is selected (see Figure 4. Click the Data Group Numerical parameters.16). several options are available.Delft3D-FLOW.43 is displayed. By default the water depth in each of the grid cell faces (velocity points) of a computational grid cell is determined to decide if the flow should be blocked by a (temporary) thin dam. the tab Tidal forces is available in Data Group Physical parameters.7. The bottom depth in the cell centre (water level point) is not uniquely defined. Selecting this tab Figure 4. You can select the components that are taken into account by the tidal force generating routine by ticking off the check box.5. 4.5. or yo have not modelled salinity or another constituent. In combination with the water level in the cell centre it is determined if the cell is being dried or flooded.8 Numerical parameters In the Data Group Numerical parameters you can specify parameters related to drying and flooding and some other advanced options for numerical approximations. An additional check can be done on the water depth at the centre (water level point). Note that this does not guarantee the water depth at the grid cell centre (water level point) to be positive definite.43: Sub-window Tidal forces 4. This results in Figure 4. see Chapter 10.

44: Data Group Numerical parameters Depth specified at depth at the cell centres (water level points). the depth at Deltares 91 . Upwind The depth at the u-velocity point is the depth at water level point (m. Min The depth at the cell centre is the minimum of the 4 surrounding depths at the cell corners. Select if the depth values have been generated at the cell corners (depth points).8.n). Min The depth at the cell face is the minimum of the depths at the water level points (m. or only a check at the cell faces (velocity points).Graphical User Interface Figure 4.n) if the u-velocity is larger than 0. Mean The depth at the cell centre is the mean of the 4 surrounding depths at the cell corners. Depth at grid cell centres Depth at grid cell faces You can either select (see also Section 10. You can either select (see also Section 10.2): Mean The depth at the cell face is the mean of the depths at the cell corners (m. Remark: Default Delft3D-QUICKIN generates data at the grid cell corners.1): Max The depth at the cell centre is the maximum of the 4 surrounding depths at the cell corners. or at the cell centres (water level points).n) and (m+1.n+1). but you can also choose to generate data at cell centres.8.n) and (m.

Remark: Depth specified at grid cell centres and Depth at grid cell faces is Mean can not be used together. If this water depth is less than the marginal depth then: The water depth equals the bottom depth plus the water level at (m. User Manual water level point (m+1. This would result in iterations in the computation and thus a larger computational time. This can introduce large spurious waves that enter the model area. The water depth equals the bottom depth plus the maximum of the water levels at (m.n) and (m+1. the prescribed initial condition and the boundary values at the start time of the simulation do not match. Example: With a tidal amplitude of 2 m and time step of 1 minute (N = 720) we find δ ≥ 0.n) if the u-velocity is less than 0. is a measure for the change of Marginal depth the water level as a function of time with a as characteristic amplitude. For the water level at the velocity points as default the mean value is used.n) if the u-velocity is 0. during which the boundary condition will be adapted gradually starting from the prescribed initial condition value. In some cases.n) if the u-velocity is larger than 0. ∆t is the time step of the computation and N is the number of time steps per tidal period. The threshold depth must be defined in relation to the change of the water depth per time step in order to prevent the water depth to become negative in just one time step. Subsequently. As a rule of thump you can use: δ≥ ∂ζ 2π |a| 2π |a| ∆t = ∆t ≈ ∂t T N ∂ζ ∂t (4. It is also possible to determine the water levels at the cell faces with a so-called upwind approach: initially the water depth is the bottom depth at the velocity point plus the water level at the velocity point. Deltares 92 . Mor The Mor option is the same as Min. Threshold depth The threshold depth above which a grid cell is considered to be wet. and the average if the u-velocity is 0.Delft3D-FLOW. To reduce this time. the wave will be reflected at the internal boundaries and along the open boundaries of the model until the wave energy is dissipated completely by bottom or internal (eddy viscosity) friction forces. T is the tidal period.017 m.13) where δ is the threshold depth. Smoothing time The time interval used at the start of a simulation for a smooth transition between initial and boundary conditions. a smooth transition period can be specified.n) if the u-velocity is less than 0. The water depth equals the bottom depth plus the water level at (m+1. These reflections can be observed as spurious oscillations in the solutions and they will enhance the spin-up (warming-up) time of the model.

However. Advection scheme for momentum Select the type of numerical method for the advective terms in the momentum equation. The WAQUA-scheme and the Cyclic method do not impose a time step restriction. see Section 10. see Stelling and Leendertse (1992). It might still be necessary to remove wiggles generated by the central difference scheme in the vertical. Remark: The continuity equation is solved iteratively. see Chapter 10. Default the standard Cyclic method is applied. Currently. when a monotonous solution (no over. The default number of iterations is 2.5. You can add this keyword and value manually to the MDF-file or specify them under Additional parameter. then the slightly less accurate but monotonous Van Leer-2 method can be applied (see Chapter 10). you can specify the number of iterations. With the keyword and value Iter = #. The third scheme can be applied for problems that include rapidly varying flows for instance in hydraulic jumps and bores (Stelling and Duinmeijer. see Stelling (1984) and Stelling and Leendertse (1992). Select the type of numerical method for the advective terms in the advection-diffusion equation for constituents. two numerical schemes are supported: theCyclic method (accurate. See Section 10. three options are available in Delft3D-FLOW. Currently. In Relaxation factor QH-forcing Deltares 93 . The time integration is based on the ADI-method.Graphical User Interface For the spatial discretization of the horizontal advection terms.4. This factor only applies if QH-forcing is used at open boundaries. Van Leer-2 method (less accurate. see Chapter 10. The integration of the advection term is explicit and the time step is restricted by the Courant number for advection. see Section 4. The first scheme is denoted as the Cyclic method. 2003).4. The first and second option use higher-order dissipative approximations of the advection terms. the following numerical schemes are supported: the Cyclic scheme the Waqua scheme the Flood scheme Default the standard Cyclic method is applied. The second scheme is denoted as the WAQUA-scheme.11. For the Flooding scheme the accuracy in the numerical approximation of the critical discharge rate for flow with steep bed slopes. Threshold depth for critical flow limiter Advection scheme for transport For the Flood scheme you have to specify a threshold depth which determines how the water depth is calculated. but positive definite results are not guaranteed). where # is the required number of iterations. see Section 10.8 for details.1 for more details. can be increased by the use of a special approximation (slope limiter) of the total water depth at a velocity point downstream. but positive definite results guaranteed).and undershoots) in the horizontal is required.4. The scheme is denoted as the Flooding scheme and was developed for 2D simulations with a rectilinear grid of the inundation of dry land with obstacles such as road banks and dikes.

These negative concentrations can be suppressed by applying a horizontal Forester filter. If no relaxation is required. followed by the filter in the vertical direction. The horizontal Forester filter acts on the salinity. as this quantity can be negative by its nature. User Manual some cases. Owing to truncation errors. However. generated by the central differences in the vertical. Forester filter Correction for sigma co-ordinates Domain: 94 Deltares . when discharge fluctuates quickly. In the vertical direction the solution is made monotone for salinity and temperature only. Activate or de-activate a Forester filter in the horizontal or in the vertical direction. The horizontal Forester filter is not applied to the spiral motion intensity. are smoothed. the horizontal filter is applied first. A counter measure (a so-called anti-creep approach) can be activated to suppress this phenomenon. small negative sediment concentrations (smaller than 1. In the horizontal direction numerical diffusion is added in points where the quantity (concentration) is negative.e.9. the vertical filter is not applied to constituents or suspended sediments. this parameter should be set to 0. The diffusion is two-dimensional along σ -planes. under-relaxation is necessary to stabilise the forcing by a QH-relation. σ co-ordinates have a slight disadvantage in case large bottom gradients exist in the model. Activate or de-activate an anti-creep correction. This parameter in the range between 0 and 1 specifies the amount of under-relaxation. including suspended sediments. i.Delft3D-FLOW.10−3 kg/m3 ) can be found. Remark: In 3D sediment transport computations. see Appendix B. When the Forester filter is applied in both directions. the penalty can be a substantially larger computational time. artificial vertical diffusion and artificial flow may occur in this case. the temperature and the concentrations of dissolved substances. Wiggles.

Min. 4. then the layer number k has to be specified. Flood 0. In addition to the discharge rate the concentration of the constituents released and the temperature and/or the salinity. Min Mean.0 Remark: A warning will be issued if the Threshold depth exceeds 1. n and k). Upwind. Upon selecting the tab Discharges the sub-window of Figure 4. depending on the selections made in the Data Group Processes must be specified.0 m. and its position in the grid (respectively m. Discharge In Delft3D-FLOW intake stations and waste-water outfalls can be considered as localised discharges of water and dissolved substances.0 off off Unit m m min none m none none At cell faces always.5.9 Operations Upon selecting the Data Group Operations the sub-window of Figure 4.45 is displayed. Waqua. Mor 0. Cyclic. The tab Dredging and dumping is only available if the Sediments process has been selected and the man-made process Dredging and dumping has been switched on.999 0. If the discharged effluent is to take place at one point in the vertical.1 -999. Deltares 95 .0 Cyclic 0.0 10 n.45 is displayed. However. Mean. Van Leer-2 0.0 Cyclic 0. A discharge (or source) in the model domain is characterised by a name which serves as identification.0 Cyclic.a.0 on or off on or off 1. optional at cell centres grid cell corners or centres Max. then the layer number K must be set to 0. if the discharged flow is to be distributed uniformly over all layers (taking into account the layer thickness).0 0.Graphical User Interface Parameter Drying and flooding check Depth specified at Depth at grid cell centres Depth at grid cell faces Threshold depth Marginal depth Smoothing time Advection scheme momentum Threshold depth flow limiter Advection scheme transport Relaxation factor QH Forester filters Correction for sigma co-ordinates Lower limit Upper limit Default at both cell corners Max Mean 0.

45: Data Group Discharges Remarks: You can model a river as an open boundary or as a discharge. and Specify the name of the discharge in the Name input field. but not the concentrations. temperature or salinity. respectively. i. Add To Add a discharge either: Click Add. (M. The various input items you can select are discussed below.Delft3D-FLOW. Click Delete. These quantities are computed in the WAQ and PART module in combination with the other water quality processes. K). N. 96 Deltares . Specify the grid indices and the layer of the release point. or Use the Visualisation Area window. The discharge rate is communicated to the water quality and particle tracking modules WAQ and PART.e. User Manual Figure 4. Delete To Delete a discharge either: Select the discharge to be deleted in the list box. K only if you have a 3D model.

97 Deltares .46. the locations to a file with extension <src>. except for an In-out discharge: a positive rate means withdrawal at the intake. When prescribing the data the momentum and its direction relative to the North must be specified. the discharge rates to a <dis> file. Remark: A positive rate indicates discharging into the model. The intake and outlet discharge rates are the same. Remark: If you want to combine discharges read from an attribute file and inserted manually you must read the file first and then add the manually defined discharges. The discharge rate in [m3 /s]. K) of the outlet related to the just defined inlet. The same amount of water is discharged into the model at the outlet.Graphical User Interface or Use the Visualisation Area window. Walking : the discharge moves to the nearest wet grid cell if the cell in which the discharge is released becomes dry. Type The discharge can be of type: Normal : the discharge rate is released without taking specific aspects into account. Open/Save Open or save the discharge data to attribute files. Momentum: the momentum of the discharge is taken into account. Upon selecting Edit data a sub-window is displayed. the concentration of a constituent at the outlet equals the concentrations at the inlet plus the user-defined concentration (when applicable). These indices are only required and accessible for the discharge Type = In-out. Outlet location You define the grid indices (M. for more details see below under specify v -direction. Block : the last value is repeated until the next value is defined. In-out : intake and outlet are on different locations. Interpolation The interpolation of time dependent input data can be either Linear : intermediate values are determined by linear interpolation. Edit data The time dependent data are defined at time breakpoints and their intermediate value is determined by either linear or block wise interpolation as defined under Interpolation. Depending on the selections made in the Data Group Processes you must specify more or less quantities. It moves back to its original position when the grid cell is flooded again. In reverse order the manually defined discharges are overwritten by those of the file. N. In the most complete case you must specify: Time Flow Date and time of the time breakpoints. see Figure 4. To specify the discharge rate and the concentration of the discharged substances.

Only if Salinity is selected in the Data Group Processes. Only for discharge of Type = Momentum.46: Sub-window to define the discharge rate and substance concentrations Figure 4. The momentum of a discharge released in the third quadrant is decomposed over the u. The mass (of water and constituents) is released in the (water level) point with grid indices (m.and v -velocity points with grid indices of respectively. Only for discharge Type = Momentum. N) and (M. n) v -magnitude v -direction The speed with which the discharge is released in [m/s]. The direction in which the momentum is delivered to the flow. (M−1. Salinity Temperature The salinity concentration in [ppt].47. Example: the momentum of a discharge released in the first quadrant as seen from the discharge location (water level point). n). is decomposed over the u.e. measured (clockwise) from the North. The momentum released to the flow due to a discharge is distributed over the velocity points that bound the quadrant in which the discharge is released. Only if Temperature is selected in the Data Group Processes. see Figure 4. User Manual Figure 4. The direction is defined similar to the wind direction.47: Decomposition of momentum released by a discharge station in (m. i.Delft3D-FLOW. The temperature of the discharge in [◦ C]. n).and v -velocity point with grid indices (m. 98 Deltares . N−1).

0 +10.0 10. you must enter K = 0.0 0.0 0.0 60. A positive discharge represents an inflow into the model area. Reversing this order will overwrite the interactively defined discharges. -10. The first time breakpoint must be before or equal to the simulation start time. A discharge can be located in an area that becomes temporarily dry during the simulation. If the discharge is to be distributed uniformly over the layers. and contains up to 20 characters.000 100. Time breakpoints must be given in ascending order. The same amount is an inflow at the outlet. Deltares 99 . then the attribute file must be read first. The time breakpoints specified for time-series must differ from the simulation start time by an integer multiple of the time-step (Data Group Time frame). In such a case there will be no intake. within the computational domain and not at the model boundaries. the specified concentration is added to the concentration of the water withdrawn. for Type = Normal.0 -360.0 0.0 100. a negative discharge represents an outflow from the area. a positive discharge represents withdrawal at the intake. The number of discharges is limited to 500. Walking discharges move to the nearest wet grid cell by looking at the steepest descent of the surrounding bathymetry. and the last time breakpoint must be equal or later than the simulation stop time (see Data Group Time frame). Furthermore.000 0. Walking and Momentum. Remarks: Layer 1 corresponds to the surface layer.0 0.0 m3 /s m/s degree ppt ◦ v -magnitude v -direction Salinity Temperature Concentration C kg/m3 Restrictions: The discharge name must be non-blank.000 0.0 0. Only if Pollutants and/or Sediments are defined in the Data Group Processes.0 0. Domain: Parameter Location Flow Lower limit Upper limit Default Unit Anywhere in the grid. For Type = In-out. If discharges are specified both through an attribute file and interactively. unless you have specified the discharge to be of Type = Walking.0 0.Graphical User Interface Pollutants and sediments The concentration of the constituents released in [kg/m3 ].0 0.0 360. The discharge must be located within the grid enclosure.

You can apply one or more of the following options: Add To Add an observation point either: Click Add. 100 Deltares . i. Observation points represent an Eulerian viewpoint at the results.1 Dredging and dumping When sediments are modelled and the man-made process Dredging and dumping is switched on (see Data Group Processes). n).e.5. User Manual Figure 4.48: Sub-data group Dredging and dumping Figure 4. Specify the grid indices of the observation point. the tab Dredging and dumping is available.e. at water level points. Replace the name “-Unnamed-” by a useful name in the Name input field.10. drogues or cross-sections.e. 4.9. Observation points are located at cell centres. see Figure 4. velocities.10 Monitoring Computational results can be monitored as a function of time by using observation points.50 is displayed.5.48. i. temperature and concentration of the constituents.dad> with dredging and dumping data.1 Observations Observation points are used to monitor the time-dependent behaviour of one or all computed quantities as a function of time at a specific location. i.49: Sub-window for Monitoring locations 4. water elevations. salinity. Upon selecting the data and sub-data group Monitoring → Observations the sub-window of Figure 4. (m.Delft3D-FLOW. fluxes. 4. Click Select file and specify the file <∗. Monitoring points are characterised by a name and the grid indices of its location in the model area. Remark: See the Delft3D-QUICKIN User Manual on how to define dredging and dumping areas.5.

Delete To Delete an observation point either: Select the observation point to be deleted in the list box. Deltares 101 .Graphical User Interface Figure 4. Remark: If you want to combine observation points read from an attribute file and inserted manually you must read the file first and then add the manually defined observation points. Save the observation points defined in an attribute file with extension <obs>. Drogues and Cross-sections are very similar and are discussed at the end of this section. Open Open and read observation points defined in an attribute file. Click Delete.50: Sub-window for Observation points or Use the Visualisation Area window. Save The domain and restrictions for Observation points. or Use the Visualisation Area window. In reverse order the manually defined observation points are overwritten by those of the file.

5. When using the visualisation area you can only specify release points at the grid centres. their grid indices M and N can have fractional values. so if other fractional values are required you have to specify them manually. Upon selecting the data and sub-data group Monitoring → Drogues the sub-window of Figure 4. You can release and recover them at any time during the simulation (provided the time complies with the restrictions for time instances). You can apply one or more of the following options: Add To Add a drogue either: Click Add. this means you can release them at fractional grid cell positions. i. the wind affects the velocities in the surface and thus the path of drogues. i. In the vertical: drogues are released in the surface layer. Specify the grid indices of the drogue release point. Drogues are only transported due to the velocities in the surface layer.2 Drogues Drogues are used to monitor the path of a particle moving with the flow.51 is displayed. When wind applies.51: Sub-data group Monitoring → Drogues 4.Delft3D-FLOW. or Use the Visualisation Area window.10.e. There is no diffusion or random process involved. User Manual Figure 4. Replace the name “-Unnamed-” by a useful name in the Name input field. Remark: Drogues can be released inside grid cells. n). To Delete a drogue point either: Deltares Delete 102 . In the horizontal: drogues can be released anywhere in the grid.e. (m.

Save The domain and restrictions for Drogues. or Use the Visualisation Area window. N1) and end (M2. Open Open and read drogues defined in an attribute file. To define a cross-section. In reverse order the manually defined drogues are overwritten by those of the file. Click Delete. you must specify an arbitrary section name. Upon selecting the data and sub-data group Monitoring → Cross-sections the sub-window of Figure 4. Observations points and Cross-sections are very similar and are discussed at the end of this section. You can apply one or more of the following options: Add To Add a cross-section either: Deltares 103 .grid index and it must include at least two grid points. N2) indices of the section.Graphical User Interface Figure 4.5.or n. fluxes of matter (if existing) and transport rates of matter (if existing) sequentially in time at a prescribed interval. flux rates (hydrodynamic). the begin indices (M1. Save the drogues defined in an attribute file with extension <par>. 4. Remark: If you want to combine drogues read from an attribute file and inserted manually you must read the file first and then add the manually defined drogues.52: Sub-data group Monitoring → Cross-Sections Select the drogue to be deleted in the list box.52 is displayed.3 Cross-sections Cross-sections are used to store the sum of computed fluxes (hydrodynamic).10. A cross-section is defined along a constant m.

or Use the Visualisation Area window. within the computational domain and not at the model boundaries. 2 ≤ N ≤ Nmax−1. 2 ≤ M ≤ Mmax−1. you should define at least 2 sections. N1) and (M2. Save the cross-sections in an attribute file with extension <crs>. If observation points are read from an attribute file and defined interactively. so you have to specify their start and end point. When saving the results as time-series. Replace the name “-Unnamed-” by a useful name in the Name input field. A cross-section must extend over at least two grid points. Delete If you want to combine cross-sections read from an attribute file and inserted manually you must read the file first and then add the manually defined cross-sections. reversing the order will overwrite the manually defined observation points. Restrictions: The release time and the recovery time of drogues must comply with the definition of time parameters. you have to define at least 2 obsservation points.Delft3D-FLOW. In reverse order the manually defined cross-sections are overwritten by those of the file. Drogues must be released in an active grid point. i. Open Save Open and read cross-sections defined in an attribute file. this point may become dry during the computation. N2). Remark: Cross-sections typically extent over more than just one grid cell. Click Delete. the file must be read first. or Use the Visualisation Area window. Specify the grid indices (M1. When using cross-sections. 104 Deltares . Domain: Parameter Location Lower limit Upper limit Default Unit Anywhere in the grid. User Manual Click Add. All cross-sections must be saved in an attribute file upon leaving the sub-data group.e. To Delete a cross-section either: Select the cross-section to be deleted in the list box.

See Appendix B for full details of the currently available Add-ons.53 is displayed.53: Data Group Additional parameters 4. Deltares 105 . file format and file contents. you can specify the required parameters in a separate input file. The option of Additional parameters has been introduced to be more flexible in providing special functionality in the computational code without changing the FLOW-GUI. Upon selecting this data group Figure 4. for details see Appendix B.Graphical User Interface Figure 4. values. for details see Appendix A.5.11 Additional parameters Through the use of an Additional parameter you have access to additional functionalities that are not yet fully supported by the FLOW-GUI.lwl# 4 The first keyword refers to local weirs and the value is the filename in which the location and other properties of the local weir(s) are defined. If you have access to an option. Example: Keyword Value Fillwl Iter #ex-001.1. their keywords. After thorough testing it will be released to all users. Each Additional parameter is characterised by a keyword and a value: Remarks: Keywords are CASE-sensitive! The value is often the name of an attribute file and must be enclosed between #’s.3. The second keyword redefines the number of iterations used when computing the continuity equation to be 4 (instead of the default value 2).

i.5. The Output window. User Manual Figure 4. Delft3D modules use one or more Map. 106 Deltares . For this you specify the restart interval at which all information required to restart the simulation. As you (can) save all results in all grid points a typical Map file can be many hundreds of MB large. Before defining the various input fields we want to explain in short the use of the various data files used to store the results. In order not to waste computing time you can restart a simulation at a predefined time of a previously executed simulation.12 Output In the Data Group Output you can specify which computational results will be stored for further analysis or for other computations and which output shall be printed. You can also restart from a map-file which is platform independent. The amount of data is usually much smaller than for a Map file. As the results must be stored in all grid points a Communication file can typically be as large as or even larger than a Map file. This means you can not use directly. Figure 4.54: Data group Output 4. are stored in a so-called Restart file. For water quality simulations you typically store the results of the last day or a (couple of) tidal cycles and use this data cyclic to make a water quality simulation over many days or weeks.54. a restart file from a simulation executed on another hardware platform. History. and you typically store history data at a small time interval to have a smooth time function when plotting the results. Typically.Delft3D-FLOW. without some kind of conversion. Remark: The restart file is not platform independent. In a History file you store all results as function of time. So typically Map results are only stored at a small number of instances during the simulation. you only store the results in the Communication file as far as needed for the other simulations. Click the Data Group Output. In the Communication file you store data required for other modules of Delft3D. Though the printing option is hardly used it is a useful option when numerical output values are required. such as the hydrodynamic results for a water quality simulation or the wave forces for a wave-current interaction. you select a large restart interval in order not to waste disk space.e. will be displayed. but only in the specified monitoring points. Communication and Restart files to store the simulation results and other information needed to understand and interpret what is on the files: Maps are snap shots of the computed quantities of the entire area. So.

Graphical User Interface Figure 4. but they cannot be changed in this data group (see the Data Group Time frame). Store history results History interval Time interval. to write the History file for the whole simulation. In the top of the Output storage sub-window the simulation start and stop time and the timestep used in the simulation are displayed for your convenience.1 Storage The sub-data group Output storage is displayed in Figure 4. Details. in minutes. to write the Map file. when and which data will be printed. in minutes.5.55: Sub-data group Output storage The data group Output is organised in three sub-data groups: Storage. Deltares 107 . to make a selection of all possible data to be stored in the files or to be printed. to define if. to define the time intervals to various file used. 4. Date and time to stop storing the results in the Map file.55. You can apply one or more of the following options: Store map results Start time Stop time Interval Date and time to start storing the results in the Map file.12. Time interval. Print.

see below for a short description and Appendix A. Write restart files Restart interval Time interval. These maps can be compared with co-tidal maps from literature. Upon selecting Fourier analysis you specify the time period to use. in minutes. Salinity. Online animation is a very powerful tool to let you inspect the behaviour of your model. the layer number. During the simulation you can change your selection of parameter(s) displayed. Mass fluxes of water. the frequencies to use and some other more special items (see Appendix A).Delft3D-FLOW. Upon ticking off this option you can select a file with the time period and the frequencies you want to apply. Stop time Date and time to stop storing the results in the Communication file. For large scale tidal flow models. the type of view used. in minutes. Temperature. Delft3D-FLOW is equipped with a facility to perform online Fourier analysis on computational results.29 for details of the file format. When you specify a zero frequency the minimum. or as cross-section. to write the Restart file. the variable(s) to analyse. The option of Online visualisation lets you monitor the results during the simulation as an animation. 108 Deltares . User Manual Store communication file Start time Date and time to start storing the results in the Communication file. Perform Fourier analysis Fourier Analysis To activate the Fourier analysis on computational results. such as iso-lines and vector plots in plain view per layer. to write the Communication file. View results during the simulation Online visualisation When selecting this option you can inspect the results during the computation. Constituent(s). This enables the generation of computed co-tidal maps. You can select the type of parameter displayed.2. You can select one or more of the following parameters: Water level. mean and maximum level of the variable is computed in the model area over the specified time period. and per layer. Velocity components. a very convenient way to check if the tidal wave propagation is simulated correctly by the model. For a short description see below. this will help you to see anomalies in the results much better and in a much earlier stage than when using plots and figures. Interval Time interval. You can store the results to file and run this file afterwards as an animation.

Remarks: No Map.runid >. History.11) by means of the keywords described in Section B. the name of the file is composed using the prefix of the name of the MDF-file: <tri-prt. Drogues are written to a separate drogue-file for each time step between the release and recovery time. Upon selecting the sub-data group Output → Print a sub-window is displayed. You can not specify the interval to write the drogue-file. When continuing with the Water Quality modules.and mid-field water quality scenarios using these hydrodynamic results. Domain: Parameter Date and time Lower limit Reference date 0.18. 4. You can apply one or more of the following options: Deltares 109 . The latter approach is the preferred the approach but it needs to be switched on using the Additional Parameters (see Section 4.5.00 Unit Valid date and time minutes Interval Restriction: The start and stop time must differ an integer multiple from the Reference date and the time interval for storing results must be an integer multiple of the Time step. The D-Water Quality input files can either be generated using a coupling program that runs concurrently with the Delft3D-FLOW simulation or they can be written by Delft3D-FLOW directly.56.Graphical User Interface Online coupling the communication file Online coupling This option is available when you store results to the communication file. By default the start time to store the hydrodynamic results is used as the simulation start time of the far. If you did not want to create this file. both specified in the Data Group Time frame.2 Print In the sub-data group Output → Print you can specify which results will be printed.0 Upper limit Default Simulation date and time 0. you can either couple the com-file after the FLOW simulation has ended.5. or do it online. you can ignore this warning.12. see Figure 4. When verifying the MDF-file a warning will be issued that the time definition is not correct and that the corresponding file will not be created. The selected results are printed to a file. Communication or Restart file is written if you set the interval to write the data equal to zero.

in minutes. 110 Deltares . Time interval. The print times may be specified in ascending order. to write the history data to a print file. to the list of time instances to be written to the print file. Date and time to stop writing the history data to a print file.56: Sub-data group Output → Print Print history results Start time Stop time Interval Date and time to start writing the history data to a print file. By ticking the check boxes you can specify which items you want the program to store on file or to print and which not. A warning is issued if the maximum number of time instances is reached. flows and transported quantities at all points are stored in an ASCII-formatted file for printing. Select a time instance from the list box and press Delete to remove a time instance from the list box. specified in the input field below the list box.Delft3D-FLOW.3 Details In the sub-data group Output → Details you can make a selection of all possible data to be stored in the files or selected for printing.5.12. Print Map Results Add Delete Add the map time. Upon selecting the sub-data group Figure 4. the computed water elevations. 4.57 is displayed. User Manual Figure 4.

Save All to save all attribute data in the current attribute files. Save All As to save all attribute data under a new name.58. Select: Save MDF to save the MDF-data under its current name.6 Save the MDF and attribute files and exit To save the MDF-file or any of the attribute files: Click File in the menu bar of the FLOW-GUI window and select any of the Save or Save As options.58: File drop down menu 4. Save MDF As to save the MDF-data under a new name. For saving the files and exiting the FLOW-GUI only the last five options in the file drop down menu are relevant.Graphical User Interface Figure 4. see Figure 4.57: Output Specifications window Figure 4. Exit to exit the FLOW-GUI and return to the Hydrodynamics window. Deltares 111 .

If you want to save the changes to the attribute files as well. including the MDF-file.59. have been saved. as they are still available in the input fields of the FLOW-GUI. request for a filename if not yet defined and proceed to the next unsaved data item. The non-saved changes are not lost. User Manual Figure 4. Select one of the options displayed: Yes Save the first unsaved data item. In this case none of the data items nor the MDF-file itself has been saved yet.59: Save changes window Upon selecting Save MDF or Save MDF As the MDF-file is save with a reference to the attribute files as they existed at the time you started the edit session or after the last save command. These Save MDF and Save MDF As options allows you to ignore some or all of the changes you have made to some or all attribute data and restore the last saved situation. including the MDF-file. you must select Save All or Save All As.Delft3D-FLOW. With Save All As you must specify a (new) filename for all the files. Existing files are overwritten when you select Save All . you are requested to specify a filename if the attribute file does not yet exist. if not all data items have been saved yet a list of all non-saved (attribute) data will be displayed and you can select for each non-saved data item the required action. Changes to attributes are ignored if you have not explicitly saved them. However. the FLOW-GUI is closed and the Hydrodynamics (including morphology) window of Delft3D-MENU is redisplayed. Deltares 112 . To exit the FLOW-GUI: Click File → Exit. which must be saved. see Figure 4. If all data items.

all attribute files are imported automatically. If the reference to an attribute file does not exist.Graphical User Interface YesAll No NoAll Cancel Save all unsaved data items and request a filename if not yet defined. removing and exporting of data When starting a scenario definition session from an existing MDF-file. If you want to concatenate data items from two files you must do so offline with your favourable editor. Don’t save the first unsaved data item. 4. proceed to the next unsaved data item. you are referred to Appendix A for details on their format and contents. Warnings: If you want to combine attribute data from an existing file with manually defined data.7 Importing. Deltares 113 . or if you wish to apply a different attribute file you can apply the Open or Select file option. Abort this Exit action and return to the FLOW-GUI. The reference to an empty attribute file (because you have deleted all data items) is automatically removed from the MDF-file upon saving the MDF-file. Exit without saving any unsaved data. you must open the existing data before you add new data items. After you have worked through all unsaved data items the FLOW-GUI will be closed and the Hydrodynamics (including morphology) window of Delft3D-MENU will be displayed. Most attribute files are ASCII-files and free formatted.

User Manual 114 Deltares .Delft3D-FLOW.

discharges and salinity concentrations or other substances transported by the flow. a curvilinear and a spherical grid.e. the bathymetry. The land boundary is obtained from digitising a map or from GIS and the bathymetry is obtained upon using the tool Delft3DQUICKIN. the bathymetry and other files. Definition of many different grid related quantities. Finally. start and stop time and various time functions. both the boundary between wet and dry areas and the location of the open boundaries where water level or flow conditions are prescribed). the basic steps that precede the definition of an input file can be summarised as: Selection of the extent of the area to be modelled. For details you are referred to the User Manuals of Delft3D-RGFGRID and Delft3DQUICKIN. The data of the land boundary. a numerical grid must be defined on which all location related parameters are being defined. Deltares 115 . structures. such as the extent of the model area. To generate a curvilinear body fitted grid you need a file with the land boundary and preferably the bathymetry.5 Tutorial 5. is stored in an input file. release points of drogues.e. the bathymetry and the numerical grid. see Appendix A. They result in most cases in one or more files that are to be located in the project directory to be defined when starting the project (see Chapter 8). To generate a curvilinear grid you can use the grid generator program Delft3DRGFGRID .1 Introduction – MDF-file and attribute files In this Tutorial we will guide you through the process of creating a simple example of a flow simulation. Rectilinear and spherical grids are generated automatically given the grid dimensions and sizes. All information for a flow simulation. the structure of an MDF-file and the basic steps you are supposed to execute. Delft3D supports the use of a rectilinear. Most activities. provided with Delft3D. (i. For the file formats of the land boundary. geometrical details of the area such as breakwaters. i. such as open boundaries. but the last two. monitoring points. i. but also information on boundary conditions. must be executed before starting the FLOW Graphical User Interface (GUI). discharges and grid related quantities such as monitoring points. also called scenario. discharges and the definition of which and where results of the simulation need to be stored for later inspection or for other simulations. Generation of a numerical grid. wind speed and direction. Definition of the time frame of your scenario. velocities or discharges. such as the open boundary conditions. The project directory is also referred to as the working directory. before starting this input definition process we want to explain in short the basics of a model definition. So. and no tool is required.e. Generation of the bathymetry defined on the numerical grid. water level. no tools are available to support these steps others than (GIS-based) maps and (digitised) charts. Definition of the land(-water) boundary. For this tutorial these files are provided. Definition of location and extent of open boundaries and the type of boundary conditions to be prescribed. The first two steps are based on experience in solving similar problems and on engineering judgement. To execute a flow simulation for a specific area we need various kinds of information. discharge locations. also known as Master Definition Flow file (MDF-file). However.

2 Filenames and conventions The names of the MDF-file and of attribute files have a specific structure. The area modelled concerns the entrance between two islands in the northern part of The Netherlands and is called “Friesian Tidal Inlet”. Example: <c:\projects\panama\r01-calibration. On the outer side of the area the flow is driven by a semi-diurnal tide.mdf>. two attribute files are provided for use in this tutorial.and two numeric-characters followed by a useful name of your project. Furthermore. The extension <mdf> is mandatory. The complete MDF-file and all attribute files created in this tutorial are also provided in the Tutorial directory. But. To create or modify an MDF-file you use the FLOW-GUI you became acquainted with in Chapter 3. but instead of using them you will build your own attribute files and MDF-file. We use this basic example to guide you through most of the input definition part of a flow simulation. some aspects are obliged while others are only advised or preferred. so-called attribute files. In this tutorial we provide an existing MDF-file with attribute files for a specific example called ‘fti’. The grid enclosure describes the extent of the numerical model. The name may consist of up to 252 alpha-numeric characters and may contain (sub-)directories: <path\runid>. the curvilinear grid file called <fti 02. viz. The runid of the filename is used in the names of the result files to safeguard the link between an MDF-file and the result files. We suggest to use a combination of one alpha.enc>.grd> and the grid enclosure file <fti 02. points outside the grid enclosure will not take part in the simulation (are outside the open boundaries or are always on-land). so all input fields when displayed are empty or only have default values. are stored in separate. You will start with a new (empty) input file. In the MDF-file only a reference is made to these files instead of including all data in the MDF-file itself. 116 Deltares . 5. While using the FLOW-GUI you will create new attribute files which contain data like locations of monitoring points or time-series needed for the computation. while the inner side the area enclosed by land. The big advantage of using attribute files is that the data can be used in many scenarios but is stored only once on the system disks. the screens displayed in this tutorial display the situation after you have typed in the required data. but merely to show a number of features of the FLOW-GUI and to make you familiar with it. MDF-file The name of an MDF-file must have the following structure: <name.Delft3D-FLOW.mdf> This file could indicate the flow-input file of the first calibration run of a project named Panama in the directory <c:\projects>. Remarks: The choices you are going to make are not based on physical relevance or other practical considerations. User Manual discharges etc.

2.. The names of attribute files are basically free.def> and <trim-runid.3. This will reset the internal data structure.3. such as monitoring points or time dependent input data such as wind.def> and <trid-runid.dat>. Figure 3.def> and <com-runid. Drogues file: <trid-runid. i. Deltares 117 . To open an existing input file: Select File → Open. If you have selected File → Open you can return to the initial situation right after starting the FLOW-GUI by selecting File → New. go to the working directory <. see Chapter 3 for details: Click the Delft3D-MENU icon on the desk-top (Windows) or execute the delft3d-menu command on the command line (Linux).\tutorial\flow\friesian tidal inlet>.1 FLOW Graphical User Interface Introduction To start the FLOW Graphical User Interface (GUI) you must execute the following commands.dat>. 5. Next the main window of the FLOW-GUI is displayed. see Figure 5. The result files are stored in the working or project directory. you are going to define all input data from scratch and save your data into an input file called <tut fti.dat>.1. Map file: <trim-runid.e. Instead. such as the water quality module. By selecting an existing MDF-file all attribute files and their related data are loaded into memory and displayed in the input fields.def> and <trih-runid. Communication file: <com-runid. but their extension is mandatory as indicated below. Attribute files Attribute files contain certain input quantities. The history file contains results of all computed quantities in a number of user-defined grid points at a user-defined time interval.3 5.Tutorial Result files The results of a Delft3D-FLOW computation are stored in four types of files: History file: <trih-runid. The drogues file contains the positions of all drogues released as a function of time. Change. The communication file contains results that are required by other modules.mdf>. see Figure 3.dat>. Select the menu item Flow input. The map-file contains results of all quantities in all grid points at a user-specified time interval. Select the menu item FLOW.

rgh> <name.dis> <name.Delft3D-FLOW.enc> <name.bch> <name. Quantity Astronomic correction factors Bathymetry or depth Bottom roughness Constituents boundary conditions Cross-sections Curvilinear grid Discharge locations Discharges rates Dredge and dump characteristics Drogues or floating particles Dry points Flow boundary conditions (astronomic) Flow boundary conditions (harmonic) Flow boundary conditions (QH-relation) Flow boundary conditions (time-series) Fourier analysis input file Grid enclosure Horizontal eddy viscosity and diffusivity Initial conditions Morphology characteristics Observation points Open boundaries Sediment characteristics Temperature model parameters Thin dams Wind Filename and mandatory extension <name.bct> <name.thd> <name.mor> <name.obs> <name.bcq> <name.src> <name.tem> <name.crs> <name.sed> <name.edy> <name. User Manual Table 5.bnd> <name.ini> <name.wnd> 118 Deltares .bca> <name.cor> <name.bcc> <name.1: Overview of attribute files.dep> <name.fou> <name.dry> <name.dad> <name.par> <name.grd> <name.

1: Starting window of the FLOW Graphical User Interface Deltares 119 .Tutorial Figure 5.

this tutorial may be somewhat tedious to work with.4 Description In the Data Group Description you can identify the scenario you are going to define by giving a comprehensive description of the project. you should save the just entered data. 120 Deltares .2. The description is only used for identification and has no influence on the simulation itself. Next select Exit to close the FLOW-GUI or continue with your data definition. they will be discussed later on. enabling you to save the unsaved data in an existing or in a new file. Figure 5. Neglect the other options of the File drop down menu. so that later on you can pick up the thread where you left it. 5. In order to save you from re-entering again all the data another time up to the moment where you had stopped.Delft3D-FLOW. Rather then going on until the end you may want to stop somewhere during the exercise.2: Data Group Description sub-window You are now ready to start defining your own scenario.3. To save your data: Select File → Save All.2 Saving the input data Initially. Type the description as displayed in. User Manual Figure 5. For each Data Group with changed data for which you have not yet defined an attribute filename and for the MDF-data (if you have not yet defined an MDF-filename) a file selection window will be opened. 5. even if the input data is incomplete. the application domain and the specific selections to be made in this scenario. but before doing so you will learn how to save your input data.

3: Sub-data groups of Domain Figure 5.3. Figure 5.4. Click Open grid.. 5.1 Grid parameters In the sub-data group Grid parameters you can open the grid to be used. The resulting window is given in Figure 5.4: Sub-data group Grid parameters 5.5. its location and the number and thickness of the vertical layers. to define the bottom topography. Select the folder <. to define dams along grid lines that block the flow but have no influence on the storage. A file selection window is displayed.5 Domain The Data Group Domain supports four sub-data groups: Grid parameters. one for the grid and one for the grid enclosure. to define the numerical grid.3. Bathymetry. Dry points.\tutorial\flow\friesian tidal inlet> if not already selected at the start Deltares 121 . There are two Open buttons. click Grid parameters. to define grid points that are excluded from being flooded during a simulation. See Figure 5. Click the Data Group Domain to show these four sub-data groups.Tutorial Figure 5. Thin dams.

A warning will be displayed when you try to define a grid related quantity before you have defined your grid. see Figure 5. Orientation is the angle (clockwise is positive) between true North and the y -direction. filenames. Set the Latitude to: “55. Maintain the currently inserted value 1 (for a depth averaged computation). Select the file <fti 02.6.3. The dimensions of the grid Mmax (number of grid points in ξ -direction) and Nmax (the number of grid points in η -direction) have been set (in this case to 15 and 22) in the grid generation program and cannot be changed. This information is contained in the grid file. It is therefore obvious first to select or define the grid you are going to use. the type of co-ordinate system (Cartesian or Spherical) is displayed together with the grid dimensions. However. its default value is zero and can be left unchanged. Repeat the same actions for the grid enclosure file <fti 02.Delft3D-FLOW.5: Sub-data group Grid parameters. The file selection window is closed and the filename <fti 02. to keep it (at least partly) 122 Deltares . Click the item View in the menu bar and select the item Visualisation Area in the drop down list. Move the Visualisation Area window to the right of the screen.1. For the current purpose the grid is ready for use. Furthermore.grd> appears to the right of the Open grid button. you should visualise and inspect the area to be modelled and the numerical grid used. see Section 5.5. you can still select the number of vertical layers.00” [degrees N]. type of co-ordinate system and grid dimensions of this tutorial.enc>. Many of the data items to be specified in the coming windows are related to and defined at a numerical grid. Before you continue with the other sub-data groups.grd> and open it. This will result in the window as shown in Figure 5. Remarks: You cannot modify the Grid Dimensions after loading a grid file. User Manual Figure 5.

For the tutorial we leave the Visualisation Area window. i. but as no quantities are yet defined only the grid is displayed. not all quantities are defined at the same location in the numerical grid. Before continuing with the sub-data group Bathymetry you should make yourself acquainted with the concept of the staggered grid applied in Delft3D.Tutorial Figure 5.6 the numerical grid is shown with the active grid cells enclosed by the grid enclosure. When defining grid related quantities it is important you have familiarised yourself with the concept of staggered grids. Select the menu items in the menu bar of the Visualisation Area window to inspect the various options. n) Deltares 123 . In the Visualisation Area window all grid related quantities can be viewed. In Delft3D-FLOW we apply a staggered grid.e. The staggered grid applied in Delft3D is given in Figure 5.7.7: Staggered grid used in Delft3D-FLOW uncovered. In Figure 5. this window is described in more details in Chapter 4. with the following legend: full lines the numerical grid grey area items with the same grid indices (m. defined.6: Visualisation Area window Figure 5. moved and deleted.

The location of other grid related quantities are given when appropriate. temperature horizontal velocity component in x.and m-direction) horizontal velocity component in y .or water level (zeta-)points depending on the type of boundary condition such as velocity or water level.or v -points.or η -direction (also called v . open boundaries through u-. Select the file <fti ch02. Remark: The bathymetry is defined relative to the reference level and is counted positive downward to the bottom. The sub-window should look like Figure 5. Activate the File option.8: Sub-data group Bathymetry + − | • water level.and n-direction) depth below mean (still) water level (reference level). User Manual Figure 5. salinity. The tool Delft3D-QUICKIN provides an option to compose a bathymetry from measured (sampled) data and to reverse the sign of grid related data in case the data is provided with the opposite sign. concentration of constituents. Closed boundaries are defined through u. Another option would have been to Open a (free or unformatted) bathymetry file. 124 Deltares .Delft3D-FLOW.or ξ -direction (also called u.2 Bathymetry The next step in preparing the MDF-file is to set up the bathymetry of the area for which you just have loaded the grid: Click the sub-data group Bathymetry. v . 5. Depth is defined positive downward.8.5.dep> after pressing the Open button.

9: Location of a dry point at grid indices (m. Figure 5.10 in several other sub-data groups. To add a dry point: Deltares 125 . Dry points are located at the water level point.5. n) indices needs to be specified.10: Sub-data group Dry Points 5. For other quantities you can specify a line of quantities and then the indices of the start and end point must be specified. Next the Dry points sub-window is displayed. Click the sub-data group Dry points. so only one pair of (m. The name of a dry point is composed of its grid indices.10. Figure 5. For some quantities you can specify only one item at a time. A couple of input fields to specify the location (and for some quantities a name field).9.3 Dry points Dry points are grid cells that are permanently dry during a computation. You can select an item by clicking on its name in the list box. so it is worthwhile to discuss this window in some more details. You will see the layout of Figure 5.Tutorial Figure 5. but for other quantities (such as discharges) you can specify a name. irrespective of the local water depth. n) Figure 5. Buttons Add and Delete for adding or deleting items from the list box. In this window you can see the following items: A list box with the names of the quantities defined so far. Buttons Open and Save for opening the items from an existing attribute file or saving the items to an attribute file.

. Minimise this window after inspection to keep it near by.14). i. this implies that dry points have been defined at these locations and are selected. you have defined a line of dry points along the grid line N=14 and extending over several M-grid points.14) and click Delete. Move the mouse pointer to the grid cell with indices M=13 and N=2. In short this process will be called selecting. note the name is composed of the grid indices of start and end point. before you move on to the next item you will delete the six defined dry points and single dry point. To add a dry point by point-and-click in the Visualisation Area window: Select Edit Mode → Add and select Edit → Dry Point in the menu bar. A window is displayed with a legend of the symbols that can be displayed in the Visualisation Area window. hence: Select the dry points (2. Click in the list box to confirm the input. Please note that while moving the cursor over the visualisation area the position of the cursor is displayed just below the menu bar in both world and grid indices.14). . M2 = “7”. As a result the earlier defined six dry points in the grid have turned yellow. In the list box a dry point with default name –Unnamed– appears. and N2 = “14”.14). N1 = “14”. As dry points can be defined as a line of dry points you must specify the indices of the start and end point of the line. Now. . In the list box the name of the line of dry points is changed to (2.Delft3D-FLOW. M=2. M=7. the dry points are removed from the list (and from the Visualisation Area). To see the just defined dry points: Select in the GUI menu View → Visualisation Area. User Manual Click Add. Click Save next to the list box.(7.(7. .. In the Visualisation Area window you can see the indicated fields light up in red. the newly defined dry point is displayed in red and in the Dry points list box the indices of the newly defined dry point is given. Remain in the Visualisation Area window and select View → Legend. 126 Deltares . or set the indices the same for a single dry point. Select this point. Now. Remark: Selecting a menu item or a grid related quantity actually consists of two actions: the first action is to move the cursor to the required item or position and next confirm the selection with clicking the left mouse. Select and enter one after the other the following values: M1 = “2”. You can also define dry points by means of point-and-click in the Visualisation Area window..e.

10. When selecting File → New.g.or v -direction. You can also define a single thin dam or a line of thin dams by point-and-click in the Visualisation Area window: Deltares 127 . The selected direction of the thin dam is indicated in the check box of u.11: Sets of thin dams blocking v -velocities (left) and blocking u-velocities (right) Specify the name <tut fti 1. Define a line of thin points in u-direction by entering the indices M1 = “14”.11 for the location of u. 5.5. there appears a sub-window for Thin dams.4 Thin dams Next you are going to define a Thin dam. Thin dams are infinitely thin objects defined along the grid lines. You have just defined two thin dams with a length of one grid cell each. see Figure 5. Click the sub-data group Thin dams in the Domain window.12. breakwaters or dams).dry>.dry> appears. Below Save the filename <tut fti 1. Please confirm. or when exiting the FLOW-GUI you will be notified that not all data has not been saved and you can either continue and discard this data or save the data to an existing or new file. In the list box a name composed of the indices of start and end point is displayed. M2 = “14” and N2 = “3”. The purpose of a thin dam is to represent small obstacles (e.dry> in the working directory of the opened file selection window and save the file. See Figure 5. Save the thin dams in the file <tut fti 1. which prohibit flow exchange between two computational cells at the two sides of the dam without reducing the total wet surface and the volume of the model.and v-thin dams. The Thin dams sub-window is displayed in Figure 5.Tutorial Figure 5. a preferred way of working is to save data items when defined. Similar to what happened in the previous section. Remark: You don’t need to save the data in each sub-data group. If the file already exists the program will ask you whether you wish to overwrite the old file <tut fti 1. The FLOW-GUI keeps track of all changed data fields. The selected grid section turns red in the visualisation area to indicate thin dams have been defined and are currently selected. However.thd>. N1 = “4”.

Delft3D-FLOW. The selected grid line section turns yellow. User Manual Figure 5. understand why a line of thin dams along one grid line changes to a set of single thin dams along the other grid direction when changing from uto v -direction and vice versa. the easiest way to do this is to neglect the additional thin dams and go on to the next data group. change the direction in which the thin dam is defined by clicking the u. Remark: You can define lines or sets of thin dams along the grid lines but also at angles of a multiple of 45 degrees with the grid. So: Open the thin dam file <tut fti 1.13 for u. so you can move on to the next data group. or to overwrite the currently defined thin dams by the saved thin dams in <tut fti 1. Familiarise yourself with this. Move the mouse pointer to the required length and release the mouse button.12: Sub-data group Thin dams Select Edit Mode → Add and Edit → Thin Dams in the Visualisation Area menu bar. as you already have saved the firstly defined thin dams.and v -types of thin dams with the same grid indices.or v -direction check box and see how the change is visualised in the Visualisation Area window. 128 Deltares . See Figure 5.thd>. Before you go on to the next data group you should undo the random changes. Add a couple of more thin dams. thus indicating a thin dam has been defined. this would stimulate a wrong working habit and so you are advised to delete all the additional thin dams before you leave the data group. Click a grid location.thd>. However. but keep the mouse button pressed. You have defined all the required items in the Domain window.

Set the Simulation start time to “05 08 1990 00 00 00”.14: Window Time frame 5.6 Time frame In the Data Group Time frame you can specify the time frame of your computation composed of the reference date. Click the Reference date input field and type the date “05 08 1990”. i. Set the time step to “5” minutes. the simulation stop time and the time step used in the numerical simulation. Deltares 129 .e. Figure 5. in the FLOW-GUI all time related input is specified in a real time format. Click the Data Group Time frame. The Reference date prescribes the (arbitrary) t = 0 point for all time-series as used in the simulations. the simulation start time.14: Remark: When this window is first activated it always shows the current calendar date of your computer. However.13: Equivalence of v -type thin dams (left) and u-type thin dams (right) with the same grid indices. [dd mm yyyy] for the date and [hh mm ss] for the time. Set the Simulation stop time to “06 08 1990 01 00 00”. n) Figure 5. (m − 1 to m + 1.Tutorial Figure 5. In the canvas area the following sub-window is displayed.

In fact all single data items are stored in the MDF-file. You can select constituents: Salinity Temperature Pollutants and tracers Sediments or physical processes: Wind Wave Secondary flow (2D mode only and no Z -model) Online Delft3D-WAVE Tidal Forces (spherical co-ordinates only) or man-made processes: Dredging and dumping (only if sediments are selected) To select the required processes: Click the Data Group Processes.15. see Figure 5.7 Processes In the Data Group Processes you specify which processes or quantities that might influence the hydrodynamic simulation are taken into account. depth and time step is about 11. The Processes sub-window is displayed with the list of possible constituents and processes. and the physical process Wind. The buttons Edit are only displayed after you have selected the constituents Pollutants and tracers or Sediments.Delft3D-FLOW. Neglect the other input field. or about two semi-diurnal tidal cycles.16. 5. in this tutorial we select the constituents: Salinity and Pollutants and tracers. the Processes: Pollutants and tracers window is displayed. Remarks: All data items of the Data Group Time frame are stored in the MDF-file and not in an attribute file. User Manual The simulation time thus covers exactly 25 hours. The maximum Courant number given the grid. 130 Deltares . see Figure 5. Immediately after ticking off the Pollutants and tracers checkbox or when the Edit is pressed (if enabled). Activate the required processes by ticking off their check box.

It is currently implemented as a special feature. You can specify a simple decay rate for each of the substances. Deltares 131 . See Chapter 11 for the use of cohesive and non-cohesive sediments. see Appendix B.16: Processes: Pollutants and tracers sub-window Now type into the Name input field the words “Conservative Spill” and click Add.Tutorial Figure 5. Remarks: The first time you select Processes . You have just defined one conservative constituent that is transported by the flow. Figure 5.Pollutants and tracers.15: Processes window Figure 5. For changing the constituent names later on you must click the aforementioned Edit button.2 for details. For interacting constituents you have to apply the water quality module D-Water Quality.15 is automatically displayed. Close the sub-window.

Open boundaries are defined in sections of one or more grid cells. 132 Deltares . end-point B of section one and start point A of section two are located in consecutive grid cells.17. The sub-window Initial conditions is displayed. Figure 5. salinity and one constituent. The initial conditions can be very simple. so if you define two boundary sections along an open boundary. their location.0” kg/m3.9” m.Delft3D-FLOW. In the dropdown list Uniform values has already been set. Click Add to the right of the list box. so in this case for the processes water level (by default). If not: Select Uniform values. User Manual Figure 5. Firstly. Enter the values: Water Level: “1. type and all input data related to forcing the simulation. Click the Data Group Initial conditions. called begin and end.9 Boundaries In the Data Group Boundaries you can define the open boundaries. Salinity: “30. or.8 Initial conditions In the Data Group Initial conditions you can specify the initial values used when starting the computation.0” ppt and Conservative Spill: “1. Boundary conditions for intermediate points along a boundary section are determined by linear interpolation. Boundary sections may not be overlapping. 5. Input fields are shown for the processes selected in the Data Group Processes.17: Initial conditions sub-window 5. Shown are some option boxes and input fields. you are going to define the location and extent of the open boundaries: Click the Data Group Boundaries. Boundary conditions are prescribed at both ends of a section. A and B. such as a uniform value in the whole area or taken from a previous computation. The initial conditions are stored in the MDF-file when saving the input.

Remark: Values are read from the input field upon losing the focus. when you select another input field or select a boundary section in the list box.e.Tutorial Figure 5. i. Now define the begin and end point of the open boundary: Type the following values of the grid indices in the appropriate input fields: M1 = “2”. M2 = “14” and N2 = “22”. N1 = “22”. Retain the value 0. Enter instead: “Sea Boundary”. In the Visualisation Area window the open boundary just defined is displayed as a thick red line.18: Boundaries sub-window In the input field Section name appears the word –Unnamed–.0 in the Reflection parameter alpha input field. To save these settings: Deltares 133 .

Figure 5. Concentrate for the moment on the Boundary Definitions attribute file and discard the others. For open boundaries there is one additional condition. User Manual Figure 5. Enter the filename <tut fti 1. select the begin point. i.Delft3D-FLOW.19. the easiest way is to import the previously saved open boundary locations. N) indices displayed in the Visualisation Area 134 Deltares . Click Save (Boundary definitions) in the pop-up sub-window. The (M. i. the open boundary must be defined at the edge of the grid as shown in the Visualisation Area. and release the mouse button.e.e. Make sure you undo all changes. drag the mouse pointer to the end point of the boundary section along the grid line. In the visualisation area all open boundaries are visualised at grid lines. i. keep the mouse button pressed.19: Open/Save Boundaries sub-window Click Open/Save in the Boundaries sub-window. Remarks: After pressing Open/Save a sub-window is displayed with references to all type of attribute files related to open boundaries. You can define an open boundary in the Visualisation Area window in the same way as you define a line of thin dams or a line of dry points.e. To define an open boundary by point-and-click: select the Edit Mode → Add and Edit → Open Boundaries in the menu bar of the Visualisation Area window and continue as prescribed. at the edge of computational grid cells.bnd> in the working directory and close the Open/Save Boundaries menu window.

Tutorial Figure 5.5 = 28. In this tutorial we are going to specify harmonic boundary conditions for a depth averaged computation (only one layer in the vertical. Hence. The harmonic components must be specified for both ends of an open boundary section.e. N1) and the End point has the indices (M2. see Figure 5. right or upper) from the grid line. In this tutorial you will use the default type of open boundary i. Click Edit flow conditions.20: Harmonic boundary conditions refers to the computational grid cell.20. lower.1 m and phase -34. Remark: The frequency is specified in degrees per hour. A mean sea level is defined with a frequency of 0 degrees/hr. For a detailed discussion on the computational grid and the location of open and closed boundaries see Appendix E.18. After filling the input fields the table should look like Figure 5.1 m and phase 0 degrees at the Begin point and amplitude 1. Deltares 135 . Set the forcing type to Harmonic.8 for the semidiurnal tide. 360/12. The Begin point has the indices (M1. To define an open boundary by point and click you should select and drag inside the first (or last) grid cell so as to select the first (or last) water level or velocity point.26 degrees in the End point. N2). This rule also applies to open boundaries in side the computational domain.Grid parameters). see Data Group Domain . The begin and end point are referred to as ‘Begin’ or ‘point A’ and ‘End’ or ‘point B’. so several options are either invisible or insensitive. You are going to prescribe a semi-diurnal tide with amplitude of 1. Water level and the Harmonic type of forcing (the default is Astronomic ). This grid cell number differs by one in M or N depending on the location of the boundary (left.

0 Next select from the dropdown list the constituent Conservative Spill.Delft3D-FLOW. You are back in the Boundaries window.18. Save the harmonic flow conditions into file <tut fti 1. Close the Open/Save Boundaries window. Click Edit transport conditions. Next.0 30. User Manual Figure 5.0 30. see Figure 5.bch>. Next you will provide time-series for the concentrations of the constituent and salinity at the open boundary. fill out the table form with the following time breakpoints and parameter values at both ends (Begin and End): Date and time 05 08 1990 00 00 00 06 08 1990 01 00 00 Begin 30. To define the boundary conditions for the process salinity: Select from the Constituents dropdown list: Salinity. Select Open/Save in the Boundaries window. and fill out the table with the following time breakpoints and parameter values at both ends.0 End 30. Set the Thatcher-Harleman time lag to “100” minutes. 136 Deltares .Flow Conditions sub-window.21: Boundaries: Transport Conditions window Now save the data to the attribute file: Close the Boundaries .

0 1. representing a tracer that is present in the inflowing sea water. for instance zero or a small value. Ignore all warnings and error messages that might be displayed (remember you might not have specified all kind of data required for a real 3D simulation).0 End 1.Save and save the transport boundary conditions in the working directory in the file <tut fti 1.Edit transport conditions .Tutorial Figure 5.0 For this constituent you set all concentrations to 1.Transport conditions . Close the Open/Save Boundaries window. Deltares 137 .22: Physical parameters sub-data groups Date and time 05 08 1990 00 00 00 06 08 1990 01 00 00 Begin 1. you must use the Constituents drop down list to select the other selected constituents.0 1. To inspect this you can go back to the Data Group Domain . they do not need to be the same. Remarks: The time breakpoints for Salinity and the constituent Conservative Spill and for any other process selected can be different. Close this window to return to the Boundaries window.Grid parameters and increase the number of layers to two (with a distribution of 50% for each layer) and return to the current window (Data Group Boundaries . For a tracer to be washed out of the system you should enter the sea-background concentration. After inspection make sure to return the number of layers to one. Select in the Boundaries window Open/Save . You have now finished the boundary conditions for all processes and constituents and you are ready to save the data.Constituents .Salinity ). Remark: In a 3D model you can specify a profile.bcc>.0. Recall you have specified the initial condition (Data Group Initial conditions) to 1.21 only the input data for Salinity is visible.0 for this constituent. In Figure 5.

22. • Type into the input fields of the third line “0.00” [m/s]. Figure 5. Roughness. The third two coefficients specify the wind drag used from the specified wind speed and higher.00” [m/s]. As temperature was not selected in the Data Group Processes the sub-data group Heat flux model is invisible. grouped in several subdata groups. 138 Deltares . 5. Temperature: “10.0” ◦ C.0” kg/m3 . Click the Data Group Physical parameters. Next the sub-data groups Constants.00” [m/s]. the water and air density and constants that determine the wind drag force. There are three Wind drag coefficients: • Type into the input fields of the first line “0. As a result you can specify a constant wind drag.10 Physical parameters In this Data Group you can define a number of physical parameters.0025” [-] and “0.23.0025” [-] and “100. The three wind drag coefficients determine breakpoints in the piece-wise linear function of wind drag and wind speed: The first two coefficients determine the wind drag value that is used from zero wind speed up to the wind speed specified by the second constant. • Type into the input fields of the second line “0. a linearly increasing wind drag or a piecewise linear function of wind speed. To specify the constants: Click the tab Constants.1 Constants The sub-data group Constants concerns items such as the acceleration of gravity.0025” [-] and “100. Viscosity and Wind are displayed as tabs.81” m/s2 (default value). User Manual 5.10. In the next sections the other sub-data groups will be treated. Set the values by entering into their respective input fields: • • • • Gravity : “9. Water density : “1024. Air density : “1.00” kg/m3 . see Figure 5.Delft3D-FLOW.

Click the tab Roughness.05” for both U. The option Uniform has already been activated by default. you only have to specify the constant value that is to be used. ´ : Three different friction formulations are given.10.Constants sub-window Figure 5. i. Deltares 139 . Manning.24: Roughness sub-window 5.and V-direction.e.23: Physical parameters .24 is displayed. The bottom roughness can be specified as a constant value over the whole area or as a space varying value (read from an attribute file). next Figure 5. Enter the value “0.Tutorial Figure 5.2 Roughness In the sub-data group Roughness you can specify the bed roughness and for specific situations the roughness of the side walls. White-Colebrook or Chezy Select White-Colebrook.

As this is a depth averaged model.10. and the horizontal eddy diffusivity to “10. Click the tab Viscosity. next Figure 5.and V-direction.Delft3D-FLOW. In this tutorial we will not use this model. Remark: You specify one value for both the Horizontal Eddy Viscosity and the Horizontal Eddy Diffusivity in U.0” m2 /s. User Manual Figure 5.25: Viscosity sub-window Ignore the Slip condition dropdown box (and the roughness value related to it). 140 Deltares . The model for 2D turbulence computes viscosities and diffusivities which will be added to the background values. only the background Horizontal eddy viscosity and the background Horizontal eddy diffusivity are displayed and require a value: Set the horizontal eddy viscosity to “2. Space varying values can be read from a file. these files must be prepared offline using some kind of editor or a program.3 Viscosity In the sub-data group Viscosity you can specify the eddy viscosity and the eddy diffusivity in the horizontal and vertical directions.0” m2 /s. 5.25 is displayed. respectively. such as the tool Delft3D-QUICKIN provided with Delft3D.

and keeping that value for the rest of the computation. You are going to define a linearly increasing wind speed from 0 m/s to 10 m/s during the first 12:30 hr of the simulation.27. To add or insert (before the current row) a row: Select Table → Copy row above. “10”.4 Wind Next you are going to specify the wind speed and direction: Click the Wind tab. Next Figure 5. select Space varying wind and pressure. “90” “06 08 1990 01 00 00”. Now for the sake of curiosity. The check box Uniform is set. A new row has been inserted and you can modify the data. the direction changing from 0 degrees to 90 degrees in the same time interval. Now save the wind data: Deltares 141 . After editing the table should look like Figure 5. indicating you are specifying a uniform.26: Wind sub-window 5. but possibly time dependent wind field. “0” “05 08 1990 12 30 00”.26 is displayed. “90” For operations on a table use the menu item Table in the menu bar. “10”. In the time table. “0”. fill in the following data: “05 08 1990 00 00 00”.Tutorial Figure 5. But then go back by activating again the Uniform check box. and see what happens next.10.

Delft3D-FLOW.27: Filled time table in the Wind sub-window Figure 5. 142 Deltares . The options for Drying and flooding check at determine where the check should be performed: at grid cell centres (water level points) and grid cell faces (velocity points). Although drying and flooding will not take place in this model. User Manual Figure 5.28 is displayed. for the sake of getting a good overview of what is going on. 5.wnd>. you will include a drying and flooding procedure.11 Numerical parameters In the Data Group Numerical parameters you can specify parameters related to drying and flooding and some other advanced options for numerical approximations. The first selection to be made concerns drying and flooding.28: Numerical parameters sub-window Save the wind data in the working directory in the file <tut fti 1. Click the Data Group Numerical parameters. next Figure 5.

5. The Depth at grid cell centres is set to MAX by default. its release rate must be defined. Set the Threshold depth to “0. The Discharges tab. The Marginal depth is used to change the approximation of the advective terms in the momentum equations to allow for local super-critical flows.29.05” m. For instance: Max: the maximum depth of the four depth values at the cell corners surrounding the cell centre (water level point) is used. For the Depth at grid cell centres (water level points) several options are available. The Smoothing time determines the time interval in which the open boundary conditions are gradually applied. Delft3D-QUICKIN generates the data at the cell corners by default. Set the Smoothing time to “60” minutes. but in addition you must specify the rate and concentrations of the discharged substances. Define the discharge location by point-and-click in the Visualisation Area window at grid cell M = “14” and N = “2”. or at the Grid cell centres.Tutorial or at only grid cell faces (velocity points). Figure 5. Skip for this tutorial the Marginal depth input field.12 Operations In the Data Group Operations you can specify the location and all related properties of discharges. The Depth specified at selection reflects if the depth values are generated at the Grid cell corners.2) in the Name input field with “Outfall”. See Chapter 10 for details of this drying and flooding procedure. Ignore the other options or input fields. Replace the text (14. Deltares 143 . Click the Data Group Operations. A grid cell is flooded if the water depth is larger than the specified Threshold depth and dried if the water depth is smaller than half this value. This smoothing of the boundary conditions prevents the introduction of short wave disturbances into the model. starting at the specified initial condition to the specified open boundary conditions. similar for the Depth at grid cell faces (velocity points). Mean: the mean depth of the four depth values at the cell corners surrounding the cell centre (water level point) is used. For this tutorial we have introduced one constituent (Conservative Spill). has a similar layout as the Dry points or Thin dams subwindow.

The discharge is located at the centre of the computational cell (m. Specify the following data in this table: 144 Deltares . The table to define the discharge rate and concentrations is displayed in Figure 5.30.29: Data Group Operations.Delft3D-FLOW.src> in the working directory. Close the Open/Save Discharges window. n). To save the discharge location: Click Open/Save to open the Open/Save Discharges window. Click Save in group Discharge definitions to save the discharge location in the file <tut fti 1. Discharges sub-window In the Visualisation Area window discharges are represented by a lozenge at a water level point. User Manual Figure 5.31. Next you must define the time dependent discharge data: Click Edit data. see Figure 5.

Tutorial Figure 5.31: Discharge Data sub-window Deltares 145 .30: Representation of a discharge in the Visualisation Area window Figure 5.

next Figure 5.17) with “Obs1”. Click the Data Group Monitoring to display the sub-data groups. Scroll the table horizontally.Spill kg/m3 1.0 Cons.Delft3D-FLOW.0 3.13. i.0 0.e. N = 17) in the Visualisation Area window. if needed. 5.e. Click the sub-data group Observations. To add an observation point: Select the location (M = 4. User Manual Date and time Flow m3 /s 3. Figure 5. the time breakpoints are the same for all time dependent quantities related to the discharge. Close the Open/Save Discharges window.1 Observation points Observation points are grid locations where all quantities computed during the simulations are stored at a user-defined time interval (Data Group Output → Storage).32: Monitoring sub-window 5.32. Figure 5.0 Salinity ppt 0.13 Monitoring The Data Group Monitoring consists of three sub-data groups.0 05 08 1990 00 00 00 06 08 1990 01 00 00 Remarks: The data specified represents a fresh water discharge (salinity = 0 ppt) of 3 m3 /s. Click Open/Save and click on Save in datagoup Discharge data to save the discharge data in the attribute file <tut fti 1.dis> in the working directory. Obviously. i. Replace the text (4. Observation points are located at the computational grid cell centres. Drogues and Cross-sections. Repeat this for the observation points Obs2 to Obs5 as specified in the table below: 146 Deltares . Observations.0 1. to display the Conservative Spill data input fields.33 is displayed. at the water level points. Close the window if all input is specified.

Deltares 147 .obs> in the working directory.34. The observation point is defined at (m. n).34: Representation of an observation point in the Visualisation area window Name Obs1 Obs2 Obs3 Obs4 Obs5 M index 4 9 12 9 9 N index 17 17 17 11 7 Save the Observation Points just defined in the file <tut fti 1.33: Observation points sub-window Figure 5.Tutorial Figure 5. Figure 5. In the Visualisation Area window observation points are represented by a cross through the computational grid cell centre.

35 is displayed. Figure 5. The drogue is located at (m. They are released and recovered at a userspecified time.50 8.13. However. in the Visualisation Area window you can only select grid centre location.50 16. Drogues can be located anywhere in a grid cell. Figure 5.50 8.35: Drogues sub-window 5.50 16. Click the sub-data group Drogues.50 8.50 11.36.2 Drogues Drogues are floats that move with the flow. n). so if other decimal values are required you must enter these indices manually in the FLOW-GUI. In the Visualisation Area window drogues are represented by a ‘+’ through the computational grid centre. 148 Deltares . For all drogues change them to: Release Time : “05 08 1990 12 30 00” Recovery Time : “06 08 1990 01 00 00” Save the drogue data in the file <tut fti 1. i.50 N index 16. User Manual Figure 5.50 The Release Time and Recovery Time are by default the same as the start and stop times of the simulation.par> in the working directory. Their location is stored at the same time interval as the computational time step (see Data Group Time frame).50 10. To add drogues: Click Add and enter the following drogues: Name Dr1 Dr2 Dr3 Dr4 Dr5 M index 3.50 6. a drogue can be released at decimal indices.e.Delft3D-FLOW.

M2 = “14” and N2 = “11”. next Figure 5. For defining cross-sections you can apply the same procedures as for Thin dams or Dry points in the Data Group Domain. N1 = “11”.38. either as U. n) to (m+1. To add a cross-section: Click Add.38 the V-cross-section runs from (m−1.or as V-cross-section.37 is displayed. In Figure 5.13. Type in the Name field the word “CS1” to replace –Unnamed–. In the Visualisation Area window a cross-section is represented by a thick line along one of the grid lines.Tutorial Figure 5.36: Representation of a drogue in the Visualisation Area window Figure 5. To enter the sub-data group Cross-sections: Click Cross-sections.3 Cross-sections Cross-sections are sections along one of the grid directions through which the various kinds of transport fluxes are determined and stored as a function of time. n).37: Cross-sections sub-window 5. Deltares 149 . Enter the indices M1 = “2”. Figure 5. Cross-sections are defined along the grid lines.

which are not yet supported by the FLOW-GUI. 5.14 Additional parameters In the Data Group Additional parameters you can specify keywords and their value. often the name of an attribute file.39: Output sub-window Remark: Cross-sections must have a minimum length of two grid cells.Delft3D-FLOW. Make sure to undo the additional definitions (for instance by opening the just saved attribute file) before you move on to the next data group. Save the data in file <tut fti 1. To save the just defined cross-section: Click Save. User Manual Figure 5. Remark: You are encouraged to define a cross-section along either of the grid lines by pointdrag-and-click. 150 Deltares .crs> in the working directory. but for this tutorial we ignore this Data Group. For details you are referred to Appendix B.38: Representation of a v-cross-section in the Visualisation Area window Figure 5.

for water quality simulations you typically store the results of the last day or a (couple of) cyclic tidal cycle and use this data repeatedly to make a water quality simulation over many days or weeks.Tutorial 5.39 is displayed. So typically map results are only stored at a small number of instances during the simulation. Print : to define the time interval to store results to a printable file. Typically you select a large restart interval in order not to waste disk space. Upon selecting Output → Storage the Output storage sub-window.15 Output In the Data Group Output you can specify which computational results will be stored for further analysis or other computations and which output shall be printed. but only in the specified Monitoring Points and Cross-Sections. such as the hydrodynamic results for a water quality simulation or the wave forces for a wave-current interaction. As the results must be stored in all grid points a communication file can be as large as or even larger than a Map file. and you typically store history data at a small interval to have a smooth time function when plotting the results. To enter the data group: Click the Data Group Output. Basically. The Data Group Output supports three sub-data groups: Storage: to define the time interval to store results to disk. will be displayed. Details: to define which quantities will be stored or printed. So typically you only store the results in the Communication file as far as needed for the other simulations. Communication file The communication file contains results required by other modules of Delft3D. As you save all results in all grid points a typical Map file can be many hundreds of Mbytes large. Before defining the various input fields we want to explain in short the use of the various files used to store the results.40. Now continue to define the various times and time intervals: Deltares 151 . Map file The map file contains snap shots of the computed quantities of the entire area. Next Figure 5. Figure 5. Though the printing option is hardly used it is a useful option when numerical output values are required. The amount of data is usually much smaller than for a map file. For this you specify the Restart Interval at which all information required to restart the simulation. Drogue file The drogue file contains the (x. In order not to waste computing time you can restart a simulation at a predefined time of a previously executed simulation (Data Group Initial conditions). y )-position of all drogues at each computational time step in the time interval between release and recovery time. Delft3D-FLOW uses four types of files to store results: History file The history file contains all quantities as a function of time.

it is discussed in detail in Chapter 4. User Manual Figure 5.Delft3D-FLOW. this results in viewing the results during the simulation. 152 Deltares . defined in the Data Group Time frame. Set Store communication file to: Start time: “05 08 1990 12 30 00” Stop time: “06 08 1990 01 00 00” Interval time: “15” min Set History interval to: “5” minutes. For this tutorial we ignore the item Fourier analysis. To control the quantities that are actually stored on files or printed: Click the sub-data group Details.41 is displayed. Set Restart int.: “1500” min Remark: The interval to write the various files must be an integral multiple of the computational time step. Tick off Online visualisation. The restart interval as set results in a restart file at the end of the simulation. For this tutorial we ignore the sub-data group Print. next Figure 5. By ticking the check boxes you can specify which items you want the program to store on file or to print and which not. details of the quantities covered by each group are given in Chapter 4.40: Output storage sub-window Set Store map results to: Start time: “05 08 1990 00 00 00” Stop time: “06 08 1990 01 00 00” Interval time: “150” min.

41: Output details sub-window You have now gone through all Data Groups and you have defined in most of them one or more data items.bnd> Wind <tut fti 1.dep> Deltares 153 .dry> Thin Dams <tut fti 1.wnd> Flow boundary conditions (harmonic) <tut fti 1.thd> Open Boundaries definitions <tut fti 1.crs> In addition you have used three predefined attribute file: 1 Curvilinear Grid <fti 02.obs> Drogues <tut fti 1.bch> Transport boundary conditions <tut fti 1.par> Cross sections <tut fti 1. To summarise: 1 2 3 4 5 6 7 8 9 10 11 Dry Points <tut fti 1. 5.grd> 2 Grid Enclosure <fti 02.enc> 3 Bathymetry <fti ch02. some of which are already saved in attribute files.bcc> Discharge locations <tut fti 1.16 Save MDF-file In the preceding sections you have entered many data items.Tutorial Figure 5. Before continuing you should save this data to make sure you can always return to this status of the FLOW-GUI.src> Discharge rates <tut fti 1.dis> Observation points <tut fti 1.

A warning was issued on leaving a Data Group without saving its related attribute file(s). Upon selecting File → New.17 Additional exercises You are encouraged to go back to the start of the tutorial and explore some of the options not used or discussed in this first example. In previous versions of the FLOW-GUI you were obliged to save the attribute data in their respective Data Group. 5. All unsaved Data Groups are ignored. All attribute data is saved in attribute files (if they exist) and the references to existing attribute files are stored in the new MDF-file.Delft3D-FLOW.mdf> in the working directory. File → Open or Exit you will be notified if some data is not saved and you can choose to ignore the data. If you have defined all attribute files in their respective Data Group. 154 Deltares . the MDF-file is complete. Save the MDF-data in the file <tut fti. User Manual Figure 5. move) any of the grid related quantities. A very worthwhile exercise is to define. or to save the data.42: Select scenario to be executed To save the reference to all these attribute files and other (single) input data in the MDF-file: Click File → Save MDF As in the menu bar of the FLOW-GUI. To exit the program: Click File → Exit. delete or modify the definition of grid related quantities using the Visualisation Area window only. but you can postpone it to the moment when you save the MDF-data.e. Starting this version of the FLOW-GUI you can save attribute data in their respective Data Group. Remarks: Do not use blanks in the name of the MDF-file. You will be prompted for unsaved data. With the menu item Edit Mode → Modify you can drag-and-drop (i.

warnings and information of the simulation. MDF-file). In foreground the status of the simulation and possible messages are displayed in the active window.e. To see this report: Select Reports in the Hydrodynamics (including morphology) window. To reproduce these plots you should start the general post processing program Delft3D-GPP (see the Delft3D-GPP User Manual GPP (2006)).18 Execute the scenario If you have an error free scenario you can start the simulation: Select Start in the Delft3D-MENU to start the computation in foreground. see Figure 5. 5. Figure 5.43.44 to Figure 5. If <tut fti.43: Part of the <tri-diag> report file The simulation is executed without errors or warnings. Deltares 155 . After the simulation is finished you are strongly advised to inspect at least some of the report files generated during the simulation to check if all went according to plan.mdf> is not selected already.mdf>.19 Inspect the results In Figure 5.46 some results are shown of the simulation just finished. In Figure 5.42 the selection window is displayed for the scenario to be executed (i.Tutorial 5. Select Flow to inspect the report file. Confirm your selection with OK. This concerns especially the <tri-diag> file (located in the working directory). Especially the end of this file is of importance as it summarises errors. use Select file to navigate to the working directory and select <tut fti.

In the main window of Delft3D-GPP select Plots and select from the list of possible plots the one you would like to inspect. In a <∗. In the file selection menu select and open the file <tut fti. In the main window of Delft3D-GPP select Session → Open. By calling this scenario file you can inspect the same plots after repeating the simulation with (other input data) of the FLOW scenario tut fti. in this case the result files of the tut ftiscenario. 156 Deltares .Delft3D-FLOW. To close Delft3D-GPP and return to Delft3D-MENU: Select in the main window of Delft3D-GPP Session → Exit. User Manual Select GPP in the Hydrodynamics (including morphology) window or Utilities . Select Exit.GPP in the main Delft3D-MENU. For details of using Delft3D-GPP you are referred to the User Manual of Delft3D-GPP.ssn>.ssn> file the references are stored to data sets. To return to the main window of Delft3D-GPP while viewing a plot: Select Plot → Close. To close Delft3D-MENU: Select Return. You can select another plot as described above. and the definition of earlier defined plots and their layout.

44: Computed time-series of the water level.Tutorial Figure 5. current and salinity in observation point Obs4 Deltares 157 .

45: Computational grid with drogue Dr4. User Manual Figure 5. and contours of water level on 6 August 1990 01:00 hr 158 Deltares .Delft3D-FLOW.

Tutorial Figure 5.46: Vector velocities and contours of salinity on 6 August 1990 01:00 hr Deltares 159 .

User Manual 160 Deltares .Delft3D-FLOW.

To see which reports are generated: Select Reports in the Hydrodynamics (including morphology) window. see Figure 6. you can either execute the computation in foreground or in batch. whereas in background all messages are written to a file and you can continue working in the current window.6 Execute a scenario 6.1. see Figure 6.2. upon which Figure 6. Figure 6. To start Delft3D-FLOW: Select Start.1: MENU-window for Hydrodynamics A new window is displayed in which you can select the input file to be used. Here we discuss to execute the Delft3D-FLOW simulation in foreground.3 After the simulation is finished you are strongly advised to inspect at least some of the report files generated during the simulation to check if all went according to plan. Apply Select file to select the required MDF-file.4 is displayed. In foreground the status of the computation and possible messages are displayed in a separate window. In foreground the status of the simulation and possible messages are displayed in the active window. Figure 6.1 Running a scenario After you have defined the input for the Delft3D-FLOW hydrodynamic simulation. Deltares 161 . After pressing OK the simulation will start and a window will appear.

Delft3D-FLOW, User Manual

Figure 6.2: Select the MDF-file to be verified

Figure 6.3: Part of the report to the output window

Select either one to inspect: The report file of the simulation: <tri-diag.∗>. The print file with printed output results: <tri-prt.∗>. The report file of the WAVE simulation (if applicable): <swn-diag.∗>. The report file of the Hydrodynamic coupling (if applicable): <couplnef.out>. A random file, upon selecting this option an editor window is opened that allows you to inspect or modify a file without having to leave the FLOW-GUI. The report file <tri-diag.∗> contains information on the total simulation time, the time spent in certain parts of the simulation and an indication of the performance (seconds processor time per grid point per time step). In the next section we shall give some guidelines to estimate the simulation time and the required disk space for the various result files.

162

Deltares

Execute a scenario

Figure 6.4: Select a report file for inspection

6.2

Run time The actual run time of a model can vary considerably depending on a variety of factors such as: The problem being solved, characterised by the number of active grid points, the number of layers in the vertical or the number of processes taken into account. The length of the simulation in time and the time step being used. The hardware configuration that is used and the work load of the processor. For this reason, only some general considerations are given to determine the run time of a hydrodynamic simulation. On a PC or a workstation without separate I/O-processors the CPU time is the sum of the processor time and the I/O time. The processor time required for a simulation is primarily determined by: The model definition, i.e. the number of active grid points and the number and type of the processes taken into account. The length of the simulated period in terms of the number of time steps executed. The I/O time is determined by: The number of times the computed data are written to history, map, print and communication files. The number of observation points, cross-sections and the number of output parameters. The simulation performance is defined as the CPU time per grid point per time step per con-

Deltares

163

Delft3D-FLOW, User Manual
Table 6.1: Simulation performance on different operating systems

Computation Without Domain decomposition With Domain decomposition

WIN 0.5E-05 0.5E-04 to 0.1E-03

LINUX 0.4E-05 0.5E-04 to 0.1E-03

stituent:

simulation performance =

CPU time [system seconds] N · Mmax · Nmax · Kmax · Lmax

where: N Mmax Nmax Kmax Lmax is the number of time steps executed is the number of grid points in x-direction is the number of grid points in y -direction is the number of layers in the vertical is the number of constituents

In Table 6.1 the simulation performance is given for a PC with a Windows and a Linux operating system respectively. The Windows and Linux PC have similar processors (AMD Athlon 64 X2 Dual Core Processor 4200+, 2.19GHz). Remarks: The simulation performance is almost linearly dependent on the processor speed. The simulation performance can be improved with about 20% by minimizing the output written to the result files. The simulation performance is written to the diagnosis file at the end of a calculation. When using Domain decomposition, each domain produces its own simulation performance index. The difference between these indexes will become very large when the domain load is unbalanced. To improve the time estimates we suggest to keep record of the simulation performance on your hardware environment including all site specific aspects. In case the simulation performance is very bad on similar hardware and the workload is not excessively high, you are advised to contact your system manager to check the general settings of your hardware such as swap and stack space. For system requirements you are referred to the Installation Manual.

6.3

Files and file sizes For estimating the required disk space the following files are important: history file map file print file communication file drogues file 164 Deltares

Execute a scenario Remark: Drogues files are generally small and are not discussed further.

6.3.1

History file The size of the history file is determined by: The number of monitoring points (observation points + cross-sections): H1. The number of quantities stored: H2. The number of additional process parameters, such as salinity, temperature, constituents and turbulence quantities, taken into account in the simulation: H3. The number of time the history data is written to the history file: H4. You can estimate the size of a history file (in bytes) from the following equation: size history file = H1 · (H2 + H3) · H4 · 4 bytes. As a first approximation you can use H2 = 15. Example: For a 2D simulation with density driven currents (salinity and temperature), a simulated period of 12 hrs 30 min, a time integration step of 5 minutes, 30 monitoring points and each time step being stored, the size of the history file will be of the order of 360 Kbytes. For the same model but now with 10 layers in the vertical the file size will increase to about 4 Mbytes. These estimates show that history files are rather small. Unless the number of monitoring points is excessively large the history files are typically much smaller than the map output files.

6.3.2

Map file The size of the map file is determined by: The size of the model, i.e. the number of active grid cells multiplied by the number of layers (Mmax · Nmax · Kmax): M1. The number of quantities stored: M2. The number of process parameters taken into account, such as salinity, temperature, constituents and turbulence quantities: M3. The number of time steps for which the map file is written: M4. Remark: For a more refined estimate you should distinguish between parameters that depend or not on the number of layers used (such as the water level). For a 3D simulation the latter quantities can be neglected, for a 2D simulation they must be accounted for. As a first estimate we double the number of quantities M2 in a 2D simulation. As a first approximation you can use M2 = 15 for a 3D simulation and M2 = 30 for a 2D simulation. You can estimate the size of a map file (in bytes) from the following equation:

Deltares

165

Delft3D-FLOW, User Manual size map file = M1 · (M2 + M3) · M4 · 4 bytes. Example: For a 3D simulation with 50 by 50 points and 5 layers, simulated with density driven currents (salinity and temperature), simulation results stored for a period of 12 hours and 30 minutes, and the file is written with an interval of 30 minutes the size of the map file will be about 31 Mbytes. For larger models the map file can easily become excessively large, as result the map file is less frequently written, for instance every 2 or 3 hours.

6.3.3

Print file The print file is a formatted ASCII file with a size that can be large, varying from hundreds of Kbytes to hundreds of Mbytes. Its size depends on: The number of grid points: P1. The number of layers in the vertical: Kmax. The print flags for the map data requested: P2. The number of times the map data is written to the file: P3. The number of monitoring stations: P4. The print flags for the history data requested: P5. The number of times the history data is written to the file: P6. You can estimate the size of a print file (in bytes) from the following equation: size print file = 2 · (P1 · Kmax · P2 · P3 + P4 · P5 · P6) · 4 bytes The factor of 2 is applied for all explaining text included in the file. Example: For a similar simulation as in the former section and all print flags checked, i.e. roughly 15 items, the print file is found to be in the order of 36 Mbytes; its size is found to be determined by the map data.

6.3.4

Communication file The size of the communication or COM-file (e.g. for the other Delft3D modules, such as the water quality module WAQ) from the hydrodynamic simulation is determined by: The number of grid cells in horizontal and vertical direction, i.e. Mmax · Nmax · Kmax: C1. The number of quantities stored: C2. The number of process parameters taken into account, such as salinity, temperature, constituents and turbulence quantities: C3. The number of time steps, for which the communication file is written: C4. You can estimate the size of a communication file (in bytes) from the following formula: size communication file = C1 · (C2 + C3) · C4 · 4 bytes.

166

Deltares

Execute a scenario As a first approximation you can use C2 = 15. Example: A COM-file size of 20.0 Mbytes should be expected for a model containing 50 by 50 points by 5 layers, simulated with density driven currents and simulation results stored for a period of 12 hrs 30 min, and the file is written with an interval of 15 minutes. For a more refined estimate you should account for parameters that are stored only once and parameters that are stored for each layer. However, the contribution of the latter to the file size easily dominates by far the contribution of the other parameters. Remark: The file sizes given are indicative and the figures may not be linearly extrapolated to determine the file size when the number of grid points is enlarged.

6.4

Command-line arguments In the following sections the comand line arguments for the executables <deltares hydro.exe> (see Section 6.4.1) and <delftflow.exe> (see Section 6.4.2) are given. There are several reasons to change the executable name and the command line options, such as: to have clear separation between the executables based on open source and closed source; and preparing the Delft3D-FLOW module for Delta Shell as in that environment the Delft3D-FLOW module will run as a Dynamic Link Library (dll) or as Shared Object (Linux).

6.4.1

Command-line arguments deltares hydro The main program <deltares hydro.exe> requires one command-line argument referring to a configuration file:

deltares_hydro.exe <configuration file>
The configuration file has deafult name <config flow2d3d.ini>. Configuration file Depending on the type of simulation the <config flow2d3d.ini> can contain several options, which grouped together and specified by a keyword
Table 6.2: Description of the configuration file for deltares hydro

Keyword File Information

Record description

FileCreatedBy FileCreationDate

Refers to the user or program which created this file Creation date and time of this configuration file continued on next page

Deltares

167

Delft3D-FLOW, User Manual Table 6.2 – continued from previous page Keyword Record description Version number of theis file

FileVersion
Component

Name

Name of the component which need to run (here flow2d3d) Refers to the Delft3D-FLOW MDF-file (then the keyword DDBFILE is not allowed in this whole configuration file) Refers to the Delft3D-FLOW DDB-file (then the keyword MDFFILE is not allowed in this whole configuration file)

MDFfile

DDBfile

RemoteOLV

JREpath JARpath Wait

Path to the jre subdirectory Path to the <DelftOnline.jar> file

yes
At time step zero the simulation will wait on a connection with the RemoteOLV module. The RemoteOLV module will then take over the simulation control. If the keyword is missing or has a different value the simulation will start without waiting on the RemoteOLV module. When the simulation is running it is still possible to connect the RemoteOLV module. See also the RemoteOLV user manual (RemoteOLV, 2010).

Example single domain simulation

[FileInformation] FileCreatedBy = Hyd-online Tcl script FileCreationDate = Mon Mar 14 13:00:51 +0100 2011 FileVersion = 00.01 [Component] Name = flow2d3d MDFfile = tut_fti [RemoteOLV] JREpath = C:/Delft3D/w32/plugins/jre JARpath = C:/Delft3D/w32/plugins/dol Wait = yes

168

Deltares

Execute a scenario 6.4.2 Command-line arguments delftflow The preprocessing program <tdatom.exe> and the computational program <trisim.exe> have been merged into one program <delftflow.exe>. The verify program <md-ver.exe> has been rendered obsolete. Instead, an option has been added to <delftflow.exe> to perform a simulation without actually performing all computations, a so-called DryRun, see Appendix B.19. The program <delftflow.exe> requires the following command-line arguments:

delftflow.exe inputfile outputfile computational-engine
The different options for the arguments are summarised below. Inputfile Depending on the type of simulation, the input file for <delftflow.exe> can contain several options:

-B -c -C

-d -J

-L -l -M -O -P -r -S -? -w

Use iterator implementation of MinimumBarrier <file> Specify name of configuration file <class path> Reference to the <DelftOnline.jar> file. Default class path is <$D3D HOME/$ARCH/plugins/dol/DelftOnline.jar> integer Specify debug/trace level (default=0=nothing, 9=everything) <Java path> Reference to the <java> directory. Default Java path is <$D3D HOME/$ARCH/plugins/jre> prefix Use global log file with given prefix prefix Use per-node log files with given prefix <hostlist> Comma-separated list of TCP/IP hosts on which to run simulation <urlfile>; recommended <name.url> Run in DelftOnline mode, write URL to specified file Run with MPI runid Specifiy the Delft3D-FLOW runID for a single domain simulation nodeid /masterhandle Start up in slave mode with specified TCP/IP master Print this usage synopsis Has only effect if also option -O is used, otherwise this option is skipped. The simulation will wait at the zero-th time step on connection with the RemoteOLV module, which will take over the simulation control.

Deltares

169

Delft3D-FLOW, User Manual Outputfile This functionality is not available yet. One should therefore specify a dummy value for the output file, see the example below.

Computational engine The last argument that is needed to run <delftflow.exe> is the computational engine that is used to run the simulation. Here one must specify delft3d.

Examples The command to start a Delft3D-FLOW simulation is:

delftflow.exe d3d-input.txt dummy delft3d
where the contents of the file <d3d-input.txt> depends on the kind of simulation. 1 For a single domain simulation with runid f34 the file <d3d-input.txt> has the contents:
-r f34

2 For a multi-domain simulation on single host (multi)processor the file <d3d-input.txt> has the contents, where the domain decomposition boundaries are given in file <f34.ddb>:
-c f34.ddb

3 For a multi-domain simulation on distributed cluster using TCP/IP the file <d3d-input.txt> has the contents, where the host list is given in the file <hostlist.txt>:
-c f34.ddb -M hostlist.txt

4 For a multi-domain simulation on distributed or SMP system using MPI the file <d3dinput.txt> has the contents, where the host list is given in the file <hostlist.txt>:
-c f34.ddb -P hostlist.txt

6.5

Frequently asked questions This chapter aims to help you with common questions that may arise while using Delft3DFLOW. 1 Question What to do if the message No convergence in UZD appear often in the diagnostic file Answer In general this message appear in the beginning of your simulation and it means that the Red-Black Jacobi iteration is not converged. If this iteration process is not converged the computed momentum has not reached the desired accuracy, after the maximum number of iterations (i.e. hard coded on 50 iterations). As no convergence is reported for the first few time steps (tidal cycles) there is no need to adjust the input parameters of your simulation. The simulations of Delft3D-FLOW are boundary driven and the mismatch of momentum should disappears after a few tidal cycles. But if the message No convergence in UZD still appears after a few tidal cycles you have to decrease the timestep. 2 Question What to do if the message Water level gradient too high > x m (per 0.5 DT) after nst timesteps in the following points: m, n appear in the diagnostic file Answer In general this message appear in the beginning of your simulation and it means that the water level changes are too high for the given time step (nst) and at location (m, n).

170

Deltares

Execute a scenario The simulations of Delft3D-FLOW are boundary driven and the mismatch should disappears after a few tidal cycles. But if the message Water level gradient too high . . . still appears after a few tidal cycles you have to decrease the timestep.

Deltares

171

Delft3D-FLOW, User Manual

172

Deltares

7 Visualise results
7.1 Introduction Visualisation and animation of results is an essential part of presenting the results of your project. In Delft3D-FLOW you have several options: Online Visualisation: to inspect your results during the computation, see Section 4.5.12. Delft3D-GPP: to display and print a large variety of graphs, store the defined sets of figures in a session file and use the same definitions in a next computation. Delft3D-QUICKPLOT: to display quickly and print a large variety of graphs, save the defined sets of figures, etc. Delft3D-MATLAB1 interface: to load, process or display your results in the MATLAB environment. Delft3D-GISVIEW interface: to load, process or display your results in the ArcGIS2 environment. The interfaces to MATLAB and ArcGIS are additional utilities in Delft3D and must be acquired separately. In the following sections we will briefly discuss each of these options; for full details you are referred to the respective User Manuals.

7.2

Working with Delft3D-GPP Delft3D-GPP offers a comprehensive selection and plotting facility to visualise or animate simulation results, to import and visualise other data such as measurements, or to export selected data sets of the results for use in other programs. You can define a single figure or a set of figures and inspect it on screen or make a hardcopy of it on one of the supported hard copy devices. The figures can be processed in an interactive manner or in the background (batch) mode. In this section we only give a very concise description of the post-processor. For a detailed description of its use and functionalities you are referred to the Delft3D-GPP User Manual.

7.2.1

Overview When executing a project with many simulations the amount of data from which a set will be visualised can be enormously large; also the files in which the data are stored can be very large. Therefore it is not optimal to search the original result files over and over again for each parameter or for each new figure. Delft3D-GPP, instead, provides a mechanism to make a selection of the various results and parameters before starting the actual visualisation process and makes a kind of reference list to these sets of data. Next you can define one or more graphs and fill them with data from these data sets. As Delft3D-GPP knows were to find this data, retrieving the data is executed very efficiently. This efficiency is further increased by the option to select all observation points for a certain quantity or to select all time instances at which a quantity is stored in the map file and let you make the final selection when producing the figure.
1 2

MATLAB is software for PC desktop and is a trademark of The MathWorks, Inc, Natick, MA, USA. ArcGIS is GIS-software for desktop and is a trademark of ESRI, Inc. Redlands, CA, USA.

Deltares

173

Delft3D-FLOW, User Manual

M eta-data

Data

m odels

file types

files

presentation

data sets

Session file

layouts

plots

Figure 7.1: Hierarchy of Delft3D-GPP

Delft3D-GPP has access to the communication file and to the result files of all Delft3D modules and in fact to many other programs of Deltares, so you can combine almost any kind of data in a figure. Delft3D-GPP uses a certain hierarchy in the data and the meta-data, see Figure 7.1. We distinguish meta-data to specify the definitions of a figure at a high level of abstraction (left part of Figure 7.1) and the actual data (right part of Figure 7.1).

Meta-data: models

file types

presentations layouts

Defines the set of models the results of which can be visualised. You can change this set of models to limit the options presented in Delft3D-GPP menus. Defines the set of file types that can be used. You can change this set of file types to limit the options presented in Delft3D-GPP menus. Defines the set of data-presentation methods, such as contour maps or xy-graphs. Defines the set of layouts that can be used in a figure. You specify the general set-up of a figure by defining the appropriate layout, the size of the graph, the plot areas, their position, additional text etc.

174

Deltares

not the data itself. 7. Deltares 175 . For the data only the references to the data is stored in the session file. The basic functions are shortly described below. You can give selected data sets a useful name. Session Description Datasets To load an existing session file or to save the settings and selections of the current session in a session file for later use.GPP and next Figure 7. List of pre-selected data sets to be used in the current plot session. which will be presented on your monitor or printed on paper. To give a short description of a session file. The actual data sets selected from the files and to be used in the visualisation. including the data. The actual figures.2: Main window of Delft3D-GPP Data: files data sets plots session file The actual files to be used in your plot session. An ASCII file containing all information that defines the figures.Visualise results Figure 7.2 Launching Delft3D-GPP To start Delft3D-GPP: Select from the Delft3D-MENU FLOW .2 is displayed.2. for full details you are referred to the Delft3DGPP User Manual. this information is used only for reference. At start-up the selections are displayed of the previous plot session in the current directory.

3. To have a quick view on a data set: Select in Figure 7. Select the required data file in the file selection window that is being displayed.Add in Figure 7. To add a data set or plot layout.4 some of the options are displayed to change attributes of the figure.Delft3D-FLOW. In Figure 7. To export the selected single or combined data set to an ASCII file or GIS-file (for single data sets only). divide. To combine any of the available (single) data sets to a new data set. and save the new data set under a unique name. The selected data set and a default plot layout will be displayed in the Plot window. depending which function on the left side of the list box has been selected.2. 176 Deltares . To preview a selected data set or plot layout from the list displayed in the list box. Add Preview Combine Export Delete To add a data set of a specific result file to the Available data sets in Figure 7.2 the required data set and click Preview. You can make as many selections from a specific result file. The parameters and locations (or time in case of map-results) available in a selected result file are displayed in the Add dataset window.3. Remark: Delft3D-GPP recognises the type of data selected and uses an appropriate default presentation method to display the results. To delete the selected data set or plot layout. see Figure 7. At start-up the selections are displayed of the previous plot session in the current directory. or from different results files (to combine results from different computations or models) as you like.3. User Manual Plots List of pre-selected plot layouts to be used in the current plot session. such as multiply. take the maximum value etc. see Figure 7. You are referred to the User Manual of Delft3D-GPP for full details on how to use Delft3DGPP. Figure 7.2: Select Datasets . Click Select File in the Add dataset window.

dat> file Deltares 177 .Visualise results Figure 7.3: Parameters and locations in the <trih-f33.

4: Some options to change the plot attributes 178 Deltares .Delft3D-FLOW. User Manual Figure 7.

k) indices.4 Delft3D-GISVIEW interface To access. 2009). After making a selection of time steps. The interface can be started from the Delft3D-MENU or by typing d3d qp at the MATLAB command prompt.3 Working with Delft3D-QUICKPLOT A graphical interface.Visualise results Figure 7. see Figure 7-5. The interface allows to open NEFIS files and to select data fields. an interface between Delft3D and ArcGIS was developed. For full details of the Delft3D-GISVIEW interface you are referred to the Delft3D-GISVIEW User Manual (Delft3D-GISVIEW. visualise and process Delft3D map-results in a GIS-environment. 7. you can either load the data (Open a data file button) or plot the data (Quick View button).5: Delft3D-QUICKPLOT interface to Delft3D result 7. Deltares 179 . See the Delft3D-QUICKPLOT User Manual for full details and a description of the routines and their use. is provided to make a basic graphical representation of the data or make a quick animation. This interface is provided as an extension to ArcGIS. From a user point of view three menu-options are added to the standard menu-bar of ArcGIS. stations and (m. Delft3D-QUICKPLOT. n.

User Manual 180 Deltares .Delft3D-FLOW.

i. such as the bed stress or wind field during a calibration phase. A typical hierarchy of steps for a hydrodynamic – water quality assessment study might be: Hydrodynamics: Inspection and analysis of the problem to be solved. both as time-series and as maps. This is necessary to guarantee the quality of the results and the consistency between input data and result. Determining the simulation period to be stored for the water quality study and generating the hydrodynamic database for the water quality simulations. crosssections and verticals. We suggest to organise your project directory along the hierarchy of steps typical for your project. such as the curvilinear grid. Setting up of the hydrodynamic model. Remark: The number of water quality computations can be very large due to the number of processes that must be simulated and the number of parameters that must be investigated in a sensitivity analysis.8 Manage projects and files 8. Organising the files in your project in a logical sense is important to keep track of all simulations and changes you have investigated. Typically. definition of the model extent. while other files may change appreciably for each simulation. Determining and executing the verification runs and accessing the final accuracy of the simulations. Deltares 181 . Collecting and processing the bathymetric data and generating the model bathymetry. the related computations each with its specific input. Accessing the results. performing a (in most cases several) sensitivity analysis and iterating on the calibration process if necessary. Defining the calibration runs to be executed. Determining if the same or a coarser grid shall be used for the water quality simulations and transfer the hydrodynamics defined on the hydrodynamic grid to the water quality grid. generation of the land boundary outline and the curvilinear grid. Preparing boundary conditions.1 Introduction When executing a project generally a large number of files are created. attribute and results files must be organised in some way to control the large amounts of files. the grid size.e. Processing field measurements for calibration and verification data in stations. Water quality: A similar set of steps can be defined for the water quality assessment study. This hierarchy of steps. the number of layers (in case of a 3D simulation). some of these files have a rather general use and will not or hardly change during the project. Determining the water quality processes and parameters that must be involved. of the available data and information of the area and of the processes involved. Defining a couple of simulations to make a first assessment of the model behaviour.

Furthermore. etc. Organise the directory in-line with the various steps mentioned in the introduction. Fortunately.mdf>. Do not use blanks in the runid. such as preliminary assessment. Currently. Online compression is only a partial solution as a typical compression factor for result files is only 2 or 3. Execute each dominant step in the project. all files used in a scenario must be in the current working directory.1 Managing projects It is of utmost importance to structure your project in a logical way. 182 Deltares . they may easily result in full file systems. User Manual 8. and give them useful names. This problem is solved to a large extent if the FLOW-GUI can handle long file names and paths. 8.2 Managing files The length of the MDF-file name is limited to 256 characters. i. the FLOW Graphical User Interface (GUI) only supports short file names in a working directory. and has the form <runid.1. the costs or mass storage devices is decreasing at a rapid rate. Below we give some general ideas to structure your projects.1. A probably more urgent problem is to handle the large result files. You are advised to prefer the structuring of your simulations above optimising the amount of disk space used for the attribute files. the bathymetry. define a project directory. Organising your project as sketched above means that many unchanged files must be copied to each (sub-)directory and so will reside many times on your file system.Delft3D-FLOW. thereby increasing your disk requirements. The result files are usually the real big ones so the storage problem is relieved quite a bit by moving these files. There are many ways to structure a project and there is probably not just one best way. The structure you want to apply may depend on company rules or personal preferences. it helps to retrieve results when after some time you have to answer additional questions. An option in case of a full file system is to move the large result files to a background storage facility and only load them back on the file system when necessary. the grid. the field data.e.e. Remark: Never combine the program files with project data. This option will be available in a next release. i. This will make it easy to understand its structure by colleagues working on the same project. However. To manage your projects you might apply the following steps: Define for each project a separate directory or file system. calibration and verification in separate sub-directories and execute sensitivity studies within these steps in sub-sub-directories. the alternative of ‘all scenarios in one and the same directory’ is not attractive either as you easy will have hundreds of files. prepare sub-directories for preparing the boundary conditions.

It aims to model flow phenomena of which the horizontal length and time scales are significantly larger than the vertical scales. For steady and non-steady modelling of the far-field water quality and ecology. However. sediment transport and morphology can be obtained from appropriate mathematical models. Delft3D is the integrated flow and transport modelling system of Deltares for the aquatic environment. Whether the problem is related. The flow module of this system. storm surges. for example. Examples in which the two-dimensional.9 Conceptual description 9. it is coupled with the far-field water quality module D-Water Quality. rivers and lakes. harbour oscillations (seiches) and transport of pollutants in vertically well-mixed flow regimes. The numerical algorithms are described in Chapter 10.1 General background Range of applications of Delft3D-FLOW The hydrodynamic module Delft3D-FLOW simulates two-dimensional (2DH. which corresponds to solving the depth-averaged equations. Non-steady modelling of the mid-field water quality is performed by coupling the flow module to the particle tracking simulation module Delft3DPART. to the stability of a hydraulic structure. provides the hydrodynamic basis for other modules such as water quality. Delft3D-FLOW. Most of the concepts and algorithms are applicable to both the σ -model and Z -model. Reliable information on water flow. water quality. a depth-averaged approach is appropriate. the specifics of the Z co-ordinate system will be presented in more detail in Chapter 12. estuaries. For the interaction between waves and currents the flow module may be coupled with the short-waves model Delft3D-WAVE. ecology. In general the first step in such modelling activities concerns the simulation of the flow itself. Delft3DFLOW is able to run in two-dimensional mode (one computational layer). to salt intrusion. viz. including the effect of density differences due to a non-uniform temperature and salinity distribution (density-driven flow). flow simulations usually form the basis of the investigations to be carried out. tsunamis.2 9.1 Introduction Increasing awareness of environmental issues has focused the attention of scientists and engineers on the problem of predicting the flow and dispersion of contaminants in water systems. coastal areas. If the fluid is vertically homogeneous. to the dispersion of pollutants or to the transport of silt and sediment. waves and morphology. depthaveraged flow equations can be applied are tidal waves. In the vertical direction Delft3D-FLOW offers two different vertical grid systems: a so-called σ co-ordinate system (σ -model) introduced by Phillips (1957) for ocean models and the Cartesian Z co-ordinate system (Z -model). Three-dimensional modelling is of particular interest in transport problems where the horiDeltares 183 .2. The flow model can be used to predict the flow in shallow seas. lagoons. This chapter gives some background information on the conceptual model of the Delft3DFLOW module. 9. depth-averaged) or three-dimensional (3D) unsteady flow and transport phenomena resulting from tidal and/or meteorological forcing. waves.

The effect of the Earth’s rotation (Coriolis force). Drying and flooding of tidal flats. Tidal forcing at the open boundaries.2 Physical processes The numerical hydrodynamic modelling system Delft3D-FLOW solves the unsteady shallow water equations in two (depth-averaged) or in three dimensions.2. fresh water river discharges in bays and thermal stratification in lakes and seas. upwelling and downwelling of nutrients. Horizontal density gradients in the pressure (baroclinic effects). heat and other conservative constituents. pressure gradients due to free surface gradients (barotropic) or density gradients (baroclinic). whereas in spherical co-ordinates the reference plane follows the Earth’s curvature. Wave induced stresses (radiation stress) and mass fluxes. river discharges). Space and time varying wind shear-stress at the water surface. bed topography or density differences. and the transport equations for conservative constituents. Evaporation and precipitation. Space and time varying atmospheric pressure on the water surface. In Delft3D-FLOW models with a rectangular grid (Cartesian frame of reference) are considered as a simplified form of a curvilinear grid. Lateral shear-stress at wall. Turbulence induced mass and momentum fluxes (turbulence closure models). Flow through hydraulic structures. Influence of waves on the bed shear-stress (2D and 3D). 9.g. wind stress at the free surface. Source and sink terms are included in the equations to model the discharge and withdrawal of water. Transport of salt. Space varying shear-stress at the bottom.Delft3D-FLOW. Time varying sources and sinks (e. Tide generating forces. bed stress. Water with variable density (equation of state). The system of equations consists of the horizontal equations of motion. User Manual zontal flow field shows significant variation in the vertical direction. Coriolis force. The flow is forced by tide at the open boundaries. the continuity equation. Wind driven flows including tropical cyclone winds. Heat exchange through the free surface. 184 Deltares . This variation may be generated by wind forcing. Vertical exchange of momentum due to internal waves. In curvilinear co-ordinates. The Delft3D-FLOW model includes mathematical formulations that take into account the following physical phenomena: Free surface gradients (barotropic effects). Examples are dispersion of waste or cooling water in lakes and coastal areas. The equations are formulated in orthogonal curvilinear co-ordinates or in spherical co-ordinates on the globe. the free surface level and bathymetry are related to a flat horizontal plane of reference. salt intrusion in estuaries. Effect of secondary flow on depth-averaged momentum equations.

the basic equations are Reynolds-averaged introducing so-called Reynolds stresses. we refer to Chapter 12 of this User Manual. The formulation for the enhanced bed shear-stress due to the combination of waves and currents is based on a 2D flow field. The horizontal eddy viscosity and diffusivity coefficients should combine both the effect of the 3D turbulent eddies and the horizontal motions that Deltares 185 . In case of small-scale flow (partial slip along closed boundaries). For details. Thus. These stresses are related to the Reynolds-averaged flow quantities by a turbulence closure model. At the bottom a slip boundary condition is assumed. L is the mixing length and ε is the dissipation rate of turbulent kinetic energy. a quadratic bottom stress formulation is applied. The boundary conditions for the turbulent kinetic energy and energy dissipation at the free surface and bottom assume a logarithmic law of the wall (local equilibrium). The coefficients may be specified (constant) or computed by means of an algebraic. These equations are derived from the three dimensional Navier-Stokes equations for incompressible free surface flow. In spherical co-ordinates the inertial frequency depends on the latitude. which means that the vertical momentum equation is reduced to the hydrostatic pressure relation. In Delft3D-FLOW the 3D turbulent eddies are bounded by the water depth.3 Assumptions underlying Delft3D-FLOW In Delft3D-FLOW the 2D (depth-averaged) or 3D non-linear shallow water equations are solved. where k is the turbulent kinetic energy. vertical accelerations are assumed to be small compared to the gravitational acceleration and are therefore not taken into account. When this assumption is not valid then Delft3D provides an option to apply the so-called Non-hydrostatic pressure model in the Z -model. In a Cartesian frame of reference. In agreement with the aspect ratio for shallow water flow. the Coriolis parameter is assumed to be uniform unless specifically specified otherwise. Therefore. Their contribution to the vertical exchange of horizontal momentum and mass is modelled through a vertical eddy viscosity and eddy diffusivity coefficient (eddy viscosity concept). In Delft3D-FLOW vertical density differences are taken into account in the horizontal pressure gradients and in the vertical turbulent exchange coefficients. the immediate effect of buoyancy on the vertical flow is not considered. The coefficients are assumed to be proportional to a velocity scale and a length scale.Conceptual description 9. The following assumptions and approximations are applied: In the σ co-ordinate system the depth is assumed to be much smaller than the horizontal length scale. but usually the hydrodynamic grids are too coarse to resolve the fluctuations. The eddy viscosity is an-isotropic. Furthermore. k -L or k -ε turbulence model. In the σ co-ordinate system. So the application of Delft3D-FLOW is restricted to mid-field and far-field dispersion simulations of discharged water. The effect of variable density is only taken into account in the pressure term (Boussinesq approximation). the horizontal gradients are included in the production term. the production of turbulence is based on the vertical (and not the horizontal) gradients of the horizontal flow. For a dynamic online coupling between morphological changes and flow the 3D sediment and morphology feature is available. The equations of Delft3D-FLOW are capable of resolving the turbulent scales (large eddy simulation). generated from the velocity near the bed using a logarithmic approximation. the effect of the Earth’s curvature is not taken into account. For such a small aspect ratio the shallow water assumption is valid.2.

The flux of matter through a closed wall and through the bed is zero. we will present in detail the governing equations.Delft3D-FLOW. In case of small-scale flow the complete Reynold’s stress tensor is computed. The drying and flooding procedure leads to a discontinuous movement of the closed boundaries at tidal flats. The velocity point is set wet again when the local water depth is above the threshold. n) near-bed peak orbital excursion reference height as defined by Van Rijn artificial compression coefficient buoyancy flux term in transport equation for turbulent kinetic energy 186 Deltares . convection. Below an overview of the symbols that are used in the equations is presented. The heat loss through the bottom is always zero. the horizontal viscosity terms are reduced to a bi-harmonic operator along co-ordinate lines. The shear-stress at the side walls is calculated using a logarithmic law of the wall. the basic equations are averaged in analogy with turbulence introducing socalled radiation stresses. It is assumed that a velocity point is set dry when the actual water depth is below half of a user-defined threshold. Without specification of a temperature model. In the σ co-ordinate system.3 Governing equations In this section. then the velocity at that point is set to zero.n) ˆδ A a α Bk m2 m m m2 /s3 area of computational cell at location (m. If the total heat flux through the water surface is computed using a temperature excess model the exchange coefficient is a function of temperature and wind speed and is determined according to Sweers (1976). and heat loss due to evaporation are modelled separately. A continuity cell is set dry when the four surrounding velocity points at the grid cell faces are dry or when the actual water depth at the cell centre is below zero (negative volume). These stresses are related to the wave quantities of Delft3DWAVE by a closure model. In case of small-scale flow partial slip is applied along closed boundaries. 9. Therefore. User Manual cannot be resolved by the horizontal grid. The natural background temperature is assumed constant in space and may vary in time. the heat exchange through the free surface is zero. atmospheric and back radiation. List of Symbols Symbol Units Meaning A(m. The effect of precipitation on the water temperature is accounted for. In the other heat flux formulations the fluxes due to solar radiation. The horizontal eddy viscosity is generally much larger than the vertical eddy viscosity. The hysteresis between drying and flooding is introduced to prevent drying and flooding in two consecutive time steps. the tangential shear-stress at lateral closed boundaries can be neglected (free slip). For large-scale flow simulations. If the point is set dry. The pressure is hydrostatic and the model is not capable of resolving the scales of short waves. Delft3D-FLOW solves the so-called long wave equation. For large-scale flow simulations.

d positive downward) root-mean-square wave height spiral motion intensity (secondary flow) equilibrium intensity of spiral motion due to curvature of stream lines 187 Deltares .Conceptual description Symbol Units Meaning Be C. C2D C3D Cd c ca cD cp cµ cµ D D DP SED D∗ d Ds D50 D90 E E ea es Fc Fx Fy Fξ Fη f f (Fc ) fFIXFAC fMORFAC fw fw fwind Gξξ Gηη g H Hrms I Ibe m2 /s4 m1/2 /s m1/2 /s kg/m3 kg/m3 J/(kg◦ C) kg/s3 kg/(m2 s) m m m m m m/s kg/(m2 s) mbar mbar m/s2 m/s2 m/s2 m/s2 1/s m m m/s2 m m m/s m/s buoyancy flux term in transport equation for the dissipation of turbulent kinetic energy ´ coefficient 2D Chezy ´ coefficient 3D Chezy wind drag coefficient mass concentration mass concentration at reference height a constant relating mixing length. turbulent kinetic energy and dissipation in the k -ε model specific heat of sea water calibration constant constant in Kolmogorov-Prandtl’s eddy viscosity formulation dissipation rate wave energy deposition rate cohesive sediment depth of sediment available at the bed non-dimensional particle diameter depth below some horizontal plane of reference (datum) representative diameter of suspended sediment median diameter of sediment sediment diameter evaporation erosion rate cohesive sediment vapour pressure at a given air temperature saturated vapour pressure at a given temperature cloudiness factor radiation stress gradient in x-direction radiation stress gradient in y -direction turbulent momentum flux in ξ -direction turbulent momentum flux in η -direction Coriolis parameter (inertial frequency) attenuation function for the solar and atmospheric (long wave) radiation fixed layer proximity factor user-defined morphological acceleration factor total wave friction factor grain-related wave friction factor dependency of λ of the wind speed in temperature excess model coefficient used to transform curvilinear to rectangular coordinates coefficient used to transform curvilinear to rectangular coordinates acceleration due to gravity total water depth (H = d + ζ .

ky ks.n equilibrium intensity of spiral motion due to Coriolis turbulent kinetic energy increased (apparent) bed roughness height felt by the current in the presence of waves Nikuradse roughness length current related bed roughness height wave number wave related bed roughness height mixing length index number of sediment fraction depth-averaged mass flux due to Stokes drift in x-direction depth-averaged mass flux due to Stokes drift in y -direction source or sink of momentum in ξ -direction source or sink of momentum in η -direction Manning’s coefficient hydrostatic water pressure precipitation production term in transport equation for turbulent kinetic energy gradient hydrostatic pressure in ξ -direction gradient hydrostatic pressure in η -direction production term in transport equation for the dissipation of turbulent kinetic energy global source or sink per unit area atmospheric radiation (long wave radiation) net atmospheric radiation back radiation (long wave radiation) heat loss due to convection heat loss due to evaporation solar radiation (short wave radiation) solar radiation with a clear sky net solar insolation heat flux through free surface local source per unit volume local sink per unit volume radius of the Earth gradient Richardson’s number radius of curvature of a streamline in the secondary flow model reflection coefficient for solar and atmospheric (long wave) radiation relative air humidity at a given temperature salinity representative surface area in heat model maximum salinity for flocculation calculations additional bedload transport vector. in the down slope direction 188 Deltares .Delft3D-FLOW.c kx . User Manual Symbol Units Meaning Ice k ka ks ks.w L S Mx S My Mξ Mη n P P Pk m/s m2 /s2 m m m 1/m m m kg m/s kg m/s m/s2 m/s2 m−1/3 s kg/(ms2 ) m/s m2 /s3 kg/(m2 s2 ) kg/(m2 s2 ) m2 /s4 m/s J/(m2 s) J/(m2 s) J/(m2 s) J/(m2 s) J/(m2 s) J/(m2 s) J/(m2 s) J/(m2 s) J/(m2 s) 1/s 1/s m m % ppt m2 ppt Pξ Pη Pε Q Qa Qan Qbr Qco Qev Qs Qsc Qsn Qtot qin qout R Ri Rs r rhum S Sarea Smax Sb. The direction of this vector is normal to the unadjusted bedload transport vector.

or ξ -direction total velocity due to flow and Stokes drift in x.or η -direction velocity of water discharged in y .or ξ -direction fluid velocity in the y .or ξ -direction amplitude of the near-bottom wave orbital velocity Stokes drift in x.uu s s T ¯ T T Ta Ta Tback Tp TTHbottom TTHsurface Ts Tξξ .or η -direction velocity at bed boundary layer in y .cr ub u∗ u∗ u∗ u∗b ub ∗ us ∗ U U10 ˆ U ˆ U orb uS u ˜ v vb V ˆ V vS v ˜ w Deltares kg/(ms) kg/(ms) kg/(ms) ppt ◦C K ◦C ◦C s s s ◦C kg/(ms2 ) s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s m/s magnitude of the unadjusted bedload transport vector (adjusted for longitudinal bed slope only) mass bedload transport rate magnitude of the bedload transport vector computed bedload sediment transport vector in u-direction. V )T averaged wind speed at 10 m above free surface velocity of water discharged in x.or η -direction total velocity due to flow and Stokes drift in y .or η -direction Stokes drift in y . Tηη . held at the u-point of the computational cell at location (m.or η -direction fluid velocity in z -direction 189 . n) relative density of sediment fraction (= ρs /ρw ) salinity temperature (general reference) temperature (general reference) non-dimensional bed-shear stress non-dimensional bed-shear stress for reference concentration air temperature natural background temperature of water in the model area peak wave period Thatcher Harleman return time for the bed layer Thatcher Harleman return time for the surface layer water temperature at free surface contributions secondary flow to shear stress tensor time depth-averaged velocity in ξ -direction velocity of water discharged in ξ -direction peak orbital velocity at the bed flow velocity in the x.n) Sb.Conceptual description Symbol Units Meaning |Sb | Sb |Sb | (m. Tξη t U ˆ U ˆδ U u ub.or η -direction depth-averaged velocity in y .or ξ -direction critical (threshold) near-bed fluid velocity near-bed fluid velocity vector friction velocity due to currents or due to current and waves effective bed shear velocity vertically averaged friction velocity friction velocity at the bed modified friction velocity near bed friction velocity at the free surface magnitude of depth-averaged horizontal velocity vector (U.

z γ γ φ φ φ ϕ ϕe ϕeo 190 .y .n) ∆ zb ∆ zs ∆σb εε ε ε εf εf.n) ∆y (m. εf. n) cell width in the y -direction. User Manual Symbol Units Meaning ws.z εs εs. y. z -directions. respectively extinction coefficient (measured) Bowen’s constant latitude co-ordinate in spherical co-ordinates exchange coefficient for the heat flux in excess temperature model internal angle of friction of bed material (assumed to be 30◦ ) equilibrium tide perturbation of the equilibrium tide due to earth tide perturbation of the equilibrium tide due to tidal load Deltares αbn β βc δm δw χ ∆r (m. held at the v -point of cell (m.y . y. respectively sediment diffusion in the z -direction sediment diffusion coefficients in the x. held at the u-point of cell (m. z z0 z ˜0 z0n ∂zb ∂s ∂z(u) ∂x ∂z(v) ∂y ∂zb ∂n m/s m/s m m m m m m deg m kg/m2 s m m m m m2 /s4 m2 /s3 m2 /s m2 /s m2 /s m2 /s m−1 mbar/◦ C deg J/(sm2 ◦ C) deg m m m particle settling velocity in clear water (non-hindered) particle (hindered) settling velocity in a mixture Cartesian co-ordinates bed roughness length enhanced bed roughness length due to current and waves roughness length normal to the boundary bed slope in the direction of bedload transport bed slope in the positive x-direction evaluated at the downwind u-point bed slope in the positive y -direction evaluated at the downwind v -point bed slope in the direction normal to the bedload transport vector user-defined coefficient ratio of sediment and fluid mixing coefficient to account for secondary flow in momentum equations thickness of wave boundary mixing layer thickness of wave boundary layer astronomical argument of a tidal component ripple height change in quantity of bottom sediment at location (m. εs. n) thickness of the bed layer thickness of the surface layer thickness of the bed boundary layer in relative co-ordinates dissipation in transport equation for dissipation of turbulent kinetic energy dissipation in transport equation for turbulent kinetic energy emissivity coefficient of water at air-water interface fluid diffusion in the z -direction fluid diffusion coefficients in the x. y. εf.n) ∆SED ∆t ∆x(m. z -directions. n) computational time-step cell width in the x-direction. εs.x .0 ws x.Delft3D-FLOW.x .

cr τb. σ = d +ζ . σ = −1) Prandtl-Schmidt number Prandtl-Schmidt number for constituent (0. (surface.and η -direction) background vertical eddy viscosity for momentum equations background horizontal eddy diffusivity (ξ .cw τb. σ = 0.7 for salinity and temperature.0 for suspended sediments) Prandtl-Schmidt number for molecular mixing (700 for salinity.d τcr. bed level.and η -direction) background vertical eddy diffusivity for transport equation vertical eddy viscosity part of eddy viscosity due to horizontal turbulence model part of eddy viscosity due to turbulence model in vertical direction relative availability of the sediment fraction in the mixing layer efficiency factor current density of water density of air reference density of water Stefan-Boltzmann’s constant z −ζ scaled vertical co-ordinate.c τb. 1.7 for temperature) bed shear stress due to current and waves bed shear stress in ξ -direction bed shear stress in η -direction magnitude of the bed shear stress due to current alone magnitude of the at bed shear stress due to waves alone magnitude of the wave-averaged at bed shear stress for combined waves and current bed shear stress due to current critical bed shear stress bed shear stress due to current in the presence of waves bed shear stress due to waves critical bed shear stress user-defined critical deposition shear stress user-defined critical erosion shear stress mean bed shear stress due to current and waves maximum bottom shear stress with wave-current interaction mean (cycle averaged) bottom shear stress with wave-current interaction shear stress at surface in ξ -direction shear stress at surface in η -direction velocity in the σ -direction in the σ -co-ordinate system angular frequency waves angular frequency of tide and/or Fourier components 191 .w τcr τcr. 6.e τcw τmax τmean τsη τsη ω ω ω Deltares m deg 1/s m2 /s m2 /s m2 /s m2 /s m2 /s m2 /s m2 /s m2 /s – – kg/m3 kg/m3 kg/m3 J/(m2 s K4 ) – – – – N/m2 kg/(ms2 ) kg/(ms2 ) kg/(ms2 ) kg/(ms2 ) kg/(ms2 ) N/m2 N/m2 N/m2 N/m2 N/m2 N/m2 N/m2 N/m2 N/m2 N/m2 kg/(ms2 ) kg/(ms2 ) m/s 1/s deg/hour perturbation of the equilibrium tide due to oceanic tidal load longitude co-ordinate in spherical co-ordinates first order decay coefficient kinematic viscosity (molecular) coefficient background horizontal eddy viscosity (ξ .Conceptual description Symbol Units Meaning ϕo λ λd νmol back νH back νV back DH back DV νV ν2D ν3D η µc ρ ρa ρ0 σ σ σc σ c0 σmol τb τbξ τbη τc τw τm τb.

The boundary becomes irregular and may introduce significant discretization errors. WGS84). The velocity scale is in physical space. In the vertical direction Delft3D-FLOW offers two different vertical grid systems: the σ coordinate system (σ -model) and the Cartesian Z co-ordinate system (Z -model). see Chapter 10 for the numerical aspects. Spherical co-ordinates are a special case of orthogonal curvilinear co-ordinates with: ξ = λ.1) in which λ is the longitude. η ζ ζb ζe ζ eo J/(m2 s) m m m m heat flux through free surface horizontal. In 3D models the vertical velocities are computed from the continuity equation. To reduce these errors boundary fitted orthogonal curvilinear co-ordinates are used. φ is the latitude and R is the radius of the Earth (6 378. an estuary or a coastal sea are in general curved and are not smoothly represented on a rectangular grid. The grid transformation introduces curvature terms in the equations of motion. In Delft3D-FLOW the equations are formulated in orthogonal curvilinear co-ordinates. The set of partial differential equations in combination with an appropriate set of initial and boundary conditions is solved on a finite difference grid. In the horizontal direction Delft3D-FLOW uses orthogonal curvilinear co-ordinates. under the shallow water and the Boussinesq assumptions. which leads to the hydrostatic pressure equation.1 Hydrodynamic equations Delft3D-FLOW solves the Navier Stokes equations for an incompressible fluid. 192 Deltares .137 km.3. Two coordinate systems are supported: Cartesian co-ordinates (ξ. The equations for the Z co-ordinate system are similar. The hydrodynamic equations described in this section are valid for the σ co-ordinate system. curvilinear co-ordinates water level above some horizontal plane of reference (datum) bottom tide earth tide tidal loading 9. η = φ. η ) Spherical co-ordinates (λ. Curvilinear co-ordinates also allow local grid refinement in areas with large horizontal gradients. but the components are perpendicular to the cell faces of the curvilinear grid. φ) The boundaries of a river. Gηη = R. Gξξ = R cos φ. User Manual Symbol Units Meaning ψ ξ. (9.Delft3D-FLOW. Equations or relations that are fundamentally different will be discussed separately and in detail in Chapter 12. In the vertical momentum equation the vertical accelerations are neglected.

the free surface elevation above the reference plane (at z = 0).g. heat exchange with the atmosphere) and the near bed area (sediment transport). the total water depth. For each layer a set of coupled conservation equations is solved.2) z ζ d H the vertical co-ordinate in physical space. 1994).1). estuaries and lakes. The co-ordinate lines intersect the density interfaces that may give significant errors in the approximation of strictly horizontal density gradients (Leendertse. the layer interfaces are chosen following planes of constant σ .1: Definition of water level (ζ ).Conceptual description Figure 9. stratified flow can occur in combination with steep topography. 1990. depth (h) and total depth (H . Therefore. The σ co-ordinate system The σ -grid was introduced by Phillips (1957) for atmospheric models. see Figure 9. Although the σ -grid is boundary fitted (in the vertical). d+ζ H (9. The flow domain of a 3D shallow water model consists in the horizontal plane of a restricted (limited) area composed of open and closed (land) boundaries and in the vertical of a number of layers. The vertical grid consists of layers bounded by two σ -planes. wind-driven flows. recently a second vertical grid coDeltares 193 . Cartesian co-ordinate system in the vertical (Z -model) In coastal seas.2).e. The distribution of the relative layer thickness is usually non-uniform. The partial derivatives in the original Cartesian co-ordinate system are expressed in σ co-ordinates by the chain rule introducing additional terms (Stelling and Van Kester. given by d + ζ . it will not always have enough resolution around the pycnocline. 1994). In a σ co-ordinate system the number of layers is the same at every location in the horizontal plane. which are not strictly horizontal but follow the bottom topography and the free surface. At the bottom σ = −1 and at the free surface σ = 0 (see Figure 9. Stelling and Van Kester. This allows for more resolution in the zones of interest such as the near surface area (important for e.2. i. Because the σ -grid is boundary fitted both to the bottom and to the moving free surface. The σ co-ordinate system is defined as: σ= with: z−ζ z−ζ = . a smooth representation of the topography is obtained. irrespective of the local water depth (see Figure 9. The number of layers over the entire horizontal computational area is constant. the depth below the reference plane.

3) with U and V the depth avaraged velocities U= V = 1 d+ζ 1 d+ζ ζ 0 u dz = d ζ −1 0 u dσ v dσ −1 (9. see Figure 9. respectively. a power plant is a withdrawal of water and should be modelled as a sink. for example. We remark that the intake of. At the free surface there may be a source due to precipitation or a sink due to evaporation. The bottom (and free surface) is usually not a co-ordinate line and is represented as a staircase (zig-zag boundary. User Manual Figure 9. The Z -grid model has horizontal co-ordinate lines that are (nearly) parallel with density interfaces (isopycnals) in regions with steep bottom slopes. The Z -grid is not boundaryfitted in the vertical. P the non-local source term of precipitation and E non-local sink term due to evaporation.2: Example of σ .5) v dz = d and Q representing the contributions per unit area due to the discharge or withdrawal of water.and Z -grid ordinate system based on Cartesian co-ordinates (Z -grid) was introduced in Delft3D-FLOW for 3D simulations of weakly forced stratified water systems. and is given by: ∂ζ + ∂t 1 Gξξ ∂ (d + ζ ) U ∂ξ Gηη Gηη + 1 Gξξ ∂ (d + ζ ) V ∂η Gηη Gξξ = (d + ζ )Q. (9. 194 Deltares . taken into account the kinematic boundary conditions at water surface and bed level. This is important to reduce artificial mixing of scalar properties such as salinity and temperature.Delft3D-FLOW.2).4) (9. Continuity equation The depth-averaged continuity equation is derived by integration the continuity equation for incompressible fluids (∇ • u = 0) over the total depth. precipitation and evaporation: 0 Q= −1 (qin − qout ) dσ + P − E. (9.6) with qin and qout the local sources and sinks of water per unit of volume [1/s].

Computation of the physical vertical velocities is only required for postprocessing purposes. Density variations are neglected.Conceptual description Momentum equations in horizontal direction The momentum equations in ξ . except in the baroclinic pressure terms. (9. The forces Fξ and Fη in the momentum equations represent the unbalance of horizontal Reynold’s stresses. These velocities can be expressed in the horizontal velocities.8) The vertical eddy viscosity coefficient νV is defined in Equation 9.9) ∂σ At the surface the effect of precipitation and evaporation is taken into account. The “physical” vertical velocities w in the Cartesian co-ordinate system are not involved in the model equations. The vertical velocity ω is defined at the iso σ -surfaces.or downwelling motions. water levels and vertical ω -velocity according to: w=ω+ 1 Gξξ Gηη u Gηη σ ∂H ∂ζ + ∂ξ ∂ξ +v Gξξ σ ∂H ∂ζ + ∂η ∂η ∂H ∂ζ + ∂t ∂t + . It may be interpreted as the velocity associated with up. (9. wave stresses. Vertical velocities The vertical velocity ω in the adapting σ -co-ordinate system is computed from the continuity equation: ∂ζ + ∂t 1 Gξξ ∂ (d + ζ ) u Gηη + ∂ξ Gηη 1 Gξξ + ∂ (d + ζ ) v ∂η Gηη Gξξ + ∂ω = (d + ζ ) (qin − qout ) . water depths. Pξ and Pη represent the pressure gradients. etc.20. Mξ and Mη represent the contributions due to external sources or sinks of momentum (external forces by hydraulic structures.7) and ∂v + ∂t − u ∂v + Gξξ ∂ξ u2 Gξξ Gηη ∂ v ∂v ω ∂v + + Gηη ∂η d + ζ ∂σ uv Gξξ Gηη + ∂ξ Gξξ 1 Pη + Fη + + fu = − ∂η ρ0 Gηη + 1 ∂ 2 ∂σ (d + ζ ) νV ∂v ∂σ + Mη .). (9. ω is the vertical velocity relative to the moving σ -plane.and η -direction are given by: ∂u + ∂t + u ∂u + Gξξ ∂ξ uv ∂ Gηη v ∂u ω ∂u + − ∂η d + ζ ∂σ Gηη v2 Gξξ Gηη ∂ Gηη + ∂ξ Gξξ Gξξ 1 − fv = − Pξ + Fξ + ∂η ρ0 Gξξ + ∂ 1 2 ∂σ (d + ζ ) ∂ Gηη νV ∂u ∂σ + Mξ . discharge or withdrawal of water.10) + σ Deltares 195 .7. (9. The effects of surface waves on the flow as modelled in Delft3D-FLOW are described in Section 9.

The atmospheric pressure is included in the system for storm surge simulations.4.3. the hydrostatic pressure is given by: 0 (9.15 and 9.11) P = Patm + gH σ ρ ξ. The atmospheric pressure gradients dominate the external forcing at peak winds during storm events. ∂σ After integration. the pressure gradients read: 1 Pξ = ρ0 Gξξ 1 Pη = Gηη 1 g ∂ζ ∂Patm + . Vertical accelerations due to buoyancy effects and due to sudden variations in the bottom topography are not taken into account. η. Space and time varying wind and pressure fields are especially important when simulating storm surges.12) Firstly. see also Stelling and Van Kester (1994). In case of a non-uniform density. Gηη ∂η ρ0 Gηη ∂η (9. 9. we consider water with a constant density and next with a non-uniform density. see Section 9. σ . In order to avoid artificial flow the numerical approximation of the baroclinic pressure terms requires a special numerical approach. In estuaries and coastal seas the vertical grid may deteriorate strongly in case of steep bottom slopes. t dσ . In the horizontal gradient a vertical derivative is introduced by the σ co-ordinate transformation. (9. the local density is related to the values of temperature and salinity by the equation of state.Delft3D-FLOW. 196 Deltares . The discretisations of Delft3D-FLOW are discussed in Section 10.10.16 represents the barotropic pressure gradient (without atmospheric pressure gradients) and the second term the baroclinic pressure gradient.15) 0 σ ρ0 dσ (9. Leibnitz’ rule is used to obtain the following expressions for the horizontal pressure gradients: 1 Pξ = ρ0 Gξξ 1 Pη = Gηη g ∂ζ d+ζ +g Gξξ ∂ξ ρ0 Gξξ g ∂ζ d+ζ +g Gηη ∂η ρ0 Gηη 0 σ ∂ρ ∂ρ ∂σ + ∂ξ ∂σ ∂ξ ∂ρ ∂ρ ∂σ + ∂η ∂σ ∂η dσ (9. the vertical momentum equation is reduced to a hydrostatic pressure equation.14) The gradients of the free surface level are the so-called barotropic pressure gradients. ∂ξ Gξξ ρ0 Gξξ ∂ξ g ∂ζ 1 ∂Patm + . For water of constant density and taking into account the atmospheric pressure.16) The first term in Eqs.13) ρ0 (9. User Manual Hydrostatic pressure assumption (for σ -grid) Under the shallow water assumption. So: ∂P = −gρH.

For a curvilinear grid you should specify the space varying Coriolis parameter. generically called ’ship’ (dship ): Patm = gρ0 dship (9. Coriolis force The Coriolis parameter f depends on the geographic latitude and the angular speed of rotation of the earth. This boundary condition is implemented by reducing the storage area of the grid cell with a factor α (the parameter Riglid in Appendix B.6): α ∂ζ ∂t (9. Rodi (1984)).1. For 3D shallow water flow the stress tensor is an-isotropic. The horizontal viscosity coefficient may be a superposition of three parts: 1 a part due to “sub-grid scale turbulence”. (for details e. Ω: f = 2Ω sin φ. The wave speed below the floating structure increases to gH/α. In 2D simulations this dispersion is not simulated as the vertical profile of the horizontal velocity is not resolved. The pressure of a floating structure at the free surface is dependent on the draft of the structure. the free surface cannot move freely.17) If the free surface is covered by a floating structure.Conceptual description Floating structures To simulate the influence of floating structures. νH . To prevent numerical inaccuracy for wave propagation α must be larger than zero.3. The Reynolds stresses are modelled using the eddy viscosity concept.g.18) with 0 < α ≤ 1. The meaning and the order of the eddy viscosity coefficients differ for 2D and 3D. the redistribution of momentum and matter due to the vertical variation of the horizontal velocity is denoted as dispersion. (1992) and 3 a part due to dispersion for depth-averaged simulations. Reynold’s stresses The forces Fξ and Fη in the horizontal momentum equations represent the unbalance of horizontal Reynolds stresses. such as ships or pontoons. Then this dispersive effect may be modelled as the product Deltares 197 . 2 a part due to “3D-turbulence” see Uittenbogaard et al. In general the eddy viscosity is a function of space and time. The horizontal eddy viscosity coefficient. both the pressure term in the momentum equations and the continuity equation are modified. is much larger than the vertical eddy viscosity νV (νH νV ). We suggest α ≈ 0. In simulations with the depth-averaged momentum and transport equations. The water level should be fixed. This concept expresses the Reynolds stress component as the product between a flow as well as grid-dependent eddy viscosity coefficient and the corresponding components of the mean rate-of-deformation tensor. using a “rigid lid” boundary condition. using a suitable projection. for different horizontal and vertical turbulence length scales and fine or coarse grids.

User Manual of a viscosity coefficient and a velocity gradient. we refer to Section 9. 198 Deltares . The background horizontal viscosity.7 and 9. see Section 9.8) and is not used in the vertical turbulence models (see for example Eqs. νV Delft3D-FLOW a constant (space and time) background mixing coefficient may be specified by you. Consequently. Therefore. The dispersion term may be estimated by the Elder formulation. viscosity νH back has to be specified by you in the GUI in The background horizontal eddy viscosity νH addition to the Elder formulation for 3D-turbulence and dispersion in 2D simulations. (9.128). (9. due to stratification or due to forcing by wind) then a 3D-model for the transport of matter is recommended. user-defined through the input file is represented by back .g.115. which is a background value for the vertical eddy viscosity in the momentum Eqs. For the description of the (vertical) turbulence models for 3D simulations. 9. it may be convenient to specify a background or “ambient” vertical mixing coefficient in order to account for all other back .20) with νmol the kinematic viscosity of water.8.5. The horizontal eddy viscosity is mostly associated with the contribution of horizontal turbulent motions and forcing that are not resolved by the horizontal grid (“sub-grid scale turbulence”) or by (a priori) the Reynolds-averaged shallow-water equations. the vertical eddy viscosity coefficient is defined by: back νV = νmol + max(ν3D .7 and 9. For the former we introduce the sub-grid scale (SGS) horizontal eddy viscosity νSGS and for the latter the horizontal eddy visback . For turbulence closure models responding to shear production only. 9. For details of this approach. νV ). one or two contributions.2 shows that this background horizontal eddy back either contains zero. in Delft3D-FLOW the horizontal eddy viscosity coefficient is defined by νH back νH = νSGS + νV + νH . back is only It is important to emphasize that the lower limit of vertical eddy viscosity νV used in the momentum equations (see Eqs. For the transport equation there is similar lower limit.5. since in Delft3D-FLOW several combinations of horizontal and vertical eddy viscosity are optional. Summarizing. Remarks: We note that the “background horizontal eddy viscosity” represents a series of complicated hydrodynamic phenomena. 9.127 and 9.2 presents an overview of these combinations. see Appendix B. The associated horizontal viscosity coefficient νSGS will then be computed by a dedicated SGS-turbulence model. 9.8. Table 9.Delft3D-FLOW. The 3D part ν3D is computed by a 3D-turbulence closure model. Delft3D-FLOW simulates the larger scale horizontal turbulent motions through a cosity νH methodology called Horizontal Large Eddy Simulation (HLES). including the Elder contribution if requested. If the vertical profile of the horizontal velocity is not close to a logarithmic profile (e.19) The 3D part νV is referred to as the three-dimensional turbulence and in 3D simulations it is computed following a 3D-turbulence closure model. Table 9. Consequently. in addition to all turbulence closure models in forms of unresolved mixing.

1994). the forces Fξ and Fη are of the form: Fξ = 1 ∂τξξ + Gξξ ∂ξ 1 ∂τηξ + Gξξ ∂ξ 1 ∂τξη . computed by vertical turbulence model. when the shear stresses along the closed boundaries may be neglected. (9. (9. see Section 9. no HLES νSGS computed by HLES - back νH (represents) ν3D computed by vertical turbulence model. τξη .24) τηη = 2νH Gηη ∂v ∂v ∂σ + ∂η ∂σ ∂η . Then. the shear stresses τξξ ..Conceptual description Table 9.21) Fη = (9.25) For large-scale flow simulated with coarse horizontal grids. This approach guarantees a positive definite operator. The σ co-ordinate system rotates the Cartesian stress tensor and introduces additional derivatives (Stelling and Van Kester.e. (9. when the shear stresses at the closed boundaries must be taken into account. Thus. Delft3D-FLOW contains a formulation to account for the effect of this spiral motion. i. the horizontal stress tensor is extended with additional shear stresses. also on the numerical grid (Beckers et al. 1998). to model secondary flow in river bends in depth-averaged simulations. The horizontal gradients are taken along σ -planes. with HLES computed by HLES - For the sake of completeness. τηξ and τηη are determined according to: τξξ = 2νH Gξξ ∂u ∂u ∂σ + ∂ξ ∂σ ∂ξ . In Delft3D-FLOW. the stress tensor is redefined in the σ co-ordinate system assuming that the horizontal length scale is much larger than the water depth (Blumberg and Mellor. The horizontal Deltares 199 . 1985) and that the flow is of boundary-layer type.2: Overview of eddy viscosity options in Delft3D-FLOW model description 2D.23) τξη = τηξ = νH 1 Gηη ∂u ∂u ∂σ + ∂η ∂σ ∂η + 1 Gξξ ∂v ∂v ∂σ + ∂ξ ∂σ ∂ξ .6. the forces Fξ and Fη are simplified. i. Gηη ∂η 1 ∂τηη . back νV 2D-turbulence + dispersion coefficient 3D-turbulence + dispersion coefficient 2D-turbulence background vertical viscosity background vertical viscosity 3D. Gηη ∂η (9.e. no HLES 2D.22) For small-scale flow. with HLES 3D.

when momentum is taken into account. Eqs. (9. For more complex processes e.25.27 depends on the application (velocity scales) and can be specified by you. Their numerical implementation is explicit and this introduces a stability condition. the turbulent eddy viscosity at the interface reduces to zero and the vertical mixing reduces to molecular diffusion (laminar flow).26 and 9.21 to 9. the transport of matter and heat is modelled by an advectiondiffusion equation in three co-ordinate directions. then the flow is decelerated. they increase the production of turbulence and generate vertical mixing of momentum. In Delft3D-FLOW. and coastal seas often transport dissolved substances.and η -direction. the curvature of the grid in the horizontal and vertical plane has been neglected. Source and sink terms are included to simulate discharges and withdrawals. 9. the effect of internal waves may be assigned to a constant (in space and time) background eddy viscosity at input. Also first-order decay processes may be taken into account.28) (9.27) where the eddy viscosity has been assumed to be a constant.25 are used in combination with rough side walls (partial slip). If U 9. (9. salinity and/or heat. respectively. A first-order decay process corresponds to a numerical solution which is exponentially decreasing.29) ˆ and V ˆ the velocity components of the momentum discharged in ξ .3. In strongly stratified flows. User Manual viscosity terms in Delft3D-FLOW are then reduced to the Laplace operator along grid lines: Fξ = νH 1 Gξξ ∂2u + Gξξ ∂ξ 2 1 Gηη ∂2u Gηη ∂η 2 . The transport equation here is formulated in a conservative form in orthogonal curvilinear 200 Deltares .g. or Eqs. which is added to the momentum equations only. 9. estuaries. The choice between the formulations represented by Eqs. These internal waves are generated continuously by various sources. Because the effect of internal waves is not explicitly taken into account in the present implementation of the turbulence models.26) Fη = νH 1 Gξξ ∂2 v + Gξξ ∂ ξ 2 1 Gηη ∂2 v Gηη ∂ η 2 . Discharge or withdrawal of momentum The discharge of water. This is physically not realistic because in strongly stratified flow random internal waves exist. eutrophication.2 Transport equation (for σ -grid) The flows in rivers.Delft3D-FLOW.21 to 9. biological and/or chemical processes the water quality module D-Water Quality should be used. with U ˆ and V ˆ are set to zero. In the expressions for the horizontal viscosity terms. 9. see Chapter 10. gives an additional term in the U and V momentum equation: ˆ − u) Mξ = qin (U ˆ − v) Mη = qin (V (9.

the horizontal eddy diffusivity should also include dispersion by the vertical variation of the horizontal flow (so-called Taylor-shear dispersion). the horizontal eddy diffusivity DH exceeds the vertical eddy diffusivity DV .4). The that in Equation 9. the diffusion tensor is an-isotropic. For vertical mixing.32 a user defined diffusion coefficient for the horizontal is used DH reason is that you may use this horizontal diffusion coefficient for calibration independent of the horizontal eddy viscosity.31) (9. 2 The contribution due to 3D turbulence. in case of a constant vertical eddy viscosity) or computed by a 3D-turbulence model. which is related to the turbulent eddy viscosity and is computed by a 3D turbulence closure model. Delft3DFLOW has an option (so-called HLES approach) solving explicitly the larger scale horizontal Deltares 201 . (9. back representing the Reynolds-averaged equations and/or accounting for other unre3 DH solved horizontal mixing. The total horizontal diffusion coefficient DH is defined by back DH = DSGS + DV + DH (9. Typically.g. For depth-averaged simulations. (1992). see Uittenbogaard et al. The horizontal diffusion coefficient is assumed to be a superposition of three parts: 1 The 2D part DSGS (sub-grid scale turbulence).Conceptual description co-ordinates in the horizontal direction and σ co-ordinates in the vertical direction: ∂ (d + ζ ) c + ∂t d+ζ Gξξ Gηη 1 Gξξ ∂ ∂ξ Gηη DH ∂ ∂ Gηη (d + ζ ) uc + ∂ξ + ∂ ∂η DH DV Gξξ (d + ζ ) vc ∂η + + ∂ωc = ∂σ Gηη ∂c Gξξ ∂ξ + Gξξ ∂c Gηη ∂η ∂c ∂σ 1 ∂ d + ζ ∂σ − λd (d + ζ ) c + S. DV .32) with DSGS the diffusion due to the sub-grid scale turbulence model (Equation B. due to motions and mixing that are not resolved by the horizontal grid. the eddy viscosity νV is defined by you at input via the GUI (e. Delft3D-FLOW will also compute the vertical diffusion coefficients by taking into account the turbulence Prandtl-Schmidt number. We remark back . see Equation 9.33) with D3D the diffusion due to turbulence model in vertical direction (Equation 9. The vertical diffusion coefficient DV is defined by DV = νmol back + max D3D . For shallow-water flow. D3D .20. DV the horizontal diffusion coefficient.30) with DH the horizontal diffusion coefficient. σmol (9. This eddy diffusivity depends on the constituent. λd representing the first order decay process and S the source and sink terms per unit area due to the discharge qin or withdrawal qout of water and/or the exchange of heat through the free surface Qtot : S = (d + ζ ) (qin cin − qout c) + Qtot .103) and with νmol the kinematic viscosity of water and σmol is either the (molecular) Prandtl number for heat diffusion or the Schmidt number for diffusion of dissolved matter.

which is the case in most of the simulations. Then the 2D part DSGS is associated with mixing by horizontal motions and forcing that cannot be resolved on the grid.5 for a more detailed description.33 DV back should For two-dimensional depth-averaged simulations. For details of the HLES-approach.3 presents an overview of these options. If you have not switched on the Ozmidov length scale.2L2 oz − g ∂ρ ρ ∂z . then the Ozmidov length scale is equal to zero. back and it is specified by The horizontal background diffusion coefficient is represented by DH you at input and it should account for all other forms of unresolved mixing (in most cases it is a calibration parameter). Instead. This part is discarded by depth averaging the 3D advection diffusion equation. The default value of 10 m /s is too large for small scale applications. In strongly stratified flows.Delft3D-FLOW. The vertical eddy diffusivity D3D is the maximum value of the Ozmidov length scale and the vertical eddy viscosity computed by the turbulence model. For the application of a depth-averaged transport model.2). We note that this is rather similar to the eddy viscosity options (see Table 9. Artificial vertical diffusion The σ co-ordinate system rotates the Cartesian stress tensor and introduces additional derivatives (Stelling and Van Kester. In addition to the vertical eddy diffusivity estimated by a 3D turbulence model you may specify a background or “ambient” vertical mixing coefficient accounting for all other forms of unresolved mixing. 1985) and that the flow is of boundary-layer type. Remarks: back and preThe horizontal background diffusion-type coefficient is represented by DH 2 scribed by you in the GUI. In Delft3D-FLOW the diffusion tensor is redefined in the σ co-ordinate system assuming that the horizontal length scale is much larger than the water depth (Blumberg and Mellor. you can specify the so-called Ozmidov length scale Loz .101): D3D = max D3D . Remark: We remark that for small scale applications the default value set by the GUI of 10 m2 /s is too large. Table 9. 0. Then.34) For all turbulence closure models in Delft3D-FLOW a vertical background mixing coefficient back has to be specified by you via the GUI. the water column should be wellmixed and the Taylor shear dispersion should be of secondary importance. see Section B. the horizontal eddy diffusivity DH also contain a contribution due to the vertical variation of the horizontal flow (Taylor shear dispersion). (9.8. 1994). Equation 9. in Delft3D-FLOW many options are available for the eddy diffusivity. see Section 9. breaking internal waves may be generated which are not modelled by the available 3D turbulence models of Delft3D-FLOW. otherwise a 3D model should be used for the simulation of the transport of matter. Summarizing. the 3D-turbulence part D3D is defined as the maximum of the vertical eddy diffusivity computed by the turbulence model and the Ozmidov length scale (see also Equation 9. User Manual turbulent motions. The horizontal 202 Deltares .

back DV 2D-turbulence + dispersion coefficient 3D-turbulence + dispersion coefficient 2D-turbulence background eddy diffusivity 3D. with HLES 3D.Outfall combination are added to the values at the intake.e.38) where the increase in temperature. i. salinity or concentration due to the power plant can be negative. Deltares 203 .3: Overview of eddy diffusivity options in Delft3D-FLOW model description 2D. salinity and concentrations of substances prescribed as a function of time for the Intake .3 Coupling between intake and outfall For a power plant. 9. (9. For steep bottom slopes in combination with vertical stratification.37) (9. Strict horizontal diffusion along Z -planes gives a more realistic description of the physical transport process. Remark: The temperature..Conceptual description Table 9. also on the numerical grid (Beckers et al. no HLES DSGS 0 computed by HLES 0 back DH (represents) D3D maximum of value of turbulence model and the Ozmidov length scale. Numerical implementation of the complete transformation behaved poorly (Blumberg and Mellor. i.35) The temperature. the horizontal diffusion along σ -planes introduces artificial vertical diffusion (Huang and Spaulding.: qout = qin . The computation is continued. 1996). or concentration at the outfall is set to zero when its value would become negative. 1998). with HLES computed by HLES - background eddy diffusivity gradients are taken along σ -planes. This approach guarantees a positive definite operator. but a warning is written to the diagnostic file. 1985). they represent the influence of the power plant on their value or concentration: Tout = Tin + ∆Tdischarge . see Section 10. Cout = Cin + ∆Cdischarge . Stelling and Van Kester (1994) developed an algorithm to approximate the horizontal diffusion along Z -planes in a σ -co-ordinate framework. salinity.e.3. maximum of value of turbulence model and the Ozmidov length scale. the discharge at the outfall is set equal to the inflow at the intake. Sout = Sin + ∆Sdischarge . no HLES 2D. (9.36) (9.10.

25t − 0. but at low pressures the effect on density can be neglected. User Manual 9.0899 · 10 −7 3 −3 t + 7. This is an important physical aspect when simulating the thermocline formation in lakes or stagnant basins.3.47) Remarks: 204 Deltares . P0 = 5890 + 38t − 0.536332 · 10−9 t5 A = 8. Therefore also the so-called UNESCO formulation is implemented in Delft3D-FLOW.24493 · 10 −1 (9.0227 · 10 t − 1. with the salinity s in ppt and the water temperature t in ◦ C.01t) s. 0 < s < 40 ppt ρ= where: 1.120083 · 10−6 t4 + 6. 2 (9.43) ρ0 = 999.3875 · 10 −4 t −3 + 1. Eckart formulation The Eckart formulation is given by (Eckart.793952 · 10−2 t − 9.5 + 11.001685 · 10−4 t3 − 1.4 Equation of state The density of water ρ is a function of salinity (s) and temperature (t).375t + 3s.8314 · 10−4 t + 5.40) (9.80 + 0.6438 · 10 −9 4 −5 2 t + (9.39) λ = 1779.45) − 8.42) For fresh water the Eckart formulation does not give the maximum density at the temperature of 4 degrees Celsius. UNESCO formulation The UNESCO formulation is given by (UNESCO.0745t2 − (3.72466 · 10 C = 4. 1958): Range: 0 < t < 40 ◦ C. 000P0 .095290 · 10−3 t2 + + 1. In Delft3D-FLOW you can choose between two different formulations for the equation of state (Eckart or UNESCO). In the original equation the pressure is present. 0.41) (9.842594 + 6. The first empirical relation derived by Eckart (1958) is based on a limited number of measurements dating from 1910 (only two salinities at 5 temperatures).Delft3D-FLOW. α0 = 0.46) (9. λ + α0 P0 (9. 1981a): Range: 0 < t < 40 ◦ C.44) − 4.5 < s < 43 ppt ρ = ρ0 + As + Bs3/2 + Cs2 where (9.2467 · 10 B = −5.6546 · 10 −6 2 t (9.6980.

45 is known as the International Equation of State for Seawater (EOS80) and is based on 467 data points. densities derived with the methods based on practical salinity measurements and the International Equation of State may deviate measurably from the true densities. However. we then recommend to contact the Helpdesk for further assistance such as the implementation of a more dedicated density formulation.6 · 10−3 kg/m3 (Millero and Poisson. Recommendation The UNESCO formulae serve as an international standard. a variable mineral content of the water may create density differences not detected by just temperature and salinity.Conceptual description Equation 9. If you want to apply the UNESCO formulation you do not need to change the MDF-file. How to specify the equation of state The equation of state cannot yet be specified from the FLOW-GUI. Further. these equations should be used with caution in waters that have a chemical composition different from standard seawater. Therefore we recommend the application of the UNESCO formulae and to select the Eckart formulation only for consistence with previous projects in which it has been used. You have to edit the MDFfile yourself: If you want to apply the Eckart formulation. In case of deviations or other constituents determining the water density. However. since the default formulation is UNESCO. 3. you may add DenFrm = #UNESCO# if you want to. The Practical Salinity Scale (UNESCO. you are warranted to check the accuracy of the UNESCO formulae against experimental densityrelations derived from the (lake) water. Deltares 205 . 1981). The standard error of the equation is 3. 1981b. the UNESCO formulae show the correct temperature of 4 degrees Celsius where fresh water has its maximum density. However. typically lakes. In such waters.2) and the International Equation of State are meant for use in all oceanic waters. At this moment the default is UNESCO. Par. the UNESCO formulae have their limitations as they are based on the general properties of seawater mixed with fresh water. in water masses different in composition from standard seawater the differences in densities derived by the new equations involve only very small errors. add DenFrm = #Eckart#. Nevertheless. Particularly in cases where marginal density differences play a role. The latter is of importance for thermal stratification in deep lakes in moderate climate zones. For such dedicated cases.

The influence of this built-in boundary condition is in most cases restricted to only a few grid cells near the open boundary. or die out due to bottom friction and viscosity. At inflow. a set of initial and boundary conditions for water levels and horizontal velocities must be specified. The number of boundary conditions specified at any particular point of the boundary should be equal to the number of characteristics entering the region at that point. For tidal flow the number of required boundary conditions varies between ebb and flood.75.48) is smaller than one. The steady state solution depends on the boundary conditions and the forcing terms while the transient solution follows from the deviation between the initial condition and the steady state solution at the start of the simulation. which means that the magnitude of the flow is smaller than the velocity of wave propagation. The first boundary condition is an external forcing by the water level. the discharge rate or the Riemann invariant specified by you. The second boundary condition is a built-in boundary condition. Open boundaries are always artificial “water-water” boundaries. When disturbances are reflected at the open boundaries and are trapped in the model area. gH (9.Delft3D-FLOW. The velocities normal to a closed boundary are set to zero. we have to specify two boundary conditions and at outflow we have to specify one boundary condition. inflow and outflow. At inflow the velocity component along the open boundary is set to zero.4 Boundary conditions The 3D and 2D depth-averaged shallow water or long-wave equations applied in Delft3DFLOW represent a hyperbolic (inviscid case) or parabolic (viscid case) set of partial differential equations. the normal velocity. Sub-critical flow means that the Froude number. viz. 9. see Courant and Hilbert (1962) or Vreugdenhil (1989). For sub-critical flow we distinguish two situations. which is defined by: |U | Fr = √ . The reflection should be minimal. It is reached after the transient solution died out. They are situated as far away as possible from the area of interest. It would be better to specify the tangential velocity component. In the present implementation 206 Deltares . To get a well-posed mathematical problem with a unique solution. User Manual 9. Usually it is assumed that initially the model is at rest (“cold start”). Long waves propagating out of the model area should not be hampered by the open boundaries. The spin-up time of a model is dependent on the time-scale with which the transient solution dies out. In a numerical model open boundaries are introduced to restrict the computational area and so the computational effort. The solution of the shallow water equations can be split in a steady state solution and a transient solution. The start of the simulation will generate short progressive waves with a wavelength coupled with the length of the model area (eigen oscillations). The transient solution may propagate out of the model area through the open boundaries (Riemann boundary condition) or be reduced at the open boundaries (low pass filter) by use of the αcoefficient. The steady state solution is completely dependent on the boundary conditions. Delft3D-FLOW assumes that the flow at the open boundaries is sub-critical. the reflection at the open boundaries and the amount of dissipation. The contour of the model domain consists of parts along “land-water” lines (river banks. Eqs.74 and 9. The tangential velocity may be determined from measurements or from model results of a larger domain (nesting of models). A water level boundary acts as a nodal point and a velocity or closed boundary acts as an anti-nodal point. a standing wave may be generated. coast lines) which are called closed boundaries and parts across the flow field which are called open boundaries. Closed boundaries are natural boundaries.

4. or z = −d) are σ co-ordinate surfaces.Conceptual description of the open boundary conditions in Delft3D-FLOW it is not possible yet to specify the tangential velocity component at input.74 and 9. The vertical velocity profile is determined by the vertical eddy viscosity and the boundary conditions at the bed (bed stress) and at the free surface (wind stress). When the grid size near the boundaries is larger than the thickness of the boundary layers occurring in the flow. A water level or velocity boundary condition can be made weakly reflective for short wave components originating from the initial conditions or the eigen frequencies by adding the time derivative of the Riemann invariant see Eqs. In the vertical direction the momentum equations are parabolic. the free surface (σ = 0. To reduce the reflections at the open boundaries a so-called weakly reflecting boundary condition may be applied. The impermeability of the surface and the bottom is taken into account by prescribing the following kinematic conditions: ω |σ=−1 = 0 and ω |σ=0 = 0 For the Z -grid the kinematic conditions read: (9.1. If we do not prescribe exactly the incoming Riemann invariants at an open boundary. 9.1 9.49) w|z =−d = 0 and w|z =ς = 0 (9. If these Riemann invariants are not used then these short wave components may disturb the solution for a long time. the shear-stresses along the lateral boundaries can be neglected. incident waves should be known and are prescribed at the open boundaries. Usually information on both water level and velocity is not available and consequently we cannot specify the Riemann invariant. the outgoing wave will reflect at the boundary and propagate back as a disturbance into the model area. it is suggested to define the model boundaries at locations where the grid lines of the boundary are perpendicular to the flow. 1984). ω is the vertical velocity relative to the σ -plane. The boundary condition is non-reflective for waves.50) Deltares 207 . which pass normal to the boundary (wave front parallel with open boundary). A so-called free slip boundary condition is applied at all lateral boundaries. The spin-up time of the vertical profiles is dependent on the vertical eddy viscosity.1 Flow boundary conditions Vertical boundary conditions Kinematic boundary conditions In the σ co-ordinate system.75. For the simulation of laboratory flumes the effect of side wall friction can be added to the system of equations in Delft3D-FLOW by a so-called partial slip boundary condition. To get a realistic flow pattern near the open boundary. For modelling the region outside the computational domain.4. 9. This boundary condition is derived using the Riemann invariants of the linearised shallow water equations without Coriolis and bottom friction (Verboom and Slob. or z = ζ ) and the bottom (σ = −1.

7. C2D = Chezy Manning’s formulation: (9. is the Manning coefficient [m−1/3 s].52) with τbξ and τbη the components of the bed stress in ξ . Specific functionality has been implemented to make the translation of bed and 208 Deltares .54) √ 6 C2 D = where: H n (9.56) H ks is the total water depth is the Nikuradse roughness length.Delft3D-FLOW. At first we discuss the formulations for 2D depth-averaged flow. White Colebrook’s formulation: H n C2D = 18 10 log where: 12H ks (9. (9. see Section 9. followed by the formulations for 3D flow. Depth-averaged flow For 2D depth-averaged flow the shear-stress at the bed induced by a turbulent flow is assumed to be given by a quadratic friction law: ρ0 g U U τb = 2 C2 D . In this section we restrict ourselves to the resistance due to flow only. ´ The 2D-Chezy coefficient C2D can be determined according to one of the following three formulations: ´ formulation: Chezy ´ coefficient [m1/2 /s]. Alluvial bed properties and land use data can be used to determine the roughness and flow resistance. ρ0 (9. The bed stress may be the combined effect of flow and waves.51) = σ =−1 (9. For the parameterisations for the combination of flow and waves. User Manual Bed boundary condition At the seabed. ρ0 1 τbη .55) is the total water depth [m]. respectively. the boundary conditions for the momentum equations are: νV ∂u H ∂σ νV ∂v H ∂σ = σ =−1 1 τbξ .53) where U is the magnitude of the depth-averaged horizontal velocity.and η -direction.

For rough walls Nikuradse (1933) found: z0 = ks .Conceptual description land use properties into roughness and flow resistance characteristics. (9. Three-dimensional flow For 3D models. Using Eqs. The first grid point above the bed is assumed to be situated in the logarithmic boundary layer.54. The bed shear stress in 3D is related to the current just above the bed: τb3D = gρ0 ub |ub | . the bottom is positioned at z0 .58) where z0 is user-defined. equality of bottom stress and the assumption z0 H .59) √ C3 D = g ∆ zb ln 1 + κ 2z0 . The magnitude of the bottom stress is defined as: |τb | = ρ0 u∗ |u∗ | . ´ Often a 3D-calculations are preceded by depth-averaged calculations. a quadratic bed stress formulation is used that is quite similar to the one for depth-averaged computations. In the numerical implementation of the logarithmic law of the wall for a rough bottom. 2 C3 D (9. so: ub = u∗ ∆ zb ln 1 + κ 2 z0 . The contribution of the vertical velocity component to the magnitude of the velocity vector is neglected. Under the assumption of a logarithmic velocity profile Equation 9.2 for details. 9. See Section 9. −e (9. C3D can be expressed in the roughness height z0 of the bed: (9.62) ´ The Chezy coefficient C2D can be converted into the bed roughness height z0 using the relation: z0 = H e κC D 1+ √2 g . the Chezycoefficients C2D may be used for calibration of the 3D model.57) with |ub | the magnitude of the horizontal velocity in the first layer just above the bed. the magnitude of the depth-averaged velocity is given by: U = |u∗ | H ln 1 + κ ez0 . Let ∆zb be the distance to the computational grid point closest to the bed.59.57 to 9.63) Deltares 209 . 30 (9. Then.61) where ks is known as the Nikuradse roughness length scale and has been determined experimentally.60) The parameter z0 can be linked to the actual geometric roughness as a fraction of the RMSvalue of the sub-grid bottom fluctuations. (9.11. (9.

the wind speed 10 meter above the free surface (time and space dependent). Equality of the bed stress in 2D leads to the relation: √ C2D = g H ln 1 + κ ez0 . because in these cases the vertical profile will deviate a lot from the logarithmic profile Equation 9. Local winds vary both in space and time. The magnitude is determined by the following widely used quadratic expression: 2 |τs | = ρa Cd U10 (9. see Smith and Banke (1975). The magnitude of the wind shear-stress is defined as: |τs | = ρ0 u∗s |u∗s | .and the 3D-Chezy pends on the relative thickness of the computational bed layer. so also Equation 9.64 and 9. reflecting increasing roughness of the sea surface with increasing wind speed.60 deNote: that the difference between the 2D. Global wind corresponds to uniform in space and varying in time. Surface boundary condition At the free surface the boundary conditions for the momentum equations are: νV ∂u H ∂σ = σ =0 1 |τs | cos (θ) . You may 210 Deltares .67) (9.63 denotes the actual water depth. As a result the roughness height z0 depends on the horizontal co-ordinates ξ .68) where: ρa U10 Cd the density of air.Delft3D-FLOW.54. ρ0 1 |τs | sin (θ) . ρ0 (9.64) ´ coefficient 9. Without wind. Using a 2D simulation for the calibration of waterlevels is only accurate if the vertical velocity profile of the horizontal velocity is almost logarithmic. Examples are discharge waves in rivers.58 is not correct. tidal flows in estuaries and seas. User Manual The depth H in Eqs. η and time t.65) νV ∂v H ∂σ = σ =0 (9. 9.62 and 9. the stress at the free surface is zero. Local wind is applied in combination with space and time varying atmospheric pressure. Delft3D-FLOW offers the possibility to prescribe either global or local wind. the wind drag coefficient. The method is not accurate for stratified flows and wind driven flows. dependent on U10 .66) where θ is the angle between the wind stress vector and the local direction of the grid-line η is constant. The wind drag coefficient may be dependent on the wind velocity. (9.

1984. speed U10 are user-defined wind speeds (i = A. B. waves can cross these boundaries unhampered and without reflections.Conceptual description 0 Figure 9. They are introduced to obtain a limited computational area and so to reduce the computational effort. (9. 1986) derived a so-called zero and first order weakly reflecting boundary condition based on the work of Engquist and Majda (1977. Cd + Cd − Cd B − UA U10 10 Cd (U10 ) = B  C B + C C − C B U10 − U10 . 9. B. Verboom and Segal. the outgoing waves will reflect at the boundary and propagate as a disturbance into the area. also the tangential velocity component should be specified. tide tables or from a larger model. which encloses the model at hand (nesting). If we do not prescribe exactly the incoming waves at an open boundary. In nature. 1979). Assuming zero flow along the boundary. specify an empirical relation following:  A Cd . The data needed for the boundary conditions can be obtained from measurements.3: Examples wind drag coefficient depending on wind speed. the tangential velocity component is set to zero. To reduce the reflections at the open boundary (Verboom and Slob.     A   A B A U10 − U10  . For an inflow boundary.  d d d  C − UB  U10  10   C Cd .69) C B ≤U U10 10 ≤ U10 .2 Open boundary conditions Open boundaries are virtual “water-water” boundaries. the zero order boundary condition may also be obtained using the so-called Riemann invariants for the linearised 1D equation normal to the open boundary: R=U ±2 Deltares gH. U10 ≤ U10 A ≤U B U10 10 ≤ U10 . C ).70) 211 .3. At an open boundary the water level. where: i Cd i U10 A. C ).69 several relations for the wind drag coefficient are possible.4. It is assumed that the flow is normal to the open boundary. With Equation 9. the normal velocity component or a combination should be prescribed to get a well-posed mathematical initial-boundary value problem. see Figure 9. In the present implementation. (9. C ≤U . U10 10 are the user-defined wind drag coefficients at respectively the wind i (i = A.1.

d |ζ | d 1. Meteorological forcing (wind set-up).74 and 9. The boundary conditions should allow these disturbances to propagate out of the area or the number of points where the boundary condition is specified should be extended. The linearised Riemann invariant is given by: U +2 gH = U + 2 g (d + ζ ) ≈ U + 2 gd + ζ g . the barotropic forcing at the open boundaries is approximated by a linear superposition of: Free surface gradient for long-term circulation (geostrophic currents). which have an oblique incidence. On water level boundaries.75. Tidal fluctuations. The boundary condition based on the Riemann invariant is weakly reflective in 1D. it is possible to specify an average pressure (paverage ) — which should correspond to your input signal on the open boundaries — which is then used to determine 212 Deltares . To this end. Usually the signal corresponds to Mean Sea Level. Velocity (in normal direction): U = FU (t). One actually wants to prescribe an input signal corresponding to the local pressure prescribed by the space varying meteo input. We restrict ourselves to the positive sign (left boundary). Nevertheless. ∂ζ Neumann ∂n = f (t). It is assumed that the reference plane is chosen such that the mean water level is zero. (9. If measured elevation data is not available. the incoming wave should be specified.71) the term 2 gd The boundary condition which should be specified by you is: f (t) = U + ζ √ is computed from the known depth-field and added in the computational part. 9. In the computational part the following type of boundary conditions are distinguished (for the sake of simplicity only a description for the U -direction is given here): Water level: ζ = Fζ (t) + δatm . g d. User Manual The two Riemann invariants are two waves moving in opposite direction with propagation √ speed R = U ± gH . But in 2D this boundary condition is only weakly reflective for waves which pass normal to the boundary and if the Coriolis force and the bottom friction are negligible. Riemann invariant U ± ζ g d = FR (t). Usually we do not have information on both water level and velocity so we do not know the incoming Riemann invariant and another type of boundary should be chosen. the boundary conditions are specified in a limited number of boundary points.Delft3D-FLOW. At the open boundary. This interpolation can generate physical unrealistic flows in the region close to the open boundary. Linear interpolation is used to generate the boundary conditions at the intermediate points along the boundary. A water level or velocity boundary condition can be made weakly reflective for short wave components originating from the initial conditions or the eigen frequencies by adding the time derivative of the Riemann invariant see Eqs. The sign is dependent on the direction of propagation. In general. Discharge (total and per cell): Q = FQ (t). an input signal is prescribed that is consistent with some average pressure. in practice it reduces the reflection also for waves. Waves (wave set-up).

This reduces the spin-up time of a model from a cold start: Water level boundary: ζ + α ∂ U ±2 ∂t gH = Fζ (t) .2. This is a global flag. For salinity and heat the concentration boundary conditions influence the flow by the baroclinic pressure term. δatm = where paverage − patm ρg (9. using PavBnd: Average Pressure on Boundaries. the flow is assumed to be perpendicular to the open boundary.73) with d the local water depth.72) paverage patm ρ g average pressure prescribed with the keyword PavBnd. the local atmospheric pressure. and is calculated as follows: FR (t) = FR (t) ± paverage − patm ρg g d (9. The baroclinic pressure is switched off for all open boundaries. to make the boundaries less reflective for disturbances with the eigen frequency of the model area.Conceptual description local pressure gradients that need to be applied along the open boundaries to obtain an input signal that is consistent with the local atmospheric pressure.4. This functionality need to be specified in the Data Group Additional parameters. The following values are advised: Water level boundary: α = Td H . [s2 ] g (9. and the concentration should be free at outflow. discharge and Riemann type of boundary condition. given by the meteo module (see Section B.74) Velocity boundary: U + α ∂ U ±2 ∂t gH = FU (t) . Using this keyword one can specify an average pressure that is used on all water level open boundaries. For the velocity. concentrations should be specified at inflow.76) Deltares 213 . this may lead to strange circulation cells along a long open boundary. and the sign depends on the orientation of the open boundary related to the grid. At slack water. Substitution of the Riemann boundary condition in the 1D linearised continuity equation leads to the well-known radiation boundary condition. The default value is #YES#.75) The reflection coefficient α should be chosen sufficiently large to damp the short waves introduced at the start of the simulation. (9. independent of the type of wind input.1. Therefore an additional flag (keyword) has been implemented to switch off the baroclinic pressure term at the open boundary BarocP = #NO#. The pressure must be specified in N/m2 . density of water acceleration due to gravity The same correction can also be used on Riemann boundaries. Stelling (1984) added the time-derivative of the Riemann invariant to the water level and velocity boundary conditions. see Section 9. when the barotropic pressure gradients are small. (9.7). For the transport equation.

(9. discharge. These values can be derived with Fourier analysis for the 1D linear long wave equation without advection by substituting a wave with period Td .78) −H The velocity of layer k . A logarithmic profile may be applied for velocity. with (for σ -grid): zk = (1 + σk−1 ) H. H and z0 . This file can be generated with the utility program Delft3D-NESTHD. see Section 4. In ocean and sea models. as harmonic components. i. time-series or astronomical components. In 3D models you can prescribe: a uniform profile a logarithmic profile or a so-called 3D profile in the vertical.e.Delft3D-FLOW.6 for details. For the velocity and discharge type of boundary condition. The vertical co-ordinate z is measured relative to the bed. User Manual Velocity boundary: α = Td [s] (9. the flow is assumed to be perpendicular to the open boundary. is specified as: uk = u∗ zk−1 κ (zk−1 − zk ) zk u∗ = (zk−1 + z0 ) ln κ (zk−1 − zk ) ln z + z0 z0 dz. uk . Remark: A 3D-profile can only be prescribed by importing a file containing the boundary condition at each layer. The depth-averaged velocity U is used to determine the friction velocity: u∗ = U 1 κH H 0 ln z + z0 z0 = dz U 1 κH (H + z0 ) ln H +z0 z0 (9. z k −1 + z 0 z0 − 1 − (zk + z0 ) ln z k + z0 z0 (9. In that case α must be set to zero. the period Td is of the same order as the period of the tidal forcing. 214 Deltares .77) where Td is the time it takes for a free surface wave to travel from the left boundary to the right boundary of the model area. This profile is described for a velocity boundary condition in direction perpendicular to the open boundary. or Riemann boundaries. otherwise effectively the amplitude of one of the components in the boundary condition is reduced. The logarithmic profile depends on the local roughness height z0 and the local water depth H .79) −1 .80) After substitution of u∗ in this equation the layer velocity uk is expressed in terms of U .5. A 3D-profile means that the velocity at each σ -layer is specified as any of the forcing types.

For the boundary conditions to match this distribution the solution has to be known beforehand. if not. in case of a total discharge boundary condition. depth averaged velocity and water depth of grid cell i.k ∆zi. u ¯i and hi are the width. One of the main problems is the specification of suitable boundary conditions at the open boundaries. which is imposed uniformly at the boundary. water depth and roughness of grid cell i respectively. only in tidal cases or in cases where a storm surge travels along a coast the alongshore gradient varies in time. and Q is the total discharge imposed. hi and Ci are the width. the alongshore gradient of the water level does not vary much in cross-shore direction. groynes. For simple cases this is possible but for more complex combinations of forcing conditions it is cumbersome. a certain water level or velocity distribution will develop in cross-shore direction. boundary disturbances will develop. k and ∆zi. In case a QH-relation is used at a water level boundary the total discharge is computed from the discharges per grid cell N K N Q= j =1 Bi k=1 ui. The alongshore water level gradient can be assumed to be zero in case of a steady wind/wave field. N is the number of boundary points and K is the number of layers.5 B i h1 i Ci N 1. (9. In the latter case. ui .82) where Bi . There are two ways to overcome this problem. the alongshore gradient varies in time at the boundary.k = j =1 Bi u ¯ i hi (9.81) where Bi . A constant periodical water level gradient for a boundary can be applied to models with a limited cross-shore extent.Conceptual description The discharge is. The total discharge Q is used to derive a water level from the QH-relation specified by you. Neumann boundary conditions Application of 2D and 3D hydrodynamic models to study the impact of engineering solutions such as the design of nourishments. but in model with a limited cross-shore extent. at each time step distributed ´ over the active points of the open boundary in the following manner derived from the Chezy equation: qi = . offshore breakwaters on approximately uniform coasts is often hampered by the fact that they are considered to be difficult to set up and difficult to calibrate. A better solution suggested by Roelvink and Walstra (2004) in the paper ‘Keeping it simple by using complex models’ is to let the model determine the correct solution at the boundaries by imposing the alongshore water level gradient (a so-called Neumann boundary condition) instead of a fixed water level or velocity. Neumann boundaries can only be applied on cross-shore boundaries in combination with a Deltares 215 . In many cases the gradient can be assumed to be zero. The first is to try and predict the water level setup or the current velocity along the lateral boundary by solving a 1DH or 2DV problem along the boundary and to impose this. N is the number of boundary points. or vary periodically for a tidal wave travelling along the coast.5 j =1 Bj hj Cj Q. under the assumption that the alongshore gradient of the water level does not vary much in cross-shore direction.k are the velocity and thickness of layer k at grid cell i in case of a 3D simulation. This is due to a combination of processes acting on the model domain.

4). with cj the velocity of the tidal wave. Let dAB where: ζ Tj Lj be the distance between point A and B. kj = 2π wave number [rad/m]. User Manual B A A' B' Figure 9. Defining boundary conditions with Neumann boundaries Consider the example below.4). denoted as Neumann Boundary ∂ζ = f (t).4. describing a small coastal region with a tidal wave travelling along the coast in eastward direction. which is needed to make the solution of the mathematical boundary value problem well-posed (see Figure 9. B–B’ in Figure 9. ωj = 2π frequency [rad/h]. The phase difference (ϕj ) between points A and B is 2π then given by kj dAB = L dAB (to get degrees multiply it by 180 π ). the water level gradient as a function of time.84) In Table 9. The offshore boundary is forced by a harmonic/astronomic water level according to a progressive wave given by: N N ζ= j =1 ˆj cos(ωj t − kj x) = ζ j =1 ˆj cos(ωj t − ϕj ) ζ (9.4: Hydrodynamic model of coastal area with three open boundaries with offshore boundary (A–B at deep water) and two cross shore boundaries (A–A’. For the forcing of the two cross-shore boundaries a gradient (Neumann) type of boundary condition is used. At the offshore boundary (A–B).4 we specify the relation between the tidal forcing at the offshore boundary (water level type) and the forcing at the cross-shore boundaries (Neumann type) for the coastal model of Figure 9. the water level has to condition will be imposed: ∂x be prescribed to make the solution fully determined. On the cross-shore boundaries (sections A–A’.Delft3D-FLOW. Lj is the tidal wave length j Lj = 2π ωj cj .83) ˆj = amplitude [m]. and B– B’) water level boundary at the offshore boundary. At the water level boundary we set the reflection coefficient α to zero. The forcing term f (t) is related to the offshore forcing by the relation: ∂ζ = ∂x N N ˆj sin (ωj t − kj x) = kj ζ j =1 j =1 ˆj cos ωj t − ϕj + π kj ζ 2 (9. For shallow water the velocity is independent of the frequency and is given by √ gH with H a characteristic water depth. 216 Deltares .

Roelvink and Walstra (2004) or be approxiπ√ mated using Lj = 2 ωj gH with H a characteristic depth along the offshore boundary. Otherwise the baroclinic pressure gradient can causes severe water level gradients.4. For each tidal component. B–B’: [-] Neumann boundaries A–A’. phases and amplitudes for the cross-shore Neumann boundaries WL–boundary A–B Frequency [◦ /h] Phase [◦ ] Amplitude A–B: [m] A–A’. the boundary condition of these quantities should match their initial value. phase difference between boundaries A and B and amplitude. Advection terms at the offshore boundary may generate an artificial boundary layer along the boundary. the corresponding component of water level gradient can be determined on the basis of its frequency. phases and amplitude on alongshore waterlevel boundary and corresponding frequenties. the phase difference between the water level and Neumann boundary is π/2 (≡ +90◦ ). the boundary conditions often do not match with the initial conditions. B–B’ ωj (A) = ωj (B ) ϕj (A) = 0◦ ϕj (B ) = ϕj ˆj (A) = ζ ˆj (B ) ζ ωj (A) = ωj (A ) = ωj (B ) = ωj (B ) ϕj (A) = ϕj (A ) = ±90◦ ϕj (B ) = ϕj (B ) = ±90◦ + ϕj ∂ζ ∂x j (A) = ∂ζ ∂x j (B ) = ∂ζ ∂x j (A ) ∂ζ ∂x j (B ) = = 2π ˆ Lj ζj 2π ˆ Lj ζj General remarks on open boundaries At the start of the simulation. the phase difference is −π/2 (≡ −90◦ ). The smoothing time is the last moment at which interpolation takes place. 2π The tidal wave length Lj may be estimated from the phase difference ∆ϕAB = L dAB j between tidal stations along the boundary. Deltares 217 . If the tidal wave component is travelling in positive x-direction. The advection terms containing normal gradients have to be switched off. radians should be used. To define the phase difference in Delft3D-FLOW the unity degrees is used. If the tidal wave component is travelling in negative x-direction. This is done by adding an additional keyword to the MDF-file: Cstbnd =#YES# When modelling salinity and/or heat transport. These larger water level gradients generates waves which will propagate into the domain and enlarging the spin-up time of the model. to reduce the spin up time of the model.4: Frequencies. see Table 9. especially on deep water. The Neumann boundaries are interpreted as gradient type of boundaries for water level. in Delft3D-FLOW at the open boundaries a linear interpolation is applied between the actual field value and the boundary condition. To determine the amplitude of the water level gradient.Conceptual description Table 9. Therefore.

2 9.1. which does not completely fix the 218 Deltares . the tangential shear stress is calculated based on a logarithmic law of the wall: τξη = τηξ = ρu2 ∗. has a wave character. As described before. estuaries. two boundary conditions have to be prescribed.4. At a closed boundary. For the shear stress along the boundary one of the following conditions can be prescribed: 1 Zero tangential shear-stress (free slip). a special boundary condition based on the concentration in the interior area in combination with a return time is used. For large-scale simulations.Delft3D-FLOW. This option has been implemented for closed boundaries and dry points but not for thin dams. Along the side walls. and coastal seas is dominated by advection.4. The consequence is that the transport equation is advection dominated. User Manual 9.86) 9. the influence of the shear-stresses along closed boundaries can be neglected. with side wall roughness length z0n and the velocity in the first grid point near the side wall.1 Transport boundary conditions Open boundary conditions for the transport equation The horizontal transport of dissolved substances in rivers. open boundaries have to limit the computational area. Because usually only the background concentration is known from measurements or from a larger model area.2. and from a mathematical point of view is of hyperbolic type. The boundary condition for flow normal to the boundary is: No flow through the boundary. experiences a constant concentration. Without diffusion. One boundary condition has to do with the flow normal to the boundary and the other one with the shear-stress along the boundary. laboratory scale). (9. the influence of the side walls on the flow may no longer be neglected. an observer moving with the flow. 2 Partial slip.85) The friction velocity u∗ is determined by the logarithmic-law for a rough wall. During outflow (ebb) the concentration must be free. Let ∆xs be the grid size normal to the sidewall.g.4. Free slip is then applied for all closed boundaries. The horizontal diffusion in flow direction is of secondary importance. Then: |usidewall | = u∗ ∆xs ln 1 + κ 2z0n . The reason is that the shear stress at a thin dam is dependent on the side (double valued). At inflow we have to specify the concentration which may be determined by the concentration at outflow of the former ebb period. which is parallel to the flow. For simulations of smallscale flow (e. (9. At an open boundary during inflow (flood) a boundary condition is needed.3 Shear-stresses at closed boundaries A closed boundary is situated at the transition between land and water. The equation has only one characteristic.

87) ck = cbottom + d + zk d+ζ (csurface − cbottom ) (9. The intermediate values are then determined using linear interpolation: ck = cbottom + (1 + σk ) (csurface − cbottom ) or similarly for the Z -grid: (9. The horizontal transport is advection dominated and the equation is of hyperbolic type.4.91) 219 .88) 3 Step profile The concentration value must be specified both for the surface layer and the bed layer. cbottom . This is the so-called Thatcher-Harleman boundary condition (Thatcher and Harleman. then the position of the jump will be determined using the water elevation read from this file. no boundary condition is allowed. Gξξ ∂ξ (9.or Z -layer using a time-series: ck = Fc (t) . The layer index of the discontinuity is kept fixed during the simulation. zk > zd . sediment.2 Thatcher-Harleman boundary conditions The transport of dissolved substances such as salt. (9. 4 3D profile For a nested model Delft3D-FLOW offers the possibility to prescribe the concentration at every σ . At inflow. The concentration for an intermediate layer is determined following: ck = csurface . may occur. the discontinuity is moving with the free surface. 2 Linear profile The concentration value must be specified at the surface and near the bed. one of the following four vertical profiles may be specified at open boundaries: 1 Uniform profile The concentration value specified is applied uniformly over the vertical. (9.90) 9. and heat is described by the advection-diffusion equation. At outflow. The concentration is determined by pure advection from the interior area: ∂C + ∂t Deltares U ∂C = 0. which is discussed in the next section. In Delft3D-FLOW. just inside the open boundary. Remark: If the initial conditions are read from a restart file (hot start). zk ≤ z d . following the position of the corresponding σ -plane. 1972). In case of σ -grid. a mismatch between your boundary condition and the concentration distribution in the vertical.Conceptual description concentration at the background value.2. one boundary condition is needed and the concentration is specified.89) The zd co-ordinate specifying the position of the discontinuity is used only initially to determine the index k of the related layer. together with the initial vertical position of the discontinuity. This mismatch can produce oscillations in the simulation results. Consequently.

The functional relationship describing the variation in concentration from the slack-water value to the present value is arbitrary. 1972). This depends on the refreshment of the water in the boundary region. For the intermediate layers the return time is determined using linear interpolation: TTHk = TTHbottom + (1 + σk ) (TTHsurface − TTHbottom ) .5: Illustration of memory effect for open boundary In Delft3D-FLOW the dispersive fluxes through the open boundaries at both inflow and outflow are zero: DH ∂C = 0. (9. Thatcher and Harleman. The transition of the concentration at the boundary from the outflow value to the inflow value may take some time. the boundary value remains constant until outward flow begins (Leendertse and Gritton. The mathematical formulation of this memory effect is given as follows: 1 Tret − tout C (t) = Cout + (Cbnd − Cout ) cos π 2 Tret +1 . or similarly for the Z -grid: (9. 0 ≤ tout ≤ Tret . 1971. there is a discontinuity in the concentration at the turn of the flow. the concentration will remain Cbnd .Delft3D-FLOW.5. During the interval 0 ≤ tout ≤ Tret the concentration will return to the background concentration Cbnd .94) TTHk = TTHbottom + 220 d + zk d+ζ (TTHsurface − TTHbottom ) (9.93) where Cout is the computed concentration at the open boundary at the last time of outward flow.95) Deltares . Gξξ ∂ξ (9. When the flow turns inward (tout = 0). In Delft3D-FLOW. tout is the elapsed time since the last outflow and Tret t is the constituent return period. the concentration is set equal to C out . After the return time. Cbnd is the background concentration described by you. a half-cosine variation is used. For a stratified flow the return time of the upper layer will be longer than for the bottom layer. After that period. Delft3D-FLOW offers the opportunity to prescribe return times for the surface layer and the bed layer. The transition time is called the return time. The mechanism is illustrated in Figure 9.92) If the concentration at outflow differs from the boundary condition at inflow. User Manual Figure 9.

The mathematical formulations for the heat exchange at the free surface are given in Section 9. respectively in Z . 9. local depth value Therefore an additional flag (key word) has been implemented to switch off the baroclinic pressure term at open boundaries: BarocP = #NO#.4. The default value is #YES#. return time surface layer. σ k d return time bed layer. Usually the grid (horizontal and/or vertical) is too coarse and the time step too large to resolve the turbulent scales of motion.8: DV ∂c H ∂σ DV ∂c H ∂σ = 0. νH νV and DH DV . The 3D part is referred to as the three-dimensional turbulence and is computed following one of the turbulence closure models. The horizontal eddy viscosity coefficient νH and eddy diffusivity coefficient DH are much larger than the vertical coefficients νV and DV . i.e.3 Vertical boundary conditions transport equation The vertical diffusive flux through the free surface and bed is zero with exception of the heat flux through the free surface. Deltares 221 . back and eddy diffusivity coefficient D back The background horizontal viscosity coefficient νH H (constant or space-varying) must be specified by you via the Delft3D-FLOW GUI. σ =0 (9. vertical position. (1992) and The 2D part is associated with the contribution of horizontal motions and forcings that cannot be resolved (“sub-grid scale turbulence”) by the horizontal grid (Reynolds averaged or eddy resolving computations). described in this section. The horizontal coefficients are assumed to be a superposition of three parts: 1 a part due to molecular viscosity. For 3D shallow water flow the stress and diffusion tensor are an-isotropic. The primitive variables are spaceand time-averaged quantities. HLES for 2D-turbulence (see Appendix B.5 Turbulence Delft3D-FLOW solves the Navier-Stokes equations for an incompressible fluid. Filtering the equations leads to the need for appropriate closure assumptions.8). σ =−1 (9.Conceptual description with: TTHbottom TTHsurface zk .96) = 0. The baroclinic pressure is switched off for all open boundaries.97) 9.2. For 2D depth-averaged simulations. the horizontal eddy viscosity and eddy diffusivity coefficient should also contain a contribution due to the vertical variation of the horizontal flow (Taylor shear dispersion).and σ co-ordinate system. 2 a part due to “2D-turbulence”. Delft3DFLOW additionally has a sub-grid scale model. 3 a part due to “3D-turbulence” see Uittenbogaard et al. The turbulent processes are “sub-grid”. This is a global flag.

defined by: Ri = ρ 222 −g ∂ρ ∂z ∂u 2 ∂z + ∂v 2 ∂z . In case of vertical density gradients. κ ≈ 0. Furthermore the horizontal coefficients are an order of magnitude larger than the vertical coefficients determined by the turbulence closure model specified by you. four turbulence closure models have been implemented to determine νV and DV : 1 2 3 4 Constant coefficient. cµ = 0. We remark that a constant eddy viscosity will lead to parabolic vertical velocity profiles (laminar flow). For a homogeneous flow this leads to a logarithwith κ the Von Karm mic velocity profile. This model uses analytical (algebraic) formulas to determine k and L. On the other hand high velocity gradients. Stratification stability can be described in the interaction between gravitational forces and turbulent shear production.99) ´ an ´ constant. (9. the turbulent exchanges are limited by buoyancy forces and the mixing length L of Equation 9. and/or the mixing length L.100) Deltares . 1984). This so-called zero order closure scheme is a combination of two algebraic formulations.98) where: cµ L k a constant determined by calibration. In Delft3D-FLOW. The turbulence closure models differ in their prescription of the turbulent kinetic energy k . 1/4 The algebraic eddy viscosity model (AEM) does not involve transport equations for the turbulent quantities.Delft3D-FLOW. the dissipation rate of turbulent kinetic energy ε. H (9. The eddy viscosity has the following form: √ ν3D = cµ L k. Ri. The eddy viscosity is related to a characteristic length scale and velocity scale. The first turbulence closure model is the simplest closure based on a constant value which has to be specified by you. L. and is the turbulent kinetic energy. the following function of the depth is taken (Bakhmetev. k -ε turbulence closure model. The stratification is characterised by the gradient Richardson number.41. The other three turbulence closure models are based on the so-called eddy viscosity concept of Kolmogorov (1942) and Prandtl (1945). 1932): L = κ (z + d) 1− z+d . derived from the empirical constant cµ in the k -ε model. The turbulent kinetic energy k depends on the (friction) velocities or velocity gradients and for the mixing length. increase turbulent mixing and can weaken the stratification.09 (Rodi. User Manual These coefficients can also be used to damp small-scale noise introduced by the advection terms. They must be chosen dependent on the grid size. Algebraic Eddy viscosity closure Model (AEM). (9. k -L turbulence closure model. is the mixing length. cµ = cµ .99 must be corrected.

0.25 (1 − 14Ri) . H (9. In Delft3D-FLOW the algebraic eddy viscosity model (AEM) is extended to stratified flows by the formulation of Busch (1972): FL (Ri) = e−2. The k -ε model has been used in tidal simulations by Baumert and Radach (1992) and Davies and Gerritsen (1994). The mixing length is now a property of the flow. the effect of the orbital velocities on the production of turbulence is modelled indirectly by the enhancement of the bed roughness.101) These damping functions have been determined by fitting mathematical functions which fulfil the limiting conditions to laboratory data sets. However.Conceptual description see e. including damping functions. is used as for the AEM turbulence model. (9.. The fourth model is the k -ε model that is a second order turbulence closure model. Miles (1987).3Ri . A brief description of each of these turbulence closure model will be given further on in this section. Different formulations have been suggested.102) The third closure model for the eddy viscosity involves one transport equation for k and is called a first order turbulence closure scheme. Stable stratification leads to damping of turbulent mixing while unstable stratification leads to higher mixing. The turbulence closure models do not take into account the vertical mixing induced by shearing and breaking of short and random internal gravity waves. Taylor (1931). Mathematically this is described by a so-called damping function FL (Ri) that depends on the gradient Richardson number Ri (Simonin et al. for stratified flow of a river plume by Postma et al. For Ri ≥ 0 the stratification is stable whereby for Ri < 0 the stratification is unstable. This internal-wave-induced mixing can be determined using an additional transport equation for Internal Wave Energy (IWE-model) (Uittenbogaard and Baron. The mixing length L is prescribed analytically and the same formulation.7. to find the kinetic energy k . Uittenbogaard. 1995). (1992).g. For strongly stratified flows it is important to introduce suitably chosen constant ambient (background) mixing coefficients. For the combination of flow and waves. From k and ε the mixing length L and viscosity are determined. see Section 9. The IWE model describes the energy transfer from internal waves to turbulence as well as the excitation of internal waves by turbulence. In this model both the turbulence energy k and dissipation rate of turbulent kinetic energy ε are calculated by a transport equation. a transport equation is solved. Ri ≥ 0. (1999) and for the evolution of a thermocline by Burchard and Baumert (1995). This turbulence closure model is known as the k -L model. and in the case of stratification no damping functions are needed. The simplest approximation of the effect of internal waves is the introduction of constant vertical ambient mixing coefficients of momentum ν . and/or heat and matter D . Ri < 0. Richardson (1920). for more details we refer to Uittenbogaard et al. because the mixing coefficients computed by turbulence models Deltares 223 . 1989. The shear production is determined only from the wave-averaged velocities. 1989): L = κ (z + d) 1− z+d FL (Ri) .

regardless the turbulence closure model.3. and for the transport of turbulent kinetic energy dissipation ε in the k -ε model σc0 = 1. For large values of the density gradients and small values of the velocity gradients.100.Delft3D-FLOW. User Manual with shear production only. the vertical diffusion equation becomes mathematically ill-posed Barenblatt 224 Deltares . A physical adjustment time of the turbulence to the variations of the vertical gradients. It is a constant. Its numerical value depends on the substance C .103) Parameter σc is the Prandtl-Schmidt number. (9.102 and 9. for small temperature gradients the heat flux increases when the temperature gradient increases but for large temperature gradients the heat flux decreases because the vertical eddy diffusivity is damped. In the latter situation the vertical layers are completely de-coupled (frictionless).0. In case the algebraic turbulence closure model is applied. it is affected by the stratification. σc (9. For suspended sediment concentrations in online sediment transport computations. and tracer concentrations. This value corresponds with field measurements in the Rotterdam Waterway.7 for the transport of heat. For the transport equation of heat. In Delft3D-FLOW the following settings of σc0 and Fσ (Ri) are used: In all cases. see Simonin et al. (9. the damping function Fσ (Ri) is the following Munk-Anderson formula (Munk and Anderson. Ri ≥ 0 Fσ (Ri) = 1 + 10Ri  1. For the transport of turbulent kinetic energy k in the k -L model and k -ε model σc0 = 1.104) The σc0 in this expression depends on the substance.105) The damping functions Eqs. Based on our experience with highly stratified flows we suggest applying an ambient or background vertical eddy viscosity in the order of 10−4 m2 /s for the vertical exchange of momentum. In the mathematical formulation. σc0 = 1. the fluxes are instantaneously influenced by changes in the vertical gradients of velocity and density. In general the expression for the Prandtl-Schmidt number reads: σc = σc0 Fσ (Ri). for zero order models (for example. reduce to zero. Fσ (Ri) is a so-called damping function that depends on the density stratification via the gradient Richardson’s number Ri defined in Equation 9.33Ri)1.5  (1√ . Eddy diffusivity The vertical eddy diffusivity is a scaled form of the eddy viscosity according to: D3D = ν3D .0. 9. the algebraic turbulence closure model). The fluxes are not a monotone function of the gradients. It slightly widens the applicability of an algebraic model for weakly stratified flows. salinity. σc0 = 0. is not taken into account. Moreover. Ri < 0. The Netherlands. (1989). 1948):  + 3. Disturbances are hardly damped and also the erosion of the vertical stratification is reduced to molecular diffusion.105 were obtained by measurements of steady-state stable stratified turbulent shear flows.

Thus: Fσ (Ri) = 1 for all Ri. (9. In Delft3D-FLOW for stable stratified flows.2L2 oz − g ∂ρ ρ ∂z . In the k -ε turbulence model no damping function is used since the influence of stratification on the mixing length is taken into account by the buoyancy terms in the transport equations for k and ε. the minimal eddy diffusivity may be based ¨ al ¨a ¨ frequency of internal on the Ozmidov length scale Loz . a logarithmic velocity profile is assumed. 225 . The combination of the two models was made to broaden the applicability of the algebraic turbulence closure. The numerical scheme for the vertical advection of heat and salt (central differences) may introduce small vertical oscillations. the turbulent eddy viscosity at the interface reduces to zero and the vertical mixing reduces to molecular diffusion.Conceptual description et al.1. 9. calibrated for local-equilibrium shear layers (Rodi. 0. we suggest to apply an ambient eddy diffusivity in the order of 10−4 to 10−5 m2 /s dependent on the Prandtl-Schmidt number. Both models will be briefly described. may generate unphysical “stepwise” profiles as observed by Deleersnijder and Luyten (1994). These models will be denoted by ALG (Algebraic closure model) and PML (Prandtl’s Mixing Length model). 9.5. This computational noise may enhance the turbulent mixing. and the computed vertical profiles may become discontinuous (stepwise). (1992).5.106) For a detailed description of the turbulence closure models of Delft3D-FLOW we refer to Rodi (1984) and Uittenbogaard et al. This leads to a linear relation between the turbulent kinetic energy at the bed and the turbulent kinetic energy at the free surface: 1 k=√ cµ with: ub ∗ 2 1− z+d H + u2 ∗s z+d . This is one of the main advantages of the k -ε turbulence model. In strongly-stratified flows.1 Algebraic closure model (ALG) In the algebraic closure model. 1984).09. see Chapter 10.1 Algebraic turbulence model (AEM) The Algebraic Eddy viscosity Model (AEM) in Delft3D-FLOW is a combination of two zero order closure schemes. (1993). Damping functions. ALG. To account for the vertical mixing induced by shearing and breaking of short and random internal gravity waves. For more details. The friction velocity at the free surface.5 turbulence closure model of Mellor and Yamada (1974. For strongly stratified flows (Ri > 1). 1982). The number of “steps” is dependent on the vertical grid. the k -ε model is advised.67.107) cµ u∗s Deltares A constant. see Equation 9. Delft3D-FLOW has a vertical filtering technique to remove this noise and to reduce the undesirable mixing. specified by you and the Brunt-Vais waves: DV = max D3D . as used in the standard Level 2. cµ ≈ 0. H (9.

This closure scheme is known as Prandtl’s Mixing Length model (PML).5. velocities higher in the water column have to be used in the determination of the bed friction velocity. this is physically not correct. Under the assumption that the velocity profile is logarithmic in the discrete model for layer with index k : u∗k = κ ln 1 + zk +d z0 uk .108) In tidal flows.107 and L of Equation 9. The friction velocity at the free surface is dependent on the wind velocity at 10 m above the free surface Eqs. In deep areas. 9.67 to 9.111) is used. Higher in the water column. a double logarithmic vertical velocity profile. Equation 9. u∗b ) (9.110) In the linear profile for the turbulence energy. (9. the turbulence intensity may still be large.98. see Baumert and Radach (1992). under the assumption of a logarithmic velocity profile: u∗b = κ ln 1 + ∆ zb 2 z0 ub . k=1 (9.107. User Manual ub ∗ A modified form of the bed friction velocity u∗b .2 Prandtl’s Mixing Length model (PML) A second algebraic expression for the eddy viscosity is obtained by assuming instantaneous local equilibrium between production and dissipation in the k -L model. the turbulent kinetic energy generated at the free surface and at the bed must be transported along the water column by vertical diffusion.Delft3D-FLOW. To avoid a zero eddy viscosity. The eddy viscosity is parabolic and gives in the case of wind-driven circulation. leading to: 1 k = √ L2 cµ 226 ∂u ∂z 2 + ∂v ∂z 2 . In the production term the horizontal derivatives are left out. (9. This will lead to phase differences in the time-series of the turbulent energy for the different layers. The bed friction velocity u∗b is determined from the magnitude of the velocity in the first grid point above the bed. 9. they are small compared to the vertical derivatives. ub ∗ = max (u∗ .112) Deltares . The turbulent kinetic energy k at the bed and at the free surface is determined basing on the shear stresses.69. (9.101 the eddy viscosity νALG is prescribed by Equation 9.1. Given k of Equation 9.109) and the average value is defined as: u∗ = 1 K K u∗k . at slack water the velocities are almost zero near the bed. In this algebraic turbulence closure model the wind stress and the bed stress affect the eddy viscosity over the total water depth instantaneously.

The transport equation for k is as follows: ∂k + ∂t u ∂k + Gξξ ∂ξ v ∂k ω ∂k + = Gηη ∂η d + ζ ∂σ 1 ∂ ∂k + 2 ∂σ Dk ∂σ (d + ζ ) + Pk + Pkw + Bk − ε. Equation 9. buoyancy and dissipation terms are the dominating terms. For robustness. The PML-model leads to a zero eddy viscosity and eddy diffusivity at the position in the vertical where the vertical gradients of the velocity are zero (e.98.g.114) The eddy diffusivity is derived from the eddy viscosity. For the PML model there is a phase difference in k . 9.5. (9.Conceptual description The mixing length L is again prescribed by Equation 9.113) see e. The second assumption leads to simplification of the production term. boundary layer type of flows).101.2 k -L turbulence model A so-called first order turbulence closure scheme implemented in Delft3D-FLOW is the k -L model.g. The conservation of the turbulent quantities is less important and the transport equation is implemented in a non-conservative form. νV (9. 9. The phase differences depend on the total water depth and the eddy viscosity: T ≈ H2 . The horizontal length scales are much larger than the vertical ones (shallow water.116) 227 . Given k of Equation 9. Note that for a logarithmic velocity profile. 1994). Baumert and Radach (1992). in Delft3D-FLOW the turbulence closure models PML and ALG are combined to one algebraic model.112 and L of Equation 9.115) with Dk = Deltares ν3D νmol + σmol σk (9. Tidal simulations using this parameterisation have been performed for the Irish Sea (Davies and Gerritsen.107 and 9. This is physically incorrect. using the Prandtl-Schmidt number.101 the eddy viscosity νPML is prescribed by Equation 9. introduced by phase differences in the vertical gradient of the horizontal velocities. in wind-driven flows). The following two assumptions are made: The production. In this model the mixing length L is prescribed analytically. The eddy viscosity is calculated following: ν3D = max (νALG . Eqs. the turbulent kinetic energy over the whole water column is in phase with the bed stress.112 give the same linear distribution of k . a buoyancy term and a production term. The turbulent kinetic energy k follows from a transport equation that includes an energy dissipation term. denoted as the AEM model. For the turbulence closure scheme ALG. The velocity scale is based on the kinetic energy of turbulent motion. without wind.101. νPML ) (9.

∂y κy (9.7.120) with the Prandtl-Schmidt number σρ = 0.Delft3D-FLOW. 9. you can switch on a more extended production term Pk of turbulent kinetic energy (option “partial slip”. In Eqs. L (9. ν3D is the vertical eddy viscosity.98.117 and 9.118 it has been assumed that the gradients of the vertical velocity w can be neglected with respect to the gradients of the horizontal velocity components u and v .121) where cD is a constant determined by calibration.117) For small-scale applications (e. This is represented by a buoyancy flux Bk defined by: Bk = ν3D g ∂ρ ρσρ H ∂σ (9. The turbulent energy production due to wave action is given by Pkw . turbulent kinetic energy is converted into potential energy. it is assumed that the dissipation ε depends on the mixing length L and kinetic turbulent energy k according to: √ k k ε = cD .7 for salinity and temperature and σρ = 1.g. the exact formulation is in Section 9. (9. derived from the constant cµ in the k -ε model: /4 cD = c3 µ ≈ 0. The horizontal and vertical (σ -grid) curvature of the grid has also been neglected. User Manual In the production term Pk of turbulent kinetic energy. prescribed by Equation 9.122) Deltares 228 .1925.0 for suspended sediments. the horizontal gradients of the horizontal velocity and all the gradients of the vertical velocities are neglected. In the k -L model. (9. (9. The production term is given by: Pk = ν3D 1 (d + ζ )2 ∂u ∂σ 2 + ∂v ∂σ 2 .118) In this expression. rough side wall) given by:  1 Pk = 2ν3D  2 (d + ζ )2  1 + 2ν3D  2 ∂u ∂σ 2 + ∂v ∂σ 2 + 1 ∂v Gξξ ∂ξ 1 ∂u Gξξ ∂ξ 2 2  + 1 ∂u + Gηη ∂η + 1 ∂v Gηη ∂η 2   .4. Near the closed walls the normal derivative of the tangential velocity is determined with the law of the wall: ∂u u∗ = .119) In stratified flows. simulation of laboratory flume).

3 k -ε turbulence model In the k -ε turbulence model. the following two assumptions are made: (9.7. H (9. Because of the first assumption. In case of wind forcing.5. cµ In the absence of wind.124) At open boundaries. transport equations must be solved for both the turbulent kinetic energy k and for the energy dissipation ε. (9. the conservation of the turbulent quantities is less important and the transport equation is implemented in a non-conservation form. cµ (9.58. A local equilibrium of production and dissipation of kinetic energy is assumed at the bed which leads to the following Dirichlet boundary condition: u2 k |σ=−1 = √∗b . see Section 9.123) The friction velocity u∗b at the bed is determined from the magnitude of the velocity in the grid point nearest to the bed.107 without horizontal advection. boundary layer type of flows). under the assumption of a logarithmic velocity profile. The bed roughness (roughness length) may be enhanced by the presence of wind generated short crested waves. the turbulent kinetic energy k at the surface is set to zero. see Equation 9.Conceptual description To solve the transport equation. buoyancy. ε In the transport equations. the turbulent energy k is computed using the equation for k Equation 9. The transport equations for k and ε are non-linearly coupled by means of their eddy diffusivity Deltares 229 . boundary conditions must be specified. For a logarithmic velocity profile this will approximately lead to the following linear distribution based on the shear-stress at the bed and at the free surface: 1 z+d k (z ) = √ u2 ∗b 1 − cµ H + u2 ∗s z+d . a similar Dirichlet boundary condition is prescribed for the turbulent kinetic energy k at the free surface: u2 k |σ=0 = √∗s . The mixing length L is then determined from ε and k according to: √ k k L = cD .126) The production. and dissipation terms are the dominating terms.125) 9. The horizontal length scales are larger than the vertical ones (shallow water.

117 and 9. and the buoyancy term Bk is defined in Equation 9.131) c1ε = 1. are given in Section 9.120.136) 230 Deltares .4.0.132) (9. c3ε = 0.135) cµ = cD cµ . c2ε = 1.129) The production term Pk is defined in Eqs. Dε and the dissipation terms.0 unstable stratification stable stratification (9. 1984): (9. the terms Pkw and Pεw in Eqs. The production term Pε and the buoyancy flux Bε are defined by: ε Pε = c1ε Pk .134) In Delft3D-FLOW in the ε-equation for stable stratification the buoyancy flux is switched off. so c3ε = 1. The energy production and energy dissipation due to waves. (9.0 and for unstable stratification the buoyancy flux is switched on c3ε = 0. 9.127 and 9.44. k ε Bε = c1ε (1 − c3ε ) Bk .92. the k -ε turbulence closure model is not available for you. User Manual Dk .130) (9. (9.126 and the calibration constants by (Rodi.Delft3D-FLOW.127) ∂ε + ∂t u ∂ε + Gξξ ∂ξ Dε ∂ε ∂σ + Pε + Pεw + Bε − c2ε ε2 . k with L prescribed by Equation 9. 9.133) (9. (9. The vertical eddy viscosity ν3D is determined by: √ k2 ν3D = cµ L k = cµ . ε with: (9.128. The transport equations for k and ε are given by: ∂k + ∂t u ∂k + Gξξ ∂ξ v ∂k ω ∂k + =+ Gηη ∂η d + ζ ∂σ 1 ∂ ∂k + 2 ∂σ Dk ∂σ (d + ζ ) v ∂ε ω ∂ε + = Gηη ∂η d + ζ ∂σ ∂ 1 2 ∂σ (d + ζ ) with + Pk + Pkw + Bk − ε. therefore for depth-averaged simulations. 1984). The coefficients of the 3D k -ε turbulence closure model as implemented in Delft3D-FLOW are not the same as in the depth-averaged k -ε turbulence closure model (Rodi.118.128) k Dk = νmol ν3D + σmol σk and Dε = ν3D σε (9.0 1.7.

Rehigh ) in which (9.). Our experience (Uittenbogaard. we extended the k -ε turbulence model in Delft3D-FLOW by a continuous damping function reducing its eddy viscosity.141) f (ReT ) = 1 − e−p Deltares 2 where p= ReT ν3D = 100 100cµ νmol (9. For a logarithmic velocity profile this will approximately lead to a hyperbolic distribution which is the superposition of two hyperbola.. the flow is nearly laminar. For the transport equation of the dissipation ε.142) 231 . the dissipation ε is set to zero at the free surface. 1 2 κ∆zs (9. Therefore. on the basis of the shear-stresses at the bed and at the free surface: ε (z ) = u3 u3 ∗s ∗b + . κ (z + d) κ (H − z − d) (9. In Delft3D-FLOW it is possible to switch on this so-called Low Reynolds turbulence model by adding the keyword LRdamp in the MDF-file. The turbulence Reynolds number is defined as ReT = k2 νmol ε = ν3D . the energy dissipation ε is computed using Equation 9. then the eddy viscosity and the eddy diffusivity computed from the k-ε model are corrected by the damping function ν3D (Relow ) = f (ReT )ν3D (k − ε. At open boundaries.4 Low Reynolds effect In laboratory flumes or highly stratified flows e.139) where z denotes the vertical co-ordinate. is that these models require large resolution and switch from laminar to turbulent flow conditions at too large Renumbers.123 and 9. 9.140) If the turbulence Reynolds number is low. however. pers. κz0 (9. the following boundary condition is used at the bed: ε|σ=−1 = u3 ∗b . corresponding to a double logarithmic velocity profile.5. in literature so-called low-Re k -ε models are proposed. cµ νmol (9. see below. by mud. comm. The formulation implemented is given below.137) and at the surface the dissipation ε is prescribed by: ε|σ=0 = u3 ∗s .Conceptual description For the transport equation of the turbulent kinetic energy k the same Dirichlet boundary conditions are used as for the k -L model described earlier.124.g. see Equation 9. 1999) and verified for the seasonal evolution of the thermocline (Burchard and Baumert.138) In case of no wind.128 without horizontal advection. Then the vertical eddy viscosity computed by the standard k-epsilon model is over-estimated because this model lacks viscous damping terms. Instead. The k -ε turbulence model was successfully applied for the simulation of stratified flow in the Hong Kong waters (Postma et al. 1995).

144) A vertical distribution for a river bend is given in Figure 9. In a 3D model the secondary flow is resolved on the vertical grid. It strongly varies over the vertical but its magnitude is small compared to the characteristic horizontal flow velocity. The flow in a river bend is basically three-dimensional.6: Spiral motion in river bend (from Van Rijn (1990)) No critical value for the Reynolds number needs to be specified. σ -model only This feature is only available in the σ -model. (9.6 Secondary flow. The spiral motion intensity I leads to a deviation of the direction of the bed shear stress from the direction of the depth-averaged 232 Deltares . but in 2D depth-averaged simulations the secondary flow has to be determined indirectly using a secondary flow model.999877) is applied. We remark that function f rapidly converges to 1 for large values of ReT . The secondary flow will be defined here as the velocity component v (σ ) normal to the depthaveraged main flow.7. The velocity has a component in the plane perpendicular to the river axis. see Kalkwijk and Booij (1986): ν (σ ) = f (σ ) I. see Figure 9.Delft3D-FLOW. 9.143) The vertical distribution of the secondary flow is assumed to be a universal function of the vertical co-ordinate f (σ ). (9. User Manual Figure 9. This so-called ’secondary flow’ (spiral motion) is of importance for the calculation of changes of the riverbed in morphological models and the dispersion of matter. The actual local velocity distribution originates from a multiplication of this universal function with the spiral motion intensity.6. because a value of ReT = 300 (f = 0. This component is directed to the inner bend near the riverbed and directed to the outer bend near the water surface. The spiral motion intensity of the secondary flow I is a measure for the magnitude of this velocity component along the vertical: 0 I= −1 |v (σ )| dσ.

(9.Conceptual description Figure 9. Depth-averaged continuity equation The depth-averaged continuity equation is given by: ∂ζ + ∂t 1 Gξξ ∂ (d + ζ ) U ∂ξ Gηη Gηη + 1 Gξξ ∂ (d + ζ ) V ∂η Gηη Gξξ = Q (9.156 and U is the magnitude of the depth-averaged velocity. The component of the bed shear stress normal to the depth-averaged flow direction τbr reads: τbr = −2ρα2 1 − α 2 U I. the depthaveraged shallow water equations have to be extended with: An additional advection-diffusion equation to account for the generation and adaptation of the spiral motion intensity. Deltares 233 .146) where U and V indicate the depth-averaged velocities on an orthogonal curvilinear grid.145) where α is defined in Equation 9. To take into account the effect of the secondary flow on the depth-averaged flow.7: Vertical profile secondary flow (V ) in river bend and direction bed stress flow and thus affects the bedload transport direction. This effect can be taken into account in morphological simulations. Additional terms in the momentum equations to account for the horizontal effective shearstresses originating from the secondary flow.

modelled as: Tξξ = −2βU V.Delft3D-FLOW.150) ∂ Gξξ 2Tηη + Gξξ Gηη ∂η ∂ Gηη 2Tηξ Gξξ Gηη ∂ξ with the shear-stresses.151) . (9. To close the equations these stresses are coupled to parameters of the depth-averaged flow field. (9.149) ∂ Gξξ 2Tξη + Gξξ Gηη ∂η ∂ Gηη 2Tξξ Gξξ Gηη ∂ξ and: Fsη = 1 d+ζ 1 ∂ [(d + ζ ) Tηξ ] + ∂ξ Gξξ + 1 ∂ [(d + ζ ) Tηη ] ∂η Gηη + . see Kalkwijk and Booij (1986).153) 234 Deltares . The main flow is assumed to have a logarithmic velocity profile and the secondary flow originates from a multiplication of a universal function with the spiral motion intensity. (9. (9.147) C2 ρ0 Gξξ D V ∂U + Gηη ∂η UV Gξξ ∂ Gξξ − ∂η V2 ∂ Gηη + ∂ξ and ∂V + ∂t U ∂V + Gξξ ∂ξ V ∂V + Gηη ∂η ρ0 UV Gξξ ∂ + fU = − Gηη Gξξ Gηη √ 2 2 1 gV U + V Pη − 2 (d + ζ ) + Fη + Fsη + Mη .and η -direction are given by: ∂U + ∂t U ∂U + Gξξ ∂ξ Gηη Gξξ Gηη √ 1 gU U 2 + V 2 − fV = − Pξ − 2 (d + ζ ) + Fξ + Fsξ + Mξ .148) C2 Gηη D Gηη − ∂ξ U2 ∂ Gξξ + ∂η The fourth term in the right-hand side represents the effect of the secondary flow on the depthaveraged velocities (shear stresses by depth-averaging the non-linear acceleration terms). User Manual Momentum equations in horizontal direction The momentum equations in ξ . Effect of secondary flow on depth-averaged momentum equations To account for the effect of the secondary flow on the depth-averaged flow. Depth averaging of the 3D equations leads to correction terms in the depth-averaged momentum equations for the effect of spiral motion: Fsξ = 1 d+ζ 1 ∂ [(d + ζ ) Tξξ ] + ∂ξ Gξξ + 1 ∂ [(d + ζ ) Tξη ] ∂η Gηη + . (9. Tξη = Tηξ = β U − V Tηη = 2βU V. the momentum equations have to be extended with additional shear stresses. resulting from the secondary flow. 2 2 (9.152) (9.

Equation 9. Ta Ie = Ibe − Ice . For βc = 0.165) Deltares 235 .161) (9. 1] . κC2D 2 (9.157. is described by a depthaveraged advection-diffusion equation: ∂ [(d + ζ ) I ] + ∂t + 1 Gξξ Gηη ∂ (d + ζ ) U ∂ξ Gηη I + ∂ (d + ζ ) V ∂η Gξξ ∂I Gηη ∂η Gξξ I = d+ζ Gξξ Gηη ∂ DH ∂ξ Gηη ∂I ∂ + DH ∂η Gξξ ∂ξ + (d + ζ ) S. 2 S=− U = Ta = La = U 2 + V 2. The limitation on α.160) (9.163) (9.157) with: I − Ie . is set to ensure that the length scale La in Equation 9. d+ζ U . La U (1 − 2α) (d + ζ ) .154) (9.155) (9.164) and Rs the radius of curvature of the stream-line defined by: Us ∂Ur =− . Ibe = Rs d+ζ Ice = f . correction coefficient specified by you.Conceptual description and: (d + ζ ) .164 is always positive. the depth-averaged flow is not influenced by the secondary flow.5α3 .156 effectively means a lower limit on C2D . The spiral motion intensity is computed by the spiral motion and κ the Von Karm Equation 9.158) (9. (9. The depth averaged transport equation for the spiral motion intensity The variation of the spiral motion intensity I in space and time.159) (9.156.6α2 + 37. (9. Remark: Equation 9.156) ∗ the effective radius of curvature of a 2D streamline to be derived from the intensity of with Rs ´ an ´ constant. β = β∗ βc ∈ [0. √ g 1 α= < . ∗ Rs β ∗ = βc 5α − 15. Rs ∂s (9.162) (9. 2κ2 α .

(9.168 is used.7 Wave-current interaction.168) Boundary conditions for spiral motion At open boundaries. The local equilibrium value Equation 9. this affects the stirring up of sediments and increases the bed friction. Delft3D-FLOW has an option to compute the spiral intensity I based on an algebraic expression assuming local equilibrium: I = Ibe + Ice . In the surf zone long-shore currents and a cross-shore set-up is generated due to variations in the wave-induced momentum flux (radiation stress). (9. The effective radius of curvature to be used for the evaluation of the coefficient β . For steady state simulations. The bed shear stress is enhanced. bathymetry strong circulations may be generated (rip currents). σ -model only This feature is only available in the σ -model.Delft3D-FLOW. while others have a (certain) distribution over the vertical. For perpendicular inflow the radius of curvature goes to infinity and the spiral intensity reduces to zero. (9. These processes are accounted for in a wave-averaged manner. The solution of the advection-diffusion equation will take some computer time. Equation 9.167) The above formulas account for two sources of secondary flow: The centrifugal force in case of curved streamlines. reads: ∗ Rs = (d + ζ ) U I . such as the energy 236 Deltares .155. In relatively shallow areas (coastal seas) wave action becomes important because of several processes: The vertical mixing processes are enhanced due to turbulence generated near the surface by whitecapping and wave breaking. such as the enhanced bed shear-stress or wave breaking at the surface. 9. Ibe . especially in cross-shore direction. In case of an irregular surf zone. Some processes basically act at a specific location or interface. and near the bottom due to energy dissipation in the bottom layer. User Manual with Us and Ur the components along and perpendicular to the streamline.166) To guarantee stability the effective radius of curvature is bounded by the following empirical relation: ∗ Rs ≥ 10 (d + ζ ) . A net mass flux is generated which has some effect on the current profile. The effect of the Coriolis force. Ice . during inflow the spiral motion intensity must be prescribed.

In SWAN. and wave length or wave period. j = {1. 3}. Delft3D-FLOW and SWAN can both apply a rectilinear or curvilinear grid.Conceptual description dissipation due to bottom friction in the wave boundary layer. the wave length and the wave period. For details and background you are referred to the Delft3D-WAVE User Manual (WAVE. The momentum equation in x-direction. 2008) and the references cited above. The numerical grids used in the flow and wave computation often differ. Delft3DWAVE supports currently one wave model: SWAN. such as the wave energy.. Express the residual terms that remain compared to the case without waves in terms of the wave properties. In Delft3D-FLOW. ∂xj (9. these terms have been implemented in a curvilinear co-ordinate system. The general procedure to derive the wave-induced forces is: Average the continuity equation and the momentum equations over the wave period. + g − = Fj . In most practical cases this does not apply. A vertical distribution can. Then the only way to split the mean and oscillating Deltares 237 . in a 2D computation such a process is accounted for in a depth averaged form. of course. The wave-induced force is computed as the gradient of the radiation stress terms. To use the required grid resolution in the area of interest you can use nested grids. u ¯j is the wave-mean velocity component. and Fj is the wave-induced force that remains after averaging the momentum equation over the wave period. i. the phase and group velocity. the wave averaged quantity is ¯ is the wave-mean indicated by an overhead bar. The above given procedure can only be applied when the mean motion is uniform with depth. you can use the same curvilinear grid as used in Delft3D-FLOW and obtain in one computation the required resolution in the area of interest. For a full wave-current interaction the currents from Delft3D-FLOW are used in Delft3D-WAVE (current refraction). A straightforward approach is to define the mean motion by time averaging the equations over a wave period.170) Using wave propagation models we can express Sij in terms of the wave parameters. 2. such as the wave energy. ζ free surface elevation. averaged over the wave motion and expressed in Cartesian co-ordinates is given by: ¯ ∂u ¯j ∂u ¯j ¯ij ∂ζ 1 ∂τ +u ¯i + . ∂t ∂xj ∂xj ρ ∂xi (9. however. only be accounted for in a 3D computation. there are various averaging procedures. a third generation wave model (Ris. the phase and group velocity. 1997).169) where for i and j the summation rule applies. For the sake of simplicity. This wave induced force can generally be written as the gradient of the radiation-stress tensor S : Fi = − ∂Sij . SWAN computes the full (directional and frequency) spectrum on a rectilinear or curvilinear grid with many processes formulated in detail. The transformation of results between the nested grids and the grids used in the flow and in the wave computations is executed automatically and is fully transparent to you. τ ¯ij are the components of the wave-averaged normal stress tensor. we will describe in this section all wave-induced effects on a rectangular grid. The computation of waves and wave-induced effects is the domain of wave models.. Obviously.

Remarks: For brevity. (9. the GLM. just the wave-mean water level must be prescribed. At water level boundaries. +u ¯L i ∂t ∂xj ∂xj ρ ∂xi (9. Groeneweg and Klopman (1998) and Groeneweg (1999) the (depth averaged and 3D) flow equations written in a co-ordinate system moving with the Stokes drift are very similar to the ordinary Eulerian formulation.172 contains a term related to a Stokes correction of the shear stresses. the hydrodynamic equations are written and solved in GLM-formulation. but now in GLM co-ordinates.. The relation between the GLM velocity and the Eulerian velocity is given by: uL = uE + uS . A super-script S is used to indicate the Stokes drift. velocity and not the Eulerian velocity. j = {1. in this section a velocity without super-script refers to the GLM velocity.Delft3D-FLOW. i. we discuss the wave-current interaction terms and their possible profile in the vertical.. 2. and the quantities have the same meaning as in Equation 9.171) where uL is the GLM-velocity vector. The Eulerian velocities are written to the Delft3D-FLOW result files to be used in comparisons of computational results with measurements. Quantities moving with the flow are transported by the total.e. In the following sections. unless explicitly mentioned differently. The velocity at the bottom is not the total velocity but the Eulerian velocity. As shown by Groeneweg (1999) the right-hand side of Equation 9. uE is the ordinary Eulerian-velocity vector and uS is the Stokes-drift vector. The momentum equation in x-direction. the total velocity must be prescribed.172) where for i and j the summation rule applies. At lateral velocity boundaries. However. Where applicable we distinguish between the depth-averaged and the 3D formulations. So in short: In Delft3D-FLOW. As shown by Andrews and McIntyre (1978). So. The only difference in the computational procedure as compared to the ordinary Eulerian formulation occurs at the boundaries. including the wave-current interactions. the implementation is such that the Eulerian velocity is prescribed in the boundary conditions (by you) and the Stokes drift is added by Delft3D-FLOW. The GLM velocities are written to the communication file and are used in all (sediment) transport computations. User Manual motion is through the Generalised Lagrangian Mean.169. the wave-induced driving force due to averaging over the wave motion is more accurately expressed in wave properties. GLM method of Andrews and McIntyre (1978). 3}. 238 Deltares . In the current implementation this term is neglected. i. averaged over the wave motion and expressed in GLM co-ordinates is given by: L ¯ ∂u ¯L ∂u ¯L ¯ij ∂ζ 1 ∂τ j j + . + g − = FjL . so the bed-shear-stress is corrected for the Stokes drift. we suppress the index L for GLM quantities.

In SWAN. ω (9. = . using the gradients of the radiation stresses in numerical models can result in spurious currents.. For linear current refraction the expression can be derived analytically. ..174) (9. . i. . Deltares 239 . the dissipation rate may be computed from the bottom friction (orbital motion).176) Fy = − (9. ∂t C2 D (9. = .. ∂x ∂y ω (9. To account for wave dissipation due to for instance bottom friction. 2D implementation For a depth averaged model the momentum equations in x. ∂x ∂y ω ∂Sxy ∂Syy ky − =D . (1987). depth-induced breaking and whitecapping. wave breaking and whitecapping and wave growth due to wind one can rely on mild slope formulations with dissipation terms.e. .175) where Fx and Fy are the depth averaged wave-induced forcings and given by the gradients of the radiation stress tensor S . (1987) approximated by wave energy dissipation: Fx = − kx ∂Sxx ∂Syx − =D . i. As shown by Dingemans et al. . . ∂t C2 D √ ∂V gV U 2 + V 2 + . or following Dingemans et al. + 2 (d + ζ ) + .173) where D is the total energy dissipation due to waves.7. The remaining part of the radiation stress gradients is closely related to the wave energy dissipation. can be expressed in the wave parameters of the wave model that is being applied. Dingemans et al. the right-hand side of Equation 9. (1987) showed that the divergence free part of the radiation stress is not capable of driving currents and can therefore be neglected if one is primarily interested in wave-driven currents. You can choose to apply the radiation stress or the dissipation rate to determine the waveinduced forces. .177) The dissipation rate D (a negative quantity) is computed by the wave model and read from the communication file. the right-hand side of Equation 9.e. ki is the wave number in i-direction and ω is the wave frequency. + Fx . + Fy . see Dingemans (1997) for many details and discussions on this subject.172..1 Forcing by radiation stress gradients The wave-induced force. .Conceptual description 9. leaving out most of the terms.172 can be written as: Fi = Dki . + 2 (d + ζ ) + . can be written as: √ gU U 2 + V 2 ∂U + .and y -direction.

9. The mean drift velocity is a second order quantity in the wave height.7. A particle at the top of the orbit beneath a wave crest moves slightly faster in the forward direction than it does in the backward direction beneath a wave trough. which may be used in comparisons with measurements at a fixed location.3 the dissipation due to bottom friction is added to the momentum equations. is always in the direction of wave propagation. In Section 9.180) S and M S are computed by an interface program and are written to the The mass fluxes Mx y communication file. As the energy dissipation due to bottom friction is usually much smaller than due to depth-induced breaking. The net horizontal displacement for a fluid particle is not zero. but it is more important to have its influence right near the bottom than the small error made in the wave-induced force in the top layer. The Eulerian velocities.178) (9. the so-called Stokes-drift. fluid particles describe an orbital motion. User Manual 3D implementation In version 3. This wave induced drift velocity. but the bottom friction should be applied in the wave boundary layer near the bottom. and D-Water Quality are based on the total flux velocities. whitecapping and bottom friction. but the wave-induced force is only applied in the top layer. Remarks: The mass flux effect is only taken into account when Delft3D-FLOW is used from within Delft3D-MOR.179) ρ0 v S dz = −d with E the wave energy defined as: 1 2 E = ρ0 gHrms . are written to the hydrodynamic map and history files.10 (and higher) of Delft3D-FLOW the 3D implementation is based on the same formulation as the 2D implementation.2 Stokes drift and mass flux In surface waves.7. So the bottom friction (due to the orbital motion) is counted twice. the error in the wave-induced force is small. Delft3D-WAVE. The drift leads to additional fluxes in the wave averaged mass continuity equation. 8 (9.Delft3D-FLOW. 240 Deltares . Remarks: The dissipation rate as provided by the wave model includes the dissipation due to depth-induced breaking. The first two indeed must be applied in the top layer. The velocities written to the communication file for use in Delft3D-MOR. S and M S are found by integration of the components of The wave-induced mass fluxes Mx y the Stokes drift uS and v S over the wave-averaged total water depth: ζ S Mx = S My = ρo uS dz = −d ζ E kx ω E ky ω (9.

187) A= Deltares uorb .184) 9. pg. given by Equation 9. The dissipation due to bottom friction Df is given by: 1 Df = √ ρ0 fw u3 orb . (1993b) given by: fw = min 0. sin φ)T .3 Streaming Streaming (a wave-induced current in the wave boundary layer directed in the wave propagation direction) is modelled as a time-averaged shear-stress which results from the fact that the horizontal and vertical orbital velocities are not exactly 90◦ out of phase. The friction factor fw is according to Soulsby et al.39 with: A z0 −0.7.Conceptual description 2D implementation The depth-averaged Stokes drift is given by: US = VS S Mx .182) 3D implementation The Stokes drift is computed from the wave theory. (9.188) 241 .185) The left-hand side of Equation 9. ρ0 (d + ζ ) (9.3.186) where uorb is the orbital velocity near the bed. 2 2 sinh (kH ) (9. eq. My . ω (9. see Dean and Dalrymple (1991.204.181) (9. (9. 1992): − ∂u ˜w ˜ ∂ Df k cos φ d+ζ −z = 1− ∂z ∂z ω δ Df k cos φ d+ζ −z = 1− for d + ζ − δ ≤ z ≤ d + ζ ωδ δ (9. 1.52 .8): uS (z ) = ωka2 cosh (2kz ) (cos φ. It is based on the wave bottom dissipation and is assumed to decrease linearly to zero across the wave boundary layer (Fredsøe and Deigaard.185 is the residual term after averaging a vertical derivative of the advection terms over the wave period.183) The angle between the current and the waves is computed from the mass fluxes which are read from the communication file: S S φ = tan−1 Mx . 287. ρ0 (d + ζ ) S My = . 10. 2 π (9.

189) 9.191) where c1ε is a calibration constant (c1ε = 1.09 H H A ks 0.8: Vertical distribution of turbulent kinetic energy production and φ is given by Equation 9. 0. Due to the breaking of waves the boundary conditions of k and ε have to be changed. 20 max ez0 ks . there is a production of turbulent energy directly associated with the energy dissipation due to breaking (Deigaard.4 Wave induced turbulence The vertical mixing processes are enhanced by the wave actions.5. which is computed in the wave model (e. The total depth-averaged contribution of wave breaking is given by Dw [W/m2 ].190) The source term Pεw in the ε-equation is coupled to Pkw according to: Pεw z ε = c1ε Pkw z . These processes are incorporated by introducing source terms in the turbulent kinetic energy (for both the k -L and the k -ε turbulence model) and in the turbulent kinetic energy dissipation equation (k -ε turbulence model).184. In the case of breaking waves. SWAN). The 242 Deltares . This can best be accounted for by adding the wave energy production and dissipation terms in the turbulence model. User Manual MWL ½ H rm s d+ ζ Z’ δ Figure 9.44). Hrms /2 . 2 (9. Wave energy dissipation due to bottom friction is also considered to produce turbulent kinetic energy.g.Delft3D-FLOW. The thickness of the wave boundary layer δ is given by: δ = H min 0. (9.7. Breaking waves In a 3D-model the contribution due to wave breaking is linearly distributed over a half wave height below the mean water surface. The two main sources of wave energy decay that are included are wave breaking and bottom friction due to the oscillatory wave motion in the bottom boundary layer.82 . This is described by the following expression for the turbulent kinetic energy distribution: Pkw z = 4Dw ρw Hrms 1− 2z Hrms for 1 0 ≤ z ≤ Hrms . k (9. 1986).

192) (9. Bijker (1967). by the adaptation of the bottom roughness height. and the bed shear-stress due to current τc . Huynh-Thanh and Temperville (1991). O’ Connor and Yoo (1988). as there are minor. Figure 9. Soulsby et al. Df represent wave energy dissipation due to bottom friction. τw . Van Rijn (2004). because the extra turbulence generated close to the bed by the waves appears to the current as being equivalent to an enhanced bottom roughness. The boundary conditions at the bottom. In addition the current profile is modified.9 gives a schematic overview of the bed shearstresses for wave current interaction. but specific differences in determining certain quantities. All these methods have all been implemented in Delft3D-FLOW and can be applied in 2D and 3D modelling. often very complex. (9. Various. Davies et al. such as determining the shear-stress at the bottom.Conceptual description boundary conditions at the free surface of k and ε are given by: k |free surface ε|free surface 2Dw κ = kwind + kwaves = kwind + ρw cD 4Dw = εwind + εwaves = εwind + ρw Hrms 2 3 (9. Grant and Madsen (1979). Deltares 243 . (1988). methods exist to describe the bottom boundary layer under combined current and wave action and the resulting virtual roughness. However.8.186. The bed shear-stress due to the combination of waves and current is enhanced beyond the value which would result from a linear addition of the bed shear-stress due to waves. (1993b).7.5 Enhancement of the bed shear-stress by waves The boundary layers at the bed associated with the waves and the current interact non-linearly.194) due to wave energy decay in the bottom boundary layer. (1993a) developed a parameterisation of these methods allowing a simple implementation and comparison of various wave-current interaction models: Fredsøe (1984). due to the non-linear interaction of the current and wave boundary layer is directly taken into account. see Figure 9. Pkw z = 2Df δ 1− d+ζ −z δ for d + ζ − δ ≤ z ≤ d + ζ. see Equation 9. Here z is the vertical co-ordinate with its origin in the (wave averaged) water level and positive downwards. we prefer to discuss the 2D and 3D implementation separately. Df is given by Equation 9. 9. This has the effect of enhancing both the mean and oscillatory bed shear-stresses. Christoffersen and Jonsson (1985). Myrhaug and Slaattelid (1990). τmax .212. For sediment transport modelling it is important to predict the maximum bed shear-stress.193) Bottom friction The contribution due to bottom friction is linearly distributed over the thickness of the wave boundary layer. 2D implementation Following Soulsby et al. while the current velocity and the turbulent diffusion are determined by the combined wave-current bed shear-stress τm .

9: Schematic view of non-linear interaction of wave and current bed shearstresses (from Soulsby et al. q . 244 Deltares .Delft3D-FLOW. The parameterisation of Soulsby for the time-mean bed shear-stress is of the form: |τm | = Y (|τc | + |τw |) . User Manual Figure 9.197) Z = 1 + aX m (1 − X )n .198) X= |τc | . |τc | + |τw | (9.195) Y = X {1 + bX p (1 − X )q } . each model having its own fitting coefficients.199) The value of the parameters a. and: (9. (1993b) fitted one standard formula to all of the models.196) |τmax | = Z (|τc | + |τw |) . with (9. Remark: The stresses τm and τmax are assumed to have the same direction as τc . m and n depends on the friction model which is parameterised. with (9. 89)) Soulsby et al. and: |τc | |τw | |τm | |τmax | magnitude of the bed stress due to current alone magnitude of the bed stress for waves alone magnitude of the mean bed stress for combined waves and current magnitude of the maximum bed stress for combined waves and current. and for the maximum bed shear-stress: (9. b. p. Figure 16. p. (1993b.

45 -0.10 -0.67 0.22 -0.50 0.70 1.00 -0. OY88=O’ Connor and Yoo (1988).50 -0.25 q4 0.06 Y = 1.19 -0.60 -0.00 0.52 0.67 0.82 1. (9.09 -0. VR04=Van Rijn (2004) Following Soulsby et al.00 0.17 0.71 m2 -0.33 -0.22 0.00 -0.17 0. DS88=Davies et al.50 0.77 b1 FR84 MS90 HT91 GM79 DS88 BK67 CJ85 OY88 VR04 1 b2 0.08 b4 -0.05 0.29 -0.29 0.05 0.59 -0.26 b3 -0. p.75 0.62 2.202) 245 .29 0.00 -0.19 0.78 0.65 0.38 q2 0.50 -0.5: Fitting coefficients for wave/current boundary layer model Model1 FR84 MS90 HT91 GM79 DS88 BK67 CJ85 OY88 a1 -0.25 -0.15 -0.87 1.19 1.49 -0.82 0.27 -2.35 0.66 0.70 0.47 -0.06 0.18 -0.75 -0.70 3. b.67 0.34 0.29 -0.19 0.07 -0.09 0.03 0.03 p1 -0.27 0.10 0. n) and J (= I.51 0.23 -0.59 0. m.00 0.75 0.64 0. 9.82 0.80 0.15 m4 0.00 0.15 -0.84 1.01 -0.26 q1 0. CJ85=Christoffersen and Jonsson (1985). BK67=Bijker (1967).34 0.54 to 9.18 -1. see Eqs.04 -0.19 0.00 -0.00 0.30 -0.58 -0.40 0.66 n3 0.00 p2 0.60 1.03 -0.50 0.00 0.42 -0.89 1.56.11 0.04 -0.27 m3 0. (1993b).21 -0.28 0.07 0.22 0.86 -0.00 -0.05 0.16 a4 0. q.31 p3 0.13 n4 -0.12 -0.89 1.65 0. see Table 9. p.66 J 3.02 -0.08 0.77 -0.00 0.91 1.201) Values of the parameters a.0 and Z = 0.50 0.09 0.00 0.07 0.26 0.23 -0.Conceptual description Table 9.200) C2D fw φ drag coefficient due to current wave friction factor the angle between the current direction and the direction of wave propagation.29 -0.200 have been optimised by Soulsby et al. GM79=Grant and Madsen (1979).11 0.40 -0.19 n2 -0. HT91=Huynh-Thanh and Temperville (1991). q and J in Equation 9. MS90=Myrhaug and Slaattelid (1990).50 2.12 I 0.02 0.09 m1 0.25 0. (1993b) the expressions for the parameters χ (= a.12 -0. As the radiation stress is always in the wave direction.13 0.24 0. we can derive φ from: |cos φ| = |U Fx + V Fy | U F .55 0.27 0.00 0.10 -0.08 0. (1988). (9.09 -0.24 -0.71 0. b.10 0.65 1.05 0.28 -0.25 0.06 0.13 0.72 0.21 -0. also depending on the friction model) have the form: χ = χ1 + χ2 |cos φ|J in which: + χ3 + χ4 |cos φ|J 10 log fw C2D .0 FR84=Fredsøe (1984).27 -0.29 0.45 a2 1.13 0.24 a3 -0.49 -0.06 -0.34 -0.14 -0.12 0.50 2.24 -0.34 2.42 0.33 0. J .58 2.19 q3 0.03 n1 0.22 -0.30 -0.68 -0.47 1. The bed shear-stress due to flow alone may be computed using various types of formulations ´ like Chezy. The bed shear-stress due to current alone can be written in the form: gρ0 U U τc = Deltares 2 C2 D .63 0.00 0.68 0.65 0.02 -0.73 0.14 -0.27 0.5 and Figure 9.10.23 0.25 p4 0.05 0.01 -0.12 -0. Manning or White Colebrook.95 1.56 2.28 0.00 0.55 0.60 0.00 -0.00 1.14 1.70 -1.73 0.00 -0.63 -0. (9.50 0.38 0.12 0.21 -0.32 0.

Figure 17. (1993b.10: Inter-comparison of eight models for prediction of mean and maximum bed shear-stress due to waves and currents (from Soulsby et al.Delft3D-FLOW. 90)) 246 Deltares . p. User Manual Figure 9.

≤π 2.205) A= uorb .116 and 11.208) U2D = Deltares 1 d+ζ ζ udz.Conceptual description The magnitude of the wave-averaged bed shear-stress due to waves alone is related to the wave orbital velocity near the bottom uorb and the friction coefficient fw : 1 |τw | = ρ0 fw u2 orb .204) where the root-mean-square wave height Hrms and the wave period T (= 2π/ω ) are read from the communication file. The total or effective bed shear stress is given by: τb = |τm | U U − US . The bed shear-stress corrected for the Stokes drift is given by: τb = with: |τm | U2D u − uS .157):  0. we must correct the bed shear-stress used in the momentum equations for the Stokes drift. 9.182. with: A ks −0. −d (9. 2 The orbital velocity is computed from the linear wave theory and is given by: (9. The magnitude of the depthaveraged velocity in Equation 9. As the bed is in rest and the equations are formulated in GLM co-ordinates. 11.181 and 9.202 is determined from the velocity near the bed.00251 exp 5.207) where the components of the depth-averaged Stokes drift U S are given by Eqs. A ks A ks >π 2.19 . assuming a logarithmic velocity profile.3.21 fw =  0. (9. The variation of the wave friction factor with relative orbital excursion at the bed under purely oscillatory flow is given by Swart (1974) (≡ Equations 11. 4 sinh (kH ) (9. 3D implementation For 3D simulations the enhanced bed shear-stress is computed along the same lines. (9. ω (9.209) 247 .206) ks is the Nikuradse roughness length-scale and ω is the wave angular frequency.203) uorb = 1 √ Hrms ω π . (9.77. but remark that the bed shear-stress is now a boundary condition.

The increased roughness length for waves and current z ˜0 satisfies: z ˜0 = ∆z b exp κ |ub | ˜∗ | |u −1 . along with the heat flux mechanisms at the water surface.11.211) with ub the horizontal velocity in the first layer just above the bed. On average neither the atmosphere nor the earth accumulates heat.8 Heat flux models The heat radiation emitted by the sun reaches the earth in the form of electromagnetic waves with wavelengths in the range of 0.212) This increased roughness length is used in the turbulence closure schemes. 9. with a logarithmic profile for the mean flow outside the wave boundary layer: ˜∗ ˜∗ u τm = ρ0 u with u ˜∗ the shear-stress velocity for waves and current and following Equation 9. (9. evaporation and convection. Matching the velocity profile corresponding with the increased mean bed stress τm for waves and current.58: (9. The wavelengths of these emitted radiations are longer (between 4 and 50 µm) due to the lower prevailing temperature in the atmosphere and on Earth. Schematically the radiation process.15 to 4 µm.183 and 9. (9. reflection and absorption by air. User Manual The Stokes drift and the angle between near bed current and the waves are given by Eqs. Most formulations differ in the dependency of the exchange on the meteorological parameters such as wind 248 Deltares . In the atmosphere the radiation undergoes scattering. is shown in Figure 9. cloud.184.210) ub = u ˜∗ ∆z b ln 1 + κ 2z0 . and heat loss due to back radiation. dust and particles. = Qs − Qsr atmospheric radiation (long wave radiation) in [J/m2 s] net incident atmospheric radiation (long wave) reflected atmospheric radiation in [J/m2 s] back radiation (long wave radiation) in [J/m2 s] heat loss due to evaporation (latent) in [J/m2 s] In Delft3D-FLOW the heat exchange at the free surface is modelled by taking into account the separate effects of solar (short wave) and atmospheric (long wave) radiation.11: Qsc Qco Qsr Qs Qsn Qa Qan Qar Qbr Qev radiation (flux) for clear sky condition in [J/m2 s] heat loss due to convection (sensible) in [J/m2 s] reflected solar radiation in [J/m2 s] solar radiation (short wave radiation) in [J/m2 s] net incident solar radiation (short wave).Delft3D-FLOW. which implies that the absorbed heat is emitted back again. Legend for Figure 9. 9. In literature there is a great variability of empirical formulations to calculate these heat fluxes across the sea surface.

The incoming radiation is absorbed as a function of depth. cloudiness and humidity. free convection of latent and sensible heat is computed by the model. This model formulation typically applies for large water bodies. Some formulations were calibrated for coastal seas others for lakes. The terms related to heat losses due to evaporation. when air and water densities and/or temperatures are such that free convection occurs. Additionally. back radiation and convection are computed by the model. 3 Excess temperature model The heat exchange flux at the air-water interface is computed.Conceptual description Figure 9. Deltares 249 . 4 Murakami heat flux model The net (short wave) solar radiation is prescribed. only the background temperature is required. In Delft3D-FLOW five heat flux models have been implemented: 1 Heat flux model 1 The incoming (short wave) solar radiation for a clear sky is prescribed. The evaporative heat flux is calibrated for Japanese waters. The net atmospheric (long wave) radiation and the heat losses due to evaporation. The effective back radiation and the heat losses due to evaporation and convection are computed by the model. 5 Ocean heat flux model The fraction of the sky covered by clouds is prescribed (in %).11: Overview of the heat exchange mechanisms at the surface speed. back radiation and convection are computed by the model. 2 Heat flux model 2 The combined net (short wave) solar and net (long wave) atmospheric radiation is prescribed. The effective back radiation and the heat losses due to evaporation and convection are computed by the model.

214) where Qtot [J/m2 s] is the total heat flux through the air-water surface. This may lead to over-prediction of the water temperature in shallow areas. this is only implemented in heat flux models 4 and 5.215) In Equation 9. Also the effect of precipitation on the water temperature is not taken into account. ρw is the specific density of water (in kg/m3 ) and ∆zs (in m) is the thickness of the top layer. cp (= 3930 J/kgK) is the specific heat capacity of sea water.8. the heat fluxes through the water surface by incoming radiation. This can be prevented by absorbing the incoming radiation as a function of depth. Currently. we refer to Octavio et al. as combined net solar 250 Deltares . The input parameters for the heat flux models are model dependent. to Murakami et al. ∂t ρw cp ∆zs (9.15. User Manual For the physical background of the heat exchange at the air-water interface and the definitions. (1977) for model 1 and 2.213) Qsn Qan Qbr Qev Qco net incident solar radiation (short wave) net incident atmospheric radiation (long wave) back radiation (long wave) evaporative heat flux (latent heat) convective heat flux (sensible heat). the water temperature near the free surface. The excess temperature model computes the heat fluxes through the water surface in such a way that the temperature of the surface layer relaxates to the natural background temperature specified by you.1 Heat balance The total heat flux through the free surface reads: Qtot = Qsn + Qan − Qbr − Qev − Qco . They are related by: lute temperature T ¯ = T + 273. in some formulas the abso¯ in K is used. T (9. the heat exchange at the bed is assumed to be zero. evaporation and convection are computed separately. Each of the heat fluxes in Equation 9. (1985) for model 4 and to Gill (1982) and Lane (1989) for model 5. The change in temperature in the top layer Ts (in ◦ C) is given by: ∂Ts Qtot = . The subscript n refers to a net contribution. Evaporation and convection depend on the air temperature. For example. In Delft3D-FLOW. with: (9. 9. the input for the solar radiation has to be specified (measured data) as incoming (shortwave) solar radiation for a clear sky (Qsc ) for heat flux model 1. However. and wind speed.213 will be discussed in detail. Remarks: The temperature T is by default expressed in ◦ C.Delft3D-FLOW. With the exception of model 3 the excess temperature model. to Sweers (1976) for model 3. This may result in an unrealistically high surface temperature when the top layer is thin. relative humidity.214 the total incoming heat flux is assumed to be absorbed in the top layer. back radiation. The heat transfer coefficient mainly depends on the water temperature and the wind speed.

(1 − β ) Qsn . 2 and 4 the solar radiation has to be specified by you.8. total water depth. by setting the coefficient of free convection cfrcon to zero. Applying Stefan-Boltzmann’s law for radiation from a black-body: ¯4 Q = σT (9. This can be switched off when needed. also related to the so-called Secchidepth γ = H 1. A part (approximately 50 %) is transmitted to deeper water. extinction coefficient (measured) in m−1 .06. In the heat flux models 1. In the excess temperature model (model 3) the solar radiation does not play an explicit role. which is absorbed at the surface and the remainder part. 9. while the relatively longer waves are absorbed at the surface.7 Secchi distance to the water surface in meters.216) ¯ the (absolute) temwith σ = Stefan-Boltzmann’s constant = 5. depending on the clarity of the water. In heat flux model 5 the flux is computed dependent on the geographical position and the local time. Equation 9.67 · 10−8 J/(m2 s K4 ) and T perature in K. the free convection of latent and sensible heat is also determined. The absorption of heat in the water column is an exponential function of the distance h from the water surface: Qsn (h) = (1 − β ) Qsn (0) with: e−γh . it is convenient to separate the incoming solar insolation into two portions: the longer wave portion. which is absorbed in deeper water. The incoming energy flux at the water surface depends on the angle (declination) Deltares 251 . Remark: The exponential decay function. Not all of the radiation is absorbed at the water surface.Conceptual description radiation and net atmospheric radiation (sum of Qsn and Qan ) for heat flux model 2. Therefore. The default value of β in Delft3D-FLOW is 0.2 Solar radiation The short-wave radiation emitted by the sun that reaches the earth surface under a clear sky condition can be evaluated by means of: Direct measurements. βQsn . In the other heat flux models the incoming radiation is expected to be absorbed in the top layer. as net solar radiation for heat flux model 4. but is part of the background temperature. When using heat flux model 5. Empirical formulae. In the heat flux model 5 the solar radiation is computed by the program and is dependent on the geographical position at the earth and the local time.217 has only been implemented in the heat flux models 4 and 5.217) β γ h H part of Qsn absorbed at the water surface which is a function of the wavelength. or is computed by the heat flux model itself (heat flux models 3 and 5). 1 − e−γH (9. Short waves can penetrate over a distance of 3 to 30 meters.

by the rotation of the Earth the solar radiation also varies during the day. see Figure 9. The fraction reflected or scattered (surface albedo) is dependent on latitude and season. For an overview of the angles used to determine the solar elevation angle γ . 1982): Qsc = 0. 23. Cloud cover will reduce the magnitude of the radiation flux that reaches the sea surface.12: Co-ordinate system position Sun between the incoming radiation and the Earth’s surface. (9. 0.220) The solar constant S = 1 368 J/(m2 s) or W/m2 . the declination is maximal. This tilting (angle δ ) varies with the time of the year and it leads to a seasonal variation of the radiation flux. 180 (9. sin(γ ) ≥ 0.Delft3D-FLOW. Near twelve o’clock local time. User Manual Figure 9.218) δ= 23. The cloudiness is expressed by 252 Deltares .76S sin(γ ). Of course. the sun elevation above the horizon is maximal.95).0. The incoming short-wave solar radiation through a clear sky at ground level Qsc is about 0. A part of the radiation that reaches the water surface is reflected. φ is the latitude. The Earth axis is not perpendicular to the line connecting the Sun with Earth. At June 21. sin(γ ) < 0. This is the average energy flux at the mean radius of the Earth.5π cos(ω0 t − 2. The temporal and latitude-dependent solar elevation angle γ is estimated by: sin (γ ) = sin (δ ) sin with: πφ 180 − cos (δ ) cos πφ 180 cos (ω1 t) (9. This declination depends on the geographical position on the Earth and the local time.219) where ω0 is the frequency of the annual variation and ω1 the frequency of the diurnal variation.12.76 of the flux incident at the top of the atmosphere (Gill.5 degrees.

with: (9.. the relation for the net atmospheric radiation Qan reads (Octavio et al. 1977): Qan = (218.221) Qsn Qs Qsr Qsc α Fc f (Fc ) net heat radiation (flux) from the Sun solar radiation (short wave radiation) in [J/m2 s] reflected solar radiation in [J/m2 s] radiation (flux) for clear sky condition albedo (reflection) coefficient fraction of sky covered by clouds (user-defined input) function of Fc The cloud function f (Fc ) and the albedo coefficient α used in Delft3D-FLOW are given in Table 9. This is expressed in the cloud function g (Fc ). The absorption of solar radiation is calculated (Gill.. the fraction of the sky covered by clouds. Taking into account the effect of reflection by the surface and reflection and absorption by clouds. Deltares (9.5. Details on this functionality can be found in Section B.0. The emission spectrum of the atmosphere is highly irregular. The correction factor for cloud cover is an empirical formula.09 0.6.Conceptual description Table 9. In heat flux model 1 a simple linear formula for the net atmospheric radiation is used (Octavio et al.222) ¯a is the air temperature (in K) and the reflection coefficient r = 0.0 − 0.8. For the Ocean heat flux model the net solar radiation can also be specified directly by the user instead of letting the model determine it from the cloud coverage. 9.223) 253 . The emissivity of the atmosphere varies between 0.0 − 0. The emissivity where T factor of the atmosphere ε may depend both on vapour pressure and air temperature. Qan = (1 − r) εσ T a (9. carbon dioxide and ozone in the atmosphere.65Fc2 - 1.3 Atmospheric radiation (long wave radiation) Atmospheric radiation is primarily due to emission of absorbed solar radiation by water vapour. The presence of clouds increases the atmospheric radiation.6: Albedo coefficient and cloud function Heat flux model 1 2 3 4 5 albedo (α) 0.38Fc2 a cloud cover fraction Fc .0 + 6.03. 1977): ¯4 g (Fc ) .3Ta ) g (Fc ) .06 0. The amount of atmospheric radiation that reaches the earth is determined by applying the Stefan-Boltzmann’s law that includes the emissivity coefficient of the atmosphere ε.7 for clear sky and low temperature and 1. 1982) as the product of the net downward flux of short wave-radiation in cloudless conditions and factors correcting for reflection and cloud cover: Qsn = Qs − Qsr = (1 − α) Qsc f (Fc ) .06 f (Fc ) 1.4Fc − 0.

9.227) The atmospheric radiation depends on the vapour pressure ea . 1977) and the reflection coefficient for the air-water interface r = 0. 9. the so-called effective back radiation. In heat model 3 the back radiation is part of the heat exchange coefficient (Sweers.985 of water (Sweers.8.222 is carried out around Ta = 15 ◦ C.8. (9. 1976.228) Deltares . In heat flux model 2 the atmospheric radiation is part of the (measured) net solar radiation flux specified by you. User Manual with Ta the air temperature (in ◦ C).225) ¯s the (absolute) water surface temperature in K. The back radiation depends on the surface temperature Ts .0 + 0.5.03: 4 ¯s Qbr = (1 − r) εσ T .8.058√ea ) 1. (9.65F 2 + 4εσ T ¯3 T ¯s − T ¯a . see Section 9. Octavio et al. In heat flux models 4 and 5 the back radiation is part of the total long-wave radiation flux. 9. The cloud function g (Fc ) in Eqs. Qeb = εσ T a c a 254 (9.2Ts .222 and 9.8. (9.0 − 0.226) Remark: In heat flux models 1 and 2 the back radiation from the water surface is computed explicitly.39 − 0. so the heat radiated back by the water can be described by Stefan-Boltzmann’s law of radiation. The fraction of the sky covered by clouds Fc is fixed for the whole simulation period in the heat flux models 1 and 4. In heat flux models 4 and 5 the atmospheric radiation is part of the total long-wave radiation flux.224) Remarks: In heat flux model 1 the atmospheric radiation is computed.6. This is called the effective back radiation: Qeb = Qbr − Qan . with T For heat flux models 1 and 2 Equation 9. (9. the air temperature Ta and the cloud cover Fc .Delft3D-FLOW.17Fc2 .8.5 Effective back radiation In heat flux models 4 and 5 the total net long wave radiation flux is computed.4 Back radiation (long wave radiation) Water radiates as a near black body.223 is given by: g (Fc ) = 1. the so-called effective back radiation.225 has been linearised around Ta = 15 ◦ C: Qbr = 303 + 5. The linearisation of Equation 9.. Fc is not used in heat flux models 2 and 3 and can be prescribed as a function of time in heat flux model 5.5. In heat flux model 4 the effective back radiation is computed using Berliand’s formula: ¯4 (0. see Section 9. see Section 9. corrected by an emissivity factor ε = 0. 1976).

Qeb = εσ T s c with the actual vapour pressure ea given by Equation 9. (9. In heat flux model 5 the effective back radiation Qeb is computed following: ¯4 (0.229) 9.0+0. 2 and 4 use vapour pressures defined as: 5 303. Heat flux models 1.5 · 106 − 2. Deltares 255 . It is computed differently for different heat flux models.38 exp 18. In heat flux model 5 it can also be specified varying in space. .00412Ta Here rhum is the relative humidity in [-].236.00412Ts .234) for the actual and remote vapour pressure. the vapour pressures are calculated using air and water surface temperatures in ◦ C. It should be specified in the input in percentages.232) where the actual vapour pressure ea and the saturated vapour pressure es . (9.7859+0. In heat flux models the evaporative heat flux Qev is defined by: Qev = Lv E.39 − 0.235) (9.0+0. respectively. The evaporation depends on meteorological factors (wind-driven convection) and vapour pressures.7859+0.6F 2 .234.230) Lv = 2. Remarks: The relative humidity rhum is specified as a function of time in all heat flux models but the excess temperature model (model 3).3 · 103 Ts .6 Evaporative heat flux Evaporation is an exchange process that takes place at the interface between water and air and depends on the conditions both in the water near the surface and the air above it.0 − 0.233) .Conceptual description with the actual vapour pressure ea given by Equation 9.3 ea = rhum 23. For heat flux model 5.8.3 . with Lv the latent heat of vaporisation in J/kg water: (9. (9.231) The evaporation rate E is defined as the mass of water evaporated per unit area per unit time [kg/(m2 s)]. (9. while for the other models the vapour pressures are computed using temperatures in K.03477Ta 1. can be given by different relations. ¯s T 5 303.1 − (9.1 − ¯ Ta es = 23. (9.05√ea ) 1. Most heat flux models use some form of Dalton’s law of mass transfer: E = f (U10 ) (es − ea ) .236) ea = rhum 10 0. Heat flux model 5 uses definitions for the vapour pressures as: es = 10 0.03477Ts 1.38 exp 18.

241) The saturated and remote vapour pressures es and ea are given by Eqs. 9.05 .237 are based on the wind speed 10 meter above the water level.62es .237) where Sarea is the exposed water surface in m2 . but E can also be prescribed as a function of time.6 on how to prescribe a time dependent evaporation.235 and 9. By default E is computed from the heat flux model formulation selected. the contribution to the evaporative heat flux is split in a contribution by forced convection and a contribution by free convection.forced = LV ρa f (U10 ) {qs (Ts ) − qa (Ta )} . 1985) to be 1. In heat flux model 4 the wind speed function following Murakami et al.0 · 106 Sarea 0.242) Deltares . 2 and 3 the wind speed function f (U10 ) following Sweers (1976) is used: f (U10 ) = (3. In heat flux model 5 the evaporation rate is computed from the difference in relative humidity. In the latter case this user-defined evaporation overrules the computed evaporation both in the heat flux model and in the continuity equation. User Manual The evaporation rate E computed from a heat flux model can be added as a source term to the right hand side of the continuity equation Equation 9. 256 (9.238) with cmur a constant. (9. (9.3. Patm − 0. Another difference between heat flux model 5 and the others is that. for heat flux model 5. These are discussed separately.0U10 ) 5. Forced convection of latent heat The latent heat flux due to forced convection for heat flux model 5 reads: Qev. It corresponds to a so-called Dalton number ce of 0.5 + 2. This was calibrated for Japanese waters (Murakami et al.Delft3D-FLOW.236. The coefficients calibrated by Sweers were based on the wind speed at 3 meter above the free surface.0016. (9.. see Appendix B. For models 1. The wind function in Equation 9.38ea qs (Ts ) = (9.239) with qs and qa the specific humidity of respectively saturated air and remote air (10 m above water level): 0. rather than from the difference in vapour pressure.239 is defined as: f (U10 ) = ce U10 .62ea qa (Ta ) = . Patm − 0. The keyword required is Maseva = #Y#. without the influence of free convection.38es 0. defined in the input and fixed for the whole simulation. (1985) is used: f (U10 ) = cmur U10 .2 · 10−9 .240) (9. the coefficients in Equation 9.

247) the remote air density (10 m above the water level): + ρa10 = 100ea Rvap Tair + 273.246) σ with σ = 0.236. The molecular diffusivity of air α m2 /s is defined as νair . Free convection is driven by buoyant forces due to density differences (by temperature and/or water vapour content) creating unstable conditions in the atmospheric boundary layer (ABL). 9.15 100Patm −100ea Rdry .243. The viscosity of air νair is assumed to have the constant value 16. In Equation 9.245) where the coefficient of free convection cf r. 1974) developed a correction to the wind function. (Ryan et al. the Dalton number ce in model 5 was calibrated for the North Sea to be ce = 0. wind driven. The specific humidity of respectively saturated air and remote air (10 m above the water level).05 J/(kg mol) and Rvap is the gas constant for water vapour: 461 495 J/(kg mol).0 · 10−6 m2 /s. see (Ryan et al. The total heat flux due to evaporation then results from adding the forced convection of latent heat in Equation 9. Free convection of latent heat Loss of heat due to evaporation occurs not only by forced convection. 2  0 cf r.free .235 and 9. but also by free convection. (9.249) Deltares 257 . The saturated and remote vapour pressure es and ea are defined in Eqs. (9..14.243) ρa = ρa0 + ρa10 .forced + Qev. Evaporation due to free convection is important in circumstances where inverse temperature/density gradients are present and wind speeds are almost negligible so that the amount of forced convection is small.240 and 9. 1974). Free convection is only included in temperature model 5 and the latent heat flux due to free convection reads: Qev. qs and qa are given by Eqs.free = ks LV ρa (qs − qa ) . with the average air density: (9.Conceptual description Without the influence of free convection.239 and the free convection of latent heat in Equation 9.241.15 .0015. (9. accounting for free convection. (9. 9. Neglecting free convection in this situation will lead to underestimating the heat loss.244) and with the heat transfer coefficient defined as: ks = if ρa10 − ρa0 ≤ 0 (ρa10 − ρa0 ) 1/3 if ρa10 − ρa0 > 0 (9. The derivation of evaporation by just free convection is based on the analogy of heat and mass transfer. the saturated air density is given α= by: 100Patm −100es Rdry + ρa0 = 100es Rvap Ts + 273.conv gα2 νair ρa (9.conv was calibrated to be 0.248) where Rdry is the gas constant for dry air: 287.7 (for air) the Prandtl number.243: Qev = Qev..

235 and 9. Forced convection of sensible heat The sensible heat flux due to forced convection is computed by: Qco.forced + Qco.free = ks ρa cp (Ts − Ta ) . just as the evaporative heat flux.255) 9. 258 Deltares . The convective heat flux is divided into a contribution by forced convection and a contribution by free convection. It is considered constant and taken to be 1 004. Equation 9.233 and 9.free .234 for heat flux models 1.8 Overview of heat flux models Heat flux model 1 For heat flux model 1 the relations used for the computation of the heat balance. (9.254) The total heat flux due to convection then results from adding the forced convection of sensible heat in Equation 9.251) γ = 0. In heat flux model 5 the convective heat flux is split into two parts.forced = ρa cp g (U10 ) (Ts − Ta ) . 9. The wind-speed function g (U10 ) is defined following Gill (1982): g (U10 ) = cH U10 .252 and the free convection of sensible heat in Equation 9.7. (9. (9.7 Convective heat flux Assuming that the turbulent exchange of heat at the air-water interface equals the turbulent exchange of mass.66 for heat flux models 1 and 2 for heat flux model 4.254: Qco = Qco.Delft3D-FLOW. 9.250) (9.8.252) with cp the specific heat of air.245.213.61 γ = 0. are given in Table 9. es − ea with γ the so-called Bowen’s constant: (9.00145.253) with cH the so-called Stanton number. User Manual 9. 2 and 4 and by Eqs. Without the influence of free convection.0 J/(kg K). the Stanton number was calibrated for the North Sea to be cH = 0.8. Free convection of sensible heat Qco. with the heat transfer coefficient ks given by Equation 9. Ts − Ta Rb = γ . the convective heat flux can be related to the evaporative mass flux by the Bowen ratio: Qco = Rb Qev . (9.236 for heat flux model 5. The saturated and actual vapour pressures are calculated by Eqs.

Excess temperature model .237 Eqs. The heat exchange coefficient λ is a function of the surface temperature Ts and the wind speed U10 . It is derived by linearization of the exchange fluxes for back radiation. (9. Table 9. The remaining terms (back radiation. 9. 9. 9.223 and 9.226 Eqs.230 to 9.256) with Ts the water temperature at the free surface and Tback the natural background temperature.Conceptual description Table 9. the heat exchange flux is represented by a bulk exchange formula: Qtot = −λ (Ts − Tback ) . 9.7: Terms of the heat balance used in heat model 1 Quantity Description short wave solar radiation for clear sky net solar radiation net atmospheric radiation back radiation heat loss due to evaporation heat loss due to convection According to measured Equation 9. evaporation and convection.049Ts + f (U10 ) 1. (9.230 to 9. evaporation. 9.226 Eqs.257) For the wind function f (U10 ) Equation 9.00158Ts .251 Qsc Qsn Qan Qbr Qev Qco Table 9.224 Equation 9. convection) are computed similarly as for heat flux model option 1.221 Eqs.251 Qsn + Qan Qbr Qev Qco Heat flux model 2 Heat flux model 2 is based on formulations given by Octavio et al.018Ts + 0.237 Eqs. The source of data for the combined net incoming radiation terms (Qsn + Qan ) must come from direct measurements. The following relation was derived by Sweers (1976): 2 λ = 4.8.8: Terms of the heat balance used in heat model 2 Quantity Description net solar and net atmospheric radiation back radiation heat loss due to evaporation heat loss due to convection According to measured Equation 9. both in ◦ C. Deltares 259 .heat flux model 3 The excess temperature model 3 is based on Sweers (1976).48 + 0.12 + 0.237 is applied.250 and 9. (1977).250 and 9.

251 Qsn Qeb Qev Qco Table 9..229 Eqs.252 to 9.221 Equation 9. 9. Ocean model . 1985).228 Eqs. The required time dependent input data for each of the heat flux models is summarised in Table 9.heat flux model 5 The heat flux model 5 following Gill (1982) and Lane (1989) was calibrated for the North Sea and successfully applied for great lakes.9. 260 Deltares .9: Terms of the heat balance used in heat model 4 Quantity Description net solar radiation effective back radiation heat loss due to evaporation heat loss due to convection According to measured Equation 9.heat flux model 4 The heat flux model 4 was calibrated for the Japanese waters (Murakami et al.10: Terms of the heat balance used in heat model 5 Quantity Description net solar radiation effective back radiation heat loss due to evaporation heat loss due to convection According to Eqs. User Manual Table 9.250 and 9. The following relations are used for the computation of the heat terms.234 Eqs.10.239 to 9.218 and 9.11. 9. 9.Table 9. Table 9. 9. The following relations are used for the computation of the heat terms. 9.230 to 9.Delft3D-FLOW.255 Qsn Qeb Qev Qco Murakami model .249 Eqs.

For coastal areas. ζ (idealised situation). Earth Tide. then the ocean response is called the equilibrium tide. Assuming that the earth is covered entirely with water and neglecting any secondary effects (see next section). In the numerical models of these areas the contribution of the gravitational forces on the water motion increases considerably and can no longer be neglected.13. The need to model larger seashore areas with sections of the deep ocean or large closed basins has increased. the prescription of tidal forcing along open boundaries is sufficient in generating the appropriate and accurate tidal motion.Conceptual description Table 9. Moon and Earth) on the water mass. Deltares 261 . to generate an accurate tidal motion. The tide generating forces originate from the Newtonian gravitational forces of the terrestrial system (Sun. The amount of water mass in these models is relatively small and the effect of these forces on the flow can be neglected.9 Tide generating forces Numerical models of tidal motion in coastal seas generally do not account for the direct local influence of the tide generating forces.11: Summary of time dependent input data of the heat flux models Quantity model 1 model 2 Excess temperature Murakami Ocean Qsc Qsn Qsn + Qan Tback Ta rhum Fc (optional) (optional) 9. see Figure 9. Displacement the earth’s crust due to the elasticity of the earth. Based on their origins the tide generated by these forces can be distinguished into: Equilibrium tide.

(9. Gηη ∂η (9.1 Tidal potential of Equilibrium tide The equilibrium tide is modelled by including the tide generating potential terms for equilibrium tide in the momentum equations.UTC Table 9.5 cos2 φ − 1 cos (ωi t + 2λ + χi ) .259) consisting of the following major tidal species: Long-Period species ν = 0 : ϕ0 = Ki 1.9.2 ϕν (λ. Semi-diurnal species ν = 2 : ϕ2 = Ki cos2 (φ) cos (ωi t + 2λ + χi ) . φ Ki ωi χi t geographical co-ordinates amplitude of equilibrium tide (component dependent) frequency of equilibrium tide (component dependent) astronomical argument of equilibrium tide (component dependent) relative to UTC midnight universal standard time . Gξξ ∂ξ and g ∂ϕ . with: (9. t) . 9. 9.13: Effect of tide generating force on the computed water elevation at Venice 9.7 and 9.1.260 to 9.260) (9.Delft3D-FLOW. 262 Deltares . φ. User Manual Figure 9. the tidal potential ϕ for equilibrium tide is decomposed into the series: ϕ= ν =0.261) (9. The following terms are added to the right-hand side of the momentum equations Eqs.262. Diurnal species ν = 1 : ϕ1 = Ki sin (2φ) cos (ωi t + λ + χi ) . respectively: g ∂ϕ .8.258) Following Schwiderski (1980).262) λ.12 provides an overview of frequencies ωi and constants Ki for each of the 11 tidal constituents considered in the species Eqs.

User Manual.984104237 30.046398 0.046843 0.082137278 0.439729554 30.544374684 0.022026 0.12: Constants of major tidal modes Tidal constituent Semi-diurnal Species M2 principal lunar S2 principal solar N2 elliptical lunar K2 declination lunar-solar Diurnal Species K1 declination lunar-solar O1 principal lunar P1 principal solar Q1 elliptical solar Long-period Species Mf fortnightly lunar Mm monthly lunar Ssa semi-annual solar Frequency ωi (degrees/hr) Constant Ki 28. If the computations are to be carried out in local time. this imposes a restriction on the simulation period to approximately one year. all other constituents fall below 4% of the dominating semidiurnal principal moon (M2) tide and need not be considered in global ocean tide models.030704 15. Since tidal models can be used in predictive mode (predicting the actual water elevations and currents) the long term variations (18.082137278 0. with regard to the accuracy of the generated tide.141565 0.000000000 28.398660914 0.262 a module of Delft3D-TIDE is applied.019256 1. then the time difference between the local time and UTC must be specified.958931361 13. For the determination of the astronomical arguments χ in Eqs.098033041 0.260 to 9. This difference is referred to as the local time zone (LTZ) in the input.100514 0.041742 0.041068639 13.943035598 14. In Delft3DFLOW the correction is carried out only once at the initialisation of the simulation. TIDE (2008)). 9.019446 Following Schwiderski (1980).6 years cycle) in the tidal components are accounted for by including nodal corrections (see Delft3D-TIDE.Conceptual description Table 9.242334 0.113033 0. Consequently. Deltares 263 .

than the inclusion of proper eddy viscosity and bottom friction coefficients for the hydrodynamics of an ocean basin.9. These perturbations are subsequently regrouped in the final equations according to their causes and are approximated as follows: ϕe − ζ e = 0. (9. 2 ϕe . it assumes that earth is a stiff body) and/or the interaction between the two. yielding a 31 % reduction of the originally derived equilibrium tide potential 264 Deltares .264) However the improvement due to the second correction term is considerably less significant (10% improvement in the difference between observed and simulated phases). Eqs. in the momentum equations.10ζ. and: (9.e.7 and 9. the following net tidal potential terms are applied. the following two interaction effects should be considered: 1 Earth tide (ζ e ) and tidal loading (ζ eo = ζ e − ζ b ). Consequently. ϕeo and ϕo are perturbations of the equilibrium tide ϕ in response to ζ e . see Schwiderski (1980). Therefore.14: Earth ocean tidal interaction (after Schwiderski (1980)) 9. User Manual Figure 9. 9. in Delft3D-FLOW only the first correction term has been included. are the earth tidal responses to the equilibrium tide ϕ and oceanic tidal load ζ . It neglects the effect of these forces on terrestrial tide (i.Delft3D-FLOW.2 Tidal potential of Earth tide The equilibrium tide approach concentrates on the influence of gravitational forces on ocean tides. ζ eo and ζ .31ϕ. Following Schwiderski.263) ϕo + ζ eo − ϕeo = 0.8. respectively.

the wet flow-through area and and the upstream and downstream water level.69 g ∂ϕ . 9. Furthermore.1. The direction of the forces: U or V should be specified at input.1 3D gate In 3D simulations.10.2 Quadratic friction The flow rate Q through a hydraulic structure can often be described by a so-called Q-H relation.267) with µ the contraction coefficient (0 < µ ≤ 1). respectively. the flow in a computational layer is blocked or a quadratic energy or linear loss term is added to the momentum equation. the velocity and the bathymetry. so it has no influence on the water volume. (9. groynes. 9. To model the force on the flow generated by a hydraulic structure. should be parameterised. sluices. The forces due to obstacles in the flow which are not resolved (sub-grid) on the horizontal grid. Examples of these obstacles in civil engineering are: gates. In the numerical model the hydraulic structures should be located at velocity points of the staggered grid.9. Gξξ ∂ξ and 0. a free slip boundary conditions has been implemented at the transition from an open layer to a closed layer (representing a gate): νV ∂U ∂z = 0.69 g ∂ϕ .265) 9. z =a (9.266) with z = a the top or the bottom of the gate. The hydrostatic pressure assumption may locally be invalid. constructions which partially block the horizontal flow can be modelled as so-called “gates”. Its horizontal and vertical position can be specified. The obstacles generate energy losses and may change the direction of the flow. Upstream of the gate the flow is accelerated due to contraction and downstream the flow is decelerated due to expansion.Conceptual description terms): 0. For the numerical implementation we refer to Section 10. see Section 10.10. They relate the flow rate to the difference between the upstream and downstream water levels: Q = µA 2g |ζu − ζd |. bridge piers and porous plates. This generates a vertical structure of the horizontal flow around a hydraulic structure which is more realistic. The obstacles are denoted in Delft3D-FLOW as hydraulic structures. weirs.10 Hydraulic structures Obstacles in the flow may generate sudden transitions from flow contraction to flow expansion. There is no transport of salt or sediment through the computational layers of a gate. The grid resolution is often low compared to the gradients of the water level. The width of a gate is assumed to be zero. barriers. The mathematical formulations are given in this section. Gηη ∂η (9. The contraction coefficient in the so-called Q-H Deltares 265 . The horizontal velocities of the computational layers at position of the gate are set to zero.

closs−u u u2 + v 2 . ∆x closs−v Mη = − v u2 + v 2 . Delft3D-FLOW has been extended with a couple features. such the floodplains of rivers and in the intertidal areas of estuaries. mega-ripples and dunes. Following Van Rijn (2007) we distinguish between roughness heights for ripples.3 Linear friction To reduce the flow rate through a certain cross-section Delft3D-FLOW offers also the functionality of a linear friction term. each type influences the hydro.268) (9. 9. this includes bedform roughness predictors and vegetation models.22) or to compute a dune roughness height which can 266 Deltares . ∆y Mξ = − (9.2.271) 9.4.2) and in a 3D model by a combination of bed resistance formulations.11. (9. the flow resistance is determined by a combination of vegetation and an alluvial bedforms or even a non-alluvial bed.and morphodynamics in its own way.10. Manning or White Colebrook as discussed in Section 9. ∆x closs−v v Mη = − . Delft3D contains a module to predict dune characteristics (height/length) which may be used in determining the threshold level for dredging (see keyword UseDunes in Section A. we start with a description of the bedform roughness height predictor. This kind of hydraulic structure is denoted as a rigid sheet : closs−u u . which should be specified by you at input.1. For the following types of hydraulic structures.Delft3D-FLOW. To accurately represent such conditions in the numerical model.270) (9.1 Bedform heights There are various types of bedforms. First.269) with closs the energy loss coefficient. ∆y Mξ = − with closs the energy loss coefficient. a quadratic formulation for the energy loss is implemented in Delft3D-FLOW: barrier bridge pier Current Deflecting Wall (CDW) local weir or 2D weir porous plate culvert 9.11 Flow resistance: bedforms and vegetation The terrain and vegetation excert shear stresses on the passing flow. Within the main stream flow the shear stresses are largely determined by the local conditions of the alluvial bed (bed composition and bedform characteristics). The magnitude of the shear stress of the bed is often characterised by means of roughness coefficient of type ´ Chezy. The contraction coefficient µ can be used to determine the additional quadratic friction term in the momentum equation.11. In other areas. User Manual relation is dependent on the kind of hydraulic structure. These types of flow resistance may be resolved in a 2D numerical model using the trachytope approach (see Section 9.

11.276 are approximations for the dune height under equilibrium conditions. Fredsøe (1982) based on Meyer-Peter and Muller (1948) or Engelund and Hansen (1967).2. In case of Van Rijn (1984c) the dune height hd is computed as hd = 0.277) By default these effects are neglected. (ii) constant relaxation distance LH .272–9. this effect has been implemented by means of an advection relaxation equation.11εH D50 H 0. Both approaches are described in the two sections below.eq .3 (1 − e−T /2 )(25 − T ) (9. k ) depending on depth-averaged flow magnitude U .Conceptual description be used as the bed roughness in the flow computation via trachytope formulae 105 (Equation 9. the dune height will need time to adjust.272) where ε is a calibration coefficient and T is the dimensionless bed shear parameter. Under time and spatially varying conditions.274) where µ is the efficiency factor (C/Cg. The corresponding keywords are described in Appendix B. By means of the keyword BdfRlx one can select four options for computing the variables uH . such that hd = hd. If one applies the formulation of Fredsøe (1982) to the transport formula by Engelund and Hansen (1967) one obtains 1 hd = εH 5 (9. (iii) relaxation distance proportional to Deltares 267 .1.eq − hd ∂hd ∂hd ∂hd + uH + vH = ∂t ∂x ∂y TH (9.311) or 106 (Equation 9. H. the power relation dune height predictor reads hd = aH H bH (9. One can also determine the dune.276) The dune height formulations given by equations 9.273) If one applies this approach to the transport formula by Meyer-Peter and Muller (1948) one ¨ obtains hd = 24 θcr εH max(1 − . 9.16. water depth H and Nikuradse bed roughness k as hd = εsH / U ∂s ∂s ∂s −H −k ∂U ∂H ∂k (9. One ¨ of these predictors can be selected using the keyword BdfH. 0) 63 µθ (9.275) The latter two formulations have been implemented in Delft3D. hd. and ripple roughness heights directly according Van Rijn (2007). vH and TH : (i) constant relaxation time scale TH .1 Dune height predictor Four dune height predictors have been implemented: Van Rijn (1984c).312) as described in Section 9. Fredsøe (1982) formulated an approach to obtain a dune height predictor for a general sediment transport formula s = f (U. and a power relation.90 )1.5 based on a grain related bed roughness of 3D90 and θcr is the critical mobility parameter based on the Shields curve or given by the user.11. Finally. mega-ripple.

In this case the bedform migration direction is assumed to be equal to the direction of the bedload transport vector sb .282) where sb is the total bedload transport rate. bcH and γmin are constants that can be set.3H otherwise it will be computed using the power relation (9.e. and cH is the bedform celerity (magnitude) given by a power relation derived for a generic transport formula approximation s = as U bs as cH = bs s as bs U bs aC U bC = = H H H (9.283) Ld = aL H bL (9. φmax Ld (9. bLH . Fr2 is the squared Froude number U 2 /gH and Hmax . i. you can choose either one of the following two options using keyword BdfL. γmin sb H (1 − Fr2 ) (9. U the flow velocity magnitude u2 + v 2 .281) and the bedform celerity is given by cH = acH max H Hmax bcH .280) while the relaxation time scale TH is either given. This value will only be used if you use the trachytope option for specifying bed roughness and include either one of alluvial roughness formulae 105 and 106.278) (9. Finally. If you choose to compute the dune height according Van Rijn (1984c) then it will be computed as Ld = 7.284) Note that only the dune height is subject to the advective and relaxation behaviour formulated in Equation 9. the direction of the bedform migration is assumed to be equal to the direction of the depth averaged velocity in all cases. acH .277. φmax . Van Rijn (2007) or a power relation. In case of Van Rijn (1984c) the roughness height related to dunes is computed as ks.d = 1. and (iv) relaxation distance dependent on dune length Ld and water depth H . you may select one of three methods for computing the bedform roughness height by means of the keyword BdfRou. uH = cH vH u U v = cH U (9. In the fourth advection case TH is computed again using TH = LH /cH but in that case the relaxation distance LH is given by LH = min Hmax H bLH . these methods are based on the formulations of Van Rijn (1984c). For the dune length Ld used in some of the formulations above. or computed from TH = LH /cH where the relaxation distance LH is either given or proportional to the dune length LH = fLH Ld . User Manual dune length Ld .Delft3D-FLOW.1hd 1 − e−25hd /Ld 268 (9. for the dune length always these equilibrium formulations are used.285) Deltares .279) where u and √v are the x and y components of the depth averaged velocity. In the first three advection cases.

286) The roughness height related to ripples is set in both cases equal to 3D90 . 9.d otherwise (9. In case of Van Rijn (2007) the roughness heights neither depend on the computed dune height nor on the dune length.5 − 0. Dsilt = 32 µm and ks.mr otherwise (9.02    200D 50 if ψ ≤ 50 if 50 < ψ ≤ 550 if ψ > 550 and D50 ≥ Dsand if ψ > 550 and Dsilt ≤ D50 < Dsand (9. in that case the equations given in the following section are used.25Dgravel /D50 )1. significant wave height Hs .r = αr 20Dsilt if D50 < Dsilt ∗ ks. (6a)–(6d) by Van Rijn (2007) the roughness height predictor for mega-ripples is given by ks. The value of ks.000062 m.d = αd Deltares 0 if D50 < Dsilt ∗ ks.11.1.65ψ )fcs D50 if 50 < ψ ≤ 250   20fcs D50 if ψ > 250 (9.2 m. water depth H and factor fcs = (0.0002ff s ψH    (0.r otherwise (9.mr is limited to values smaller than 0.2 Van Rijn (2007) bedform roughness height predictor As proposed by Eq.mr = αmr 0 if D50 < Dsilt ∗ ks.289) ∗ is given by where αmr is a calibration factor and ks. (7a)–(7d) by Van Rijn (2007) the roughness height predictor for dunes is given by ks. wave length L derived from (L/Tp ± uc )2 = gL tanh(2πh/L)/(2π ).r   if ψ ≤ 50 150fcs D50 = (182. wave number k = 2π/L.287) ∗ is given by where αr is a calibration factor.d = aR Hd (9.5Dsand ) and Dsand = 0.291) 269 .Conceptual description and in case of the power relation this quantity is computed as bR ks.00002ψ )f H fs =  0.mr ∗ ks.011 − 0.290) in which ff s = (D50 /1.r ∗ ks. Uw s w w s r c depth-averaged current velocity uc . Following Eq.r is limited to values ranging from D90 to 2% of the water depth. (5a)–(5d) in Van Rijn (2007) the roughness height predictor for ripples is given by ks. relative wave period Tr . peak orbital velocity near bed U = πH /[T sinh(2kh)].288) 2 /[(s − 1)gD ] using U 2 = in which ψ is the current-wave mobility parameter defined as Uwc 50 wc 2 +u2 and relative density s = ρ /ρ .5 in which Dgravel = 0.002 m. peak wave period Tp . Following Eq. The value of ks.mr   0.

old ks.r = (1 − αr )ks.293) −∆t/Td ks.Delft3D-FLOW. User Manual ∗ is given by where αd is a calibration factor and ks. 9.10.old where αr = e−∆t/Tr (9. and Tr . the corresponding relaxation factor α∗ will be zero and the relaxation effect disappears. 0].2 Trachytopes This functionality allows you to specify the bed roughness and flow resistance on a sub-grid level by defining and using various land use or roughness/resistance classes.mr = (1 − αmr )ks.17.r. 1 Mξ = − λm.4.d.n v u2 + v 2 2 (9. At every time step (or less frequent as requested by the user) the trachytopes are converted into a representative bed roughness C .r.mr.mr.294) (9.1) and optional linear flow resistance coefficient λm. Tmr and Td are different relaxation time scales for the three bedform types.new + αd ks.old where αmr = e−∆t/Tmr where αd = e where ∆t is the hydrodynamic time step of bed roughness updating. Warning: Until Delft3D release 3.277 cannot be included these Van Rijn (2007) formulations. The input parameters and files to use the trachytopes functionality are described in Section B. 270 Deltares . Note that if a relaxation time scale T∗ becomes zero. k or n (as described in Section 9.292) Since these formulations do not depend on the equations for the dune characteristics given in the previous section.28. the advection and relaxation effects represented by Equation 9.295) To save computational time the user may choose to update the computed bed roughness and resistance coefficients less frequently than every time step.d = (1 − αd )ks.0004ff s ψH = (0. Non-erodible layers and partial transport effects are not taken into account to reduce the roughness height.048 − 0. a separate relaxation option has been implemented for all three bedform roughness heights according Van Rijn (2007) formulations as given by ks. further referred to as trachytopes after the Greek word τραχυ ´τ ης for roughness.new + αr ks. Remarks: In case of a simulation with multiple sediment fractions.d ∗ ks.d   0.n per velocity point.00008ψ )ff s H   0 if ψ ≤ 100 if 100 < ψ ≤ 600 if ψ > 600 (9.d. the D50 of the grain mixture in the surface layer is used. the relaxation factors α∗ are not defined as indicated above but as max[1 − (5∆t/T∗ ). See Section B.new + αmr ks.n u u2 + v 2 2 1 Mη = − λm.17 for a description of the keywords and input files associated with this feature.11. Therefore.

9. The next six sections provide an overview of the various trachytope formulae implemented. The area classes (type range 51–200) basically cover the whole area.1 Trachytope classes Three base classes of trachytopes are distinguished: area classes.Conceptual description The following two sections describe the various classes of trachytopes distinguished and the way in which they are combined. line classes and point classes. the effect on the flow resistance should be taken into account. Deltares 271 . The point class (type range 251–300) represents a set of point flow resistance elements. they are generally the dominant roughness factor. FormNr Name Formula Special classes (1–50) 1 flood protected area area fraction shows up as fb in Eqs. Special classes (1–50) In addition to the three base classes two special trachytope classes have been defined: a flood protected area and a composite trachytope class. respectively.11. however.350 fraction α of type T1 and fraction β (generally β = 1 − α) of type T2 2 composite trachytope Area trachytope classes (51–200) The class of area trachytopes is subdivided into three types: simple (51–100). The first class represents a sub-grid area that is protected from flooding and thus does not contribute to the bed roughness. it will add anisotropy to the roughness field. Four simple area trachytopes have been implemented representing the four standard roughness types of flow module. 9. alluvial (101– 150) and vegetation (151–200). FormNr 51 52 53 54 Name White Colebrook value ´ value Chezy Manning value z0 value Formula k C √ C = 6 h/n k = 30z0 Six alluvial trachytopes have been implemented. therefore.2.347 and 9. The second class can be used to make derived trachytope classes that are a combination of two other trachytopes: an area fraction α of trachytope type T1 and an area fraction β (often equal to 1 − α) of trachytope type T2 . The line classes (type range 201–250) may be used to represent hedges and similar flow resistance elements.

The Van Rijn (1984c) alluvial roughness predictor reads k = k90 + 1.307 to 9.296) it is obtained from Equation 9.Delft3D-FLOW.300) where h is the local water depth and the transport stage parameter T is given by T = 2 u2 ∗ − u∗.301) where u ∗ is the bed shear velocity given by 2 u ∗ = gu2 /Cg.296 Equation 9.10 0.298 by noting that hb ∝ h0.7 and Lb ∝ h and ignoring the grain related roughness.cr u2 ∗.312 The first alluvial roughness formula is a simplified version of the Van Rijn (1984c) alluvial roughness predictor k = Ah0. The second formula implemented is a straightforward general power law C = AhB (9.7 1 − e−Bh −0.90 = 18 10 log(12h/k90 ) and k90 = 3D90 and u∗.304)  0.013D∗    0.310 Equation 9. User Manual FormNr 101 102 103 104 105 106 Name simplified Van Rijn power relation Van Rijn (1984c) Struiksma bedforms quadratic bedforms linear Formula Equation 9.3h D50 h 0.24/D∗    −0. The parameters A and B can be calibrated by the user.29  0.311 Equation 9.055 272 if D∗ ≤ 4 if 4 < D∗ ≤ 10 if 10 < D∗ ≤ 20 if 20 < D∗ ≤ 150 if 150 < D∗ (9.11h Lb = 7.04D∗ θc =  0.cr is the critical bed shear velocity according Shields given by (9.299) (9.306 Equations 9.3 1 − e−T /2 (25 − T ) (9.3 (9.cr = g ∆D50 θc given (9.298 to 9.303) u2 ∗.64  0  .302) where Cg.305) Deltares .297) where A and B are calibration coefficients.1hb 1 − e−25hb /Lb where the bedform height hb and length Lb are given by (9. 90 2 (9.14D∗ −0.cr (9.297 Equations 9.298) hb = 0.

313 – 9.mr + ks. mega-ripples kmr and dunes kd .307) C90 = A1 and 10 log(A2 h/D90 ) (9.328. but uses a linear addition h k = min(ks.319.11. ) 2 (9.d .313 – 9. CD = 1. 9.r s. 9. it reads 1 1 1 = (1 − ξ ) 2 + ξ 2 2 C C90 Cmin where (9. These roughness heights are based on Van Rijn (2007) formulae as described in Section 9.d .309) which varies from 0 at θg ≤ θc to 1 at θg = θm where θg = u2 2 ∆D C90 50 (9.311) It uses the roughness heights of ripples kr .) allows for a lot of adjustments.308) ξ= 2 −θ θ max(0. these formulae depend on D50 and D90 data which may be either specified as part of the roughness type or obtained from the morphology module. comm. Cmin are coefficients that the user needs to specify. 9.331 273 . FormNr 151 152 153 154 Deltares Name Barneveld 1 Barneveld 2 Baptist 1 Baptist 2 Formula Eqs. θg − θc ) θm c g θm − θc ( θm − θc ) θg (9. pers.1 combined with trachytope number 106.330 and 9. This formula requires also D50 and D90 data from the morphology module.11. (1996. 9. θc . Two formulae (referred to as ‘Barneveld’) are based on the work by Klopstra et al. 9.322.1.312) Four vegetation based area trachytopes have been implemented. The sixth formula is similar.327 Eqs. It does not include the advective and relaxation behaviour that is available by explicitly simulating the dune height as described in Section 9.306) This predictor does not contain any calibration coefficients but requires D50 and D90 data from the morphology module. A2 .mr + ks.326 and 9.r + ks. The fifth formula is based on Van Rijn (2007) and reads h 2 2 + k2 k = min( ks.325 Eqs. 1997) and two on the work by Baptist (2005). 9. θm .Conceptual description where D∗ = D50 g∆ ν2 1/3 (9.323 – 9.65 Eqs. The second alluvial roughness predictor proposed by (Struiksma. ) 2 (9.310) and A1 .

320) √ u2 v0 = 2g CD n (9.315 is given by α = max(0. (1997) reads C= 1         √2 2A C3 ehv uv0 √ 2A √ 2A + u2 v0 + √ 2A +u2 −u )( v0 v0 √ 2A +u2 +u )( v0 v0 h3/2        ln ( ( C3 ehv C3 e2 √ √ C3 +u2 v 0 +uv 0 ) √ 2 C3 +uv0 −uv0 ) h−(hv −a) z0         + − a ln a z0 g (h−(hv −a)) κ (h − (hv − a)) ln − (h − hv ) (9.315) 1+ a= and 1+ 2 κ2 (h−h ) 4E1 v g 2 κ2 2E1 g (9. hv is the vegetation height.321) and i is the water level gradient.313)        where A= nCD 2α 2 g ( h − hv ) √ 2A (9. For the first implementation the parameter α in Equation 9.001. and n = mD where m is the number of stems per square metre and D is the stem diameter. User Manual The formula by Klopstra et al.316) z0 = ae−F where (9. h is the water depth. 0.314) C3 = α 2A(ehv √ + e−hv √ 2A ) (9. For emerged vegetation the first implementation reads 1 CD nh = 2 C 2g 274 (9.322) Deltares .317) √ E1 = 2 and 2AC3 ehv √ √ 2A (9.01 hhv ) and the velocity within the vegetation is approximated by uv0 i where (9.319) Here.Delft3D-FLOW.318) C3 ehv 2A + u2 v0 κ F = C3 ehv √ 2A + u2 v0 g (h − (hv − a)) (9.

the presence of vegetation will lead to a higher bed roughness and thus to a higher bed shear stress and larger sediment 2 transport rates in case of morphological computations. The drawback of the three vegetation based formulations above is that they parameterize the flow resistance by means of the bed roughness. for second implementation you’ll also need to specify the drag coefficient CD and the alluvial bed roughness kb (Cb in Equation 9. The second term goes to zero at the transition from submerged to emerged vegetation.65). The first implementation of the roughness predictor by Baptist reads for the case of submerged vegetation C= 1 1 2 Cb √ CD nhv 2g + + g h ln( ) κ hv (9.325 is computed as 18 10 log(12h/kb )). (1996) is identical except for the following modifications to Eqs. At that transition the formula changes into the formula for non-submerged vegetation which reads C= 1 1 2 Cb + CD nh 2g (9.327) which is identical to the non-submerged case of the second implementation of the work by Klopstra et al. However.7 α = 0.322.315 is now given by .323) and the velocity within the vegetation is approximated by uv0 i where √ u2 v0 = hv CD hv n 2g + 1 2 Cb (9. 9. Therefore. we can split the terms such that the bed shear stress computed using Deltares 275 . For the case of non-submerged vegetation h < hv the flow resistance and bed roughness are strictly separated C = Cb and λ = CD n (9. The first implementation requires vegetation height hv and density n as input parameters (the drag coefficient CD is equal to 1.0227h0 v (9.Conceptual description The second implementation based on Klopstra et al. The parameter α in Equation 9. The main difference between the two implementations is the inclusion of the roughness Cb of the bed itself (without vegetation). (1996) (see Equation 9. For emerged vegetation the second implementation reads CD nh 1 1 = + 2 2 C 2g Cb (9.325) For large values of Cb the latter two equations simplify to the corresponding equations of the first implementation. Consequently. we have included a − λ 2u term in the momentum equation where λ represents the flow resistance of the vegetation.326) where n is the vegetation density (n = mD where m is the number of stems per square metre and D is the stem diameter).328) In the case of submerged vegetation h > hv the two terms can’t be split in an equally clean manner.324) and i is the water level gradient.325).320 – 9.

Linear trachytope classes (201–250) Two formulae have been implemented for linear trachytopes such as hedges or bridge piers.335 to 9. The second implementation reads CD nLhedge h 1 = 2 C 2gLcell Wcell 276 (9. Wcell and Lcell are the width and length of the grid cell.335) Deltares . The second ratio is thus the inverse of the average distance between these hedges within the grid cell.326 as (9.332) where Lhedge is the projected length of the hedge.328 for h = hv .331) which simplify to Equation 9.337 The first implementation reads 1 h Lhedge 1 − µ2 = C2 2g Wcell Lcell µ2 (9. density n. 9.329) √ C = Cb + and 2 g CD nhv Cb h ln( ) 1 + κ hv 2g (9.333) if the hedge extends above the water level (hv > h) and is given by µ = 1 − 0. which is determined by the hedge pass factor µ given by µ = 1 + 0.Delft3D-FLOW.334) if the hedge is fully submerged (h > hv ) where n is a dimensionless hedge density.175n h −2 hv (9. u2 u2 v = 2 C2 Cb With this additional requirement we can rewrite Equation 9.332 to 9. Both formulae by Baptist require vegetation height hv .175n h hv (9.330) λ = CD n 2 hv C b h C2 (9. The ratio Lhedge /Wcell may be interpreted as the number of hedges that the flow encounters per unit width. User Manual the depth averaged velocity u and the net bed roughness C equals the bed shear stress computed using the velocity uv within the vegetation layer and the real bed roughness Cb . FormNr 201 202 Name hedges 1 hedges 2 Formula Eqs. drag coefficient CD and alluvial bed roughness Cb as input parameters. The last term may be loosely referred to as the drag of the hedge. 9.334 Eqs.

The generalization of Equation 9. The formula is identical to Equation 9. It may be used to represent groups of individual trees or on a smaller scale plants.75). Note that the formula for computing the loss coefficient for a bridge explicitly includes the reduction in the flow area and the resulting increase in the effective flow velocity. FormNr 251 Name trees Formula Eqn.338 to the submerged case (h > hv ) lacks the extra term in Equation 9.326. roughnesses are accumulated as White Colebrook roughness values and as Chezy for the latter values both the linear sum (“parallel”) and the sum of inverse of squared values Deltares 277 .2.338) where n = mD with m the number of trees per unit area and D the characteristic tree diameter.i (9. Point trachytope classes: various (251–300) One formula for point trachytopes has been implemented. We recognize the same ratio Lcell Wcell /Lhedge that represents the average distance between hedges. These ´ values.322).335 can be directly compared to similar equations for area trachytopes (Equation 9.337) for submerged conditions.338 The implemented formula reads C= 2g CD n min(hv .Conceptual description or equivalently C= 2gLcell Wcell hLhedge 1 CD n (9.340) 1 2 = Clin 1 i 2 Clin .2 Averaging and accumulation of trachytopes ´ Point and linear roughnesses are accumulated by summing the inverse of the squared Chezy values Ci . 1 2 Cpnt = i 1 2 Cpnt .336) for non-submerged conditions and C=  2gLcell Wcell  hv hLhedge h 1 CD n + m0 1 h−hv 2 h  hv 2 − h− h  (9.339) (9.11.338) and bridge resistance (Equation 10. whereas the above mentioned trachytope formulae don’t. 9.i The area roughnesses are accumulated weighted by the surface area fraction fi . Equation 9.327 except for the fact that the point trachytope formula has no bed roughness and area associated with it. 9. h) (9. point trachytopes (Equation 9. hv is the vegetation height and h is the local water depth.

12 − 0. 0. 151 and 201 corresponds to the NIKURADSE option of the WAQUA/TRIWAQ flow solver.344) The final effective bed roughness of the grid cell may be computed by either one of the following two methods.843. This method together with trachytope classes 1.348) Deltares .342) (9.25fb − 0.345) where klin = 12h10−Clin /18 and kpnt = 12h10−Cpnt /18 .p = i fi Ci For the fraction of the grid cell area for which no roughness class is specified the default roughness is used.p 278 (9. km = karea + klin + kpnt (9. fb = max(min(0.99 fb (9. Roughness values are converted into each other as needed based on the local water depth using Eqs.s = 1 Ci2 Carea . Method 1 The total mean roughness is computed by summing the White Colebrook values for the areas and line and point resistance features.54–9. karea = i fi ki fi i (9. fb ).341) (9. 9.347) The resulting Ctotal value is used in the computation. λ= i fi λi (9. 51. For the fraction of the grid cell area for which no flow resistance is specified.014) Ctotal = Cm 1. 101.56. The effect of the water free area fraction fb is taken into account by means of the following empirical relation in which Cm = ´ value corresponding to the total mean White–Colebrook 18 10 log(12h/km ) is the mean Chezy roughness value obtained from Equation 9.Delft3D-FLOW. The flow resistance coefficients are also accumulated proportionally to the surface area fraction fi associated with the trachytope considered.346) (9.s + (1 − αs )Carea . User Manual (“serial”) are computed.345. Method 2 The total roughness is computed by first averaging over the serial and parallel averages of the ´ values according Chezy Carea = αs Carea .343) 1 2 Carea . obviously none is used.

0. 53.843.014) Carea . the inverses of the squared Chezy 1 1 1 1 = 2 + 2 + 2 2 Carea .corr Ctotal Clin Cpnt (9. fb ).corr =Carea 1.347 of method 1). 51.350) ´ value representing the total bed roughness is computed by accumulating Finally the Chezy ´ values. 52. fb = max(min(0. Deltares 279 .351) The resulting Ctotal value is used in the computation.Conceptual description where αs = 0. This method together with trachytope classes 1. 101. Subsequenty the effect of the water free area fraction fb is taken into account by means of the following empirical relation (identical to Equation 9. 152.99 fb (9. 202 and 251 corresponds to the ROUGHCOMBINATION option of the WAQUA/TRIWAQ flow solver.6 by default.349) (9.25fb − 0.12 − 0.

Delft3D-FLOW. User Manual 280 Deltares .

Delft3D-FLOW is applied for modelling a wide range of different flow conditions: e. The geometrical quantities Gξξ and Gηη introduced in the transformed equations.1 Staggered grid The numerical method of Delft3D-FLOW is based on finite differences.10 Numerical aspects of Delft3D-FLOW The mathematical formulations of the 3D shallow water flow and transport model Delft3DFLOW were presented in Chapter 9. robustness had high priority. To discretise the 3D shallow water equations in space.2 and Figure 10. time integration and numerical solution methods applied in Delft3D-FLOW will be described. 10. Staggered grids have several advantages such as: Boundary conditions can be implemented in a rather simple way. w) describe the flow. This particular arrangement of the variables is called the Arakawa C-grid. the model area is covered by a curvilinear grid. Stelling (1984). v . In both cases a curvilinear grid. a file with curvilinear grid co-ordinates in the physical space. The primitive variables water level and velocity (u. Deltares 281 . To solve the partial differential equations the equations should be transformed to the discrete space. Such a file may be generated by a grid generator. For all these applications (with complete different length scales) Delft3D-FLOW should give a solution. In this chapter the space discretisation. Numerical aspects that are specifically valid only for the Z -model system will be described separately in Chapter 12. density driven flows by thermal discharges. The water level points (pressure points) are defined in the centre of a (continuity) cell. wind driven flows in lakes and ocean dynamics. The grid co-ordinates can be defined either in a Cartesian or in a spherical co-ordinate system. The velocity components are perpendicular to the grid cell faces where they are situated.3 to 9. has to be provided. see the User Manual of Delft3D-RGFGRID.8. It is possible to use a smaller number of discrete state variables in comparison with discretizations on non-staggered grids. The pattern is called a staggered grid. see e. have to be discretised on the computational grid. 9. Staggered grids for shallow water solvers prevent spatial oscillations in the water levels. rapidly varying flows in rivers. To discretise the 3D shallow water equations.g. see Eqs. to obtain the same accuracy. see Figure 10. It is assumed that the grid is orthogonal and well-structured.2. see Figure 10. The numerical grid transformation is implicitly known by the mapping of the co-ordinates of the grid vertices from the physical to the computational space.g. see Figure 10. tidal flow in estuaries and seas. For the choice of the numerical methods. The description provided here applies in most cases for both the vertical σ co-ordinate system (σ -model) and the vertical z co-ordinate system (Z -model). turbulent flows in laboratory flumes. the variables are arranged in a special way on the grid.1.3.

v or w) Figure 10.Delft3D-FLOW.3: Grid staggering.1: Example of a grid in Delft3D-FLOW Figure 10. 3D view (left) and top view (right) 282 Deltares . User Manual Figure 10.2: Mapping of physical space to computational space Legend: + → water level (ζ ) / density (ρ) point velocity point (u.

Numerical aspects of Delft3D-FLOW 10. There are two types of boundaries: closed boundaries along “land-water” lines (coastlines. (Mmax. This allows for more resolution in the zones of interest such as the near surface area (important for e. riverbanks) and open boundaries across the flow field. Nmax).3 Definition of model boundaries The horizontal model area is defined by specifying the so-called computational grid enclosure (automaticly enerated by Delft3D-RGFGRID). For the σ co-ordinate grid.g.4 is defined by the following polygon of line pieces: Deltares 283 . High resolution around pycnocline is in this situation preferred. the grid enclosure and the position of open and closed boundaries.5 shows the corresponding grid of control volumes which may be generated by the grid generator Delft3D-RGFGRID. Figure 10. The Z -grid is not boundary-fitted in the vertical. If a computational grid enclosure is not specified. wind-driven flows. The polygons consist of line pieces connecting water level points on the numerical grid. recently a second vertical grid co-ordinate system based on Cartesian co-ordinates (Z -grid) was introduced in Delft3D-FLOW for 3D simulations of weakly forced stratified water systems. will generally not have enough resolution around the pycnocline which is strictly horizontal in the physical space. A default enclosure is spanned by the lines connecting the water level points (1. (Mmax. The Z -grid has horizontal co-ordinate lines that are (nearly) parallel with density interfaces (isopycnals) in regions with steep bottom slopes. with a direction parallel to the grid lines or diagonal (45 degrees) through the grid. 1).4 shows for a horizontal model area the grid staggering. However. the number of layers over the entire horizontal computational area is constant. occasionally this grid may not be sufficient to solve problems where stratified flow can occur in combination with steep topography. The computational grid enclosure for the grid of control volumes shown in Figure 10. irrespective of the local water depth. then a default rectangular computational grid is assumed. The distribution of the relative layer thickness is usually non-uniform. The open boundaries are artificial and chosen to limit the computational area. heat exchange with the atmosphere) and the near bed area (sediment transport). The σ -grid is commonly used in Delft3D-FLOW. Figure 10.2 σ -grid and Z -grid In the vertical direction. The bottom (and free surface) is usually not a co-ordinate line and is represented as a staircase (see previous chapter). Nmax) and (1. two types of grid are available: a boundary-fitted σ co-ordinates and a grid that is strictly horizontal also called the Z -grid. The computational cells on the grid enclosure are land points (permanent dry) or open boundary points. This is important to reduce artificial mixing of scalar properties such as salinity and temperature. The σ -grid. The computational grid enclosure consists of one or more closed polygons that specify the boundaries of the model area. 1). This example will now be explained in detail. Therefore. though boundary fitted. An island may be removed from the computational domain by specifying a closed polygon (“land-water line”) as part of the grid enclosure. 10.

3) (8. respectively.n)=(1.3) (8.6) (1.1) (7. with Mmax-2 and Nmax-2 the number of grid cells in ξ .6) (1.and η -direction.n)=(8.5).4: Example of Delft3D-FLOW model area (1. which can be used by Delft3D-FLOW.4.1) (7.2. After generating a curvilinear grid with the grid generator Delft3D-RGFGRID .3) and at (m. User Manual 6 5 4 3 2 1 1 2 3 4 5 6 7 8 Closed boundary Grid enclosure Water level point Open boundary point u-velocity point v-velocity point Grid cell/Computational cell/Control volume Figure 10. the grid enclosure and the location of boundary points see Appendix E.6) (3. as output file there is also a computational grid enclosure file. We remark that the dimensions of the grid in Delft3D-RGFGRID are Mmax-1 and Nmax-1.Delft3D-FLOW. 284 Deltares . the grid lines and corresponding control volumes drawn by the grid generator Delft3D-RGFGRID are shown in Figure 10.4) (1.1) The array dimensions Mmax=8 and Nmax=6. Open boundaries are located at the line pieces (m. counting the grid corners.3) (8. For a detailed discussion of the grid. in both directions the dimension of the grid in Delft3D-RGFGRID is one less than the dimension specified for the model area in the MDF-file.3) (8. This has to do with the staggered grid and is often a source of misunderstanding.5.4) - (5.2) to (1. see <∗. For this example.dep> files and Appendix A.4) to (8. Thus.6) (3.1) (5. We remark that the depth files (files with extension <dep>) should have entries for all Mmax by Nmax values.

For the linearised depth-averaged shallow water equations without advection and bottom friction. a robust solver for the shallow water equations has to satisfy the following demands: Robustness (unconditionally stable). However.2) U = (u. this is not acceptable. Accuracy (at least second order consistency). g is the acceleration of gravity.5) 285 .4 Time integration of the 3D shallow water equations Following Stelling (1984).5: Example of Delft3D-FLOW grid 10.1) where ∆t is the time step.Numerical aspects of Delft3D-FLOW Figure 10. Computationally efficient. 2 ∆x ∆y 2 (10. an implicit method is needed. For many practical applications. this requires a time step of only a few seconds to simulate tidal propagation. An explicit time integration of the shallow water equations on a rectangular grid is subject to a time step condition based on the Courant number for wave propagation: CFLwave = 2∆t gH 1 1 + < 1. Exceeding the time step would generate instability and from the view of robustness. ζ )T . the Crank Nicholson method in vector form reads: U with: +1 −U 1 1 + AU + AU ∆t 2 2 +1 = 0. an implicit scheme can be uneconomic in computer time and storage if the inversion of a large matrix is required. 0 (10. A straightforward implicit finite difference approximation is the Crank-Nicholson method.3) 0  f A= ∂ H ∂x and:  −f 0 ∂ H ∂y  ∂ g ∂x ∂  g ∂y . v. Suitable for both time-dependent and steady state problems. Therefore. Deltares (10.and η -direction of the physical space. (10. (10.4) H = d + ζ. H is the total water depth and ∆x = Gξξ and ∆y = Gηη are the smallest grid spaces in ξ .

4. We discuss the time integration for scheme WAQUA and method Cyclic. 10. Solving the equations would require a large computational effort. For the complete time step.7) 286 Deltares .Delft3D-FLOW.1 ADI time integration method Leendertse (1967). The second scheme is denoted as the Cyclic method.6) After discretization. The third scheme can be applied for problems that include rapidly varying flows for instance in hydraulic jumps and bores (Stelling and Duinmeijer. back substitution of the water levels in the discrete momentum equations gives the velocities. For the Flooding scheme. each separate term is still integrated second-order accurate in time. For the water level gradient and the advection terms. 2003). if in one stage a term is taken implicitly in time. The integration of the advection term is explicit and the time step is restricted by the Courant number for advection. The WAQUA-scheme and the Cyclic method do not impose a time step restriction. After computing the water levels. see Stelling (1984) for full details. the advection term is taken at the previous time level and moved to the right-hand side. Substitution of the discrete momentum equations in the continuity equation leads to a tri-diagonal system of equations for the water levels. this term will be taken explicitly in time in the other stage. For the spatial discretization of the horizontal advection terms. which is applied for the σ -grid in Delft3D-FLOW. Leendertse et al. The time integration is based on the ADI-method. The scheme is denoted as the Flooding scheme and was developed for 2D simulation with a rectilinear grid of the inundation of dry land with obstacles such as road banks and dikes. the time levels are alternating. In both stages. 2 (10. three options are available in Delft3D-FLOW. Leendertse and Gritton (1971). In the next section a computationally efficient time integration method will be described. The ADI-method splits one time step into two stages. the spatial differential operators in both grid directions generate a system of difference equations with a band matrix. The band width is relatively large and all grid points are implicitly coupled. corresponding to the grid cells surrounding the water level point. all the terms of the model equations are solved in a consistent way with at least second order accuracy in space. see Stelling (1984) and Stelling and Leendertse (1992). In each row there are at least five non-zero entries. Each stage consists of half a time step. see Stelling and Leendertse (1992). User Manual The solution after one time step is given by U +1 = 1 I + ∆tA 2 −1 1 × I − ∆tA U . The first and second option use higher-order dissipative approximations of the advection terms. (10. The first scheme is denoted as the WAQUA-scheme. In vector form the ADI-method is given by: Step 1: U +1 2 −U 1 2 ∆t 1 + Ax U 2 +1 2 1 + Ay U + B U 2 +1 2 = d. (1973) introduced an Alternating Direction Implicit (ADI) method for the shallow water equations. The advantage of the ADI-method is that the implicitly integrated water levels and velocities are coupled along grid lines leading to systems of equations with a small bandwidth.

followed by the U -momentum equation.7. Eq. 1 and the simulation time from t = ∆t In the first stage the time level proceeds from to + 2 1 to t = + 2 ∆t. 2νV 2νV Deltares (10. To improve stability the bottom friction is integrated implicitly for each stage. (10. In the second stage the time level proceeds from + 1 2 to + 1. the advection terms and viscosity terms are integrated explicitly.Numerical aspects of Delft3D-FLOW Step 2: U +1 −U 1 2 ∆t 1 +2 1 + Ax U 2 +1 2 1 + Ay U 2 +1 + BU +1 = d. followed by the V -momentum equation which is implicitly coupled with the continuity equation by the free surface gradient.11) with λ the linearized bottom friction coefficient. 0 0 λ  (10. 9. is solved. only the sign of the Coriolis term is dependent on the computational direction. ∂ v ∂y and:  λ 0 0 B =  0 λ 0 . In this stage first the U -momentum equation is solved. In Delft3D-FLOW the similarity of the difference equations is used in the implementation. d is the right-hand side containing external forcings like wind and atmospheric pressure. The grid coefficients.10) ∂  g ∂y . the advection terms and viscosity terms are integrated implicitly. water level gradient) is integrated implicitly. The time integration of the horizontal viscosity terms is discussed in Section 10. Eq.e. The only principal difference between the u. The second stage in the ADI-method is almost similar to the first stage. by the free surface gradient. water level gradient) is integrated explicitly. direction dependent roughness coefficients and the u-velocity and v -velocity are interchanged. Similarly in the stage.3. ∂ u ∂x (10.9) ∂ ∂ u ∂x + v ∂y  f Ay =  0 0 0 ∂ H ∂y 0  (10. 9.and v -momentum equations is the sign of the Coriolis term. in which the barotropic pressure term (i. For the 3D shallow water equations the horizontal velocity components are coupled in the vertical direction by the vertical advection and viscosity term. The same subroutine is used for the momentum equation in both directions. 9.4. In this stage first the V -momentum equation. In the stage. An explicit time integration of the vertical exchange terms on a σ -co-ordinate or a Z -co-ordinate grid would in shallow areas lead to very severe time step limitations: (∆σH )2 (∆z )2 ∆t ≤ or ∆t ≤ .8. in which the barotropic pressure term (i. Eq. which is implicitly coupled with the continuity equation.8) with: 0 Ax =  0 ∂ H ∂x   −f ∂ ∂ u ∂x + v ∂y 0 ∂  g ∂x 0 .12) 287 .e.4 and is dependent on the formulation.

13) ω w where ∆σH and ∆Z are the vertical grid size and ω is the transformed vertical velocity and w ∆t ≤ the vertical velocity. (1982) and Weare (1979). The wave propagation is related to the Courant number. Therefore. which is first-order accurate in time and leads to tri-diagonal systems of equations. the ADI-method is inaccurate for Courant numbers larger than 4 2. see Eq. and through “zig-zag”. (10. Water levels and velocities are implicitly solved along grid lines.4. see Figure 10.6. Benque boundaries. You are advised to carry out sensitivity tests in order to determine the largest time step for which the ADI-method still yields accurate results. see e. see Figure 10.Delft3D-FLOW. It is known. User Manual P Figure 10.1.2 Accuracy of wave propagation The standard time integration method in Delft3D-FLOW for the σ -model is the ADI-method described in the previous section. flow over tidal flats.6. “Zig-zag channel” ∆σH ∆z or ∆t ≤ . This inaccuracy is called the ADI-effect and is introduced by the splitting of the spatial operator in two directions. namely in case of a narrow channel (width of few grid sizes) that makes an angle of 45 degrees with the computational grid. this is a rough estimate. in the vertical direction a fully implicit time integration method is applied. A free surface wave cannot travel through more than two bends of 90 degrees in one complete ADI time step (Stelling. √ 288 Deltares .6: Numerical region of influence for one time step.g. The numerical region of influence does not correspond to the mathematical region of influence following the theory of characteristics for hyperbolic systems of equations. However. 1984). The upper bound for the Courant number of 4 2 occurs in the most critical situation. that for flow along irregular (staircase) closed e. 10. In practical situations the Courant number should not exceed a value of 10. see ´ et al. 10. flow around islands. The ADI-method may lead to inaccurately predicted flow patterns. Stelling (1984). The vertical coupling of the discretized momentum equations is eliminated by a double sweep algorithm.g. channels where the grid lines do not smoothly √ follow the geometry.

the water depth at velocity points may be discontinuous. Eq.23–9.n This factor converges to one.3 Iterative procedure continuity equation The non-linearity in the coupled continuity equation. 9.16) Hm.26 and 9. The iterative procedure in Delft3D-FLOW is mass conservative after each iteration. the complete Reynolds stress tensor is used.1.25. For rapidly varying flows.4. In this way positive water depths are guaranteed. 9.1 the horizontal turbulent fluxes of momentum are described.3.n Um+21 . no slip at the closed boundaries.Numerical aspects of Delft3D-FLOW 10. Deltares 289 . This procedure is unconditionally stable.15) The iterative procedure in Delft3D-FLOW removes the non-linear term by multiplying the free surface gradient in the discretized momentum equation by the factor: r [q ] = Hm. Two iterations are enough both for accuracy and stability.n 2 2 [q ] − ∆ξ Gηη Hm− 1 .n 2 2 [q ] . This approach is less stable in shallow areas.17) This linearization of the continuity equation is used in combination with the Flooding scheme for advection and the Z -model.4 Horizontal viscosity terms In Section 9. For the U -momentum equation: ∂ ∂ξ Gηη HU +1 2 Gηη HU ≈ ξ [q ] m+ 1 . see Section 10.n 2 2 ξ +1 − ∆ξ Gηη Hm− 1 . 9. 10.14) [q ] ≈ Gηη Hm+ 1 . (10. for the following cases: partial slip at closed boundaries. (10. If the horizontal viscosity terms are simplified.27. The number of iterations is set in the MDF-file by the keyword Iter. Eqs.4. In this case. Eqs.n Um+ 1 .5. this term is integrated fully implicitly by using operator splitting. The same approach is used for the Z -model: ∂ ∂ξ Gηη HU +1 2 ≈ Gηη Hm+ 1 . Eqs.n 2 ξ +1 2 − ∆ξ Gηη HU m− 1 . there results a Laplace operator along grid lines.3. To obtain accurate approximations of the local water depth at discontinuities slope limiters are used. This requires local linearization of the continuity equation.7 and 9.n 2 ξ ξ +1 2 (10.n Um− 1 . In the momentum equations. 9.n 2 2 ξ +1 (10.n Um−21 .8 requires an iterative procedure. and momentum equation. The u-momentum equation involves only second-order derivatives of the u-velocity.n [q ] [q −1] .

∆y ) ≤1 HLES-model for sub-grid viscosity. which leads to the following additional stability condition: ∆t ≤ 1 2νH 1 1 + 2 ∆x ∆y 2 −1 .4. Tξη in the depth averaged U -momentum equation Eq. For the complete Reynolds stress tensor the shear stress τξη in the u-momentum equation contains derivatives of the v -velocity.Delft3D-FLOW. The Reynold stresses are integrated explicitly.5 Spatial discretizations of 3D shallow water equations 290 Deltares . Stability baroclinic mode internal wave propagation (Z -model only) Stability horizontal viscosity term (HLES.5 Overview time step limitations In this section we give an overview of the time step limitations due to stability and accuracy for the time integration of the shallow water equations in Delft3D-FLOW. The stresses due to secondary flow Tξξ . (10. velocity scale. In case of a curvilinear grid ∆x = Gξξ and ∆y = Gηη . ∆y ) with β defined in Equation 9. Let ∆x and ∆y be horizontal grid sizes.18) In case of a curvilinear grid ∆x = Gξξ and ∆y = Gηη .and V -velocity components. no slip) Secondary flow (with β defined in Equation 9.155) ∆t ≤ T 40 1 ∆ x2 Cf = 2∆t gH ∆t|u| ∆x + 1 ∆y 2 √ <4 2 <2 (ρbottom −ρtop ) H g4 ρtop 1 ∆ x2 1 ∆y 2 1 ∆ x2 ∆t + 1 ∆y 2 <1 2∆tνH + <1 2∆tβ |u| min(∆x. which leads to the following stability condition: 2∆tβ |u| ≤1 min (∆x. 9. (10.19) 10.1 the time step limitations are given for the shallow water code Delft3D-FLOW. In Table 10. Which of the limitations is most restrictive is dependent on the kind of application: length scale. User Manual Table 10. 10. partial slip.155.1: Time step limitations shallow water solver Delft3D-FLOW Points per wave period T Accuracy ADI for barotropic mode for complex geometries Explicit advection scheme “Flooding scheme” and for the Z -model. These stresses are integrated explicitly. with or without density-coupling etc.147 contain the depth averaged U .

20 and Eq.k + um. This higher order viscosity suppresses the wiggles without smoothing the solution too much. this artificial viscosity dominates the physical viscosity and the computed solution is much smoother than the correct one.n. (Stelling. Central differences are often second order accurate. v ¯m.k            ξη v ¯m.k 2∆η −3um. which are successively used in both stages of the ADI-scheme.n.n−1.k 2∆η .k <0 ξη v ¯m.k Gηη Gηη m.n 3um.n+1. For the cross advection term the spatial discretization is given by: Stage 1: v ∂u Gηη ∂η and Stage 2: = m. the so-called “wiggles” (Gresho and Lee.k − 4um.n.n+2. These wiggles arise in the vicinity of steep gradients of the quantity to be resolved. based on the dissipative reduced phase error scheme. On the other hand. which has the form of a second-order artificial viscosity term (Vreugdenhil.n+1.n. In Delft3D-FLOW the following advection schemes are implemented: 1 2 3 4 WAQUA-scheme Cyclic method Flooding-scheme Multi directional upwind (Z -model only) WAQUA-scheme The first scheme (WAQUA) based on Stelling (1984) is an extension of the ADI-method of Leendertse with a special approach for the horizontal advection terms.k − um.k ξη v ¯m. 1994). It is a splitting of a third order upwind finite difference scheme for the first derivative into two second order consistent discretizations: a central discretization Eq.n um.k ≥0 ξη . Deltares 291 . first order upwinding is unconditionally wiggle-free or monotone. 10. but introduces a truncation error.n. ξη v ¯m. but may give rise to non-physical spurious oscillations. In advection-dominated flows.n.n−2.20) v ∂u Gηη ∂η = m.5.n. monotony and efficiency of the computational method.k m.Numerical aspects of Delft3D-FLOW 10.k 2∆η .21. Spatial oscillations in the velocities are suppressed by means of fourth-order dissipation.1 Horizontal advection terms The choice of the spatial discretization of the advective terms has great influence on the accuracy.n−1.k + 4um. The normal advection ∂u term u ∂u ∂ξ is discretised with central differences and the cross advection term v ∂η . In shallow water flow these wiggles may also be introduced near closed boundaries and thin dams. (10. 10.k Gηη m.n.k − um. Higher order upwinding is not free from numerical oscillations and introduces fourth-order artificial viscosity.n (10.21) The advective terms for the v -momentum equation are discretized in a similar way. 1981) in the solution.n. thus promoting the stability of the solution process.

see Figure 10. For stability the vertical terms are integrated implicitly in both stages. the matrix is not diagonally dominant for large time steps (Courant number advection more than one) and this may lead to bad convergence behaviour. conservation properties become crucial. For sufficiently smooth conditions. The conservation of momentum in the present implementation has been derived only for a Cartesian rectangular grid and depth averaged velocities. 1992). (Cyclic ). For the Cyclic method the matrix is diagonally dominant and the iterative scheme converges well. The system is solved efficiently by a Red Black Jacobi iterative scheme in the horizontal direction and a double sweep in the vertical direction. In the Flooding scheme the bottom is approximated by a staircase of tiles (DPUOPT=MIN).20 and 10. Flooding-scheme The third scheme. In combination with the local invalidity of the hydrostatic pressure assumption. The momentum conservative approximation for the normal advection term u ∂u ∂ξ derived for a Control Volume around a velocity point is given by 292 Deltares . the advection terms are integrated implicitly in the stage of the ADI-method in which the free surface gradient is at the old time level. based on Stelling and Duinmeijer (2003) can be applied to rapidly varying depth-averaged flows for instance the inundation of dry land or flow transitions due to large gradients of the bathymetry (obstacles). centred around the water level points. 1992).21 for both the normal advection term u ∂u ∂ξ and the cross advection term v ∂u ∂η (Stelling and Leendertse. The scheme can also be used in combination with obstacles represented by only one point on coarse grids. The limiter prevents the generation of wiggles. The WAQUA scheme has little dissipation and is used for the accurate prediction of water levels along the Dutch rivers. In flow expansions a numerical approximation is applied that is consistent with conservation of momentum and in flow contractions a numerical approximation is applied that is consistent with the Bernouilli equation. User Manual 1984.Delft3D-FLOW. For the WAQUA scheme. The local order of consistency depends on the solution. is based on the dissipative reduced phase error scheme Eqs. The upwind discretization is used in the stage in which both the horizontal advection and vertical viscosity term are integrated implicitly. The approximations are second-order. 2003). which is the default method. Stelling and Leendertse. and a fine grid size. both approximations converge to the same solution. due to the central difference scheme for the normal advection. but the accuracy reduces to first order near extreme values by the use of the so-called Minmod slope limiter (Stelling and Duinmeijer. For WAQUA and the Cyclic method. For negative flow direction the discretizations are defined accordingly. We describe the discretizations for positive flow direction. Cyclic-scheme The second method.7b. 10. The resulting linear system of equations has eleven diagonals. In the stage in which the free surface gradient is integrated implicitly the advection terms are at the old time level.

27) (10.n m− .28) q |m+ 1 .n m+ .n 2 2 ru = um− 1 .n 2 (10.n − um− 1 .n 2 ∆ξ (10.24) ξ ≥ q ¯m.n u ∆ξ − um+ 1 .n 2 2 For negative flow direction.n 2 ξ ξ ¯m.22) 1 m+ 2 . the limiter is defined accordingly.7: (a) Control Volume for mass for the Flooding scheme.n q ¯m +1.n = qm+ 1 .n m− .n 2 2 2  u 1 + 1 ψ (ru ) u 1 − u 3 m− .n ξ q ¯m. min(ru . (b) Control Volume for momentum in horizontal and vertical direction for the Flooding scheme (using q = Hu): u ∂u Gξξ ∂ξ = 1 m+ 2 .n − um− 3 . (10.25) (10.n 2 2 2 2 2 2 2 (10.n + Hm+1.n q ¯m +1.n = 2 (Hm.n = u 1 + 1 ψ (ru ) u 1 − u 3 m+ .n = u|m+ 1 . 1)) um+ 1 .n ) × m+ 1 .n 1 H Gξξ ∂qu ∂q −u ∂ξ ∂ξ Gξξ (10.n m+ .n − q ξ ξ u ˜m.26) ψ (ru ) = max(0.n 2 2 Hm.n 2 The momentum conservative approximation for the cross advection term v ∂u ∂η is given by Deltares 293 .n + qm− 1 .Numerical aspects of Delft3D-FLOW (a) (b) Figure 10.n <0 u ˜m.23) ξ q ¯m.n + Hm+1.n − q ¯m.n ˜m+1.

n+1 − um+ 1 . 10.n 2 = 2 × (Hm. the limiter is defined accordingly. q2 2 Gξξ 1 (Hm+1.n +Hm+1.n = v ¯ξη 2 m+ 1 .n )2 Energy Head Conservation = 1 Gξξ Hm+1.n− 2 2 2  um+ 1 . see Figure 10.n+2 2 2 2 p ¯ξ <0 m+ 1 . To avoid an artificial boundary layer or instabilities.n < Hm+1.n−1 − um+ 1 . 1)) um+ 1 .n 2  ξ  ξ  p ˜m+ 1 .n− 1 = 2 2  1 um+ 1 .36.n ∧ ∧ − 1 Hm.n ) λ= 1 (Hm.n (10. 294 Deltares .n− 1 u ˜m+ 1 .n + Hm+1.n ) Gηη m+ 1 . User Manual (using p = Hv ): v ∂u Gηη ∂η = 1 m+ 2 .n ) 2 − 1 (Hm.29) m+ 1 . min(ru .n−2 2 ≥0 p ¯ξ m+ 1 .n− 1 2 2 pm.n− 1 2 (10.7b can be derived under steady state conditions (constant discharge q ) in 1D (along a streamline in 2D).n 1 H Gηη ∂pu ∂p −u ∂η ∂η (10. Stelling (1984) developed the numerical boundary treatment implemented in Delft3D-FLOW. Near the boundaries the higher order discretization stencils for the advection terms contain grid points on or across the boundary.31) 2 2 u ˜m+ 1 .n 2 (10.35) From the momentum conservative formulation a so-called energy head conservative discretization for the same Control Volume. p|m+ 1 .n−1 − um+ 1 .Delft3D-FLOW.32) ψ (ru ) = max(0.36) q>0 q>0 Hm.n )2 4Hm.n Hm.n − um+ 1 .37) The energy conservative discretization is applied for contractions in both directions.n−2 2 2 (10.n−1 2 2 ru = um+ 1 .n− 1 2 2 (10.n+1 + 2 ψ (ru ) um+ 1 .n  2  ∆η 2 2 (10.n+ 1 − p 1 1 m+ 2 . For 2D flow the direction of the grid lines do not always coincide with streamlines and this will generate small head losses.n+ 1 − p ¯m+ 1 .n (10.n− 1 + pm+1.n+ 1 u 2 2 2 2 2 2 2 2  −  ∆η   ξ  ¯ξ p ¯m+ 1 .n > Hm+1.n Hm+1.n 2 Hm. the discretizations are reduced to lower order discretizations with smaller stencils. see Eq.n + Hm+1.n Momentum Conservation 2q 2 λ (Hm.34) For negative flow direction.n−1 + 2 ψ (ru ) um+ 1 .n− 1 2 2 1 u 1 m+ .n + Hm+1.33) (10.n− 1 ¯m+ 1 .30) p ¯ξ = m+ 1 .

3 Viscosity terms The approximation of the vertical viscosity terms are based on central differences. It is an extension to two dimensions of the first-order upwind method.n and H the total water depth or hm.k + hm. Deltares 295 .n.8: Layer numbering in σ -model Multi directional upwind (Z -model only) The fourth scheme.k = ∆zm. with hm.38) where hm.n.k denotes the thickness of the computational layer with index k defined by hm.n.n. For the space discretization of the vertical advection term.5.n.k in the Z -model.n. In a shallow water model the horizontal length scale is much larger than the vertical length scale.n.n.5.n.n.k − 1 2 um. 10. Only for stratified flows where the turbulent exchange is reduced.n. advection may be dominant.k−1 + hm.k−1 − um.k νV |m.n.n.k m.k um. a fully implicit time integration is used for the vertical exchange terms.n.n.k in the Z -model.k+1 1 1 2 hm.k − um. In the vertical direction the eddy viscosity term dominates the advection term.n. This results in tridiagonal systems of equations in the vertical. (10. The σ co-ordinate system can lead to very thin layers in shallow areas. To prevent instabilities.n.k (hm. However there is the Courant number stability constraint.k ) 1 2 + . which is only available and at the same is at present the only advection scheme in the Z -model is a multidirectional upwind scheme.k+1 ) The vertical eddy viscosity is computed at the layer interface.k−1 − um. The vertical viscosity term in the u-equations is discretized as: 1 ∂ H 2 ∂σ νV ∂u ∂σ = m.n.k = ∆σk Hm.39) νV |m.k = ∆zm. (10.2 Vertical advection term The horizontal velocities of adjacent vertical layers are coupled by the vertical advection and the vertical viscosity term.n.k = ∆σk Hm.k um.k hm.n. The method is described by Bijvelds (2001).n. 10.k+1 .n.n.k−1 hm. a second order central difference is used: ω ∂u H ∂σ ξσ =ω ¯ m.k−1 + hm. It is a positive and monotone scheme.nk+1 (hm.k + 2 hm.n.n or hm.n.n.Numerical aspects of Delft3D-FLOW hk-1 hk uk-1 uk layer k-1 layer k hk+1 uk+1 layer k+1 Figure 10.

41) For the 3D transport equation the scalar concentrations are coupled in the vertical direction by the vertical advection and diffusion term. ω (10. In case of a local minimum or maximum the first order upwind scheme is applied.42) (10. Monotony (positive solution). The vertical coupling of the discretized transport equations is removed by a double sweep algorithm. Computationally efficient. It combines two numerical schemes. (10.6 Solution method for the transport equation A robust and accurate solver for scalar transport has to satisfy the following demands: Mass conservation. Suitable for both time-dependent and steady state problems. 1974). For the spatial discretization of the horizontal advection terms. ∆x ∆y ≤ 1. whereas the upwind scheme of Fromm is used in case of a smooth numerical solution. The Van Leer-2 scheme is slightly less accurate than the scheme of Stelling and Leendertse. which is first order in time and leads to tridiagonal systems of equations.43) Therefore in the vertical direction a fully implicit time integration method is applied. (10. Accuracy (at least second order consistency). An explicit time integration of the scalar advection-diffusion equation on a rectangular grid has a time step limitation based on the Courant number for advection: Cadv = max u∆t v ∆t .40) with ∆x and ∆y the grid spaces in the physical space. namely a first order upwind scheme and the second order upwind scheme developed by Fromm. two options are available in Delft3D-FLOW. The time integration of the Van Leer-2 scheme is explicit and therefore a CFL condition for advection and diffusion 296 Deltares . Explicit integration of the horizontal diffusion term yields an upper limit of: ∆t ≤ 1 DH 1 1 + 2 ∆x ∆y 2 −1 . User Manual 10. This scheme is denoted as the Cyclic method. The second option is an explicit scheme that belongs to the class of monotonic schemes: the so-called Van Leer-2 scheme (Van Leer.Delft3D-FLOW. An explicit time integration of the vertical exchange terms on the σ -co-ordinate grid would lead to very severe time step limitations: ∆t ≤ (∆σH )2 . see Stelling and Leendertse (1992). 2DV ∆σH ∆t ≤ . To ensure that the total mass is conserved the transport equation in Delft3D-FLOW is discretized with a mass conserving Finite Volume approach (flux form). The first (and default) option is a finite difference scheme that conserves large gradients without generating spurious oscillations and is based on the ADI-method. The Cyclic method of Stelling and Leendertse is based on an implicit time integration of both advection and diffusion and does not impose a time step restriction.

k 2 2 2 Gηη 1 m+ 2 .k 1 Gξξ ξη m.k ≥ 0 2 2 um+ 1 . see Section 10.k hm+ 1 . Owing to the explicit time integration the Van Leer-2 scheme requires per time step less computation time than the Cyclic method of Stelling and Leendertse. the system can be solved effectively by a Red Black Jacobi iterative scheme in the horizontal direction and a double sweep in the vertical direction. In the second stage the directions for explicit and implicit integration are interchanged.n. sink terms are integrated fully implicit.Numerical aspects of Delft3D-FLOW must be fulfilled.n.n. In order to avoid negative concentrations and instabilities. The coupling with the flow is weak and in Delft3D-FLOW the transport equation is solved independently of the flow for each half time step. The temporal variations in salinity are slow compared to the variations in the flow and therefore the baroclinic term in the momentum equations is treated explicitly.n. 10.4.n.k at the U -velocity point the interpolation is given by: 2 Fm+ 1 .n. If the spatial discretization of the horizontal diffusion term is based on a Cartesian grid using the Finite Volume approach of Stelling and Van Kester (1994) the integration is explicit.k = um+ 1 .n. For the Cyclic method the upwind discretization of the horizontal advective fluxes in ξ -direction is described by: 1 ∂ huc Gηη ∂ξ Gξξ = m. Source terms are integrated explicitly.16 and Section 9.k +cm−2.n 10cm. For the Cyclic method the time integration follows the ADI-method for the continuity equation. see Eqs. the resulting linear system of equations has thirteen diagonals but the matrix is diagonally dominant.n. The transport scheme for the Z -model is described by Bijvelds (2001). (10.5. because of the fact that a first order upwind discretization is applied in case of a local maximum or minimum.n.n Fm+ 1 .k −5cm−1.3.n. 9. Thus.k .4.1 Cyclic method To keep the numerical diffusion as small as possible the horizontal advection terms in the scalar transport equation are approximated by the sum of a third-order upwind scheme and a second-order central scheme.k < 0 (10.n. introducing a stability condition for internal gravity waves (baroclinic mode).k −5cm+2.n. In the first stage all space derivatives with respect to ξ are taken implicitly and all derivatives in the η -direction are taken explicitly.n.15 and 9. A second order central scheme is applied for the approximation of the vertical advection term.k 2 2 ∆ξ . The transport equation is coupled with the momentum equations by the baroclinic pressure term. For the time integration of the horizontal diffusion term along σ -planes the Crank-Nicholson method is applied. If the upwind discretization is used in the stage in which both the horizontal advection and vertical viscosity term are integrated implicitly.k − Fm− 1 .k .6.45) Deltares 297 . 6∆ξ um+ 1 .n. the Van Leer-2 scheme produces a more diffusive numerical solution. However.n. 6∆ξ 10cm+1.k +cm+3.44) For the scalar flux Fm+ 1 .

n.n Gm.n.n. 10.k−1 − Fm. 2 +1. (10.k  cm+1.50 consist of a diffusive first-order upwind term and a higher order anti-diffusive term.k −cm−1.n.k + α (1 + CF Ladv−u ) (cm. or min.k  .n.n.n.n+ 1 .47) Near open and closed boundaries the approximations for the fluxes are reduced to lower order.k Gηη 2 2 2 m+ 1 . (monotone).k +cm−1.k − Gm.k (10.k ≥ 0.k 1 Gηη ξη m.k +cm−1.  1.n. ≤ 1.k = um+ 1 .48) with: CFLadv−u = and: ∆t |u| ∆x cm+1.Delft3D-FLOW.k + cm.n.k cm+1. the interpolation formula for the horizontal fluxes is given by: × Fm+ 1 .n.k ) cm  cm.n. > 1.k 2 2 ∆η .k cm+1.k + α (1 − CF Ladv−u ) (cm.n.46) with the scalar flux Gm+ 1 .n.48 to 10.n.k 1 Hm.6.n.n.k hm+ 1 .2 Van Leer-2 scheme For the second option in Delft3D-FLOW.k ) cmm  +1.n+ 2 cm.n.k ∆σ .k cm+1.n.n.n.n.k −cm.n.n.k = ωm.k = vm+ 1 .n. 10. 10.k−1 .n.n.k at the V -velocity point determined by: Gm+ 1 . 1 m+ 2 .k 2 2 2 Gξξ 1 m.n.n.50) In η -direction a similar discretization is applied. (local max.52) Deltares .n Fm.6. Eqs.n+1.n 2  c +1.n.k .n.k .n.n.k −cm+2.n.k − cm−1.n. cm.n.).k −cm+2.k determined by: Fm.k 298 cm.k −2cm.n. The time integration of the Van Leer-2 scheme is explicit.k    when um+ 1 .n.k − cm−1.k −2cm.51) with the flux Fm. 2 (10.n.n.n.k −cm−1.49) α=   0. The Courant number for advection should be smaller than 1. namely the Van Leer-2 scheme.    when u < 0.n. (10. (10.k −cm−1. User Manual In the first stage in η -direction a central scheme is applied: 1 ∂ hvc Gξξ ∂η Gηη 2 = m.k (10.3 Vertical advection In the vertical direction the fluxes are discretized with a central scheme: 1 ∂ωc H ∂σ = m. 2 (10.n.k + cm.n.n− 1 .k hm+ 1 .

n.53. The superscript p denotes the iteration number.k Vm. so-called “wiggles” (Gresho and Lee. 1979).4 Forester filter It is well-known that second or higher order advective difference methods (such as the Cyclic method) on coarse grids may exhibit non-physical oscillations near regions of steep gradients. This filter is applied only in grid cells with a negative concentration.n.nk m−1.n. the central differences in the vertical may give rise to non-physical spurious oscillations.n.n.k − cm. then the iterative. If concentration cm.n.54) α = min 1. For depth-averaged simulations the Van Leer-2 scheme is strictly positive.n. The difference operators do not guarantee positive solutions and consequently negative concentrations may occur. In shallow water scalar transport these wiggles may be introduced in stratified areas near closed boundaries and steep bottom slopes.k min 1. . (10. a non-linear approach which removes the computational noise without inflicting significant amplitude losses in sharply peaked solutions.k = cp m.56) Deltares 299 .n. can be interpreted as an approximation of the following advectiondiffusion equation: ∂c α − β ∆x ∂c α + β ∆x2 ∂ 2 c = + + (higher order terms). 10.n. mass conservative filtering process is described (for the sake of simplicity only in one direction. 10. The vertical advection leads to a tridiagonal system in the vertical.k 4 Vm+1. and: Vm+1.55) β = min 1.Numerical aspects of Delft3D-FLOW The time integration in the vertical direction is fully implicit.k min 1. The filter smoothes the solution and reduces the local minima (negative concentrations). Eq.n.6. (10. The filtering technique in this procedure is the so-called Forester filter (Forester. In 3D.k − cm.n.k Vm. for both transport schemes.k denoting the volume of cell (m.n.53) with Vm. 1981) in the |w | ∆ z ´ solution when the vertical grid Peclet number Pe∆z = D ≤ 2.k + p cp m+1.k + Vm. The solution has an unphysical maximum or minimum scalar concentration (overshoot or undershoot). In case of negative concentrations an iterative filter procedure based on local diffusion along σ -lines followed by a vertical filter is started in order to remove the negative values.n.k p cp Vm−1. namely the ξ -direction) by: +1 cp m.n.k . k ). the central differences in the vertical may give rise to nonphysical spurious oscillations. n. a discharge of buoyant water.k . ∂t 4 ∆t ∂x 4 ∆t ∂x2 with: (10. Vm−1. Positive solutions are not guaranteed.k is negative.n. (10. These wiggles may arise in V the vicinity of steep gradients of the quantity to be resolved. due to vertical stratification in combination with upwelling or downwelling near a closed boundary or a sill. + 4 Vm. If the flow in the vertical is advection dominated.

k−1 ) + ε.k = 1− α+β 4 cp m. A negative concentration surrounded by positive concentrations.58) As both α ≤ 1 and β ≤ 1.n. and Pe∆z = local minimum: |w| ∆z ≤2 DV cm. If there is still a grid cell with a negative concentration after 100 iterations. cm+1.k (10.k+1 .n. cm. all coefficients of Eq.53 as: +1 cp m.9. User Manual Figure 10.k + cp 4 4 m−1. 10. A similar filtering technique as in the horizontal direction is applied for points with a local maximum or minimum in the vertical: local maximum: cm.k > max (cm.n.n.n.k+1 .k−1 ) + ε.n. cm. In the vertical. see Figure 10. a positive concentration will remain positive. we rewrite Eq.Delft3D-FLOW. it will not introduce negative concentrations irrespective the steepness of the concentration gradients. 4 ∆t 2∆t (10. To further understand the influence of the Forester filter.n. these wiggles may lead to unrealistic vertical profiles of temperature and/or salinity. 10. then a warning is generated. Local maxima and minima in temperature or salinity in the vertical direction. Consequently.k < min (cm. i. usually the result of ill represented steep gradients (wiggles).n.n. We should be sure that it is a wiggle generated by the numerical method and not by physical processes like heating or cooling through the free surface or the discharge of water somewhere in the vertical.k + α p β .57) Thus the filter introduces numerical diffusion but only locally. generated by the computational method may give physically unstable density profiles and can also better be removed by a numerical filter then by turbulent vertical mixing. and Pe∆z = |w| ∆z ≤2 DV Deltares 300 . The numerical diffusion coefficient of the horizontal filter is: Dnum = α + β ∆x2 ∆x2 ≤ . will be less negative after one iteration and is effectively removed after several iterations by adding enough (local) diffusion to force the concentration to become positive.58 are positive. Maximal 100 iterations are carried out.9: Illustration of wiggles in vertical direction The Forester filter introduces an artificial advection and diffusion.n.e.

n.k−1 ) 2∆zk (Tm. with ε = 10−3 . If both the horizontal and vertical filter are switched on. ∆zk−1 ) Tm.k−1 + min (∆zk . The numerical diffusion coefficient of the vertical filter is: Dnum = ∆z 2 .g.k = Tm. The transport equations of turbulent kinetic energy k .k−1 ) = Tm.n.Numerical aspects of Delft3D-FLOW the filter is applied. the entire area is water covered but as tide falls. The maximum number of filter steps in the vertical direction is 1000. and relatively flat areas separated and interleaved by deeper channels and creeks. 10. the shallow areas are exposed. Remark: The vertical Forester filter does not affect sediments and other constituents. The eddy viscosity is always based on information of the previous half time step. for temperature. On the staggered grid. (1992). 2∆t (10.127. the turbulent quantities k . and ultimately the flow is confined only to the deeper channels.n. 9. the filter only smooths salinity and temperature. This choice makes it possible to discretize the vertical gradients in the production term and buoyancy term accurately and to implement the vertical boundary conditions at the bed and the free surface.k−1 Endif with ε = 10−6 .5.n.k − Tm. then first the filter in the horizontal direction is carried out.59) Smooth but unstable vertical density profiles of salinity and temperature in the vertical direction. For turbulent boundary flows local production.n. shallow. The maximum number of filter steps in the horizontal direction is 100. and dissipation rate ε.n.7 Numerical implementation of the turbulence models The turbulence closure models in Delft3D-FLOW are all based on the eddy viscosity concept. When activated. If the maximum number of filter steps is exceeded. can sometimes also better be vertically mixed by a numerical filter technique then by the turbulence model. dissipation.k − Tm. ε and the eddy viscosity νV are positioned at the layer interfaces in the centre of the computational cell.k > Tm. For more details we refer to Uittenbogaard et al. 9. and vertical diffusion are the dominant processes.n.n. The dry tidal flats may occupy a substantial fraction of the total surface area.128 are solved in a non-conservative form. Eq. E. First order upwind differencing for the advection provides positive solutions. the algorithm is given by: If Tm.n.k − min (∆zk . The accurate reproduction of covering or uncovering of the tidal flats is an important feature of numerical flow models based on the shallow water equations.n.k−1 + ε Then Tm. a warning is written in the diagnostic file. Deltares 301 . This is followed by applying the filter in the vertical direction and hereby minimising the additional vertical mixing. Eq. see Section 9. ∆zk−1 ) 2∆zk−1 (10.60) 10. (Tm.8 Drying and flooding Estuaries and coastal embayments contain large. When water levels are high.

see Figure 10. It replaces partly the function of the old input parameter DRYFLP. the so-called depth points. The algorithms to determine the depth in a water level point from the four surrounding depth 302 Deltares . consisting of a main channel which always carries flow (the summer-bed) and one or two flood plains which only carry flow during extreme river discharges (the winter-bed). these three items will be discussed.8. The standard drying and flooding algorithm in Delft3D-FLOW is efficient and accurate for coastal regions. To determine the total water depth at water level points.n is not uniquely defined. In the following subsections. The oscillations introduced by the drying and flooding algorithm are small if the grid sizes are small and the bottom has smooth gradients. we specify the algorithms which have been used to determine the moment when a grid cell (water level point) or cell boundary (velocity point) becomes dry or wet. The way in which the water level is defined at velocity points. the so-called water level point (DPSOPT=DP). the process of drying and flooding is represented by removing grid points from the flow domain that become “dry” when the tide falls and by adding grid points that become “wet” when the tide rises. Drying and flooding gives a discontinuous movement of the closed boundaries and may generate small oscillations in water levels and velocities. MAX. Drying and flooding is constrained to follow the sides of grid cells. Criteria for setting a velocity and/or water level points wet or dry. 10. or in the cell centre. In older versions of Delft3DFLOW. The winter-bed is surrounded by much higher dikes. which are designed to protect the polders against extreme river discharges. three options were available: MEAN. estuaries. Delft3D-QUICKIN. The flooding of the flood plains increases the drainage capacity of the river and reduces the local water level gradients. The DPSOPT=DP-option implies that the position of the depth points is shifted to the water level points. and rivers. the algorithm is also effective and accurate for rapidly varying flows with large water level gradients because of the presence of hydraulic jumps or the occurrence of bores as a result of dam breaks.4. tidal inlets. MIN and DP.g. which will overtop when the river discharge increases. User Manual Many rivers have compound channels. At input the bottom depth can be specified by you at the vertices of a computational cell. Recently a new flag DPSOPT has been introduced with the following extended options: MEAN.Delft3D-FLOW. The summer bed is surrounded by low dikes. a bottom depth in the cell centre of ζ the Control Volume is required.1 Bottom depth at water level points Delft3D-FLOW uses a staggered grid. In combination with the Flooding scheme for advection in the momentum equation.10. MAX and MIN (through the value of the parameter DRYFLP). In a numerical model. The crucial items in a wetting and drying algorithm are: The way in which the bottom depth is defined at a water level point. You should consider this interpretation when generating the depth values with e. In this section. The bottom depth in a water level point dm. The algorithm used to determine this depth value from the four surrounding depth points depends on the choice made by you. see Figure 10.

n . dm−1. For the combination of flow computations with transport of dissolved substances.n−1 . Deltares 303 . dm−1. if unspecified.n = 0. For the time being.Numerical aspects of Delft3D-FLOW Figure 10.n ).n . dm.n = dm. If its value drops below the user specified threshold than all four cell interfaces are set to dry (velocities are set to zeroes).n . YES implies that an additional drying and flooding check is required based on the evaluation of the value of the total water depth.n .n + dm−1.n−1 ) DP-option: dζ m. while some of the adjacent velocity points still have a positive flow through height. dm−1. dm−1.n−1 .25(dm.n = max(dm.n−1 ) MIN-option: dζ m.n−1 ) MEAN-option: dζ m. DPSOPT value is determined automatically from DRYFLP by the program (default value of DPSOPT is equal to DRYFLP and subsequently DRYFLP is then set according to its original value).11 an example is shown in which the water level is below the bottom in the cell ζ centre (dm. DRYFLP. the control volume/retention volume should be positive.n = min(dm. In Figure 10.n + dm.10: Definition bottom depth on FLOW grid points are given by: MAX-option: dζ m. The retention volume of a dry cell is the cell area times the difference between the water level and the bottom depth at the cell centre. dm. is below the bottom. DRYFLP=NO implies that the drying and flooding check is only to be based on the individual test of the depth values at the cell interfaces. determined on basis of the average depth. is now restricted to YES or NO.n−1 + dm−1.n With the introduction of DPSOPT the value of the keyword in the input file that is related to the selection of additional drying and flooding procedure at a water level point.

It has been found that the MAX procedure is more favourable and will produce a more smooth solution than the options described by Stelling et al. (10. you can now choose from the following three For the depth values at the cell interfaces. The algorithms for flooding and drying of tidal flats have been extensively described by Stelling et al.n + ζm.2 Total water depth at velocity points Due to the staggered grid applied in Delft3D-FLOW. (1986).n = dζ m. User Manual d(m-1.62) 10.8.3.n Hζm.n ζm.n .8.n-1) d(m. (1986). The initial water level at a dry cell is determined by the depth at a water level point: ζm. MEANoption d(m. that article does not include the MAX option.63) ¯η .n = −dζ m.Delft3D-FLOW. (10.n) Therefore after solving the coupled system of continuity equation and momentum equation there is a drying check applied to the water level points. the MEAN option may lead to flooding of velocity points and afterwards drying due to a negative control volume.11: Negative control volume with two positive flow-through heights.61) the continuity cell is taken out of the computation and the half time step is repeated. In case of steep bottom slopes.n) dζm. η ξ (10. d options by setting an appropriate value for the input parameter DPUOPT: 304 Deltares . The use of MAX (default) is recommended. increasing the computational time.n-1) Figure 10. see Section 10.n =d +ζ . If the total water depth in a water level point is negative: ζ Hm. the total water depth at a velocity point for the computation of the discharge through a cell face is not uniquely defined. However. Usually (default option.n ≤ 0.n d(m-1. marginal depth DCO=-999) it is determined by the arithmetic average of the depth specified in the vertices of the cell face (side) plus the average of the water levels computed in the cell centres at each side of that cell face: U Hm.

2 In the neighbourhood of steep bottom gradients.n ) . dm+1.12: Drying of a tidal flat. dm+1.n +dm. Figure 10.63.Numerical aspects of Delft3D-FLOW Figure 10. ¯ξ as default the mean value is used: ζ = For the water level at the cell interfaces ζ ξ (ζm. η (dm.  ζ  m. averaging approach. may lead to an inaccurate determination of the flowthrough height. see Equation 10. The velocity point is set dry too early. use of the average water level to compute the total water depth at a velocity point. the velocity point is set dry. see Figure 10.13.n +ζm+1.n < 0 if Um.n .n DPSOPT is DP and DPUOPT is MEAN should not be used together. Deltares 305 . the velocity point is set dry. A large volume of water is left on the tidal flat. 10.63.n   min dζ Remark: if Um. see Eq. The flow-through height is based on the average water level.n > 0 if Um. The flow-through height is based on the average water level.13: Overtopping of a river bank (weir).n . averaging approach. 2 DPUOPT = MEAN (default option): d = η ζ DPUOPT = MIN(imum): d = min dζ m.n−1 ) .12 and Figure 10.n .n  dζ η d DPUOPT = UPW(ind): d = m+1. increasing artificially the storage capacity of the wet area.n = 0 ζ m.

The upwind approach is physically less realistic if the flow has the opposite direction as the water level gradient (wind driven flow). see Eq. which allows the water level gradient to drive a larger amount of water into the neighbouring cell during the next time step.n < 0. η ξ If (DPUOPT = UPW ∨ hydraulic structure ∨ Z -model) Then  η Um.15 shows the situation of a river which overtops its bank.  d + ζm. The upwind flow through height in a U -velocity point based on an upwind water level is given by: U Hm. User Manual Figure 10. Upwinding the water level in the determination of the total water depth (flow through height) enhances the discharge because the upwind water level is generally higher than the average water level.4.14) or for weir like situations (Figure 10.n representing the depth averaged velocity both for 2D and 3D.n = 0. leading to unrealistic water levels downstream.15). In the Z -model always upwind water levels and depth are used in velocity points (Bijvelds. see Section 10. Taking the maximum of the two surrounding water levels at a dry cell face prevents that a velocity point is artificially kept dry.n . 2001). An example of such a flow problem is the simulation of the inundation of dry land with obstacles such as road banks and dikes. is accurate for the approximation of the advection in rapidly varied flows due to sudden expansions or sudden contractions (Stelling and Duinmeijer. Um. generating a shock wave.n . The flow-through height is determined by flow direction.64) with Um. Figure 10. Due to a sudden contraction. the flow speed can become critical. suddenly the flood plains are filled with water. It is also possible to determine the water levels at the cell faces with a so-called upwind approach. The water level will rise too much in the main channel of the river. If we take the average water level to determine the total water depth at the crest.n > 0. This approach was already suggested by Stelling (1984) for shallow regions. When the river run off increases.14: Drying of a tidal flat.64. The accuracy in the approximation of the critical discharge 306 Deltares . The Flooding scheme.n in a V -velocity point is similar. The computation of V the upwind total water depth Hm. the velocity point remains dry.Delft3D-FLOW.  η d + max (ζm. The upwind approach is physically more realistic for velocity points between cells with different bottom depth at low water and falling tide (Figure 10.n ) . the velocity point remains wet. 10. Um. resulting in a larger flow area.n = d + ζm+1. η U Hm. 2003).4. upwind approach. ζm+1. In Delft3D-FLOW the choice between upwind and the average approach for the water level in a U -point is controlled by flag DPUOPT.n =d +ζ . Endif (10.n .

10. the velocity point remains wet.15: Overtopping of a river bank. centred around the water level points.Numerical aspects of Delft3D-FLOW Figure 10. The flow through height in a U -velocity point is always based on an upwind water level. Figure 10.64. For the Flooding scheme the bottom is assumed to be represented as a staircase (DPUOPT=MIN) of tiles. the user-defined marginal depth DCO is set at +999. The flow-through height is based on the maximum water level. see Figure Deltares 307 . For the Flooding scheme the accuracy in the numerical approximation of the critical discharge rate for flow with steep bed slopes.7b and Figure 10. upwind approach. see Eq. The limiter function is controlled by the user-defined threshold depth for critical flow limiter DGCUNI.16: Special limiter for critical flow due to a sudden contraction (Flooding scheme and increase in bottom larger than DGCUNI) rate is dependent on the approximation (limiter) of the total water depth at a velocity point downstream. see Figure 10. can be increased by the use of a special approximation (slope limiter) of the total water depth at a velocity point downstream.16.

If a negative control volume occurs. You can avoid this by smoothing the bottom gradients. the four velocity points at the cell sides are set dry. This method consists of two stages (half time steps). 10. In 3D simulations the velocities are set to zero for all the computational layers.n + DGCU N I < dm+1.n ) Then 2 ζ U U ζ Hm. When all four velocity points of a computational cell surrounding a water level point are dry then this computational cell will be set dry.n .4 an Alternating Direction Implicit (ADI) time integration method is used in Delft3D-FLOW. If the water level rises and the total water depth is larger than the threshold. If the bottom has steep gradients across a large area on a tidal flat. If the total water level becomes negative. In 3D simulations the velocities are set to zero for all the computational layers.16.3 Drying and flooding criteria As described in Section 10.66) In general. the points for this special approach. The drying threshold is given half the value of the wetting threshold (hysteresis) to inhibit changes of state in two consecutive time steps (“flip-flop”). The thickness of the water layer of a dry cell (retention volume) is dependent on the threshold d specified by you.n + DGCU N I ) Then U U Hm. If the total water level drops below half of a userspecified threshold. The total water depth Hm.16 and: ζ If (Um.n 3 Endif (10. the threshold value d must fulfil the following condition: ζ δ≥ ∂ζ ∆t .n at a velocity point should at least be positive to guarantee a realistic discharge across a cell face. we will only describe the drying and flooding algorithm for the first half time step.8. due to oscillations introduced by the algorithm itself. U The total water depth Hm.65) By the introduction of the user-defined threshold DGCUNI.n = min Hm.n at a water level point should at least be positive to guarantee a positive control volume. ∂t 2 (10. max Hm. At both stages the same drying and flooding algorithm is applied. 308 Deltares .Delft3D-FLOW. The computational cell is closed for the side normal to the velocity point. This will produce short oscillations. Therefore. Flooding is restricted to velocity points.n m+1.n . max 2 ζ H .n > 0 ∧ dζ m. then the velocity point is set dry.n > dm+1. the bottom topography and the time step. Therefore.n . are recognised automatically during the simulation. the half time step should be completely repeated and the computational time increases. Hζ 3 m.n < 0 ∧ dζ m. the magnitude of the disturbances generated by the drying and flooding algorithm will depend on the grid size. The disturbances are small if the grid size is small and the bottom has smooth gradients. Hm +1. the velocity point is set wet again.n = min Hm. User Manual 10. see Figure 10.n ζ Elseif (Um. a large area may be taken out of the flow domain in just one half integration time step. without having to specify all the points at input.

You may define in velocity points so-called weirs or spillways.64 is applied. a check at each water level point is carried out to avoid negative volumes. HKRU.64. A smooth flooding behaviour is obtained if the simulation is initialised at high water. 10. In step four of the drying and flooding algorithm in Delft3D-FLOW. The depth at closed boundaries should be a bottom value near the coastline and not a land height.n > δ ). If the time step is too large an unphysical water level gradient at the wet-dry interface is built up.9.n < 0. 10. independent of the marginal depth DCO.5δ ) during iterative solution for new water level. Deltares 309 . weirs and barriers) always upwinding Eq. is taken into account in the drying and flooding algorithm. The boundary of the wet area can only move one grid cell per time step. Remark: Near hydraulic structures: (discharge points. U 2 Drying check for velocity points in x-direction (Hm. the water layer may be very thin and cause problems in combination with online salt transport or off-line water quality simulations. the drying and flooding algorithm in Delft3D-FLOW consists of the following four checks: V 1 Drying check for velocity points in y -direction (Hm. the directions are interchanged. The thickness of the water layer is zero. with the initial water level at the expected maximum level. If the computational cell is flooded. respectively. In that case the water layer (retention volume) at dry points is initialised by the drying flooding algorithm. In Delft3D-FLOW the computational part is protected against “dividing by zero” by assuming that the total water depth is at least 1 centimetre. For a 2D weir the height of the crest.0).8. The threshold δ is specified by you at input.63 or Eq. Four options are available (see Section 10. Weirs are hydraulic structures causing energy losses. see Section 10. 10. Default the average approximation is used in the σ -model and the upwind approach is always used in the Z -model. In the second stage of the ADI-method.n < 0. The total water depth at velocity points is computed according to either Eq.5δ ). If a simulation is not started at high water.Numerical aspects of Delft3D-FLOW Flooding is an explicit process. These checks are based on the water level of the previous half time step. U 3 Drying check for velocity points in x-direction (Hm. ζ 4 Drying check (negative volumes) for water level points (Hm. The total water depth in water level points depends on the way in which the bottom depth is computed. the water levels at dry points are initially set at the bottom depth.1). which will generate oscillations after flooding. In the first stage of the ADI-method.n < 0. corresponding to the average and the upwind approach. Attention should also be paid to the fact that depth values at points at closed boundaries are used in the total water depth of a velocity point parallel to the boundary and for the depth at a water level point for the MEAN-option.n < 0.5δ ) and flooding check for veU locity points in x-direction (Hm.

All hydraulic structures except for culverts are located on the interface between two computational cells.9 Hydraulic structures In a Delft3D-FLOW model. the mathematical formulations and implementation of the hydraulic structures available in Delft3D-FLOW will be described in more detail. the following energy preserving upwind discretization of advection is applied: ∂U U ∂x m. locally a first order upwind scheme is used by default. 2∆x Um. porous plates. ζm. Around a hydraulic structure.67) and flooding the flooding check: U Hm. the velocity field and in the concentrations. weirs. For supercritical flow.n < 2 δ. Culverts have been implemented as two coupled discharge locations (one inlet and one outlet) which may be located in different parts of the computational grid. Only for the hydraulic structures of the types 2D weir and culvert also the supercritical flow rate is computed accurately. (10. 2∆x 2 2 Um+1.n ) + HKRUm. Culverts with a user defined discharge relation (type ‘u’) can be used to implement (non-local) discharge formulations representative of other kinds of structures. For the momentum equations.n . barriers. 10.4.n.k .n.5. The energy loss formulations presently available in Delft3D-FLOW assume subcritical flow.69) For the transport equation. A hydraulic structure generates a loss of energy apart from the loss by bottom friction.n.Delft3D-FLOW. sluices. bridges.k > 0. so-called hydraulic structures can be defined to model the effect of obstructions in the flow which can not be resolved on the horizontal grid (sub-grid) or where the flow is locally non-hydrostatic. 310 Deltares .68) The weir acts as a thin dam for water levels lower than the crest height. to parameterise the extra loss of energy.n.n <1 2 δ ∧ max(ζm−1. see Section B.n.n > δ ∧ max(ζm−1.k −Um.k 1 ∂U 2 = 2 ∂x = m. The hydraulic structures are divided into three basic types: hydraulic structures with quadratic friction.n ) + HKRUm.k .n. the downstream water level has no influence on the flow rate.k 2 2 Um. ζm. an additional force term is added to the momentum equation.n > δ. User Manual The drying check for a 2D weir point at a U -point is given by: U 1 Hm. (10.k < 0. sills. The term has the form of a friction term with a contraction or discharge coefficient. groynes. hydraulic structures with linear friction and floating structure. (10.n. At hydraulic structure points. In order to prevent unphysical oscillations in the velocities and concentrations upstream of hydraulic structure points the user can switch (option) the discretisation of the advective terms at such points to an upwind approximation. The flow condition at hydraulic structures may be supercritical.n.k − Um −1.3. Examples of hydraulic structures in civil engineering are: gates. there will be large horizontal gradients in the water level.n . In this section. Um.

respectively. 10. For 3D gates not only the begin and end co-ordinates of its horizontal position. “fixed” in the Cartesian co-ordinate system (and moving through the σ -grid). Upstream of the structure the flow is accelerated due to contraction and downstream the flow is decelerated due to expansion. In the σ -model the position of the 3D-gate can be: “fixed” in the computational grid (and moving with the water level in the Cartesian coordinate system). The reduction factors are computed on basis of the vertical position of the gate and the time-dependent vertical σ -grid. For more details on the data input requirements for the different type of 3D gates we refer to Appendix B. moving in the Cartesian co-ordinate system by lowering or raising of the gate (3D barrier). see Figure 10.17: Example of a 3D Gate (vertical cross-section) 10. so it has no influence on the water volume in the model area. (10.9. the wet flow-through area and and the upstream and downstream water level. A gate may be used to model a vertical constriction of the horizontal flow such as barriers. The flow at all intermediate layers of the gate is set to zero.2 Quadratic friction The steady-state subcritical flow rate Q through a hydraulic structure is related to the difference between the upstream and downstream water levels: Q = µA 2g |ζu − ζd |. A 3D gate is located at a velocity point and its width is assumed to be zero. For the latter two cases.Numerical aspects of Delft3D-FLOW Figure 10. The wet cross section of that computational layer is reduced. The vertical constriction of the flow may vary in time by the lowering or raising of the gate.70) with µ the contraction coefficient (0 < µ ≤ 1).9. a layer may be partially blocked near the top or the bottom of the gate. This means that some of the layers can be represented as vertical thin dams.18.1 3D Gate A 3D gate is in fact a thin dam with a limited height/depth (and positioned in the vertical). The contraction coefficient in the so-called QDeltares 311 .3. but also its vertical position can be specified.1. sluices and Current deflection walls. The layer near the top and the layer near the bottom of the gate may be partially blocked.

10. The appropriate energy loss coefficient should be specified 312 Deltares . The expansion introduces a water level jump between the upstream and downstream water level.Delft3D-FLOW.71. to model the effect of the hydraulic structure. The Q-H relation may be used to determine the coefficient closs−U in the quadratic friction term of the momentum equation. which is independent of the grid size. For a 3D barrier. a quadratic formulation for the energy loss is implemented in Delft3D-FLOW: barrier bridge Current Deflection Wall 2D weir local weir porous plate culvert 10. User Manual Figure 10.71) The resistance coefficient closs−U need to be specified by the user at input. The energy loss for a barrier is taken into account by adding an extra quadratic friction term to the momentum equations Eq. some of the layers are closed dependent on the gate height. The barrier is “sub-grid” and has no influence on the water volume in the model area.18: Computational layer partially blocked at bottom of gate H relation is dependent on the kind of hydraulic structure. In 3D models the model depth at the barrier point should be decreased to the sill depth for appropriate modelling of salt transport.n ζu − ζd Q2 = 2 2 = closs−U . For a Q-H relation at a U -velocity point: g Um. For the following types of hydraulic structures. ∆x 2µ A ∆x ∆x (10.1 Barrier In Delft3D-FLOW. the hydraulic structure barrier is the combination of a movable gate and a quadratic friction term.9. It can be used in 2D and 3D models. We assume that the hydraulic structure is “sub-grid”. Upstream of a barrier the flow is accelerated due to contraction and downstream the flow is decelerated due to expansion.2. and that there is a local equilibrium between the force on the flow due to the obstruction and the local water level gradient. A barrier is located at a velocity point.n Um.

the drag coefficient of a pier (pile) (1. 2 µ2 (10. part of the energy loss is computed directly by the discretisation of the convection terms in the momentum equations and the bottom friction term. At a bridge point the quadratic friction term is added to the momentum equation for all layers which are open. For a partial open layer k . The depth-averaged flow-rate through a barrier is given by: Q = µA 2g |ζu − ζd |. 9. dependent on the local discharge relation. Currently. 9. effective wet cross sectional area (Atot minus the area blocked by piles: Aef f = Atot − ζN dpile ).2 Bridge The flow resistance due to a jetty or a bridge is dependent on the blocking of the flow by the piers (Farraday and Charlton. The contraction coefficient is used to determine the friction coefficient in Eq.71. (10.74) with a the bottom of the gate. free slip boundary conditions has been implemented at the transition from an open layer to a closed layer (representing the bottom side of the gate): vv ∂U ∂z =0 gate z =Zbottom (10. 10. see Figure 10. see Eq. with thickness hk the same loss coefficient is used as for the other layers. Furthermore.73) For a 3D barrier.75) Atot Aef f Cdrag dpile N total cross sectional area (Atot = ζ ∆y ). 10. The discharge relation presently available at barrier points in Delft3D-FLOW assumes subcritical free surface flow. the number of piles in the grid cell.270 or Eq. Deltares 313 .9. For a row of piles perpendicular to the U -direction the energy loss coefficient closs−u perpendicular to the flow is given by (and has to be specified): closs−u = with: N Cdrag dpile 2∆y Atot Aef f 2 .Numerical aspects of Delft3D-FLOW for each barrier by the user. The quadratic friction is added to the momentum equations for all layers which are open. The contraction coefficient should be obtained from laboratory or field measurements. In 2D.18.2. 1983). The loss coefficient closs−U should be used for calibration.72) with µ the barrier contraction coefficient (0 < µ ≤ 1).271 and depends on the direction of the barrier.0 for a smooth cylindrical pile). (10. Between the gate and the sill we assume a uniform velocity profile. the diameter of a pile. you should derive the value of the loss coefficient and specify at input (see the structure Porous plate). the energy loss coefficient perpendicular to the flow has to be specified according to: closs = 1 .

summer dikes) and groynes in simulation models of rivers. In 3D some of the computational layers are closed dependent on the gate height and the upstream water level. 2D Weir The mathematical concept was developed by Wijbenga (1990). see Eq. 10. The energy loss for a CDW is taken into account by adding an extra quadratic friction term to the momentum equations Eq.71) or dependent on the flow resistance that you want to induce. 10. which is independent of the grid size. The quadratic friction is added to the momentum equations for all layers which are open.76) 314 Deltares .9. The energy loss is parameterised and added in the momentum equation as follows: Mξ = − gH ∆Eweir ∆x (10. The expansion introduces a water level jump between the upstream and downstream water level. A 2D weir is assumed to be “sub-grid”. So the energy loss generated by the weir is not computed directly by the convective terms in the momentum equations. dependent on the local discharge relation. For a 2D weir the crest height (HKRU ) is taken into account only in the drying and flooding algorithm.Delft3D-FLOW. The CDW is “sub-grid” and has no influence on the water volume in the model area. and the total water depth in the wet cross section to compute the discharge is still based on the bottom without crest. see Section 10.71. to determine if a velocity point is wet or dry. Upstream of a CDW the flow is accelerated due to contraction and downstream the flow is decelerated due to expansion.2. 10.75. 10.8. The appropriate energy loss coefficient should be specified by the user.3 Current Deflection Wall In Delft3D-FLOW. the hydraulic structure weir is a fixed non-movable construction generating energy losses due to constriction of the flow.2. They are commonly used to model sudden changes in depth (roads.4 Weir In Delft3D-FLOW. It can be used in 2D and 3D models. User Manual 10. see Eq. the hydraulic structure CDW is the combination of a fixed gate and a quadratic friction term.3. A CDW is located at a velocity point.9.

j + 2 (10. The flow at the weir crest is supercritical if: 2 ζdown + HKRU ≤ Eup 3 or Q ≥ Qsuper (10. the energy loss term is used to control the discharge at the theoretical supercritical discharge rate Eq. when the velocity is between 0.77.or supercritical.79) For supercritical flow. This depends on the flow velocity Uweir at the weir crest. Deltares 315 .82) The energy loss following Carnot is given by: ∆ECar = with (Uweir − Udown )2 2g (10.78) The flow condition depends on the water level downstream and the discharge rate.77) Eup = ζup + HKRUi+ 1 .j + = 2 2g 2g (10. the difference between the energy head upstream and downstream of the weir should be equal to the energy loss caused by the weir.5 m/s. For sub-critical flow conditions.81) In the momentum equation. a weighted average is used between ∆ECar and ∆Etable . ∆Eweir = Eup − Edown with (10. 10.84) If the flow velocity at the weir crest is less than 0. if the velocity is more than 0. 1987) ∆Etable and/or the formula of Carnot ∆ECar . assuming conservation of energy for the flow contraction.25 m/s and 0. This velocity is derived from the discharge upstream and the crest height. For supercritical flow the discharge at the weir is completely determined by the energy head upstream and the discharge is limited by: 2 Qsuper = ∆y Eup 3 with 2 gEup 3 (Uup )2 (Uweir )2 = ζweir + HKRUi+ 1 .25 m/s the energy loss is calculated according to ∆ECar . ∆Etable is used.83) Udown = Qweir ∆y (ζdown + HKRU + dsill ) (10. Uweir = Qweir ∆y (ζup + HKRU ) (10.80) Edown = ζdown + (Udown )2 2g (10. the energy loss ∆Eweir is based on experimental data “Tabellenboek van Rijkswaterstaat” (Vermaas.5 m/s.Numerical aspects of Delft3D-FLOW The flow condition at a local weir may be sub.

the model depth at a weir point should be decreased for appropriate modelling of salt transport. To prevent oscillations when the flow over the weir is just below supercritical a relaxation parameter θ has been introduced which can be set to a value between 0 and 1 by means of the ThetaW keyword. Mξ = −closs−U Um.. part of the energy loss is computed directly by the discretisation of the convection terms in the momentum equations and the bottom friction term. The crest height of a local weir is not derived from the bottom but specified at input.5 Porous plate For a so-called porous plate the energy loss coefficient should be specified by the user at input.3.Delft3D-FLOW. 316 Deltares .9. A culvert is a special intake/outlet coupling in which the discharge rate depends on the flow regime.85) (10.n Um.. The total water depth in the wet cross-section at a local weir point is based on the crest height. which is called a “power station”.8.2..2.6 Culvert In Delft3D-FLOW intake/outlet couplings have been implemented. see Section 10.5 for details on the user-input. User Manual The energy loss is assumed to be normal to the obstruction and for a U -weir it is added to the momentum equation as follows g ∆E n Ui+1/2.9. the discharge rate is specified on input.86) where the velocity in the denominator of the first term is replaced by the critical flow velocity over the weir in case of a perfect weir. The loss coefficient closs−U should be used for calibration.. − ∂t ∆xUin +1/2.+ = .j ∂U + . 10. to determine if a velocity point is wet or dry. Local weir In 3D models.n ∆x (10. In case of a normal intake/outlet coupling.j ∆E n = (1 − θ) ∆E n + θ∆E n−1/2 n+1/2 (10.87) 10. See Appendix B. For a local weir.2.3. The default value is 0 meaning no under-relaxation used for the energy loss ∆E . The crest height (HKRU ) is taken into account only in the drying and flooding algorithm.

respectively. To consider six flow classifications and their discharge charateristics some variables are defined. Furthermore. a more general formulation for the computation of the discharge through the culvert is available. (1987). Therefore. Let us first introduce some notation: ζu = max(0. Sc is the so-called critical slope. ζintake − zculvert ) and ζd = max(0. A the area (in m2 ) of the culvert opening and ζintake and ζoutlet the water levels at the intake and outlet. with different discharge relations for each type. the critical depth is defined by 3 Hc = Q2 gW 2 (10. In particular. ζintake − ζoutlet ) (10. Six flow classifications are distinguished. according to Q = µA 2g max(0.88 does not perform satisfactorily in all flow conditions. The classification was taken from French et al. which has to be specified by the user as well. which corresponds to a completely submerged culvert. This coefficient represents all energy losses due to a culvert that are not part of the other terms in the discharge relations. u is the velocity through the culvert and α is the so-called energy loss correction coefficient. in which the superscript represents the type(s) of flow condition.90) with W the width (in [m]) of the culvert.Numerical aspects of Delft3D-FLOW Definition of a culvert In Delft3D-FLOW four types of culverts are distinguished: Type ‘c’ ‘d’ ‘e’ ‘u’ Description One-way culvert with a “simple” discharge formulation Two-way culvert with a “more generalized” discharge formulation One-way culvert with a “more generalized” discharge formulation Two-way culvert/structure with user-defined discharge formulation Culvert of type ‘c’ For a culvert of type ‘c’. Deltares 317 . cD and cD ). this is the case when the downstream water level is below the so-called critical depth. (page 368). Culvert of type ‘d’ or ‘e’ It is known that Eq. ζoutlet − zculvert ) (10.88) with µ the culvert loss coefficient (dimensionless). n the Manning’s coefficient (in [m1/3 /s]) representing the energy loss due to the culvert and a discharge coefficient that is prescribed by the user 2 3 (c1 D . S0 is the slope of the culvert. the discharge rate through the culvert (in [m3 /s]) is computed by the Delft3D-FLOW program. 10. more generalized formulations were proposed by WL Borgerhout for the culvert loss coefficient and have been encorporated. However.89) in which zculvert is the vertical position of the bottom of the culvert relative to the reference level (positive upwards). Furthermore. L is the length of the culvert.

91) Type 2 (supercritical flow with critical depth at outfall.95) Type 6 (full flow free outlet) Q = µ(H ∗ )HW 2g (ζu − H ) with H ∗ = H 2gLn2 +α (R∗ )4/3 c2 D 2 µ(H ∗ ) = c2 D/ 1 + and R∗ = HW 2H + 2 W (10. namely c1 D : flow classification type 1. User Manual The six flow classifications and their discharge characteristics read: Type 1 (supercritical flow with critical depth at intake.Delft3D-FLOW. steep culvert slope) Q = µHc W 2g (ζu − Hc ) with µ = c1 D (10.96 it can be seen that three different discharge coefficients are used.92) Type 3 (tranquil flow): Q = µ(H ∗ )ζd W 2g (ζintake − ζoutlet ) with H ∗ = 0.5ζd 2gLn2 +α (R∗ )4/3 c1 D 2 µ(H ∗ ) = c1 D/ 1 + ζd H∗ 2 and R∗ = H ∗W 2H ∗ + W (10.5Hc + 0.91 to 10. 2 and 3 c2 D : flow classification types 4 and 6 c3 D : flow classification type 5 In the table below the conditions are listed for these six flow regimes.5ζu 2gLn2 +α (R∗ )4/3 c1 D 2 µ(H ∗ ) = c1 D/ 1 + Hc H∗ 2 and R∗ = H ∗W 2H ∗ + W (10.96) From Eqs.94) Type 5 (rapid flow at inlet) Q = µ(H ∗ )HW ∗ 2gζu with µ(H ∗ ) = c3 D and H = H (10.93) Type 4 (submerged flow) Q = µ(H ∗ )HW 2g (ζintake − ζoutlet ) with H ∗ = H 1+ 2gLn2 +α (R∗ )4/3 c2 D 2 µ(H ∗ ) = c2 D/ and R∗ = H ∗W 2H ∗ + W (10. 318 Deltares . mild culvert slope) Q = µ(H ∗ )Hc W 2g (ζu − Hc ) with H ∗ = 0.5ζu + 0. 10.

or (types 1. Either.96. ‘d’ and ‘e’ and variations thereof. If none of these flow conditions are satisfied. You can use this routine to reproduce the formulations for culverts of types ‘c’.8. This culvert type can be used to implement a variety of structures resulting in pointto-point fluxes (either uniform over the water depth or at specific heights). If the water level at the outlet location is higher than at the intake. Since the culvert implementation applied in Delft3D-FLOW is based on strict horizontal culverts. The discharge can be based on the water levels at both ends of the culvert. Deltares 319 . This means that for a two-way culvert flow through the culvert in two-directions is possible. see Eq. The differences are in the computation of the culvert loss coefficient µ.5H >H ≥ 1. Only the computation of the discharge through the culvert differs. For a one-way culvert (type ‘e’) the discharge is always from the intake location to the outlet location. only flow type 2 can occur and not flow type 1. then (type 4). The formulations are more or less identical to the ones applied in SOBEK-RURAL of Deltares.5H ≥ 1.88 versus Eqs.91 to 10. We note that culvert types ‘c’ and ‘e’ are comparable.5H < 1. An example routine is given in Section B. either intake or outlet location) is considered as the intake location. In case of a two-way culvert (type ‘d’) the location with the highest water level (thus. Flow type 1 corresponds to culverts with a steep slope. 10. Culvert of type ‘u’ The discharge Q through a culvert of type ‘u’ is determined by a subroutine that you provide yourself. Difference between culvert type ‘d’ and ‘e’ In the previous text it is described how the discharge through a culvert is computed. or you can implement a discharge functions that is valid for some completely different structure. (types 5 or 6). then there is no flow through the culvert.Numerical aspects of Delft3D-FLOW Type Flow regime 1 2 3 4 5 6 supercritical flow with critical depth at intake supercritical flow with critical depth at outlet tranquil flow submerged flow rapid flow at inlet full flow free outlet Conditions ζu < 1.5H ζd ≤H ≤H ≤H >H ≤H ≤H ζd ≤ Hc ≤ Hc > Hc ≤ Hc ≥ Hc other So > Sc So ≤ Sc Remarks: From this table it can be verified that all possible flow conditions can occur.5H < 1.3. 10. 2 or 3). whereas flow type 2 represents a mild culvert slope.

y = y ∗ .9.n ∆x (10.9.6 for details of the data input requirements. the vertical resolution increases automatically in shallow areas. an artificial compression coefficient may be specified. small pressure gradients at the left-hand side may be the sum of relatively large terms with opposite sign at the right-hand 320 Deltares . the resistance force is assumed to be linearly dependent on the flow. Due to truncation errors artificial vertical mixing and artificial flow may occur. the time derivative in the continuity equation Equation 9. User Manual 10. d the depth measured downward positive and H the total water depth. Leendertse (1990) and Stelling and Van Kester (1994). can be specified by the user at input. ∂x ∂x∗ ∂x ∂σ ∂x∗ H ∂x ∂x ∂σ (10. The bottom boundary and free surface are represented smoothly.98) ∂σ 10. For steep bottom slopes combined with vertical stratification. (10.97) 10. Let ζ be the position of the free surface. H (10. defined by: x = x∗ . In a σ -model.10 Artificial vertical mixing due to σ co-ordinates The σ -transformation is boundary-fitted in the vertical.4 Floating structure Floating structures can be modelled. Via an external file the depth of the floating structure has to be specified.Delft3D-FLOW.100) In case of vertical stratification near steep bottom slopes.99) the horizontal pressure gradient reads: ∂p ∂p∗ ∂σ ∂p∗ ∂p∗ 1 ∂ζ ∂H ∂p∗ = + = − + σ . σ -transformed grids introduce numerical problems for the accurate approximation of horizontal gradients both in the baroclinic pressure term and in the horizontal diffusion term. (keyword Riglid in the input) yielding: α ∂ζ + ∂t 1 Gξξ ∂ (d + ζ ) u Gηη + ∂ξ Gηη 1 Gξξ ∂ (d + ζ ) v ∂η Gηη Gξξ + + ∂ω = H (qin − qout ) . closs−U . A floating structure is simulated by assuming a local pressure that pushes the water to a depth specified by the key word Filfls. In combination with a floating structure.3 Linear friction Rigid sheet For a so-called rigid sheet.3. If we consider the transformation from Cartesian co-ordinates to σ co-ordinates. See Appendix B. This artificial vertical transport is sometimes called “creep”. The water column is divided into the same number of layers independent of the water depth. The resistance coefficient. Then. σ = z−ζ . Mξ = −closs−U Um.9 is multiplied by the artificial compression coefficient α.

The truncation errors depend on the grid sizes ∆x and ∆z . similar to the horizontal pressure gradient. it can be seen that by increasing the number of σ -levels the consistency condition will eventually be violated. In the notation used by Haney (1991) this consistency relation is given by: σ ∂H ∂σ < . also on the numerical grid (Beckers et al.103) 321 . The horizontal gradients are taken along σ -planes.Numerical aspects of Delft3D-FLOW (a) Hydrostatic consistent grid cell (b) Hydrostatic inconsistent grid cell Figure 10. ∂σ 2 ∂x ∂x∗ ∂σ ∂x2 ∂σ (10.102) For such a combination of terms it is difficult to find a numerical approximation that is stable and positive. Similarly. (b) Hδσ < σ ∂x δx side. ∂x2 ∂x∗2 Deltares (10. This approach guarantees a positive definite operator. the transformation of a simple second order derivative leads to: ∂2c ∂ 2 c∗ = + ∂x2 ∂x∗2 ∂σ ∂x 2 − ∂ 2 c∗ ∂σ ∂ 2 c∗ ∂ 2 σ ∂c∗ + 2 − + − .19: Example of a hydrostatical consistent and inconsistent grid. H ∂x ∂x (10. see also Figure 10.. Observations of this kind has led to the notion of “hydrostatic consistency”. This artificial forcing produces artificial flow. Small truncation errors in the approximation of both terms result in a relatively large error in the pressure gradient. 1985) and that the flow is of boundary-layer type. see Huang and Spaulding (1996). the transformation from Cartesian co-ordinates to σ co-ordinates leads to various cross derivatives.101) From this equation. ∂H (a) Hδσ > σ ∂H ∂x δx. If the same approach is used for the horizontal diffusion operator in the transport equation: ∂ 2 c∗ ∂2c ≈ . For a detailed description we refer to Chapter 9. truncations errors in the approximation of the horizontal diffusive fluxes in σ -co-ordinates are likely to become very large. Near steep bottom slopes or near tidal flats where the total depth becomes very small. for the horizontal diffusion term.19. In Delft3D-FLOW the stress tensor is redefined in the σ co-ordinate system assuming that the horizontal length scale is much larger than the water depth (Blumberg and Mellor. 1998). For example.

Since these boxes are not nicely connected to each other. see Figure 10.20. is not fulfilled. User Manual Figure 10. This “anti-creep” option is based upon a Finite Volume approach.5.104) 322 Deltares .Delft3D-FLOW. ∗ s1 = (c∗ 2 − c1 )/∆x and s2 = (c2 − c1 )/∆x: If s1 × s2 < 0 Then ∆c ∆x = 0 Else ∆c ∆x = sign (s1 ) × min (|s1 | . see Figure 10. an interpolation in z co-ordinates is required to compute the fluxes at the interfaces. A more accurate. even when the hydrostatic consistency condition Eq. The horizontal diffusive fluxes and baroclinic pressure gradients are approximated in Cartesian co-ordinates by defining rectangular finite volumes around the σ -co-ordinate grid points. A method has been implemented which gives a consistent. |s2 |) Endif (10.21.8 option Correction for sigma-co-ordinates.21. In Delft3D-FLOW an option is available that minimises artificial vertical diffusion and artificial flow due to truncation errors. stable and monotonic approximation of both the horizontal pressure gradient and the horizontal diffusion term. The values obtained from this interpolation ∗ are indicated by c∗ 1 and c2 respectively in Figure 10.20: Finite Volume for diffusive fluxes and pressure gradients Horizontal diffusion will lead to vertical transport of matter through vertical stratification interfaces (pycnocline) which is unphysical. a z -interpolation of the scalar concentration c is needed to compute strictly horizontal derivatives. see Section 4. Since the centres of the finite volumes on the left-hand side and right-hand side of a vertical interval are not at the same vertical level. strict horizontal discretization is needed. Stelling and Van Kester (1994) apply a non-linear filter to combine the two consistent approximations of the horizontal gradient.

By taking the minimum of the gradients. Let s4 be a consistent approximation of the horizontal gradient s4 = (s1 + s2 )/2. The horizontal fluxes are summed for each control volume to compute the diffusive transport. This limiter is fully analogous to the min-mod limiter used for the construction of monotone advection schemes (Hirsch. it is important to ensure that the difference operator is positive definite in order to get physically realistic solutions. The integration of the horizontal diffusion term is explicit with time step limitation: 1 ∆t ≤ DH 1 1 + ∆x2 ∆y 2 −1 . Slørdal (1997) suggested to take s4 as approximation of the horizontal gradient. He calls his approach the “modified Stelling and Van Kester scheme”.105) The derivatives are used in the integral for the baroclinic pressure force in the momentum equation: Px (x.106) Originally. Slørdal (1997) stated that the above approximation may sometimes produce errors of the same sign which leads to a systematic underestimation of the baroclinic pressure term. 1990). z ) = 1 ρ0 ζ g z ∂ρ (x. Since the same approximation of the horizontal gradient is used for the horizontal diffusion flux. s) ds. The maximum and minimum of a variable being transported by diffusion do not increase or decrease (min-max principle). this approach was implemented in Delft3D-FLOW. It is equivalent to linear interpolation at a Deltares 323 . Stelling and Van Kester (1994) show that. the derivative is directly set to zero. ∂x (10.21: Left and right approximation of a strict horizontal gradient If an interval has only grid boxes at one side. which selects the minimum of the two gradients under consideration. From numerical tests Slørdal (1997) concluded that the underestimation is reduced by increasing the vertical resolution. This underestimation can be ascribed to the non-linear filter. (1998) show that any nine-point consistent linear discretization of the horizontal diffusion on the σ -grid does not fulfil the min-max principle. Beckers et al. but is sometimes enhanced by increasing the horizontal resolution. the minmax principle is fulfilled. (10.Numerical aspects of Delft3D-FLOW Figure 10.

324 Deltares . Due to the related artificial mixing. unless the flow is dominated by the open boundary conditions. the stringent conditions of the minimum operator can be relaxed somewhat. The damping of the transient solution determines the spin-up time of the numerical model. The following non-linear approach presently available in Delft3D-FLOW is both consistent and assures the min-max principle: If s1 × s2 < 0 Then ∆c ∆x = 0 Elseif |s4 | < |s3 | Then ∆c ∆x = s 4 Elseif min (|s1 | . transport of matter discretised on a σ co-ordinate grid may still generate some numerical vertical diffusion by the discretisation of the advection terms. It may introduce wiggles and a small persistent artificial vertical diffusion (except for linear vertical density distributions). By introducing an additional approximation of the horizontal gradient in the filter algorithm defined by s3 = (c2 −c1 )/∆x. see Figure 10. |s2 |). User-prescribed (space varying or uniform) input fields (cold start). which is consistent for min (|s1 | . If the streamlines are strictly horizontal. The boundary conditions represent the external forcing and determine the steady state solution. In Delft3D-FLOW the initial conditions for the water level and velocities are obtained from: The results of a previous run (warm start). This will generate a transient solution consisting of waves with eigen frequencies of the model domain. The waves should be dissipated completely by bottom friction and viscosity terms or leave the domain through the open boundaries.Delft3D-FLOW. The increase in computation time is about 30%. |s2 |) Endif (10.107) The method requires a binary search to find the indices of neighbouring grid boxes. It is more accurate than taking the minimum of the absolute value of the two slopes s1 and s2 but it does not fulfil the min-max principle for the diffusion operator.11 Smoothing parameter boundary conditions The solution of the shallow water equations is uniquely determined by a set of initial and boundary conditions. These waves may be reflected at the boundaries and generate a standing wave system.22. The deviation between the initial condition and the steady state solution generates a transient (mass spring system analogy). The drawback of underestimation of the baroclinic pressure force reported by Slørdal (1997) can be minimised without loosing that the method fulfils the min-max principle. 10. This third gradient s3 . User Manual certain z -level before taking the gradient. The initial values are usually inconsistent with the boundary conditions at the start time of the simulation. |s2 |) < s3 < max (|s1 | . has point-to-point transfer properties and therefore leads to a positive scheme for sufficiently small time steps. stratification may disappear entirely for long term simulations. |s2 |) Then ∆c ∆x = s 3 Else ∆c ∆x = sign (s1 ) min (|s1 | . |s2 |) < |s3 | < max (|s1 | . which is time consuming.

5 0 50000 100000 150000 200000 250000 300000 350000 400000 Figure 10.5 -2 water levels -2. Deltares 325 . The boundary condition. 0. t < Tsmo . wave-current interaction) are modelled. The assumption is made that. the free surface waves can be propagated correctly.108) α= and: Tsmo −t Tsmo .5 -1 -1. Delft3DFLOW has an option to switch on the boundary forcing gradually by use of a smoothing period (parameter Tsmo ).5 1 0. The boundary forcing is given by: Fbsmo (t) = (1 − α)Fb (t) + αFi (t) .5 0 -0. 10. The time integration method strongly influences the wave propagation when applying a large time step. (10. The boundary condition after smoothing. t ≥ Tsmo . the geographical area. If the initial conditions are consistent with the boundary conditions at the start time of the simulation then the smoothing time should be set to zero. Smoothing is possible both for a warm and a cold start.22: Cold start with damping of eigen oscillations due to bottom friction To reduce the amplitude of the transient wave and the spin-up time of a model.Numerical aspects of Delft3D-FLOW 2 1.109) Fi (t) Fb (t) Fbsmo (t) The initial value at the boundary. and the physical processes (turbulence. with: (10. The accuracy of the solution depends not only on the numerical scheme. by restricting the computational time step.12 Assumptions and restrictions The solution of the discretized equations is just an approximation of the exact solution. but also on the way in which the bottom topography.

if the phase variation of the tidal forcing along an open boundary segment is non-linear then the number of open boundary segments should be increased so that the phases at all the segments can be specified. 326 Deltares . The open boundary can be divided into segments (sections). These reflections will be observed as spurious oscillations superimposed on the physical results. User Manual The open boundaries in a numerical flow model are artificial in the sense that they are introduced to limit the computational area that is modelled. Phase errors may generate an artificial re-circulation flow (eddy) near the open boundary. one for each segment end. calibration on processed field data obtained from a tidal analysis or Fourier analysis. In the numerical model. due to meteorological or other effects. Therefore. For steadystate simulations. weakly reflective boundaries are available which diminish these effects. The boundary conditions in Delft3D-FLOW are specified for these segments. a similar effect may be observed near the open boundaries if the effect of the Coriolis force on the water level gradient along the open boundary is not taken into account in the boundary conditions. For tidal flow computations. two values per segment are required. wave reflections may occur at the open boundaries. In Delft3D-FLOW. Care must be taken when time-series of measurements are directly prescribed as forcing functions at the open boundaries. The free surface waves should pass these boundaries completely unhindered. Measurements often contain a lot of undesired noise. The boundary condition at internal points within this segment is obtained by linearly interpolation between the end points. avoids this problem.Delft3D-FLOW.

y and εs.z ∂c( ) ∂z = 0.x . Other processes such as the effect that sediment has on the local mixture density. The formulation of several of these processes (such as. If the suspended load is known to be negligible (either due to sediment diameter or sediment transport formula chosen). the interaction of sediment fractions is important for many Deltares 327 . There are.1 General formulations Introduction The sediment transport and morphology module supports both bedload and suspended load transport of non-cohesive sediments and suspended load of cohesive sediments. and constituents.x ∂c( ) ∂x − ∂ ∂y ∂c( ) ∂y − ∂ ∂z ) ε( s.y ∂ ∂x ) ε( s. For schematisation we distinguish “mud” (cohesive suspended load transport). or vice versa. Computationally. however.1. such as salinity.1. and hence on turbulence damping. a number of important differences between sediment and other constituents. can also be taken into account. and the settling velocity of sediment under the action of gravity. Furthermore. if a net flux of sediment from the bed to the flow. The only difference between “bedload” and “sand” fractions lies in the fact that the suspended load advection-diffusion equation is not solved for the “bedload” fraction. the three-dimensional transport of sediment is computed in exactly the same way as the transport of any other conservative constituent. settling velocity. this especially applies for sand and mud. (11. 11. heat. the exchange of sediment between the bed and the flow. occurs then the resulting change in the bathymetry should influence subsequent hydrodynamic calculations.1) c( ) u. In addition.11 Sediment transport and morphology 11. These additional processes for sediment are obviously of critical importance. the “bedload” approach is more efficient. A model may contain a mixture of up to 99 suspended (i.e. sediment deposition and pickup) are sediment-type specific.1 11.2 Suspended transport Three-dimensional transport of suspended sediment is calculated by solving the three-dimensional advection-diffusion (mass-balance) equation for the suspended sediment: ∂c( ) ∂uc( ) ∂vc( ) ∂ w − ws + + + ∂t ∂x ∂y ∂z − where: ( ) c( ) + ) ε( s.z ( ) ws mass concentration of sediment fraction ( ) [kg/m3 ] flow velocity components [m/s] eddy diffusivities of sediment fraction ( ) [m2 /s] (hindered) sediment settling velocity of sediment fraction ( ) [m/s] The local flow velocities and eddy diffusivities are based on the results of the hydrodynamic computations. Sediment interactions are taken into account although especially in the field of sand-mud interactions still a couple of processes are lacking. “sand” and “mud”) fraction and an arbitrary amount of “bedload” fractions if computer memory and simulation time allows. εs. v and w ( ) ( ) ( ) εs. “sand” (non-cohesive bedload and suspended load transport) and “bedload” (non-cohesive bedload only or total load transport) fractions. for example.

2.8 and 10. The following sections describe.1. These negative concentrations can be suppressed by applying a horizontal Forester filter.4 and 11.3 Effect of sediment on fluid density By default Delft3D-FLOW uses the empirical relation formulated by UNESCO (1981a) to adjust the density of water in order to take into account varying temperature and salinity.6. 11. However. See the sections on sediment interaction (Sections 11. Remark: 328 Deltares . At the same time we discuss some of the differences in general terms and refer for the details of the mathematical formulations to Sections 11. Sections 4. This is achieved by adding (per unit volume) the mass of all sediment fractions. this can result in a substantially larger computing time. A vertical Forester filter applied in a sediment transport computation will not affect the sediments. and subtracting the mass of the displaced water. Remarks: The presence of multiple sediment fractions considerably complicates the calculation of the density of the bed and the availability of a particular class of sediment at the bed.Delft3D-FLOW.6.4. For sediment transport this relation is extended to include the density effect of sediment fractions in the fluid mixture. the differences between the suspended transport of sediments and the transport of other conservative constituents. as discussed below. You can include or neglect the effect of sediment on the fluid density by setting the DENSIN flag in the morphology input file.4). Since this filter smoothes the vertical profile and thus can have a strong influence on the vertical mixing processes.3) and bed composition models (Section 11. User Manual processes. at a conceptual level. As a mathematical statement this translates as: lsed ρmix S.3.2 and 11.2) ρw (S ) S ( ) ρs lsed specific density of water with salinity concentration S [kg/m3 ] salinity concentration [ppt] specific density of sediment fraction ( ) [kg/m3 ] number of sediment fractions Horizontal density gradients (now also due to differences in sediment concentrations) can create density currents. this vertical filter has been de-activated for sediments.5. c( where: ) = ρw (S ) + =1 c( ) 1− ρw (S ) ρs ( ) (11.4. It is suggested to accept small negative concentrations and to apply a Forester filter only when the negative concentrations become unacceptably large. Vertical density gradients can also have a significant effect on the amount of vertical turbulent mixing present. for instance the simultaneous presence of multiple suspended sediment fractions has implications for the calculation of the local hindered settling velocity of any one sediment fraction as well as for the resulting mixture density. Small negative sediment concentrations (−1 · 10−3 kg/m3 ) can be found in a computation.

A simple method of circumventing these problems can be achieved by setting DENSIN = . ( ) (11.5 Dispersive transport The eddy diffusivities εs. Our experience shows that this change in the free surface level (even if very slight) can lead to calibration problems when converting an existing 2DH model to 3D if the model is driven using water level boundary conditions. the magnitude of this error has been shown to be rather small (Lesser et al.false.. which has the effect of preventing the sediment from having any effect on the density of the water/sediment mixture. taking into account the effect of high sediment concentrations on damping turbulent exchange processes) and the influence of waves (due to wave induced currents and enhanced bottom shear stresses). the settling velocity of a single particle is reduced due to the presence of other particles.y and εs.1. The total mass concentration ctot m is the sum of the mass concentrations of the sediment fractions: lsed ctot m = =1 ) c( s .0 is the ‘basic’ sediment fraction specific settling velocity. ws.1 and 11. This reduction in bed shear stress is particularly pronounced when the k -ε turbulence closure model is used.x .Sediment transport and morphology This option is included as it has been found that a secondary effect of including sediment in the density calculations is a reduction of the flow velocity in the lower computational layers (when compared with a standard logarithmic velocity profile) and a consequent reduction in the computed bed shear stress.4) As the fall velocity is now a function of the sediment fractions concentration. In order to account for this hindered settling effect we follow Richardson and Zaki (1954) and determine the settling velocity in a fluid-sediment mixture as a function of the sediment concentration and the non-hindered settling fall velocity: ( ) ws ( ) = ctot 1− s Csoil 5 ws. see Sections 11. Delft3D-FLOW supports four so-called “turbulence closure models”: Constant coefficient.3) where Csoil is the reference density (input parameter). 11.1 for details.z depend on the flow characteristics (turbulence level.1. Remark: The process of sediment settling is computed with a first-order upwind numerical scheme. and leads to an increase in overall flow velocity and a consequent lowering of the free surface. this implies that each sediment fraction has a fall velocity which is a function of location and time.2.4 Sediment settling velocity The settling velocity ws for sand and mud are strongly different in formulation.0 . In high concentration mixtures. 2000). While use of the upwind settling formulation does slightly under-predict the mass of sediment settling. (11. 11. εs.3. Algebraic eddy viscosity closure model. Deltares 329 ( ) ( ) ( ) .

k -ε turbulence closure model.2. A constant eddy viscosity will lead to parabolic vertical velocity profiles (laminar flow). The other three turbulence closure models are based on the eddy viscosity concept of Kolmogorov (1942) and Prandtl (1945) and offer zero.1. vertical fluid mixing coefficient calculated by the selected turbulence model. Important wave effects such as streaming in the wave boundary layer and wave induced turbulence were not accounted for. The first is a simple constant value which is specified by you. This is a limitation of the present implementation. and second order closures for the turbulent kinetic energy (k ) and for the mixing length (L). 2 The production of turbulent energy associated with wave breaking is incorporated by introducing an extra source term in the kinetic energy and dissipation equations of the k -ε turbulence model. The problem of three dimensional wave effects has been studied by Walstra and Roelvink (2000). 4 Streaming (a wave induced current in the wave boundary layer directed in the wave propagation direction) is modelled as a time averaged shear stress. The output of a turbulence closure model is the eddy viscosity at each layer interface. Remarks: For cohesive sediment fractions the extra turbulent mixing due to waves is not yet included in the eddy diffusivity. first. from this the vertical sediment mixing coefficient is calculated: ) ε( s = βεf . (11.Delft3D-FLOW. See also Section 11. See Section 9. cohesive sediment fractions and fine sand (< 150 µm): 1. For non-cohesive sediment the effect of waves is accounted for by using a modified or effective ‘beta’ factor of Van Rijn (k -ε model) or by using a separate formula to compute εf (algebraic or k -L) model.6 Three-dimensional wave effects Traditionally wave effects were only incorporated in a depth-averaged manner via a (breaking) wave induced shear stress at the surface.3. See also Section 11. 11. their main suggestions for improvement are: 1 The wave induced mass flux is corrected with the second order Stokes drift. 330 Deltares .0.2.5 for a full description of the available turbulence closure models. a wave induced mass flux and an increased bed shear stress. 3 The production of turbulent energy associated with dissipation in the near-bed wave boundary layer is incorporated by introducing an extra source term in the kinetic energy and dissipation equations of the k -ε turbulence model. User Manual k -L turbulence closure model. All three of the more advanced turbulence closure models take into account the effect that a vertical density gradient has on damping the amount of vertical turbulent mixing.5) where: εs β ( ) εf vertical sediment mixing coefficient for the sediment fraction ( ) non-cohesive sediment: Van Rijn’s ‘beta’ factor or effective ‘beta’ factor.2.

7 Initial and boundary conditions To solve Equation 11. a uniform zero concentration for the non-cohesive sediment fractions is usually adequate.1. i. so in the case of a cold start where the hydrodynamic model also takes some time to stabilise. see Sections 11. which can cross this boundary). This is left unchanged for suspended sediment.7. Water surface boundary The vertical diffusive flux through the free surface is set to zero for all conservative constituents (except heat.7. 11. In many practical applications the non-cohesive sediment concentrations adapt very rapidly to equilibrium conditions. Deltares 331 .2.5 and 11. In these options cohesive and non-cohesive sediment fractions are treated in the same way.6) where z = ζ is the location of the free surface. These effects are important when computing the transport of sediment in wave and current situations. for full details you are referred to Section 9.1 Initial condition The initial conditions for the sediment fractions are handled in exactly the same manner as those for any other conservative constituent. Space-varying initial concentrations read from a user-defined input file.2 Boundary conditions For each of the model boundaries you must prescribe the boundary condition for each sediment fraction.Sediment transport and morphology These improvements have been implemented in Delft3D-FLOW.3. ∂z at z = ζ (11.e.1. 11. Space-varying initial concentrations read from a restart file generated by a previous run.7. you can specify: One global initial concentration for each sediment fraction.4 for full details regarding their effect on cohesive and non-cohesive sediments respectively. ( ) ( ) ) −ws c − ε( s. We discuss in short the general type of conditions and refer for the details to the sections to follow.1.1 you need to prescribe initial and boundary conditions for each suspended sediment fraction.z ∂c( ) = 0. 11.

To assist with modelling coarser material an additional option has been included. or step distribution over the vertical. zero concentration gradient at the boundary.7) where: D( ) E( ) sediment deposition rate of sediment fraction ( ).3. The calculated fluxes are also applied to the bed in order to update the bed level. The formulations of D ( ) and E ( ) strongly differ for cohesive and non-cohesive sediment. sediment erosion rate of sediment fraction ( ). 11. cohesive sediment than for sand-sized particles. Open inflow boundaries Delft3D-FLOW requires you to specify boundary conditions for all conservative constituents at all open inflow boundaries. EqmBc) and/or NeuBcMud to true in the morphology input file.z ∂c( ) = D( ) − E ( ) . This will generally be the desired situation if the model boundaries are well chosen. The boundary condition at the bed is given by: ( ) ( ) ) −ws c − ε( s. All of these options are also available for sediment constituents. ∂z at z = zb (11. This option allows you to specify that.2.3 and 11. These fluxes are then applied to the bottom computational layer by means of a sediment source and/or sink term in each computational cell. This feature has been implemented as a Neumann boundary condition. This feature can be activated for sand and mud fraction separately by setting NeuBcSand (previously. a near-perfectly adapted flow will enter the domain and very little accretion or erosion should be experienced near the model boundaries.Delft3D-FLOW.4 respectively. linear. for the details you are referred to Sections 11. You may also choose to specify a “Thatcher-Harleman” return time to simulate the re-entry of material that flowed out of the model after the flow reverses direction. although they are probably more appropriate for fine. effectively this means that the dispersive transport of sediment at the outflow boundary is neglected compared to the advective transport. that is. User Manual Bed boundary condition The exchange of material in suspension and the bed is modelled by calculating the sediment fluxes from the bottom computational layer to the bed. and vice versa. When modelling in three dimensions you may choose to specify boundary concentrations that have a uniform.2 Cohesive sediment 332 Deltares . the flow should enter carrying all “sand” sediment fractions at their “equilibrium” concentration profiles. at all open inflow boundaries. Open outflow boundaries No boundary condition is prescribed at outflow boundaries. By setting the the sediment concentrations at the boundary equal to those just inside model domain.

maximal salinity at which WSM is specified ws. 11. τcr. settling velocity of sediment fraction ( ) at salinity concentration SALMAX WS0.Sediment transport and morphology 11.2 Cohesive sediment dispersion The vertical mixing coefficient for sediment is equal to the vertical fluid mixing coefficient calculated by the selected turbulence closure model.e . is the settling velocity of the fraction in water having a salinity equal to SALMAX.12) cb = c( where: Deltares ( ) ) z= ∆ zb .8) where: ws. with the degree of flocculation depending on the salinity of the water.e.0 =  ( )  ws. i. These flocs are much larger than the individual sediment particles and settle at a faster rate. The settling velocity of the sediment flocs is calculated as follows: ( ) ws.f 2 ( ) ) . when S ≤ Smax 1 + cos( SπS max when S > Smax (11. is taken to be the settling velocity of the sediment fraction in fresh water (salinity = 0). In order to model this salinity dependency you must supply two settling velocities and a maximum salinity.0 ( ) ws.9) where: εs εf ( ) vertical sediment mixing coefficient for sediment fraction ( ) vertical fluid mixing coefficient calculated by the selected turbulence closure model 11.1 Cohesive sediment settling velocity In salt water cohesive sediment tends to flocculate to form sediment “flocs”.2.f S Smax ( ) Remarks: Modelling turbulence induced flocculation or the break-up of sediment flocs is not yet implemented. WS0. The influence of flocculation is disregarded by setting WSM = WS0. (11.2. ( ) (11. ( ) D ( ) = ws cb S τcw .2. max 2 w( ) .10) (11. fresh water settling velocity of sediment fraction ( ) salinity SALMAX. The first velocity. τcr. 2 333 .3 Cohesive sediment erosion and deposition For cohesive sediment fractions the fluxes between the water phase and the bed are calculated with the well-known Partheniades-Krone formulations (Partheniades.d .t . s.max ) + 1 − cos( SπS max ws.max ( ) the (non-hindered) settling velocity of sediment fraction ( ) WSM.11) (11.: ) ε( s = εf . 1965): ( ) E ( ) = M ( ) S τcw . The second velocity. WSM.

but the latter behaviour may be activated by setting the OldMudFrac keyword tot true in the morphology input file.e ( ) τcr.d ( ) = τcr. when τcw ≤ ( ) τcr.e =      0. τcr.e . ( ) (11. Remarks: 334 Deltares .e .13) deposition flux [kg m−2 s−1 ] fall velocity (hindered) [m/s] average sediment concentration in the near bottom computational layer deposition step function: S τcw . If the bed stratigraphy is modelled in detail. The formulations given in the previous section have been formulated for uniform cohesive sediment beds.d maximum bed shear stress due to current and waves as calculated by the wave-current interaction model selected by the user.d . one must assume either that the bed composition is almost uniform (in which case the erosion rate can again be assumed to be proportional to the bed composition) or that the cohesive sediment fraction considered forms a layer that covers the other sediment fractions (in this case the erosion rate of the cohesive sediment will not be reduced). In such cases the erosion rate will be affected. cw cr. see Section 9. On the other hand if the bed stratigraphy is not explicitly included in the model and only the overall characteristics of the local bed composition is known. it may be assumed that the erosion rate is proportional to the availability of the sediment fraction considered in the top-most layer of the bed stratigraphy. often the bed will be made up of a range of sediment types and sizes. τcr.d τcw ( ) ( )      τ  1 − cw  . when τcw > τcr.e erosion flux [kg m−2 s−1 ] user-defined erosion parameter EROUNI [kg m−2 s−1 ] erosion step function: S D( ) ( ) ws ( ) cb S τcw . and diffusion through the bottom of the near bottom computational cell are all set to zero to prevent double counting these fluxes.14) τcr.d ( ) ( ) τcw .Delft3D-FLOW. Advection.d    ( ) 0.7 for full details user-defined critical erosion shear stress TCEUNI [N/m2 ] user-defined critical deposition shear stress TCDUNI [N/m2 ] Remark: Superscript ( ) implies that this quantity applies to sediment fraction ( ).e − 1 .2. when τ < τ ( ) . 11. τcw ( ) τcr. (11. τcr. The former approach is nowadays the default approach for the online-morphology module. However. User Manual E( ) M( ) ( ) S τcw . particle settling.4 Interaction of sediment fractions The following notes hold only in case of multiple sediment fractions. The calculated erosion or deposition flux is applied to the near bottom computational cell by setting the appropriate sink and source terms for that cell. when τcw ≥ τcr. τcr.

2. 11.bcc> file are overruled.3. 11. This reduction factor is related to the formulations implemented for non cohesive sediment transport (see Sections 11. Although this option may sometimes be very useful.7 for details. see Section 9.6 Inclusion of a fixed layer If the thickness of the sediment layer becomes small then the erosion flux is reduced by a factor fFIXFAC as defined in Section 11. Assuming that the erosion rate is independent of the availability of the sediment fraction considered will lead to an overestimation of the erosion rate.2. So.4. For cohesive material you can specify that. this is not always possible in long-term simulations.2.7 Inflow boundary conditions cohesive sediment Although it is general good advice to locate the open boundaries sufficiently far away from the area of interest. one must be careful when applying it: the sediment concentration of the incoming flow may take on any value that does not lead to significant deposition in the first grid cell. Delft3DFLOW allows for a different approach. This is a limitation of the present implementation. a unique “equilibrium” concentration (profile) does often not exist due to differences in critical shear stresses for erosion and sedimentation.5) Deltares 335 . the values specified in the <∗.9. the concentrations to be applied at the inflow boundaries are read from the <∗. If the parameter is set to true. Van Rijn (2000) approach is followed by default. The mud concentrations are in general more loosely coupled to local morphology than the concentrations of coarser non-cohesive sediment fractions. at all open inflow boundaries. which has to be created with the FLOW User Interface. By setting NeuBcMud = false.bcc> file. For instance. 11. if the model includes two equal cohesive sediment fractions their total transport rate will be double that of the rate observed in an identical simulation carried out using the total amount of the two sediment fractions in the former simulation. In such cases it is desirable to impose some kind of equilibrium boundary conditions.4. You can also specify a number of other transport formulations (see Section 11. This implies that the extra turbulent mixing due to waves will not be included in the suspended sediment transport calculations (for these sediment fractions) except by way of the enhancement of the bed shear stress caused by wave-current interaction.5 Influence of waves on cohesive sediment transport For cohesive sediment fractions the sediment mixing coefficient will still be set following Equation 11.5 and 11.Sediment transport and morphology Assuming an erosion rate proportional to the availability of the sediment fraction considered may result in a significant underestimation of the erosion rate if the bed is modelled as a single uniformly mixed layer (default setting) and the mud contents is low. This feature is enabled by setting NeuBcMud in the morphology input file to true (Neumann boundary condition: concentration gradient perpendicular to open boundary equal to zero).3 Non-cohesive sediment For the transport of non-cohesive sediment. the flow should enter carrying the same mud concentration as computed in the interior of the model.4 for suspended and bedload transport respectively). 11.4.

3. 100 µm < Ds ≤ 1000 µm   Ds ν2       ( ) 1.5. Remark: In the case of non-uniform bed material Van Rijn (1993) concluded that. ( ) (11.24) vertical fluid mixing coefficient calculated by the selected turbulence closure model Deltares 336 .1 Non-cohesive sediment settling velocity The settling velocity of a non-cohesive (“sand”) sediment fraction is computed following the method of Van Rijn (1993). Expressed mathematically: ) ε( s = βεf .15) where: s( ) ( ) Ds ν Ds ( ) relative density ρs /ρw of sediment fraction( ) representative diameter of sediment fraction ( ) kinematic viscosity coefficient of water [m2 /s] is the representative diameter of the suspended sediment given by the user-defined sediment diameter SEDDIA (D50 of bed material) multiplied by the user-defined factor FACDSS ( ) (see also remarks).1 (s( ) − 1)gDs .1 Using the algebraic or k -L turbulence model Without waves If the algebraic or k -L turbulence model is selected and waves are inactive then the vertical mixing coefficient for sediment is computed from the vertical fluid mixing coefficient calculated by the selected turbulence closure model. This value of Ds will be overruled if IOPSUS=1 and the transport formula of Van Rijn (1993) has been selected. on the basis of ( ) measurements. from this the vertical sediment mixing coefficient is calculated using the following expressions: 11. 65 µm < Ds ≤ 100 µm    18ν    ( )3 0. User Manual 11.3. The formulation used depends on the diameter of the sediment in suspension: ws.01(s( ) − 1)gDs 10ν  = 1 + − 1 .3.0 ( )  ( )2  (s( ) − 1)gDs   .16) where: εs β ( ) εf ( ) vertical sediment mixing coefficient for sediment fraction ( ) Van Rijn’s ‘beta’ factor for the sediment fraction (see Equation 11. Ds should be in the range of 60 to 100% of D50 of the bed material. For non-cohesive sediment the fluid mixing coefficient is multiplied by Van Rijn’s ‘beta factor’ which is intended to describe the different diffusivity of a fluid ‘particle’ and a sand grain.2 Non-cohesive sediment dispersion The output of a turbulence closure model is the eddy viscosity at each layer interface. 11.Delft3D-FLOW.2. 1000 µm < Ds ( ) (11. If the bed material is very widely graded (well sorted) consideration should be given to using several sediment fractions to model its behaviour more accurately.1 for details. see Section 11.

Sediment transport and morphology Figure 11. when z < 0. when z ≥ 0. using expressions given by Van Rijn (1993) for both the current-related and wave-related vertical turbulent mixing of sediment. with a linear transition between the two hinge points.w  () ( )ˆ  εs.004D∗ δs Uδ .bed + ε( s.17) where: εs. s. The expressions used to set this distribution are: ( ) εs.5h. ( ) z − δs 0.25κβu∗.5h − ( ) ( ) δs . The wave-related mixing is also calculated following Van Rijn (1993). when z ≥ 0.5h.      () ( ) ) εs.5h.bed =    0.max − εs.bed = 0.1: Sediment mixing coefficient in non-breaking waves (Source: Van Rijn (1993)) Including waves If waves are included in a simulation using the algebraic or k -L turbulence closure model then the sediment mixing coefficient for non-cohesive sediment fractions is calculated entirely separately from the turbulence closure model.c ( ) u∗.1.c = κβu∗. The current-related mixing is calculated using the ‘parabolic-constant’ distribution recommended by Van Rijn: ) ε( s. In this case Van Rijn recommends a smoothed step type distribution over the vertical. see Figure 11. ( ) (11. (11. The turbulent mixing in the upper half of the water column is generally of little importance to the transport of ‘sand’ sediment fractions as sediment concentrations in the upper half of the water column are low.18) Deltares 337 . when δs < z < 0.c vertical sediment mixing coefficient due to currents (for this sediment fraction) current-related bed shear velocity In the lower half of the water column this expression should produce similar turbulent mixing values to those produced by the algebraic turbulence closure model. 0.max = Tp when z ≤ δs .5h.c z (1 − z /h).c h.035γbr hHs  )   ε( .

21) wave-related bed roughness (as calculated for suspended sediment transport) We calculate the total vertical sediment mixing coefficient by following Van Rijn and taking the sum of the squares: ) ε( s = εs. max (5γbr δw . ( )2 ( )2 (11.5 when when Hs h Hs h > 0.24) Deltares .c + εs. given by: ( ) δs = min [0.4 ≤ 0. (11.23) where: εs ( ) βeff ( ) vertical sediment mixing coefficient of sediment fraction ( ) the effective Van Rijn’s ‘beta’ factor of sediment fraction ( ) As the beta factor should only be applied to the current-related mixing this ( ) c is estimated as: βeff = 1 + β ( ) − 1 τwτ +τc . max {0. 1984b): β 338 ( ) =1+2 ws u∗. ( ) (11.4 (11.4 0.1.19) where: δw thickness of the wave boundary layer: ˆδ δw = 0.c ( ) 2 . using the following expression: ) ε( s = βeff εf .24 bed shear stress due to currents bed shear stress due to waves vertical fluid mixing coefficient calculated by the k -ε turbulence closure model β( ) τc τw εf Van Rijn’s ’beta’ factor is calculated from (Van Rijn.Delft3D-FLOW.072A γbr ˆδ A ks.2. 11. Equation 11.w −0. 10γbr ks.25 (11.2 Using the k -ε turbulence model In the case of the k -ε turbulence closure model the vertical sediment mixing coefficient can be calculated directly from the vertical fluid mixing coefficient calculated by the turbulence closure model. for non-cohesive sediment fractions Van Rijn’s ‘beta’ factor of the sediment fraction ( ).w )}] ( ) (11.w 1+ 1 Hs h − 0.w .3.5. User Manual where δs (the thickness of the near-bed sediment mixing layer) is estimated using Van Rijn’s formulation.22) where εs is the vertical sediment diffusion coefficient used in the suspended sediment transport calculations for this sediment fraction.20) empirical coefficient related to wave breaking: γbr = ks.

This implies that the value of β ( ) is space (and time) varying. 0. The default transport formula (Van Rijn.3.4 on bed composition models). This precaution is only likely to come into effect in very shallow areas.01h . 0.6.5. The reference concentration is adjusted proportional to the relative availability of the sediment fraction in the top-layer of the bed (see Section 11. sand). With the keyword IOPKCW you have two options to calculate ks (and kw ): ks derived from current-related effective roughness height as determined in the Delft3DFLOW module (spatially varying) and kw = RW AV E · ∆r . Remarks: In a wave and current situation Van Rijn (1993) applies the β -factor to only the currentrelated turbulent mixing. little is known about the dependence of the β -factor on flow conditions. The reference height is given by: a = min max AKSFAC · ks .5. however it is constant over the depth of the flow.25) a AksFac ks ∆r h Van Rijn’s reference height user-defined proportionality factor (morphology input file) user-defined current-related effective roughness height (see options below) wave-induced ripple height. Calculation of the reference concentration The reference concentration ca is calculated directly by the sediment transport formula or it is derived from the suspended sediment transport rate given by the sediment transport formula as ca = Ss /Hu .20h . 2 (11. 1993) includes a formula for the reference concentration (see Section 11. we follow the method of Van Rijn (1993) for the combined effect of waves and currents. due to the limited knowledge of the physical processes involved. The k -ε turbulence closure model has been extended by Walstra et al. and u∗.g. (2000) to include the three-dimensional effects of waves. whereas we apply it to the total turbulent mixing calculated by the selected turbulence closure model. this discrepancy is expected to be of little importance in practical situations. ( ) 11.025 m water depth Remark: Van Rijn’s reference height a is limited to a maximum of 20% of the water depth.1). set to a constant value of 0.Sediment transport and morphology Where ws is the settling velocity of the non-cohesive sediment fraction. However the effect of wave asymmetry on the bedload transport is not yet included. Deltares 339 .3 Reference concentration For non-cohesive sediment (e. where: ∆r . ks and kw specified by you (constant in space). we follow Van Rijn (1993) and limit β ( ) to the range 1 < β ( ) < 1. In addition.c is the local bed shear stress due to currents. However.

whereas a source term is solved explicitly.4 Non-cohesive sediment erosion and deposition The transfer of sediment between the bed and the flow is modelled using sink and source terms acting on the near-bottom layer that is entirely above Van Rijn’s reference height.Delft3D-FLOW.2: Selection of the kmx layer. The sediment concentrations in the layer(s) that lie below the kmx layer are assumed to rapidly adjust to the same concentration as the reference layer. Each half time-step the source and sink terms model the quantity of sediment entering the flow due to upward diffusion from the reference level and the quantity of sediment dropping out of the flow due to sediment settling. A sink term is solved implicitly in the advection-diffusion equation. The required sink and source terms for the kmx layer are calculated as follows. see Figure 11. User Manual KEY Standard computational cell Reference cell for “sand” sediment Concentration set equal to concentration of reference layer for ‘sand’ sediment calculations Coarse Grid Medium Grid Fine Grid kmx kmx kmx kmx Layer a BED BED BED Figure 11. 340 Deltares . This layer is identified as the reference layer and for brevity is referred to as the kmx-layer. 11.2.3: Schematic arrangement of flux bottom boundary condition Remark: The reference concentration and therefore the suspended load can be calibrated using the keyword Sus in the morphology input file.3. where a is Van Rijn’s reference height dc D eposition Flux = wsc Erosion Flux = εs dz km x layer ∆z a BED Figure 11.

29) 341 .4). (11.4: Approximation of concentration and concentration gradient at bottom of kmx layer In order to determine the required sink and source terms for the kmx layer. the exponent A( ) can be determined.Sediment transport and morphology Approximation to concentration gradient at bottom of kmx layer Approximation to concentration at bottom of kmx layer H e ig h t a b o v e b e d kmx + ckmx Rouse profile + ca ∆z a ckmxbot BED Concentration Figure 11.26) c( ) ( ) ca a h z A( ) concentration of sediment fraction ( ) reference concentration of sediment fraction ( ) Van Rijn’s reference height water depth elevation above the bed Rouse number As the reference concentration and the concentration in the centre of the kmx layer ckmx are known. the concentration and concentration gradient at the bottom of the kmx layer need to be approximated.27) The concentration at the bottom of the kmx layer is: ( ) ckmx(bot) ) c( a = a(h − zkmx(bot) ) zkmx(bot) (h − a) A( ) (11. c where: ( ) = ) c( a a(h − z ) z (h − a) A( ) . We assume a standard Rouse profile between the reference level a and the centre of the kmx layer (see Figure 11. ( ) ckmx = ) c( a a(h − zkmx ) zkmx (h − a) A( ) ⇒A ( ) ln = ln ckmx ca a(h−zkmx ) zkmx (h−a) (11.28) We express this concentration as a function of the known concentration ckmx by introducing a correction factor α1 : ckmx(bot) = α1 ckmx Deltares ( ) ( ) ( ) (11.

31) We express this gradient as a function of the known concentrations ca and ckmx by introducing another correction factor α2 : c kmx(bot) = α2 ( ) ( ) ckmx − ca ∆z ( ) ( ) (11. User Manual The concentration gradient of the Rouse profile is given by: ∂c( ) ) = A( ) c( a ∂z a(h − z ) z (h − a) A( ) −1 · z 2 (h −ah − a) (11. The second can only be evaluated implicitly and is implemented as a (positive) sink 342 Deltares .34) where: α2 ( ) εs ( ) ( ) correction factor for sediment concentration sediment diffusion coefficient evaluated at the bottom of the kmx cell of sediment fraction( ) reference concentration of sediment fraction( ) average concentration of the kmx cell of sediment fraction( ) difference in elevation between the centre of the kmx cell and Van Rijn’s reference height: ca ( ) ckmx ∆z ∆z = zkmx − a The erosion flux is split in a source and sink term: E( ) ≈ ( ) ( ) ( ) α εs ckmx α2 εs ca − 2 . (11.35) The first of these terms can be evaluated explicitly and is implemented as a sediment source term.Delft3D-FLOW.33) where εs and ( ) ∂c( ) ∂z We approximate this expression by: ) E ( ) ≈ α 2 ε( s ( ) ca − ckmx ∆z ( ) ( ) . (11.30) The concentration gradient at the bottom of the kmx layer is: ( ) kmx(bot) ) A( ) c( a c = a(h − zkmx(bot) ) zkmx(bot) (h − a) A( ) −1 · −ah 2 zkmx (bot) (h − a) (11. ∂z are evaluated at the bottom of the kmx layer. ∆z ∆z ( ) ( ) ( ) (11.32) Erosive flux due to upward diffusion The upward diffusion of sediment through the bottom of the kmx layer is given by the expression: ) E ( ) = ε( s ∂c( ) .

We set: ( ) ( ) ckmx(bot) = α1 ckmx . The deposition flux is approximated by: ( ) D( ) ≈ α1 ckmx ws .41) The total source and sink terms is given by: ) Source( ) = α2 c( a ( ) εs ∆z εs ∆z ( ) ( ) . ( ) ( ) ( ) ( ) ( ) (11.37) Deposition flux due to sediment settling The settling of sediment through the bottom of the kmx cell is given by the expression: ( ) D ( ) = ws ckmx(bot) . ( ) ( ) (11.42) Sink ( ) = α2 ( ) (11.4. Deltares 343 .Sediment transport and morphology term.40) This results in a simple deposition sink term: ( ) Sinkdeposition = α1 ckmx ws . ( ) + α1 ws ckmx . ( ) ( ) ( ) (11.43) These source and sink terms are both guaranteed to be positive.3.39) (11.36) (11.5 Inclusion of a fixed layer The bedload transport is reduced if the thickness of the sediment layer becomes small (see Section 11.38) where ws and ckmx(bot) are evaluated at the bottom of the kmx layer. ( ) (11. The same effect has been implemented as a reduction for the entrainment and deposition terms as well as the equilibrium concentration by a factor fFIXFAC if erosion is expected to occur. Thus: Sourceerosion = Sinkerosion ( ) ( ) α2 εs ca ∆z ( ) ( ) ( ) α εs ckmx = 2 ∆z ( ) ( ) ( ) (11.4). 11.

1.g. 11. the bedload transport direction ϕτ is given by the following formula: tan(ϕτ ) = in which u v − αI U Is v u − αI U Is (11. This parameter used to be called EqmBc. In the latter case the initial transport direction will be assumed to be equal to the direction of the characteristic (near-bed) flow direction. the flow should enter carrying the same concentration of sediment as computed in the interior of the model. This means that the sediment load entering through the boundaries will be near-perfectly adapted to the local flow conditions and very little accretion or erosion should be experienced near the model boundaries.bcc> file are overruled. the secondary flow/spiral flow intensity Is optionally computed by the flow module may be taken into account.1 Basic formulation For simulations including waves the magnitude and direction of the bedload transport on a horizontal bed are calculated using the transport formula selected assuming sufficient sediment and ignoring bed composition except for e.5.4 Bedload sediment transport of non-cohesive sediment Bedload (or. we have implemented the open boundary condition in a consistent manner. User Manual 11. at all open inflow boundaries. the values specified in the <∗. This feature is enabled by setting NeuBcSand in the morphology input file to true (Neumann boundary condition: concentration gradient perpendicular to open boundary equal to zero). Although equilibrium boundary conditions may be better defined for non-cohesive sediments than for cohesive sediments. for the simpler transport formulae. corrected for bed-slope effect and upwind bed composition and sediment availability. Some of the sediment transport formulae prescribe the bedload transport direction whereas others predict just the magnitude of the sediment transport.4. If the parameter is set to true. total load) transport is calculated for all “sand” and “bedload” sediment fractions by broadly according to the following approach: first. This method gives the correct results even when the turbulent mixing profile is clearly non-parabolic. the magnitude and direction of the bedload transport at the cell centres is computed using the transport formula selected (See Section 11. This will generally be the desired situation if the model boundaries are well chosen. By setting NeuBcSand = false. which has to be created with the FLOW User Interface.6 Inflow boundary conditions non-cohesive sediment Although it is general good advice to locate the open boundaries sufficiently far away from the area of interest. In the case of a depth-averaged simulation. subsequently the transport rates at the cell interfaces are determined. The default sediment transport formula is Van Rijn (1993) as documented in Section 11.Delft3D-FLOW. hiding and exposure effects on the critical shear stresses. For non-cohesive suspended material you can specify that.5).bcc> file.3. 11.45) Deltares 344 . In such cases it is desirable to impose some kind of equilibrium boundary conditions. the concentrations to be applied at the inflow boundaries are read from the <∗. this is not always possible in long-term simulations.44) √ 2 1 g αI = 2 Es 1 − κ 2 κC (11.

Meyer-Peter & Muller.Sediment transport and morphology where: g κ C U Es gravitational acceleration ´ an ´ constant Von Karm ´ roughness Chezy the depth-averaged velocity coefficient to be specified by you as Espir keyword in the morphology input file The default value of Es is 0. These three transport components can be calibrated independently by using the respective keywords Bed. Sbc bedload due to waves. for “sand” sediment fractions. Ssw . 11. i. the suspended sediment fluxes below the reference height a are derived by means of numerical integration from the suspended transport rates. Ashida & Michiue and optionally the user-defined formula. i. This suspended load correction is included in the depth-averaged suspended load written to the output files of the program. Van Rijn (1984). BedW and SusW in the morphology input file. However. This effect is only included for transport formulae that return the bedload transport rate but not its direction. General formula.4.4. 11.3 Interaction of sediment fractions The following notes hold only in case of multiple sediment fractions. Engelund & Hansen. In order to prevent double counting. The spiral flow effect is of crucial importance in a depth-averaged simulation to get pointbar formation in river bends. Sediment fractions may interacted in several ways: reference concentrations. erosion rates and sediment transport rates will be reduced proportional to the availability of sediment fraction considered in the bed (less of the fraction available for transport) sediment fractions of different sizes influence each other by means of hiding and exposure: fine sediments hide among coarse sediments and are thereby partly shielded from the main flow while the coarser sediments are more exposed than they would be among other sediments of the same size.e.2 Suspended sediment correction vector The transport of suspended sediment is computed over the entire water column (from σ = −1 to σ = 0). The Van Rijn (1993) formula distinguishes the following transport components that are all treated like bed or total load. The opposite of these fluxes are scaled with the upwind sediment availability and subsequently imposed as corrective transport. This effect is taken into account by increasing the effective critical shear stress for fine sediments while lowering it for coarse sediments. Van Rijn regards sediment transported below the reference height a as belonging to “bedload sediment transport” which is computed separately as it responds almost instantaneously to changing flow conditions and feels the effects of bed slopes. Sbw suspended load due to waves. which implies that the spiral flow effect on the bedload transport direction is not included. Deltares 345 .e without relaxation effects of an advection diffusion equation: bedload due to currents.

The hiding and exposure effect has been implemented for the following transport formulae containing a critical shear stress: Meyer-Peter & Muller. Ashida-Michiue and optionally the user-defined formula.51) Sb magnitude of the bedload transport vector (before correction for bed slope effects) Deltares 346 . If the quantity of sediment available is less than the threshold then the magnitude of the calculated bedload transport vector is reduced as follows: Sb = fFIXFAC Sb . Kellermann & Loy formulation ξ= Dm Di α . User Manual This adjustment is carried out using a multiplicative factor ξ .4 Inclusion of a fixed layer Inclusion of a fixed layer implies that the quantity of sediment at the bed is finite and may. 11.46) • Ashida & Michiue formulation  Dm   0. the troughs of the bedforms will start to expose the non-erodible layer before sediment runs out completely. (11.4. This effect is taken into account in the bedload formulations by comparing the thickness of the sediment layer available at the bed with a user-defined threshold value. This results in a gradual reduction of the transport capacity over a certain sediment thickness indicated by THRESH. become exhausted and be unavailable to supply sediment to suspended and bedload transport modes.38889 2 otherwise .47) • Parker. (11. • Wu. In case the bed is covered by bedforms. The following formulations have been implemented: • No hiding and exposure correction (ξ = 1) • Egiazaroff formulation ξ= 10 log 19 10 log 19 2 i /Dm ) + 10 log (D .49) ξ = 1 − ϕ( ϕ( ) (11. Klingeman & McLean or Soehngen. (11. if excessive erosion occurs.50) where m is given by the MWWJHE keyword in the morphology input file. where: (11. general formula.48) where α is given by the ASKLHE keyword in the morphology input file.8429 Di 10 log 19 ξ=   10 log 19 + 10 log (D /D ) i m if Di /Dm < 0.Delft3D-FLOW. Wang & Jia formulation ϕ( ) = i ( ) η (i) D ( i) D ( ) − D ( i) ) m (11.

5). see Figure 11. It is possiDeltares 347 .5 the bedload transport component (n.5.Sediment transport and morphology fFIXFAC DPSED THRESH upwind fixed layer proximity factor: fFIXFAC = range 0 ≤ fFIXFAC ≤ 1.x and the component Sb. but in general it will follow the imposed behaviour closely. a boundary condition for the bed composition is also needed at inflow boundaries. rather than at the water level points where Sb. In the latter case the bedload transport rates are known from the model input. the bedload transport vector components are actually required at the U and V velocity points.9. 11. that is at the grid cell just outside the model domain.y .2.9 for imposing various morphological boundary conditions.x and Sb.m) (n+1.uu is set equal to Sb. THRESH limited to the The equilibrium suspended load concentration in the sediment pickup term is reduced by the same fixed layer proximity factor (in this case of course the local value and not some upwind value is used since suspended sediment pickup has no associated horizontal direction).6 Bedload transport at U and V velocity points As the control volume for bed level change calculations is centred on the water level points.3. For each active velocity point the upwind direction is determined by summing the bedload transport components at the water level points on either side of the velocity point and taking the upwind direction relative to the resulting net transport direction. By default.m) Sb. 11. A consequence of this approach is that the bed level at the first grid cell inside the model domain will not exactly behave as you imposed.m) (n. we use a simple “upwind” numerical scheme to set the bedload transport components at the U and V points as this ensures that the bed will remain stable.vv is set equal to Sb. In the example shown in Figure 11. In effect.2 and B. See Appendices B. It is likely that these areas will accrete a little sediment in order to allow an equilibrium bedload transport pattern to develop. The bedload transport component at the velocity point is then set equal to the component computed at the water level point immediately “upwind” (see Figure 11. because of the upwind approach used to transfer the bedload transport components to the U and V velocity points. This effect has also been included for cohesive and non cohesive suspended sediment as indicated in Sections 11.5. One implication of the use of this rather simple approach is that a finite (although always less than the user-defined threshold) thickness of sediment is required at the bed if a non-zero magnitude of the bedload transport vector is required. if the quantity of the sediment at the bed is limited.4. depth of sediment available at the bed user-defined erosion threshold DPSED . Remarks: Areas may be initially specified as containing zero bottom sediment if non-erodible areas are required.6 and 11. whereas in the former case the effective bedload transport rates at the boundary could be derived from the mass balance at the open boundary point. The bed level boundary condition is imposed at the same location where a water level boundary condition is imposed. In case of multiple sediment fractions. this method limits the sediment that can leave a computational cell.5 Calculation of bedload transport at open boundaries At open boundaries the user may either prescribe the bed level development or the bedload transport rates.4.y are calculated.m) (n.

x (n) Depth point Bed-load transport component at velocity point Bed-load transport component at water level point S ( n . in vector component form: (11.uu ( n . m) Sb . it is less stable (less damping). User Manual (n+1) ( n +1. m+1) Sb .52) Sb. m) Sb . (11.7 Adjustment of bedload transport for bed-slope effects Bedload transport is affected by bed level gradients. m) b.y Key Water level point ( n .x . Sb. 11.x = αs Sb.y .x ( n .y = αs Sb.55) Deltares . m) Sb .vv Velocity point ( n .53) (11. although the central approach is more accurate. or.5: Setting of bedload transport components at velocity points ble to switch from upwind to central approach by setting the UpwindBedload keyword in the morphology input file to false. In this case αs is given by αs = 1 + αbs 348 cos tan−1 ∂z ∂s tan (φ) tan (φ) − ∂z ∂s −1 .54) whereas the primary effect of the transverse bed slope is a change in transport towards the downslope direction (this may be accomplished by either a pure rotation of the transport vector or by adding a transverse transport component). m) Sb . This is the default option for the bedload transport of all sediment transport formulae. The longitudinal bed slope results in a change in the sediment transport rate as given by: S b = αs S . y (n-1) Bed load transport components at velocity points are set equal to the component at the upwind water level point y x (m-1) (m) (m+1) Figure 11. (11. 1988) as presented by Van Rijn (1993) for transverse slope.Delft3D-FLOW. Two bed slope directions are distinguished: the slope in the initial direction of the transport (referred to as the longitudinal bed slope) and the slope in the direction perpendicular to that (referred to as the transverse bed slope). 1 no effect of bed slope on bedload transport 2 Bagnold (1966) for longitudinal slope and Ikeda (1982. You may choose one of the following formulations for these effects.4.

|ub | ∂n (11. perpendicular to the main bedload transport vector.x = Sb. The magnitude of this vector is calculated using a formulation based on the work of Ikeda (1982. 1988) as presented by Van Rijn (1993). (11.cr ub additional bedload transport vector.5) critical (threshold) near-bed fluid velocity near-bed fluid velocity vector ∂zb bed slope in the direction normal to the unadjusted bedload trans∂n port vector To evaluate Equation 11.cr ∂zb .n where: = Sb fnorm .x − Sb. This implies that there is no flow in the n direction: i. Sb.y . in the down slope direction magnitude of the unadjusted bedload transport vector (adjusted for longitudinal bed slope only): |Sb | = 2 2 Sb. ALFABS keyword in the morphology input file (default = 1. In this case αs is given by αs = 1 + αbs Deltares ∂z .56 we substitute: ub. |τb | ∂n (11. ∂s (11.y fnorm Sb.58) fnorm = αbn τb.x + Sb. ALFABN (default = 1.61) 349 .56 by setting the reference co-ordinates s and n aligned with and perpendicular to the local characteristic flow direction respectively.cr .cr = |ub | where: τb. The direction of this vector is normal to the unadjusted bedload transport vector.cr ∂zb .x fnorm (11.n = Sb αbn where: ub.y are the components of the required bedload transport vector.60) where Sb.y + Sb. |τb | (11.n |Sb | αbn ub.56) Sb.n = 0: Sb. (1995). user-defined coefficient.Sediment transport and morphology where αbs is a user-defined tuning parameter.y = Sb.0). ub.59) The two components of this vector are then added to the two components of the bedload transport vector as follows: Sb.cr τb resulting in: critical bed shear stress bed shear stress due to current and waves: τb = µc τb.57) τb.e.2.cw + µw τb.52) is modified to Equation 11. calculated at the water level points 3 Koch and Flokstra (1980) as extended by Talmon et al.x and Sb. An additional bedload transport vector is subsequently calculated.w . Van Rijn’s equation (7.

This completes the calculation of the bedload transport field. Longitudinal bed slope This bed slope is calculated as: ∂z(u) Sb.63) where Ash . lift-drag ratio µ and critical shields parameter θcr should be specified by you in the morphology input file as keywords CoulFri.x ∂z(v) Sb.6. 350 Deltares . (11. (11. Bshld.62) in which ϕτ is the original direction of the sediment transport and ϕs is the final direction and where f (θ ) equals: Bsh f (θ) = Ash θi Di H Csh Di Dm Dsh . FlFdRat and ThetaCr. The direction of the bedload is adjusted according to the following formulation: tan(ϕs ) = sin(ϕτ ) + cos(ϕτ ) + 1 ∂zb f (θ) ∂y 1 ∂zb f (θ) ∂x . User Manual where αbs is a user-defined tuning parameter. The transports at the U and V velocity points are then stored for use in the computation of bed level changes. a downward bed slope returns a positive value). ALFABS keyword in the morphology input file (default = 1. ∂s ∂x Sb ∂y Sb ∂zb = 0.0).64) where Coulomb friction parameter cL .66) bed slope in the direction of bedload transport bed slope in the positive x-direction evaluated at the U -point bed slope in the positive y -direction evaluated at the V -point internal angle of friction of bed material (assumed to be 30◦ ) φ Remarks: zb is the depth down to the bed from a reference height (positive down). Note that this formula includes the hiding and exposure factor ξ .9 tan (φ) . Cshld and Dshld.65) (11. (11. ∂s max where: ∂zb ∂s ∂z(u) ∂x ∂z(v) ∂y (11. Csh and Dsh are tuning coefficients specified by you in the morphology input file as keywords Ashld.y ∂zb = + . 4 Parker and Andrews (1985). The same formulae for αs and ϕs hold as in the previous case except for f (θ ) which now equals: f (θ) = cL 1 + µcL θ max 1 10 θ. In all cases the bed slope has been defined as follows. Bsh . The bed slope is calculated at the U and V points as these are the locations at which the bedload transport vector components will finally be applied. ξθcr .Delft3D-FLOW. as described in the Section 11.

Ashida–Michiue (1974) Bedload Bedload + suspended Total transport Total transport Total transport Bedload + suspended Bedload + suspended Bedload + suspended Bedload + suspended Total transport Waves Yes No No No Yes No Yes Yes No Transverse bed slope This bed slope is calculated as: ∂z(u) Sb.5.4.9.Sediment transport and morphology Table 11. Meyer-Peter-Muller (1948) 11.8. Table 11.5.67) This special feature offers a number of standard sediment transport formulations for noncohesive sediment.1 Van Rijn (1993) Van Rijn (1993) distinguishes between sediment transport below the reference height a which is treated as bedload transport and that above the reference height which is treated as suspended-load.5. Soulsby/Van Rijn 11.1: Additional transport relations Formula 11.2.5.5.5. Van Rijn (1984) 11.5. Sediment is entrained in the water column by imposing a reference concentration at the reference height. Van Rijn (1993) 11.7.015ρs D50 Ta a D∗ ( ) ( ) 1. Reference concentration The reference concentration is calculated in accordance with Van Rijn (2000) as: ) ( ) c( a = 0.5 ( ) 0. See Appendix B.3.68) where: ca ( ) mass concentration at reference height a In order to evaluate this expression the following quantities must be calculated: Deltares 351 .x ∂zb =− + . Now. Engelund-Hansen (1967) 11. you can implement your own sediment transport formula in a shared library (<dll> or <so>) and call it from Delft3D-FLOW.5 Transport formulations for non-cohesive sediment (11. Soulsby 11. let us continue with a general description of the sediment transport formulae included by default.y ∂z(v) Sb.5.5.6. General formula 11. Bijker (1971) 11.9. 11.5.3 for this option.1 gives a summary of the available additional formulae: Additionally.3 (11.1.5. ∂n ∂x Sb ∂y Sb 11.

1 8 1.w bed shear stress due to waves: 1 τb. 11.w −0.69) Ta ( ) non-dimensional bed-shear stress: ( ) Ta = (µc τb.025 and limits: (0.w = RW AV E ∆r .cw + µw τb.70) µc ( ) efficiency factor current: ) µ( c ) f( c = fc (11.74) efficiency factor waves: ) µ( w = max 0. Note that the bed shear velocity u∗ is calculated in such a way that Van Rijn’s wave-current interaction factor αcw is not required. User Manual D∗ ( ) non-dimensional particle diameter: ( ) D∗ ( ) D50 = (s( ) − 1)g ν2 1/3 (11.w ) − τcr τcr ( ) ( ) ( ) ( ) (11.5 − Hs h 2 (11.w ≤ 0.1 m) where: 352 (11.157):  fw = exp −6 + 5.71) f ( ) c gain related friction factor: f fc ( ) ( ) c = 0. with ∆r = 0.19   (11.2 ˆδ A ks. τb. the wave related roughness ks.78) Deltares .77) To avoid the need for excessive user input.cw bed shear stress due to current in the presence of waves.73) τb.Delft3D-FLOW.063.75) τb.w = ρw fw Uδ 4 fw 2 (11.76) total wave-related friction factor (≡ Equations 9.cw = ρw u2 ∗ µw ( ) (11.01 m ≤ ks. using the relationship: ks.24 10 log 12h ks −2 (11.205.w is related to the estimated ripple height.72) total current-related friction factor: ( ) fc = 0.24 10 log 12h 3D90 ( ) −2 (11.116 and 11.

79) θcr ( ) threshold parameter θcr is calculated according to the classical Shields curve as modelled by Van Rijn (1993) as a function of the nondimensional grain size D∗ .04D∗ =  0. in this expression the symbol D∗ has been used ( ) where D∗ would be more correct: ( ) ( ) θcr  −1 .  0.7 for details: ka ≤ 10ks user-defined current-related effective roughness height (space varying) wave-related roughness.055.Sediment transport and morphology RWAVE the user-defined wave roughness adjustment factor. rather than the depth-averaged velocity used by Van Rijn.   0.29 .072A ˆδ A ks. see Equation 11.78 thickness of wave boundary mixing layer following Van Rijn (1993): 3δw (and δm ≥ ka ) wave boundary layer thickness: ˆδ δw = 0. √ peak orbital velocity at the bed: 2×RMS orbital velocity at bed.5D50 water depth apparent bed roughness felt by the flow when waves are present. 0. This avoids the need for iteration. We find that this is necessary to prevent unstable oscillations developing. In a current-only situation the velocity in the bottom computational layer is used.78 velocity magnitude taken from a near-bed computational layer. height above bed of the near-bed velocity (uz ) used in the calculation of bottom shear stress due to current estimated ripple height. the velocity is taken from the layer closest to the height of the top of the wave mixing layer δ . if waves are active.24D∗    −0. for clarity.w −0. Recommended to be in the range 1-3. default = 2.64 .80) a ˆδ A D50 ( ) D90 h ka ( ) Van Rijn’s reference height ˆδ = peak orbital excursion at the bed: A median sediment diameter ( ) ks ks. see Section 9.1 . critical bed shear stress: ( ) ) ( ) τcr = (ρ( s − ρw )gD50 θcr ( ) τcr ( ) (11. All sediment calculations are based on hydrodynamic calculations from the previous half time-step.25 . 0. Deltares 353 . calculated from ripple height. We emphasise the following points regarding this implementation: The bottom shear stress due to currents is based on a near-bed velocity taken from the hydrodynamic calculations. taken from the wave module.w uz Uδ zu ∆r δm δw 90 % sediment passing size: D90 = 1. Calculated by Delft3D-FLOW using the wave-current interaction formulation selected. see Equation 11.013D∗    0. ( ) (11. Otherwise.14D∗ −0. Note that. 1 < D∗ ≤ 4 4 < D∗ ≤ 10 10 < D∗ ≤ 20 20 < D∗ ≤ 150 150 < D∗ ˆδ Tp U 2π .

64D50 ( ) ( ) = D50 1 + 0.5 Me ( ) (11.81) where Ta ( ) is given by equation 11.82) where: Sb M Me bedload transport [kg m-1 s-1 ] sediment mobility number due to waves and currents [-] excess sediment mobility number [-] M= Me = 2 vef f (s − 1) gD50 (vef f − vcr )2 (s − 1) gD50 2 + U2 vR on (11. At this time. This means that it is not possible for a user to exactly reproduce results obtained using Van Rijn’s full formulations for waves and currents.70. Van Rijn’s wave-current interaction model is not available in Delft3D-FLOW. assuming a logarithmic velocity profile [m/s] near-bed peak orbital velocity [m/s] in onshore direction (in the direction on wave propagation) based on the significant wave height Uon (and Uof f used below) are the high frequency near-bed orbital velocities due to short waves and are computed using a modification of the method of Isobe and Horikawa (1982).015 TA − 25    D( ) 50 for TA ≤ 1 for 1 < TA ≤ 25 for 25 < ( ) TA ( ) ( ) (11.83) (11. User Manual The apparent roughness felt by the flow (ka ) is dependent on the hydrodynamic wave-current interaction model applied.84) (11.85) vef f = in which: vcr vR Uon critical depth averaged velocity for initiation of motion (based on a parameterisation of the Shields curve) [m/s] magnitude of an equivalent depth-averaged velocity computed from the velocity in the bottom computational layer. This achieved by setting IOPSUS=1 the representative diameter of the suspended sediment will then be set to: ( ) ( ) Ds  ( )    0. 354 Deltares . (2003).7 |Sb | = 0.006ρs ws D50 M 0. Bedload transport rate For simulations including waves the magnitude and direction of the bedload transport on a horizontal bed are calculated using an approximation method developed by Van Rijn et al. Adjustment of the representative diameter of suspended sediment The representative diameter of the suspended sediment Ds generally given by the userdefined sediment diameter SEDDIA (D50 of bed material) multiplied by the user-defined factor FACDSS (see also remarks) can be overruled in case the Van Rijn (1993) transport formula is selected. The method computes the magnitude of the bedload transport as: 0.Delft3D-FLOW.

92) (11.w and Ss.007ρs D50 Me The three separate transport modes are imposed separately.c | where: (11.c | |ub | (11.w = 0 if r < 0.90) (11.91) Sbw.u = Sb.c = 0 if r > 100. The direction of the bedload transport vector is determined by assuming that it is composed of two parts: part due to current (Sb.Sediment transport and morphology This method is a parameterisation of fifth-order Stokes wave theory and third-order cnoidal wave theory which can be used over a wide range of wave conditions and takes into account the non-linear effects that occur as waves propagate in shallow water (Grasmeijer and Van Rijn.v = Sb. and ϕ = angle between current and wave direction for which Van Rijn (2003) suggests a constant value of 90◦ . and part due to waves (Sb.w take on the wave propagation direction.w sin φ Deltares (11.w = fSUSW γUA LT where: (11.87) r= (|Uon | − vcr )3 (|vR | − vcr )3 (11.w ) which acts in the direction of wave propagation. This results in the following transport components: Sbc.93) 355 .89) Ss.u |Sb.w cos φ Sbw.88) Sb.v ub.5 m of the bed (the bulk of the suspended transport affected by high frequency wave oscillations).c = Sb 1 + r2 + 2 |r| cos ϕ (11.2) velocity asymmetry value [m/s] = 4 −U 4 Uon of f 3 3 Uon +Uof f suspended sediment load [kg/m2 ] = 0.01. whereas the two wave related transport components Sb. Also included in the “bedload” transport vector is an estimation of the suspended sediment transport due to wave asymmetry effects. These components are determined as follows: Sb. The direction of the bedload due to currents Sb.w | = r |Sb. Sb.86) |Sb. 1998). This is intended to model the effect of asymmetric wave orbital velocities on the transport of suspended material within about 0.w fSUSW γ UA LT wave-related suspended transport [kg/(ms)] user-defined tuning parameter phase lag coefficient ( = 0.u = Sbc.v = |Sb.c is assumed to be equal to the direction of the current.c | |ub | ub.c ) which acts in the direction of the near-bed current. This wave-related suspended sediment transport is again modelled using an approximation method proposed by Van Rijn (2001): Ss.

2 Engelund-Hansen (1967) The Engelund-Hansen sediment transport relation has frequently been used in rivers and estuaries.96) q ∆ C α magnitude of flow velocity the relative density (ρs − ρw )/ρw ´ friction coefficient Chezy calibration coefficient (O(1)) The transport rate is imposed as bedload transport due to currents Sbc .u = Ss.5.v = Ss. is only used to determine the C value when the Chezy been defined. which represents the roughness height for currents alone in ´ friction in the flow has not [m]. User Manual Ssw. It reads: S = Sb + Ss.047) hiding and exposure factor for the sediment fraction considered and the Shields mobility parameter θ given by θ= q C 2 1 ∆D50 (11.95) where φ is the local angle between the direction of wave propagation and the computational grid. BedW and SusW keywords in the morphology input file.97) α ∆ µ θcr ξ calibration coefficient (O(1)) the relative density (ρs − ρw )/ρw ripple factor or efficiency factor critical mobility parameter (= 0. It reads: S = 8αD50 where: ∆gD50 (µθ − ξθcr )3/2 (11.5. This parameter. A second formula specific input parameter (rk ) is required for the Engelund-Hansen formula. 11. Remarks: The D50 grain size diameter is based on the sediment fraction considered.05αq 5 gC 3 ∆2 D50 (11.w cos φ Ssw.3): calibration coefficient α and roughness height rk .eq = √ where: 0. The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B. this parameter can thus be treated as a dummy parameter. The different transport components can be calibrated independently by using the Bed.98) 356 Deltares .Delft3D-FLOW.94) (11.w sin φ (11. 11. as it includes a critical shear stress for transport. Generally.3 Meyer-Peter-Muller (1948) The Meyer-Peter-Muller sediment transport relation is slightly more advanced than the EngelundHansen formula.9.

5 .102) in which q is the magnitude of the flow velocity. The following parameters have to be specified in the input files of the Transport module (See Appendix B.slope + Ss. Remark: The D50 is based on the sediment fraction considered. The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B.4 General formula The general sediment transport relation has the structure of the Meyer-Peter-Muller formula. The ripple factor µ reads: µ = min C Cg.asymm + Sb.5. The near-bed sediment transport (Sb ) and the suspended sediment transport (Ss ) are given by the formulations in the first sub-section. It reads: S = αD50 ∆gD50 θb (µθ − ξθcr )c (11. In general this formula should not be used. It is possible to include sediment transport in the wave direction due to wave asymmetry and bed slope following the Bailard approach.slope Ss = Ss0 Deltares (11. see Bailard (1981). 1.90 1.5 Bijker (1971) The Bijker formula sediment transport relation is a popular formula which is often used in coastal areas.3): calibration coefficient α. but all coefficients and powers can be adjusted to fit your requirements. Stive (1986). critical mobility parameter θcr . This formula is aimed at experienced users that want to investigate certain parameters settings. It is robust and generally produces sediment transport of the right order of magnitude under the combined action of currents and waves. powers b and c.90 is the Chezy Cg. The transport rate is imposed as bedload transport due to currents Sbc . Separate expressions for the wave asymmetry and bed slope components are included: Sb = Sb0 + Sb.asymm + Ss.Sediment transport and morphology in which q is the magnitude of the flow velocity [m/s].9. ripple factor or efficiency factor µ.100) with D90 specified in [m].3): calibration coefficient α and a dummy value. given by: where Cg.104) 357 .0 (11. the D90 grain size diameters is based on the composition of the local sediment mixture. Bedload and suspended load are treated separately.103) (11.90 = 18 10 log 12h D90 (11.101) where ξ is the hiding and exposure factor for the sediment fraction considered and θ= q C 2 1 ∆D50 (11. 11. 11.9.99) ´ coefficient related to grains. The transport rate is imposed as bedload transport due to currents Sbc .5.

216 rc z∗ −1 h c z∗ 1− r h 1 1−y y z∗ dy (11. respectively. User Manual where Sb0 and Ss0 are the sediment transport in flow direction as computed according to the formulations of Bijker in the first sub-section. The transport vectors are imposed as bedload transport vector due to currents Sbc and suspended load transport magnitude Ss .108) I1 = 0. (hw /h) − Cd Cs − Cd (BS − BD) (11.05) Deep water criterion (default value 0.0h Ss = 1.Delft3D-FLOW.111) Deltares .83Sb I1 ln + I2 rc Sb = bD50 where (11.216 rc z∗ −1 h c z∗ 1− r h 1 ln y rc /h 1−y y z∗ dy (11.5.1 Basic formulation The basic formulation of the sediment transport formula according to Bijker is given by: q√ g (1 − φ) exp (Ar ) C 33. min 1.27∆D50 C 2 1 + 0 .110) where BS BD Cs Cd rc and Coefficient b for shallow water (default value 5) Coefficient b for deep water (default value 2) Shallow water criterion (Hs /h) (default value 0.105) (11.107) b = BD + max 0.109) rc /h I2 = 0.5 b ψU q 2 (11.106) C h q φ and ´ coefficient (as specified in input of Delft3D-FLOW module) Chezy water depth flow velocity magnitude porosity Ar = max (−50.4) Roughness height for currents [m] Ara = µq 2 358 −0.5. Both bedload and suspended load terms are incorporated in the bedload transport for further processing. Ara )) (11. and the asymmetry and bed slope components for bedload and suspended transport are defined in the second sub-section. min (100. 11. from which the equilibrium concentration is derived.

Cs .112) z∗ = 1 + 0.194 a0 0 (11.2 Transport in wave propagation direction (Bailard-approach) If the Bijker formula is selected it is possible to include sediment transport in the wave direction due to wave asymmetry following the Bailard approach.3): BS . 11. Cd .5.41) Von Karm The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B.77 and 11.113) Ub = ωhw 2 sinh (kw h) (11. wave velocity sediment fall velocity [m/s] relative density (ρs − ρw )/ρw ´ an ´ constant (0. w.115) fw = exp −5.117) ψ= where C 0 fw 2g if wave effects are included (T > 0) otherwise (11. dummy argument. ε and T user.5. 11. BD .123 . Ub ωrc (11.116) (≡ Equations 9.9.118) C hw kw T Ub w ∆ κ ´ coefficient (as specified in input of Delft3D-FLOW module) Chezy wave height (Hrms ) wave number wave period computed by the waves model or specified by you as T user.977 + 5. rc .Sediment transport and morphology µ= C 18 10 log(12h/D90 ) w κq g C √ 1 .5 b ψU q 2 (11.205. Deltares 359 . see Bailard (1981) and Stive (1986). For a detailed description of the implementation you are referred to Nipius (1998).114) ω= 2π T (11.5 (11.157): a0 = max 2.

Both extra bedload and suspended load transport vectors are added to the bedload transport as computed in the previous sub-section: Sb = Sb0 + Sb.123) and similar for the η direction.119) where the asymmetry components for respectively the bedload and suspended transport in wave direction are written as: Sb. where: u(t) ρ ρs cf φ ϕ w εb εs near bed velocity signal [m/s] density of water [kg/m3 ] density of the sediment [kg/m3 ] coefficient of the bottom shear stress [-] (constant value of 0.asymm + Ss.ξ (t) = ρcf εb 1 ∂zb |u(t)|3 (ρs − ρ) g (1 − φ) tan ϕ tan ϕ ∂ξ ρcf εs εs ∂zb |u(t)|5 (ρs − ρ) g (1 − φ) w w ∂ξ (11.ξ (t) = (11. see also Roelvink and Stive (1989).asymm (t) = (11. see Equation 11. User Manual Separate expressions for the wave asymmetry and bed slope components are included for both bedload and suspended load. is determined by using the following averaging expressions of the near bed velocity signal (calibration coefficients included): u |u|2 = F acA u ˜ |u ˜|2 + 3F acU u ¯ |u ˜|2 (11.Delft3D-FLOW. The (short wave) averaged sediment transport due to wave asymmetry.asymm (t) = ρcf εb |u(t)|2 u(t) (ρs − ρ) g (1 − φ) tan ϕ ρcf εs |u(t)|3 u(t) (ρs − ρ) g (1 − φ) w (11.126).slope.slope.122) Ss.005) porosity [-] (constant value of 0.10) efficiency factor of suspended transport [-] (constant value of 0. but in implemented expression for suspended bed slope transport the second εs is replaced by a user-defined calibration factor. Equations 11.124) u |u|3 = F acA u ˜ |u ˜|3 + 4F acU u ¯ |u ˜|3 in which: 360 (11.slope + Ss.125) Deltares .120 and 11.63) sediment fall velocity [m/s] efficiency factor of bedload transport [-] (constant value of 0.121) from which the components in ξ and η direction are obtained by multiplying with the cosine and sine of the wave angle θ w and the bed slope components as: Sb.120) Ss.slope (11.121. These transports are determined by generating velocity signals of the orbital velocities near the bed by using the Rienecker and Fenton (1981) method.4) natural angle of repose [-] (constant value of tan ϕ = 0.02.asymm + Sb.

3 T 2. Increased values result in increased gross transports and consequently a reduced morphological response time (and vice versa). 1998) showed that the following coefficient settings yielded the best results for the Dutch coast: FacA = 0.1 for T < 3. A validation study (Nipius.4 FacU = 0.) user-defined calibration coefficient for the wave asymmetry user-defined calibration coefficient for the averaged flow The suspended transport relation due to the bed slope according to Equation 11. 11. in the wave propagation direction). The bedload transport rate is given by:   0. you have to create a separate file named <coef. Separate expressions for bedload and suspended load are given.0 50 ∗ Deltares (11.126) εsl user-defined calibration coefficient EpsSL To activate this transport option. but maintaining a constant ratio.3 T 1.c) sediment transport relation is a transport formula commonly used for fine sediments in situations without waves. increased offshore transports).0 50 ∗ Sb =  0.Sediment transport and morphology u ˜ u ¯ F acA F acU orbital velocity signal averaged flow velocity (due to tide.0 EpsSL = 0.inp> which contains on three separate lines the calibration coefficients: FacA. FacU and EpsSL. etc. The ratio between these parameters determines the balance between onshore and offshore transport and hence the shape and slope of the cross-shore bottom profile. The other parameters are read from the transport input file or are specified as general sediment characteristics.123 is implemented as: Ss.1 ∆gD3 D−0.ξ (t) = where: ρcf εs εsl ∂zb |u(t)|5 (ρs − ρ) g (1 − φ) w w ∂ξ (11. By increasing EpsSL an increased flattening of the bottom profile occurs (i. Note: the user-defined FacU value is currently treated as a dummy value. undertow.e.6 Van Rijn (1984) The Van Rijn (1984a.053 ∆gD3 D−0.0 will always be used. The associated response time of the cross-shore morphology can be influenced by modifying the values of the two parameters. The bed slope transport is in general offshore directed.slope .11 If a relatively straight coast is considered the effect of the parameters is: The wave asymmetry causes onshore directed sediment transport (i. An increased FacA results in an increased onshore transport and hence steepening of the cross-shore bottom profile.5 for T ≥ 3.5. wind.b.127) 361 .e. FacU = 0.

130) µc = (11.2 (11.132) The critical shear stress is written according to Shields: τbcr = ρw ∆gD50 θcr (11.129) fcb = (11.2 f1 (zc ) if zc = 1.134) The suspended transport formulation reads: Ss = fcs qhCa (11.131) ´ coefficient where Cg. The formulas of the shear stresses are 1 τbc = ρw fcb q 2 8 0. User Manual where T is a dimensionless bed shear parameter.133) in which θcr is the Shields parameter which is a function of the dimensionless particle parameter D∗ : D∗ = D50 ∆g ν2 1 3 (11.Delft3D-FLOW.136) f0 (zc ) = (11. h the water depth and fcs is a shape factor of which only an approximate solution exists: fcs = f0 (zc ) if zc = 1.138) Deltares .90 is the grain related Chezy Cg.137) f1 (zc ) = 362 ξc /h 1 − ξc /h ln (ξc /h) (11.24 ( 10 log (12h/ξc ))2 18 10 log (12h/ξc ) Cg.2 − zc ) 1.135) In which Ca is the reference concentration. the term µc τbc is the effective shear stress.2 (ξc /h)zc − (ξc /h)1.90 2 (11. q depth averaged velocity.128) It is normalised with the critical bed shear stress according to Shields (τbcr ). written as: T = µc τbc − τbcr τbcr (11.90 = 18 10 log 12h 3D90 (11.2 (1 − ξc /h)zc (1.

References in the following text refer to this book. 11.3 ξc D∗ (11.3): calibration coefficient α1 .while the computed suspended load transport rate is converted into a reference concentration equal to fcs Ca . p. water depth larger than 200 m or smaller than 1 cm.140) β = min 1.144) Furthermore.65 0. The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B.5 ws u∗ 0. If the wave period Tp is smaller than 10−6 s. The sediment transport is set to zero in case of velocities smaller than 10−6 m/s.5 0. the wave period is limited to values larger than 1 s. The root-mean-square wave height is limited to values smaller than 0.7 Soulsby/Van Rijn The sediment transport relation has been implemented based on the formulations provided in Soulsby (1997). where H is the water depth. 1997.015α1 D50 T 1.8 Ca 0.5.Sediment transport and morphology where ξc is the reference level or roughness height (can be interpreted as the bedload layer thickness) and zc the suspension number: zc = min 20.104): D∗ = Deltares g∆ ν2 1/3 D50 (11.4 H . 1 + 2 (11. Furthermore.145) 363 .9. the wave period Tp is set to 5 s and the rootmean-square wave height is set to 1 cm. D∗ is defined as (Soulsby. ws +φ βκu∗ (11.143) The bedload transport rate is imposed as bedload transport due to currents Sbc .141) φ = 2.142) The reference concentration is written as: Ca = 0. dummy argument.4 (11.5. reference level (bedload layer thickness) or roughness height ξc [m] and settling velocity ws [m/s]. The root-mean-square orbital velocity is computed as: Urms = √ 2 πHrms Tp sinh (kH ) (11.139) u∗ = q fcb 8 ws u∗ 2 (11.

defined by: CD = κ ln (H/z0 ) − 1 2 (11. p.3.018 2 U − Ucr CD rms 2. 1997.3): Ss = Acal Ass ξ where u2 + v 2 (11. User Manual Using the critical bed shear velocity according to Van Rijn (Soulsby.012D50 ξ −0. The direction of the bedload transport is assumed to be equal to the direction of the depth-averaged velocity in a 2D simulation and equal to the direction of the velocity at the reference height a (see Section 11. where z0 equals 6 mm and the Von Karm The bedslope correction factor is not explicitly included in this formula as it is a standard correction factor available in the online morphology module.5 mm 0 .151) a general multiplication factor ξ= U2 + 0. The sediment transport is split into a bedload and suspended load fraction.4.153) ´ an ´ constant κ is set to 0.149) Acal Asb a user-defined calibration factor bedload multiplication factor Asb = 0. 1997.2 (11.9.19D50 90 50 ≤ 0.05H Ass D50 /H ∆gD50 1.3. The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B.183): Sbx = Acal Asb uξ Sby = Acal Asb vξ (11. 364 Deltares .147) (11.5 mm < D50 ≤ 2 mm (11. p.146) larger values of D50 lead to an error and to the halting of the program.4 (11. The method is intended for conditions in which the bed is rippled.148) and the suspended transport magnitude is given by the following formula (this quantity is lateron converted to a reference concentration to feed the advection-diffusion equation for the suspended sediment transport as indicated in Section 11.5D50 log (4H/D90 ) if 0.2 (11.150) suspended load multiplication factor Ass = 0.3): the calibration factor Acal . the ratio of the two characteristic grain sizes D90 /D50 and the z0 roughness height.176): Ucr = 0.3) in a 3D simulation (Soulsby.6 D∗ (∆gD50 )1. 6 10 8.Delft3D-FLOW.1 10 log (4H/D ) if D 0.152) where U is the total depth-averaged velocity and CD is the drag coefficient due to currents.

154) For a flat.77) using r = a∗ /(60π ) where r is the relative roughness used by Soulsby.1a∗ if a∗ ≤ 30π 2 if a∗ > 30π 2 (11.8 Soulsby The sediment transport relation has been implemented based on the formulations provided in Soulsby (1997). References in the following text refer to this book.157) which corresponds to formulae 60a/b of Soulsby (p.116): fw = 0. 1997.205. The interaction of the currents and waves is taken into account using the factor Y in the following formula for mean bed shear stress during a wave cycle under combined waves and currents (Soulsby. The sediment transport is set to zero in case of velocities smaller than 10−6 m/s.5ρfw Uorb (11.94): τm = Y (τw + τc ) Deltares (11.155) The relative roughness is characterised using a∗ : a∗ = Uorb Tp z0 (11. p. p.160) as defined on Soulsby (1997.25. The root-mean-square wave height is limited to values smaller than 0. water depth larger than 200 m or smaller than 1 cm. the wave period Tp is set to 5 s and the rootmean-square wave height is set to 1 cm. eq.3 −0. the wave period is limited to values larger than 1 s. the shear stress due to currents is computed as: τc = ρCD U 2 where (11. The friction factor is used to compute the amplitude of the bed shear-stress due to waves as: 2 τw = 0.00251 exp 14. Furthermore.53–55). where H is the water depth.159) CD = κ 1 + ln (z0 /H ) 2 (11.19 0.48) where χ is a user-defined constant: z0 = D50 χ (11. If the wave period Tp is smaller than 10−6 s.4 H . 1997. The root-mean-square orbital velocity Urms and the orbital velocity Uorb are computed as Urms = √ 2Uorb = √ 2 πHrms Tp sinh (kH ) (11. 11.77 and 11.161) 365 . non-rippled bed of sand the z0 roughness length is related to the grain size as (Soulsby.Sediment transport and morphology 11.158) Furthermore.156) which is subsequently used to determine the friction factor of the rough bed according to Swart (1974) (≡ Equations 9. p.5.

25 p4 0.56 2. 1997.47 -0.162) X= τc τc + τw 10 (11.49 -0.40 -0.19 q3 0.165) Furthermore.25 q4 0.04 -0.45 -0.25 -0.40 0.86 -0.73 0.91)): Using the shear stresses given above.24 -0.70 -1.7 0.60 0.34 0.42 0.29 0.24 -0.65 0.29 0.34 2.0 0.14 1.22 0. 1993a) Model 1 (FR84) 2 (MS90) 3 (HT91) 4 (GM79) 5 (DS88) 6 (BK67) 7 (CJ85) 8 (OY88) b1 0.12 0.28 0.50 0.13 0.32 0. and the coefficients are determined by the model index modind and tables 11.89 1.23 -0.00 -0.89 1.18 -1.60 1.2: Overview of the coefficients used in the various regression models (Soulsby et al.10 -0.12 -0.24 0..27 0.26 Table 11.3: Overview of the coefficients used in the various regression models..00 0.00 -0.164) and p and q are determined using similar equations.31 p3 0.10 0.08 b4 -0.51 0.06 0.29 0.26 b3 -0.65 0. continued (Soulsby et al.50 2.60 -0. D∗ is defined as (Soulsby.7 3.50 The formula for Y is given by: Y = X [1 + bX p (1 − X )q ] where: (11.166) Deltares .30 -0.3 (related to Soulsby (1997.26 0.50 0.06 b2 0.77 -0.21 -0.33 0.00 -0. User Manual Table 11.00 -0.73 0.63 -0. p.07 0.19 1.163) and b is computed using: b = b1 + b2 |cos φ|J + b3 + b4 |cos φ|J log (fw /CD ) (11.03 p1 -0.91 1.Delft3D-FLOW.55 0.19 0.04 -0. Table 9.13 0.02 -0.66 J 3.00 p2 0.05 0.104): D∗ = 366 g∆ ν2 1/3 D50 (11.50 0.21 -0.09 0.07 -0.0 0.28 -0. p.2 and 11.50 2.68 0.10 -0.00 0.38 q2 0.27 -2.13 0. the following two Shields parameters are computed: θm = τw τm and θw = ρg ∆D50 ρg ∆D50 (11.10 0. In this formula φ equals the angle between the wave angle and the current angle.55 0.22 0.27 0.68 -0.35 0. 1993a) Model 1 (FR84) 2 (MS90) 3 (HT91) 4 (GM79) 5 (DS88) 6 (BK67) 7 (CJ85) 8 (OY88) q1 0.05 0.75 -0.27 0.14 -0.14 -0.

5.19θm θw 1. the change in the mass of bed material that has occurred as a result of the sediment sink and source terms and transport gradients is calculated. eq. From these expression are finally the actual bedload transport rates obtained: Sb.9.5 θm + ε θw + ε (11.30 + 0. At each time-step. the model index for the interaction of wave and current forces modind (integer number 1 to 8) and the D50 /z0 ratio χ (about 12).174) where ξ is the hiding and exposure factor for the sediment fraction considered and: θ= q C 2 1 ∆D50 (11. p and q .77.170) Φx2 = 12 (0.173) The transport vector is imposed as bedload transport due to currents. eq.129.19 cos (2φ)) θm sin (2φ) + 1.171) where ε is a small constant (10−4 ) to prevent numerical complications.5 (θm + ε)1.175) in which q is the magnitude of the flow velocity.2D∗ (11. 11. p.3): calibration coefficient Acal . This is one of the distinct advantages over an offline morphological computation as it means that the hydrodynamic flow calculations are always carried out using the correct bathymetry.95 + 0. Φx2 ) Φy = 12 (θw + ε) 2 0.167) The sediment transport rates are computed using the following formulations for normalised transport in current direction and normal direction (Soulsby.106): θcr = 0.168) (11.02D∗ )) 1 + 1. The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B. p. 11. θc .x = Sb.172) (11.Sediment transport and morphology with which a critical Shields parameter is computed (Soulsby. The transport rate is imposed as bedload transport due to currents Sbc . This change in mass is Deltares 367 .6 Morphological updating The elevation of the bed is dynamically updated at each computational time-step. 1997.3): α.055 (1 − exp (−0.169) (11. 1997.y = 3 g ∆D50 (Φx u − Φy v ) U 3 g ∆D50 (Φx v − Φy u) U (11.166/167): Φx1 = 12 (θm − θcr ) Φx = max (Φx1 .5 (11.9.9 Ashida–Michiue (1974) The transport rate is given by a generalised version of the Ashida-Michiue formulation: Sbc = α 3 θm ∆gD50 θc 1−ξ θ p 1− θc ξ θ q (11. m. The following formula specific parameters have to be specified in the input files of the Transport module (See Appendix B.

368 Deltares . Morphological delay Frequently. Remark: The depths stored at the depth points (which are read directly from the bathymetry specified as input) are only updated for writing to the communication file and the result files.Delft3D-FLOW. User Manual then translated into a bed level change based on the dry bed densities of the various sediment fractions. the amount of sediment available in the bed will still be updated. Morphological “switch” You can specify whether or not to update the calculated depths to the bed by setting the MorUpd (or equivalently BedUpd) flag in the morphology input file. a hydrodynamic simulation will take some time to stabilise after transitioning from the initial conditions to the (dynamic) boundary conditions. If you wish to prevent any change in both the bottom sediments and flow depths from the initial condition then this may also be achieved by either setting the morphological delay interval MorStt to a value larger than the simulation period. or an investigation of sediment transport patterns with a constant bathymetry is required. It is likely that during this stabilisation period the patterns of erosion and accretion that take place do not accurately reflect the true morphological development and should be ignored. See below for a description of these two user variables. During the MorStt time interval all other calculations will proceed as normal (sediment will be available for suspension for example) however the effect of the sediment fluxes on the available bottom sediments will not be taken into account. Use the CmpUpd flag to switch off the updating of the bed composition. These are discussed below. Both the bed levels at the cell centres and cell interfaces are updated. This is made possible by use of MorStt whereby you can specify a time interval (in minutes after the start time) after which the morphological bottom updating will begin. A number of additional features have been included in the morphological updating routine in order to increase the flexibility. Remark: The use of MorUpd or BedUpd only affects the updating of the depth values (at ζ and velocity points). or by setting the morphological factor MorFac to 0. It may be useful to turn bottom updating off if only the initial patterns of erosion and deposition are required.

From a hydrodynamical point of view this increase in morphological development rate is allowed if the hydrodynamics is not significantly influenced. the morphological factor should be interpreted as a speed-up factor for morphological development without changing the order of events. You have to edit the <∗. see Appendix B. one should note that by following this approach the order of events is changed. or phrased more correctly the hydrodynamics is simulated at a 10 times faster rate. larger time step than the hydrodynamics. This can be achieved by specifying a non-unity value for the variable MorFac in the morphology input file. months.mor> file manually. In that case the morphological development after one tidal cycle can be assumed to represent the morphological development that would in real life only have occurred after 10 tidal cycles. it means that the morphological development is simulated using a. at each computational time-step.Sediment transport and morphology Morphological time scale factor One of the complications inherent in carrying out morphological projections on the basis of hydrodynamic flows is that morphological developments take place on a time scale several times longer than typical flow changes (for example. This feature is not yet supported by the GUI. For such applications. Remark: The Morphological scale factor can also be time-varying.g. In river applications there is no such periodicity as a tidal cycle. By increasing the morphological factor to for instance 10. tests have shown that the computations remain stable in moderately morphologically active situations even with MorFac-factors in excess of 1 000. the morphological variations during a tidal cycle are often small and the hydrodynamics is not significantly affected by the bed level changes. The implementation of the morphological time scale factor is achieved by simply multiplying the erosion and deposition fluxes from the bed to the flow and vice-versa by the MorFacfactor. The interpretation of the morphological factor differs for coastal and river applications.9. Remarks: Verify that the morphological factor that you use in your simulation is appropriate by varying it (e. This allows accelerated bed-level changes to be incorporated dynamically into the hydrodynamic flow calculations. In this example the number of hydrodynamic time steps required to simulate a certain period is reduced by a factor of 10 compared to a full 1:1 simulation. the morphological changes during one simulated tidal cycle are increased by this factor. For coastal applications with tidal motion. This means Deltares 369 . whereas the morphology of a coastline will usually take weeks. tidal flows change significantly in a period of hours. or years to change significantly). and will remain a matter of judgement. Effectively. We also note that setting MorFac=0 is often a convenient method of preventing both the flow depth and the quantity of sediment available at the bottom from updating. reducing it by a factor of 2) and verify that such changes do not affect the overall simulation results. if an investigation of a steady state solution is required. possible conflicts may arise in combination with limited sediment availability and bed stratigraphy simulations. While the maximum morphological time scale factor that can be included in a morphodynamic model without affecting the accuracy of the model will depend on the particular situation being modelled. for instance 10 times. This leads to a significant reduction in simulation time. However.8. One technique for approaching this problem is to use a “morphological time scale factor” whereby the speed of the changes in the morphology is scaled up to a rate that it begins to have a significant impact on the hydrodynamic flows.

n) [kg/m2 ] computational time-step [s] user-defined morphological acceleration factor.g. MORFAC area of computational cell at location (m. In such cases the entrainment flux that generated a certain suspended sediment concentration will differ from the deposition flux that was caused by the settling of the same suspended sediment. The error may kept to a minimum by appropriately choosing the transition times. changing the morphological factor must be associated with changing all external time-varying forcings.n−1) − Sb. 11. and the resulting change in the bottom sediment mass is added to the change due to the suspended sediment sources and sinks and included in the bed compostion and bed 370 Deltares . However. in case of suspended load the entrainment and deposition may occur at time-steps governed by different morphological factors.6. but the suspended concentrations are not (since that would influence the density effects). At each time step bedload is picked-up from the bed and deposited on the bed: only the transports are increased by the morphological factor used for the time step considered.n) (m. (11.uu (m−1.n−1) . The effect of the morphological factor is different for bed and suspended load.n) (m.n) ∆y (m. For river applications. It is possible to vary the morphological factor during a simulation to speed up relatively quiet periods more than relatively active periods.uu ∆y (m. The erosion and deposition fluxes are increased by the morphological factor.1 Bathymetry updating including bedload transport The change in the quantity of bottom sediments caused by the bedload transport is calculated using the expression: ∆SED = where: (m. For coastal applications only the overall simulation time should be adjusted.n) Sb. n) [m] cell width in the y direction. A change in morphological factor during a period of non-zero suspended sediment concentrations.n) − Sb. river hydrograph) the time-scale of these forcings should be sped up: a 20 day flood peak will be compressed in 2 days.Delft3D-FLOW. n) [kg/(m s)] cell width in the x direction. one should take care that by speeding up the hydrodynamic forcings one does not substantially change the nature of the overall hydrodynamic and morphological development: a quasi-steady flood period should not become a short. will thus lead to a mass-balance error in the order of the suspended sediment volume times the change in morphological factor. held at the U point of cell (m.n) Sb. However.n) Sb.n) ∆y (m−1.uu ∆x(m. However.176) ∆SED ∆t fMORFAC A(m.n) + ∆x(m.n) (m.n) change in quantity of bottom sediment at location (m.n) (m.n) ∆tfMORFAC A(m. n) [m2 ] computed bedload sediment transport vector in u direction. n) [m] This calculation is repeated for all ‘sand’ and ‘bedload’ sediment fractions. Note that the combination of a river-like flood peak and a tidal motion will cause problems when interpreting morphological factor not equal to 1.vv (m. Such changes in the morphological factor will not influence the mass balance of a bed or total load simulation since pickup and deposition are combined into one time step. if more than one is present. dynamic flash flood. in case of suspended load there is a time-delay between the time of erosion and the time of deposition.vv ∆x(m. User Manual that in case of time-varying boundary conditions (e. held at the V point of cell (m. held at the u point of the computational cell at location (m.

The actual factor to be transferred is equal to Thet . even when a steep scour hole would develop right next to the beach. If ThetSD equals zero the standard scheme is used. m) Sb . m−1) Sb . A modification to this method may be activated by specifying a parameter HMaxTH larger than the threshold depth SedThr for computing sediment transport. Deltares 371 . Therefore a scheme has been implemented that allows the (partial) redistribution of an erosion flux from a wet cell to the adjacent dry cells. m) b . h1 is the local water depth. In this case. m) Sb(.e. ThetSD ) (11. i.uu Velocity point Depth point Control volume (Positive) bed-load transport component (n-1) −1. the factor ThetSD is used as upper limit for the fraction of the erosion to be transferred to adjacent dry cells. If ThetSD equals 1 all erosion that would occur in the wet cell is assigned to the adjacent dry cells. which determines the fraction of the erosion to assign (evenly) to the adjacent cells. If erosion of the dry beach is desired. 11. respectively. so HMaxTH should be set less than SedThr.6.6: Morphological control volume and bedload transport components level updating scheme.Sediment transport and morphology (n+1) Key S (n) ( n . this option is not recommended. The purpose of this formulation is to allow erosion of parts that are inactive in terms of transport but still wet.177) Here.2 Erosion of (temporarily) dry points In the case of erosion near a dry beach or bank.nvv y x (m-1) (m) (m+1) Figure 11.vv Water level point ( n .uu ( n . The distribution is governed by a user-defined factor ThetSD. which is computed as: Thet = (h1 − SedThr )/(HMaxTH − SedThr ) ∗ ThetSD where Thet = min(Thet . all erosion occurs at the wet cell. more and less than the threshold depth SedThr for computing sediment transport. The ‘wet’ and ‘dry’ cells in the paragraph above are defined as cells at which the water depth is. while limiting the erosion of the dry beach. the standard scheme will not allow erosion of the adjacent cells.

11. by default grid cells are considered to lie within a polygon if their centre lies within the polygon.g. The dredging and dumping feature allows you to specify dredging and dumping areas as x. before the first timestep. The volume of dredged material is summed over all cells in a dredge area and distributed over the dump areas.g. The <mdf> file should contain a keyword Fildad referring to the file used. 372 Deltares . using the link percentages or the link order (up to the dump capacity).6. for a description of this attribute file see Section A. the sequence of dredging (e.Delft3D-FLOW. Remark: Dredging and dumping may also performed during initialization. It is possible to distribute the dredged material over multiple dumping locations. The <∗. Within each dredging polygon the bed levels are lowered to a user-defined depth.dad> file refers to the file containing the polygons.22. The sediment may be distributed equally or non-uniformly (e.3 Dredging and dumping If the bed levels are updated. you may also include some dredging and dumping activities at the end of each half time step. If the dredging capacity is less than the volume to be dredged. the erosion from the adjacent cells will replenish the eroded cell with different sediment fractions than those that were eroded. Depending on the availability of individual sediment fractions at the central ‘wet’ cell and the surrounding ‘dry’ cells.2.y polygons. User Manual Remark: The overall erosion flux is redistributed to the adjacent cells. For each dump area: the bed level is raised and the bed composition is adjusted based on the volume and characteristics of material to be dumped. In simulations with multiple sediment fractions the sediment composition is tracked.dad> file. top first or uniform) determines which grid cells are dredging at the current point in time. Dredging and dumping is performed at this stage in the following order: For each dredge area: if the bed level exceeds a threshold level (or the water depth drops below a certain level) then the bed level is lowered based on the dredging option and the corresponding volume of sediment is removed. The dredging and dumping activities should be specified in a <∗. Warning: Dredging large amounts of material may harm the stability of the calculation. deepest points first) over the grid cells in the dump area.

The flux terms affected are slightly different for cohesive and non-cohesive sediments.6.Sediment transport and morphology 11. CDryB is constant in time and space for each individual sediment fraction.sed> file will overruled. 1 Deltares 373 . The default bed composition model is the uniformly mixed one. At input you must specify the amount of sediment available at the bed as the total (dry) mass of all sediment fractions in [kg/m2 ]. This may be a constant value for the entire model or. as described below. the sediment flux terms will be reduced. There is no bookkeeping of the order in which sediments are deposited and all sediments are available for erosion. Only sediments in the top-most layer are available for erosion.9. When the model almost runs out of sediment at a particular location. Non-cohesive sediment fractions In the case of non-cohesive sediments all bedload transport rates out of a grid cell are reduced by the upwind ratio of available sediment thickness over Thresh. If you have more detailed information on the bed stratigraphy. the erosive sediment source term is reduced proportionally to the ratio of available sediment thickness over Thresh. If the creation of a new layer would exceed the maximum number of layers specified by you. The initial bed composition is assumed to be uniformly mixed. sediments are pushed towards the bookkeeping layers beneath it.9. A user-defined number of bed composition bookkeeping layers may be included to keep track of sediment deposits. you have to specify dummy values though. layers at the bottom of the stratigraphy stack will be merged.4 Bed composition models and sediment availability The morphology module currently implements two bed composition models: A uniformly mixed bed (one sediment layer). The deposition term is never reduced. The source and sink terms of the advection-diffusion equation are not reduced unless the erosive sediment source term The uniformly mixed bed can be used as input for both bed composition models.9). The top of these sediment deposits will coincide with the initial bed level.1 The thickness of the total sediment layer is calculated from the sediment mass by dividing by the user-defined dry bed density CDryB. A layered bed stratigraphy (multiple sediment layers). See Appendix B. Currently only the default bed composition model is supported by the user interface. When sediments are deposited.2 on how to select the other bed composition model. they are initially added to the top-most layer. you may use the bed stratigraphy model and specify an initial layering of the bed composition by means of the IniComp keyword (see Appendix B.9. if this threshold is exceeded a new layer is created. Cohesive sediment fractions In the case of cohesive sediment. In that case the bed composition given in the <∗. After erosion. Currently. The reduction starts when the available sediment thickness drops below a user-defined threshold Thresh. the top-most layer is replenished from below.2) and associated initial bed composition file (see Appendix B. a space-varying initial sediment file (values to be specified at cell centres). The bookkeeping layers are filled up to a user-defined maximum thickness. After mixing in the top layer. alternatively. Below the bottom of these deposits the model assumes a non-erodible bed (sometimes referred to as a fixed layer).

180) where ∆sed is the thickness of sediment at the bed. e. you must ensure that the user-defined threshold depth for drying and flooding. to prevent sudden bursts of sediment from occurring when computational cells are flooded. is not set too large. see Section 4. User Manual is predicted to be larger than the deposition (sink) term. the bed accretes more than the water depth.1 . rather than using the computational layer closest to the top of the sediment mixing layer. The cell will then be set dry. then deposition is expected. then the sediment source and sink terms and bedload transport are not calculated for ‘sand’ and ‘bedload’ sediment fractions. where fr is a reduction factor determined by: (11.5. The reference velocity in the bottom computational layer is adjusted to the top of the sediment mixing layer using the apparent bed roughness ka before being used to compute the bed shear velocity using the physical bed roughness ks . Sinktotal = Sinktotal ∗ fr .g. 11. This situation is checked for and. Thresh (11. specified in the morphology input file. Threshold depth for sediment calculations introduced If the water depth in a cell is less than SedThr.178) (11.7 Specific implementation aspects Negative water depth check In rare situations (with high morphological acceleration factors) it is possible that. and fr is set to 1. Remark: In areas with very shallow water depths and sediment sources and sinks. in one timestep.Delft3D-FLOW. This restriction has been included in order to prevent numerical problems during the computation of the reference concentration.8. the water surface level for the cell is set equal to the new bed level.0 so that deposition close to the fixed layer is not hindered. in that case both terms are reduced by the ratio of available sediment thickness over Thresh as shown by the following equations: Sourcetotal = Sourcetotal ∗ fr . If the sink term is greater than the source term. 374 Deltares . The likelihood of erosive conditions occurring is assessed by calculating the total sediment source and sink terms using the concentration from the previous time-step to evaluate the implicit sink term. The bed shear is always calculated using the velocities computed in the bottom computational layer. Calculation of bed shear in wave and current situations altered The calculation of the bed shear velocity u∗ has been simplified in situations with waves and currents.179) fr = min ∆sed . If this occurs the water depth will become negative (water surface level is below the bed level). if it occurs.

bathymetrical changes are expected to occur at some point during the simulation period).Sediment transport and morphology Depth at grid cell faces (velocity points) During a morphological simulation the depth stored at the U and V velocity points must be updated to reflect the bed level changes calculated in the water level points. This change significantly improves the smoothness of flooding dry cells. The program still requires the depth at velocity points to be set to MOR for morphological simulations.4. the vertical Forester filter is always de-activated for sediment. A vertical Forester filter applied in a sediment transport computation will not affect the sediments.5. If this condition is not met then the depths at the velocity points do not need to be updated during the course of the simulation. Deltares 375 . 11. Since it smoothes the vertical profile and thus can have a strong influence on the vertical mixing processes. However. It is suggested to accept small negative concentrations and to apply a Forester filter only when the negative concentrations become unacceptably large. Lesser et al.e. using a simple upwind numerical scheme. These are reported in Lesser et al.true. Sections 4. Roelvink (2003). Lesser (2003). this can result in a substantially larger computing time. These negative concentrations can be suppressed by applying a horizontal Forester filter. This used to be performed by setting the new depth for the velocity point by copying the new depth held at the water level point. (2004). we advise you to use the MIN procedure during the calibration of a hydronamic model that will later on be converted into a morphological model. (2000). Ruessink and Roelvink (2000). This anticipated that this restriction is lifted in a coming release. this method has been replaced by setting the depth at U and V points equal to the minimum of the adjacent depths in water level points. Remarks: The setting of depths at velocity points as the minimum of the adjacent water level points only comes into effect if sediment is present and the user-defined flag MORUPD is . As this may introduce instabilities in the flow computation.6. Since the MOR and MIN procedures for computing the depth at cell interfaces are equivalent.8 and 10.8 Validation To test and validate the formulations and implementation of the 3D sediment transport feature many simulations have been executed. Remarks: Small negative sediment concentrations (−1 · 10−3 kg/m3 ) can be found in a computation. (i. especially near drying and flooding and in tidal simulations.

User Manual 376 Deltares .Delft3D-FLOW.

which are (nearly) parallel with density interfaces (isopycnals) in regions with steep bottom slopes. This is important to reduce artificial mixing of scalar properties such as salinity and temperature. the vertical index increases from top to bottom. The σ co-ordinate gives significant errors in the approximation of strictly horizontal density gradients (Leendertse. In a finitedifference model. The σ coordinate system is boundary fitted but will not always have enough resolution around the pycnocline. Therefore. A transport flux along the bed is split into a horizontal and vertical part. In other words. The vertical grid should: resolve the boundary layer near the bottom to allow an accurate evaluation of the bed stress be fine around pycnocline avoid large truncation errors in the approximation of strict horizontal gradients. The Z -model is not boundary-fitted in the vertical. The existence of vertical stratification influences the turbulent exchange of heat. the σ - Deltares 377 . The relative layer thickness ∆σ does not depend on the horizontal co-ordinates x and y . The accuracy of the discretisation of the vertical exchange processes is determined by the vertical grid system. Moreover.12 Fixed layers in Z -model 12. The Cartesian Z co-ordinate system has horizontal co-ordinate lines. The number of grid cells in the vertical varies for each horizontal grid point. see Figure 12. stratified flow occurs in combination with steep topography. 2001). This makes it impossible to locally refine the grid around pycnocline in regions with steep bed topography. the vertical index of the free surface cell is “kmax ” and the vertical index of the bottom layer is “kmin”. The staircase representation of the bottom. estuaries and lakes. referred as Z -model in this manual. Stelling and Van Kester. in 2003 a second vertical grid co-ordinate system based on Cartesian co-ordinates (Z -grid) was introduced in Delft3D-FLOW for 3D simulations of weakly forced stratified water systems.1 leads to inaccuracies in the approximation of the bed stress and the horizontal advection near the bed (Bijvelds. 1990. 3D numerical modelling of the hydrodynamics and water quality in these areas requires accurate treatment of the vertical exchange processes. due to the σ -transformation. The commonly used σ co-ordinate system does not meet all the requirements.1 Background In coastal seas. In the σ -model of Delft3D-FLOW the vertical index of the free surface cell is always “1” and the vertical index of the bottom layer is “kmax ”. Grid spacing in the σ co-ordinate model is constructed by lines of constant σ . which leads to numerical cross-wind diffusion in the transport equation for matter. The difference between the numbering in both grid systems has a historical background. 1994) in areas with steep bottom topography. In the Z co-ordinate system. These indices are dependent on the horizontal index. The vertical layer index in the Z -model decreases from top to bottom. salinity and passive contaminants. The inaccuracies related to the staircase boundary representation of the bed in the Z -model are removed by simple adjustments of the determination of the bed shear stress and the advection near solid vertical walls. The grid co-ordinate lines intersect the density interfaces. the number of control volumes in the vertical direction is constant over the entire computational domain. The bottom (and free surface) is usually not a co-ordinate line and is represented as a staircase (zig-zag boundary). which is independent of the horizontal index.

k > 0. are based on almost the same numerical methods. The vertical index k of the top layer of neighbouring horizontal grid cells may vary. The numerical scheme may become non-convergent in these areas due to hydrostatic inconsistency (Haney. fluxes may be defined at cell faces that do not necessarily have a “wet” neighbouring grid cell. y ) and the first horizontal surface above the bed.1. The variation of the free surface and bed topography is smooth in most areas. The layer thickness here is defined as the distance between two consecutive grid surfaces and is independent of space and time for an intermediate layer. The thickness of the bottom layer is the distance between the bottom z = −d(x. Formally. zk−1 ] (12. User Manual ∆ z k Figure 12. y. even approaching zero as the top cell becomes dry. At tidal flats at low tide. In that case. velocity points on the staggered grid of two adjacent grid cells may be situated at different vertical positions. A computational cell is set “wet” when ever ∆zi. The layer thickness may vary in space and time. which justifies the neglect of the cross terms involved. t). see Figure 12. The grid points that are “wet” are determined every half time step. 378 Deltares . The 3D shallow-water models in Delft3D-FLOW using σ co-ordinates and z co-ordinates respectively. this leads to additional terms in the discretized equations but these terms are not taken into account. The concept of layers used here should not be confused with layers of constant density in stratified flows. The layer thickness of the top and bottom cells can be very small. The layer thickness of the top layer ∆zkmax is defined as the distance between the free surface and the first horizontal surface. high grid resolution in shallow areas (tidal flats) and possibly insufficient grid resolution in deeper parts (holes) of the computational domain. Let z = zk be strict horizontal surfaces.1: Irregular representation of bottom boundary layer in the Z -model transformation gives rise to. 1991). Since the grid spacing near the bed and free surface may vary as a function of space and time. zk ] − max [−d(x. y ). the vertical grid spacing ∆zk is defined by: ∆zk (x. t) = min [ζ (x. y. 1992).Delft3D-FLOW. not always required. where k is an integer indicating the layer index.j. The free surface moves through the vertical grid (Casulli and Cheng. The vertical grid system of a Z co-ordinate model is based on horizontal surfaces with constant z co-ordinate value intersecting the water column. the mapping may even become singular.1) Taking into account variable grid sizes near the bed and allowing the free surface to move through the vertical grid introduces a lot of book keeping and makes the free surface boundary elaborate to treat in the numerical method. In the present model.

The time step is restricted by the Courant-Friedrichs-Lewy condition for horizontal advection. 1984).2: Vertical computational grid Z -model (left) and σ -model (right) The 3D shallow-water equations are discretized on a staggered grid (Arakawa C-grid). implicit time integration is used for the vertical derivatives in the transport equation. The horizontal advective terms are approximated by upwind discretisation along streamlines (Van Eijkeren et al. It uses two partial differential equations to compute the transport of turbulent kinetic energy and energy dissipation rate. u and v . The production term only depends on the vertical gradients of the horizontal velocity. Apart from the free surface. In the Z -model near the free surface. The vertical grid space may vanish due to drying and flooding of shallow areas. which ensures mass conservation. 1993). To circumvent time step restrictions imposed by the small vertical grid size in the σ -model in drying areas. a threshold for the vertical grid spacing is introduced that allows the use of explicit schemes for vertical advection. Horizontal advective terms are computed by the Van Leer II TVD scheme and vertical advection is computed using a first order upwind method in order to obtain positive and monotonic solutions. The presence of stratification is taken into account by the buoyancy flux. Central differences are employed for the discretisation of advection in vertical direction. This is favourable since it allows for the use of accurate positive monotonic schemes in vertical direction that are easy to implement. it was found to be much less diffusive than the standard first order upwind method (Bijvelds. For horizontal derivatives. diffusion is discretized using central differences. A finite volume approach is used for the discretisation of the scalar transport equation. The vertical advection and viscosity term are integrated fully implicitly in order to avoid an excessive small time step imposed by the relatively small vertical grid spaces. The vertical eddy viscosity and eddy diffusivity are computed by the standard k -ε turbulence closure model (Rodi. Both the horizontal components of the velocity vector. the vertical grid spacing is not a function of time.Fixed layers in Z -model Figure 12. The shallow-water equations (SWE) are solved by an ADI-type of factorization for the barotropic pressure (Stelling. Although this method is formally of first order accuracy. Deltares 379 .. explicit approximations are used. are computed once in a full time step ∆t. 1997). In horizontal and vertical direction. The horizontal advection and viscosity terms are integrated explicitly. It is noted that in the Z -model the time step restriction due to an explicit treatment of the vertical advection is not as severe as in the σ -model. 1984).

external forcing like wind and atmospheric pressure.2 12. The vertical grid space may vanish due to drying and flooding of the top layer.4) 0 0  f 0 Ay =  ∂ 0 H ∂y and: ∂  g ∂y . the ADI-method is given by: Step 1: U +1 2 −U 1 2 ∆t 1 + Ax U 2 +1 2 1 + Ay U + B U 2 +1 2 = d.Delft3D-FLOW. the time level proceeds from to + 2 and the simulation time from t = ∆t 1 to t = + 2 ∆t. The time step is restricted by the Courant-Friedrichs-Lewy condition for horizontal advection. (12.5. 1984). In this stage the U -momentum equation.1 Time integration of the 3D shallow water equations ADI time integration method The 3D shallow-water equations are discretized on a staggered grid (Arakawa C-grid). ∂ v ∂y (12. 1 In the first stage.  (12. is solved. In vector form. The time integration of the horizontal viscosity terms is discussed in Section 10. 0 0 λ with λ the linearised bottom friction coefficient. User Manual 12.5)  λ 0 0 B =  0 λ 0 . The horizontal advection and viscosity terms are integrated explicitly.1 and is dependent on the formulation. Both the horizontal components of the velocity vector. d is the right-hand side containing the horizontal advection terms. The shallow-water equations (SWE) are solved by an ADI-type of factorization for the barotropic pressure (Stelling. u and v . (12.2. followed by which is 380 Deltares .2) Step 2: U +1 −U 1 2 ∆t 1 +2 1 + Ax U 2 +1 2 1 + Ay U 2 +1 + BU +1 = d. are computed once in a full time step ∆t. The vertical advection and viscosity term are integrated fully implicitly in order to avoid an excessive small time step imposed by the relatively small vertical grid spaces near the bottom and the free surface.3) with: 0  0 Ax = ∂ H ∂x   −f 0 0  ∂ g ∂x 0 .6) To improve stability the bottom friction is integrated implicitly for each stage. ∂ u ∂x 0  (12.

In the second stage. The V -momentum equation is solved which is implicitly coupled with the continuity equation by the free surface gradient.n 2 ≈ m+ 1 . In the vertical direction a fully implicit time integration method is applied. the time level proceeds from + 1 2 to + 1. which is first-order accurate in time and leads to tridiagonal systems of equations.26 and 9. This will result in bed stress term that is discontinuous. y -space and its value can locally be quite small.7 and 9. To avoid this. For the 3D shallow water equations. For the terms containing U velocity in the continuity equation. ∆zb can vary strongly in the x-.3. In the Z -model this situation is more likely to occur than in the σ -model because the grid distance of the first grid point above the bed. Equation 9.4 Horizontal viscosity terms The horizontal viscosity term is simplified (see Eqs.Fixed layers in Z -model implicitly coupled with the continuity equation. In this expression the grid distance of the first grid point above the bed. by the free surface gradient.n 2 ∆ξ (12. 9. The simplification yields a Laplace operator along grid lines. The vertical coupling of the discretised momentum equations is eliminated by a double sweep algorithm. 9.8) 12.2.8. 9. the horizontal velocity components are coupled in the vertical direction by the vertical advection and viscosity term. ∆zb is used to determine the bed stress. the bed stress term is computed using the velocity at one grid point above the bed (unless the number of active layers equals one): u∗ = ln 1 + ubottom+1 κ ∆zbottom+1 +∆zbottom 2 z0 (12. The u-momentum equation involves only second-order derivatives of the u-velocity. This is caused by local maxima of the turbulent energy level computed by the turbulence closure model that affects the vertical viscosity term and vertical velocity.1.3 Bed stress term The bed stress term is computed using the logarithmic boundary layer relation expressed by Equation 9.3. given by Eqs.27). is removed by linearisation of the fluxes in time.7) 12.58. This linearisation of the continuity equation is also used in the σ -model in combination with the Flooding scheme for advection 1 Deltares 381 . the linearisation leads to1 : ∂ ∂ξ Gηη HU +1 2 ≈ ∂ ∂ξ Gηη H U Gηη H U 1 +2 1 +2 ≈ − Gηη H U +1 2 m− 1 .2 Linearisation of the continuity equation The non-linear terms in the coupled continuity equation and momentum equations. When the distribution of the layer thickness at the bottom show large variation then large error in the water level gradient may be introduced. see Figure 12. 12.

. the shear stress τξη in the u-momentum equation contains derivatives of the v -velocity. which leads to the following additional stability condition: ∆t ≤ 1 2νH 1 1 + 2 ∆x ∆y 2 −1 .23 to 9.5.5 12. In the relations below ∆x and ∆y are horizontal grid sizes. 1997).Delft3D-FLOW. For a curvilinear grid ∆x = Gξξ and ∆y = Gηη . it was found to be much less diffusive than the standard first order upwind method (Bijvelds. velocity scale. Which of the limitations is most restrictive is dependent on the kind of application: length scale. For the complete Reynolds stress tensor.25. 9. with or without density-coupling etc. HLES-model for sub-grid viscosity. 12. The discretisation stencil is dependent on the direction of the flow. 12. see Figure 382 Deltares .9) In case of a curvilinear grid ∆x = Gξξ and ∆y = Gηη . User Manual Table 12.1: Time step limitations shallow water solver Delft3D-FLOW Points per wave period T Accuracy ADI for barotropic mode for complex geometries Explicit advection scheme Stability baroclinic mode Explicit algorithm flooding Stability horizontal viscosity term ∆t ≤ T 40 Cf = 2∆t gH 1 ∆x2 + 1 ∆y 2 √ <4 2 ∆t|u| ∆x <2 ∆ρ gH ρ 4 1 ∆ x2 ∆t ∆t|u| ∆x + 1 ∆y 2 <1 <2 1 ∆ x2 ∆t2νH + 1 ∆y 2 <1 In the momentum equations. The horizontal viscosity term is integrated explicitly.4. (12.1 Overview time step limitations In Table 12.1 an overview of the time step limitations due to stability and accuracy for the time integration limitations are given for the shallow water code Delft3D-FLOW. 1993). given by Eqs. is used in the following cases: partial slip at closed boundaries. Although this method is formally of first order accuracy. no slip at the closed boundaries. the complete Reynolds stress tensor.1 Spatial discretisations of 3D shallow water equations Horizontal advection terms The horizontal advective terms in the Z -model are approximated by upwind discretisations along streamlines (Van Eijkeren et al.

k >0 .2 Vertical advection term The horizontal velocities of adjacent vertical layers are coupled by the vertical advection and the vertical viscosity term.k Near the boundaries.n.k −um−1.k   (√G ) ηη m.n.n. For the space discretisation of the vertical advection term.k ξη um.n.k   ( Gηη )m.n um.n. To prevent instabilities.k > 0 v ¯m.10) v ∂u Gηη ∂η um.n−1.k−1 1 1 2 hm.n.n. (12. (12. This results in tridiagonal systems of equations in the vertical.k m.n. the discretisation stencils for the advection terms may contain grid points on or across the boundary. The discretisation for flow with a positive U and V -component is given by the following two equations.n.k ∆ξ um.n.n−1.n.n.k = zk − zk−1 .k ∆η um−1.5.n−1.n.n.n. In a shallow water model the horizontal length scale is much larger than the vertical length scale.12) where hm.3: discretisation along streamlines.11) m.n. To avoid an artificial boundary layer or instabilities. the discretisations are reduced to smaller stencils. Deltares 383 . a fully implicit time integration is used for the vertical exchange terms.k   √G ( ηη )m.n. u ∂u Gξξ ∂ξ and: m.n. 12.n.n.k−1 + hm. In the vertical direction the eddy viscosity term dominates the advection term. Grid points in difference stencil dependent on flow direction 12.n  ξη v ¯m.k+1 − um.n−1.k > v ¯m.k denotes the thickness of layer with index k defined by hm.3.k um.n = u m.k > v ¯m.k −um−1.n.k    (√G ) ηη m.k  um.n.k ∆η ξη > um.n.k −um. (12.k ξη >0 um.n.n = ξη v ¯  √ m.n.n.n.k −um−1. a second order central difference is used: w ∂u ∂z ξz =w ¯m.k .k > 0 v ¯m.k+1 .k + 2 hm.k ∆ξ ξη > um.Fixed layers in Z -model Figure 12. except in stratified flows where the turbulent exchange is reduced and advection may be dominant. The Z -model can have a very thin layer near the bottom or near the free surface.

k (hm. 12.13) − um.k ) νV |m.3 Viscosity terms The approximation of the viscosity terms are based on central differences. accuracy. To ensure that the total mass is conserved.n.n. The time integration of the Van Leer-2 scheme is explicit and therefore the CFL condition for advection gives a stability condition: Cadv = max 384 u∆t v ∆t .n.k ∂c ∂c + ∂y DH ∆x∆zm. User Manual 12.Delft3D-FLOW. ∆x and ∆y are the grid spaces in the physical space.k − (wc)m.n.n. The vertical grid spacing of the top layer ∆zkmax changes in time as the free surface moves.n.k−1 with λd representing the first order decay process and S the source and sink terms per unit area.n. 1974) is used for the approximation of the horizontal fluxes.k c) 1 + {∂x (uc∆y ∆zm.n.n.n.n.n. (12.k ∂z m.n.n.k 1 2 um. kmin and kmax are the indices of the bottom and top cells respectively.n.5.k−1 = + 1 ∆x∆y ∂x DH ∆y ∆zm.n.14) ∂z m.n.k 1 2 + . ∆x ∆y ≤ 1.k c + S. There is no scalar flux through the free surface and the bottom. (12.k hm.k+1 hm.n.k − um.n. The index of the computational layer k increases in the positive Z -direction.k )} ∂t ∆x∆y + (wc)m.6 Solution method for the transport equation A robust and accurate solver for scalar transport has to satisfy the following demands: mass conservation by consistency with the discrete continuity equation.k νV |m. the transport equation is discretised with a mass conserving Finite Volume approach (flux form). monotony (positive solution).k+1 + hm.15) Deltares . (12.k + hm.n. The range of vertical index varies over the horizontal grid.n.k−1 ) The vertical eddy viscosity is computed at the layer interface. To guarantee monotony so-called Van Leer-2 scheme (Van Leer.k−1 (hm.n.k + ∂x ∂y ∂c ∂c + DV − DV − λd ∆zm. The transport equation formulated in a conservative form in Cartesian co-ordinates is given by: ∂ (∆zm. The vertical viscosity term is discretised as: ∂ ∂z νV ∂u ∂z = m. suitable for both time-dependent and steady state problems and computationally efficient.n.k+1 − um.k ) + ∂y (vc∆x∆zm.

The water column is divided into two parts.4: Aggregation of Control volumes in the vertical due to variation position free surface Explicit integration of the horizontal diffusive fluxes yields an upper limit of: 1 ∆t ≤ 2DH 1 1 + 2 ∆x ∆y 2 −1 .k) Figure 12. see Figure 12.4.kmax) t+∆t/2 zk (m. The vertical coupling of the discretised transport equations is removed by a double sweep algorithm. 1 the free surface at the old time level t and the new time level t + 2 ∆t are in a different computational layer. The remaining “wet” cells are aggregated to one Control Volume and the horizontal fluxes are summed. we introduce an approach based on a fractional step method.18) Therefore in the vertical direction a fully implicit time integration method is applied.Fixed layers in Z -model t (m.17) (12. In this case.kmax) zk-1 (m. The concentrations at the new time level. 2DV ∆z ∆t ≤ . To combine the horizontal and vertical transport. w (12. t + 1 2 ∆t are computed. The 3D part of the water column consists of the cells (finite volumes) which are situated entirely under the free surface layer at both time levels. Deltares 385 .k) (m. If the index of free surface the cell has increased.kmax) (m. Source terms are integrated explicitly. sink terms are integrated fully implicit.16) The scalar concentrations are coupled in the vertical direction by the vertical advection and diffusion term. In order to avoid negative concentrations and instabilities. the concentration at the new time level is assumed to be constant over all the computational layers in the free surface cell. see Figure 12. which is first order in time and leads to tri-diagonal systems of equations. (12. An explicit time integration of the vertical exchange terms near the bottom and free surface would lead to very severe time step limitations: ∆t ≤ (∆z )2 . The vertical transport is computed at the layer interfaces which are situated entirely under the free surface layer both at the old and the new time level.5.

k −2cm.n.k + α (1 + CFLadv −u ) (cm.n.k−1 2 (12.k .n.n.n.n.k+1 2 −wm. (monotone).k −cm+2.n.k ≥ 0.k  cm+1.n.19) with: CFLadv −u = and: ∆t |u | ∆x (12.k−1 = wm.n.k < 0.k m a x (m )) Figure 12.k −cm−1. cmm+1.n.n.n.  1.k  c +1.n.k +cm−1.n.k + cm.k cm+1.k −cm−1. 12.n.1 Van Leer-2 scheme The Van Leer-2 scheme (Van Leer.n.n.k hm.Delft3D-FLOW.n.k + cm.).6.n.n.k cm.k −(wc)m.n. whereas the upwind scheme of Fromm is used in case of a smooth numerical solution.n.21) In y -direction.n.k ) cm  cm. = um.k−1 m.k −cm.5: Horizontal fluxes between neighbouring cells with variation in position free surface 12. In case of a local minimum or maximum the first order upwind scheme is applied.n. The interpolation formula for the horizontal fluxes is given by: Fm.n. a similar discretisation is applied. 386 Deltares . cm+1.k + α (1 − CFLadv −u ) (cm.n. cm.   when um.k ) cmm   +1.k m a x (m -1 ) ) (m . the fluxes are discretised with a central scheme: (wc)m.n. namely a first order upwind scheme and the second order upwind scheme developed by Fromm. (local max.k − cm−1. The time integration of the Van Leer-2 scheme is explicit.k −cm+2.6.n.n. It combines two numerical schemes.k cm. The Courant number for advection should be smaller than 1.k −cm−1.n.k   when um.n. 1974) is used for the approximation of the horizontal transport terms.n.k > 1.2 Vertical advection In the vertical direction. cm+1.n.20) α=   0. User Manual ( m .n.22) .k ≤ 1.n.k or min. (12.n. (12.n.k .k −2cm.n.k ∆y +1.k +cm−1.k − cm−1.

k − 2cm.k is negative.k + p p cp m+1. The filter smooths the solution and reduces the local minima (negative concentrations). will be less negative after one iteration and is effectively removed after several iterations by adding enough (local) diffusion to force the concentration to become positive. The filtering technique in this procedure is the so-called Forester filter (Forester. The superscript p denotes the iteration number. with ε = 10−3 . Maximal 100 iterations are carried out. the filter is applied.6. The scalar concentration then computed has an unphysical maximum or minimum (overshoot or undershoot).n. because there the vertical transport is large.n. The vertical advection leads to a tri-diagonal system in the vertical. cm. The wiggles in the concentration may be introduced in stratified areas near closed boundaries and steep bottom slopes.k+1 . 1979).n.n.n.23) This filter is applied only in grid cells where negative concentration occurs. a positive concentration will remain positive. These wiggles arise in the vicinity of steep gradients of the quantity to be resolved. 1981) in the solution. a discharge of buoyant water. then the iterative filtering process in the x-direction is given by: +1 p cp m. 12. If the flow in the vertical is advection dominated. then a warning is generated.k > max (cm. the central differences in the vertical may give rise to non-physical spurious oscillations. local minimum: cm. The numerical diffusion coefficient of the vertical filter is: Dnum = Deltares ∆z 2 .e. i. If there is still a grid cell with a negative concentration after 100 iterations.n.n. it will not introduce negative concentrations irrespective the steepness of the concentration gradients.3 Forester filter In 3D. If concentration cm. due to vertical stratification in combination with up welling or down welling near a closed boundary or a sill.k−1 ) + ε.Fixed layers in Z -model The time integration in the vertical direction is fully implicit. A negative concentration surrounded by positive concentrations. Positive solutions are not guaranteed. usually the result of ill-represented steep gradients (wiggles). Consequently. Local maxima and minima in temperature or salinity in the vertical direction.n. In case of negative concentrations.k−1 ) + ε.k = cm. so-called “wiggles” (Gresho and Lee.k < min (cm. (12. cm. 2∆t (12. generated by the computational method may give physically unstable density profiles and can also better be removed by a numerical filter then by turbulent vertical mixing. A similar filtering technique as in the horizontal direction is applied for points with a local maximum or minimum in the vertical: local maximum: cm.k + cm−1.n.n.n.k+1 . the central differences in the vertical may give rise to nonphysical spurious oscillations.k 4 . an iterative filter procedure based on local diffusion along Z -lines followed by a vertical filter is started in order to remove the negative values.n. a non-linear approach which removes the computational noise without inflicting significant amplitude losses in sharply peaked solutions.24) 387 .

(1992) and Bijvelds (2001).5.j  ∆zm.k−1 Endif with ε = 10−3 .n.n. the horizontal derivatives of the density can be discretised straightforward on the computational grid:  ρm+1.n.n. ε and the eddy viscosity νV are positioned at the layer interfaces in the centre of the computational cell.k − sm. ∆zk−1 ) sm. dissipation.n.26) The temporal variations in salinity and temperature are slow compared to the variations in the flow and therefore the baroclinic term in the momentum equations is treated explicitly.k = sm.1. For salinity.n.16 and Section 9. For more details we refer to Uittenbogaard et al.n. Remark: The vertical Forester filter does not affect other constituents.n. In the Z -model.n.k + ρ0  2 ∆x kmax j =k+1  ρm+1.k − sm.k > sm. see Section 9.15 and 9.8 Numerical implementation of the turbulence models The turbulence closure models in Delft3D-FLOW are all based on the eddy viscosity concept. see Table 12.n.n. First order upwind differencing for the advection provides positive solutions. Equation 9. The transport equations of turbulent kinetic energy k . the algorithm is given by: If sm.25) 12. and dissipation rate ε.k−1 + ε Then sm. and vertical diffusion are the dominant processes.k − ρm. the turbulent quantities k .n.128 are solved in a non-conservative form. see Eqs. 388 Deltares .k−1 + min (∆zk . When activated it only smooths salinity and temperature. It is followed by the filter in the vertical direction.k g 1 ∆zm.k − min (∆zk . Equation 9. The coupling with the flow is weak and in Delft3D-FLOW.Delft3D-FLOW.n. introducing a stability condition for internal gravity waves (baroclinic mode).j − ρm. can sometimes also better be vertically mixed by a numerical filter then by the turbulence model.n.127. 12. 9.j  ∆x (12.n. (sm.n. ∆zk−1 ) 2∆zk−1 (12. If both the horizontal and vertical filters are switched on. the transport equation is solved independently of the flow for each half time step. then first the filter in the horizontal direction is carried out.3. On the staggered grid. For turbulent boundary flows local production. The eddy viscosity is always based on information of the previous half time step.k−1 ) 2∆zk (sm. This choice makes it possible to discretise the vertical gradients in the production term and buoyancy term accurately and to implement the vertical boundary conditions at the bed and the free surface. User Manual Smooth but unstable vertical density profiles of salinity and temperature in the vertical direction.4.7 Baroclinic pressure term The transport equation is coupled with the momentum equations by the baroclinic pressure term. thereby minimising the additional vertical mixing.k−1 ) = sm.

DP. and MIN.1 Bottom depth at water level points The main difference between the σ -model and the Z -model is the representation of the bottom in the model. The way in which the water level is defined at velocity points.9. In Z -model however.9 Drying and flooding Just as in the σ -model. The crucial issues in a wetting and drying algorithm are: The way in which the bottom depth is defined at a water level point. For the definition and a detailed discussion on this topic we refer you to Chapter 10. see Figure 12.6. Due to different representation of the bottom depth (see following section) the procedure are somewhat simplified. the free surface can move freely through the vertical grid.n is not uniquely defined. Criteria for setting a velocity and/or water level points wet or dry.Fixed layers in Z -model Figure 12. MAX. The top layer can vanish due to flooding and drying in the vertical. The option is specified through the value of the parameter DPSOPT. a technical difference is introduced due to the nature of the grid definition in the vertical. Furthermore. These three items will be discussed below. In σ -model when a point is set to dry (or vice versa) then all the layers are deactivated simultaneously. The drying and flooding procedure in the Z -model is almost identical to the procedure applied in the σ -model. see Figure 12.1.6: Definition bottom depth on FLOW grid 12. In both the σ . due to grid staggering. the bottom depth in a water ζ level point dm. 12. Deltares 389 . in the Z -model shallow parts of estuaries and coastal seas are subject to drying and flooding during the tidal cycle. four options are available: MEAN. In the Z -model it is represented as a staircase around the depth in the water level points.and the Z -model. The manner in which this depth value can be determined from the four surrounding depth points may be influenced by the user. However. The way in which the total water depth is defined.

Um.Delft3D-FLOW. (12.2 Total water depth at velocity points Due to the staggered grid applied in Delft3D-FLOW.12) or for weir like situations.n .  d + ζm. The upwind approach is physically more realistic for velocity points between cells with different bottom depth at low water and falling tide (Figure 12.n − zkmax−1 ∧ zkmax ≥ ζm.31) with Um. Um. 12. based on the evaluation of the value of the total water depth.  η d + max (ζm.n ) .7.n > 0. see Figure 12.n = 0.n < 0.n .n ≤ 0. To activate the additional drying and flooding procedure at a water level point.8. this value for DPUOPT is fixed.6.n (12. works exactly the same as in the σ -model. The total water depth at (U ) velocity point is hence given by: U Hm. The initial water level at a dry cell is determined by the depth at a water level point: ζm. DRYFLP.28) The surface layer thickness The vertical grid size near the free surface depends on the spatial location and on time.n + ζm. the total water depth at velocity points is not uniquely defined.n  η Um. The computation of V the upwind total water depth Hm.n . DRYFLP = NO implies that the drying and flooding check is only to be based on the individual test of the depth values at the cell interfaces.n = −dζ m. (12.n in a V -velocity point is similar. centred around the water level points.n . η = d + ζm+1.n representing the depth averaged velocity both for 2D and 3D. see Figure 12. the value of DRYFLP must be set to YES. (12. For the Z -model the bottom is represented as a staircase (DPUOPT=MIN) of tiles. dm+1. If the total water depth in a water level point is negative: ζ Hm. 390 Deltares .n . So any other value will not be accepted by the program. The flow through height in a U -velocity point is always based on an upwind water level.29) 12. ζ DPUOPT = MIN : d = min dζ m.3 Upwinding of the water level in defining the total water depth The upwind flow cross-section in a U -velocity point is computed using the upwind water level.9. User Manual Drying and flooding switch The drying and flooding switch.n . see Section 10. η (12.27) the horizontal cell is taken out of the computation and the half time step is repeated.n = dζ m. the following equations are used to determine the index of the top cell and the vertical grid size: ∆zkmax = ζm.3. ζm+1.9.30) In contrast to the σ -model. Once the new free surface location has been computed.

In 3D simulations. If the water level rises and the total water depth is larger than the threshold. The drying threshold is given half the value of the wetting threshold (hysteresis) to inhibit changes of state in two consecutive time steps (“flip-flop”).n ζ H U m.n ) z = zk +1 H m.7: The flow-through height is determined by the flow direction.n Figure 12. The computational cell is closed layer is set to zero if the vertical grid space ∆zk for the side normal to that velocity point. At both stages the same drying and flooding algorithm is applied.n ζ z = zk z = zk −1 ζ dpm . both the horizontal geometry (wet versus dry) and the vertical geometry (number of vertical layers) are updated.n U z = zk max( m.n ζ m+1.Fixed layers in Z -model ζ m.9. then the velocity point is set dry. due to oscillations introduced by the algorithm itself. The initial velocity of the layer is the same as the velocity of the top layer of the previous half time level. U The total water depth Hm. 12. This may result into flip-flop behaviour of the computational cell where it is alternately set to dry and wet during the computation.n at a velocity point should at least be positive to guarantee a realistic discharge across a cell face. Deltares 391 . which allows the water level gradient to drive a larger amount of water into the neighbouring cell during the next time step. n H m+1. Taking the maximum of the two surrounding water levels at a dry cell face prevents that a velocity point is artificially kept dry. the velocity point is set wet again. Upwinding the water level in the determination of the total water depth at the velocity points as specified above enhances the discharge. we will only describe the drying and flooding algorithm for the first half time step.4 an Alternating Direction Implicit (ADI) time integration method is used in Delft3D-FLOW. If a new water level is computed. The computed water level is generally higher than the average water level. resulting in a larger flow area. The method above is physically less realistic if the flow has the opposite direction as the water level gradient (wind driven flow). This method consists of two stages (half-time steps). Therefore.4 Drying and flooding criteria As described in Section 10. If the total water level drops below half of a user-specified threshold.n ζ dpm +1. The bottom is represented as a staircase around the depth in water level points. If the vertical grid U larger than a threshold ∆z space ∆zk min the computational layer is taken into account. the velocity of a computational U vanishes.

which will generate a shock wave after flooding.n. is taken into account in the drying and flooding algorithm. U 2 Drying check for velocity points in x-direction (Hm. the four velocity points at ζ the cell sides are set dry. In the second stage of the ADI-method. In the first stage of the ADI-method the drying and flooding algorithm in Delft3D-FLOW consists of the following checks: U U < zmin ) and < 0. In 3D simulations. 2 2 and the flooding check: U Hm. see Section 10.k 1 Drying check for velocity points in x-direction (Hm. The boundary of the wet area can only move one grid cell per time step.5δ ) during iterative solution for new water level. This will produce short oscillations. The total water depth at velocity points is computed by the upwind approach.n at a water level point should at least be positive to guarantee a positive control volume. In Delft3D-FLOW the computational part is protected against “dividing by zero” by assuming that the total water depth is at least 1 centimeter.n . You may define in velocity points so-called weirs or spillways. For a 2D weir the height of the crest. If the bottom has steep gradients across a large area on a tidal flat. Therefore.0).n > δ and ∆zm. Flooding is an explicit process. The disturbances are small if the grid size is small and the bottom has smooth gradients.n ) + HKRUm. The thickness of the water layer is zero. the total water depth Hm. The thickness of the water layer of a dry cell (retention volume) is dependent on the threshold d specified by you. Weirs are hydraulic structures causing energy losses. ζ 3 Drying check (negative volumes) for water level points (Hm.n < 0.Delft3D-FLOW. the magnitude of the disturbances generated by the drying and flooding algorithm will depend on the grid size. These flooding check for velocity points in x-direction (Hm. the vertical grid space ∆zk should be positive. the bottom topography and the time step.9.k checks are based on the water level of the previous half time step.n > δ. ∂t 2 (12. If the time step is too large an unphysical water level gradient at the wet-dry interface is built up. User Manual In 2D.32) In general.n > δ or max(ζm−1. ζm. a large area may be taken out of the flow domain in just one half integration time step. The drying check for a 2D weir point at a U -point is given by: 1 1 U Hm. HKRU. If the total water level becomes negative.34) Deltares 392 . (12.n ) + HKRUm. You can avoid this by smoothing the bottom gradients.5δ and ∆zm.n U U > zmin ). the directions are interchanged. The threshold δ is specified at input.n < 0.n < δ or max(ζm−1.33) (12. the water layer may be very thin and may cause problems in combination with online salt transport or offline water quality simulations. If the computational cell is flooded.n . the threshold value d must fulfil the following condition: ζ δ≥ ∂ζ ∆t .n < δ.n. ζm.

two methods for removal of such inaccuracies are implemented in Delft3D-FLOW.Fixed layers in Z -model Figure 12. The velocity nodes are placed at the centre of the cut face of the Control Volume. We discuss the approximation of the fluxes through the boundary cells and the pressure gradients. the advection terms and the wall shear stresses are not corrected for the cut cells.8 (right) to remove the staircase. Deltares 393 .10 Cut-cell and 45 degrees closed boundaries This feature only applies for the Z -model only. One is derived for the so-called Cut Cell method.10. 12. even in curvilinear co-ordinates. The weir acts as a thin dam for water levels lower than the crest height.1 Introduction A staircase closed boundary is sometimes unavoidable when schematising land-water interface. In 3D the horizontal area is the same for all layers. Only for Cut Cells the curvature term was removed because the grid cells are no longer orthogonal. for general curved closed boundaries which do not coincide with a gridline. to reduce the effect of the zero velocities at the closed boundaries on the bottom stress. In the present implementation. Figure 12. It is necessary to relocate the velocity nodes associated with the cut boundary cells. see Kirkpatrick et al.2 Cut Cells The Cut Cell approach involves truncating the Control Volumes at the boundary surface to create new cells which conform to the shape of the boundary. The pressure (water level) points are left in the original position. In the grid generator corner points are shifted. This reduces the artificial boundary layer along closed “staircase” boundaries. This approach is called a “Cut Cell” method. 12. see Figure 12. To avoid inaccuracies introduced by the staircase closed boundaries in the Z -model. The only difference is that the horizontal area (volume) is recalculated for the truncated cells. On the staggered grid you need averaging to determine the V -velocity in a U -velocity point. For Cut cells we changed the averaging procedure. It allows the mass conservation to be discretised in the same manner as for a standard cell.8 (left). In the averaging procedure only velocity points which are not at closed boundaries are taken into account. The second one is derived for 45 degrees boundaries (1 to 1). The spatial approximation of the advection terms was not adapted. (2003).10.8: left: Cut Cell (definition) and right: defined by shifting (exaggerated) the corner point to boundary. 12. even though this may mean that they are physically outside the boundaries of the associated Control Volume.

groynes. User Manual Figure 12. 12. porous plates. Figure 12. sills.11 Hydraulic structures Also in the Z -model.9. a special approach is implemented for the advection terms.11.10. Vi. weirs. to parameterise the extra loss of energy. It is located at a velocity point and its width is assumed to be zero.j = − ∆y Ui. the mathematical formulations and implementation of the hydraulic structures available in Delft3D-FLOW has been described thoroughly. The velocities are reflected in the boundary line.3 45 degrees closed boundary For a staircase boundary of 45 degrees (1-1). so it has no influence on the 394 Deltares .1 3D Gate A 3D gate is in fact a thin dam with a limited height/depth (and position in the vertical). barriers. In Section 10. sluices.j ∆x (12.Delft3D-FLOW. In this section only the parts that deviate will be discussed. A hydraulic structure generates a loss of energy apart from the loss by bottom friction.9: Flow along staircase boundary. At these points additional force term is added to the momentum equation.35) The advection terms for these 45 degree boundary cells are discretised with an explicit first order upwind scheme using the velocities at the boundaries. 12.10: Reflection of velocities 12. Examples of hydraulic structures in civil engineering are gates. bridges. taking into account the aspect ratio of the grid cells. the so-called hydraulic structures can be defined to model the effect of obstructions in the flow that cannot be resolved on the horizontal grid (sub-grid) or where the flow is locally non-hydrostatic. The term has the form of a friction term with a contraction or discharge coefficient.

The implementation of the 3D gate has been described in Section 10.70 to 10.Fixed layers in Z -model Figure 12. we refer to Appendix B.75. 12. Upstream of the structure the flow is accelerated due to contraction and downstream the flow is decelerated due to expansion. The vertical constriction of the flow may vary in time by the lowering or raising of the gate.1. A 3D gate may be used to model a vertical constriction of the horizontal flow such as near barriers.9.2 Quadratic friction The only hydraulic structure where quadratic friction is applied that is not available within the Z -model is the so-called 2D Weir.11: Example of a 3D Gate (vertical cross-section) Figure 12. In this section. The remaining structure may be defined in the Z -model in a similar manner as the σ -model. The flow at all intermediate layers of the gate is set to zero. see Eqs.12: Computational layer partially blocked at the bottom of the 3D gate water volume in the model area. sluices and Current deflection walls.11. Deltares 395 . The layer near the top and the layer near the bottom of the gate may be partially blocked.3. the figures showing the 3D gate as defined in the Z -model will be shown. For more details on the data input requirements for the different type of 3D gates. 10.1.

Phase errors may generate an artificial re-circulation flow (eddy) near the open boundary. Rigid sheet in the Z -model is treated in a similar manner as in the σ -model.12 Assumptions and restrictions The solution of the discretised equations is just an approximation of the exact solution. The accuracy of the solution depends not only on the numerical scheme. These reflections will be observed as spurious oscillations superimposed on the physical results.11.Delft3D-FLOW. due to meteorological or other effects. see Equation 10. calibration on processed field data obtained from a tidal analysis or Fourier analysis. 12.4 Floating structure Floating structures can also be modelled in the Z -model. wave reflections may occur at the open boundaries. Measurements often contain a lot of undesired noise. avoids this problem. In Delft3D-FLOW weakly-reflective boundaries are available which diminish these effects. see Equation 10. In the numerical model. and the physical processes (turbulence.11. but also on the way in which the bottom topography. 12. User Manual 12. For tidal flow computations. one for each segment end. a similar effect may be observed near the open boundaries if the effect of the Coriolis force on the water level gradient along the open boundary is not taken into account in the boundary conditions. The boundary condition at internal points within this segment is obtained by linearly interpolation between the end points. if the phase variation of the tidal forcing along an open boundary segment is non-linear then the number of open boundary segments should be increased so that the phases at all the segments can be specified. The open boundaries in a numerical flow model are artificial in the sense that they are introduced to limit the computational area that is modelled. Therefore. The assumption is made that. It is treated in a similar manner as in the σ -model. The boundary conditions in Delft3D-FLOW are specified for these segments. wave-current interaction) are modelled. The free surface waves should pass these boundaries completely unhindered. The open boundary can be divided into segments (sections).97. two values per segment are required. by restricting the computational time step. Care must be taken when time-series of measurements are directly prescribed as forcing functions at the open boundaries.3 Linear friction The resistance force that is assumed to be linearly dependent on the flow is applied for the rigid sheet. the geographical area. The time integration method strongly influences the wave propagation when applying a large time step. the free surface waves can be propagated correctly.98. For steadystate simulations. 396 Deltares .

Tech. Rep. H.D. Feuillet. 1961. J. US government Print Office. P. 199.” Journal of Fluid Mechanics 253: 341–358. 1997. M. Coastal and Ocean Division. 1971. 357. current computation..” Journal of Geophysical Research 86 (C11): 10938–10954.D. D. 2005. “An exact theory of nonlinear waves on a Lagrangian-mean flow. A. Hauguel and F. “Another reason why simple discretizations of rotated diffusion operators cause problems in ocean models: comments on “isoneutral diffusion in a z co-ordinate ocean model”. Imberger. “An Energetics Total Load Sediment Transport Model for Plane Sloping Beaches. E. R. Monograph.. 388. ASCE 108: 396–417.References Andrews. M. 633 Deltares 397 . J. A. 1932. Burchard. Campin. 382 —. A. G. thesis. Some considerations about scales for coastal models with moveable bed. Delft University of Technology. 243. 245 Bijvelds. and G. 203. 199. A. E. 1985. L. 1981. 202. G. J. “Currents produced by an air curtain in deep water. McIntyre. 633 Bagnold. Inc. Rep. B.. Bertsch. S.. and A.. The Netherlands.” Journal of the Waterway. Delft University of Technology. and J. D. 295. WL | Delft Hydraulics. 1967.. 297. 1966. M. 273 Barenblatt. 379. Numerical modelling of estuarine flow over steep topography.” Journal of Geophysical Research 97 (C3): 3669–3677. W. “Hysteresis of Turbulent Kinetic Energy in Non-rotational Tidal Flows: A Model Study. The Netherlands. 73. Tech. Random data: analysis and measurement procedures. Hydraulics of open channels. Radach. and M. Piersol. A. H. 227 Beckers. P. 226. F. “Structure of bubble plumes in leinearly stratified medium. T. A. 590. “A mathematical model of turbulent heat and mass transfer in stably stratified shear flow.” Journal of Fluid Mechanics 89 (4): 609–646. 348 Bailard. Delft. Mathieu. 377. J. Dal Passo. J. 321 Bulson. Prostokishen and M.” Journal of Fluid Mechanics 249: 35–37. Mellor. John Wiley & Sons. “Modelling vertical and horizontal diffusivities with the sigma co-ordinate system. Ph. V. 1992. 1982. 203. An approach to the sediment transport problem from general physics. 2001.. 1998. McGraw-Hill. Ph. 288 Bijker. Soc.” American Meteorological Society 28: 1552–1559.. thesis.. Delft University of Technology. “New method for tidal Benque. Ughi. 1-97. G. 323 Bendat. 1992. E. 224 Baumert.. 70. 306. Eng. M. 50.” Dock and Harbour Authority 42: 15–22. 591 Blumberg. Holly. J. 222 Baptist. 359 Bakhmetev. Recirculating steady flows in harbours: comparison of numerical computations to laboratory experiments.” Monthly Weather Review 113 (8): 1379. 321. 238 Asaeda. Deleersnijder and P. M. and G. P. New York. 532 ´ J. P. 223. Cunge. Delft. Port. J. R. 1993. Modelling floodplain biogeomorphology. J.. M. I. 1978. M.

Dalrymple. M. Svendsen. 1997. “A numerical model of the combined wave and current bottom boundary layer. Port. M. The equation of state of water and sea water at low temperatures and pressures. 1. 2009. Deltares.5 turbulence closure applied to marine modelling. G. A. “Computation of the driving forces of wave-induced currents. “Absorbing-generating boundary conditions for shallow water models. WL | Delft Hydraulics. M. and R.” Journal of Geophysical Research 100 (C5): 8523–8540. Delft. R. V. Cheng. W. N. Math. User Manual Burchard. A. “Bed friction and dissipation in a combined current and wave motion. “On the practical advantages of the quasi-equilibrium version of the Mellor and Yamada level 2. 590 Casulli. 245 Courant. M. 1 and 2. Vol. 223 Busnelli. E. New York. 2 of Advanced series on ocean engineering. 1986.” International Journal Numerical Methods In Fluids 15 (6): 629–648. 2001.” Appl. 1985. Ph.” Ocean Engineering 12 (5): 387–423. “On the performance of a mixed-large model based on the k-epsilon turbulence closure. T. and R. 204 398 Deltares . L.04 ed. 227 Dean. World Scientific. A. 590 Casulli. R. Modelling 18: 281–287. “Properties of water. 239 Dongeren. Numerical simulation of free surface flows with steep gradients.. A.. 1994. 225 Delft3D-GISVIEW. J. vol.D. Luyten. King.” Coastal Engineering 11: 539–563.Delft3D-FLOW. Interscience. ASCE 123 (6): 303–313. B.” Journal of the Waterway. 1987. 179 Dingemans. 1992.. 231 Busch. 241 Deigaard. and H. free surface flows.” In Workshop on Micrometeorology. C. J. Gerritsen.” Journal of Geophysical Research 93 (C1): 491–508.” Tellus 46A: 200–221. L. 245 Davies.. 243.. Water wave mechanics for enginieers and scientists. G. C. World Scientific Publishing Company. A. 1999. 378 Christoffersen. The Netherlands. American Meteorological Society. London. “Semi-implicit finite difference methods for threedimensional shallow water flow. Radder and H. “An intercomparison of three-dimensional tidal hydrodynamic models of the Irish Sea. 1997. R. E. A. Not yet known.” American Journal of Science 256: 225–240. Coastal and Ocean Division. Delft3D-GISVIEW User Manual. 239 Dingemans. and P. van and I. 616 Eckart. Advanced Series on Ocean Engineering. de Vriend. 223. 1994. 1988. Methods of mathematical physics. Soulsby and H. pages 1–65. 1972. G. thesis. Vol. V. H. 70. “On the mechanics of atmospheric turbulence.. M.. 223. 13. 1991.” International Journal Numerical Methods In Fluids 30: 425–440. Jonsson. R. 70. W. and I. 242 Deleersnijder. Baumert. 1958. and H. 206 Davies. Water Wave Propagation over Uneven Bottoms. and D. 1995. 243. 1962. Part II. “A semi-implicit finite difference mehod for non-hydrostatic. Hilbert.

H. Moin and W. Madsen. ISBN 3-528-07645-3. J. 260 Golub. M. and F. J. Van Loan. D. and E. Vreugdenhil and B. “Turbulent boundary layer in wave-current interaction. K. Zseleczky and R. vol. 1983. 267 —. Hansen. 557 Engquist. ASCE 108: 932947. Copenhagen. 245 Fredsøe. H. 317 Galappatti. 1993. S. E.. 2nd edition. Van Rijn. Wallingford. T. 561 Germano. and C. S. 1983.References Eijkeren. 1982. rep. Open-Channel Hydraulics. Margolies. A depth integrated model for suspended transport.. 355 Deltares 399 .” In Proc. A. G. 70. The Netherlands. Numerical Methods for Advection-Diffusion Problems. No. 70. Teknisk Forlag. U. Oxfordshire. and O. 1979. B.” Mathematics of Computation 31 (139): 629–651. Koren. 2.. 1980. 1967. H. G. 267. C. 30 of International Geophysics Series. F. Sjoen. 387 Fredsøe. Majda. AIAA 18-th Aerospace Sci. W. Virginia. Linear upwind biased methods. 1992. “Breaker bar formation and migration. vol. 155 Grant. P. Braunschweig.. “Higher Order Monotonic Convective Difference Schemes. 3. 243.. World Scientific Publishing. 253. 1979. L. Atmosphere-Ocean dynamics. pages 55–91. F. 241 French. V.. “Radiation boundary conditions for acoustic and elastic wave calculations. 2006. chap. J. R. Matrix computations. Delft3D-GPP User Manual. 533 Gill. 379. 245 Grasmeijer. J. 1991. 630 Farraday. de Haan. C.” Physics of Fluids A 3 (7): 1760–1765. 1979. and K. A monograph on Sediment Transport in Alluvial Streams.” Journal of Hydraulic Engineering 110: 1103–1120. R.” In C. Hydraulic factors in bridge design. 250. 1984.. 243..” Journal of Computational Physics 23: 1–22. 1982. R. Charlton. McGrawHill..” Journal of Geophysical Research 84 (C1): 1797–1808. Vieweg Verlag. B. “A dynamic subgrid-scale eddy viscosity model. 382 Engelund. England. B. Tech. Deltares. B. “Shape and dimensions of stationary dunes in rivers.14 ed.” Coastal Engineering pages 2750–2758. 299. rep. eds.” Journal of the Hydraulics Division. “Absorbing boundary conditions for the numerical simulation of waves. 45. Deigaard. and L.. 211 —. ISSN 0179-9614. Academic Press. “Notes on Numerical Fluid Mechanics. Tech. 211 Fannelop. “Combined wave and current interaction with a rough bottom. 1987. J. van.” Communications on Pure & Applied Mathematics 32: 313–357. Piomelli.. 1989. 252. USA. Delft. 1998. Mechanics of Coastal Sediment Transport. eds. 1977. Cabot.. Delft University of Technology. and R. Meeting. Stelling and T. “Hydrodynamics of air-bubble plumes. 258. 594 GPP. G. and A. 3 in Advanced series on ocean engineering. van Stijn. H. Baltimore: John Hopkins University Press. New York. Hydraulics Research Station Ltd. 313 Forester.

L. Delft. and A.” In R. 232. “Changes of the mean velocity profiles in the combined wave-current motion in a GLM formulation. G. 349 400 Deltares .” In The 27th IAHR Congress.. Proceedings of Theme A. “Adaptation of secondary flow in nearly horizontal flow. Horikawa. F. “Bed level computations for curved alluvial channels. 1991. J. 234 Kirkpatrick. P. 348. Bettes. The Netherlands. 1996. In Dutch (3D rekenen rondom constructies). ASCE 108 (1): 95–114. pages 357–364. eds. 274 Koch. 245 Ikeda. and R. S. New York. The Netherlands. and K. PR051. P. and C. J. M. Q2487. 321. J. Balkema Rotterdam. “Modelling horizontal diffusion with sigma coordinate system. 291. Lee. The Netherlands. 1988. 1991. “A representation of curved boundaries for the solution of the NavierStokes equations on a staggered three-dimensional Cartesian grid. 273.” Journal of Fluid Mechanics 370: 271–296. L. Flokstra. 275 Klopstra.” Journal of the Hydraulics Division.D. P... Commissioned by Rijkswaterstaat/RIZA. Tech. Barneveld. 2-7 Feb. 70. Van Velzen. thesis.” In Proceedings of the XIXth congress of the International Association for Hydraulic Research. and G. van Schijndel. H. San Francisco.Delft3D-FLOW. they’re telling you something. The Netherlands. vol. 2000. 323 Huang. 378 Hirsch. Van Noortwijk. 1999. Armfield and J. 1982. 273. 299. 387 Groeneweg. Temperville. “Don’t suppress the wiggles. HKV consultants.. M.” Journal of Physical Oceanography 21: 610–619. A.” Journal of Computational Physics 184: 1–36. 238 Groeneweg. A. 1998. Tech. India. 349 Isobe. Delft University of Technology. M. H. R. 1996. 1986. 354 Jagers. J. M. “On the pressure gradient force over steep topography in sigma coordinate models. Rep. R. H. and R. 1981. 1982. 203. 1997.. T. Not yet known. 1981. pages 93–100. New Delhi. Delft. estuaries and the sea. 3D computations around structures. Booij. 349 —. S. Soulsby and R. New York. Rep. Klopman.” Computer and Fluids 9: 223–253. American Society of Civil Engineers (ASCE). 321 Huynh-Thanh. 2. and S. Ph. 238 Haney.. H. Analytisch model hydraulische ruwheid van overstroomde moerasvegetatie.. C. Kent. 243. W. 393 Klopstra. WL | Delft Hydraulics. 1997. M. pages 775–780. S. J. 2003. “Study on water particle velocities of shoaling and breaking waves. Sand transport in rivers. D. 348.” Coastal Engineering in Japan 25: 109–123. John Wiley & Sons. User Manual Gresho. H. H. Van Noortwijk and E. and M.” Journal of Hydraulic Research 24 (1): 19–37. Numerical computation of internal and external flows. W. Managing Water: Coping with Scarcity and Abundance. J.. Barneveld and J. “Incipient Motion of Sand Particles on Side Slopes. D. “Analytical model for hydraulic roughness of submerged vegetation. 1990. “A numerical model of the rough turbulent boundary layer in combined wave and current interaction. L. Spaulding. 534 Kalkwijk. Lelystad. 1980. Wave-current interactions in a generalized Lagrangian mean formulation.” Journal of Hydraulic Engineering 122 (6): 349–352.

2003. Akad. Muller. R-708-NYC. R.. Tech. “Mean flow in round bubble plumes. asce. A. 2004. Eng. 286 Leer. 2.. 375 McDougall. T. and A. G. 1948. Monotonicity and conservation combined in a second order scheme. Tech.” In Proceedings of the ¨ 2nd Congress IAHR. RM-5294-RR. 2000. J. 1974. 260 Leendertse. 630. S. “Development and validation of a three-dimensional morphological model.. “Bubble plumes in stratified environments. 193. 286 —. 1942. A water quality simulation model for well mixed estuaries and coastal seas: Vol. Z2899. Nauk. J. Volume II: Aspects of computation. 384. 225 Meyer-Peter. A. A. Volume III: The interim program. 329. 296. 330 Lane. and E. C. Poisson. “Turbulence modelling of surface water flow and transport: part IVa. 286 Leendertse. “Richardson’s number revisited. Stelling. A. R-2187-OWRT. 1982. and R. Mech. 225 —.” Journal of Fluid Mechanics 133: 345–376. Pasadena. Seria fizicheska Vi No. 632 Millero. 1981. 3rd International Symp.. J. 320. J.. pages 39–64.” In E. E. Alexander and S.. R-1417-OWRT. The heat balance of the North Sea. Yamada. 267 Miles. February 3-5. Roelvink.” Review of Geophysics and Space Physics 20 (4): 851–875.. 1990. 375 Lesser. van. R. wl. “Development of a turbulence closure model for geophysical fluid problems.-K. J. Rand Corporation. 1987. C.D. Roelvink.” Coastal Engineering 51: 883–915. Liu. Computation Procedures. M. 375 Lesser. 633 Mellor. Santa Monica. R-1764-OWRT. 1978. English translation: 1968 Imperial College. B. xviii. ON/6.. Rep. 386 Lesser. 1971. and T. 220.” Journal of Computational Physics 32: 101–136. “Formulas for bed load transport. Volume IV: Turbulent energy computation.” Journal of Atmospheric Science 31: 1791–1896. J. Dept. T. “Equations of turbulent motion in incompressible fluid. 1967. 1983. 223 Milgram. Tech. van Kester and G. F. 222. vol. Not yet known. A.” Journal of Hydraulic Engineering 114 (4): 603–606. 250. On-line sediment transport within Delft3DFLOW. 1973. Rand Corporation. G. Rep. California. J. 377 Leendertse. N. van Kester and J. Santa Monica. Stockholm. J. thesis. Proudman Oceanographic Laboratory. Ph. Volume I: Principles of Computations Report R-1417-OWRR.” Izv. Stratified Flows. “A hierarchy of turbulence closure models for planetary boundary layers.. List and G. 1974. Rept. R1884-OWRT. J. “Towards the ultimate conservative difference scheme II. L. II. SSR. A three-dimensional model for estuaries and coastal seas. J. 205 Deltares 401 . J. J. G. eds.. Rand Corporation. H.References Kolmogorov. Volume I: Principles of Computations.” Deep-Sea Research 28A (6): 625–629.1 2 (1-2): 56–58. J. Rep. CA.” Journal of Fluid Mechanics 85: 655–672. 8. Gritton. 1989. pages 1–7. Aspects of a computational model for long-period water-wave propagation. “International one-atmosphere equation of state of sea water. Jirka.

Bailard toegepast op de Voordelta Grevelingenmonding). Laws of flow in rough pipes. 70.00 ed. 1945. 1957. R. E. 1965.” In Proceedings of the 2nd International Symp. Delft.v. “edimentation and fluidization: Part I. 224 Murakami. 231 ¨ Prandtl. A. F. J.” Journal of the Hydraulics Division. Delft University of Technology. Transverse transport modelling using Bailard applied to Grevelingenmouth delta. VDI Verlag.” Nachrichten von der Akademie der Wissenschaften in Gottingen. Slaattelid. and W. Institution of Chemical Engineers 32: 35–53. Oonisishi and H. “Uber ein neues Formelsystem fur ¨ die ausgebildete Turbulenz. and D. Stelling and J. L. “Three-dimensional water quality and hydrodynamic modelling in Hong Kong. and E. Kunishi. F.” Proceedings of the Royal Society London A97: 345–373. N. 256.” Water Resources Research 21: 1361–1373. N. G. Boon. Rep. and E.” Coastal Engineering 14: 265–293.00 ed. 1999. 2008. “A rational approach to wave-current friction coefficients for rough. Germany. 193 Postma. Deltares. Trans. M. A. “The supply of energy from and to atmospheric eddies. 4. 254.” Coastal Engineering 12: 1–21. 1977. 1933.” I. 1948. Delft3D-RGFGRID User Manual. H. Anderson. 223. Andrews.. 222. 223 402 Deltares . Zaki. “Notes on the theory of the thermocline. pages 43–49. 243. Balkema. F. 81. Deltares.Delft3D-FLOW. Stratification and water quality. and O.. W. NACA TM 1292. Y. 250. 361 O’ Connor. B. 359. M.. Forschungsheft 361. Mathematisch-Physikalische Klasse pages 6–19.. “Erosion and Deposition of Cohesive Soils. “Sorting of bed load sediment by flow in meander bends.” Journal of Marine Research 7 (3): 276–295. 250. smooth and transitional turbulent flow. R.. 1985. 561 Phillips. December 1998. D. 79. K.” Journal of the Oceanographical Society of Japan 41: 221–224. 1988. G. 260 Myrhaug. J.b. Tech. H. 1998. 70. 4. 245 Octavio... in Dutch (Dwarstransportmodellering m. Berlin. The Netherlands. H.Sc. D. 183.. RemoteOLV User Manual. 1954. A. “A numerical simulation of the distribution of water temperature and salinity in the Seto Inland Sea. Forschung auf dem Gebiete des Ingenieurwesens. 1920. Vertical Heat Transport Mechanisms in Lakes and Reservoirs. “A co-ordinate system having some special advantages for numerical forecasting. Hong Kong. 1950. G. K. 243. “Mean bed friction of combined wave-current flow. 608 Richardson. Rep. H. 329 Richardson. 350 Partheniades. ASCE 91 (HY 1): 105–139. 333. S. 168 RGFGRID. thesis. User Manual Munk. 245 Nikuradse. In Ger¨ man (Stromungsgesetze in rauhren rohren) English translation: Laws of flow in rough pipes. Yoo. Rotterdam. 1985. 2010.. 22. 1990. 209 Nipius. Jirka and D. Tech.” Journal of Meteorology 14: 184–185. on Environmental Hydraulics. Harleman. 424. 253. L. 330 RemoteOLV. Massachusetts Institute of Technology. 259 Parker. L. G.

“Sediment transport. Principles of Sediment Transport in Rivers. J.” Journal of Hydraulic Engineering 133 (6): 649– 667. L. 360 Rijn. vi. R. 243. 349. “Dissipation in random wave groups incident on a beach. 82. 12. and J. “Turbulence models and their application in Hydraulics. 237 Rodi. 85. Not yet known. Walstra. Delft.” Journal of Geophysical Research 94 (C4): 4785–4800. W. 2004. 215. 268.” In Corpus Christi. Part II: suspended load transport. Ph. Fenton. Pan and J. 1984c.” Journal of Fluid Mechanics 104: 119–137. 1984. 86. 348. page p. Grasmeijer. 2003. 354. 361 —. VI. I: Initiation of Motion. xv. 269. 225. 361 —. F. USA. R.” Journal of Hydraulic Engineering 110 (10): 1431–1456. “Sediment transport. 273. Part I: bed load transport. B. 613. 620 Rijn. 375 Roelvink. 1997. vol. 1981. D. 555. 217 Deltares 403 . Sierra. “Keeping it simple by using complex models. “Sediment transport. J. 361 —. 267. 338. 268. A. “Sediment transport by currents and waves. M. Delft University of Technology. Spectral Modelling of Wind Waves in Coastal Areas. “A Fourier approximation method for steady water waves. 245 —. estuaries. 379 Roelvink. C. 270. 351. J. Walstra. Not yet known. 1993. and Bed-Load Transport. Part III: bed form and alluvial roughness. The Netherlands. Australia. 1984b. 355 —. 1984a. M. J. The Netherlands. general approximation formulae Coastal Sediments.” Coastal Engineering 47: 295–327. thesis. “Unified View of Sediment Transport by Currents and Waves. Principles of fluid flow and surface waves in rivers. 2003. Bed Roughness. 2001. 339. L. Brisbane. Advances in Hydro-Science and Engineering. 266.” Coastal Engineering 19: 127–150. 272. 354 Ris. and M. 360 Roelvink. 86. Not yet known. Sutherland. “Bar-generating cross-shore flow mechanisms on a beach.. and D. “The predictability of cross-shore bed evolution of sandy beaches at the time scale of storms and seasons using process-based profile models. A. 553 —. 616 —. The Netherlands. State-of-the-art paper article sur l’etat de connaissance. C.” Journal of Hydraulic Engineering 110 (11): 1613–1640.” Journal of Hydraulic Engineering 110 (12): 1733–1754. 556 —. 197. 345. 267.References Rienecker. Not yet known. 86. Communications on Hydraulic and Geotechnical Engineering. 232 —. 353. 70. 86. The Netherlands. 230. 222. van. 1990. D. 344.” IAHR Paper presented by the IAHR-Section on Fundamentals of Division II: Experimental and Mathematical Fluid Dynamics. report 97-4.” In Proceedings of the 6th International Conference on Hydro-Science and Engineering. Aqua Publications. Stive. 337. 2004. 2000. 336. 335. van. A. 351. J.D. 1989. Estuaries and Coastal Seas. 1993. S. 355 —.. J. 2007.. xvi. 2003. seas and oceans. Aqua Publications.

D. L. A. “Practical aspects of accurate tidal computations. R. Fredsøe. 281. 357. Delft. Temperville and T. 199. 377 Stelling. van Kester. 285.” In Proceedings 20th International Coastal Engineering Conference.. Rijkswaterstaat. Harleman and K. R. G.” In M. A. Stolzenbach. 1997. 324 Smith. G. 380. Huntley. 1994. 306. R. 245. T.1–1/2. 297. 262. H. D. 292.” Review of Geophysics and Space Physics 18 (10): 243–268. F. L. Myrhaug.. S. 241. Zitman.” In Abstracts-in-depth of the Marine Science and Technology G8-M overall workshop. Canada. Myrhaug. 223. Uittenbogaard. 291.” Coastal Engineering 21: 41–69. Davies. and J. 294.. Willemse. Thomas. S. J. J. 1993b. G. 202. 1984. G. London. ASCE. 304 Stive. eds. J. Validation of On-line Mud Transport within Delft3DFLOW. R.. G. M. L.. Tech.” Journal of Hydraulic Engineering 112 (9): 802–817. 263. measured in a steady stratified mixing layer. Hamm. P. 1986. R. A. 364. Tampa. G. 363. August 21-25. D. 1997. 379. The Netherlands. J. 1989. 224 Slørdal. 366. “A model for cross-shore sediment transport. pages 771–782. 2003. 323. 2000. rep. S.. A.. Jonnson. 35. 1986. 322. “Possibilities and limitations to simulate turbulence fluxes of mass and momentum. 367 Soulsby.. L. New York. 1974. F. I. 93. 243. Wiersma and J. 296 Stelling.. xv. M. 288. S. Bedford and A. 323. 196. American Society of Civil Engineers. P. “On the approximation of horizontal gradients in sigma co-ordinates for bathymetry with steep bottom slopes. F. R. “A staggered conservative scheme for every Froude number in rapidly varied shallow water flows. Klopman. Banke. xix. pages A55–A62. L. Duinmeijer. 244. Rep. R. 286.1–4. L. T. xvi. J. “Surface Heat Loss From Cooling Ponds.” International Journal Numerical Methods In Fluids 18: 915–955. 213. G. “Bed shear stresses due to combined waves and currents. 210 Soulsby.” International Journal Numerical Methods In Fluids 24: 987–1017. D. Ottawa. M. A. Dynamics of marine sands. E. Simons and G. 193. Simons.. pages 1550–1564. a manual for practical applications. “Wave-current interaction within and outside the bottom boundary layer. B. 375 Ryan. 306 Stelling. “Approximation of Convective Processes by Cyclic AOI methods.. G. 203. Baron and P. Tech. and S. W. 359 404 Deltares . 1975. E. 93. pages 2. “Variation of the sea surface drag coefficient with wind speed. 320.” Water Resources Research 10 (5): 930–938. 246 Stelling. 1980.. Blumberg. S. 264 Simonin. G. 93.” Quarterly Joournal of the Royal Meteorological Society 101: 665–673. D. R. O. S. 257 Schwiderski. D. K. On the construction of computational methods for shallow water flow problems. G. Estuarine and coastal modeling. “On Charting Global Ocean Tides. A. P. K. WL | Delft Hydraulics. 534 Stelling. Roelvink. Leendertse. Grenoble. National Research Council.Delft3D-FLOW. R. 292. 243. Proceedings 2nd Conference on Estuarine and Coastal Modelling. and E. Viollet. and J. 1993a. and J. Thomas Telford. 286. 286. 1992. User Manual Ruessink. Spaulding.” In XXIII IAHR Congress. 365. A..” International Journal Numerical Methods In Fluids 43: 1329–1354. 366 Soulsby. “The pressure gradient force in sigma-co-ordinate ocean models.

Delft Hydraulics Publ. “Laboratory measurements of the direction of sediment transport on transverse alluvial-bed slopes. 48. E. H. K. L. 657 Thatcher. 38.” In In: Shallow flows: proceedings of the international symposium. The importance of internal waves for mixing in a stratified estuarine tidal flow. B. F.” In 25th Meeting Dutch Working group on Numerical Flow Simulations. Delft University of Technology. Background papers and supporting data on the international equation of state 1980.. Slob. 532 UNESCO. G. Sidney. Burlington. Delft. 256. Tech. 223. 211 Deltares 405 . The Netherlands. 225. UNESCO. pages 21–23. “Subgrid-scale model for Quasi-2D turbulence in shallow water. van Kester and G. R. 349 Taylor. R. and D. a critical survey of some literature. Not yet known. Van Vossen. M.. 23: 730–731. WL | Delft Hydraulics. Tenth report of the Joint Panel on Oceanographic Tables and Standards (1981). 1986. Canada. 1995. 207. 1931. M. 1992.. 197. 144.. 2008. Ph. Delft3D-TIDE User Manual. 211 Verboom. 259 Talmon. Stanford. Tech. The Netherlands.D. Tech. Rep. Offshore sediment transport and equilibrium beach profiles.00 ed. Phil. 388. and A. E. “Effect of variation in density on the stability of superposed streams of fluid. Delft.D. “Weakly reflective boundary conditions for shallow water equations. (JPOTS)..C. M. 186. E. UNESCO. G. 221. 1995. van Mierlo. Stelling. The Netherlands. 1981. and B. The Netherlands. Struiksma and M. 365 Sweers. MIT School of Engineering Massachusetts Institute of Technologie. 223 Thatcher. 131. pages 575–582. L. Department of Civil Engineering.. I. 328 —. Rep. 301.” Proceedings Camb. 36. Deltares.” In 7th Turbulent Shear Flows Symp. Soc. K.” Journal of Hydraulic Research 33 (4): 495–517. 3. Also presented at “5th International Conference on Finite Elements in Water Resources. 250. 254.. L. C. August. Harleman.. 205 Verboom. A mathematical model for the prediction of unsteady salinity intrusion in estuaries. 63. G. Rep. 201. 1974. 220 TIDE. “A nomogram to estimate the heat exchange coefficient at the air-water interface as a function of windspeed and temperature. S. Delft. Z81. Model for eddy diffusivity and viscosity related to sub-grid velocity and bed topography. Implementation of three turbulence models in 3D-TRISULA for rectangular grids. 77. June 1984”.” Journal of Hydrology 30: –. and A. Report no. T. 1984. 659 Uittenbogaard. N. 532 Uittenbogaard. 2003. The Netherlands. R. M. thesis. Segal. rep. 247. 1981a. Delft University of Technology. 219. Delft University of Technology. 1998. Baron. Vermont. 1976. thesis..” Advances in water resources 7 (4): 192–197. “Weakly-reflective boundary conditions for two-dimensional water flow problems.References Swart. Delft. Ph. M. 263 Uittenbogaard. “A proposal: extension of the q2e-eps model for stably stratified flows with transport of Internal Wave Energy. 204. WL | Delft Hydraulics. 1989. A. The practical salinity scale 1978 and the international equation of state of seawater 1980. 1972. Tech. R. 1981b. E. Delft. J. Delft. 48. and F. E. Sanford University. A. 223 —. R. 223 Uittenbogaard.

2000. July 16-21. Coastal and Ocean Engineering submitted: –. User Manual Vermaas. Dienst Binnenwateren/RIZA. xvii. 339 Wang. H. Tech. J. Groeneweg. Port. Australia. 579 406 Deltares . A. D. T. B. Winterwerp. 206 —. “3D Calculation of Wave Driven Cross-shore Currents. A. Numerical methods for shallow-water flow. 532. A model to simulate the transport of fluid mud. Verweij. “Calculation of Wave-Driven Currents in a 3D Mean Flow Model. J. B. WL | Delft Hydraulics. WL | Delft Hydraulics. 1999.. J. A. MEAH-197. R. 2008. Delft. 2000. D. Waterway. 535 Vreugdenhil. M. Sydney. 3.” International Journal Numerical Methods Engineering 14: 921–931. J.” In Proceedings 27th International Conference on Coastal Engineering. Springer-Verlag. 1992. 291 Walstra. Water Science and technology library. “On the far-field impact of Water Injection Dredging.Delft3D-FLOW. Representation of extra energy losses in RIVCUR. H. Energylosses due to weirs. Delft. The Netherlands.. 1979. C. The Netherlands. WL | Delft Hydraulics.. J. Tech. R. 2000. van Kester and F. Q910. Deltares. Tech. July 16-21. Horizontal Large Eddy simulations. T. Rep. Z. Roelvink and J. In Dutch(Weergave van extra energieverlies in RIVCUR). J.” J. and J. 1987... 288 Wijbenga. 315 Vossen.03 ed. Z. 1990. B. The Netherlands.. Delft. Tech. Sydney. Rep. 314 Winterwerp. 1989. evaluation of flow computations with Delft3D-FLOW.” In Proceedings 27th International Conference on Coastal Engineering. Australia. 330 Walstra. Delft. In Dutch (Energieverliezen door overlaten: Een gewijzigde berekeningsprocedure voor WAQUA-rivieren versie). pages 1050–1063. and J. 1994. Computational Hydraulics: an Introduction. 2000. Z163. A. The Netherlands. research for Rijkswaterstaat. 237 Weare. 2000. J. Kluwer Academic Publishers. Wang. 577. Q92. WL | Delft Hydraulics. Roelvink. J. “Errors arising from irregular boundaries in ADI solutions of the shallow water equations. C. C. Rep. B. Delft3D-WAVE User Manual. 579 WAVE. 576. Rep. van.

Unit: meter. such as bathymetry.g. Represented in Delft3D-FLOW by a matrix of depth values. Tidal condition in which the tide repeats itself (e. Amplitude and phase of the tidal constituents which represent the tide. Program for the generation and manipulation of grid related data. flux rates.or η -co-ordinate. and closed boundaries representing the landwater interface. Defined in the middle of grid sides. A line defined along a fixed ξ . Given according to: Manning. Measure of the resistance of the flow to the bottom. Physical parameter for Delft3D-FLOW. where the sum of computed fluxes.Glossary of terms area astronomical tidal constituent bathymetry Geometric domain for the models. The measurement of depths of the model area. A second program for the visualisation and animation of results of Delft3D modules. A boundary section is a part of a boundary on which boundary conditions are prescribed. Program for the generation of orthogonal curvilinear grids. With the coupling program. Program for the analysis of observed and simulated tide.0. See also Delft3D-TIDE. each value is defined in the right upper corner of the corresponding grid cell. Program for the visualisation and animation of results of Delft3D modules. to be used in the FLOW Input Processor at simulation start time if no MDF-file has been opened. Boundaries can be divided in open boundaries. Prepared as an attribute file for Delft3D-FLOW or entered in the MDF-file as a uniform value for the whole grid. Delft3D-FLOW itself can now also perform these actions (See B. Prepared as an input file for Delft3D-FLOW (containing non-uniform values in U and V direction) or entered in the MDF-file as a uniform value. 407 bottom roughness coefficient boundary section boundary conditions calibration coupling program cross-section current cyclic tide default Delft3D-GPP Delft3D-QUICKIN Delft3D-QUICKPLOT Delft3D-RGFGRID Delft3D-TRIANA Deltares . Chezy Boundaries are separation lines between the model area and the outside world. In case of 2D computation: Speed and direction of the hydrodynamic depth-averaged flow. White Colebrook. Tuning of model parameters such that the simulation results match an observed data set within a prescribed accuracy interval. cyclic semi-diurnal tide which repeats itself after it’s period of approximately 24 hours and 50 minutes). located in open water. fluxes of matter (if exist) and transport rates of matter (if exists) are stored sequentially in time at a prescribed interval. Default: 0. Program which performs some operations on Delft3D-FLOW output files in order to create input files for D-Water Quality or Delft3DPART. Boundary conditions describe the influence of the outside world on the inside of the model area. Initial value for a parameter. aggregation in time and/or space is possible.18). In case of 3D computation: Speed and direction of the hydrodynamic flow in a layer. ´ [m1/2 /s]. or initial conditions. used during calibration. Presentation: table or single value.

Structured set of virtual points covering the model area in the vertical direction on which the simulation results are obtained. In Delft3DFLOW two types of vertical grids can be distinguished: a σ -grid and a Z -grid. Volume of fluid passing a cross-section per unit of time. i. East = 90◦ . Represented by a collection of cells in the grid (dry points) which are either temporarily or permanently dry. Amplitudes and phases that constitute a time dependent signal. Location where water and possibly constituents dissolved in the water are released into or subtracted from the model area. A process in which points or sub-areas of the model area are becoming dry or wet depending on the local water depth. Speed and direction of a water particle. Part of the model area that is not flooded. Generally these components are obtained as a result of Fourier analysis. In Delft3D-FLOW two types of horizontal grid co-ordinate systems can be applied: a Cartesian or a spherical co-ordinate system. including the effect of density differences due to a non-uniform heat and salt concentration distribution. North = 0◦ . Direction in which the flow of a water particle is moving. Presentation of the bathymetry by iso-lines of depth values.e.Delft3D-FLOW. South = 180◦ and West = 270◦ . Fourier components are often prescribed as model α β grid (horizontal) grid (vertical) harmonic components 408 Deltares . Range of values a parameter can have for meaningful results. the velocity components are not averaged over the velocity points on both sides of the water level point before being used. The amount of water and possibly constituents dissolved in the water released into or subtracted from the model area per unit of time. Floating objects moving with the flow. In the staggered grid the speed is computed at the water level point using only the velocity components with the same grid co-ordinates as the water level point. A domain is represented by its lower and upper limit. averaged over the depth. Previously known as TRISULA. The convention is: depth averaged speed depth contours discharge discharge rate domain drogues dry area/dry point drying and flooding flow rate flow velocity flow direction flow direction β = 90◦ − α with: mathematical angle angle according Nautical convention Structured set of virtual points covering the model area in the horizontal direction on which the simulation results are obtained. Hydrodynamic speed. User Manual Delft3D-FLOW Simulation program for tidal and wind driven flow. In both systems the grid is curvilinear and orthogonal. such as a tide.

that may be in the order of magnitude of (or greater than) the horizontal motion. Also called observation point. File that contains the results of a simulation in monitoring stations as a function of time. Computed with the module 3DMOR feature of FLOW. When the vertical motion of fluid is small compared to the motion in the horizontal direction we may still apply this principle in the computation. The forcing of the outer world on the model area is described by boundary conditions and external forces such as wind. The vertical motion. Longitudinal and latitudinal co-ordinates in degrees. Monitoring point for current. File that contains a sub-set of the results in ASCII format that can be listed on a printer. Verification of input and output data on both its domain and its internal consistency. File that contains the results of a simulation in all grid points at specific instances of time. the water level and/or the concentration of constituents are monitored as a function of time.Glossary of terms forcing when a cyclic tidal movement is required. water level and/or temperature and salinity. The model area is connected to the outer world through closed and open boundaries. Change of bathymetry due to sedimentation and erosion. Set of data which determines the input for Delft3D-FLOW. Tide with a small tidal range. Delft3DFLOW writes the results of the simulation in this point to a history file. Observation points are defined at the centre of grid cells. Remark: the velocity in a layer is in a σ -layer and is not in a horizontal plane. Recommended term is bottom high water hydrostatic pressure history file horizontal velocity hydrodynamic conditions integrity check logarithmic speed/velocity profile lon/lat map file model area monitoring station morphological dynamics neap tide non-hydrostatic pressure observation station print file restart restart file river outflow/run-off roughness Deltares 409 . may be induced by buoyancy or by obstacles or a hydraulic jump. In case of 3D computation: Speed and direction of the hydrodynamic flow in a layer. where computational results. Pressure exerted by the fluid due to its (relatively large) vertical motion. In case of 2D computation: Speed and direction of the hydrodynamic depth-averaged flow. Time of the tide with maximum water level. to be used as initial conditions in a restart run. minutes and seconds. Equivalent to ‘bottom roughness’. Start of a simulation using the results of a previous run as initial conditions. Virtual point in the model area. An expression for the velocity distribution over the depth. File with the simulation results at the last time step of a previous simulation. A part of the physical space (the world) that is (schematically) represented in the simulation. Pressure exerted by a fluid due to its weight. The velocity is assumed to be a logarithmic function of the water depth and depends on the depth averaged speed. such as the current. Amount of water flowing from the river into the model area.

all time variations are absent. The arrow represents the speed and direction of the flow. Time period between simulation start and stop time. Elevation of the free water surface above some reference level. Set of conditions which determine the hydrodynamic simulation completely. transient time or warming-up time. velocity components and depth points are defined at different locations. Constituent of (sea) water. positive angle measured clock wise. Simulation run to check the hydrodynamic behaviour of the model as described by the scenario. Flow data in all grid cells expressed as arrows at a certain instance in time. Measure of the resistance of the fluid to the flow. A scenario is defined (stored) in an MDF-file and its attribute files. Program for the analysis of observed or simulated water levels or flows in terms of astronomical tidal components. Also known as initial period. Evaluation of the quality of simulated results by comparison with observed data. including writing the results to file. Salinity causes a density induced flow additional to the hydrodynamic flow. Tide with a large tidal range. Amplitude and phase of the tidal constituents which represent the tide. Sequence of numbers giving the value of one or more parameters used in the model input. User Manual roughness. Magnitude and direction of wind. Grid in which the water level. salinity scenario simulation time spin-up time spring tide staggered grid steady state temperature test run thin dam threshold tidal constants tidal cycle TIDE time frame time history time-series time step vector field verification viscosity water level wind velocity 410 Deltares . Start and stop time of the simulation and the forcing of all (sub)processes in the simulation. A virtual dam along the side of a grid cell across which no flow exchange is possible. In Delft3DFLOW the temperature is treated as a constituent. Temperature causes a density induced flow additional to the hydrodynamic flow. Thin dams are defined in the middle of the grid sides. Thermodynamic condition of the material concerned. expressed as real time or in the number of time steps.Delft3D-FLOW. Defined as a uniform value in the MDF-file or as non-uniform values in every grid point in an attribute file. about 12 hours and 25 minutes for a semi-diurnal tide and about 24 hours and 50 minutes for a diurnal tide. Time required by the model to adjust itself to match the prescribed boundary and initial conditions. Time interval at which the results of the simulation are computed. Water depth above which a dry grid cell is becoming wet. Time period of the dominant tidal component. or of the simulation results in an observation point at sequential moments. Equilibrium situation. usually defined relative to true North. Also called validation. See also Delft3D-TIDE. in order to calculate a geometric series to represent the tide. Sequence of numbers giving the value of one or more parameters in an observation point at sequential moments.

the socalled σ -plane. A 3D model that has been defined with a vertical grid that is strictly horizontal. Deltares 411 .e.Glossary of terms Z -grid Z -model σ -grid σ -model Vertical grid in a 3D model that is strictly horizontal. Index k = 1 in the Z-grid refers to the bottom layer and index k = kmax to the surface layer. i. A 3D model that has been defined with a vertical grid using σ -planes. Index k = 1 in the σ -grid refers to the surface layer and index k = kmax to the bottom layer. Vertical grid in a 3D model that follows the depth profile.

User Manual 412 Deltares .Delft3D-FLOW.

the hydrodynamic simulation program and the FLOW-GUI do not necessarily have to reside on the same hardware platform. The MDF-file has the following general characteristics: Each line contains a maximum of 300 characters. It is.1. Each (set of) input parameter(s) is preceded by a Keyword. followed by an equal sign “=”. it can be transported to an arbitrary hardware platform. you need not to bother about the internal layout or content of the MDF-file.1 MDF-file Introduction File contents Filetype File format Filename Generated The Master Definition FLOW file (MDF-file) is the input file for the hydrodynamic simulation program.mdf> FLOW-GUI or manually offline The Master Definition FLOW file (MDF-file) is the input file for the hydrodynamic simulation program. Generally. A. The user-definable attribute files are listed and described in Section A. This is especially useful when parameters contain a large number of data (e. Consequently. the central column contains the dimension (if useful) and the right column contains a short description. A Keyword is at most 6 characters long (a combination of numerical and alpha-numerical characters. In the MDF-file you can define attribute files in which relevant data (for some parameters) are stored.A Files of Delft3D-FLOW A. sometimes useful to be able to inspect the file and/or make small changes manually. it contains all the necessary information about the model concerned. The left column contains the keyword and its value(s). The MDF-file is an intermediate file between the FLOW-GUI and the hydrodynamic simulation program. time-dependent or space varying data). however. but starting with an alpha-numeric character).1.2 Example In this section an example MDF-file is listed and described.g. but the scenario applies to a 3D computation. The MDF-file is self contained.1 A. The same area is used as in Chapter 5. It contains all the necessary data required for defining a model and running the simulation program. Therefore the MDF-file is an ordinary ASCII-file which you can inspect and change with your favourite ASCII-editor. As it is an ASCII-file. ASCII Free formatted <name.2.e. i. It can therefore be used as model archive by printing the file. Record description: Deltares 413 .

C*256 C*2 1R 1R C*256 C*2 3I 1R 1R 1R C*256 C*2 File with depth values <∗.bcc# FmtbcC= #FR# C*256 C*2 C*256 C*2 File with indices of dry points <∗.wnd# Fmtwnd= #FR# Wndint= #Y# Commnt= Zeta0 = 1.12# Commnt= Runtxt= #Demonstration model Delft3D # #Friesian Tidal Inlet model # #3 layers # Filcco= #fti 02.5000000e+001 Grdang= 0.dry> Format of dry points file File with indices of thin dams <∗.bch> Format of harmonic conditions file File with transport boundary conditions <∗.5000000e+003 Dt = 5.enc> Format of grid enclosure file Number of grid points 3D simulation Layer thickness from top to bottom in percentage of total water depth Ident = #Delft3D-FLOW .90000 U0 = [.grd> Format of grid file Latitude of model centre Angle between true North and y-axis Grid enclosure file <∗.0000000e+000 Filgrd= #fti 02.bcc> Format of transport conditions file continued on next page 414 Deltares .dry# Fmtdry= #FR# Filtd = #tut fti 1.bnd# Fmtbnd= #FR# FilbcH= #tut fti 1.39.0000 35.] S0 = 3.Delft3D-FLOW.enc# Fmtgrd= #FR# MNKmax= 15 22 3 Thick = 50.0000000e+001 C01 = 1. User Manual Keyword and value Format C*28 Description Identification string FLOW-GUI Comment line (not used) C*20 Up to ten lines of free text to clarify the purpose of the simulation and to indicate specific parameter selections.0000000e+000 Commnt= Filbnd= #tut fti 1. one value for each layer Initial concentration constituent 1 C*256 C*2 C*256 C*2 C*256 C*2 File with boundary locations <∗.wnd> Format of wind data file Wind data interpolation flag 1R 1R 1R 1R 1R Initial condition water level Initial condition x-velocity Initial condition y-velocity Initial condition salinity.0000 Fildep= #fti ch02.000000 Tstop = 1.bch# FmtbcH= #FR# FilbcC= #tut fti 1.03.0000 15.] V0 = [.dep# Fmtdep= #FR# Commnt= Fildry= #tut fti 1.dep> Format of depth file Name of the grid file <∗.00000 Tzone = 0 Commnt= Sub1 = #S W # Sub2 = #PC # Namc1 = #Conservative Spill # Wnsvwp= #N# Filwnd= #tut fti 1.grd# Fmtcco= #FR# Anglat= 5.02 3.thd# Fmttd = #FR# Commnt= Itdate= #1990-08-05# Tunit = #M# Tstart= 0.bnd> Format of boundary file File with harmonic flow boundary conditions <∗.thd> Format of thin dam file C*10 C*1 1R 1R 1R 1R Reference date for all time functions Time unit of time dependent data Start time after Itdate in Tunits Stop time after Itdate in Tunits Time step in Tunits Local time zone C*4 C*3 C*20 C*1 C*256 C*2 C*1 Flags to activate the processes in two groups of four and three characters Name of first constituents Flag for space varying wind and pressure File with wind data <∗.

src# xR 1R 1R Gravitational acceleration Density of water at background temperature and salinity Reflection coefficient 1R 1R C*4 6R 1R 1R C*1 C*12 1I 1R 1R 1I C*1 Background water temperature Background salinity Bottom stress formulation due to wave action Wind stress and wind speed coefficients Air density Parameter spiral motion Flag for computation spiral motion Type of turbulence closure model Selection flag for heat model Percentage sky cloudiness Surface area in heat model Vapour pressure user specified.000000 0.0000 Vicoww= 1.0500000 Dco = -999.000 0.00000e-006 Dicoww= 1.Files of Delft3D-FLOW continued from previous page Keyword and value Format xR Description Thatcher-Harleman return time at surface (for x open boundaries) Thatcher-Harleman return time at bed level (for x open boundaries) Rettis= 1.999 Tlfsmo= 0. Ozmidov length scale Flag for HLES sub-grid model Uniform horizontal eddy viscosity Uniform horizontal eddy diffusivity Uniform vertical eddy viscosity Uniform vertical eddy diffusivity Flag to activate partial slip conditions Roughness length vertical side walls 1I C*3 C*3 C*4 1R 1R 1R 1R C*1 C*1 C*1 C*13 C*6 Number of iterations in cont.1000000e+001 Rouwav= # # Wstres= 0.0000000e+002 Commnt= Ag = 9.000000 Rhoa = 1.000000 Sarea = 0.eq.500000 Equili= #N# Tkemod= #Algebraic # Ktemp = 0 Fclou = 0.0000 Salw = 3.00000e-006 Irov = 0 Z0v = [.0000000e+002 1.0000000e+000 Forfuv= #Y# Forfww= #Y# Sigcor= #N# Trasol= #Cyclic-method# Momsol= #Cyclic# Commnt= Filsrc= #tut fti 1.00250000 0.00000 Dicouv= 10. Flag for extra drying and flooding Option for check at water level points Option for check at velocity points Threshold depth drying and flooding Marginal depth in shallow area’s Time interval to smooth the hydrodynamic boundary conditions Relaxation parameter for QH relation Flag horizontal Forester filter Flag vertical Forester filter Flag to activate anti-creep Numerical method for advective terms Numerical method for momentum terms C*256 File with discharge locations <∗.0000 Xlo = 0.000000 Htur2d= #N# Vicouv= 2. Only for heat model 4.0000000e+002 1.000000 ThetQH= 0.81300 Rhow = 1030.00250000 0.00250000 100.src> continued on next page Deltares 415 .] Tempw = 15.0000000e+002 Rettib= 1.000000 Ivapop = 0 Temint= #Y# Commnt= Roumet= #C# Ccofu = 45.00 Alph0 = [. Uniform bottom roughness in v-dir.] Commnt= Iter = 2 Dryflp= #YES# Dpsopt= #MAX# Dpuopt= #MEAN# Dryflc= 0.00000 Betac = 0. Interpolation flag for temperature data C*1 1R 1R 1R C*2 1R 1R 1R 1R 1I 1R Type of bottom friction formulation Uniform bottom roughness in u-dir.0000 Ccofv = 45.

par> Format of drogues file Filcrs= #tut fti 1.00 10.1.crs# Fmtcrs= #FR# Commnt= PMhydr= #YYYYYY# PMproc= #YYYYYYYYYY# PMderv= #YYY# PHhydr= #YYYYYY# PHproc= #YYYYYYYYYY# PHderv= #YYY# PHflux= #YYYY# SMhydr= #YYYYY# SMproc= #YYYYYYYYYY# SMderv= #YYYYY# SHhydr= #YYYY# SHproc= #YYYYYYYYYY# SHderv= #YYYYY# SHflux= #YYYY# Commnt= attribute file fourier analysis Filfou= # # Online= #YES# Prmap = [. cross sections: 4 C*256 C*2 C*256 C*2 File with observation points <∗.00 Flrst = -999.00 Flpp = 2190.000000 5. observation points: 5 Filsta= #tut fti 1.obs> Format of observation points file File with drogues <∗.000 2940. User Manual continued from previous page Keyword and value Format C*2 C*256 C*2 Description Format of discharge locations file File with discharge data <∗.00 120.par# Fmtpar= #FR# Commnt= Eps = [.] Prhis = 750.obs# Fmtsta= #FR# Filpar= #tut fti 1.0000 Flmap = 2190.dis> Format of discharge data file Fmtsrc= #FR# Fildis= #tut fti 1.00 Flhis = 0.crs> Format of cross-sections file C*6 C*10 C*3 C*6 C*10 C*3 C*4 C*5 C*10 C*6 C*4 C*10 C*5 C*4 Flags print map output hydrodynamic quantities Flags print map output constituents and turbulence Flags print map output derived quantities Flags print time history hydrodynamic quantities Flags print time history constituents and turbulence Flags print time history derived quantities Print flags time history fluxes through cross-sections Flags store map output hydrodynamic quantities Flags store map output constituents and turbulence Flags store map output derived quantities Flags store time history output hydrodynamic quantities Flags store time history output constituents and turbulence Flags store time history output derived quantities Flags store time history output fluxes through cross-sections C*256 C*3 File with quantities to be Fourier analysed <∗.000 0. Default = #Y# A.000000 60.999 Commnt= BarocP = #Y# C*256 C*2 File with cross-sections <∗.0000 2940.00000 2940.] Commnt= Commnt= no.fou> Flag for online visualisation Time instances to print map output Time information to print history output Time information to store map output Time information to store history output Time information to write to the communication file Time interval to write restart file {1 R} 3R 3R 3R 3R 1R C*1 Flag for activating barocline pressure term at open boundaries.Delft3D-FLOW .3 Physical parameters 416 Deltares .dis# Fmtdis= #FR# Commnt= no.

0000000e+001 Rettib = 1.0000000e+002 1.0000000e+000 5. The layout of Tidfor is defined as follows: Tidfor = #M2 S2 N2 K2# #K1 O1 P1 Q1# #Mf Mm Ssa # A. Here is an example specifying a return time of 100 minutes for all constituents.0000000e+002 2.Files of Delft3D-FLOW A. Remarks: Rettis and Rettib can be combined with the additional keywords to specify return times for all constituents. you have to add them with a text editor to the mdf-file. Example with 3 open boundaries: Rettis = 1.0000000e+002 2.0000000e+000 5.0000000e+001 It is also possible to specify return times for individual constituents.3.0000000e+002 RetbT = 2.2 Thatcher-Harleman Conditions The Thatcher-Harleman conditions are by default specified by a return time (in minutes) for the surface and the bed level for each open boundary. the constituent specific variants are not.3. using the following keywords: RetsS RetbS RetsT RetbT Rets01 Retb01 Return time for Salinity at surface Return time for Salinity at bed level Return time for Temperature at surface Return time for Temperature at bed level Return time for Constituent number 1 at surface Return time for Constituent number 1 at bed level Idem for constituents with other numbers. except for temperature.1.0000000e+002 1.0000000e+002 0. having a return time of 200 minutes: Rettis = 1.1.1 Tide Generating Forces For tide generating forces you can specify the tidal components that are taken into account.0000000e+002 RetsT = 2.0000000e+002 Rettib = 1.0000000e+002 Rettis and Rettib are supported by FLOW-GUI. Deltares 417 . with varying values for specific constituents.0000000e+002 0.

SMderv to select the quantities and Flmap to select the output times.1. The keywords related to 2D/3D MAP-file output are SMhydr. PH*. Table A. PHflux to select the quantities and Prhis to select the output times. Richardson number if either or both are selected Density PMhydr = #YYYYYY# PMproc = #YYYYYYYYYY# PMderv=#YYY# 3 Table A. PHproc. so the flags for each option are described in the tables below. SM* and SH* keywords are strings of Y and N characters representing flags for output of different quantities.2: Print flags for map-data Keyword and value Number/Description 1 2–5 6 1 2 3–7 8 9 and 10 1 and 2 Water level U and V-velocities. The keywords related to history station and cross-section ascii output are SHhydr. The keywords related to 2D/3D ascii output are PMhydr. SMproc. the quantities written to the com-file cannot be changed. PMproc.and V-velocities.6 lists a couple of options for additional output to map. SHproc. magnitude and direction ω -velocities relative to σ -plane and wvelocities PHhydr = #YYYYYY# 418 Deltares . User Manual A. The values to be specified for the PM*. PMderv to select which quantities to write and Prmap to indicate the time steps at which to write these fields.4 Output options The MDF-file contains a number of keywords for selecting output options to print (ascii) files and binary NEFIS map-. The ascii output options are generally not used except for debugging purposes. PHderv. his.and his-files and the tri-diag file which can be switched on in the Additional parameters section of the user interface. SHflux to select the quantities again and Flhis to select the times. Finally Table A. The keyword Flpp specifies the output times for the com-file. magnitude and direction ω -velocities relative to σ -plane and wvelocities Concentration salinity Temperature Concentration constituents Intensity spiral motion Turbulent energy and dissipation Vertical eddy viscosity and vertical eddy diffusivity.and com-files.3: Print flags for history-data Keyword and value Number/Description 1 2–5 6 Water level U. SHderv. These quantities differ slightly for storing results to file and for printing. The keywords related to history station and cross-section ascii output are PHhydr.Delft3D-FLOW.

Richardson number if either or both are selected Density Filtered U.4: Storage flags for map-data Keyword and value Number/Description 1 2 and 3 4 and 5 1 2 3–7 8 9 and 10 1 and 2 3 and 4 Water level U.and V-velocities ω -velocities relative to σ -plane and wvelocities Concentration salinity Temperature Concentration constituents Intensity spiral motion Turbulent energy and dissipation U. Richardson number if either or both are selected Density Total flux through cross-setions Momentary flux through cross-sections Advective transport through cross-sections Dispersive transport through cross-sections PHproc = #YYYYYYYYYY# PHderv=#YYY# 3 PHflux = #YYYY# 1 2 3 4 Table A.and V-velocities W-velocities SHhydr = #YYYY# Deltares 419 .and V-velocities of HLES model and horizontal eddy viscosity SMhydr = #YYYYY# SMproc = #YYYYYYYYYY# SMderv=#YYYYYY# 5 6 Table A.and V.3: Print flags for history-data Keyword and value Number/Description 1 2 3–7 8 9 and 10 1 and 2 Concentration salinity Temperature Concentration constituents Intensity spiral motion Turbulent energy and dissipation Vertical eddy viscosity and vertical eddy diffusivity.Files of Delft3D-FLOW Table A.5: Storage flags for history-data Keyword and value Number/Description 1 2 and 3 4 water level U.bed stress components Vertical eddy viscosity and vertical eddy diffusivity.

User Manual Table A. This can be used to check whether the (combined) prescribed meteo input resulted in the expected model forcing.Delft3D-FLOW.5: Storage flags for history-data Keyword and value Number/Description 1 2 3–7 8 9 10 1 and 2 3 and 4 Concentration salinity Temperature Concentration constituents Intensity spiral motion Turbulent kinetic energy Turbulent dissipation U. is defined as: Item First real number Second real number Third real number Description Start time in minutes after the computation start time Time interval in minutes Stop time in minutes after the computation start time Table A.bed stress components Vertical eddy viscosity and vertical eddy diffusivity. ´ values to map-file. Write roughness as Chezy Default #N# AirOut HeaOut Chezy #Y# or #N# #Y# or #N# #N# #N# continued on next page 420 Deltares .and V. The layout of the time information to store results.6: Optional output flags under Additional parameters Keyword Value #Y# or #N# Description Write meteo input to map-file. Flag for output of heat fluxes determined by temperature models. such as Prhis and Flmap.and V-direction in ζ -point Total flux through cross-sections1) Momentary flux through cross-sections1) Advective flux through cross-sections1) Dispersive flux through cross-sections1) SHproc = #YYYYYYYYYY# SHderv=#YYYYY# 5 SHflux=#YYYY# 1 2 3 4 1) if cross-sections are defined. Richardson number if either or both are selected Density Flux in U.

Files of Delft3D-FLOW Table A. Threshold value in [m/s] for reporting velocity component change messages to tri-diag file. If the centre point is located in multiple polygons. the grid cell will be assigned to the first polygon only. Flag for additional performance timing output to tri-diag file.6 – continued from previous page Keyword Value #Y# or #N# Description Write roughness in input unit (as specified Roumet keyword) to map-file. Default #N# Rough AdvFlx #Y# or #N# #N# CumAFl #Y# or #N# #N# LayOut SMVelo AddTim SgrThr UgrThr Filbal #Y# or #N# #glm# or #euler# #Y# or #N# #N# #euler# #N# 25 m 5 m/s - 10−3 to 103 10−3 to 103 #filename# continued on next page Deltares 421 . constituent and sediment flux data. The values will be accumulated over grid cells for which the centre point is located within the polygon. Flag for output of vertical coordinates of layer interfaces to the map-file. Note 1: anticreep and Forester filter fluxes are not included in these fluxes. Triggers mass balance output for the polygons on the history-file: volume. Flag for output of cumulative horizontal fluxes of the advection diffusion equation per substance. Flag for velocity output to his.2. This is particularly useful when using the water depth dependent roughness formulations offered by trachytopes. area. Note 2: AdvFlx will be automatically switched on if CumAFl is activated. Flag for output of instaneous horizontal fluxes of the advection diffusion equation per substance. Note: anticreep and Forester filter fluxes are not included in these fluxes. average concentration and average bed level data as well as total water. The polygon file should match the file format described in Section A.24.and map-files. Threshold value in [m] for reporting water level change messages to tri-diag file.

This specification may be useful when output reduction is needed. For each file we give the following information: File contents. depth averaged velocity.2. Most of the attribute files can be generated by the FLOW-GUI after defining an input scenario. Most of these files contain the quantities that describe one specific item. or time-series for wind speed and direction by client specific tools. Example(s). vertical profiles). with Delft3DQUICKPLOT) may be unable/awkward when necessary information is missing. Still. By default all layers are written. Default - SHlay 136 SMlay 2578 - A. Restrictions on the file contents. unless explicitly stated differently. By default all layers are written. (e.Delft3D-FLOW. Filename and extension. or time dependent data of fluxes discharged in the model area by discharge stations.6 – continued from previous page Keyword Value Description Specification of the layers for which the output parameters are written to history-file.g. Postprocessing (e. Postprocessing (e. how to generate the file). Specification of the layers for which the output parameters are written to map-file. such as the location of open boundaries. such as astronomic boundary conditions.g. we describe both type of files as it might be useful to know how the input data is structured to be able to generate (large) files. In the examples the file content is printed in font Courier and comment (not included in the file) between curly brackets font.g.g. vertical profiles). with Delft3DQUICKPLOT) may be unable/awkward when necessary information is missing. This specification may be useful when output reduction is needed. 422 Deltares . Some files can almost only be generated by utility programs such as the curvilinear grid generated by Delft3D-RGFGRID.e. Generated by (i. depth averaged velocity.2 A. Remarks: The access mode of all attribute files is sequential. Filetype (free formatted.1 Attribute files Introduction In the following sections we describe the attribute files used in the input file (MDF-file) of Delft3D-FLOW. (e. fix formatted or unformatted). User Manual Table A.

...000000. 23:19:22 * Coordinate System = Cartesian 9 7 0 0 0 Eta= 1 0. Sep 30 2008. A label and record number.000000. Restrictions: The grid must be orthogonal.and n-direction (2 integers).2. ASCII Free formatted <name.. Example: * * Deltares. File contents Filetype File format Filename Generated Record description: Record Record description Preceding description records. This set of records is repeated for each row until n = nmax... 5.00000000000000000E+02 6.00000000000000000E+00 1.00000000000000000E+02 7. will be ignored.000000.00000000000000000E+02 2.000000. K+4 to 2K+3 K is the number of records to specify for all grid points a set of x. the x-component of the world coordinates of all points in m-direction.or y -co-ordinates. A similar set of records for the y -component of the world coordinates. 1 Record with The co-ordinates of the orthogonal curvilinear grid at the depth points. Input items in a record are separated by one or more blanks. Three real values (not used).Files of Delft3D-FLOW A.grd> Delft3D-RGFGRID Co-ordinate System = Cartesian or value Spherical 2 3 4 to K+3 The number of grid points in m. The label and record number are suppressed on the continuation lines. Delft3D-RGFGRID Version 4.2 Orthogonal curvilinear grid The orthogonal curvilinear grid file can be specified in the FLOW-GUI in Data group Domain . with as many continuation records as required by mmax and the number of co-ordinates per record.00000000000000000E+02 6. starting with an asterisk (∗). starting with row 1 to row nmax. Eta= 2 0. 5.16..00000000000000000E+02 2. Deltares 423 ..Grid parameters.01.00000000000000000E+02 7.00000000000000000E+00 1. 23:32:27 * File creation date: 2008-10-01.4531.

the file itself is generated by Delft3D-RGFGRID..00000000000000000E+02 3.00000000000000000E+02 1... 2.00000000000000000E+02 2.00000000000000000E+02 2..000000..00000000000000000E+02 4.00000000000000000E+02 4.00000000000000000E+02 1.000000...000000. 2.00000000000000000E+02 7.000000.000000.000000..000000..00000000000000000E+02 0. File contents The indices of the external computational grid enclosure(s) and optionally one or more internal computational grid enclosures that outlines the active computational points in a Delft3D-FLOW computation. 1.. 2. 7..00000000000000000E+02 6. 7. 4.00000000000000000E+02 4. 135.00000000000000000E+02 4. A line segment may not intersect or touch any other line segment.2..00000000000000000E+02 3..000000.00000000000000000E+02 5...00000000000000000E+02 7.....00000000000000000E+00 5. 90.00000000000000000E+02 0..00000000000000000E+02 2. 3....000000.00000000000000000E+02 3..00000000000000000E+02 1.00000000000000000E+02 2.00000000000000000E+02 6. see RGFGRID (2008). 7.000000. 225... Restrictions: A polygon must be closed.000000.00000000000000000E+02 1.00000000000000000E+02 6.00000000000000000E+02 5. 424 Deltares ..00000000000000000E+02 3..00000000000000000E+02 6.00000000000000000E+02 5.000000.. 6.000000.. The angle formed by consecutive line segments (measured counter clock-wise) can have a value of: 45. The first point of the polygon is repeated as last point..000000.000000.000000..00000000000000000E+02 6. 5.00000000000000000E+00 5..00000000000000000E+00 5. 2. 5..000000.3 Computational grid enclosure The computational grid enclosure file need to be specified in the FLOW-GUI in Data Group Domain .. 270 or 315 degrees.000000.00000000000000000E+00 5.00000000000000000E+02 1.Delft3D-FLOW. 4.00000000000000000E+02 0.00000000000000000E+02 6...000000. Input items in a record are separated by one or more blanks.Grid parameters.. 7.000000.00000000000000000E+00 5.00000000000000000E+02 1. 7. 7. The file is strongly related to the curvilinear grid file. A.000000.00000000000000000E+02 1.00000000000000000E+02 7.00000000000000000E+02 7.000000. 2..00000000000000000E+02 1.. 6..00000000000000000E+02 6. In a row or column there should be at least two active computational grid cells. 2. 3.00000000000000000E+02 6..00000000000000000E+02 2. but not 0..00000000000000000E+02 5.enc> Delft3D-RGFGRID Filetype File format Filename Generated Record description: Record All Record description One pair of M and N indices representing the grid co-ordinates where a line segment of the computational grid enclosure (polygon) changes direction. ASCII Free formatted <name.. 1.000000. User Manual Eta= Eta= Eta= Eta= Eta= Eta= Eta= Eta= Eta= Eta= Eta= Eta= 3 4 5 6 7 1 2 3 4 5 6 7 0.. 7.000000.00000000000000000E+02 1.00000000000000000E+02 6....00000000000000000E+02 0. 180 and 360 degrees.

+ .+ .+ .+ .+ .d ir ec tio n 1 9 + . location for velocity or discharge open boundaries.+ .+ .Files of Delft3D-FLOW N.+ .+ .+ .+ .+ .+ . represented by depth values (in metres) for all grid points.2.+ .+ .+ .+ .+ .Bathymetry.+ .+ .+ .+ . 1 6 8 9 9 16 19 19 17 4 1 1 13 14 14 13 13 1 1 3 3 1 1 4 6 8 8 5 1 4 4 5 5 4 begin external polygon end external polygon begin internal polygon end internal polygon A.+ .+ .+ .+ .+ .1.+ .+ .+ .+ . File contents Filetype Deltares The bathymetry in the model area.+ .+ .+ .+ .+ .+ .+ .+ . ASCII 425 + .+ .+ .+ .1: Example of computational grid enclosures .+ .+ - + .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ - + .+ .+ .+ . Figure A.+ .4 Bathymetry The bathymetry file can be specified in the FLOW-GUI in Data Group Domain .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 M.+ .+ .+ .+ .+ .+ .+ .+ .+ .+ - + .+ - + .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ - water level point v-velocity point u-velocity point grid enclosure and (for the external polygon only) location of water level open boundaries.+ .+ .+ .+ .+ .+ - + .+ .+ .+ .+ .d irectio n 8 7 6 5 4 3 2 1 Legend: + | − Full thick line Full thin line Example: Model area with (one) external and one internal polygon.+ - + .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ .+ . see Figure A.+ .+ .

0 18.5 Thin dams The thin dams file can be specified in the FLOW-GUI in Data Group Domain .0 14.0 16.Thin dams.0 -999.0 8.0 15.0 -999.0 10. User Manual File format Filename Generated Free formatted or unformatted <name.0 -999.0 13.0 18.0 5.0 17.0 -7.0 11.0 15.0 17.0 9.0 2.0 -999.0 13. Record description: Filetype Free formatted Record description Depth values per row.0 -999.0 -999.0 -999.0 -7.0 13.0 19.0 9. The maximum record length in the free formatted file is 132.0 -7.0 17.0 16.0 3.0 15.0 18. Mmax depth values per row for N = 1 to N = Nmax. The default missing vale is: −999.0 13.0 19.0 8.0 17.0 12.0 -999.0 -8.0 10.0 14.0 15.0 -999.0 -999.0 11.0 19.0 10.0 -5.0 3.0 17.0 A.0 15.0 -999.0 -5.0 9.0 13.0 9.0 Example: File containing 16 ∗ 8 data values for a model area with 15 ∗ 7 grid points (free formatted file). File contents Filetype File format Filename 426 Location and type of thin dams.0 17.0 4.0 18.dep> FLOW-GUI (only for uniform depth values).0 6. separated by one or more blanks.0 16.0 12.0 15.0 18.thd> Deltares .0 19.0 -999.0 17.0 12.0 16.0 -999.0 -999.0 15.Delft3D-FLOW.0 13.0 14.0 19.0 -999.0 16.0 -999.0 20.0 16. Offline with Delft3D-QUICKIN and data from digitised charts or GISdatabase.0 -5.0 18.0 12.0 17.0 -6.0 10.0 10. 1. starting at N = 1 to N = Nmax.0 12.0 16.0 6. Depth values from the file will not be checked against their domain.0 -999.0 14.0 4.0 13.0 8.0 7.0 14.0 -8.0 14.0 16. Unformatted Restrictions: The file contains one M and N line more than the grid dimension.0 7.0 18. The input items are separated by one or more blanks (free formatted file only).2.0 14.0 20.0 11.0 11.0 19. The number of continuation lines is determined by the number of grid points per row (Mmax) and the maximum record size of 132.0 19.0 12.0 5.0 19.0 7.0 -6.0 -999.0 17. ASCII Free formatted <name.0 -999.0 18.0 15.0 19.0 -999.0 -5.0 -999.0 -999.0 20.0 -999.

Input items are separated by one or more blanks.Dry points.Files of Delft3D-FLOW Figure A. File contents Filetype Deltares Index location of (permanently) dry points. A character indicating the type of thin dams (U or V).2: Example of thin dams in a model area Generated Record description: Record each record Delft3D-QUICKIN or FLOW-GUI Record description The grid indices of the begin and end point of a line of thin dams (4 integers).2. Thin dams can not be defined along the model boundaries (which by default lie along the lines M = 1. Therefore.2. the indices of thin dams must lie between M = 2 and Mmax-1 and N = 2 and Nmax-1 respectively. see Figure A. ASCII 427 . M = Mmax or N = Nmax). 6 7 12 2 4 3 6 10 12 4 V 7 U 7 U A.6 Dry points The dry points file can be specified in the FLOW-GUI in Data Group Domain . Restrictions: The angle of the line segment and the horizontal numerical grid axis may be an integer multiple of 45 degrees. the direction of the dam is perpendicular to the velocity direction over which the dams are superimposed!! Example: Three (sets of) thin dams in model area of 19 ∗ 8 grid points. N = 1.

Dry points may not be defined along the model boundaries (which by default lie along the lines M = 1. respectively. M = Mmax or N = Nmax). The input items are separated by one or more blanks. N = 1.dry> Delft3D-QUICKIN or FLOW-GUI Record description The grid indices of the begin and end point of a dry section (4 integers). User Manual Figure A.3. Example: Five sets of dry points in a model area of 19 ∗ 8 grid points. 2).Delft3D-FLOW. Therefore. 5 8 13 13 14 3 4 3 4 6 5 10 14 14 14 6 6 3 4 6 428 Deltares . Restrictions: The angle of a line of dry points and the horizontal numerical grid axis can be an integer multiple of 45 degrees.3: Dry points in model area File format Filename Generated Record description: Record each record Free formatted <name. see Figure A. The most lower-left dry point has indices (2. the indices of these points must lie between M = 2 and Mmax-1 and N = 2 and Nmax-1.

A. 0.Files of Delft3D-FLOW A. 610.0 0. 235.8 Space varying wind and pressure In this section the different options are described for specifying space varying meteo data (wind. 30.7 Time-series uniform wind Time-series for wind speed and direction for a uniform wind can be specified in the FLOW-GUI in Data Group Physical parameters . i. starting as a north wind.wnd> FLOW-GUI.7 4.4 is about +60 degrees. The wind speed in m/s and the wind direction relative to north.2 5. 10. 315. ASCII or binary Free formatted or unformatted <name. The input items are separated by one or more blanks. 0. 600.4.4. Example: Time-series for uniform wind field.2. 900. see Figure A. 225. atmospheric pressure. Deltares 429 . More details on the compatibility of the meteo input files and the conversion from one version to another can be found in Appendix B. The contents of the file will not be checked on its domain.7 4.0 0. 150. File contents Time-series for wind speed and direction for a uniform wind. WAVE-GUI.Wind. The wind direction in Figure A. relative to north and positive in clock-wise direction. 270. or manually offline Filetype File format Filename Generated Record description: Record each record Record description The time in minutes after the Reference date 00:00:00 hr (1 real). 0. positive clock wise (2 reals).) as input for your simulation.e. The descriptions in this section are applicable to meteo input files of a certain version.0 1. turning to south-west and back to north. etc. The wind direction is defined according to the nautical convention.2 3. 225. see Figure A.2. Restrictions: Times must be an integer multiple of the simulation time step.7.

Free formatted or unformatted.2. defined on the computational grid. File contents Time-series for space varying wind velocity components (east-west and south-north) and atmospheric pressure. Filetype File format Filename Generated Header description: 430 Deltares . The header specifies the type of file and the input it contains using a number of keywords. keyword based.4: Definition sketch of wind direction according to Nautical convention A. The keywords are case insensitive and the order of the keywords is not fixed.wnd> Some offline program. User Manual Figure A. followed by datablocks containing the wind and pressure fields at times specified using a standardised time definition above each datablock.Delft3D-FLOW. The file consists of a header. ASCII or binary. <name.1 Defined on the computational grid Time-series for space varying wind velocity components (east-west and south-north) and atmospheric pressure can be specified in the FLOW-GUI in Data Group Physical parameters Wind.8.

used to completely determine the time at which a dataset is valid. The time definition and the data block — for all three quantities — are repeated for each time instance of the time-series. The time definition is followed by the datablock of input values corresponding to the specified time. 1) on the grid. meters/second unit of quantity3. the velocity component in N-direction and the atmospheric pressure. Every next line in the dataset then corresponds to a row on the grid. Deltares 431 . Pa or millibar FileVersion Filetype NODATA value free n quantity 3 quantity1 quantity2 quantity3 unit1 x wind y wind air pressure m s-1 unit2 m s-1 unit3 Pa or mbar Time definition and data block description Keywords Value fixed format described below Description time definition string Time The time definition string has a fixed format.g.Files of Delft3D-FLOW Keywords Value 1. where the first value in the dataset corresponds to cell (1. No extra spaces or tabs can be added between the different parts of the definition.03 meteo on computational grid Description version of file format meteo input on computational grid value used for input that is to be neglected number of quantities specified in the file wind in x-direction wind in y-direction air pressure unit of quantity1. 360 minutes since 2008-07-28 10:55:00 +01:00 The format of the string is completely fixed. The time definition string has the following format: TIME minutes/hours since YYYY-MM-DD HH:MM:SS TIME ZONE. All three quantities are given for Nmax by Mmax points. e. meters/second unit of quantity2. respectively. The data block consists of three subsequent blocks containing the velocity component in M-direction.

Delft3D-FLOW, User Manual File version and conversion The current description holds for FileVersion 1.03. The table below shows the latest modifications in the file format (and version number). FileVersion Modifications No changes for this meteo input type, but for the meteo types meteo on equidistant grid and meteo on curvilinear grid No changes for this meteo input type, but for the meteo type meteo on spider web grid Changed keyword MeteoType to FileType Changed fixed value of input type (Keyword Filetype) from Svwp to meteo on computational grid (meteo on flow grid is also allowed) Restrictions: Keywords are followed by an equal sign ’=’ and the value of the keyword. When a keyword has value free the value of this keyword is free to choose by the user. When only one value is given for a keyword, this keyword has a fixed value and when 2 or more options are shown, the user can choose between those values. Times must be specified exactly according to the time definition. See the examples shown in this section. The contents of the file will not be checked on its domain. The wind components are specified at the cell centres (water level points) of the computational grid. Input items in a data block are separated by one or more blanks (free formatted file only). Remarks: The time definition in the meteorological file contains the number of minutes or hours since a reference data and time in a certain time zone. The reference time and time zone may differ from those of the simulation. The computational engine will search in the meteo file for the simulation time and interpolate between neighbouring times if necessary. Possible differences in time zone will be accounted for by shifting the meteo input data with the difference. The reference times within the time definition string may vary in a meteo file, i.e. it is possible to attach new input with a different reference time, behind the last data block. Comments can be added after #’s.

1.03

1.02

1.01

432

Deltares

Files of Delft3D-FLOW

Figure A.5: Definition wind components for space varying wind

Example Model area of 25 ∗ 33 grid points (Mmax = 25; Nmax = 33). The input data is printed in Courier, comments are printed behind #’s.
Time = 0.0 minutes since 2008-09-20 10:30:00 +01:00 {33 records with 25 values each} {33 records with 25 values each} {33 records with 25 values each} Time = 340.0 minutes since 2008-09-20 10:30:00 +01:00 {33 records with 25 values each} {33 records with 25 values each} {33 records with 25 values each} Time = 600.0 minutes since 2008-09-20 10:30:00 +01:00 {33 records with 25 values each} {33 records with 25 values each} {33 records with 25 values each} Time = 1240.0 minutes since 2008-09-20 10:30:00 +01:00 {33 records with 25 values each} {33 records with 25 values each} {33 records with 25 values each}
# Time definition # Wind component west to east # Wind component south to north # Atmospheric pressure # Time definition # Wind component west to east # Wind component south to north # Atmospheric pressure # Time definition # Wind component west to east # Wind component south to north # Atmospheric pressure # Time definition # Wind component west to east # Wind component south to north # Atmospheric pressure

Remarks: To obtain the wind direction according to the nautical convention, the wind direction is reversed. The wind is specified in terms of its components along the west-east (x wind) and south-north (y wind) co-ordinate system, see Figure A.5. These definitions differ from the nautical convention (used for uniform wind), which is specified relative to true North, see Figure A.4. Remark: On open boundaries, an input signal is prescribed that is consistent with some average pressure. Usually the signal corresponds to Mean Sea Level. One actually wants to prescribe an input signal corresponding to the local pressure prescribed by the space varying meteo input. To this end, it is possible to specify an average pressure - which should correspond to your input signal on the open boundaries - which is then used to determine local pressure gradients that need to be applied along the open boundaries to obtain an input signal that is consistent with the local atmospheric pressure. This

Deltares

433

Delft3D-FLOW, User Manual functionality used to be specified in the wind file but it should now be specified in the Master Definition File in the Data Group Additional parameters, using PavBnd: Average Pressure on Boundaries. Using this keyword one can specify an average pressure that is used on all open boundaries, independent of the type of wind input. The pressure must be specified in N/m2 . An example:
PavBnd= 101300.0

A.2.8.2

Defined on an equidistant grid Time-series for space varying wind velocity components (east-west and south-north) and atmospheric pressure on an equidistant grid (other than the computational grid) can be specified in the FLOW-GUI in Data Group Additional parameters. File contents Time-series of a space varying wind and atmospheric pressure defined on an equidistant rectilinear or spherical grid (other than the computational grid). ASCII. Free formatted, keyword based. <name.amp> for atmospheric pressure in [Pa] or [millibar], <name.amu> for the wind speed component in east-west-direction in [m/s], <name.amv> for the wind speed component in north-south direction in [m/s]. Some offline program.

Filetype File format Filename

Generated

Remark: Space varying wind and pressure on an equidistant grid is implemented as a special feature. You must specify some additional keywords and values in Data Group Additional parameters or in the MDF-file; see Section B.7.1 for details.

A.2.8.3

Defined on a curvilinear grid Time-series for space varying wind velocity components (east-west and south-north) and atmospheric pressure on a separate curvilinear grid (other than the computational grid) can be specified in the FLOW-GUI in Data Group Additional parameters. File contents Time-series of a space varying wind and atmospheric pressure defined on a curvilinear (Cartesian or spherical) grid (other than the computational grid). ASCII. Free formatted, keyword based. <name.amp> for atmospheric pressure in [Pa] or [millibar], <name.amu> for the wind speed component in east-west-direction in [m/s], <name.amv> for the wind speed component in north-south direction in [m/s], <name.grd> for the curvilinear grid on which the wind and pressure are specified. Some offline program.

Filetype File format Filename

Generated Remark: 434

Deltares

Files of Delft3D-FLOW Space varying wind and pressure on a curvilinear grid is implemented as a special feature. You must specify some additional keywords and values in Data Group Additional parameters or in the MDF-file; see Section B.7.2 for details.

A.2.8.4

Defined on a Spiderweb grid Time-series for space varying wind velocity components (east-west and south-north) and atmospheric pressure on a Spiderweb grid can be specified in the FLOW-GUI in Data Group Additional parameters. File contents Time-series of a space varying wind and atmospheric pressure defined on a Spiderweb grid. The grid can be rectilinear or spherical. This type of wind input is used to describe (rotating) cyclone winds. ASCII. Free formatted, keyword based. <name.web> containing the wind speed in [m/s], wind (from) direction in [degree] and atmospheric pressure in [Pa] or [millibar]. Some offline program.

Filetype File format Filename Generated

Remark: Space varying wind and pressure on a Spiderweb grid is implemented as a special feature. You must specify some additional keywords and values in Data Group Additional parameters or in the MDF-file; see Section B.7.3 for details.

A.2.9

Initial conditions File contents

File format Filename Generated

Initial conditions for the hydrodynamics, the transported constituents (if any) and the secondary flow intensity (if any) at all points (Data Group Initial conditions). Free formatted or unformatted. <name.ini> Offline by some external program.

Initial conditions for separate quantities can be generated with the program Delft3D-QUICKIN. Record description: Record each record Record description A record contains a row of Mmax values (Mmax reals).

The file contains a matrix of dimensions (Mmax∗Nmax) for each quantity and if relevant for each layer for which an initial condition is required. The matrices are given in the following order: 1 2 3 4 5 Water elevation (one matrix). U-velocities (Kmax matrices). V-velocities (Kmax matrices). Salinity, only if selected as an active process (Kmax matrices). Temperature, only if selected as an active process (Kmax matrices). 435

Deltares

Delft3D-FLOW, User Manual 6 Constituent number 1, 2, 3 to the last constituent chosen, only if selected (Kmax matrices). 7 Secondary flow (for 2D simulations only), only if selected as an active process (one matrix). In total there will be (Cmax+2)∗Kmax + 1 matrices of Mmax∗Nmax where: Mmax Nmax Kmax Cmax number of grid points in M (U or ξ ) direction. number of grid points in N (V or η ) direction. number of layers. number of constituents (including temperature, salinity, secondary flow).

Restrictions: The maximum record length in the free formatted file is 132 The contents of the file will not be checked on its domain. Input items in a record are separated by one or more blanks (free formatted file only). Example 1: A model of 1 layer, Mmax = 25 and Nmax = 33, with two constituents, TEST1 and TEST2, and secondary flow. The input data is printed in Courier; comment (not in the file) is printed between brackets.
{33 {33 {33 {33 {33 {33 records records records records records records with with with with with with 25 25 25 25 25 25 values values values values values values each} each} each} each} each} each} {Water elevation} {U-velocity component} {V-velocity component} {Concentrations constituent TEST1} {Concentrations constituent TEST2} {Secondary flow intensity}

Example 2: A model with 2 layers, Mmax = 25 and Nmax = 33, with salinity, temperature and one constituent, denoted by TEST. The input data is printed in Courier; comment (not in the file) is printed between brackets.
{33 {33 {33 {33 {33 {33 {33 {33 {33 {33 {33 records records records records records records records records records records records with with with with with with with with with with with 25 25 25 25 25 25 25 25 25 25 25 values values values values values values values values values values values each} each} each} each} each} each} each} each} each} each} each} {Water elevation} {U-velocity component in layer 1} {U-velocity component in layer 2} {V-velocity component in layer 1} {V-velocity component in layer 2} {Salinity in layer 1} {Salinity in layer 2} {Temperature in layer 1} {Temperature in layer 2} {Concentrations constituent TEST in layer 1} {Concentrations constituent TEST in layer 2}

Remarks: A record for the free formatted file is to be interpreted as a logical record. The length of a physical record is limited to 132, so a logical record consists of as many physical records as required by Mmax and the number of values per (physical) record. The velocity components are the computational U and V components, not the east and north components. 436 Deltares

Files of Delft3D-FLOW A.2.10 Open boundaries The open boundaries file(s) can be specified in the FLOW-GUI in Data Group Boundaries. File contents Filetype File format Filename Generated Record description: Record each record Record description Name of the open boundary section (20 characters). Type of boundary (1 character). Z C N Q T R water level current Neumann discharge per grid cell total discharge for boundary section Riemann The location and description of open boundaries. ASCII Fix formatted for text variables; free formatted for real and integer values. <name.bnd> FLOW-GUI

Type of data (1 character). A H Q T astronomic harmonic QH tables (only for water level boundaries) time-series

Grid indices of the begin and end point of the boundary section (4 integers). Reflection coefficient (1 real), not for Neumann or Riemann. Vertical profile (three strings); only for 3D simulations and velocity type boundaries (C, Q, T and R). Uniform Logarithmic 3D profile Two labels (each 12 characters, no blanks in the label name) referencing to the blocks in the amplitude and phase file <∗.bca>; only if the type of data is A.

Restrictions: Maximum record length in the free formatted file is 132. The boundary section name must start at position one in a record. The value of the reflection coefficient will not be checked on its domain. All input items in a record must be separated by one or more blanks. Astronomic and harmonic forced boundaries must be specified before QH-relation forced Deltares 437

Delft3D-FLOW, User Manual boundaries, which in turn should be specified before time-series forced boundaries. Astronomic and harmonic forced boundaries cannot be combined. Example: Two boundary sections with data type A(stronomic) and one with type T(ime series).
Paradise Bay 1 Paradise Bay 2 Sea Entrance Q A C A Z T 1 16 4 1 3 8 1 16 14 5 0.0 Uniform 6 0.0 Logarithmic 8 0.0 Paradise_1A Paradise_2A Paradisee_lB Paradisee_2B

Remarks: A label may not contain blanks between non-blank characters. For the labels 12 characters are read. Be sure the second label starts at least 13 positions after the start of the first.

A.2.11

Astronomic flow boundary conditions File contents Boundary conditions for open boundary sections of type Astronomic (Data Group Boundaries - Flow conditions) in terms of amplitudes and phases for the astronomic components. Filetype ASCII File format Fix formatted for text variables, free formatted for real and integer values. Filename <name.bca> Generated FLOW-GUI or offline by program Delft3D-TRIANA. Record description: Record 1 Record description Label for end point A of open boundary section (12 characters, no blanks) For each component its name (8 characters), amplitude and phases (2 reals). Label for end point B of open boundary section (12 characters, no blanks). For each component its name (8 characters), amplitude and phases (2 reals).

2 to 2+NCOM-1

2+NCOM

2+NCOM+1 to 2+2∗NCOM where: NCOM

number of tidal components.

These records are repeated for each open boundary section. Restrictions: The labels for the end points of an open boundary section are defined in the <name.bnd> 438 Deltares

Files of Delft3D-FLOW file. The name of the label may not contain blanks between non-blnak characters. All component names are written in upper case. The label and component name must start in position one. The number of components and the components used may differ between boundary sections. At both ends of a section the same set of components must be defined. Between sections these sets may differ. Example: A model with 3 open boundary sections (with astronomical boundary conditions). The input data is printed in Courier; comment (not in the file) is printed between brackets.
East bound A A0 0.02 0.0 M2 1.87 314.3 S2 0.32 276.4 O1 0.21 14.3 East bound B A0 0.03 0.0 M2 1.89 264.7 S2 0.29 220.9 O1 0.19 38.3 West bound A A0 0.06 0.0 M2 1.71 122.5 S2 0.18 46.4 West bound B A0 0.06 0.0 M2 1.69 110.3 S2 0.19 22.4 Sea bound A A0 0.07 0.0 M2 1.67 300.9 S2 0.32 76.2 K1 0.05 33.1 Sea bound B A0 0.07 0.0 M2 1.69 324.1 S2 0.29 110.1 K1 0.09 6.1 {section name, end point A} {mean value} {component name, amplitude and phase}

{section name, end point B} {mean value} {component name, amplitude and phase}

{section name, end point A} {mean value} {component name, amplitude and phase} {section name, end point B} {mean value} {component name, amplitude and phase} {section name, end point A} {mean value} {component name, amplitude and phase}

{section name, end point B} {mean value} {component name, amplitude and phase}

Deltares

439

Delft3D-FLOW, User Manual A.2.12 Astronomic correction factors File contents The file contains corrections to the astronomical components of the open boundary sections with data type astronomic. These corrections may be applied during calibration. This file avoids a large processing effort of your tidal data. Just specify the appropriate boundary section and the component(s) that you wish to alter and how much it needs be changed (Data Group Boundaries - Flow conditions). The amplitude factor is a multiplicative factor and the phase factor is an additive factor. Filetype ASCII File format Fix formatted for text variables, free formatted for real and integer values. Filename <name.cor> Generated FLOW-GUI Record description: Record 1 2 to 1+NCOMR Record description Label for begin point (12 characters). Component name (8 characters), amplitude and phase correction factors (2 reals). Label for end point (12 characters). Component name (8 characters), amplitude and phase correction factors (2 reals).

2+NCOMR 3+NCOMR to 2+2∗NCOMR where: NCOMR

the number of tidal components that requires corrections.

The records (1 to 2+2∗NCOMR) may be repeated for the number of open boundary sections. Restrictions: The name of the label may not contain blanks between non-blnak characters. Label names for the begin and end points are defined in the <name.bnd> file. All component names are written in upper case. The label and component name must start in record position one. The number of components and the components used may differ per boundary section. The astronomical component A0 cannot be corrected. Example: Model for which two open boundary sections, with astronomical boundary conditions, need corrections. The input data is printed in Courier; comment (not in the file) is printed between brackets.

440

Deltares

Files of Delft3D-FLOW
East bound A M2 0.90 10.0 S2 0.95 -7.5 East bound B M2 0.90 10.0 S2 0.95 -7.5 Sea bound A M2 0.95 7.0 S2 0.90 -3.0 Q1 1.10 15.0 K1 1.10 10.0 Sea bound B M2 0.95 7.0 S2 0.90 -3.0 Q1 1.10 15.0 K1 1.10 10.0 {section name, end point A} {component name, amplitude and phase} {section name, end point B} {component name, amplitude and phase} {section name, end point A} {component name, amplitude and phase}

{section name, end point B} {component name, amplitude and phase}

Remark: In the example the correction factors are the same for both end points; but this is not mandatory.

A.2.13

Harmonic flow boundary conditions File contents The frequencies, amplitudes and phases for all open boundary sections with data type equal to H(armonic) (Data Group Boundaries Flow conditions). Filetype ASCII File format Free formatted Filename <name.bch> Generated FLOW-GUI or manually offline Record description: Record 1 2 3 to NTOH+2 Record description Frequencies (including 0.0 for the the mean value) (reals). Blank Amplitudes at the begin of each boundary section for all frequencies (reals). to Amplitudes at the end of each boundary section for all frequencies (reals). Blank to Phases at the begin of each boundary section for all frequencies (blanks for mean value) (reals). Phases at the end of each boundary section for all frequencies (blanks for mean value) (reals).

NTOH+3 2∗NTOH+2 2∗NTOH+3 2∗NTOH+4 3∗NTOH+3 3∗NTOH+4 4∗NTOH+3 where:

to

Deltares

441

Delft3D-FLOW, User Manual NTOH number of open boundary sections driven with harmonic frequencies.

Restrictions: Maximum record length is 132. The input items will not be checked on their domain. Input items in a record are separated by one or more blanks. Remark: The phase values at intermediate points are interpolated from the values specified at the begin and end of the opening section. You should take care for a good representation of phases at transition points (e.g. an interpolation between 356 and 13 degrees). Example: Model area with 3 open boundary sections (NTOH = 3) with H data type and two harmonic frequencies including the mean value (which has frequency 0.0). The input data is printed in Courier; comment (not in the file) is printed between brackets.
0.0 -1.6 -1.5 1.6 -1.7 -1.7 1.8 30.0000 2.1 2.3 2.2 2.2 2.1 2.2 120.5 79.6 245.8 125.6 88.9 283.7 {KC = 2, including zero frequency for the mean value} {blank record} {amplitudes at end A, section 1} {amplitudes at end A, section 2} {amplitudes at end A, section 3} {amplitudes at end B, section 1} {amplitudes at end B, section 2} {amplitudes at end B, section 3} {blank record} {phase at end A, 1-st component, section 1} {phase at end A, 1-st component, section 2} {phase at end A, 1-st component, section 3} {phase at end B, 1-st component, section 1} {phase at end B, 1-st component, section 2} {phase at end B, 1-st component, section 3}

A.2.14

QH-relation flow boundary conditions File contents QH relations at water level boundaries for boundary sections with the data type Q (Data Group Boundaries - Flow conditions). File format Fix format for header information; free format for time series data. Filename <name.bcq> Generated FLOW-GUI or manually offline Record description: For each water level boundary segment with data type Q a data block must be prescribed consisting of: Header records containing a number of compulsory and optional keywords accompanied by their values. A set of records containing the discharge/water level data. Each record contains a discharge in [m3 /s] and a water level in [m]. Restrictions: 442 Deltares

Files of Delft3D-FLOW Maximum record length is 5000. Position, format of the keywords and the format of keyword-values in the header are fixed (see example). All keywords have a length of 20 characters. Header in each block must be ended with the (compulsory) keyword: ‘records in table’ accompanied by the number of data records to follow. Positive discharges indicate flow in positive M/N direction. If the model flows in negative M/N direction negative discharges must be specified. Discharges must be specified in increasing order. That is, Q = 100.0 should be specified before Q = 200.0. For negative discharges Q = −200.0 must be specified before Q = −100.0. QH boundaries should only be specified as outflow boundaries. The sequence of blocks must be consistent with the sequence of water level boundary sections with data type Q. Example: Model with open boundary sections with data type Q. Flow in negative M or N direction.
table-name contents location1 xy-function interpolation2 parameter parameter records in table -2000.0000 -1500.0000 -1000.0000 -500.0000 -300.0000 Boundary Section : 2 uniform open boundary number 2 ’equidistant’ linear ’total discharge (t) ’ unit ’[m**3/s]’ ’water elevation (z) ’ unit ’[m]’ 5 2.1455 1.7711 1.3516 0.8514 0.6057
a20,1x,a40 a20,1x,a20 a20,1x,a20 a20,1x,free a20,1x,a20,1x,a20,free a20,1x,a20,1x,a20,free a20,1x,i6

A.2.15

Time-series flow boundary conditions File contents Time-series for flow boundary conditions of all open boundary sections with data type T (Data Group Boundaries - Flow conditions). Filetype ASCII File format Fix format for header information; free format for time-series data. Filename <name.bct> Generated FLOW-GUI, program Delft3D-NESTHD or manually offline Record description: For each open boundary segment with boundary data of type T (time-series) the data is given in two related blocks: 1 A header block containing a number of compulsory and optional keywords accompanied by their values. 2 A data block containing the time dependent data.
1 2

sequence must follow sequential order of open boundaries in pre-processor Optional Keywords and values

Deltares

443

Delft3D-FLOW, User Manual Description header block: Record 1 2 Text table-name contents Format a20, 1x a20,1x Value ’Boundary Section: ##’ ’Logarithmic ’ ’Uniform ’ ’3d-profile ’ ’name boundary section’ (see <name.bnd> file) ’non-equidistant’ yyyymmdd (must be equal to itdate) ’minutes’ ’linear’ ’parameter name and location’, ’layer and location’, units ’[ccc]’ Format a a

3 4 5 6 7 8 to 8+NPAR+1

location time-function reference-time time-unit interpolation parameter

a20, 1x a20, 1x a20, 1x a20, 1x a20, 1x a20, 1x

a a i8 a a a, ’units’, a integer

10+NPAR

records-in-table

a20, 1x

number of records in the data block

Remark: NPAR is the number of parameters for which a time varying boundary condition is being specified. Description data block: Record each record Record description Time in minutes after the Reference Date and NPAR∗{2 of Kmax} values representing the parameter for which a time varying boundary condition is being specified (all reals). The number of values to be specified for each end point of a boundary section depends on the type of profile in the vertical: 2 real values for either a uniform or logarithmic profile Kmax∗2 real values for 3D-profile where the end points for each layer are prescribed.

Restrictions: Maximum record length is 5000. Position, format of the keywords and the format of keyword-values in the header are fixed (see example). All keywords have a length of 20 characters. Header in each block must be ended with the (compulsory) keyword: ‘records in table’

444

Deltares

The vertical profile is logarithmic. table-name ’Boundary Section : 1’ contents ’logarithmic ’ location ’East Boundary ’ time-function ’non-equidistant ’ reference-time 19941001 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’ ’ parameter ’flux/discharge (q) end A’ unit ’ ’ parameter ’flux/discharge (q) end B’ unit ’ ’ records in table 2 0. the time series is assumed to be non-equidistant and the time is given in minutes. the interpolation method linear. Example 1: Model with 2 open boundary sections with time-series as boundary conditions. The vertical profile is uniform.Files of Delft3D-FLOW accompanied by the number of data records to follow. i. All open boundaries that have ‘3D-profile’ must precede other open boundaries.0000 50000.16 for the corresponding bcc-file.60000 8000. i. the times specified will be checked on their domain.0 and 8 000. at 0. The computation is depth averaged.0 and 8 000.0000 1. one section is Riemann controlled and one section is water level controlled. the interpolation method linear.0 minutes after the Reference Date. Times must be multiples of the integration time step. The second boundary section concerns a current boundary for which the boundary condition is given at two time breakpoints.0 100000. table-name ’Boundary Section : 1’ Deltares 445 . one section is velocity controlled. The first boundary section concerns a discharge boundary for which the boundary condition is given at two time breakpoints.e. table-name ’Boundary Section : 2’ contents ’uniform ’ location ’West boundary ’ time-function ’non-equidistant ’ reference-time 19941001 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’current (c) end A’ unit ’[m/s]’ parameter ’current (c) end B’ unit ’[m/s]’ records in table 2 0. at 0. 8000.50000 1.0 133000.e.2. the time-series is assumed to be non-equidistant and the time is given in minutes.80000 Example 2: The second example concerns an open sea boundary consisting of three sections.0000 1.0000 75000.75000 1. See Section A.0 minutes after the Reference Date. The sequence of blocks must be consistent with the sequence of open boundary sections with data type T(ime).

104 ...... 3555...341 3570..........651 -0..000 0.......0 2.0 2..192 3600.783 table-name contents location time-function reference-time time-unit interpolation parameter parameter parameter records-in-table 840.... but now for a 3D computation with 5 layers in the vertical..0 1...116 ...120 0.0 -0.Delft3D-FLOW ..756 -0..0 0.0 0..589 -0.0 -0.......0 -0. 3585. table-name contents location time-function reference-time time-unit ’Nested flow BC west-vel-1 ’3d-profile ’ ’west-vel-1 ’ ’non-equidistant’ 19960718 ’minutes’ ’ 446 Deltares .. .808 .. 3555.0 0.197 2..186 2. User Manual contents location time-function reference-time time-unit interpolation parameter parameter parameter records-in-table 840.000 855..848 1..0 -0..301 table-name contents location time-function reference-time time-unit interpolation parameter parameter parameter records-in-table 840......16 for the bcc-file. See Section A............183 ’Uniform ’ ’west-vel-1 ’ ’non-equidistant’ 19960718 ’minutes’ ’linear’ ’time ’ unit ’[min]’ ’current (c) end A’ unit ’[m/s]’ ’current (c) end B’ unit ’[m/s]’ 185 ’Boundary Section : 2’ ’Uniform ’ ’west-vel-2 ’ ’non-equidistant’ 19960718 ’minutes’ ’linear’ ’time ’ unit ’[min]’ ’riemann (r) end A’ unit ’[m/s]’ ’riemann (r) end B’ unit ’[m/s]’ 185 ’Boundary Section : 3’ ’Uniform ’ ’west-wl ’ ’non-equidistant’ 19960718 ’minutes’ ’linear’ ’time ’ ’water elevation (z) end A’ ’water elevation (z) end B’ 185 unit ’[min]’ unit ’[m]’ unit ’[m]’ Example 3: The third example concerns the same open boundary of the second example..000 855..845 3570..000 0..2..858 855.0 1.815 -0..113 0..899 1.0 0...

..557 -.0 -.101 ...........000 .. 3585.104 .403 -..177 Deltares 447 .899 1.365 3600.109 .450 -.103 ....000 ..766 -.0 .0 1.454 -......124 .651 -.... 3585.0 -..636 -..491 -...409 -..000 ..000 .....349 -..705 -.454 -.0 1.Files of Delft3D-FLOW interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’current (c) end A layer: 1’ unit ’[m/s]’ parameter ’current (c) end A layer: 2’ unit ’[m/s]’ parameter ’current (c) end A layer: 3’ unit ’[m/s]’ parameter ’current (c) end A layer: 4’ unit ’[m/s]’ parameter ’current (c) end A layer: 5’ unit ’[m/s]’ parameter ’current (c) end B layer: 1’ unit ’[m/s]’ parameter ’current (c) end B layer: 2’ unit ’[m/s]’ parameter ’current (c) end B layer: 3’ unit ’[m/s]’ parameter ’current (c) end B layer: 4’ unit ’[m/s]’ parameter ’current (c) end B layer: 5’ unit ’[m/s]’ records-in-table 185 840...000 ....804 -....374 -..447 -.583 -........000 ....339 -. 3585...185 2.112 ..119 ..788 -.999 .000 855.000 .118 .286 table-name ’Nested flow BC west-wl ’ contents ’uniform ’ location ’west-wl ’ time-function ’non-equidistant’ reference-time 19960718 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’water elevation (z) end A ’ unit ’[m]’ parameter ’Water elevation (z) end B ’ unit ’[m]’ records-in-table 185 840..856 855.641 -.000 855.776 -.115 .682 -....000 .109 .....000 .390 -..000 .107 .0 ..524 -.....0 ...0 2.529 -.000 ...000 .106 ...000 .000 ..0 -..000 .883 1...736 -.422 -.000 ...407 -..857 .109 .......0 -..........668 -...634 -.102 ......549 -............186 2...126 .....000 .181 3600..113 ......691 -......728 -..0 ...444 -.0 2.415 -.106 ..841 3600....612 -.104 .119 .741 table-name ’Nested flow BC west-vel-2 ’ contents ’3d-profile ’ location ’west-vel-2 ’ time-function ’non-equidistant’ reference-time 19960718 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’current (c) end A layer: 1’ unit ’[m/s]’ parameter ’current (c) end A layer: 2’ unit ’[m/s]’ parameter ’current (c) end A layer: 3’ unit ’[m/s]’ parameter ’current (c) end A layer: 4’ unit ’[m/s]’ parameter ’current (c) end A layer: 5’ unit ’[m/s]’ parameter ’current (c) end B layer: 1’ unit ’[m/s]’ parameter ’current (c) end B layer: 2’ unit ’[m/s]’ parameter ’current (c) end B layer: 3’ unit ’[m/s]’ parameter ’current (c) end B layer: 4’ unit ’[m/s]’ parameter ’current (c) end B layer: 5’ unit ’[m/s]’ records-in-table 185 840...499 -.000 .

2. 1x a20.Transport conditions). program Delft3D-NESTHD or manually offline Record description: For each open boundary segment the data is given in two related blocks: 1 A header block containing a number of compulsory and optional keywords accompanied by their values. units ’[ccc]’ number of records in the data block Format a a 3 4 5 6 7 8 to 8+NPAR+1 location time-function reference-time time-unit interpolation parameter a20. 1x a20. 2 A data block containing the time dependent data. Description header block: Record 1 2 Text table-name contents Format a20. Description data block: 448 Deltares .bnd> file ’non-equidistant’ yyyymmdd (must be equal to itdate) ’minutes’ ’linear’ ’parameter name and location’.bcc> Generated FLOW-GUI. 1x a20. Filetype ASCII File format Fix format for header information. a integer 10+NPAR records in table a20. ’layer and location’. 1x a20. User Manual A.1x Value ’Boundary Section : ##’ ’Logarithmic ’ ’Uniform ’ ’Step ’ ’3d-profile ’ ’name boundary section’ (see <name. 1x a a i8 a a a.Delft3D-FLOW. 1x a20. 1x Remark: NPAR is the number of parameters for which a time-varying boundary condition is being specified. ’units’. free format for time-series data Filename <name.16 Time-series transport boundary conditions File contents Time-series for transport boundary conditions of all open boundary sections (Data Group Boundaries . 1x a20.

Header in each block must be ended with the (compulsory) keyword: ‘records in table’ accompanied by the number of data records to follow. Not all values of other parameters from the file will be checked on their domain. A 3D simulation (KMAX >1) has been specified. For a boundary section called East BOUNDARY. All open boundaries that have ‘3D-profile’ must precede other open boundaries. The number of values to be specified for each end point of a boundary section depends on the type of profile in the vertical: uniform: 2 real values for each parameter specified linear: 4 real values for each parameter specifying the parameter in the surface and in the bottom layer for both section end points.e.5 metres below the surface and a constituent called ’test constituent’ has a Uniform profile in the vertical. at 0. Times must be multiples of the integration time step. Salinity and one constituent are included in the simulation. Position. 4 reals for each parameter specifying the parameter in the surface and in the bottom layer for both section end points and the location of the discontinuity/jump in the vertical in metres below the surface 3D-profile: Kmax∗2 real values for each layer in both end points of the boundary section being prescribed. the times specified will be checked on their domain.0 and 8000. i. Example 1: Model area with two open boundary sections.e.0 minutes after midnight on the Reference Date. i. at 0. The boundary condition is specified as follows: For a boundary section called West BOUNDARY.0 and 8000. table-name contents location time-function reference-time time-unit ’T-series BC for process run: 123’ ’step’ ’West BOUNDARY’ ’non-equidistant’ 19941001 ’minutes’ Deltares 449 . the profiles in the vertical are prescribed as functions.Files of Delft3D-FLOW Record each record Record description Time in minutes after the Reference Date and NPAR values representing the parameter for which a time varying boundary condition is being specified (all reals). format of the keywords and the format of keyword-values in the header are fixed (format in example).0 minutes after the Reference Date 00:00:00 hr. Restrictions: Maximum record length is 5000. step: 5 real values. The sequence of blocks must be consistent with the sequence of open boundary sections and the sequence of constituents. The boundary conditions are given for two time breakpoints. All keywords have a length of 20 characters. the salinity has a ‘Step’ profile with the discontinuity occurring at 2. the salinity has a linear profile in the vertical and a constituent called ’test constituent’ has a Uniform profile in the vertical. The boundary conditions are given for two time breakpoints.

60000 8000.000 35.2000 table-name ’T-series BC for flow run: 123’ contents ’uniform’ location ’East BOUNDARY’ time-function ’non-equidistant’ reference-time 19941001 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’test constituent end A ’ unit ’[-]’ parameter ’test constituent end B ’ unit ’[-]’ records in table 2 0.75000 1.2000 2.2.000 34.0000 32.5000 8000.0000 36.50000 1.0000 32.80000 table-name ’T-series BC for process run: 123’ contents ’linear’ location ’East BOUNDARY’ time-function ’non-equidistant’ reference-time 19941001 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’salinity end A surface’ unit ’[ppt]’ parameter ’salinity end A bed ’ unit ’[ppt]’ parameter ’salinity end B surface’ unit ’[ppt]’ parameter ’salinity end B bed ’ unit ’[ppt]’ records in table 2 0.0000 36.15 for the corresponding bct-file.75000 1.000 35.5000 table-name ’T-series BC for flow run: 123’ contents ’uniform’ location ’West BOUNDARY’ time-function ’non-equidistant’ reference-time 19941001 time-unit ’minutes’ interpolation ’linear’ parameter ’time ’ unit ’[min]’ parameter ’test constituent end A ’ unit ’[-]’ parameter ’test constituent end B ’ unit ’[-]’ records in table 2 0.0000 1.50000 1. table-name contents location ’Boundary Section : 1’ ’Uniform ’ ’west-vel-1 ’ 450 Deltares .0000 8000.7000 34.0000 2.80000 Example 2: The second example concerns an open sea boundary consisting of three sections for which the salinity concentrations are prescribed.0000 1.0000 35.60000 8000.0000 1.7000 34.0000 33. User Manual parameter ’time ’ unit ’[min]’ parameter ’salinity end A surface’ unit ’[ppt]’ parameter ’salinity end A bed ’ unit ’[ppt]’ parameter ’salinity end B surface’ unit ’[ppt]’ parameter ’salinity end B bed ’ unit ’[ppt]’ parameter ’discontinuity ’ unit ’[m]’ records in table 2 0.0000 33.0000 36.Delft3D-FLOW . The computation is depth averaged.0000 36.0000 35. see Section A.0000 1.000 34.

..8 3585...3 17.0 20.2 21..8 ....0 20.. 3555.2 21..0 21.0 21..5 3600..0 20.6 900.8 885.0 885..8 3570.0 20. 3555..9 855.0 21.1 17.0 20....7 12..4 900.2 3570.2 855.....0 20.0 3585.6 22..9 870.8 900.2 17.0 22.1 870.5 .1 21.0 16.0 20..0 20..0 23..0 16.4 3570.2 3585.0 20.8 3600.9 21... 3555.7 21..0 16....8 21...9 table-name contents location time-function reference-time time-unit interpolation parameter parameter parameter records-in-table 840..4 ..9 22.0 20.4 21.7 13.0 21.0 23.9 23...0 23.5 21...0 22..0 20.9 ’non-equidistant’ 19960718 ’minutes’ ’linear’ ’time ’Salinity ’Salinity 185 ’ end A uniform’ end B uniform’ unit ’[min]’ unit ’[ppt]’ unit ’[ppt]’ ’Boundary Section : 2’ ’Uniform ’ ’west-vel-2 ’ ’non-equidistant’ 19960718 ’minutes’ ’linear’ ’time ’ ’Salinity end A uniform’ ’Salinity end B uniform’ 185 unit ’[min]’ unit ’[ppt]’ unit ’[ppt]’ ’Boundary Section : 3’ ’Uniform ’ ’west-vel-3 ’ ’non-equidistant’ 19960718 ’minutes’ ’linear’ ’time ’ ’Salinity end A uniform’ ’Salinity end B uniform’ 185 unit ’[min]’ unit ’[ppt]’ unit ’[ppt]’ Deltares 451 .9 22.9 23.0 16.7 13..0 20.1 17.0 21..9 23...8 13.