You are on page 1of 96

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/331984335

APPLICATION OF TWO-DIMENSIONAL HYDRAULIC MODELING IN RIVERINE


SYSTEMS USING HEC-RAS

Thesis · May 2017

CITATIONS READS

7 2,507

1 author:

Abdulaziz Alzahrani
University of Akron
2 PUBLICATIONS   7 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Two-Dimensional Hydraulic Modeling View project

All content following this page was uploaded by Abdulaziz Alzahrani on 25 March 2019.

The user has requested enhancement of the downloaded file.


APPLICATION OF TWO-DIMENSIONAL HYDRAULIC MODELING IN RIVERINE

SYSTEMS USING HEC-RAS

Thesis

Submitted to

The School of Engineering of the

UNIVERSITY OF DAYTON

In Partial Fulfillment of the Requirements for

The Degree of

Master of Science in Civil Engineering

By

Abdulaziz Saeed Alzahrani

UNIVERSITY OF DAYTON

Dayton, Ohio

May, 2017
APPLICATION OF TWO-DIMENSIONAL HYDRAULIC MODELING IN RIVERINE

SYSTEMS USING HEC-RAS

Name: Alzahrani, Abdulaziz Saeed

APPROVED BY:

_________________________ _______________________
Donald V. Chase, Ph.D., P.E. Catherine Kublik, Ph.D.
Advisory Committee Chairman Committee Member
Visiting Professor Assistant Professor
Department of Civil and Environmental Department of Mathematics
Engineering and Engineering Mechanics

________________________
Shuang-ye Wu, Ph.D.
Committee Member
Associate Professor
Department of Geology

_______________________________ _________________________
Robert J. Wilkens, Ph.D., P.E. Eddy M. Rojas, Ph.D., M.A., P.E.
Associate Dean for Research and Innovation Dean, School of Engineering
Professor
School of Engineering

ii
ABSTRACT

APPLICATION OF TWO-DIMENSIONAL HYDRAULIC MODELING IN RIVERINE

SYSTEMS USING HEC-RAS

Name: Alzahrani, Abdulaziz Saeed


University of Dayton

Advisor: Dr. Donald V. Chase

Flood simulation models have a wide variety of approaches that are available to

compute the water surface elevations associated with a flood event. Some of these

models use a One-dimensional (1D) approach, others use a Two-dimensional (2D), and

there are others allow the use of integrated 1D and 2D simulations. In 2015, the US Army

Corps of Engineering Hydrologic Engineering Center (HEC) released HEC-RAS Version

5.0.3 which performs 1D steady and unsteady flow calculation, as well as 2D unsteady

flow calculation. 2D modeling is likely to become more common due to HEC-RAS. This

paper is the comparison of 1D and 2D models for a water level in the channel and the

floodplain inundation. Two treatments of the problem are explored: (1) a 1D model based

upon discretization of floodplain units into storage areas; (2) a 2D model for a channel

and a floodplain surface. The two models were tested on the Great Miami River and Bear

Creek. The models were assessed by comparison with measured inundation extent. Also,

the study reviews the academic basis for modeling floodplain flow based on a two-

dimensional analysis.

iii
The goal of the proposed research effort will be to include the application of 2D

unsteady-state models in future releases of the HEC-RAS 2D model.

iv
DEDICATION

To my parents, my wife, my kids, and the rest of my family.

v
ACKNOWLEDGMENTS

First, I would like to thank my advisor, Dr. Donald V. Chase, for his patience and

persistent support throughout this study. Without him, this research would not have been

possible.

The data for this study was provided by Ohio Geographically Referenced

Information Program (OGRIP) and Montgomery County, Ohio. This study would not

have been possible without this data.

vi
TABLE OF CONTENTS

ABSTRACT ....................................................................................................................... iii


DEDICATION .................................................................................................................... v
ACKNOWLEDGMENTS ................................................................................................. vi
LIST OF FIGURES ........................................................................................................... ix
LIST OF TABLES ............................................................................................................. xi
ABBREVIATIONS .......................................................................................................... xii
CHAPTER 1 INTRODUCTION ........................................................................................ 1
Background: ........................................................................................................................ 1
Purpose of Study: ................................................................................................................ 2
Scope of Report: ................................................................................................................. 3
CHAPTER 2 LITERATURE REVIEW ............................................................................. 4
Brief History of Floodplain:................................................................................................ 4
HEC-RAS History: ............................................................................................................. 6
One-Dimensional Steady and Unsteady Modeling: ............................................................ 7
Two-Dimensional Unsteady Modeling: .............................................................................. 8
Brief Cases Studies of 1D and 2D Models: ...................................................................... 12
CHAPTER 3 DEVELOPING AND SOLVING THE FINITE VOLUME EQUATION. 18
Introduction: ...................................................................................................................... 18
The Cell Vertex: ................................................................................................................ 18
Generalized Finite Volume Formulation: ......................................................................... 20
Application of Finite Volume Method to 2D Floodplain Modeling: ................................ 22
CHAPTER 4 DESCRIPTION OF STUDY AREA .......................................................... 27
Introduction: ...................................................................................................................... 27
Description of Study Area: ............................................................................................... 27
Light Detection and Ranging Elevation Data: .................................................................. 29

vii
Create Terrain Layer with Actual Channel: ...................................................................... 29
Flow Data: ......................................................................................................................... 30
Roughness and Cross-sectional: ....................................................................................... 31
CHAPTER 5 APPLICATION OF 1D AND 2D UNSTEADY STATE MODEL ........... 34
Introduction: ...................................................................................................................... 34
Terrain Layers: .................................................................................................................. 34
Creating 2D Flow Area Mesh Only: ................................................................................. 37
Land Use Classification: ................................................................................................... 38
1D Unsteady State with Storage Area Model: .................................................................. 39
Flood Inundation Map: ..................................................................................................... 42
CHAPTER 6 RESULTS ................................................................................................... 48
Introduction: ...................................................................................................................... 48
WSEL Comparison Between 1D and 2D Model: ............................................................. 48
The Inundation Area and the WSEL In Floodplain Area: ................................................ 52
Findings: ........................................................................................................................... 59
CHAPTER 7 CONCLUSION AND RECOMMENDATIONS ....................................... 60
Conclusion: ....................................................................................................................... 60
Recommendations: ............................................................................................................ 60
REFERENCES ................................................................................................................. 62
APPENDIX A ................................................................................................................... 66
APPENDIX B ................................................................................................................... 78

viii
LIST OF FIGURES

Figure 1: Study Area Location. ........................................................................................... 3


Figure 2: Cross-Section View of a Natural Channel Characteristics. ................................. 5
Figure 3: An Unsteady Flow Routing Concept (Chase, 2016). .......................................... 8
Figure 4: Grid Cells and GIS Cells. .................................................................................... 9
Figure 5: Ground Geometry at Cell Face. ......................................................................... 10
Figure 6: Cell Vertex Method and Secondary Mesh. ....................................................... 19
Figure 7: Dual Cell Finite Difference and Volume Formulation (Brunner,2016). ........... 20
Figure 8: Unit Normal ṅ Components. ............................................................................. 21
Figure 9: Cartesian Mesh. ................................................................................................. 23
Figure 10: Cell Centered Structured Finite Volume Mesh. .............................................. 25
Figure 11: A Portion of Moraine, West Carrollton and Farm Located in Floodplain. ..... 28
Figure 12: (A) Original Terrain. (B) A New Terrain with Channel Data.
(C) Combination Between Two Terrain Model and the New Created Terrain. .............. 30
Figure 13: RAS Mapper with a Terrain Data Layers. ....................................................... 35
Figure 14: RAS Mapper with Terrain and Geometry Layers. .......................................... 36
Figure 15: Example of a Trimmed Cross-Section. ........................................................... 36
Figure 16: Creating 2D Flow Area Mesh and Break-Lines. ............................................. 38
Figure 17: GMR Model Layout for the Proposed of Floodplain Storage Area. ............... 40
Figure 18: Bear Creek Model Layout for the Proposed of Floodplain Storage Area. ...... 41
Figure 19: A Flow Hydrograph for Great Miami River and Bear Creek. ......................... 42
Figure 20: Unsteady Flow Analysis Window. .................................................................. 43
Figure 21: 2D Model Flow Runs Through Great Miami River and Bear Creek n=0.035. 44
Figure 22: 2D Model Flow Overtopping to the Floodplain Area n=0.035. ...................... 45
Figure 23: 1D Model Flow Runs Through Great Miami River and Bear Creek n=0.035. 47
Figure 24: 1D Model Flow Overtopping to Storage Area n=0.035. ................................. 47
Figure 25: Maximum WSEL Profile Along GMR Computed with 1D and 2D Model. ... 49
Figure 26: Maximum WSEL Profile Along Bear Creek Computed with 1D and 2D
Model. ............................................................................................................................... 51
Figure 27: (A) Create a Storages Areas. (B) Create the Weirs or Laterals Structures. .... 53
Figure 28: (A) Concepts of 1D Model. (B) Concepts of 2D Model. ................................ 54
Figure 29: A Comparison Inundation Map in GMR n=0.035. ......................................... 54
Figure 30: A Comparison Inundation Map in GMR n=0.06. ........................................... 55
Figure 31: A Comparison Inundation Map in GMR n=0.1. ............................................. 56
Figure 32: A Comparison of Inundation Map in Bear Creek n=0.035. ............................ 57
Figure 33: A Comparison of Inundation Map in Bear Creek n=0.06. .............................. 58
Figure 34: A Comparison of Inundation Map in Bear Creek n=0.1. ................................ 58
ix
Figure 35: The Cell (A) That HEC-RAS Will Compute the WSEL Example 1. ............. 66
Figure 36: Cell Volume Elevation Relationship Example 1. ............................................ 69
Figure 37: The Cell (A) That HEC-RAS Will Compute the WSEL Example 2. ............. 70
Figure 38: Cell Volume Elevation Relationship Example 2. ............................................ 73
Figure 39: The Cell (A) That HEC-RAS Will Compute the WSEL Example 3. ............. 74
Figure 40: Cell Volume Elevation Relationship Example 3. ............................................ 77
Figure 41: The 25 Cross-Sections Location in 1D and 2D Model. .................................. 78
Figure 42: Large Difference of WSEL Between 1D and 2D Model n=0.035. ................. 79
Figure 43: Average Difference of WSEL Between 1D and 2D Model n=0.035. ............. 79
Figure 44: Minimum Difference of WSEL Between 1D and 2D Model n=0.035. .......... 80
Figure 45: Largest Difference of WSEL Between 1D and 2D Model n=0.06. ................ 80
Figure 46: Average Difference of WSEL Between 1D and 2D Model n=0.06. ............... 81
Figure 47: Minimum Difference of WSEL Between 1D and 2D Model n=0.06. ............ 81
Figure 48: Largest Difference of WSEL Between 1D and 2D Model n=0.1. .................. 82
Figure 49: Average Difference of WSEL Between 1D and 2D Model n=0.1. ................. 82
Figure 50: Minimum Difference of WSEL Between 1D and 2D Model n=0.1. .............. 83

x
LIST OF TABLES

Table 3.1: Flow Data Assigned for a Great Miami River and Bear Creek. ...................... 31
Table 6.1: A Differences of Maximum WSEL in GMR. .................................................. 51
Table 6.2: A Differences of Maximum WSEL in Bear Creek. ......................................... 52
Table 6.3: Comparison of WSEL in Floodplains Areas Around GMR. ........................... 56
Table 6.4: Comparison of WSEL in Farm Area Around Bear Creek. .............................. 57
Table A-1: WSEL for Such Cells at Time (t1) for Example 1. ......................................... 67
Table A-2: Computes the Discharge Through Each Face Example 1. ............................. 67
Table A-3: WSEL for Such Cells at Time (t1) for Example 2. ......................................... 71
Table A-4: Computes the Discharge Through Each Face Example 2. ............................. 71
Table A-5: WSEL for Such Cells at Time (t1) for Example 3. ......................................... 75
Table A-6: Computes the Discharge through Each Face Example 3................................ 75

xi
ABBREVIATIONS

1D One-Dimensional

2D Two-Dimensional

GMR Great Miami River

FEMA Federal Emergency Management Agency

USACE United States Army Corps of Engineers

HEC-RAS Hydrologic Engineering Center-River Analysis System

GIS Geographic Information System

FVM Finite Volume Method

WSEL Water Surface Elevation Level

LiDAR Light Detection and Ranging

OGRIP Ohio Geographically Referenced Information Program

USGS U.S. Geological Survey

ODNR Ohio Department of Natural Resources

DEM Digital Elevation Model

XS Cross-Section

xii
CHAPTER 1

INTRODUCTION

Background:

Hydraulic modeling and flood inundation mapping are performed to provide

important information from a flood event including the level of inundation and water

surface elevations within the study area. A hydraulic simulation model is a mathematical

representation of the physical hydraulic processes that occur during a flood event. Such

processes can be described by Conservation of Mass, Conservation of Momentum, and

Conservation of Energy equations posed in either one, two or three dimensions.

As one might expect, as the dimensionality of the problem increases the complexity

associated with solving the problem also increases. As a result, many different

simplifications and assumptions have been made to create models capable of providing

suitable accuracy without requiring a large amount of computing power or input data. A

river system is a combination of the main river channel and adjacent floodplain areas.

When the water surface elevations of a waterway during a flood event exceeds the depth

of the main channel, then the flow expands into floodplains. 1D or 2D models with

steady and unsteady-state assumptions are used to simulate high flows. 2D models use

the terrain as a continuous surface, whereas 1D models only consider the river and

floodplain geometry at discrete locations along the length of the channel. The use of

continuous terrain enables two-dimensional models to more

1
accurately describe the lateral interaction of flow between the main channel and the

floodplain as well as the storage effects associated with floodplain areas.

In 1D hydraulic modeling, the flow is assumed to move in the longitudinal

direction only, that is, downstream. The terrain in a1D model is represented as a system

of cross-sections and the results are an estimate of the average velocity and water depth at

each cross-section. There are some physical constraints in a 1D model that can be

overcome in a 2D model namely simultaneous flow in both the longitudinal and lateral

directions.

Purpose of Study:

The objective of the study documented in this thesis is to demonstrate two-

dimensional modeling using the 2D modeling feature available in HEC-RAS Version 5.0.

The 2D modeling feature is new to HEC-RAS. Members of the thesis advisory committee

feel that including 2D modeling in HEC-RAS will only promote its wide-spread use

possibly in the very near future. Thus, this thesis provides the author with the opportunity

to gain a greater understanding of the theoretical basis and practical application of two-

dimensional floodplain modeling. Also, the objectives of this thesis are accomplished by

comparing the results of the one-dimensional unsteady state and the floodplains modeling

as storage areas, with the two-dimensional model by using HEC-RAS (Hydrologic

Engineering Center – River Analysis System) in cross-sections and flood inundation.

This study uses LiDAR topographic datasets and different cross-sectionals. Both one-

dimensional and two-dimensional simulations are performed on a reach of the Great

Miami River (GMR) and the Bear Creek in Montgomery County, Ohio as shown in

Figure 1.

2
Figure 1: Study Area Location.

Scope of Report:

This thesis is organized in six chapters. Chapter 1 presents an introduction that

provides a background of some of the basic differences between 1D and 2D floodplain

modeling. Chapter 2 presents a literature review comprising a brief history of floodplain

modeling, history of HEC-RAS and an overview of steady and unsteady state condition

of 1D and 2D modeling. Chapter 3 describes how the fundamental equations of

Conservation of Mass and Conservation of Momentum are developed and solved using

the solution methodology employed by HEC-RAS. Chapter 4 presents the study area and

data. In this section, important information for geometric data as well as the flow data

and roughness is presented. Chapter 5 presents the methodology used to apply the model

of a 2D in unsteady-state flow and result. In Chapter 6 a comparison is made between the

1D unsteady with storage area model and the 2D model. Finally, Chapter 7 presents a

conclusion and recommendation.

3
CHAPTER 2

LITERATURE REVIEW

Brief History of Floodplain:

On April 1, 1979, President Jimmy Carter signed legislation that established the

Federal Emergency Management Agency (FEMA). FEMA’s mission is to protect and

serve the American people from natural and man-made disasters including flood events.

Floods are one of the most common natural disasters affecting humankind. In fact, of the

44 Major Disaster Declarations issued by FEMA during the first 10 months of 2016, 60%

of the events were considered a Flooding type of disaster (FEMA, 2016). Clearly

protecting the public from the dangers of floods is of paramount concern to civil

engineers.

Floodplains are geographic areas located adjacent to rivers and streams that carry

large volumes of water during high flow events. As illustrated in Figure 2, when water

enters a floodplain, it can pose a safety and personal property risk to individuals who

reside in and structures that are located in the floodplain. Consequently, as part of

FEMA’s mission, the agency has rated these floodplains and has given them a zonal

designation describing their severity of risk. Zone designations are, low-to-moderate

risks, high risks, coastal areas, and undetermined risks.

One of the first widely used simulations models available to examine the hydraulic

characteristics of channels and adjacent floodplain area was HEC-2. HEC-2, written by

4
the U.S Army Corps of Engineers, was developed with the aim of calculating water

surface profiles for the purpose of floodplain management as well as to evaluate areas

that have the potential for flooding (Moore, 2010). The program also evaluates the effects

of improving water channels and building artificial embankments (Dunn, 2014).

Figure 2: Cross-Section View of a Natural Channel Characteristics.


The origin of HEC -2 goes back to 1964 when Bill Eichert started developing a

software package on river hydraulics. He wrote the software in WIZ, which is an older

version of BASIC, on a GE 225 system while he was employed at the Tulsa District of

the U.S. Army Corps of Engineers (USACE) (Dunn, 2014). The USACE Hydrologic

Engineering Center (HEC) officially released the first FORTRAN version of the HEC-

2program in 1966 under the name Backwater Any Cross Section (Moore, 2012).

The program was unique in that, it could calculate the surface profiles of irregular

channel cross sections, which was not so with previous backwater programs. As such, it

became the platform for the development of modern hydraulic computer engineering

software.

5
Before long, the program was improved, expanded and its name changed to HEC -2

Water Surface Profiles in 1968 (Dunn 1). HEC further improved the program by adding

new features that led to the consequent release of new versions in the years that followed.

The package became so successful that engineers around the globe adopted it as the

standard for open channel hydraulic computations, eventually leading to the migration of

HEC-2 to the microcomputer in 1984 by Alfred Montalvo (HEC, 1989). It was the first

HEC software adapted to the PC, and it facilitated the development of PC based support

programs such as the PLOT 2 and SUMPO. One of the major applications of HEC -2 was

in the National Flood Insurance Program, which employed the program to facilitate its

mapping procedures (FEMA, 2016). HEC has expanded its software offerings over the

years to include software development throughout Water Management and Water

Resource Systems (HEC, 1989).

HEC-RAS History:

HEC-RAS was the first computer program in a suite of Windows-based software

applications developed by the Hydrologic Engineering Center starting in the early 1990’s.

Released in July 1995, HEC-RAS is HEC’s flagship program, replacing HEC-2 as the

simulation program capable of performing calculations to determine water surface profile

computations.

The capabilities of HEC-RAS have grown significantly since its inception. The

earliest versions only permitted 1D steady-state analysis. Version 3.0, published in

January 2001, included 1D unsteady flow computations (Dewberry, 2016). The latest

version of HEC-RAS is Version 5.0 released in 2016 which permits two-dimensional

unsteady flow calculations.

6
One-Dimensional Steady and Unsteady Modeling:

Early versions of HEC-RAS had the capability to calculate water surface profiles for

steady-state gradually varied flow in channels. Steady-state conditions describe the case

where the discharge in the channel remains constant over time. Gradually varied flow

conditions represent the case where water surface elevations do not change appreciably

over distance. The water surface profiles in 1D steady-state are found by applying

Conservation of Energy from one cross-section to the next cross-section along the length

of the channel. This is known as the Standard Step Method.

It is unusual for natural channel flow to be steady or constant. Rather natural channel

flow is unsteady, meaning that the flow rate in the channel actually varies with time.

Unsteady flow routing is the process of determining depths and flows at various locations

along within a channel at various times. This means that velocity, discharge and depth are

functions of location (distance along the stream channel) and time. Figure 3 below

illustrates this condition. At time t there is a certain discharge and depth at Cross-Section

A. At time t+∆t, there can be a different discharge and depth at the same location.

Unsteady flow routing permits the depth and discharge at numerous locations to be found

at different points in time.

7
Figure 3: An Unsteady Flow Routing Concept (Chase, 2016).
1D unsteady flow routing solves the Saint-Venant equations. Barre de Saint-Venant

developed the Saint-Venant equations, also called the shallow water equations, in the late

1800’s. The Saint-Venant equations were developed from Conservation of Mass

(equation 2.1) and Conservation of Momentum (equation 2.2) applied to a small control

volume of fluid (Brunner, 2016).

𝑑𝐴𝑇 𝑑𝑄
+ −𝑞 =0 Equation 2.1
𝑑𝑡 𝑑𝑥

𝑑𝑄 𝑑𝑄𝑉 𝑑𝑧
+ + 𝑔𝐴 (𝑑𝑥 + 𝑆 𝑓 ) = 0 Equation 2.2
𝑑𝑡 𝑑𝑥

Where A is the area of cross-section, 𝑡 is a time, Q is the flow, X is the distance along

channel, q is a source or sink term, 𝑉 is a velocity of flow, g is the gravity acceleration,

𝑑𝑧
𝑆 𝑓 is the friction slope and is the water surface slope.
𝑑𝑥

Two-Dimensional Unsteady Modeling:

The fundamental concept underlying 2D modeling is to discretize the river and

adjacent floodplain areas into a collection of individual cells called grid cells, 2D flow

cells, or computational grid cells. Each grid cell contains elevation and roughness data to

represent the ground surface elevation and friction effects along the ground surface.

8
HEC-RAS uses the sub-grid bathymetry approach. With the sub-grid bathymetry

approach, each grid cell is composed of multiple GIS cells as shown in Figure 4. Each

GIS cell has a unique elevation. A collection of grid cells, it is actually a collection of

GIS cells that make up the terrain model. It is the terrain model that describes the

continuous ground geometry that is so critical when analyzing the two-dimensional

behavior of floodplains.

Figure 4: Grid Cells and GIS Cells.


The interface between two grid cells is called a cell face. The ground geometry at a

face is composed of the ground elevations found from the GIS cells. Hydraulically, a cell

face is the same as across-section as illustrated in Figure 5 below. The ground geometry

is known since it is provided by the GIS cell information. As a result, hydraulic

properties (cross-sectional area, wetted perimeter, hydraulic radius and conveyance) can

be computed for any water surface elevation. In addition, since the topography within a

grid cell is known, then a relationship between the storage volume in the grid cell and the

water surface elevation can be developed. This is known more commonly as a stage-

9
storage curve. A water surface elevation is computed at each grid cell for each point in

time. The size of the grid cells defines the resolution of the model and model results.

Figure 5: Ground Geometry at Cell Face.


If the user wishes, HEC-RAS permits a channel and its floodplain to be examined

using a 2D flow mesh only. HEC-RAS also has the capability to link 1D and 2D areas.

The 2D flow area elements can be connected to 1D elements in several ways. 1D and 2D

areas can be directly connected to one another at the downstream end or the upstream end

of a river reach; they can be connected via lateral structures, or they can be connected to

one another through storage areas (Brunner, 2016). A 2D flow area can be directly

connected to storage of water area behind dams and barriers dam or levee.

10
HEC-RAS will combine continuity and the Diffusion-Wave form of the

momentum equation to compute the water surface elevation at a point in time, i.e. WSEL

(t).

The unsteady differential form of the Mass Conservation (Continuity) equation is:

𝜕𝐻 𝜕(𝑢ℎ) 𝜕(𝑣ℎ)
+ + +𝑞 =0 Equation 2.3
𝜕𝑡 𝜕𝑥 𝜕𝑦

Where t is time, H is a water surface elevation, h is the water depth, q is a source or sink

term, and u and v are the velocity components in the X and Y direction. In vector form,

the Continuity equation takes the form:

𝜕𝐻
+ ∇. ℎ𝑉 + 𝑞 = 0 Equation 2.4
𝜕𝑡

Where V= (u,v) is the velocity and (∇) is the vector of the partial derivative given by ∇=

∂ 𝜕
( ∂x , 𝜕𝑦).

The Diffusion-Wave form of the Momentum Equation can jest regarded the barotropic

pressure gradient and bottom friction.

−𝑔∇𝐻 = 𝐶𝑓𝑉 Equation 2.5

Where g is gravity acceleration and Cf is the bottom friction. HEC-RAS use Manning’s

formula, the Diffusion-Wave Equation results:

𝑛2 𝑔⃒ 𝑉
−𝑔∇𝐻 = ( 4 )𝑉 Equation 2.6
𝑅3

Where n is the Manning’s Roughness Coefficient and R is Hydraulic Radius.

The velocity will have determined by a balance between barotropic pressure gradient and

bottom friction.
2
(𝑅(𝐻))3 (∇𝐻)
𝑉=− 𝑛 1 Equation 2.7
(∇𝐻)2

11
Now, the Diffusion Wave can be direct substitution in the Mass Conservation.
2
𝜕𝐻 𝑅(𝐻)3
− ∇. 1 ∇𝐻 + 𝑞 = 0 Equation 2.8
𝜕𝑡
𝑛 (∇𝐻)2

Note that the only unknown in Equation 2.6 is water surface elevation H. The

derivation show above can be found in the HEC-RAS Hydraulic Manual (Brunner, 2016).

In Chapter 3 we will provide more details of the specific solution methodology employed

by HEC-RAS.

Brief Cases Studies of 1D and 2D Models:

One-dimensional and two-dimensional hydrodynamic models in HEC-RAS can

be used to simulate floods in Riverine system. However, each modeling approach has its

own advantages and limitations. The main objective of 1D model is to compute water

surface elevation at locations of interest for a given boundary condition. The flow

underlying 1D can be steady and unsteady state condition. Main inputs to the model

include flow data, manning’s n values, cross-section geometry and reach length. Model

can handle bridges, weir flow and storage areas i.e. The computational procedure is based

on solution of the one-dimensional energy equation using the standard step method.

Nevertheless, 1D can use kinematic, diffusive, or fully dynamic, vertically integrated

mass and momentum equations the “Saint Venant” equations. The Solution of continuity

and momentum equations is based on an implicit finite difference scheme. The 1D model

just computes how deep that water is going to get. It does not determine the direction.

That is really down to the engineer to actually determine which direction that water's

going to go, which the 2D modeling can be very useful to actually determine which

direction the flow going to go. HEC-RAS 2D has a GIS interface and applies the finite

12
volume method to solve unsteady flow equations that describe the two-dimensional

(Brunner, 2016).

There are many of case studies using one and two dimensional models to show

the abilities of the model being used. Some of these case studies have been performed to

confirm results while some have been performed to develop flood maps of flood levels.

Here a case study designed to Compare of One-Dimensional and Two-Dimensional

hydrodynamic modeling approaches (Ahmad, Simonovic, 1999), a study performed on

the Red River valley in Manitoba and North Dakota is very flat with direction of flow

from south to north. The tool that used to build 1D modeling is MIKE 11 and MIKE 21

on a 25 meters’ resolution DEM. MIKE 11 is an implicit finite difference model for 1D

model based on the Saint-Venant equations, and MIKE 21 is a 2D implicit finite

difference scheme used for unsteady flows. The results show that the error between

observed and 2D modeled in peak water levels at different locations is in the range of 0.2

to 1.1 ft. In other hand, the error in peak water levels in MIKE 11 (1D) range between 0.0

to 0.3 ft. The difference between the peak observed and simulated discharges in 2D at the

floodway inlet is 6.4 % of recorded flow. However, the 2D models require more time

and computational resources to simulate same hydrological event. Also, 2D requires a

significant amount of additional data (especially topographic data) and time to set up and

run. The lack of this study is that the comparison between 1D with storage area model

and 2D model. This study just focused on the different results between 1D and 2D at

specific location and Two-dimensional models, compared with 1D models, require a

significant amount of additional data (especially topographic data) and time to set up and

run.

13
Recent research has been conducted on comparing one-dimensional models with

two dimensional models, including using HEC-RAS, LISFLOOD-FP and TELEMAC-

2D. This study performed on a 60 kilometer reach of the River Severn, UK (Horritt and

Bates, 2002). HEC-RAS is a 1D model solving unsteady state condition. LISFLOOD-FP

is a combined of 1D-2D model solving 1D unsteady state in the floodway and 2D in the

floodplain and TELEMAC-2D solves just the 2D shallow water equation. In this study, it

was shown that HEC-RAS and TELEMAC-2D are both capable of being calibrated

against discharge or inundated area while LISFLOOD-FP must be calibrated against an

independent inundated area to produce acceptable results.

Another case study that is done by The U.S. Geological Survey about

determination the estimated extent of flood inundation using 2-D hydrodynamic

modeling in the Blue River valley from flooding on the lower Blue River and from

Missouri River backwater (Kelly and Rydlund, 2005). The Blue River flows through of

Kansas City to its mouth at the Missouri River and has been a source of flood damage in

Kansas City for many years. Flooding in the Blue River Basin has caused millions of

dollars of damage and has resulted in the deaths of more than 26 people in the last 25

years. Finite Element Surface-Water Modeling System (FESWMS) was chosen to

simulate steady-state flood flows. The FESWMS simulations provided information as to

the needed design of hydraulic structures and produced flood inundation maps at 2-ft

water level intervals. The results show that difference between observed and 2D modeled

peak water levels at different locations is in the range 0.2 to 0.53 ft. The result of this

study is the flood inundation maps created represent a substantial increase in the

14
capability of public officials and residents to minimize flood deaths and damage in

Kansas City.

Similarly, a study performed on the J¨okuls´a ´a Fj¨ollum River in Northeast

Iceland, compared two models; a 1D hydraulic model using HEC-RAS with a 2D

hydraulic model using TELEMAC-2D for the simulation extreme glacial outburst floods

(Petteri Alho and Juha Aaltonen, 2007). This study compares 1D unsteady state with

levees with 2D flow area. Digital elevation model (DEM) was 25-meter resolution. The

results show that HEC-RAS was more rapid inundation of slack water areas in floodplain

and there are significant differences in flow depth between HEC-RAS and TELEMAC-

2D and the total inundation of 1D model was 89% of the 2D model.

Similarly, a study of a 6 kilometer reach of the River Wharf, UK compared

inundation extent from HEC-RAS with a two-dimensional diffusion wave model (Tayefi

et al., 2007). This study has a three model: 1D unsteady state without levee, 1D unsteady

state with storage area and 1D connection to 2D. The results from this study show that,

the manning’s n value is more sensitive in both of 1D models and the 2D model produces

the best results of floodplain inundation. Also, 1D unsteady state with storage area is only

7% less effective than the 2D model on other way was 93% of the 1D connect to 2D

model.

In another study done by (Chandranath Chatterjee et al., 2008) to investigate the

effectiveness of a proposed flood emergency storage area at the middle Elbe River,

Germany by comparison of 1D unsteady state with storage area model using MIKE 11

with 1D connection 2D model using MIKEFLOOD. The storage area is divided into two

basins by an already existing and the two basins are connected by a gate of 50 m width

15
and 2 m depth. The DEM grid size used for MIKEFLOOD is 50 m. The results of a

comparison between MIKE11 and MIKEFLOOD simulation runs shows that there is

absolutely no difference in the water level and discharge reduction in the Elbe River.

In New Zealand, a study done by (Ian G. Jowett and Maurice J. Duncan, 2011) to

compare the depth and velocity predictions made using 1D unsteady state model

(RHYHABSIM) with predictions made by two 2D model (River2D and Hydro2de). This

study was applied a 1D model and two 2D models to a section of Braided River called

(the Hurunui River) and compare measured and predicted water depths and velocities.

The results show that the 1D depth predictions and cross section profiles were very

similar to the observed values of the hydraulic variables but 2D has lower fit between the

observed and simulated. 1D was an overestimation of depth by 3–6% and an

underestimation of velocity by 10–11%. Hydro2de (2D) overestimated depths by 7–17%

and underestimated velocities by 2–12% and River2D underestimated depth by 0–8% and

underestimated velocities by (−3% and 4%).

Shallow water equation hastily developed in the last years and changing from 1D

to 2D. Also, there was a developed of knowledge of the hydraulic phenomena, and

developing in an availability of high-resolution topographic data. Here a case study done

by (Pierfranco Costabile et al., 2015) to realize how the flood mapping using LiDAR

DEM data can reduce the difference between 1D and 2D models. A study performed on

two rivers, namely Crati and Corace, in the Calabria, Italy. This study used ORSADEM

software to approach 1D model and 2D computational grid model is obtained using a

commercial software able to provide a TIN, the cross sections and 2D flow area of the

case studies were extracted using the 2m resolution DEM. Flood events with a return

16
period of 500 years, for both the case studies have a 20-min time step, was chosen to

simulate Unsteady-state flood flows. The first companion between 1D and 2D was the

water surface elevation at a railway bridge crossing the Corace River. However, the result

shows differences between the maximum water elevations computed by 1D and 2D

models are in the range -0.25 m to +0.25 m. However, there are 3 approaches to flood

prone areas mapped. First, 1D extending of cross sections in to the urban area (1D

unsteady state without levee); second, 1D approach was to limit the cross section to the

main channel and to evaluate the outflowing discharge considering the levees as broad-

crested weirs (1D unsteady state with levee). Third model is 2D model. The result is

Referring to the 2D flooded area, the first approach underestimates the flooded area of 27

%, while the 1D with levee underestimation reduces to 3.5 %.

Unfortunately, there is a lack of studies in the literature showing the comparison

between the 1D unsteady state with storage area with 2D model. The level of

improvement associated with the use of LIDAR data makes the 1D and 2D model easier

than before. In order to analyze this, it is important to recall that 2D models represent the

reality in a more simplified way than 1D unsteady state with storage area models.

Therefore, the comparison between 1D unsteady state with storage area and 2D results,

aimed at the evaluation of 2D model performances and created a guideline, seems to be

the most natural choice. In this paper, the compared analysis mentioned above was

carried out.

17
CHAPTER 3

DEVELOPING AND SOLVING THE FINITE VOLUME EQUATION

Introduction:

The finite volume method is a robust method extensively used in computational

fluid dynamics (Moukalled, 2016). The finite volume method is commonly abbreviated

as FVM. There are two methods of finite volume, the first is cell vertex and second is

cell-centered.

The Cell Vertex:

The cell vertex method and so called by (Dual) uses a secondary mesh made up of

a cell for every vertex. The grid cell is shown in black color in Figure 6. The HEC-RAS

generate a new cell for every vertex by joining the centroids of the grid cell, which is

drawn in a blue color. The finite volume formulation is applied on these newly generated

cells. The advantage of the cell vertex method is that the application of boundary

condition becomes easy because the cell centroids lie on the boundaries. However, the

construction of secondary mesh and data structures associated with it can get a little

messy, if not handled properly.

18
Figure 6: Cell Vertex Method and Secondary Mesh.

HEC-RAS uses a hybrid method to compute the ∇H gradient term (Brunner,

2016). When the direction of the hydraulic gradient is normal to the cell face as shown in

Figure 7(a), then HEC-RAS uses a finite difference solution to find ∇H as shown in

Equation 3.1 below. In this case, cells are said to be orthogonal. H2 and H1 are the water

surface elevations at cell 2 and cell 1 respectively. ∆L is the distance between the center

of cell 2 and cell 1.

𝐻2−𝐻1
∇H = Equation 3.1
∆𝐿

When the direction of the hydraulic gradient is not perpendicular to the cell face –

as shown in Figure 7(b) – then HEC-RAS uses a finite volume solution. With the aid of

the Gauss Divergence Theorem, the value of a gradient term at a grid face is

approximated as the sum of the normal component of the water surface elevation at face

k times the length of face k divided by the area of the dual cell.

∑𝑘 𝐻𝑘 .𝑛𝑘 𝐿𝑘
∇𝐻 = Equation 3.2
𝐴𝑗

19
(a)

(b)

Figure 7: Dual Cell Finite Difference and Volume Formulation (Brunner,2016).

Generalized Finite Volume Formulation:

To derive the finite volume formulation let us consider the partial differential

equation shown in Equation 3.3, which is representative for a general governing equation.

This equation is unsteady as shown by the t term where t is time and is two dimensional

as evidenced by the x and y terms where x and y represent distance. The quantities 𝑚,

𝑓, 𝑔 𝑎𝑛𝑑 𝑆 can be scalars or vectors (Hirsch, 1989).

𝜕𝑚 𝜕𝑓 𝜕𝑔⃒
+ + =𝑆 Equation 3.3
𝜕𝑡 𝜕𝑥 𝜕𝑦

Integrating the equation over a single cell, we can write the equation as shown in

Equation 3.4 and A is the area of the cell and Ω is the volume of the cell:

𝜕𝑚 𝜕𝑓 𝜕𝑔⃒
∬Ω 𝑑𝐴 + ∬Ω (𝜕𝑥 + 𝜕𝑦) 𝑑𝐴 = ∬Ω 𝑆 𝑑𝐴 Equation 3.4
𝜕𝑡

The cell can be of any shape like a triangle, quadrilateral or some other polygon.

HEC-RAS limits the grid cell to no more than eight sides.

20
Now, the simplification is the next step, which leads us to the finite volume

formulation. We simplify the first term by using the average of 𝑚, which is defined as

the integral of 𝑚 divided by an area of the cell. Similarly, the integral of the source term

gets simplified by using the definition of the average of the source term as shown:

𝜕𝑚 𝜕𝑚′
∬Ω 𝑑𝐴 = 𝐴 Equation 3.5
𝜕𝑡 𝜕𝑡

∬Ω 𝑆 𝑑𝐴 = 𝑆′𝐴 Equation 3.6

However, the most interesting term is the area integral of the space derivative

terms. These terms get simplified by using the Gauss-Divergence Theorem, which

converts the area integral over the cell footprint to a line integral over the closed

boundary of the cell (Hirsch, 1989). The lines enclosing the cell open up and then a line

integration is out by summing the contributions of these integrals over all faces as shown

in Equation 3.9.

Figure 8: Unit Normal ṅ Components.


The unit normal ṅ has components (nx and ny), which are going to be different for

each of the faces as shown above in Figure 8. The unit normal vector points outward of

the cell (and not inward). Now apply the integrals to each of the faces to obtain Equation

21
3.7 below. This expression can be further simplified obtain this Equation 3.9 using the

definition of averages.

∮Γ(𝑓𝑛,𝑥 + 𝑔𝑛,𝑦 )𝑑𝐿 Equation 3.7

∑𝐹𝑎𝑐𝑒𝑠 ∫Γ𝐹𝑎𝑐𝑒(𝑓𝑛,𝑥 + 𝑔𝑛,𝑦 )𝑑𝐿 Equation 3.8

∑𝐹𝑎𝑐𝑒𝑠(𝑓𝑛,𝑥 + 𝑔𝑛,𝑦 )Δ𝐿 Equation 3.9

The Finite Volume formulation is the sum of the fluxes exchanged between

neighboring cells, when the source terms is absence (Hirsch, 1989). Collect the equations

3.5, 3.8 and 3.9, which is almost the final step of Finite Volume Method. The steps of

FVM depend on the scheme used for calculation of fluxes and time integration.

𝜕𝑚′
𝐴 + ∑𝐹𝑎𝑐𝑒𝑠(𝑓𝑛,𝑥 + 𝑔𝑛,𝑦 )Δ𝐿 = 𝑆′𝐴 Equation 3.10
𝜕𝑡

Application of Finite Volume Method to 2D Floodplain Modeling:

We use a simple Cartesian mesh to solve a 2D equation as shown in Figure 9.

This equation looks like the governing equation for which have just obtained a finite

volume formulation.

𝜕ℎ 𝜕(𝑢ℎ) 𝜕(𝑣ℎ)
+ + +𝑞 =0 Where h= h(x,y,t) Equation 3.11
𝜕𝑡 𝜕𝑥 𝜕𝑦

Where 𝑡 is time, h is the water depth, q is a source or sink term, and u and v are the

velocity components in the X and Y direction.

22
Figure 9: Cartesian Mesh.
Now, Continuing the procedure of FVM as shown earlier. Then, ∆X∆Y refers to

the area of the rectangular grid cell.

𝜕ℎ 𝜕𝑓 𝜕𝑔⃒
+ 𝜕𝑥 + 𝜕𝑦 = 0 Equation 3.12
𝜕𝑡

𝜕ℎ 𝜕𝑓 𝜕𝑔⃒
∬Ω 𝜕𝑡 𝑑𝑥𝑑𝑦 + ∬Ω (𝜕𝑥 + 𝜕𝑦) 𝑑𝑥𝑑𝑦 = 0 Equation 3.13

𝜕ℎ′
∆X∆Y + ∮Γ(𝑓𝑛,𝑥 + 𝑔𝑛,𝑦 )∆X∆Y = 0 (Using Gauss-Divergence) Equation 3.14
𝜕𝑡

For a Cartesian mesh, it easily simplifies further to as single term inside the line

integral. In Cartesian mesh, the vertical faces have ny= 0, therefore the term 𝑔n,y goes

away and for horizontal faces, nx = 0, therefore, 𝑓n,x becomes zero. Also, simplify this

further by using the average values of flux (flowrates) over the faces.

𝜕ℎ′
∆X∆Y = − ∮Γ(𝑓𝑛,𝑥 )∆X − ∮Γ(𝑔𝑛,𝑦 )∆Y Equation 3.15
𝜕𝑡

𝜕ℎ′
∆X∆Y = − ∑𝐹𝑎𝑐𝑒𝑠(𝑓′𝑛,𝑥 )∆X − ∑𝐹𝑎𝑐𝑒𝑠(𝑔′𝑛,𝑦 )∆Y Equation 3.16
𝜕𝑡

Substitute the value of f and g:

𝜕ℎ′
∆X∆Y 𝜕𝑡
= − ∑𝐹𝑎𝑐𝑒𝑠 𝑢′ . ℎ′ . ∆𝑋𝑛,𝑥 − ∑𝐹𝑎𝑐𝑒𝑠 𝑣 ′ . ℎ′ . ∆𝑌𝑛,𝑦 Equation 3.17

23
Where Ω (H) = ∆X∆Yh is the cell volume, A(H) = ∆Y ∗ h and ∆X ∗ h is the area of face

cell and h=H-Z where H is the water Surface Elevation and Z is the DATUM elevation

(Brunner, 2016).

𝜕Ω(𝐻)′
= − ∑𝐹𝑎𝑐𝑒𝑠 𝑢′ 𝐴(𝐻)𝑛,𝑥 − ∑𝐹𝑎𝑐𝑒𝑠 𝑣 ′ 𝐴(𝐻)𝑛,𝑦 Equation 3.18
𝜕𝑡

𝜕Ω(𝐻)′
= − ∑𝐹𝑎𝑐𝑒𝑠(𝑉𝐴(𝐻)) ∙ 𝑛 Equation 3.19
𝜕𝑡

1
However, get rid of the integral as well. So, call the right-hand side face as i + 2 ,

1 1 1
the left-side face i - 2, the top face as j + 2 and the bottom face as j - 2. Now recognizing

the fact that nx is equal to positive One for the right face and negative One for the lift.

Similarly, ny is positive One for top face and negative One for the bottom face (Hirsch,

1989). Finally, will get the finite volume formulation.

𝜕Ω(𝐻)′ 1 1
= − ∆𝑋 (𝑢′ 𝐴(𝐻)( 𝑖+1) − 𝑢′ 𝐴(𝐻)(𝑖−1) ) − ∆𝑌 (𝑣 ′ 𝐴(𝐻)(𝑗+1) − 𝑣 ′ 𝐴(𝐻)(𝑗−1) )
𝜕𝑡 2 2 2 2

Equation 3.20

For a central scheme, there are alternatives can be considered for Finite Volume

method. The average of fluxes is one of these considered (Hirsch, 1989) as shown in

Figure 10.

(𝑓(𝑖,𝑗) +𝑓(𝑖+1,𝑗) )
𝑓𝐴𝐵 = Equation 3.21
2

(𝑔⃒(𝑖,𝑗) +𝑔⃒(𝑖,𝑗+1) )
𝑔𝐵𝐶 = Equation 3.22
2

24
Figure 10: Cell Centered Structured Finite Volume Mesh.
Now, we should define how to calculate the flux components at the cell centers by

using Equation 3.21 and 3.22.

𝜕Ω (𝐻)′ 𝑖,𝑗 1 1
= − 2∆𝑋 (𝑢 𝐴(𝐻)𝑖+1,𝑗 − 𝑢 𝐴(𝐻)𝑖−1,𝑗 ) − 2∆𝑌 (𝑣 𝐴(𝐻)𝑖,𝑗+1 − 𝑣 𝐴(𝐻)𝑖,𝑗−1 )
𝜕𝑡

Equation 3.23

Final, which may have already observed, is that this equation looks a lot the 1D

formula, but with this additional term for Y direction shows us 2D. The conclusion, the

FVM for Cartesian mesh behaves similarly 2D problems.

As mentioned in chapter 2, HEC-RAS will combine Continuity and Diffusion-

Wave form of the Momentum Conservation to compute WSEL (t). After we directly

substitute Diffusion Wave in the Mass Conservation, we will integrate the Equation 2.6

and the result of this integral will be Equation 3.19.

𝜕Ω(𝐻)′
+ ∑𝐹𝑎𝑐𝑒𝑠 𝑉. 𝐴(𝐻). 𝑛 = 0 Equation 3.24
𝜕𝑡

2
𝜕Ω(𝐻)′ 𝐴𝑘(𝐻)𝑅(𝐻)3 ∇𝐻
− ∑𝐹𝑎𝑐𝑒𝑠 1 .𝑛 = 0 Equation 3.25
𝜕𝑡 𝑛
∇𝐻 2
Where the conveyance channel (K) is a permanent waterway, designed to convey storm
water runoff and equal to:

25
2
1.486𝐴𝑘(𝐻)𝑅(𝐻)3
K (U.S) = Equation 3.26
𝑛
1
𝜕Ω(𝐻)′
− ∑𝑓𝑎𝑐𝑒𝑠 𝐾∇𝐻 2 . 𝑛 = 0 Equation 3.27
𝜕𝑡

The Flow (Q) is equal:

(Q) = 𝐾√∇𝐻 Equation 3.28


𝜕Ω(𝐻)′
− ∑𝐹𝑎𝑐𝑒𝑠 𝑄 . 𝑛 = 0 Equation 3.29
𝜕𝑡

In Appendix A contains a detailed example of computation a water surface elevation

(WSEL) using FVM in cell for a HEC-RAS 2D model.

26
CHAPTER 4

DESCRIPTION OF STUDY AREA

Introduction:

This section gives an overview of the study area, which includes the Great Miami

River in West Carrollton and Moraine and another is Bear Creek, that will be used to

simulate the floodplain using one and two dimensional hydrodynamic models. This study

used topographic datasets that are publicly available through the Ohio Geographically

Referenced Information Program (OGRIP). The topographic data sets were in raster form

and were developed using Light Detection and Ranging Elevation Data (LiDAR)

techniques. Presented in the following sections are Geometry Data and cross section data,

land use classifications in terms of the Manning’s n value, and flow data at various river

stations.

Description of Study Area:

Moraine area is a city in Montgomery County, Ohio, United States. The

population is 6,307 at the 2010 census (Moraine, 2014). Moraine has special flood hazard

areas that are subject to periodic inundation which may result in loss of life and property,

and create health and safety hazards. Also, north of West Carrollton (storage area 2) has a

risk from flood hazard as shown in Figure 11. In the terrain figure (right image), it is

clear that a portion of Moraine area and West Carrollton is located in a wide flat

floodplain of the Great Miami River. Moraine area contains a small local airport as well

27
as residential areas. Presently all commercial and residential area as well as the air

park are protected from floodwaters by a levee that runs parallel to the Great Miami

River. Also, the study area around the Bear Creek contains farm as shown in Figure 11.

Figure 11: A Portion of Moraine, West Carrollton and Farm Located in Floodplain.
The study area along the Great Miami River is a 7.38-mile-long and 0.15 mile-

long for Bear Creek. The Great Miami River is characterized by meandering curves and a

relatively flat floodplain with levees located on both sides of the river. The average width

for the Great Miami Rive is around 400 feet with 4 feet average depth and the average

slope is 0.00066 ft/ft. In Bear Creek, the average width is around 100 feet and the average

slope is 0.00066 ft/ft.

28
Light Detection and Ranging Elevation Data:

OGRIP provided all LiDAR data for Montgomery County used in this study.

LiDAR is a remote sensing method that uses an aircraft fitted with a laser which

measures its distance to the ground (NOAA, 2015). The aircraft is also equipped with

high-precision GPS equipment so that the exact spatial location of the aircraft is known.

Each point on the targeted area has three-dimensional coordinate (latitude, longitude,

Elevation) and these data called terrain layers. LiDAR data for Montgomery county is

characterized each “point” is a 2.5 x 2.5 square foot. Ground surface elevations obtained

by LIDAR are available from OGRIP in raster format.

Create Terrain Layer with Actual Channel:

As mentioned earlier, RAS Mapper is a software application that runs in HEC-

RAS that allows display and some management of GIS data. RAS Mapper can import

terrain data that have several formats including raster format. Most application of LiDAR

use lasers that are incapable of penetrating water. A two-laser system must be used in

order to penetrate the water and hence obtain channel bathymetry information (Fugro

Pelagos). Adjustments to the terrain model must be made if the terrain data provided by

LiDAR, does not contain the actual main channel bathymetry. Fortunately, this can be

accomplished within HEC-RAS. From RAS Mapper, the user turn on the geometry layer

for the geometry data to use in creating the channel terrain model. The channel terrain

model is created by right clicking on the geometry layer and selecting Export Layer, then

Create Terrain GeoTiff From XS’s (Channel Only) and XS means cross-sections. HEC-

RAS will ask the user for the raster cell size to use for this terrain layer and the user can

use a same cell size of the base terrain. Now, user has a terrain layer from the channel

29
data and has a base terrain (the terrain with floodplain data). HEC-RAS will have

combined between these two-terrain layers and the new created terrain will have a

channel data and floodplain by create a new terrain layer data and will chose these two

layers and HEC-RAS will combined two layer together automatically as shown in Figure

12.

Figure 12: (A) Original Terrain. (B) A New Terrain with Channel Data. (C) Combination
Between Two Terrain Model and the New Created Terrain.

Flow Data:

We used USGS to get flow data. The USGS station number is 03271500 and

station name Great Miami River at Miamisburg OH. This station is located 3.918 miles

downstream of the study reach. In Bear Creek, the flow data was assumed. Table 3.1

presents the flow data assigned at Great Miami River and Bear Creek. The flow data for

30
open channels is typically unsteady since the discharge through the channel will usually

vary with time and the reason for this is the temporal nature of the storm event that

produced the flooding event. From the table below it is clear that, a storm has occurred

and is causing high flows in the river.

Table 3.1: Flow Data Assigned for a Great Miami River and Bear Creek.

Great Miami River Bear Creek


Time Flow Time Flow
No. No.
(hr) (cfs) (hr) (cfs)
1 0 528 1 0 320
2 24 528 2 24 731.43
3 48 558 3 48 1142.86
4 72 709 4 72 1554.29
5 96 743 5 96 1965.71
6 120 1130 6 120 2377.14
7 144 16100 7 144 2788.57
8 168 40400 8 168 3200
9 192 50900 9 192 2672
10 216 39000 10 216 2144
11 240 23200 11 240 1616
12 264 16800 12 264 1088
13 288 13700 13 288 560

Roughness and Cross-sectional:

In 1D model, Manning’s n value is a very important to the solution of the

computed water surface elevations. There are several factors, which influence the value

of Manning’s n including surface roughness, vegetation, channel irregularities, size and

shape of the channel, seasonal changes, temperature, and suspended material. In the most

recent 10 years, DEM data has made available a huge amount of vegetation data,

preferring the development of new methods for a good estimating roughness (as cited in

31
Pierfranco Costabile et al., 2015). As mention in chapter 2, the 1D modeling is using the

St Venant equations, the roughness parameter commonly appears through a friction slope

relationship that represents bed roughness Equation 2.2. The roughness value computes

the water surface elevation and the water surface elevation determines the area of flow

then will compute the flow crosses this area. So, if we have a high roughness value will

compute a high WSEL then we will have a high flow as well as a low roughness value

will compute a low WSEL then the flow will be low value. The 1D models tend to have

high sensitivity to roughness, and this makes roughness the main key of parameterization

in 1D models.

The roughness parameter, Manning’s n value is also an important parameter that

can be used in calibrating the two-dimensional model. For 2D model, RAS Mapper has

the ability to create a land cover layer and associate this layer with terrain data. RAS

Mapper also allows modelers to specify manning’s roughness values with various land

use categories that, in turn, are defined in the land cover layer. The end result of this is

that RAS Mapper will associated a Manning’s roughness value with each computational

cell faces. On other hand, 1D model will set the manning n value for the channel (NC)

and left and right overbank (LOB and ROB). Finally, in both 1D and 2D model small

changes in Manning’s n can produce significant changes in the WSEL. In this paper, we

will comparison between a three Manning’s n value 0.1, 0.06, 0.035 and we will see what

is a difference between 1D and 2D models result.

A cross-section is simply a set of ground points with each ground point consisting

of a station and an elevation. The cross-section of a river can be determined by measuring

the width of the river and riverbank as well as the depth of the water. In the past, People

32
could be using hand tools such as a tape measure and surveying pole to measure the

cross-section of a river. However, HEC-RAS can provide a cross-section for the river by

using terrain model. The information compiled from multiple river cross-sections can be

used for research projects to determine hydraulic flow and floodplains inundation areas.

33
CHAPTER 5

APPLICATION OF 1D AND 2D UNSTEADY STATE MODEL

Introduction:

HEC-RAS 5.0 can perform a simulation with only a 2D mesh (i.e. no cross

sections). The objective of the research effort is to gain a greater understanding of the 2D

modeling capabilities in HEC-RAS and comparison between 1D unsteady state with

storage area and 2D results, aimed at the evaluation of 2D model performances and

created a guideline for 2D model, to assess the accuracy of 2D results and to identify any

difficulties in building a 2D model. This chapter provides a detailed description for

creating a 1D and 2D floodplain model using HEC-RAS for the Great Miami River.

Terrain Layers:

The RAS Mapper can import floating-point grid format (*.fit), GeoTIFF (*tif) and

other formats (Brunner, 2016). The terrain layers used to construct the terrain model for

this study consisted of 12 raster images or tiles each 5,000 ft x 5,000 ft. Each 5,000 ft x

5,000 ft raster image contained 4,000,000 pixels with each pixel or GIS cell having

dimensions of 2.5 ft x 2,5 ft. Thus, the terrain model used in this study contains a total of

48,000,000 individual GIS cells with each GIS cell having a unique elevation. Figure 13

shows the terrain model used in the study.

34
Figure 13: RAS Mapper with a Terrain Data Layers.
The terrain data does not often include the actual channel bathymetry underneath

the water surface. RAS Mapper now can modify the terrain data to include channel

bathymetry by using the individual HEC-RAS cross-sections geometry and the Cross-

Section Interpolation feature. The result from this step is to generate the channel terrain

layer. The channel terrain layer is created by taking the channel bathymetry data from the

cross sections and using the interpolation feature to interpolate an elevation for each grid

cell between any two-cross sections (Brunner, 2016) as shown in Figure 14. Channel

bathymetry data for this study was obtained by modifying the ORGRIP terrain data.

Specifically, all ground elevations located between the main channel bank station were

lowered by 4 ft. The amount by which ground elevations were lowered was based on

knowledge of the depth of the Great Miami River in the vicinity of the study area as

shown in Figure 15.

35
Figure 14: RAS Mapper with Terrain and Geometry Layers.

Channel Bathymetry Data


740

735

730
Elevation, feet

725 The Terrain Data

720 Channel Bathymetry


Data
715

710

705
0 200 400 600 800 1000 1200 1400
Station, feet

Figure 15: Example of a Trimmed Cross-Section.

36
Creating 2D Flow Area Mesh Only:

Central to the concept of 2D floodplain modeling is use a computational mesh. As

discussed in Chapter3, HEC-RAS uses a combination of a finite-difference and a finite-

volume method to compute water elevation at the center of each computational grid cell

for each computational time step. 2D modeling features in HEC-RAS allow a user to

create computational mesh. In the Geometric Data Editor, the modeler can define the

limits of the computational mesh that envelopes the channel itself plus any adjacent

floodplain areas.

Spatial details describing the polygon can be defined with 2D Flow Area Editor

button. Spatial details include the size of the individual 2D flow cells as well as

Manning’s roughness values for each cell. As will discussed later, Manning’s roughness

values can be defined for specified land use using GIS techniques. In some cases, the

modeler needs to add break-lines. The break lines are any high ground that users want to

align the mesh faces along such as levees and roads and we can add these lines manually.

HEC-RAS will keep water out of the “dry” side of a break line until the water surface

elevation is greater than the elevation of the break line. The 2D Area Break Lines button

can be used add these lines into the terrain background as shown in Figure 16.

After the spatial details of the computational mesh and after any break lines have

been defined, then detailed information describing the computational grid including

hydraulic property table can be generated. There are tolerance input boxes that allow the

user to have some control of the 2D grid. Finally, boundary conditions at the upstream

and downstream ends using must be defined using 2D Area BC Lines as shown in Figure

16.

37
Figure 16: Creating 2D Flow Area Mesh and Break-Lines.

Land Use Classification:

HEC-RAS uses Manning’s formula to compute friction losses along the ground

surface. Friction losses are used in the solution methodology employed by HEC-RAS

specifically the Conservation of Momentum equation. Ohio Department of Natural

Resources (ODNR) provided all land use classification for Montgomery County used in

this study. To associated roughness values to the two-dimensional mesh, a land use shape

file created in ArcGIS is opened in the RAS Mapper. In RAS mapper, polygons are

created for each land use classification based on a Manning’s n value. The Land Cover

Editor allows the users to associate a specific land classification with a unique Manning’s

n value. RAS Mapper can associate between land cover data and specific geometry data

set. In geometry data, Table Menu allows the users to show all the land cover identifiers

and can editor.

38
1D Unsteady State with Storage Area Model:

Storage area is a series of areas contain some of the natural obstacles or urban

areas. The storage cell boundaries were used to extract the Lidar data associated with

each storage area. These data were used to generate a cell-by-cell volume–elevation

relationship that describes the volume of water that is stored in a given cell at a given

water level. The 1D unsteady state with storage area model has assumed that the

floodplain comprises a series of discrete areas acting as storage cells, which connect with

the main channel. However, Overflow weirs control the filling and emptying process of

the storage area. When the water level is over the lateral weir the flow will be computed

using the standard equation for a broad-crested weir equation:

Q = CLH3/2 Equation 5.1

Where Q is discharge, C is the weir flow coefficient, L is the length of weir crest

and H is the upstream energy head above the weir crest. Also, Cell volume–elevation

relationship is required to capture the filling of each storage cell and to define the

inundated area in each storage cell during filling and draining. The very important

assumptions about the process of the filling of the storage cell that is the water surface are

horizontal in storage area. Also, this model just uses the continuity equation to be solved

to describe water levels within storage area.

The 1D unsteady state with storage area model approaches to flood inundation are

now more applied as high-resolution LiDAR data are increasingly available for defining

and parameterizing storage cells without extensive field data collection. HEC-RAS has

the ability to create storage areas and simulates of the floodplain flow using storage areas

that hydraulically connect with other neighboring storage areas and/or the main channel.

39
The 1D model is set up to represent a 7.38-miles reach of the Great Miami River, which

is described by a series of 25 cross-sections. The storage area is organized in the model

by two storage and these storage areas area described by their volume-elevation curves.

Both of cross-sections and the volume-elevation curves are provided from a high-

resolution digital elevation model (DEM) that was obtained from LiDAR data using

HECRAS as shown in Figure 17 and 18.

Figure 17: GMR Model Layout for the Proposed of Floodplain Storage Area.

40
Figure 18: Bear Creek Model Layout for the Proposed of Floodplain Storage Area.

Boundary Condition:

With the 2D mesh created and 1D geometry data, an unsteady state simulation is

conducted. The most commonly used boundary condition available in HEC-RAS will be

a flow hydrograph. A flow hydrograph is the variation of discharge with respect to time

as shown in Figure 19. Discharge is the volume of water flowing past a location per unit

time (usually in cubic feet per second (cfs)). HEC-RAS allows a flow hydrograph to be

assigned at either the upstream or downstream end of the 2D flow area or the cross-

section in 1D. However, for most applications, the hydrograph will reflect the upstream

boundary. At the downstream boundary, the normal depth option commonly be used.

This option is simple and the users must have an estimate of the friction slope at the

41
downstream boundary. Typically, the slope of the channel will be used in lieu of the

friction slope. At the downstream boundary, HEC-RAS will use Manning’s equation to

compute the stage based on the flows obtained from the unsteady flow analysis. From

ArcGIS, The Great Miami River bottom slope and Bear Creek at downstream end for 1D

and 2D equal 0.00066 ft/ft. As well as, the input flow data for 1D model is the same that

have been introduced by the 2D model.

Bear Creek
Flow Data for GMR

Figure 19: A Flow Hydrograph for Great Miami River and Bear Creek.

Flood Inundation Map:

To run the model, the unsteady flow analysis may be selected from the Run menu

item on the main HEC-RAS window. The modelers create a plan by selecting the

geometry that contains the 2D and 1D geometry data and an unsteady-flow data file for

the flood event to be examined. There is a box for Flood mapping under the Programs to

Run area. The option will work, when the modeler has been set up correctly for RAS

Mapper data. The resulting from this option is generated a depth grid of the maximum

inundation that occurred at all locations in a 2D and 1D model. A new feature also under

the Computational Setting called Mapping Output Interval that allows modelers to set a

mapping interval for creating Dynamic maps and This form is shown in Figure 20. For

42
this study, a simulation duration of 13 days of computational time step of 2 minutes and 4

minutes for Mapping Output was used. In GMR model, the time required to complete the

simulation was nearly 12 minutes for 2D model and 20 second for 1Dmodel but Bear

Creek model was 48 Seconds for 2D model and 10 seconds for 1D model.

Figure 20: Unsteady Flow Analysis Window.


In 2D model, HEC-RAS pre-processor calculates the stage-storage curve base on

the terrain data for each cell. The stage-storage curve is developed at the preprocessor

step; also, we can compute 2D flow area hydraulic table from RAS Mapper. As discussed

in Chapter 3, each cell has a center and the water surface elevation is computed at these

centers. The cell face is evaluated similar to cross section and computed all hydraulic

property table an example, elevation versus - Area, Roughness, wetted-perimeter. The

flow across the face based on this hydraulic property table. In Figure 21, the

computational cells in HEC-RAS contain enough hydraulic detail such that the flow can

move through a channel. However, the flow remains in the channel until the stage higher

43
than the bank elevation of the channel, then it spills out into the overbank areas as shown

in Figure 22.

Figure 21: 2D Model Flow Runs Through Great Miami River and Bear Creek n=0.035.

44
Floodplain Area

Figure 22: 2D Model Flow Overtopping to the Floodplain Area n=0.035.


As mentioned in the previous chapter, the current version of HEC-RAS allows

users to create 2D flow area model. Currently, the result of 2D model can be observed in

RAS Mapper. The dynamic mapping in RAS Mapper became the main way to express at

a result. There are many option results layers but the important three layers will be Depth,

Velocity, and Water Surface Elevation.

The accuracy of this dynamic mapping depends on several factors, but the most

important factors are cell size. The cell size should be appropriate for terrain layer and

flow over the terrain. Each cell in HEC-RAS is an included elevation volume relationship

that represents the details of the terrain layer. Also, cell faces are detailed cross section

which gets processed into hydraulic properties tables. Because the accuracy represents

the underlying terrain, the modeler can use larger cell size. Additionally, the cell size

describes the water surface slope. If the water surface slope changes rapidly, the modeler

can use the small cell size to have enough computational point to describe the changing

45
water surface. As result, accuracy HEC-RAS resulting depend on the accuracy of terrain

layers.

In Figure 23, flow across the cross-sections based on a hydraulic property that is

input in geometry data. The computational of water surface elevation at each cross-

section in HEC-RAS show that the flow can move through a channel. In the 1D unsteady

state with storage area model, when the water is on the floodplain it is assumed that fills

the storage area from the lowest ground elevation to upwards and flow will leave the

channel when the elevation of the water exceeds the lateral structure into a storage area as

shown in Figure 24. Also, flow cannot leave the storage area until the elevation of the

water exceeds the hydraulic connection into another storage area. So, the only control on

the flow between that storage area connects to the next storage cell is the nature of the

hydraulic connection. There is no explicit representation of momentum effects in this

situation, except as parameterized through an application of a weir equation at each

hydraulic connection.

46
Floodplain Area

Figure 23: 1D Model Flow Runs Through Great Miami River and Bear Creek n=0.035.

Figure 24: 1D Model Flow Overtopping to Storage Area n=0.035.

47
CHAPTER 6

RESULTS

Introduction:

The results from 1D unsteady with storage area modeling and 2D modeling are

presented in this chapter for Great Miami River and Bear Creek. We will comparison

between a three simulations Manning’s n value 0.1, 0.06 and 0.035. For both models, we

produce a comparison for stage hydrographs for each manning’s value, water surface

elevation of floodplain, water surface elevation along river and flow into storage area for

each manning’s value. To compare between the 1D and 2D models, three cross-sectionals

configurations for Great Miami River and Bear Creek are evaluated by determining the

largest, average and the minimum differences of WSEL for both 1D and 2D models in

the three different simulations. Also, to compare the results of floodplain area on 1D and

2D model, created a mesh computational resolutions of 80×80 feet in GMR and 20×20

feet in Bear Creek for 2D model and created two storages areas (Moraine area and

Storage area 2) in GMR and one storage area (Farm area) in Bear Creek for the 1D model

to determine the WSEL. The results from the HECRAS simulations are then compared.

WSEL Comparison Between 1D and 2D Model:

The terrain grid size used for 2D flow area model is 2.5-ft and the volume–

elevation curves and cross-sections for 1D model are also extracted from 2.5-ft grid

48
terrain. The output from 1D model is in the form of level at each cross-section calculation

point along the main channel. In 2D model, at each cell will compute the levels of water.

However, in Great Miami River model the first major point to make from these

result is that the comparison of time series of WSEL (stage hydrographs) between 1D and

2D model. We will select three cross-sections to compare stage hydrographs. The first

stage hydrograph represents the largest different between the 1D unsteady state with

storage area with 2D flow area. The second stage hydrograph represents the average of

the different between two models. The third stage hydrograph explains the minimum

different between 1D and 2D model. We will repeat these comparisons for all the three

manning’s n value simulations as shown in Appendix B. A comparison of the results

obtained from two models simulation runs shows that there is absolutely no big

difference in the water level in Great Miami River as shown in Figure 25.

Figure 25: Maximum WSEL Profile Along GMR Computed with 1D and 2D Model.

49
The figure above shows that it is clearly the solution of the 1D model approaches

the 2D solution if the cross-sections are properly located in order to capture all hydraulics

situations. If the positioning of the cross sections were wrong, will result in a diffusion of

the water surface over the bed step. From 8,000 feet station to 15,000 feet station, it is

clearly that there is a big change in ground geometry (Channel bottom) and Caused the

occurrence in the largest differences of maximum WSEL between 1D and 2D models.

The 2D model will take into consideration this change in ground geometry because at

each grid cell will compute all hydraulic properties (cross-sectional area, wetted

perimeter, hydraulic radius, and conveyance) and can be computed a water surface

elevation at each grid cell and there are more than 450 grid cells along this distance.

However, the 1D model just computes WSEL at Cross-Sections and between them use

interpolation technique and does not take into consideration any change in the

characteristics of the channel. Also, the 1D model just has three cross-sections along of

this ground geometry changing. In this consideration, we should understand that the

water level in the 1D model is constant along the cross section, but the 2D model results

may show a significant variability of the water level across the section. Also, this

difference in WSEL between 1D and 2D model in the river causes a difference in

amounts of flows that fill the storage area 2. We will consider the cross sections, where

the differences of maximum WSEL between the 1D and 2D models at each roughness

values, are the large, average, and minimum difference. The result shows the differences

between the two models overall studied reach are not higher than 1.19-feet as shown in

Table 6.1.

50
Table 6.1: A Differences of Maximum WSEL in GMR.

Max. WSEL Max. WSEL ∆H, WSEL


N XS
1D (ft) 2D (ft) {1D-2D) (ft)
180 Largest 718.5 717.31 1.19
0.035 130 Average 711.12 711.39 0.27
150 Minimum 714.05 714.09 0.04
150 Largest 716.32 717.27 0.95
0.06 100 Average 710.89 711.11 0.22
190 Minimum 722.81 722.77 0.04
180 Largest 723.4 722.82 0.58
0.1 110 Average 717.53 717.28 0.25
120 Minimum 718.01 717.97 0.04

In Bear Creek, a comparison of the results obtained from two models simulation

runs shows that there is absolutely no difference in the water level as shown in Figure 26.

Figure 26: Maximum WSEL Profile Along Bear Creek Computed with 1D and 2D Model.

51
We will consider the cross sections, where the differences of maximum WSEL

between the 1D and 2D models at each roughness values. The result shows the

differences between the two models overall studied reach are not higher than 0.17-feet

(2.0 Inch) as shown in Table 6.2.

Table 6.2: A Differences of Maximum WSEL in Bear Creek.

Max. WSEL Max. WSEL ∆H, WSEL


N XS
1D (ft) 2D (ft) {1D-2D) (ft)
10 709.28 709.34 0.06
0.035 15 709.61 709.54 0.07
20 710.76 710.86 0.1
10 712.05 712.1 0.05
0.06 15 712.25 712.25 0
20 712.37 712.54 0.17
10 715.09 715.03 0.06
0.1 15 715.3 715.15 0.15
20 715.33 715.38 0.05

The Inundation Area and the WSEL In Floodplain Area:

In the following part, attention is focused on a floodplain area on the Great Miami

Rive particularly in Moraine area and West Carrollton (storage area 2) and on Bear Creek

in Farm area. This thesis is discussed to show the different accuracy in the simulation

inundation extent in a floodplain for both the 1D unsteady state with storage area and 2D

model. In the 1D model, a LIDAR data obtains all details of the storage area and the

lateral structure or weir. Weir is a low dam or wall built across a stream to raise the

upstream water level and measuring flow. Sometimes described as a measuring weir or

gauging weir. The types of weirs include broad-crested weirs, sharp-crested weirs, ogee

weirs, and V-notched weirs. Figure 27 (A) shows the simplicity of creating a storage area

as a floodplain area and how easily using HECRAS to compute the volume-elevation

52
relationship as depending on terrain data. Also, HECRAS can provide all details of the

weir or lateral structure to connect main channel with storage area as shown in Figure 27

(B). In Bear Creek, we also used same technique to create a storage area and weir.

Figure 27: (A) Create a Storages Areas. (B) Create the Weirs or Laterals Structures.
It is clear to us that there are no buildings around or link with the levee at Moraine area.

Thus, the weir around the Moraine area represents very well but the weir around the

Storage Area 2 does not represent very well because there is a buildings link with it.

In 2D model, it is conceptually an extension of the 1D unsteady with storage area

approach: rather than discretizing the floodplain into several large storage areas, the

floodplain surface and channel are discretized into a large number of small storage cells

as shown in Figure 28. But in practice, the 2D model approach is based upon explicit

53
mass conservation (as with the 1D unsteady with storage area model) additionally to

momentum conservation.

Figure 28: (A) Concepts of 1D Model. (B) Concepts of 2D Model.

1D Model 2D Model

Figure 29: A Comparison Inundation Map in GMR n=0.035.


54
In Figure 29 above, the comparison between the 1D with storage area and 2D in

inundation extent map in Great Miami River results of roughness value (n=0.035) is

shown. The figure 27 (A) represents the solution of the 1D model: It significantly differs

from the results obtained using the 2D model figure 27 (B). This is due to the filling

process that, in practice, in 1D with storage area model the flow will start to fill the

lowest ground point base on mass conservation but in the 2D model, the results show

how the flow moves around the floodplain area base on momentum conservation until

arrive in the lowest ground point. Also, due to the techniques used by the two models for

the management of the lateral structure (weir). In situations like these, the 1D model

provides simply a rough approximation of the 2D results. However, there is a similarity

result for both of models in inundation map at the floodplain area for the two-roughness

value (n= 0.06 and 0.1) as shown in Figure 30 and 31.

1D Model 2D Model

Figure 30: A Comparison Inundation Map in GMR n=0.06.

55
1D Model 2D Model

Figure 31: A Comparison Inundation Map in GMR n=0.1.


In Table 6.3, the comparison between the 1D with storage area model and 2D

model in the maximum water levels in floodplain area (Moraine Area and Storage Area

2) results is shown. The result shows the differences between the two models overall

studied reach are not higher than 0.7-feet (8.4 Inch). As mentioned earlier, because the

weir was represented very well at Moraine area the difference of WSEL between 1D and

2D models was very small. But the poor presentation for the weir around the Storage

Area 2 and the largest Difference WSEL in the river was near the Storage Area 2, Cause

a high difference of WSEL between both 1D and 2D models at Storage Area 2.

Table 6.3: Comparison of WSEL in Floodplains Areas Around GMR.

Moraine Area Storage Area2


N Max. Max. ∆H, WSEL Max. Max. ∆H, WSEL
WSEL in WSEL in (1D-2D) WSEL in WSEL in (1D-2D)
1D Model 2D Model (ft) 1D Model 2D Model (ft)
0.035 - - - 706.5 705.8 0.7
0.06 722.07 722.05 0.02 715.21 714.57 0.64
0.1 724.42 724.25 0.17 717.82 717.4 0.42

56
In Table 6.4, the comparison between the 1D with storage area model and 2D

model in the maximum water levels at Bear Creek in floodplain area (Farm Area) results

is shown. The result shows the differences between the two models overall studied reach

are not higher than 0.09-feet (1.0 Inch).

Table 6.4: Comparison of WSEL in Farm Area Around Bear Creek.

Farm Area
N Max. WSEL in Max. WSEL in ∆H, WSEL
1D Model 2D Model (1D-2D) (ft)
0.035 711.1 711.07 0.03
0.06 712.12 712.11 0.01
0.1 714.61 714.52 0.09

The comparison between the 1D with storage area model and 2D model in

inundation extent map in Bear Creek results of roughness value (n=0.035, n= 0.06 and

0.1) is shown in Figure 32, 33 and 34. The result shows that there is a similarity result for

both of models in inundation map at the floodplain area.

1D Model 2D Model

Figure 32: A Comparison of Inundation Map in Bear Creek n=0.035.

57
1D Model 2D Model

Figure 33: A Comparison of Inundation Map in Bear Creek n=0.06.

1D Model 2D Model

Figure 34: A Comparison of Inundation Map in Bear Creek n=0.1.

58
Findings:

Our summary is that If there is a good definition or presentation of the storages

areas and weirs and the cross-sections are properly located in order to capture all

hydraulics situations, the results are largely convergent similarly between 1D unsteady

with storage area model and 2D flow area model as shown in Bear Creek model, while if

the storages areas and weirs are not defined clearly and the positioning of the cross

sections were wrong the results 2D Model will be better and more realistic. The results of

those applications show that the use of 1D models requires greater hydraulic skills than

the use of 2D model. As expected, the value of roughness plays key role in 1D with

storage area model and 2D model. This is reflected in the inundation area and the WSEL

at each cross section, where higher values of roughness were produced a high WSEL and

inundation area.

59
CHAPTER 7

CONCLUSION AND RECOMMENDATIONS

Conclusion:

Building 2D flow area by using HEC-RAS not only provides us with the accurate

floodplain mapping but also reduces the time and effort required by modelers for building

the 1D hydrodynamic model. For 1D model with storage area, we must create cross

sections and create weirs structures and create storage areas that require a lot of time and

effort. Based on the result, to use the 1D model with storage area, we must represent

storages areas very well and use terrain data to compute volume-elevation relationship.

Also, we must represent weirs very well and clearly capture all filling and emptying

processes locations. Then, we must select a Manning's n value very carefully. Unlike the

1D model with storage area, it is a lot easier to create a 2D flow area model because the

2D setup already is pre-configured to contain such information; for that reason, we

recommend using the 2D model. Finally, the steps to create 2D modeling are providing

terrain map/model with projection coordinate system, creating the computational mesh,

providing a flow inputs and exit stage, exploring a land cover with a spatial bed

roughness, setting a turbulence, running Simulation and Validation the results.

Recommendations:

From this study, a contribution has been made to the modeling world by

comparison between 1D unsteady with storage area model and 2D model by using HEC-

60
RAS. This study finds a limitation to edit terrain data and we recommend creating

a pixel editor in RAS Mapper that helps users to modify and add in terrain data such as

modifying a Bathymetry geometry for the channel for 2D model. This study can be

further developed to go deep in manning's n value and how to select the best value for 1D

and 2D model.

61
REFERENCES

Alho P,Aaltonen J. “Comparing a 1D hydraulic model with a 2D hydraulic model for the

simulation of extreme glacial outburst floods”. 2008. Hydrological Processes

22(10): 1537–1547. DOI: 10.1002/hyp.6692.

Ahmad, S., Simonovic, S. P. Comparison of One-Dimensional and Two-Dimensional

Hydrodynamic Modeling Approaches For Red River Basin. Report to the

International Joint Commission-Red River Basin Task Force, Ottawa,

Washington. 1999. 1-51.

Bates, P.D., Horritt, M.S., Hunter, N.M., Mason, D., Cobby, D. Numerical modelling of

floodplain flow. Computational Fluid Dynamics: Applications in Environmental

Hydraulics, 2005 John Wiley & Sons, Ltd. 271 – 297.

Brunner W. Gary. HEC-RAS, River Analysis System Hydraulic Reference Manual

version 5.0. Printed and Distributed by US Army Corps of Engineers Hydrologic

Engineering Center (HEC), Feb. 2016, Chapter 2, 55.

Brunner W. Gary. HEC-RAS, River Analysis System Hydraulic Reference User’s

Manual version 5.0. Printed and Distributed by US Army Corps of Engineers

Hydrologic Engineering Center (HEC), Feb. 2016, Chapter 3, p 36.

Chase, Donald V., Chapter 17, HEC-RAS Course Notes, 2016.

62
Chatterjee C, Forster S, Bronstert A. Comparison of hydrodynamic models of different

complexities to model floods with emergency storage areas. 2008. Hydrological

Processes 22(24): 4695–4709.

Costabile, P., Macchione, F., Natale, L., Petaccia, G. Flood mapping using LIDAR DEM.

Limitations of the 1-D modeling highlighted by the 2-D approach. 2015. Nat.

Hazard. http://dx.doi.org/10.1007/s11069-015-1606-0.

Dewberry and LLC Davis. HEC-RAS Procedures for HEC-2 Modelers. Printed and

Distributed by Federal Emergency Management Agency (FEMA), April 2016.

Dunn, Christopher N. HEC A Personal View (Celebrating 50 Years). (2014): n. pag.

Web. 11 Nov. 2016.

FEMA.gov. N.p., n.d. Disaster Declarations for 2016. Disaster Declarations for 2016 |

Web. 23 Nov. 2016.

FEMA, Hydraulic Numerical Models Meeting the Minimum Requirement of National

Flood Insurance Program. Federal Emergency Management Agency (FEMA)

Web, 2016.

Fread, D.L., Lewis J.M. (1998). NWS FLDWAV MODEL. Printed and Distributed by

the Hydrologic Research Laboratory Office of Hydrology National Weather

Service (NWS), NOAA 1984, pp. V.

Fema.gov., Floodplain Modeling Manual: Hydrologic Engineering Center-River Analysis

System Procedures For HEC-2 Modelers | FEMA.Gov. Fema.gov. N.p., 2016.

Web. 11 Nov. 2016.

63
Fugro Pelagos. Airborne Lidar Bathymetry Services by Fugro Pelagos, Inc. Airborne

Lidar Bathymetry Services by Fugro Pelagos, Inc. FUGRO PELAGOS, n.d. Web.

26 Nov. 2016.

Horritt M. S. and Bates P. D. Evaluation of 1D and 2D numerical models for predicting

river flood inundation. Journal of Hydrology, 268:87–99, 2002.

Hirsch, Ch. Numerical Computation of Internal and External Flows. Vol. 1. Chichester:

John Wiley & Sons, 1989. Print.

Hilldale, Rob, MS PE, and Dave Mooney, PhD PE. One-Dimensional Hydraulic

Modeling of the Yakima Basin. U.S. Department of the Interior Bureau of

Reclamation Pacific Northwest Region. 2007. (2007): 1-26. Web. 31 Jan. 2017.

HEC. Hydrologic Engineering Center: A Quarter Century, 1964-1989:

Histories. Cdm16021.contentdm.oclc.org. N.p., 1989. Web. 11 Nov. 2016.

Jowett, I.G., and M.J. Duncan. Effectiveness of 1D and 2D hydraulic models for instream

habitat analysis in a braided river. 2011. Ecol. Eng. Published online DOI:

10.1016/j.ecoleng.2011.06.036.

Kelly, Brian P., and Paul H. Rydlund. Estimated flood-inundation mapping for the Lower

Blue River in Kansas City, Missouri, 2003-2005. USGS Publications Warehouse

RSS. N.p., 2005. Web. 01 Feb. 2017.

Moraine.oh.us., History of the City. History of the City. N.p., 17 Nov. 2014. Web. 11

Nov. 2016.

Moore, Walter P. HEC-2: Flood Plain Analysis. Dodson-hydro.com. N.p., 2012. Web. 11

Nov. 2016.

64
Moukalled, Fadl Hassan, L. Mangani, and M. Darwish. The Finite Volume Method in

Computational Fluid Dynamics: An Advanced Introduction with OpenFOAM®

and Matlab®. Cham: Springer, 2016. Web.

NOAA. What Is LIDAR. US Department of Commerce, National Oceanic and

Atmospheric Administration. N.p., 29 May 2015. Web. 12 Nov. 2016.

Tayefi, V., Lane, S.N., Hardy, R.J., Yu, D. A comparison of one- and twodimensional

approaches to modelling flood inundation over complex upland floodplains. 2007.

Hydrological Processes 21 (23), 3190–3202.

65
APPENDIX A

This section provides a simple example showing how mass conservation is used

for computing WSEL. This example follows the same general algorithm that is carried

out by the HEC-RAS. For this an example, we will refer Figure 35 and we will compute

the water surface elevation at Cell A for time period t+∆t. Notice that Cell A has four

cells that bound it: Cell B, Cell C, Cell D and Cell E.

Figure 35: The Cell (A) That HEC-RAS Will Compute the WSEL Example 1.

We know the WSEL all cells at time= t1 as shown in Table A-1. The length

shown in Table A-1 is the distance from the center of Cell A to the center of the Cells

that bound it. It is important to note that HEC-RAS Computes property tables for each

grid cell before any water surface elevation computations are carried out. Hydraulic

66
property tables generated for each cell include those shown in the list below.

Notice that each of the hydraulic properties are a function of the water surface elevation.

• Cell Volume vs. Elevation (stage-storage curve).

• Face Area vs. Elevation.

• Face Wetted Perimeter vs. Elevation.

Table A-1: WSEL for Such Cells at Time (t1) for Example 1.

WSEL(t1)
Cell Length
At 4-Mar-63-9:24:00
A 715.58 80
B 715.50 80
C 715.66 80
D 715.64 80
E 715.55 80

WSEL at t1 = 4-Mar-63-9:24:00, the hydraulic table will provide all information

of Area, Wetted Perimeter, and Roughness. HEC-RAS will compute the conveyance for

each face cell and computes the discharge through each face as shown in Table A-2.

Table A-2: Computes the Discharge Through Each Face Example 1.

Wetted Hydraulic
Area Manning's Conveyance Discharge
Face Perimeter Radius ∆H/L Direction
(Ft2) Roughness (cfs) (CFS)
(Ff) (Ft)
A-B 18 20 0.9 0.015 1662.245 0.001 52.56 OUT

A-C 32 66 0.48 0.015 1956.508 0.001 61.87 IN

A-D 15 82 0.18 0.015 478.854 0.00075 13.11 IN

A-E 24 74 0.32 0.015 1122.342 0.000375 21.73 OUT

67
Sample calculation for Table A-2:

𝐴𝑟𝑒𝑎 18
Hydraulic Radius (R) (A-B face) = 𝑊𝑒𝑡𝑡𝑒𝑑 𝑃𝑒𝑟𝑖𝑚𝑒𝑡𝑒𝑟 = 20 = 0.9 Ft.

2 2
1.486 ∗ 𝐴𝑟𝑒𝑎 ∗ Hydraulic Radius 3
=
1.486∗18∗(0.9)3
Conveyance (A-B face) = = 1662.245 Cfs.
𝑛 0.015

We assume that the dual grid direction is an orthogonal between a grid cell an example

Cartesian Mesh, HEC-RAS will have used finite difference method:

(715.58−715.50)
∆H/L = = 0.001.
80

Discharge (Q) (A-B face) = 𝐾 √∆H/L = 1662.245 * √0.001 = 52.56 Cfs.

The program uses the mass conservation method:

𝑑𝑣
∑QIN-∑QOUT = 𝑑𝑡

𝑉(𝑡+∆𝑡)−𝑉(𝑡)
(61.87+13.11) - (52.56+21.73) = 0.69 = ∆𝑡

V(4-Mar-63-9:24:00) from Stage-storage curve = 2450 Ft3.

At time t2= t1+∆t = (4-Mar-63-9:28:00), the volume of the cell will be:

𝑉(𝑡 + ∆𝑡) =2450 + (0.69 * ∆t (240 second)) = 2614.48 Ft3.

This means that 2614.48 Ft3 of water will be in Cell A after a 4-minute period.

HEC-RAS will then use the Elevation-Volume Relationship to find the new WSEL for

Cell A at t2 as shown in Figure 36. From the stage-storage curve for Cell A, the WSEL

after 4 minutes is equal 715.60. in other words, the 2560.68 Ft3 of water that will be in

Cell A causes a water level change of 0.02 Ft.

68
Figure 36: Cell Volume Elevation Relationship Example 1.

69
Figure 37 is a second example and we will compute the water surface elevation at

Cell A for time period t+∆t. Notice that Cell A has four cells that bound it: Cell B, Cell

C, Cell D and Cell E.

Figure 37: The Cell (A) That HEC-RAS Will Compute the WSEL Example 2.
Assume that we know the WSEL all cells at time= t1 as shown in Table A-3. The

length shown in Table A-3 is the distance from the center of Cell A to the center of the

Cells that bound it. It is important to note that HEC-RAS Computes property tables for

each grid cell before any water surface elevation computations are carried out. Hydraulic

property tables generated for each cell include those shown in the list below. Notice that

each of the hydraulic properties are a function of the water surface elevation.

• Cell Volume vs. Elevation (stage-storage curve).

• Face Area vs. Elevation.

• Face Wetted Perimeter vs. Elevation.

70
Table A-3: WSEL for Such Cells at Time (t1) for Example 2.

WSEL(t1)
Cell Length
At 9-Mar-63-10:24:00
A 714.725 80
B 714.723 80
C 714.739 80
D 714.729 80
E 714.724 80
WSEL at t1 = 9-Mar-63-10:24:00, the hydraulic table will provide all information

of Area, Wetted Perimeter, and Roughness. HEC-RAS will compute the conveyance for

each face cell and computes the discharge through each face as shown in Table A-4.

Table A-4: Computes the Discharge Through Each Face Example 2.

Wetted Hydraulic
Area Manning's Conveyance Discharge
Face Perimeter Radius ∆H/L Direction
(Ft2) Roughness (cfs) (cfs)
(Ft) (Ft)
A-B 48 31 1.548 0.015 6364.366 2.5E-05 31.822 OUT
A-C 50 65 0.769 0.015 4158.487 1.8E-04 55.012 IN
A-D 25 81 0.309 0.015 1131.108 5.0E-05 7.998 IN
A-E 75 81 0.926 0.015 7058.401 1.2E-05 24.955 OUT

Sample calculation for Table A-4:

𝐴𝑟𝑒𝑎 48
Hydraulic Radius (R) (A-B face) = 𝑊𝑒𝑡𝑡𝑒𝑑 𝑃𝑒𝑟𝑖𝑚𝑒𝑡𝑒𝑟 = 31 = 1.548 Ft.

2 2
1.486∗ 𝐴𝑟𝑒𝑎 ∗ Hydraulic Radius 3
=
1.486∗48∗(1.548)3
Conveyance (A-B face) = = 6364.37 Cfs.
𝑛 0.015

We assume that the dual grid direction is an orthogonal between a grid cell an example

Cartesian Mesh, HEC-RAS will have used finite difference method:

71
(714.725−714.723)
∆H/L = = 2.5E-05.
80

Discharge (Q) (A-B face) = 𝐾 √∆H/L = 6364.37* √2.5E − 05 = 31.822 Cfs.

The program uses the mass conservation method:

𝑑𝑣
∑QIN-∑QOUT = 𝑑𝑡
𝑉(𝑡+∆𝑡)−𝑉(𝑡)
(55.01 + 7.99) – (31.82 + 24.95) = 6.23 = ∆𝑡

V (9-Mar-63-10:24:00) from Stage-storage curve = 6300 Ft3.

At time t2= t1+∆t = (9-Mar-63-10:28:00), the volume of the cell will be:

𝑉(𝑡 + ∆𝑡) = 6300 + (6.577* ∆t (240 second)) = 7795.85 Ft3.

This means that 7795.85 Ft3 of water will be in Cell A after a 4-minute period.

HEC-RAS will then use the Elevation-Volume Relationship to find the new WSEL for

Cell A at t2 as shown in Figure 38. From the stage-storage curve for Cell A, the WSEL

after 4 minutes is equal 714.90. in other words, the 7795.85 Ft3 of water that will be in

Cell A causes a water level change of 0.175 Ft.

72
Figure 38: Cell Volume Elevation Relationship Example 2.

73
Figure 39 is the third example and we will compute the water surface elevation at

Cell A for time period t+∆t. Notice that Cell A has four cells that bound it: Cell B, Cell

C, Cell D and Cell E.

Figure 39: The Cell (A) That HEC-RAS Will Compute the WSEL Example 3.

Assume that we know the WSEL all cells at time= t1 as shown in Table A-5. The

length shown in Table A-5 is the distance from the center of Cell A to the center of the

Cells that bound it. It is important to note that HEC-RAS Computes property tables for

each grid cell before any water surface elevation computations are carried out. Hydraulic

property tables generated for each cell include those shown in the list below. Notice that

each of the hydraulic properties are a function of the water surface elevation.

• Cell Volume vs. Elevation (stage-storage curve).

• Face Area vs. Elevation.

• Face Wetted Perimeter vs. Elevation.

74
Table A-5: WSEL for Such Cells at Time (t1) for Example 3.

WSEL(t1)
Cell Length
At 3-Mar-63-19:00:00
A 709.808 80
B 0 80
C 710.159 80
D 0 80
E 0 80

WSEL at t1 = 3-Mar-63-19:00:00, the hydraulic table will provide all information

of Area, Wetted Perimeter, and Roughness. HEC-RAS will compute the conveyance for

each face cell and computes the discharge through each face as shown in Table A-6.

Table A-6: Computes the Discharge through Each Face Example 3.

Wetted Hydraulic
Area Manning's Conveyance Discharge
Face Perimeter Radius ∆H/L Direction
(Ft2) Roughness (cfs) (cfs)
(Ft) (Ft)
A-B 0 0 - 0.015 - - - -
A-C 2 23 0.087 0.015 38.89 0.0043 2.58 IN
A-D 0 0 - 0.015 - - - -
A-E 0 0 - 0.015 - - - -

Sample calculation for Table A-6:

𝐴𝑟𝑒𝑎 2
Hydraulic Radius (R) (A-C face) = 𝑊𝑒𝑡𝑡𝑒𝑑 𝑃𝑒𝑟𝑖𝑚𝑒𝑡𝑒𝑟 = 23 = 0.087 FT.

2 2
1.486 ∗ 𝐴𝑟𝑒𝑎 ∗ Hydraulic Radius 3
=
1.486∗2∗(0.087)3
Conveyance (A-B face) = = 38.89 Cfs.
𝑛 .015

We assume that the dual grid direction is an orthogonal between a grid cell an example

Cartesian Mesh, HEC-RAS will have used finite difference method:

(710.159−709.808)
∆H/L = = 0.0043.
80

75
Discharge (Q) (A-B face) = 𝐾 √∆H/L = 38.89 * √0.0043 = 2.58 Cfs.

The program uses the mass conservation method:

𝑑𝑣
∑QIN-∑QOUT = 𝑑𝑡
𝑉(𝑡+∆𝑡)−𝑉(𝑡)
2.58 = ∆𝑡

V (3-Mar-63-19:00:00) from Stage-storage curve = 150 Ft3.

At time t2= t1+∆t = (3-Mar-63-19:04:00), the volume of the cell will be:

𝑉(𝑡 + ∆𝑡) = 150+ (2.58 * ∆t (240 second)) = 768.22 Ft3.

This means that 768.22 Ft3 of water will be in Cell A after a 4-minute period.

HEC-RAS will then use the Elevation-Volume Relationship to find the new WSEL for

Cell A at t2 as shown in Figure 40. From the stage-storage curve for Cell A, the WSEL

after 4 minutes is equal 710.95. in other words, the 768.22 Ft3 of water that will be in Cell

A causes a water level change of 1.14 Ft.

76
Figure 40: Cell Volume Elevation Relationship Example 3.

77
APPENDIX B

The output from 1D model is in the form of levels at each cross-section

calculation point along the main channel. In 2D model, at each cell will compute the

levels of water. However, the comparison of stage hydrograph between 1D and 2D

results in 25 cross-sections as shown in figure 41.

Figure 41: The 25 Cross-Sections Location in 1D and 2D Model.


We will select three cross-sections to compare stage hydrographs. The first stage

hydrograph represents the largest different between the 1D unsteady state with storage

area with 2D flow area. The second stage hydrograph represents the average of the

different between two models. The third stage hydrograph explains the minimum

different between 1D and 2D model. We will repeat these comparisons for all the three

manning’s n value simulations as shown in these figures below.

78
XS 180

Figure 42: Large Difference of WSEL Between 1D and 2D Model n=0.035.

XS 130

Figure 43: Average Difference of WSEL Between 1D and 2D Model n=0.035.

79
XS 150

Figure 44: Minimum Difference of WSEL Between 1D and 2D Model n=0.035.

XS 150

Figure 45: Largest Difference of WSEL Between 1D and 2D Model n=0.06.

80
XS 100

Figure 46: Average Difference of WSEL Between 1D and 2D Model n=0.06.

XS 190

Figure 47: Minimum Difference of WSEL Between 1D and 2D Model n=0.06.

81
XS 180

Figure 48: Largest Difference of WSEL Between 1D and 2D Model n=0.1.

XS 110

Figure 49: Average Difference of WSEL Between 1D and 2D Model n=0.1.

82
XS 120

Figure 50: Minimum Difference of WSEL Between 1D and 2D Model n=0.1.

83

View publication stats

You might also like