Professional Documents
Culture Documents
Technical Reference
© 2008 Landmark Graphics Corporation
R5000.0.1
© 2008 Landmark Graphics Corporation
All Rights Reserved Worldwide
This publication has been provided pursuant to an agreement containing restrictions on its use. The publication is also
protected by Federal copyright law. No part of this publication may be copied or distributed, transmitted, transcribed,
stored in a retrieval system, or translated into any human or computer language, in any form or by any means,
electronic, magnetic, manual, or otherwise, or disclosed to third parties without the express written permission of:
Trademarks
3D Drill View, 3D Drill View KM, 3DFS, 3D Surveillance, 3DView, Active Field Surveillance, Active Reservoir Surveillance, ADC,
Advanced Data Transfer, ARIES, ARIES DecisionSuite, AssetConnect, Asset Decision Solutions, Asset Development Center,
AssetDirector, Asset Journal, AssetLink, AssetLink Advisor, AssetLink Director, AssetLink Observer, AssetObserver, AssetObserver
Advisor, Asset Performance, AssetPlanner, AssetSolver, AssetSolver Online, AssetView, BLITZPAK, CasingLife, CasingSeat,
COMPASS, Corporate Data Archiver, Corporate Data Store, Data Analyzer, DataManager, DataStar, DBPlot, Decision Management
System, DecisionSpace, DecisionsSpace 3D Drill View KM, DecisionSpace AssetLink, DecisionSpace AssetPlanner, DecisionSpace
AssetSolver, DecisionSpace AssetView 2D, DecisionSpace AssetView 3D, DecisionSpace Atomic Meshing, DecisionSpace Decision
Management Systems(DMS), DecisionSpace for Production, DecisionSpace Infrastructure, DecisionSpace Nexus, DecisionSpace
PowerModel, DecisionSpace PrecisionTarget, DecisionSpace Reservoir, DecisionSpace TracPlanner, DecisionSpace Well Seismic
Fusion, DecisionSpace WellSolver, DecisionSuite, DepthTeam, Depth Team, DepthTeam Explorer, Depth Team Explorer, DepthTeam
Express, Depth Team Express, DepthTeam Extreme, Depth Team Extreme, DepthTeam Interpreter, Depth Team Interpreter, Desktop
Navigator, DESKTOP-PVT, DESKTOP-VIP, DEX, DIMS, Discovery, Discovery Asset, Discovery FrameBuilder, DMS, Drillability Suite,
Drilling Desktop, DrillModel, Drill-to-the-Earth Model, Drillworks, Drillworks ConnectML, DSS, Dynamic Reservoir Management,
Dynamic Surveillance System, EarthCube, EDM, EDT, eLandmark, Engineer’s Data Model, Engineer's Desktop, Engineer’s Link,
ESP, Event Similarity Prediction, ezFault, ezModel, ezSurface, ezTracker, FastTrack, FG+, FieldPlan, Field Scenario Planner, FZAP!,
GeoAtlas, GeoDataLoad, GeoGraphix, GeoGraphix Exploration System, GeoLink, GeoProbe GF DataServer, GES, GES97,
GESXplorer, GMAplus, GMI Imager, GRIDGENR, Handheld Field Operator, HHFO, I2 Enterprise, iDIMS, IsoMap, iWellFile, i Wellfile,
KnowledgeSource, Landmark, Landmark Decision Center, Landmark & Design, Landmark Logo and Design, Landscape, Lattix,
LeaseMap, LMK Resources, LogEdit, LogM, LogPrep, Make Great Decisions, MathPack, MIMIC, MIMIC+, Model Builder,
MyLandmark, Nexus, Object MP, OpenBooks, OpenJournal, OpenSGM, OpenVision, OpenWells, OpenWire, OpenWorks,
OpenWorks Development Kit, OpenWorks Well File, OpenWorks Production, PAL, Parallel-VIP, PetroBank, PetroBank Explorer,
PetroBank Master Data Store, PetroWorks, PetroWorks Asset, PetroWorks Pro, PetroWorks ULTRA, PlotView, Point Gridding Plus,
Pointing Dispatcher, PostStack, PostStack ESP, PostStack Family, PowerCalculator, PowerExplorer, PowerExplorer Connect,
PowerGrid, PowerHub, Power Interpretation, PowerJournal, PowerModel, PowerView, PrecisionTarget, Presgraf, Pressworks,
PRIZM, Production Asset Manager, Production, PROFILE, Project Administrator, ProMAGIC, ProMAGIC Connect, ProMAGIC
Server, ProMAX, ProMAX 2D, ProMAX 3D, ProMAX 3DPSDM, ProMAX 4D, ProMAX Family, ProMAX MVA, ProMAX VSP, pSTAx,
Query Builder, Quick, Quick+, QUICKDIF, QuickWell, QuickWell+, QUIKRAY, QUIKSHOT, QUIKVSP, RAVE, RAYMAP, RAYMAP+,
Real Freedom, Real Time Asset Management Center, Real Time Decision Center, Real Time Operations Center, Real Time
Production Surveillance, Real Time Surveillance, Real-Time View, Reference Data Manager, Reservoir Framework Builder, RESev,
ResMap, RightTime, RTOC, SCAN, SeisCube, SeisMap, SeisModel, SeisSpace, SeisVision, SeisWell, SeisWorks, SeisWorks 2D,
SeisWorks 3D, SeisWorks PowerCalculator, SeisWorks PowerJournal, SeisWorks PowerSection, SeisWorks PowerView,
SeisXchange, Semblance Computation and Analysis, Sierra Family, SigmaView, SimConnect, SimConvert, SimDataStudio,
SimResults, SimResults+, SimResults+3D, SIVA+, SLAM, smartSECTION, Spatializer, SpecDecomp, StrataAmp, StrataMap,
StrataModel, StrataSim, StrataWorks, StratWorks, StratWorks 3D, StreamCalc, StressCheck, STRUCT, Structure Cube, Surf &
Connect, SynTool, SystemStart, SystemStart for Clients, SystemStart for Servers, System Start for Servers, SystemStart for Storage,
Tanks & Tubes, TDQ, Team Workspace, TERAS, The Engineer’s Desktop, Total Drilling Performance, TOW/cs, TOW/cs Revenue
Interface, TracPlanner, TracPlanner Xpress, Trend Form Gridding, Turbo Synthetics, VESPA, VESPA+, VIP, VIP-COMP, VIP-CORE,
VIPDataStudio, VIP-DUAL, VIP-ENCORE, VIP-EXECUTIVE, VIP-Local Grid Refinement, VIP-THERM, WavX, Web Editor, Wellbase,
Wellbore Planner, Wellbore Planner Connect, WELLCAT, Well Cost, WELLPLAN, Well Seismic Fusion, WellSolver, WellXchange,
WOW, Xsection, You're in Control. Experience the difference, ZAP!, and Z-MAP Plus are trademarks, registered trademarks or
service marks of Landmark Graphics Corporation.
All other trademarks, service marks and product or service names are the trademarks or names of their respective owners.
Note
The information contained in this document is subject to change without notice and should not be construed as a commitment by Landmark
Graphics Corporation. Landmark Graphics Corporation assumes no responsibility for any error that may appear in this manual. Some states
or jurisdictions do not allow disclaimer of expressed or implied warranties in certain transactions; therefore, this statement may not apply to you.
Landmark acknowledges that certain third party code has been bundled with, or embedded in, Landmark’s
software. The licensors of this third party code, and the terms and conditions of their respective licenses,
may be found at the following location:
..\Nexus-VIP5000.0.1\help\com\lgc\dspx\Third_Party.pdf
❖
Table of Contents
List of Figures
About This Manual
Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii
Audience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii
Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii
Related Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxv
Chapter 1
Aquifer Modeling
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-37
1.2 Analytic Model — The Carter-Tracy Aquifer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-38
1.3 Analytical Model-The Fetkovich Aquifer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-41
1.4 Numerical Aquifer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-43
Chapter 2
Boundary Flux Option
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-45
2.2 Flux Model Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-46
2.2.1 Determining Program Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-46
2.2.2 Coarse Model Run (Flux OUTPUT Mode) . . . . . . . . . . . . . . . . . . . . . . 2-48
2.2.3 Fine Model Run (Flux INPUT Mode) . . . . . . . . . . . . . . . . . . . . . . . . . . 2-48
2.3 Special Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-50
2.3.1 Flux Rates Based on Cumulative Fluxes . . . . . . . . . . . . . . . . . . . . . . . . 2-50
2.3.2 Vertical Distribution of Influx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-50
2.3.3 Efflux Partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-52
2.3.4 Output Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-52
Chapter 3
Corner-Point Geometry Option
3.1 Corner-Point Position Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-53
R5000.0.1 v
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 4
Dual Porosity Models
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-61
4.2 Matrix-Fracture Flow Exchange Term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-62
4.3 Matrix-Fracture Diffusion Term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-63
4.4 Treatment of Imbibition and Gravity Drainage . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-65
4.4.1 Pseudo Capillary-Pressure Treatment of Exchange Between Matrix and
Fractures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-66
4.4.2 Coats’ Method for Treatment of Gas-Oil Exchange Between Matrix and
Fractures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-68
4.4.3 The Case of lz Greater than ΔZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-70
4.5 Dual Permeability Versus Single Permeability . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-71
4.6 Partially Fractured Reservoirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-73
4.7 Dual Porosity Benchmarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-75
4.8 Simulator Input Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-82
4.8.1 Fracture Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-82
4.8.2 Effective Fracture Permeability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-83
4.8.3 Pore Compressibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-87
4.8.4 Matrix Block Size and Matrix-Fracture Transmissibility . . . . . . . . . . . . 4-88
4.8.5 Relative Permeabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-88
4.8.6 Pressure Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-89
4.8.7 Well Permeability-Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-90
4.9 Example of Dual Porosity Model Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-91
Chapter 5
End-Point Scaling
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-97
5.2 Normalized Saturation End Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-98
5.2.1 Two-Point Scaling Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-99
5.2.2 Three-Point Scaling Option (Default Option) . . . . . . . . . . . . . . . . . . . 5-100
5.2.3 Comparisons of Two- and Three-Point Scaling Options . . . . . . . . . . . 5-101
5.2.4 End-Point Consistency Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-103
vi R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 6
Equilibration
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-109
6.2 Gravity-Capillary Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-110
6.3 Saturation Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-113
6.3.1 GBC Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-113
6.3.2 INTSAT Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-113
6.3.3 VAITS Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-114
6.4 Capillary Pressure Adjustment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-117
6.5 Supercritical Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-118
Chapter 7
Faults
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-121
7.2 Assumptions and Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-123
7.3 Input Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-125
7.4 Method of Specifying Arbitrary Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-126
7.5 Fault Transmissibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-128
7.6 Noncontiguous Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-129
7.6.1 Connectivity Search Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-129
7.7 Input Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-132
7.8 Conductive (Leaky) Fault Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-134
7.8.1 Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-134
7.8.2 Input Requirement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-137
Chapter 8
Gas-Water, Water-Oil, and Black-Oil Options
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-139
8.2 Simplified Gas-Water Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-139
8.2.1 Data Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-140
8.2.2 Special Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8-140
R5000.0.1 vii
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 9
Governing Equations
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-143
9.2 IMPES Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-143
9.2.1 Hydrocarbon Species Mass Balance Equations . . . . . . . . . . . . . . . . . . 9-143
9.2.2 Overall Hydrocarbon Mass Balance Equation . . . . . . . . . . . . . . . . . . . 9-144
9.2.3 Water Mass Balance Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-144
9.2.4 Fugacity Equality Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-144
9.2.5 Overall Component Mole Fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-144
9.2.6 Mole Fraction Constraint Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-145
9.2.7 Saturation Constraint Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-145
9.2.8 Reduction of Simultaneous Equations and Selection of Primary Unknowns .
9-145
9.3 Fully Implicit Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-147
9.3.1 Relaxed Volume Balance Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9-147
Chapter 10
Grid Coarsening
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-149
10.2 Coarse Block Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10-149
Chapter 11
Horizontal and Inclined Well Model
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-153
11.2 Modeling Considerations - Pressure Loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-155
11.3 Modeling Considerations - Permeability-Thickness and Well Index . . . . . . . 11-158
11.4 Required Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-162
11.5 Conventions That Users Must Follow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-164
11.6 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-168
11.7 Available Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-172
Chapter 12
Hydraulically Fractured Well Option
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-173
viii R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 13
Hysteresis
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-179
13.2 Relative Permeability Hysteresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-180
13.2.1 Description of the Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-181
13.2.2 Input Data Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-184
13.3 Capillary Pressure Hysteresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-185
13.3.1 Input Data Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-190
Chapter 14
IMPES Stability
14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-191
14.2 Three-Phase Immiscible Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-191
14.2.1 Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-191
14.2.2 IMPES Difference Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-192
14.2.3 Total Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-192
14.2.4 Difference Equations in Terms of Fractional Mobility . . . . . . . . . . 14-193
14.2.5 Stability Analysis for Two-Phase Flow . . . . . . . . . . . . . . . . . . . . . . 14-193
14.2.6 Stability Analysis for Three-Phase Flow . . . . . . . . . . . . . . . . . . . . . 14-195
14.3 Two-Phase Compositional Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-200
14.3.1 Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-200
14.3.2 Difference Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-201
14.3.3 Stability Analysis of Total Flow Equation . . . . . . . . . . . . . . . . . . . . 14-204
14.3.4 Stability Analysis for Other Dependent Variables . . . . . . . . . . . . . . 14-204
14.3.5 Final Note . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14-210
Chapter 15
Local Grid Refinement
15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-211
15.2 Major Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-212
15.3 Standard VIP-EXECUTIVE Features Available with LGR . . . . . . . . . . . . . . 15-213
15.4 Types of Refinements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-214
15.4.1 Cartesian Refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-214
R5000.0.1 ix
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 16
Miscible Options
16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-263
16.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-264
16.3 Effective Relative Permeabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-265
16.4 Miscible-Immiscible Transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-267
16.5 Effective Viscosities, Densities, And Capillary Pressures . . . . . . . . . . . . . . . 16-269
16.6 PVT and VLE Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-271
16.6.1 Three-Component Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-271
16.6.2 Four-Component Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16-271
Chapter 17
Non-Darcy Gas Flow
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17-273
17.2 Pressure-Dependent Gas Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17-274
x R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 18
Numerical Solution
18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-281
18.2 Discretization of the Mass Balance Equations . . . . . . . . . . . . . . . . . . . . . . . . 18-281
18.2.1 Overall Hydrocarbon Mass Balance Equation . . . . . . . . . . . . . . . . . 18-283
18.2.2 Water Balance Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-284
18.3 Linearization of the Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-286
18.4 Solution of Non-Linear Systems of Equations . . . . . . . . . . . . . . . . . . . . . . . . 18-289
18.5 Solution of Linear System of Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-292
18.5.1 Gauss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-292
18.5.2 BLITZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-292
18.5.3 EXCEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18-295
Chapter 19
Optimal Material Balance Option
19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-297
19.2 Unknown Update and Convergence Control . . . . . . . . . . . . . . . . . . . . . . . . . . 19-298
19.3 Reordering Gridblocks Based On Fluid Type . . . . . . . . . . . . . . . . . . . . . . . . . 19-304
19.4 Partial Jacobian Update . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-306
19.5 Criteria For Single-phase Stability Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-308
19.6 Vectorization of Saturation Pressure Calculation and Flash . . . . . . . . . . . . . . 19-309
19.7 Input Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-311
Chapter 20
Parallel Computing
20.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-313
20.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-313
20.2.1 Local Grid Refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-314
20.2.2 Parallelization Via Local Grid Refinement . . . . . . . . . . . . . . . . . . . 20-314
20.2.3 Automatic Grid Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-316
20.2.4 Well Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-318
20.2.5 The Parallel Linear Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-319
20.2.6 Parallel Computers Used for Verification of PARALLEL-VIP . . . . 20-324
R5000.0.1 xi
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 21
Phase Equilibrium Calculations
21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-345
21.2 The Standard VIP-EXECUTIVE Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-347
21.2.1 Determining Number of Phases in a Gridblock . . . . . . . . . . . . . . . . 21-347
21.3 Gibbs Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-349
21.3.1 Phase Stability Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-350
21.3.2 Gibbs Energy Minimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21-351
21.3.3 Implementation of the Gibbs Option in VIP-EXECUTIVE . . . . . . . 21-351
Chapter 22
Polymer Modeling Option
22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-353
22.2 Description of Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-354
22.3 Material Conservation Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-355
22.3.1 Polymer Solution Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-356
22.3.2 Polymer Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-358
22.3.3 Permeability Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-359
22.3.4 Polymer Inaccessible Pore Volume . . . . . . . . . . . . . . . . . . . . . . . . . 22-360
22.3.5 Cation Exchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-360
22.3.6 Effective Polymer Viscosity Used in the Well Calculations . . . . . . 22-361
22.4 Instantaneous Gel Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22-363
Chapter 23
PVT Representation
23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-365
23.2 Rock Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-365
23.3 Water Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-365
23.4 Hydrocarbon Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-366
23.4.1 Compositional Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-366
23.4.2 Black Oil Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-375
23.4.3 Gas Condensates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-380
23.4.4 Miscible Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-380
23.5 Equation of State Interpolation Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23-381
xii R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 24
Relative Permeability and Capillary Pressure Adjustments Near the
Critical Point
24.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24-407
Chapter 25
Saturation Function
25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-409
25.2 Water Saturation Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-410
25.3 Gas Saturation Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-412
25.4 Three-Phase Oil Relative Permeability Models . . . . . . . . . . . . . . . . . . . . . . . . 25-413
25.4.1 Stone’s Model I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-413
25.4.2 Stone’s Model II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-414
25.4.3 Saturation Weighted Interpolation Model . . . . . . . . . . . . . . . . . . . . 25-414
25.4.4 Guidelines for Selecting the Models . . . . . . . . . . . . . . . . . . . . . . . . . 25-415
25.5 Consistency Checks For Saturation Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-417
25.6 Directional Relative Permeability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-418
25.6.1 Input Data Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-418
25.7 Gas Remobilization Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-420
25.7.1 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-420
25.7.2 Input Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25-422
Chapter 26
Separators
26.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26-425
26.2 Separator Battery Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26-426
26.3 Definition of Separator Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26-427
26.4 Mass Balance Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26-428
26.5 Phase Equilibrium Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26-429
26.5.1 Compositional Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26-429
R5000.0.1 xiii
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 27
Simulator Performance
27.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-433
27.2 Timestep Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-433
27.2.1 Primary Timestep Controls (DT) . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-434
27.2.2 Control Convergence Failures And Timestep Cuts (TCUT) . . . . . . 27-434
27.2.3 IMPES Stability Controls (IMPSTAB) . . . . . . . . . . . . . . . . . . . . . . 27-435
27.2.4 Optimal Material Balance Option (OPTMBL) . . . . . . . . . . . . . . . . . 27-435
27.3 Non-Linear Iteration Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-436
27.3.1 Outer Iteration Controls (ITNLIM) . . . . . . . . . . . . . . . . . . . . . . . . . 27-436
27.3.2 Convergence Tolerance (TOLD) . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-437
27.3.3 Convergence Tolerance (TOLR) . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-438
27.3.4 Maximum Allowable Material Balance Error (ABORT) . . . . . . . . . 27-438
27.3.5 Minimum BHP Damping Factor (CBHPMN) . . . . . . . . . . . . . . . . . 27-438
27.3.6 Gas Percolation Control (GASPERC) . . . . . . . . . . . . . . . . . . . . . . . 27-438
27.4 Simulator Performance Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-439
27.4.1 Timestep Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-439
27.4.2 Simulation Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-441
27.4.3 Non-Linear Iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-442
27.4.4 Linear Iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-443
27.5 Run Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-445
27.5.1 Timestep and Iteration Control Parameters . . . . . . . . . . . . . . . . . . . 27-445
27.5.2 Selection of Formulation and Solver . . . . . . . . . . . . . . . . . . . . . . . . 27-446
27.5.3 Auto-adjustable Linear Tolerance . . . . . . . . . . . . . . . . . . . . . . . . . . 27-446
27.5.4 Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-447
27.5.5 Analyzing a Run . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-448
27.6 Estimating Run Durations and Memory Requirements . . . . . . . . . . . . . . . . . . 27-450
27.6.1 Simulator Run Durations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-450
27.6.2 Performance of the Simulator on Various CPUs . . . . . . . . . . . . . . . 27-452
27.6.3 Memory Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-452
27.6.4 Using the Simulator to Determine Memory Requirements (STORAGE) . .
27-454
Chapter 28
Single-Well Gridded Wellbore Simulation
28.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28-455
xiv R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
28.2 Calculation of Vertical Flow Coefficients from the Wellbore Flow Equations 28-456
28.3 Modifications for Liquid Slippage or Holdup . . . . . . . . . . . . . . . . . . . . . . . . . 28-458
28.4 Application of Minimum Lift Velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28-459
28.5 Data Structure and Definition Changes for VIP-CORE . . . . . . . . . . . . . . . . . 28-460
28.6 Data Structure and Definition Changes for the Simulation Module . . . . . . . . 28-461
Chapter 29
Surface Pipeline Network Options
29.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-463
29.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-464
29.1.2 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-465
29.2 Hydraulic Models of Flow Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-466
29.2.1 General Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-466
29.2.2 Flow Modeling in Horizontal, Vertical, or Inclined Pipes . . . . . . . . 29-467
29.2.3 Application of Hydraulic Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-472
29.2.4 Application of Look-Up Tables for Pressure Gradient Determination In
Pipes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-473
29.2.5 Valve Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-474
29.3 Surface Pipeline Network Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-475
29.3.1 Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-475
29.3.2 Pressure and Rate Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-481
29.3.3 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-482
29.3.4 Solution Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-483
29.3.5 Simultaneous Modeling of Multiphase Fluid Flow in Reservoir and Surface
Pipeline Network System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-484
Chapter 30
Total Compressibility Checks
30.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30-487
30.2 Reasonableness Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30-488
30.3 Saturated Oil Compressibility Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30-489
30.4 Undersaturated Oil Compressibility Check . . . . . . . . . . . . . . . . . . . . . . . . . . . 30-490
Chapter 31
Tracking Calculations In VIP-EXECUTIVE
31.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31-491
31.2 Tracking Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31-492
31.3 Tracking in a Reservoir . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31-493
31.3.1 Variables to Describe Equity Fluids . . . . . . . . . . . . . . . . . . . . . . . . . 31-493
R5000.0.1 xv
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 32
Tracer Analysis Option
32.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-499
32.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-500
32.3 Tracer Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-500
32.4 Simulation of Tracer Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-502
32.4.1 Particle Tracking Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-503
32.5 Implementation in VIP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-504
32.5.1 Particle Unit Cube Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-505
32.5.2 Particle Velocity Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-505
32.5.3 Particle Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-508
32.6 Validation of Tracer Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-510
32.6.1 One-Dimensional Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-510
32.6.2 Two-Dimensional Areal Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-512
32.7 Interpretation of Tracer Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-516
Chapter 33
Transmissibility Calculations
33.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-517
33.2 Cartesian Coordinate System Transmissibility . . . . . . . . . . . . . . . . . . . . . . . . 33-517
33.2.1 Standard Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-517
33.2.2 Fault Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-519
33.2.3 Nine Point Transmissibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-520
33.3 Radial System Transmissibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-524
33.3.1 Standard Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-524
33.3.2 Fault Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-525
33.4 Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-525
Chapter 34
Unit Conventions
34.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34-529
34.2 Conversion Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34-531
xvi R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 35
Velocity-Dependent Relative Permeabilties
35.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35-533
35.2 Capillary Number Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35-534
35.3 Forchheimer (Non-Darcy) Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35-537
35.4 Velocity Effects around the Production Well . . . . . . . . . . . . . . . . . . . . . . . . . 35-540
8. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35-545
Chapter 36
Vertical Equilibrium
36.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36-549
36.2 VE Option in Rectangular or Radial Grid Systems . . . . . . . . . . . . . . . . . . . . . 36-549
36.2.1 Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36-549
36.2.2 Two-Phase Gas-Oil Pseudo Functions . . . . . . . . . . . . . . . . . . . . . . . 36-553
36.2.3 Two-Phase Water-Oil Pseudo Functions . . . . . . . . . . . . . . . . . . . . . 36-554
36.2.4 VE Directional Relative Permeability . . . . . . . . . . . . . . . . . . . . . . . 36-555
36.3 VE Option in Corner-Point Grid System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36-556
36.3.1 Features of the Corner Point VE Options . . . . . . . . . . . . . . . . . . . . . 36-556
36.3.2 Enhanced VE Procedure with Segregated Fluids (VEWO,VEGO) . 36-557
36.3.3 The VE (VEWO, VEGO) Simulation Procedure . . . . . . . . . . . . . . . 36-560
36.3.4 The Capillary-Gravity Equilibrium Option (VEITS) . . . . . . . . . . . . 36-561
36.3.5 The VEITS Simulation Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . 36-562
Chapter 37
Water Tracking Option
37.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37-565
Chapter 38
Well Inflow Performance
38.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-569
38.2 Wellbore Flow Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-570
38.3 Well Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-572
38.4 Bottom-Hole Pressure Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-575
38.5 Use of Inflow and Outflow Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-576
38.6 Tubing Performance Curve Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-578
R5000.0.1 xvii
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 39
Well Management Features
39.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-579
39.2 Well Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-579
39.2.1 Minimum Data Requirements for Production Well . . . . . . . . . . . . . 39-579
39.2.2 Minimum Data Requirements for Injection Wells . . . . . . . . . . . . . . 39-580
39.2.3 Vertical and Deviated Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-581
39.3 Well Completion Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-582
39.3.1 FPERF Card . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-582
39.4 Production Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-585
39.4.1 Type of Producers and Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-585
39.4.2 Production Well Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-585
39.4.3 ONTIME Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-587
39.4.4 Pressure Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-587
39.4.5 Well Status Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-588
39.4.6 Testing Shutin Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-589
39.4.7 Special Features for Gas Producers . . . . . . . . . . . . . . . . . . . . . . . . . 39-590
39.5 Injection Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-591
39.5.1 Type of Injectors and Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-591
39.5.2 Injection Well Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-592
39.5.3 Water Injectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-592
39.5.4 Gas Injectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-592
39.5.5 WAG Injectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-593
39.5.6 ONTIME Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-593
39.5.7 Well Status Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-594
39.5.8 Testing Shutin Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-594
39.6 Wellbore Gradient Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-595
39.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-595
39.6.2 Mobility Weighted Averaging Method . . . . . . . . . . . . . . . . . . . . . . . 39-595
39.6.3 Volume Balancing Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-596
39.6.4 Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-599
39.7 Wellbore Crossflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-600
39.7.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-600
39.7.2 Crossflow Implementation in BLITZ . . . . . . . . . . . . . . . . . . . . . . . . 39-602
39.8 Well Management Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-603
39.8.1 Production Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-603
39.8.2 Minimum Production Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-604
39.8.3 Injection Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-605
39.8.4 Minimum Injection Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-605
39.9 Gas Reinjection and Gas Handling Features . . . . . . . . . . . . . . . . . . . . . . . . . . 39-606
39.9.1 Shrinkage, Fuel, and Sales Gas Specification . . . . . . . . . . . . . . . . . . 39-606
39.9.2 Makeup Gas Specification and Composition . . . . . . . . . . . . . . . . . . 39-606
xviii R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
R5000.0.1 xix
VIP-EXECUTIVE® Technical Reference Guide Landmark
Appendix A
Well Models
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-649
A.2 Definition of Well Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-650
A.2.1 Alternative Definition of Well Index . . . . . . . . . . . . . . . . . . . . . . . . . A-650
A.2.2 Comparison with Productivity Index . . . . . . . . . . . . . . . . . . . . . . . . . A-651
A.3 One-Dimensional Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-651
A.3.1 Linear Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-651
A.3.2 Radial Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-652
A.4 Two-Dimensional Areal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-654
A.4.1 Well in Center of Square Gridblock . . . . . . . . . . . . . . . . . . . . . . . . . A-654
A.4.2 General Definition of rb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-656
A.4.3 An Approximate Derivation of rb for Square Well Block . . . . . . . . . A-656
A.4.4 Well in Center of Rectangular Gridblock . . . . . . . . . . . . . . . . . . . . . A-658
A.4.5 Well in Center of Block in Anisotropic Rectangular Grid . . . . . . . . A-659
A.4.6 Single Well Arbitrarily Located in Isolated Well Block . . . . . . . . . . A-661
A.4.7 Multiple Wells in Same Isolated Well Block . . . . . . . . . . . . . . . . . . A-662
A.4.8 Two Wells With Same Rate in Adjacent Blocks . . . . . . . . . . . . . . . . A-664
A.4.9 Single Well in Edge Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-665
A.4.10 Single Well Exactly on Edge of Grid . . . . . . . . . . . . . . . . . . . . . . . A-666
A.4.11 Single Well Exactly at Corner of Grid . . . . . . . . . . . . . . . . . . . . . . . A-667
A.4.12 Single Well Arbitrarily Located in Corner Block . . . . . . . . . . . . . . A-668
A.5 Incorporating Skin into Well Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-669
A.5.1 Derivation of Skin Due to Altered Permeability . . . . . . . . . . . . . . . . A-669
A.5.2 Including Mechanical Skin In Well Index . . . . . . . . . . . . . . . . . . . . . A-670
A.5.3 Skin Due to Restricted Entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-671
A.5.4 Effects of Restricted Entry on Well Index . . . . . . . . . . . . . . . . . . . . . A-673
A.6 Well Index from Productivity Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-675
A.7 Non-Darcy Gas Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-678
A.7.1 Effect of Pressure-Dependent Gas Properties on Well Model . . . . . . A-678
A.7.2 Rate-Dependent Skin Factor in Well Model . . . . . . . . . . . . . . . . . . . A-679
A.8 Horizontal Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-680
A.9 Inclined Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-682
xx R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Appendix B
Corner-Point Geometry
B.1 Mapping of Gridblock to Unit Cube . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-683
B.1.1 Two-Dimensional Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-683
B.1.2 Three-Dimensional Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-685
B.2 Volumetric Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-687
B.3 Integration by Gaussian Quadrature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-690
B.4 Transmissibility Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-691
B.4.1 Calculation of Transmissibility in 2D . . . . . . . . . . . . . . . . . . . . . . . . . B-691
B.4.2 Calculation of Transmissibility in 3D (HARINT) . . . . . . . . . . . . . . . . B-698
B.4.3 Calculation of Transmissibility in 3D by the NEWTRAN Option . . . B-702
B.4.4 Calculation of Full Transmissibility Between Gridblocks . . . . . . . . . . B-703
B.4.5 Choice Between HARINT and NEWTRAN Options . . . . . . . . . . . . . B-705
Appendix C
References
Subject Index
R5000.0.1 xxi
VIP-EXECUTIVE® Technical Reference Guide Landmark
xxii R5000.0.1
❖
List of Figures
About This Manual
Chapter 1
Aquifer Modeling
Figure 1-1: Aquifer Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-38
Figure 1-2: Natural Extension of Reservoir Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-43
Figure 1-3: Aquifer with Arbitrary Connected Blocks . . . . . . . . . . . . . . . . . . . . . . . . 1-43
Chapter 2
Boundary Flux Option
Figure 2-1: Schematic Representation of Boundary Flux Feature . . . . . . . . . . . . . . . 2-46
Chapter 3
Corner-Point Geometry Option
Figure 3-1: Fault Block Connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-57
Figure 3-2: Rectangle Used In Computing A(λ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-58
Chapter 4
Dual Porosity Models
Figure 4-1: Idealization of Naturally Fractured Reservoir . . . . . . . . . . . . . . . . . . . . . 4-61
Figure 4-2: Fluid Exchange Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-65
Figure 4-3: Matrix Blocks and Fractures in a Gridblock . . . . . . . . . . . . . . . . . . . . . . 4-67
Figure 4-4: Two Dual Porosity Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-71
Figure 4-5: Well Perforations to Rock and Fracture . . . . . . . . . . . . . . . . . . . . . . . . . . 4-72
Figure 4-6: Representation of a Partially Fractured Reservoir . . . . . . . . . . . . . . . . . . 4-74
Figure 4-7: Gas/Oil Drainage Recovery for a 10x10x10 ft. Matrix Block
(Coats’ Figure 15) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-75
Figure 4-8: Gas/Oil Drainage Recovery for a 10x10x10 ft. Matrix Block
(Coats’ Figure 16) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-76
R5000.0.1 xxiii
VIP-EXECUTIVE® Technical Reference Guide Landmark
Figure 4-9: Gas/Oil Drainage Recovery for a 1x1x1 ft. Matrix Block
(Coats’ Figure 17) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-76
Figure 4-10: Gas/Oil Gravity Drainage for a 4 ft. Matrix Block
(Coats’ Figure 19) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-77
Figure 4-11: Gas/Oil Gravity Drainage for a 1 ft Matrix Block
(Coats’ Figure 20) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-77
Figure 4-12: Sixth SPE Comparative Solution Project
Single Block - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-78
Figure 4-13: Sixth SPE Comparative Solution Project
Depletion - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-78
Figure 4-14: Sixth SPE Comparative Solution Project
Depletion - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-79
Figure 4-15: Sixth SPE Comparative Solution Project
Depletion - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-79
Figure 4-16: Sixth SPE Comparative Solution Project
Reinjection - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-80
Figure 4-17: Sixth SPE Comparative Solution Project
Reinjection - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-80
Figure 4-18: Sixth SPE Comparative Solution Project
Reinjection - Zero PCF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-81
Figure 4-19: Fracture and Matrix Continuum Porosities in VIP . . . . . . . . . . . . . . . . 4-82
Figure 4-20: Kelton’s Permeameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-84
Figure 4-21: Simplified Layered Fracture Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-84
Figure 4-22: Pore Compressibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-87
Figure 4-23: Matrix Block Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-88
Figure 4-24: Fracture Relative Permeabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-89
Figure 4-25: Permeability Dependence on Confining Pressure . . . . . . . . . . . . . . . . . 4-89
Figure 4-26: Well Perforations to Rock and Fracture . . . . . . . . . . . . . . . . . . . . . . . . . 4-90
Chapter 5
End-Point Scaling
Chapter 6
Equilibration
Figure 6-1: Gravity-Capillary Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-110
Figure 6-2: Multiple Water-Oil Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-111
Figure 6-3: Block Representation in the INTSAT Option . . . . . . . . . . . . . . . . . . . . 6-114
xxiv R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 7
Faults
Figure 7-1: Part of a Structure Map of a Petroleum Reservoir . . . . . . . . . . . . . . . . . 7-121
Figure 7-2: Different Ways of Modeling Faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-122
Figure 7-3: Fault Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-123
Figure 7-4: Input Structure of Typical Fault Data Input . . . . . . . . . . . . . . . . . . . . . . 7-125
Figure 7-5: Specifying a Single Fault Connection . . . . . . . . . . . . . . . . . . . . . . . . . . 7-126
Figure 7-6: Arbitrary Connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-127
Figure 7-7: Transmissibility for Standard Connection . . . . . . . . . . . . . . . . . . . . . . . 7-128
Figure 7-8: Example of Faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7-132
Chapter 8
Gas-Water, Water-Oil, and Black-Oil Options
Chapter 9
Governing Equations
Chapter 10
Grid Coarsening
Chapter 11
Horizontal and Inclined Well Model
Figure 11-1: Vertical and Horizontal Wells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-153
Figure 11-2: A Well Segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-155
Figure 11-3: Inclined Well Segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-159
Figure 11-4: Ellipsoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-161
Figure 11-5: A Production Well Example Input . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-162
Figure 11-6: Inclined Well Segment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-164
Figure 11-7: ANGLV, q, in various situations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-165
Figure 11-8: Example of Multiple Well Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11-166
Figure 11-9: Well With Noncontiguous Perforations and Sample VIP Input . . . . 11-168
Figure 11-10: Areal and Cross Sections of the Model . . . . . . . . . . . . . . . . . . . . . . 11-169
Figure 11-11: Block Number and Well Section Length . . . . . . . . . . . . . . . . . . . . . 11-170
R5000.0.1 xxv
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 12
Hydraulically Fractured Well Option
Figure 12-1: Areal View of Vertically Fractured Well Centered in a Rectangular Drainage
Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12-173
Figure 12-2: Areal Finite-difference Grid of One Quadrant of the Drainage Area. 12-174
Figure 12-3: Vertical Cross-section of the Hydraulic Fracture (at Y=1). . . . . . . . . 12-174
Chapter 13
Hysteresis
Figure 13-1: Nonwetting Phase Relative Permeability Curves for Hysteresis . . . . 13-180
Figure 13-2: User-Defined Nonwetting Phase Imbibition-Drainage Curve . . . . . . 13-183
Figure 13-3: Water-Oil Capillary Pressure Bounding and Scanning Curves . . . . . 13-185
Figure 13-4: Water Saturation Reversal During Secondary Drainage . . . . . . . . . . 13-186
Figure 13-5: Water Saturation Reversal During a Primary Drainage . . . . . . . . . . . 13-187
Figure 13-6: Second Water Saturation Reversal . . . . . . . . . . . . . . . . . . . . . . . . . . . 13-189
Chapter 14
IMPES Stability
Chapter 15
Local Grid Refinement
Figure 15-1: Example of Cartesian Refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-214
Figure 15-2: Example of Radial Refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-215
Figure 15-3: Radial Grids - RADZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-216
Figure 15-4: Radial Grids - RADX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-216
Figure 15-5: Radial Grids - RADY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-217
Figure 15-6: Typical GRIDGENR Display with Radial Refinement . . . . . . . . . . . 15-220
Figure 15-7: Cartesian Refinement with Imbedded Radial Refinement . . . . . . . . . 15-221
Figure 15-8: Levels of Refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-224
Figure 15-9: Termination of Grid Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-225
Figure 15-10: Grid Conformance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-225
Figure 15-11: Nesting of Refinements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-226
Figure 15-12: Isolation of Refinements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-227
Figure 15-13: Case 1 - No Refinements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15-234
Figure 15-14: Case 2 - 3x3 Cartesian Refinements Around Each Well . . . . . . . . . 15-235
xxvi R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Figure 15-15: Case 3 - 4x4 Radial Refinements Around Each Well . . . . . . . . . . . 15-236
Figure 15-16: Comparison of Gas-Oil Ratio Performance . . . . . . . . . . . . . . . . . . . 15-238
Figure 15-17: Comparison of Water-cut Performance . . . . . . . . . . . . . . . . . . . . . . 15-239
Chapter 16
Miscible Options
Figure 16-1: Damping Function for Miscible-Immiscible Switch . . . . . . . . . . . . . 16-267
Chapter 17
Non-Darcy Gas Flow
Chapter 18
Numerical Solution
Chapter 19
Optimal Material Balance Option
Figure 19-1: GridblockReordering Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19-305
Chapter 20
Parallel Computing
Figure 20-1: Communication Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-315
Figure 20-2: LGR Grid Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-316
Figure 20-3: Example Grid For Automatic Grid . . . . . . . . . . . . . . . . . . . . . . . . . . 20-317
Figure 20-4: Automatically Decomposed Grid Decomposition . . . . . . . . . . . . . . . 20-318
Figure 20-5: Strip Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-320
Figure 20-6: LGR Multilevel Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20-321
Figure 20-7: Strip Decomposition Linear Solver Results . . . . . . . . . . . . . . . . . . . . 20-322
Figure 20-8: Composite Grid Linear Solver Results . . . . . . . . . . . . . . . . . . . . . . . 20-322
Figure 20-9: Effect of Decomposition on Strip and Composite Grid Linear Solvers . 20-
323
Figure 20-10: Grid and Structure for Small Homogeneous Example Problem . . . 20-326
Figure 20-11: Grid and Structure for 3D Salt Dome . . . . . Heterogeneous Model 20-328
Figure 20-12: Grid and Structure for Sloping Fault Heterogeneous Model . . . . . . 20-328
Figure 20-13: Grid and Structure for Compositional Load-Balancing Example . . 20-329
Figure 20-14: Performance of T3D on Small Homogeneous Model . . . . . . . . . . . 20-331
Figure 20-15: Performance of SGI Power Challenge on Small Homogeneous Model
20-332
R5000.0.1 xxvii
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 21
Phase Equilibrium Calculations
Chapter 22
Polymer Modeling Option
Chapter 23
PVT Representation
Figure 23-1: Comparison of EOS and EOSINT Model Results - Fifth SPE Comparative
xxviii R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 24
Relative Permeability and Capillary Pressure Adjustments Near the
Critical Point
Chapter 25
Saturation Function
Figure 25-1: Typical Saturation and Relative Permeability Path of a Gridblock Subject to
Gas Injection followed by Water Injection and Pressure Blowdown . . . . . . . . . . . 25-422
Chapter 26
Separators
Figure 26-1: Sample Separator Battery Configuration . . . . . . . . . . . . . . . . . . . . . . 26-426
Chapter 27
Simulator Performance
Figure 27-1: Convergence Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27-436
Chapter 28
Single-Well Gridded Wellbore Simulation
Chapter 29
Surface Pipeline Network Options
Figure 29-1: Pressure Gradient in Pipes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-469
Figure 29-2: An Example of a Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29-477
Figure 29-3: An Example of a Surface Pipeline Network System . . . . . . . . . . . . . 29-478
R5000.0.1 xxix
VIP-EXECUTIVE® Technical Reference Guide Landmark
Chapter 30
Total Compressibility Checks
Chapter 31
Tracking Calculations In VIP-EXECUTIVE
Chapter 32
Tracer Analysis Option
Figure 32-1: Comparison With Analytical Solution . . . . . . . . . . . . . . . . . . . . . . . . 32-511
Figure 32-2: Effect of the Number of Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-511
Figure 32-3: Zero Dispersion Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-512
Figure 32-4: Effect of the Grid System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-513
Figure 32-5: Effect of the Number of Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-514
Figure 32-6: Effect of Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32-515
Chapter 33
Transmissibility Calculations
Figure 33-1: Bedding Plane Lengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33-518
Figure 33-2: Shared Thickness Calculation Example . . . . . . . . . . . . . . . . . . . . . . . 33-520
Figure 33-3: Nine-Point Transmissibilities Deriavation . . . . . . . . . . . . . . . . . . . . . 33-521
Chapter 34
Unit Conventions
Chapter 35
Velocity-Dependent Relative Permeabilties
Chapter 36
Vertical Equilibrium
Figure 36-1: Block Span THVE Versus Block Thickness TH . . . . . . . . . . . . . . . . 36-550
Figure 36-2: Block Angles Produced by the Old Dip Angle Method . . . . . . . . . . 36-551
Figure 36-3: Block Angles Produced by the New Dip Angle Method . . . . . . . . . . 36-552
Figure 36-4: Gridblock Saturation Distributions in the VEWO, VEGO Option . . 36-558
Figure 36-5: The VE Initialization Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36-559
Chapter 37
Water Tracking Option
Figure 37-1: Variation of Fractional Flow of Extraneous Water Types
xxx R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Chapter 38
Well Inflow Performance
Figure 38-1: Inflow and Outflow Curve for a Production Well Showing Intersection Point
38-569
Chapter 39
Well Management Features
Figure 39-1: Schematic Representation of Well Management Levels . . . . . . . . . . 39-603
Figure 39-2: Schematic Diagram of a Gas Handling Loop with the Major Gas Sales Option
39-610
Figure 39-3: Intersection of the Inflow Performance Curve (-PI) and the Lift Curves for the
Various Gaslift Gas Rates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39-624
Figure 39-4: Calculated Gaslift Performance Curve. . . . . . . . . . . . . . . . . . . . . . . . 39-625
Appendix A
Well Models
Figure A-1: Fine Grid Around A Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-649
Figure A-2: One-Dimensional, Linear, Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-651
Figure A-3: One-Dimensional, Radial, Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-653
Figure A-4: Well in Square Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-654
Figure A-5: Numerical Solutions for Pressure Plotted vs Radius . . . . . . . . . . . . . A-655
Figure A-6: Well in Edge Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-665
Figure A-7: Well on Edge of Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-666
Figure A-8: Well at Corner of Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-667
Figure A-9: Well in Corner Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-668
Figure A-10: Radial Flow With Zone of Altered Permeability . . . . . . . . . . . . . . . A-669
Figure A-11: Pseudo Skin Factor, after Brons and Marting . . . . . . . . . . . . . . . . . . A-671
Figure A-12: Examples of Partial Well Completion . . . . . . . . . . . . . . . . . . . . . . . A-672
Figure A-13: Partial Well Completion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-673
Figure A-14: Partial Well Completion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-674
Figure A-15: Typical Plot of ρg/μg vs Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . A-679
Figure A-16: Effect of αe and z/Δz on rbe for Centered Well . . . . . . . . . . . . . . . . . A-681
R5000.0.1 xxxi
VIP-EXECUTIVE® Technical Reference Guide Landmark
Appendix B
Corner-Point Geometry
Figure B-1: Mapping of Quadrilateral to Unit Square . . . . . . . . . . . . . . . . . . . . . . . B-683
Figure B-2: The Eight Corners of a Gridblock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-685
Figure B-3: Elemental Tube for Determining Thickness . . . . . . . . . . . . . . . . . . . . . B-687
Figure B-4: Elemental Tube for Determining DX . . . . . . . . . . . . . . . . . . . . . . . . . . B-688
Figure B-5: Orthogonal Gridblocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-691
Figure B-6: Nonorthogonal Gridblocks with Parallel Sides . . . . . . . . . . . . . . . . . . . B-692
Figure B-7: Gridblockwith Non-Parallel Sides . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-693
Figure B-8: Slice Parametrized by u . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-694
Figure B-9: Collection of Tubes and Slices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-694
Figure B-10: Tube, Slice, and Chunk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-695
Figure B-11: Gridblockwith Nonparallel Sides . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-696
Figure B-12: Lines a and b Defining Plane Tangent to Slice . . . . . . . . . . . . . . . . . . B-700
Figure B-13: Projections of Right Face onto the Three Coordinate Planes . . . . . . . B-703
Figure B-14: Fault Block Connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-704
Figure B-15: Radial problem, with Δθ = 45° . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-705
Figure B-16: Comparison of HARINT and NEWTRAN for Angular Transmissibility Be-
tween Blocks 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-707
Figure B-17: Comparison of HARINT and NEWTRAN for Angular Transmissibility Be-
tween Blocks 3 and 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-708
Figure B-18: Comparison of HARINT and NEWTRAN for Radial Tranmissibility . . . B-
709
Appendix C
References
Subject Index
xxxii R5000.0.1
Preface
Purpose
This document provides detailed technical and application information on the
VIP-EXECUTIVE® simulation software package; i.e., the initialization module
VIP-CORE® and the simulation modules VIP-ENCORE®, VIP-COMP®, VIP-
DUAL®, and VIP-POLYMER™. It is written for VIP-EXECUTIVE® users who
wish to know more about the underlying calculations performed by the simulator
and want to understand better how data are being processed.
Audience
This manual is intended to be used by reservoir simulation engineers who require
a more detailed discussion of the features of VIP-EXECUTIVE than provided in
the Keyword Reference and Getting Started manuals.
Organization
The information in this manual is arranged in a logical manner for maximum ease
of use. The main technical topics have their own chapter which are arranged in
alphabetical order. The following chapters are included:
■ Aquifer Modeling.
R5000.0.1 xxxiii
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ End-Point Scaling.
■ Equilibration.
■ Faults.
■ Governing Equations.
■ Grid Goarsening
■ Hysteresis.
■ IMPES Stability.
■ Miscible Options.
■ Numerical Solution.
■ Parallel Computing
■ PVT Representation.
■ Saturation Function.
■ Separators.
■ Simulator Performance
xxxiv R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ Transmissibility Calculations.
■ Unit Conventions.
■ Vertical Equilibrium.
■ Appendix C - References.
Related Documentation
The following manuals provide more information related to the material in this
manual. For more information, please consult the appropriate manual listed below.
R5000.0.1 xxxv
VIP-EXECUTIVE® Technical Reference Guide Landmark
xxxvi R5000.0.1
Chapter
1
Aquifer Modeling
1.1 Introduction
When modeling aquifers, VIP-EXECUTIVE always treats outer boundaries of a
reservoir model as sealing barriers to flow. The influx/efflux of fluids from
outside the grid is treated by source/sink terms.
ra
A Aquifer
re
Reservoir
θ B
The water influx from the aquifer into the reservoir is the response to pressure
changes at this boundary. Water flow in the aquifer is described by Equation 1-1
with the initial and boundary conditions:
φμ w c t ∂p ∂- ∂p
-------------- ------ = 1--- ---- r ------ (1-1)
k ∂t r ∂r ∂r
p t=0 = pi (1-2)
p r = re
= pe ( t ) (1-3)
∂p
------ = 0 (1-4)
∂r r = ra
Van Everdingen and Hurst1 derived the following superposition form of the
solution of the above problem:
td
dp e ( t d1 )
W ( t ) = – B ∫ -------------------- Q ( t d – t d1 )dt d1 (1-5)
dt d1
0
where
t d = ---t- (1-6)
t0
2
φ μw ct re
t 0 = ----------------------
- (1-7)
k
2
B = r e hφ c t θ (1-8)
In Equation 1-5, W(t) is the cumulative water influx from the aquifer and Q(td) is
the standard function derived by Van Everdingen and Hurst, which gives the
cumulative influx in the case of the unit pressure change at the boundary r = re.
Parameter t0, which is given by Equation 1-7, is the aquifer time constant. It is
used to define the dimensionless time td according to Equation 1-6. Parameter B,
as defined by Equation 1-8, is the aquifer capacity parameter. Equation 1-8
contains two constants: h is the aquifer thickness and θ is the angle (in radians)
subtended by the aquifer (Figure 1-1).
Carter and Tracy2 derived an approximation of the solution (Equation 1-5). This
approximate solution of the problem (Equations 1-1 through 1-4) is given by:
B [ p i – p ( t + Δt ) ] – W ( t )P' d ( t d + Δt d )
W ( t + Δt ) = W ( t ) + ------------------------------------------------------------------------------------------ Δt d (1-9)
P d ( t d + Δt d ) – t d P' d ( t d + Δt d )
Here, Δt is the timestep, Δtd is the dimensionless timestep, and Pd(td) is a function
derived by Van Everdingen and Hurst. Pd(td) is the dimensionless pressure
solution of the problem (Equations 1-1 through 1-4) for the case when the
condition at the boundary r = re is specified as the unit flow rate, instead of the
condition (Equation 1-3).
The Carter and Tracy solution does not require integration and therefore is more
efficient. In the VIP-EXECUTIVE implementation of the Carter and Tracy
solution, the user may provide the dimensionless pressure function Pd(td) in a
tabular form through input data. One can find tabulated pressure functions Pd(td)
for finite aquifers in the published literature.1,3 The dimensionless pressure
function for infinite radial aquifer1 is given in Table 1-1. This is the default table
that will be used if the user does not provide an input table.
dW e
---------- = J ( p a – p ) (1-10)
dt
is used where,
The water influx is evaluated using the simple aquifer material balance
We = ct Wi ( pi – pa ) (1-11)
in which,
W
p a = p i ⎛ 1 – -------e-⎞ (1-12)
⎝ W ⎠ ei
where
W ei = c t W i p i (1-13)
dW W
---------e- = J p i ⎛ 1 – -------e-⎞ – p (1-14)
dt ⎝ W ei⎠
Integrating the equation for the time interval ( t, t + Δt ) yields the following
Jp Jp
– --------i Δt – --------i Δt
p-⎞ ⎛ 1 – e W ei ⎞
+ W ei ⎛ 1 – ---
W ei
W ( t + Δt ) = W ( t )e ⎜ ⎟ (1-15)
⎝ p i⎠ ⎝ ⎠
or
Jp i
– -------- Δt
p-⎞ W – W ( t ) ⎛ 1 – e Wei ⎞
W ( Δt ) = ⎛ 1 – --- ⎜ ⎟ (1-16)
⎝ p i⎠ ei ⎝ ⎠
Aquifer
Reservoir
A model with the naturally extended grid system may require a significant number
of additional gridblocks in the model. This significantly increases memory and
CPU requirements. On the other hand, this grid system results in a matrix with an
orderly banded structure, which is an efficient solution to the problem.
Another method is to represent the aquifer by several blocks with large pore
volumes which are connected to reservoir blocks around the edges of the grid as
shown in Figure 1-3. This is done by defining nonstandard connections between
gridblocks. However, VIP-EXECUTIVE does not provide a specific input
structure that is convenient for this purpose. Therefore, it currently is a tedious
manual task to describe all connections required for complete description of the
aquifer.
Aquifer
Reservoir
The aquifer model that uses nonstandard connections results in a matrix with
many off-band terms. It is more computer intensive to solve such a matrix as
compared to the case of an orderly banded matrix.
2
00000 Boundary Flux Option
2.1 Introduction
With VIP-EXECUTIVE, users can define rectangular regions of the reservoir grid
for which boundary flux calculations will be performed in the simulation module.
The program may be run in output mode so that the calculated boundary flux is
reported in the output listing and recorded in a disk file. Subsequently, the
program may be run in input mode where the boundary flux file from an output
mode run is used as input data to the program. In input mode, the boundary flux
data are used in the same manner as sink or source information. That is, the
equivalent of a production/injection well is defined internal to the program for
each boundary block.
Figure 2-1 is a schematic representation of a coarse and fine model. The figure
shows a 6×8×1 coarse grid with an imbedded 10×9×1 fine grid model. Note that
the fine gridblocks are not allowed to overlap more than one coarse block.
1 2 3 4 5 6
2
1 2 3 4 5 6 7 8 9 10
1
3
2
3
4 4
5
6
7
5 8
9
Fine Grid
6
Coarse Grid
To make coarse and fine grid runs using the boundary flux option, set the program
dimensions properly in VIP-CORE. For the coarse model, set the required
dimension to the number of coarse blocks. If you use the boundary flux option to
write flux information for more than one rectangular region in a single run, define
the number of flux regions and the total number of coarse blocks. For the fine
model, define both the fine and coarse block numbers correctly. The dimensions
of the boundary flux model can be calculated using the following equations:
NFBLKS Total number of fine gridblock faces for all flux regions.
where
and
NXC = 4 NXF = 10
NYC = 3 NYF = 9
NZC = 1 NZF = 1
The equations for the number of block faces given above assume that the coarse
grid model has NX, NY, NZ > 2 and that the fine grid model does not touch any of
the 6 outer boundaries of the coarse model. The equations change in the obvious
manner if any of these assumptions is false. For example, if the coarse and fine
grid models are areal (NZ = 1) and the fine grid model touches the i = 1 boundary
of the coarse model, the equations become:
After completion of the base case run, compare region summaries with those from
the coarse model. Compare the production, injection, pressure, and boundary flux
information. The fine model may need further tuning to resolve any differences.
This ensures that the fine grid model has been properly set up and can be used for
further history matching or sensitivity analysis. For the example shown in
Figure 2-1 the fine grid model keywords required are:
■ Efflux partitioning
The mobility weighting option has been modified so that it is used only for efflux
terms.
In these methods, the model assumes that the proportions of oil, gas, and water
present in the coarse grid cell are maintained in each fine grid cell. Each phase is
proportionately allocated to each fine gridblock. While these methods may be
adequate for water-oil problems, neither of these methods is good for handling
segregation of fluid phases in the vertical direction. Consequently, they may not
be realistic for gas-oil problems.
1. If the VEONLY or VEAREA options are selected, all gas influx is allocated to
fine gridblocks above the gas-oil contact; similarly, all water influx is
allocated to fine gridblocks below the oil-water contact. Oil influx is
distributed between the gas-oil contact and the water-oil contact. In the
VEONLY option, the flux allocation is based on the fluid contact height only.
In the VEAREA option, flux is allocated in proportion to the product of
contact height and cross-sectional area.
2. If the SEGAREA option is used for the oil or water phase, the influx for that
phase is distributed below the gas-oil contact. The influx allocation is based
on the product of contact height and cross-sectional area.
3. The NOVEAD keyword turns off the allocation of influx on the basis of
vertical segregation for the coarse gridblocks and the phase under
consideration.
The user may choose which technique (segregated or dispersed) to use in the
model for each coarse gridblock by phase. Therefore, it is possible to use the
dispersed option for water-oil blocks and the segregated option for gas-oil blocks
in the model.
The mobility weighting option is not applicable for the distribution of influx in the
fine grid model. If the VE options are not used, influx is allocated on the basis of
the product of cross-sectional area and permeability for each fine gridblock.
In the flux input mode, the user can print the flux rate and cumulative flux for
each boundary block in the fine grid model using the FLUX parameter on the
PRINT card. The user can control the frequency of output using TIME, TNEXT,
or freq parameters on the PRINT card.
3
Corner-Point Geometry Option
■ X-, Y-, and Z-coordinates of corner points in all layers are specified using the
array input XCORN, YCORN, and ZCORN.
■ X- and Y-coordinates of corner points in all layers are specified using the
array input XCORN and YCORN. Z-coordinates of the gridblocks in the first
layer should be defined by the ZCORN array input with the LAYER or DIP
option. Layer thickness should be specified using the DZCORN, DZBCOR,
or DZVCOR array input. If the DZBCOR or DZVCOR array input is used,
then X- and Y-coordinates are adjusted so that the line from a corner point in
one layer to the corresponding point in the next layer is perpendicular to the
bedding plane. The DZCORN array input is similar to DZVCORN, but the X-
and Y-coordinates are not adjusted.
■ Users can specify the corner points along depth lines. These lines may have an
arbitrary orientation. Include the keyword LINE in the CORNER card to
invoke the LINE option. Specify each depth line by two triplets of X-,Y-, and
Z-coordinates. Define X-, Y-, and Z-coordinates of two points in depth lines
using the XCORN, YCORN, and ZLNCOR array input with the LNVAL
The program automatically checks to ensure that the projections of each block
face on the coordinate planes are convex quadrilaterals.
If any of the following cards or arrays are included in initialization data, the
program identifies positions of faults and pinchouts and calculates resulting
transmissibilities: FAULTS, PINCHOUT, ZCORNW, ZCORNE, ZCORSW,
ZBOTNW, ZBOTNE, ZBOTSW, and ZBOT. A pinchout is defined as a block
with a thickness less than the tolerance specified in the PINCHOUT card. The
pore volume of this block is set to zero. A connection is created between two
blocks with non-zero pore volumes - one above the pinchout and one below. The
fault and pinchout connections are called non-standard connections.
Vb = ∫ dV (3-1)
V
■ When the VAITS keyword is specified, the block is divided into several sub-
layers and the block bulk volume is estimated as
N
1
D c = ------ ∫ z dV (3-3)
Vb
V
1
TH = ------ ∫ DZ dV (3-4)
Vb
V
C DARCY × TMLTX i
T ij = ---------------------------------------------------
- (3-5)
right left
Ai Aj
------------------------ + -------------------- -
right left
A ij TX i A ij TX j
Ai right
I Aij
Aj left
J
where
s(λ) distance along path from the center of the block to the
right or left face
ψ(λ) angle between the path of s(λ) and the normal to the
quadrilateral of A(λ)
More complete details of both the standard and NEWTRAN options for corner-
point geometry are provided in Appendix B.
C DARCY × TMLTX i
T ij = ----------------------------------------------
- (3-7)
1 1
-------- + --------
TX i TX j
where
A x DI x + A y DI y + A z DI z
TXi = PERMX i RNTG i --------------------------------------------------------
2 2 2
- (3-8)
DI x + DI y + DI z
DIx, DIy, DIz X-, Y-, and Z-components of the distance between the
center of the i’th gridblock and the center of relevant face
of Block i.
The expressions for transmissibilities in the Y- and Z-directions are similar, but
the net-to-gross ratio is not included in the expression for the Z-transmissibility.
4
Dual Porosity Models
4.1 Introduction
Naturally fractured rocks often are observed near faults and folds. The fracture
channels are highly permeable compared to the rock matrix itself; they make the
reservoir productive even though the matrix has poor permeability. Oil in rock
matrix is drawn into the fracture by imbibition, gravity drainage, diffusion, or
expansion, then is carried away in the fracture network to production wells.
Matrix Fracture
Dual porosity
representation Km
⊕ Kf
These continua are superposed, and the model assumes the two continua exchange
fluids. In this approach, two finite-difference grid systems represent both
continua. Consequently, users must supply the fracture properties (fracture
porosity, etc.), flow exchange terms, and pseudo-capillary pressures, as well as
data for the rock matrix.
where subscript α indicates the phases (liquid, vapor, and water); subscripts f and
m indicate the fracture and rock matrix, respectively; krα, μα, and Bmα are the
relative permeability, viscosity, and formation volume factor for each phase α; P
is pressure; D is depth; and Tmf is matrix-fracture transmissibility. The matrix-
fracture transmissibility is defined by
where ΔX, ΔY, and ΔZ are the finite-difference gridblock dimensions, and
effective permeability kmf is defined by
1 k mx k fx k my k fy k mz k fz ⎞
k mf = --- ⎛ --------------------
- + --------------------
- + -------------------
- (4-3)
3 k mx + k fx k my + k fy k mz + k fz⎠
⎝
1
k mf = --- ( k mx + k my + k mz ) (4-4)
3
Coats53 defines k mf σ as
⎛ k mx k my k mz⎞ (4-6)
k mf σ = 8 ⎜ -------
- + -------- + --------⎟ ( 1 – φ f )
⎝ l 2mx l 2my l mz ⎠
■ The σ and matrix and fracture permeabilities (SIGMA, KX, KXFEFF, etc.)
*
q do = S mo k do ρ mo ( x f – x m ) (4-7)
*
q dg = S mg k dg ρ mg ( y f – y m ) (4-8)
* ψ iv y i
x fi = ------------ (4-9)
ψ iL f
* ψ iv x i
y fi = ------------ (4-10)
ψ iL f
where Smo and Smg are matrix oil and gas saturations; ρmo and ρmg are oil and
gas densities in a matrix; xm and ym are molar fractions of oil and gas, and ψiL
and ψiv the liquid and vapor phase fugacities. Tdo and Tdg are mass transfer
coefficients of oil and gas defined by:
k do = D o σ d ΔXΔYΔZ (4-11)
k dg = D g σ d ΔXΔYΔZ (4-12)
VIP input for Do, Dg, and σd are TDIFFO, TDIFFG, and SIGMAD respectively.
Coats2 has shown that the approximate diffusion time for the concentration to
decay 90% of its initial value is:
0.85τ or l 2
t = ----------------- ⎛ ---⎞ (4-13)
D ⎝ 2⎠
where τor, l and D are a tortuosity, length of a matrix cube, and diffusion
coefficient respectively. For example, consider a gas-gas high pressure (4500 psi)
diffusion with tortuosity of 3.5, l =1’, and D=0.001 cm2/sec. Then, t =8 days. If
this is an insignificant time compared to the production period, then the diffusion
effect may be ignored in the model. Thus, the above equation can be used to
decide if a model should include the diffusion effects.
From Equation 4-7 and Equation 4-8, we observe that the diffusion flux would be
significant if the compositions in a fracture and a matrix are markedly different.
Such is the case when an injection gas composition differs greatly from that of the
resident gas.
Gas-cap expansion
Matrix fluid expansion
gas
oil
Oil imbibition
gas
oil
Diffusion
gas
oil
Water
oil
Each matrix has the dimensions of lx, ly, and lz as shown before.
Step 2. Set the location of the water-oil contact within the fracture system,
beginning at the top of the gridblock. Numerically integrate the equilibrium
distribution of water saturation in all matrices using the specified water-oil contact
as a boundary condition. The average water saturation of the matrices is the
resulting integral divided by the number of matrices. The average water saturation
strongly depends on the height of a matrix since the water column in the matrix is
snapped off at each matrix block interface. The imbibition capillary pressure
curve should be used for integration.
ΔX
ΔZ/2
lz
owc
ΔZ/2
⎛ ⎞ lz
P c = ( γ w – γ o ) ⎜ ΔZ
------- – h⎟ ------- (4-14)
⎝2 ⎠ ΔZ
where h is the height of the water-oil contact (Figure 4-3) and γo and γw are
gradients of oil and water, respectively. When defined this way for the matrix, the
matrix capillary pressure will cover the same range of values that is covered by
the vertical equilibrium capillary pressure used in the fractures. However, the
corresponding values of saturation will be quite different from the fracture values.
Step 4. Repeat Steps 2 and 3 for many different water-oil contact levels, evenly
spaced over the height of the gridblock. By doing so, the capillary pressure is
tabulated as a function of the average water saturation. This table has the property
that is equivalent to the one determined by rigorous integration of the rock
capillary pressure curve subjected to the boundary condition. This table is used for
the simulation.
4.4.2 Coats’ Method for Treatment of Gas-Oil Exchange Between Matrix and
Fractures
Coats’ method53 for describing the gas-oil exchange between matrix and fracture
is implemented in VIP-EXECUTIVE. This method is available in addition to the
pseudo-capillary pressure model for gas-oil gravity drainage used in previous
versions of VIP-EXECUTIVE and described for in the section above. The Coats
method is the default method for gas-oil gravity drainage. The pseudo-capillary
pressure method is selected using the NOCOATS parameter on the PSEUDO
card.
The matrix-fracture phase flux exchange in surface volume units is given by:
λ gm
q g = τ --------- ( ΔP o + P cgo ) (4-15)
Bg
λ om
q o = τ --------- ΔP o (4-16)
Bo
where
σ shape factor
k rocwm (4-18)
λ om = ---------------
-
μ om
τz k rgf ⎛ S gm ⎞
P cgo = ---- β g μ gf ⎛ -------
- + λ om⎞ ( l z Δγ – P ce ) ⎜ --------------
- – 1⎟ (4-19)
τ ⎝μ ⎠ ⎝ S* S ⎠
gf ge gf
⎛ 1 – φ f⎞
τ z = 0.001127k zm ⎜ ------------
-⎟ ΔXΔYΔZ (4-20)
⎝ l 2z ⎠
where
φf fracture porosity
1 l z Δγ ⁄ σ r
S ge = --------------------
l z Δγ ⁄ σ r ∫o S g ( P c )dP c (4-21)
The terms, Δγ, Pce and Sge in Equation 4-19 are functions of pressure. The
pressure dependence of Pce and Sge is modeled through the dependence of gas-oil
interfacial tension on pressure. The gas-oil capillary pressure is assumed to be a
linear function of gas-oil interfacial tension as follows:
The function
σ(p)
σ r ( p ) = -------------- (4-23)
σ ( Po )
VIP-CORE uses the SIGT table to construct tables for each gridblock. These
tables consist of three columns. The first column contains pressures read from the
SIGT tables, the second column contains Sge values at the corresponding pressure
computed using Equation 4-21, and the third column contains the quantity lzΔγ -
Pce. These tables are output to file with extension .pseudo and also to the restart
file. VIP-EXECUTIVE interpolates these tables to determine Pcgo from
Equation 4-19 and then computes the flux terms from Equation 4-15 and 4-16.
The dual porosity option is activated using DUAL in one of the utility inputs. If it
is followed by POR, then the program uses the dual porosity option with single
permeability. Without the POR card, the program uses dual permeability.
Figure 4-4 shows the schematics of these two options. The single permeability
option on the right does not have the flow exchanges among rock matrices, but the
flow exchanges between fracture and matrix take place.
Rock
Continuum
Fluid
Exchange
Fracture
Continuum
Wells that are specified will perforate both rock and fracture continua and
therefore, there will be two sets of permeability thicknesses in the dual
permeability option as illustrated in Figure 4-5. In highly fractured reservoirs, the
flow out of fractures dominates production. In such cases, ignore the flow from
matrices; this can be achieved by setting FM = 0 in the FPERF data.
If, however, the single permeability option is used, wells commmunicate only
with the fractures and FM is automatically set equal to zero.
Km*Hm
❍
Matrix
❍ ⇐
Well ❍
❍
⇐ Kf*Hf
Fracture
Any type of fractured reservoir can be solved using the dual porosity/dual
permeability option because there are two distinct flow paths - one in the fracture
and one in the matrix. So if the fractures are absent in an area the flow occurs
entirely in the matrix. However, this option can be expensive to run. The more
efficient dual porosity/single permeability option only allows flow from gridblock
to gridblock through the fractures. Fluids in the matrix can only exchange with
fluids in the fracture of a common gridblock. With this option, regions of a
partially fractured reservoir that are not fractured (regions with zero fracture
porosity) end up disconnected from the flow network.
Flow medium
Matrix
medium
In order to make this more understandable, let the grid system through which
fluids flow be called a “flow medium”, and let the medium which supplies fluids
be called a “matrix medium” as shown in Figure 4-6. Grid blocks in the “matrix
medium” do not communicate with each other in the method under consideration
here (an assumption of the dual porosity with a single permeability).
Only grid blocks of the “flow medium” communicate each other. The model in
this example runs four times faster than the dual porosity/dual permeability
model.
Results from VIP-EXECUTIVE were also compared with those presented in the
Sixth SPE Comparative Solution Project54. Shown on Figure 4-10 are results for
the single block gas-oil drainage problem of SPE 6. These results demonstrate that
the dual porosity model of VIP-EXECUTIVE matches gas-oil drainage from a
single block as accurately as a fine grid model. Results for the depletion and gas
injection problems of SPE 6 are presented on Figures 4-11 to 4-16. Since no
analytical results are available for these cases nothing can be said about the
accuracy of the results. The different results reported by VIP-EXECUTIVE
(exec22r, exec23d) are due to the use of the different gas-oil matrix fracture
exchange models available in VIP-EXECUTIVE.
Sgf = 1.0
Figure 4-7: Gas/Oil Drainage Recovery for a 10x10x10 ft. Matrix Block
(Coats’ Figure 15)
krog/3
Figure 4-8: Gas/Oil Drainage Recovery for a 10x10x10 ft. Matrix Block
(Coats’ Figure 16)
Figure 4-9: Gas/Oil Drainage Recovery for a 1x1x1 ft. Matrix Block
(Coats’ Figure 17)
Consider a bulk volume of a fractured medium as shown in Figure 4-19. The dual
porosity option in VIP replaces the bulk volume with two continua: matrix and
fracture. Porosities associated with these continua are defined so that the
porosities times a grid block volume give the respective pore-volumes for matrix
and fracture systems.
DY
= = +
Vrock
DX DX DX
Vf V φm in VIP
Vt φf ≡ Vf / Vt
Vm,pore = Vpore in matrix/ Vt
where Vt is the bulk or grid block volume. By using the above equations we can
derive the matrix porosity input data for VIP as
φ m in VIP = V m, pore ⁄ V t
= φ m in lab * V m ⁄ V t
(4-26)
= φ m in lab * ( V t – V f ) ⁄ V t
= φ m in lab * ( 1 – φ f )
A fracture porosity varies depending on the origin and size, but Teodorovich has
proposed a rough guideline6:
Typically, long fractures that span many layers have a porosity ranging from 0.01
to 0.5%, and networked small fractures within a layer have 0.01 ~ 2%.
The effective permeability can be directly determined by core tests, but it is likely
to be limited to finding a single fracture channel permeability, Kff, since cores are
too small to contain fractures with a typical fracture spacing.
↓↓↓↓↓↓↓↓
Fractured core
Impermeable
jacket
e h
Flow direction
b
e .... fracture spacing
b .... fracture width
a a ... width of sample
AL AL ... length of sample
Suppose a single fracture channel has permeability Kff, then the equivalent
continuum permeability, Kf, of this model can be expressed by7:
b
K f = K ff ⎛ ---⎞ (4-27)
⎝ e⎠
K f ( md ) = A × B × C × D (4-29)
where A, B, C and D are functions of rock type, porosity, pore size and pore shape
respectively, and they are defined in the following:
Pore Space
Description Value of A Characteristics for sub types
Type
Effective porosity
Description Value of D
VIP keywords for effective fracture permeabilities are KXFEFF, KYFEFF, and
KZFEFF. If corresponding transmissibilities are known, then one may use TXF,
TRF, TYF and TZF.
1 ∂V pore
C r = ------------
V pore ∫ ---------------
∂p
- dv (4-30)
where the integration is performed over the pore volume enclosed by a control
volume subjected to a constant confining pressure. In case of a fractured rock,
there are two types of pore volumes: fracture pore and matrix pore, and they
would have different pore compressibilities as illustrated in Figure 4-22:
1 ∂V pore 1 ∂V pore
------------
V pore ∫matrix pore ∂p
---------------
- dv +
V pore ∫fracture pore ∂p
------------ ---------------- dv ≡ C m + C f
(4-31)
Fracture pore
→ ←
Matrix pore
where Cm and Cf represent the matrix pore and fracture pore compressibilities
respectively. Jones8 gave an example of carbonate rock compressibilities:
Cm = 2x10-6 /psi
Cf = 72x10-6 /psi.
VIP input for Cm and Cf are CR and CRF arrays respectively. There is another
input, Cr , in the DWB card, and this is overridden by the above array input. This
Cr in the DWB card can be used without the presence of CR and CRF arrays only
if the compressibilities of fracture and matrix pores are identical.
≅
LZ LZ
LX LY
Real rocks do not have such a simple fracture morphology. However, this
idealization gives a theoretical estimate of shape factors, or transmissibilities. The
dimensions of an equivalent block are specified in VIP by arrays LX, LY and LZ
as shown in Figure 4-23.
These are used to compute the transmissibilities from Equations 4-2 to 4-5.
ΔX, ΔY, and ΔZ are dimensions of a finite difference grid block; kmx, kfx, etc. are
the matrix and fracture permeabilities of an element in the x-direction in Figure 4-
23. The input LX, LY and LZ correspond to lx, ly and lz in the above equation.
Optionally the shape factor, σ, may be directly specified in stead of lx, ly and lz.
and the corresponding VIP input keyword is SIGMA. Furthermore, Tmf may also
be specified directly using VIP keyword TEX. Both SIGMA and TEX are array
input data.
Figure 4-24. Thus, the fracture relative permeabilities are significantly rate-
dependent, so that experiment should be performed with a representative flow rate
during the production.
kro kro
krw krw
Sw Sw
The fracture saturation tables in VIP are specified by headings SWTF and SGTF.
Operational
range
Permeability
Confining pressure
If the operational pressure varies significantly during the life of a reservoir, the
simulation model must reflect the pressure dependence appropriately. If the
pressure dependence is significant, it can be modeled by using the compaction
option in VIP: a table of multiplication factors to porosity and permeabilities. The
table heading is ICMT, and ICMTF array is the corresponding pointer. In using
the compaction option a utility card, COMPACT, must be included.
Km*Hm
❍ matrix
❍ ⇐
Well ❍
❍
⇐ Kf*Hf
fracture
In dual porosity single permeability model wells can only communicate with the
fractures and FM is automatically set to zero.
.8 1. .00 2.
C ----------------------------------------------------------------
C FRACTURE WATER-OIL
C ----------------------------------------------------------------
SWTF 1
SW KRW KROW PCWO
.0 0. 1. 0.
.50 .50 .50 0.
.999 .999 0.001 0.
1. 1. 0. 0.
C ----------------------------------------------------------------
C FRACTURE GAS-OIL
C ----------------------------------------------------------------
SGTF 1
SG KRG KROG PCGO
.0 .0 1. .0
.1 .1 .9 .0
.2 .2 .8 .0
.3 .3 .7 .0
.4 .4 .6 .0
.5 .5 .5 .0
.7 .7 .3 .0
.999 .999 .0 .0
1. 1. 0. .0
BETAG 1.14
SIGT
P SIGR
14.7 100.0
1688.7 66.6667
2044.7 52.2222
2544.7 36.6607
3005.7 24.4444
3567.7 14.2222
4124.7 8.0
4558.7 4.9333
4949.7 2.8333
5269.7 1.7222
5559.7 1.0
7014.7 0.5556
C P SIGR DGOG
C 14.7 100.0 .7795
C 1688.7 66.6667 .5934
C 2044.7 52.2222 .5666
C 2544.7 36.6607 .5299
C 3005.7 24.4444 .4972
C 3567.7 14.2222 .4597
C 4124.7 8.0 .4247
C 4558.7 4.9333 .3986
C 4949.7 2.8333 .3760
C 5269.7 1.7222 .3570
C 5559.7 1.0 .3411
C 7014.7 0.5556 .2676
BOTAB 1
DOB WTRO
0.81918 200.
PSAT RS BO BG GR VO VG
7014.7 2259 2.1978 .600 .762 .109 .0330
5559.7 1530 1.8540 .696 .762 .210 .0274
5269.7 1413 1.7953 .720 .762 .229 .0265
FVEGOF CON
1.0
ISATEM CON
1
ISATEF CON
1
OVER TZF
1 10 1 1 1 5 *.1
C
END
RUN
DIM NWMAX NGCMAX NPRFTO NPRFMX NBATMX NSTGMX NFSMAX
2 1 10 10 1 1 1
C
IMPLICIT
C
RESTART 0 3
PLOT FORM
START
OUTPUT P SO SG
PRINT WELLS TIME
PRINT REGIONS ARRAYS TIME
WPLOT TIME 3
C ABORT 1. 1.
BLITZ
C TOLR .001 .01 RELTOL
C
WELL N NAME I J IGC
0 PROD45 10 1 1
0 INJ123 1 1 1
FPERF
WELL L WIL
INJ123 1 1.613294
X 2 1.613294
X 3 1.755122
PROD45 4 1.690117
X 5 1.690117
INJ G FSTD INJ123
PROD O STD PROD45
QMAX INJ123
.9
QMAX PROD45
1000.
ECOLIM O PROD45
QMIN 1
100.
DPBHMX PROD45
102.
ITNLIM 1 10 500 .1 .1 .1
DT -1. 1. 91. 500. .1 .1 .1
DATE 30 6 1992
DATE 31 12 1992
DATE 30 6 1993
DATE 31 12 1993
DATE 30 6 1994
DATE 31 12 1994
DATE 30 6 1995
DATE 31 12 1995
DATE 30 6 1996
DATE 31 12 1996
DATE 30 6 1997
DATE 31 12 1997
DATE 30 6 1998
DATE 31 12 1998
DATE 30 6 1999
DATE 31 12 1999
DATE 30 6 2000
DATE 31 12 2000
DATE 30 6 2001
DATE 31 12 2001
STOP
END
5
End-Point Scaling
5.1 Introduction
VIP-EXECUTIVE allows the user to model spatial differences in relative
permeability and capillary pressure end points with a single set of generic relative
permeability and capillary pressure curves. Both the saturation and relative
permeability end points for any gridblock can be different from those of the
generic rock type to which the gridblock is assigned.
The calculation of the relative permeabilities is divided into two steps: saturation
end-point scaling, then end-point relative permeability scaling. Two- and three-
point saturation scaling options are available in VIP-EXECUTIVE.
2. SWR - Residual water saturation (the highest water saturation for which
the water is immobile).
3. SWRO - Water saturation at residual oil (one minus residual oil saturation
where the residual oil saturation is the highest oil saturation for which the oil
is immobile in the water-oil system).
5. SGL - Connate gas saturation (the smallest gas saturation entry in a gas
saturation table).
6. SGR - Residual gas saturation (the highest gas saturation for which the gas
is immobile).
7. SGRO - Gas saturation at residual oil (one minus the residual oil saturation
minus the connate water saturation where the residual oil saturation is the
highest oil saturation for which the oil is immobile in the gas-oil-connate
water system).
8. SGU - Maximum gas saturation (the highest gas saturation entry in a gas
saturation table).
If any of the saturation end points are not specified, rock data will be used for that
gridblock. Both connate and residual water saturations are set to the specified
value if the following exceptions exist: if the connate and residual water
saturations are the same in the rock table and if you specify either the connate or
residual water saturation (but not both) for the gridblock.
′ ′
( S – SL ) ( SR – SL ) ′
S′ = -----------------------------------------
- + SL (5-1)
( SR – SL )
S′ is then used in the table look-up for the relative permeability or capillary
pressure.
Two-point scaling uses the connate and maximum saturations as two saturation
end points in the calculation of capillary pressures:
Two-point scaling uses the residual and maximum saturations as two saturation
end points in the calculation of relative permeabilities:
In addition to honoring the end points used in the two-point scaling option, the
three-point scaling algorithm honors the relative permeability of the mobile phase
at the residual saturation of the other phase. The following end points are used in
the relative permeability calculation:
Example:
C-----------------------------------------------------------
C SATURATION DEPENDENT TABLES
C-----------------------------------------------------------
SWT 1
SW KRW KROW PCWO
0.2 0.0 0.6 10.
0.3 0.0 0.45 8.
0.7 0.4 0.0 3.
0.9 1.0 0.0 0.
SWT 2
....
....
....
ARRAY
C------------------------------------------------------------
C SATURATION END POINTS
C------------------------------------------------------------
SWL CON
0.05
SWR CON
0.15
SWRO CON
0.85
SWU CON
1.0
The specified rock table shows that connate water saturation, SWL′, is 0.2. The
residual water saturation, SWR′, is 0.3. Water saturation at residual oil, SWRO′,
is 0.7. Maximum water saturation, SWU′, is 0.9. The specified grid end points
are SWL = 0.05, SWR = 0.15, SWRO = 0.85, and SWU = 1.0.
The water-oil capillary pressure curves for the rock and gridblock with the two-
and three-point options are compared below. The capillary pressures at SWL and
SWU are honored for the two-point option and those at SWL, SWR, SWRO, and
SWU are honored for the three-point option. Rock saturations 0.3 and 0.7 are
equivalent to grid saturations 0.1857 and 0.7286, respectively, for the two-point
option.
The water relative permeability curves for the rock and gridblock with the two-
and three-point options are compared below. The water relative permeabilities at
SWR and SWU are honored for two-point scaling, while those at SWR, SWRO,
and SWU are honored for three-point scaling. The rock saturation 0.7 (SWRO′) is
equivalent to the gridblock saturation 0.7167 for two-point scaling and 0.85
(SWRO) for three-point scaling. The relative permeability at SWRO = 0.85 is
0.6823 for two-point scaling, and 0.4 for three-point scaling.
■ Maximum gas saturation should not exceed 1.0 minus the connate water
saturation.
■ The connate gas saturation should not exceed 1.0 minus the maximum water
saturation.
■ The connate gas/water saturations should not exceed the residual gas/water
saturations.
■ The water/gas saturations at residual oil should not exceed the water/gas
maximum saturations.
1. KROLW - The oil relative permeability at the connate water saturation and
minimum gas saturation.
3. KRGRO - The gas relative permeability at the residual oil saturation. This
input option is available for the three-point option only.
Note that for the two-point option, the relative permeabilities at SWRO, and
SGRO are not honored and KRWRO and KRGRO are neglected.
′ KROLW
k row = k row × ----------------------- (5-2)
KROLW′
and
′ KROLW
k rog = k rog × ---------------------- (5-3)
KROLG′
KROLW’ is the oil relative permeability at the connate water saturation for the
generic rock curve, and KROLG’ is the oil relative permeability at the connate gas
saturation for the generic rock curve. krow′ and krog′ are the rock oil relative
permeabilities at the saturations calculated from the saturation end-point scaling.
′ KRWRO ′
k rw = k rw × ----------------------- if S w ≤ SWRO’ (5-4)
KRWRO′
′
( KRWU – k rw )
k rw = KRWU – ---------------------------------------------------- × ( KRWU – KRWRO )
( KRWU – KRWRO′ )
′
if S w > SWRO’ (5-5)
and
′ KRGRO- ′
if S g ≤ SGRO’
k rg = k rg – ---------------------- (5-6)
KRGRO′
′
( KRGU – k rg )
k rg = KRGU – --------------------------------------------------- × ( KRGU – KRGRO )
( KRGU – KRGRO′ )
′
if S g > SGRO’ (5-7)
KRWRO’ and KRGRO’ are the water and gas relative permeabilities at the
residual oil saturation for the generic rock curves. KRWU and KRGU are the
water and gas relative permeabilities at the maximum saturations, respectively.
krw’ and krg′ are the water and gas relative permeabilities at the saturations
calculated from the end-point scaling.
Example:
C------------------------------------------------------------
C END-POINT RELATIVE PERMEABILITIES
C------------------------------------------------------------
KROLW CON
0.8
KRWRO CON
0.25
The comparisons of the rock and grid-point relative permeability curves with end-
point relative permeability scaling are shown in the following figures.
KROLW’
KROLW
Example:
SWL X+ ZVAR
0.2 0.1 0.25 0.15
SWR X+ ZVAR
0.25 0.2 0.35 0.2
SWRO X+ ZVAR
0.6 0.7 0.65 0.75
SWU X+ ZVAR
0.9 0.95 0.98 1.0
Pc K
J ( Sw ) = ------------------- ---
σ cos θ c φ
00
Pc wo ( Sw ) = J ( Sw ) σ cos θ c 00
Pc wo ( Sw )
Pc = ------------------------
- 00
Kx
------
φ
6
Equilibration
6.1 Introduction
VIP-EXECUTIVE offers four reservoir initialization options:
■ Vertical equilibrium.
DEPTH
GAS
GAS-OIL
GAS-OIL CONTACT TRANSITION ZONE
(GOC)
OIL
WATER-OIL
TRANSITION ZONE
WATER-OIL CONTACT
(WOC)
WATER
SATURATION
Figure 6-1: Gravity-Capillary Equilibrium
The fluid contacts, datum liquid phase pressure and depth, and capillary pressures
at the contacts are entered in the equilibrium constant table:
Definitions:
The label IEQUIL refers to the region of the field to which the initialization
applies. VIP-EXECUTIVE permits a variable number of contacts within a field;
this is illustrated in Figure 6-2:
For each IEQUIL region, the automatic initialization of reservoir phase pressures,
saturations, and compositions proceeds as follows:
The phase pressure vs depth tables are generated by the following relationships:
dP
--------o- = ρ o ( P o, PSAT ieq )g (6-1)
dD
dP
---------w- = ρ w ( P o )g (6-3)
dD
dP
--------g- = ρ g ( P o, PSAT ieq )g (6-5)
dD
For reservoirs which have an initial oil column, the datum depth, DEPTH, must
satisfy
When the reservoir has a primary gas cap and an oil column,
For gas/gas condensate reseroirs with no initial oil column (GOC = WOC), the
datum depth, DEPTH, may be at any depth at or above the contact.
When the reservoir does not have a gas cap, GOC may be assigned a depth above
the top of the reservoir. In any case, the saturation pressure entered in the
equilibrium constants table is the saturation pressure at DEPTH.
1. Calculates average phase pressures Po, Pg, and Pw for each block.
–1
S = Pc ( Pc ) (6-9)
DMAX
1 –1
S wb = ----------------------------------------
DMIN – DMAX ∫ P c ( D )dD (6-12)
DMIN
DMAX
1
S wb = ----------------------------------------
DMIN – DMAX ∫ S w ( D )dD (6-13)
DMIN
DMIN
DTOP
MDEPTH
DMAX DBOT
Therefore, the INTSAT method may not represent gridblock geometry accurately.
For example, the calculated average saturations for the two blocks shown in
Figure 6-3 would be the same. This method is accurate only for non-dipping
rectangular grid systems.
most accurate method for calculating gridblock saturations and bulk volumes.
Average block saturations (Sbw, Sbg) and bulk volume (Vb) are defined as
DMAX
1 –1
S bw = ------
Vb ∫ P cw ( D )AREA ( D )dD (6-14)
DMIN
DMAX
1 –1
S bg = ------
Vb ∫ P cg ( D )AREA ( D )dD (6-15)
DMIN
DMAX
i=N D i + 1
1 –1
S bw = ------ ∑ ∫ P cw ( D )AREA ( D )dD (6-17)
Vb
i = 0 Di
Similar expressions are used for the average gas saturation and block bulk
volume.
3. The sublayers are not equally spaced; therefore, they may have different
thickness. The number of sublayers N and their thickness are internally
calculated to ensure that relative approximation errors are less than tolerances,
which are specified on the VAITS card.
DMIN
D1
D2
D3
.
.
AREA(D)
.
Dn-1
DMAX
D
Figure 6-4: The VAITS Initialization
In Figure 6-5, SW(Di) refers to saturation at the gridblock center and SWi refers to
the block average saturation for block i. The capillary pressure adjustment is
given by
A new capillary pressure function that ensures equilibrium for saturation SWi is
defined by
The card used to trigger the supercritical initialization is the CRINIT card in the
Utility Data section of VIP-CORE. The initial pressure, PINIT, should be entered
at DEPTH, which should be the same as the input gas-oil contact depth, GOC. The
GOC depth specification can affect the speed and stability of supercritical
initialization since dew-point calculations are performed first for gridblock depths
above DEPTH and bubble-point calculations are tried first for gridblock depths
below DEPTH. The program checks on the heavy component k-value and
switches to the appropriate routine if necessary; however, this takes time. The
density of the original compositions is then used to initialize the function of
pressure versus depth from PINIT at DEPTH input on the IEQUIL card. The
composition at each depth is interpolated from those given on the
COMPOSITION cards and the equilibrium composition is computed in the
bubble- and dew-point routines.
For each equilibrium region, the program uses the input gas-oil contact depth
(GOC in IEQUIL card) as the estimated depth of the change from gas to oil and
will perform bubble- and dew-point calculations using the input composition as a
function of depth. The fluids are sorted using the computed equilibrium
compositions of the heaviest component to determine whether the reservoir fluid
is oil or gas. The criterion that the heavy component must have a k-value less than
one is used to determine whether the results of the bubble- or dew-point
calculation should be used at that depth in the reservoir. The point at which the
switch occurs is the true gas-oil contact for the super-critical fluid. The input
PINIT value from the IEQUIL card is used to compute the original pressure
surface from DEPTH and the computed oil and gas densities. At all times, bubble-
or dew-point routines are used to compute the equilibrium phase compositions
and the original saturation pressure values as a function of depth. A switch occurs
from bubble- to dew-point fluids based on the heavy component k-value.
7
Faults
7.1 Introduction
Figure 7-1 shows a structural map in a portion of a petroleum reservoir. The
contour discontinuities indicate faults. A fault is one of the structures that traps
hydrocarbons; it may influence fluid flow significantly. Therefore, reservoir
models frequently include the faults. The fault handling option in VIP-
EXECUTIVE allows simple ways of realizing such noncontiguous layer
connections resulting from the faults.
Simple faults can be modeled without using VIP-EXECUTIVE’s fault option, but
such a model requires more gridblocks. Figure 7-2 illustrates a model with and
without the fault option.
Without
Using
Two-Layer Model Fault
Option
➡
Using
Fault Option
Consider the two-layer model on the left in Figure 7-2. Without the fault, option
one must use five layers to model the displacement due to the fault. However, the
fault option allows users to create the grid system shown in the lower right where
the grids may not be contiguous. VIP-EXECUTIVE determines the connections
among blocks across the fault plane based on user input, then computes the
transmissibilities. Solvers take into account the flows associated with the fault
connections. In this manual, assumptions, guidelines, and the method of
computing the transmissibilities are presented.
1 2 3 4
5
6 x(i)
1
FY
3
FX
y(j)
■ The fault plane is located vertically at the grid boundary; the throw (or shift)
is constant in the Z-direction. In other words, throws between (3,3,k) and
(4,3,k) are the same for all k in the fault plane FX.
■ Fault throws may vary areally. For example, the throws of fault FY at i = 4, 5,
6 may be different.
much as possible to specify depth. If the layers are no longer horizontal, for
example, because of significant layer-thickness changes, then use the corner-
point geometry option.
■ The depths of gridblocks should be consistent to the fault throws (i.e., fshift in
Section 2.5 in the VIP-CORE Reference Manual) specified in user input.
■ The fault throws have positive or negative signs depending on the relative
depths of grids across fault. If one walks in an increasing X-direction and
encounters a drop-off, the drop-off, or throw, is positive since the drop is in the
positive Z-direction. If one encounters a wall of rock, it has a negative throw.
FX 4 3 -4 20.00 *1.5
Fault FY has three throws listed, i.e., 21 ft (m), 22 ft (m), and 23 ft (m) for the
three-grid-long fault, while the first input line has only a single throw, 20 ft (m). If
only a single shift is specified, then the shift is constant along the fault.
There are variations of the input formats depending on the options, but the most
frequently used format has the structure shown in Figure 7-4:
FX 4 3 -4 20.00 *1.5
Transmissibility Multiplier
Throw
Areal Extent of Fault Plane (range of j location)
Fault Plane Location (i location)
Heading for Fault Input
Another set of input formats can be used for arbitrary gridblock connections.
However, users must explicitly specify the gridblocks that are connected as well
as the transmissibilities. As a result, this input method can become tedious.
Furthermore, the connections for matrix and fractures are specified separately in
the dual porosity model.
In describing the arbitrary connections, there are two kinds of input formats. The
first method assumes that the connections are between two neighboring columns
and that each line of input specifies a single connection as shown in Figure 7-5.
Users must specify the connecting gridblocks and transmissibility for each
connection.
FLTXC
i j kr kl tx
Transmissibility
Vertical Location of Grids
Areal Location of Faults x
i
Heading for Fault Input
kr
i-1
z kl
FLTXC should be replaced with FLTYC for Y-direction faults, FLTRC for R-
direction faults in radial coordinate systems, or FLTTC for theta-direction faults in
radial coordinate systems. To specify the fault connection similarly for fracture
porosity, the corresponding headings are FLTXCF, FLTYCF, FLTRCF, or
FLTTCF. For example, the connection between (3,4,2) and (4,4,1) in Figure 8-5
can be expressed as:
FLTXC 4 4 1 2 192.23,
The second type of fault input data for arbitrary connections also is used for the
same purpose, but can be used to connect any gridblocks within the model; i.e.,
the gridblocks need not be neighbors. One of the applications is to use ordinarily
inactive grids as an aquifer. This input format requires the block coordinates and
transmissibility shown in Figure 7-6.
FTRANS
i1 j1 k1 i2 j2 k2 trans
Transmissibility
Coordinates of the Second Grid
Coordinates of the First Grid
Heading
(i1,j1,k1) (i2,j2,k2)
trans
For the fracture porosity, the heading is FTRANSF. FTRANS or FTRANSF may
be used to override fault transmissibilities, but can do so only for faults. To
override transmissibilities of the regular connections, use OVER or VOVER
option. An example of FTRANS usage is shown later.
Left Right
DZ
DZ
DZ
DX DX
VIP-CORE uses a fault shift input and layer thicknesses to determine the
connectivities, which could result in wrong connectivities if there are gaps
between layers. In the figure below a two layer model, (a), has a gap between
layers and a fault shift. If depth information is ignored, the layers are assumed to
be contiguous, resulting in connectivities as shown in model (b).
The layer tops and bottoms of the neighbor column are displaced by the shift
amount.
VIP-CORE has 2 options: (a) the method that uses depth information to calculate
layer connectivity, and (b) the method that assumes the layers are contiguous.
Method (a) is the default. Method (b) would be used if answers from a previous
FX 2 1 2 -25 NODEPTH
FX 2 2 -25 NODEPTH
The input for the model is listed in the figure, and the resulting output by the
method (a) is:
If method (b) is used there would be 18 connections, and they are probably
incorrect.
If the top layers of the neighbor and fault columns are not horizontal, but have the
same depths, the above scheme still computes the connections based on the shift
amount as shown in the next figure.
1 2 3 4 5 6 7 8 9 10 11 x
2 #1
3 #2
6
7
8
9 #3
10
11
y
Examples:
FAULTS
C fault #1
FY 2 3 -4 220.5
FX 5 2 -2 200.1
FY 3 5 -6 190.2 180.1
FX 7 3 -4 80.0 10.0 *0.8
C fault #2
FX 10 2 -6 30.5
C fault #3
FY 10 5 -9 -59.8
C override two connection of fault #2
FLTXC
10 2 1 3 121.55
10 3 1 4 22.9
C Connect aquifer to (10,1,1)
FTRANS
10 1 1 1 8 1 211.9
10 1 2 1 8 2 331.2
10 1 3 1 8 3 249.1
10 1 4 1 8 4 10.2
The options are currently implemented for single-grid models only and are not
compatible with the dual porosity option, the tracer option, and the implicit
version of VIP.
∑ Qk = 0, (7-1)
k=1
where
Q k ≡ T Ik λ kt ( P k – P f – α fk ) (7-2)
is the total volumetric flow rate for faulted gridblock k, TIk is the transmissibility
between faulted gridblock k and the fault which is simply the I-direction half-
transmissibility of the faulted gridblock, and λkt is the total mobility of gridblock
k;
k rw k rg k ro⎞
λ kt ≡ ⎛ ------- + ------- + ------- . (7-3)
⎝μ μg μo ⎠ k
w
Here krj and μj are relative permeability and viscosity of phase j. Parameters Pk,
Pf, and αfk are pressure in gridblock k, fault pressure at a reference depth, and the
gravity term in the fault (relative to the reference depth), respectively.
For outflow gridblocks in which the pressure difference term in Equation 7-2 is
positive, the water surface rates and the hydrocarbon component molar rates can
be readily calculated:
+ T Ik k rwk
q wk = ----------------- B w ( P k – P f – α fk ), (7-4)
μ wk
Here Bw is the reciprocal formation volume factor of water, ρo and ρg are molar
densities of oil and gas phases, xi is the molar fraction of component i in the oil
phase, yi is the molar fraction of component i in the gas phase, and Nc is the total
number of hydrocarbon components.
For inflow gridblocks, the calculation of individual phase and component rates
requires detailed saturation distribution in the fault. This can be rigorously
obtained only through simulation with fault blocks explicitly represented in the
model. Otherwise, approximations based on assumptions of fluid distributions in
the fault will be needed. Two user-controlled options are currently available: a
fully-mixed option and a fully- segregated option.
In the fully-mixed option, the inflow gridblock’s phase rates are calculated using
the assumption of a fully-mixed wellbore, i.e., all inflow gridblocks have the same
inflow stream which has the composition of the combined outflow streams. With
this assumption, the water surface rates and the hydrocarbon component molar
rates for the inflow gridblocks are
+
- qw
q wk = T Ik λ kt ------+- ( P k – P f – α fk ), (7-6)
Q
+
- qi
q ik = T Ik λ kt ------+- ( P k – P f – α fk ), i = 1, N c , (7-7)
Q
where
+
+ +
qw ≡ ∑ q wk , (7-8)
+k
+ +
q i ≡ ∑ q ik , (7-9)
k
and
+
+ +
Q ≡ ∑ Qk . (7-10)
k
Summations in Equations 7-8 through 7-10 are taken over all gridblocks with
positive total volumetric rates (i.e., Qk > 0.). It can be shown that Equations 7-4 to
7-10 satisfy the material balance automatically.
The second option assumes fully segregated flow in the fault, i.e., the water
produced from all outflow gridblocks will preferentially flow into the bottom
inflow gridblocks and the gas produced from all outflow gridblocks will
preferentially flow into the top inflow gridblocks. Numerically, the segregated
flow option can be represented by the following equations:
- + -
Q wk = min ( Q w, avl , Q k ), (7-11)
+
- - qw
q wk = Q wk ------+- , (7-12)
Qw
- + - -
Q ok = min ( Q o, avl , Q k – Q wk ), (7-13)
+
∑ qok xik
- k -
q iok = Q ok --------------------
+
-, i = 1, N c , (7-14)
Qo
+
∑ qgk yik
- - k - -
q igk = ( Q k – Q wk – Q ok ) --------------------
+
-, i = 1, N c , (7-15)
Qg
- -
where q wk is the water surface inflow rate for gridblock k, q imk (m = o, g) is the
component molar inflow rate for gridblock k, and
+
+
Qm = ∑ TIk λmk ( Pk – Pf – αfk ) , m = o, g, w, (7-16)
k
+ k rmk
q mk = T Ik ---------- ρ m ( P k – P f – α fk ), m = o, g. (7-17)
μ mk
-
Symbol Q k is the total volumetric flow rate for inflow gridblock k in Equation 7-
+
2, and Q m, avl is the available volumetric outflow rate for phase m (m = w, o, g)
that has yet to be distributed to inflow gridblocks. For each conductive fault, the
inflow gridblocks are sorted based on the descending order in depth. Starting from
the inflow gridblock that has the highest depth, Equations 7-1 and 7-5 are then
used to calculate the individual phase rates gridblock-by-gridblock. It can be
readily shown that this model also guarantees that the material balance for each
component is automatically satisfied.
It should be noted that the segregated flow option presented here neglects any
mass transfer effects inside the fault. As in the fully-mixed option, the fault
pressure is calculated by Equation 7-1, and the phase and/or component rates for
outflow gridblocks are calculated from Equations 7-4 and 7-5. For the inflow
gridblocks, the phase and/or component rates are calculated from Equations 7-12,
7-14, and 7-15.
The options implemented in VIP allow the user to specify any input faults as
being conductive. For each conductive fault, an additional constraint equation
(similar to an implicit well constraint equation), Equation 7-1, is added to the
simulator and fully coupled with the reservoir equations to calculate the fault
pressure. For outflow faulted gridblocks (positive Qk), Equations 7-4 and 7-5 are
added to the water and hydrocarbon mass balance equations and their derivatives
with respect to the gridblock pressure and the fault pressure are added to the
corresponding Jacobian terms.
For inflow faulted gridblocks, the situation is more complex because Equations 7-
9 and 7-10 (applicable to the fully-mixed option), or Equations 7-15, 7-17, and 18
(applicable to the fully-segregated option) for each inflow gridblock depend not
only on the gridblock and fault pressures, but also pressures of all outflow
gridblocks. The conductive fault options implemented here have a user-controlled
feature to either lagging the calculations of Equations 7-8 to 7-10 (for fully-mixed
option), and Equations 7-8, 7-16, 7-17 (for fully-segregated option) for inflow
gridblocks by one iteration or fully implicitly evaluating these terms. By
2
implicitly calculating these terms, a maximum of n f ⁄ 4 additional non-
neighboring pressure derivative terms will be added.
The gravity term in the fault plane, αfk, is evaluated using a simple mobility
weighted averaging method. Results of two test runs using this method and a more
complex volume averaging method for a conductive fault problem indicate that
the model performance is insensitive to the gravity term.
8
Gas-Water, Water-Oil, and Black-Oil Options
8.1 Introduction
VIP-EXECUTIVE has the ability to simulate two-phase gas-water or water-oil
systems using simplified gas-water and water-oil options. When these special
options are used, only one hydrocarbon equation is solved in the IMPLICIT
formulation; thus speeding up the simulation. The coefficient generation is
unaffected. These options also enable reduced data input for saturation and PVT
tables, since only oil phase (for water-oil) or gas phase (for gas-water) properties
are required.
The simulator also has the ability to simulate a true non-volatile black-oil system
where the K-value of the oil component is near zero.
2. Printout of the pressure divided by the gas compressibility factor (P/Zg) in the
gas property tables,
■ The BGTAB table is used instead of the BOTAB table to specify the PVT
properties.
■ The SGT and SWT tables originally used to specify the relative permeability
and capillary pressure curves for the gas/oil and water/oil systems are
combined into the SGT table. The new SGT table is used to specify gas and
water relative permeability and gas/water capillary pressure curves. The SWT
table should not be specified if the simplified gas/water option is chosen.
■ The IEQUIL card is changed to specify the depth of the gas/water contact and
gas/water capillary pressure at the contact instead of the depths of the gas/oil
and water /oil contacts and the gas/oil and water/oil capillary pressures at the
contacts as the original program. The program sets the depths of the gas/oil
and water/oil contacts to the depth of the gas/water contact.
■ For end-point scaling, only gas saturation related end points are necessary.
These are SGL, SGU, SGR, and SGRW.
■ The BOTAB card used to specify the PVT properties but gas parameters are
omitted.
■ The IEQUIL card is changed to specify the depth of the water/oil contact and
water/oil capillary pressure at the contact. The gas-oil contact and capillary
pressure at gas-oil contact are not required.
The black-oil option now implemented aims at eliminating the oil vaporization
problem stated above. This option is a simplistic approach to the true conventional
black-oil model where the oil component is non-volatile. An exact black-oil
approach would be to set the heavy component K-values to zero (i.e., the heavy
component molecular weight equals the input residual oil molecular weight,
WTRO). This exact approach requires further modifications to VIP-CORE and
the simulation modules because of zero K-values. The option actually
implemented calculates the heavy component molecular weight by adding 2 x 10-
8
(as opposed to 2.0 without this option) to WTRO and calculates the light
component molecular weight by subtracting 10-8 (as opposed to 1.0 without this
option) from the constant gas molecular weight (29. x GR). The resulting heavy
component K-values are in the order of 10-10. This will eliminate any significant
vaporization of the oil component.
■ If multiple PVT tables are specified, the WTRO (molecular weight of residual
oil) values in all BOTAB tables must be the same, and the GR values in all
BOTAB tables must also be the same.
9
Governing Equations
9.1 Introduction
For a general compositional model of Nc hydrocarbon components and water, the
basic differential equations are the species mass balance equations of the first Nc -
1 hydrocarbon components, the overall hydrocarbon mass balance equation, and
the water mass balance equation. In VIP-EXECUTIVE, water is treated as an inert
phase and the formulation of the hydrocarbon mass balance equations follows that
of Young and Stephenson.5 The black oil option is treated as a special case of the
general compositional model.
∂ ( φFz i ) k k ro ρo Mo g
------------------- = ∇ ⋅ ⎛ ------------ ρ o x i⎞ ∇ ⎛ P o – ------------------------------------- D⎞
∂t ⎝ μo ⎠ ⎝ 5.6146 × 144g c ⎠
k k rg ρg Mg g
+ ∇ ⋅ ⎛ ------------ ρ g y i⎞ ∇ ⎛ P o + P cg – ------------------------------------
- D⎞ + q o x i + q g y i
⎝ μg ⎠ ⎝ 5.6146 × 144g c ⎠
i = 1, ..., Nc - 1 (9-1)
∂--------------
( φF )- k k ro ρo Mo g
= ∇ ⋅ ⎛ ------------ ρ o⎞ ∇ ⎛ P o – ------------------------------------- D⎞
∂t ⎝ μo ⎠ ⎝ 5.6146 × 144g c ⎠
k k rg ρg Mg g
+ ∇ ⋅ ⎛ ------------ ρ g⎞ ∇ ⎛ P o + P cg – ------------------------------------- D⎞ + q o + q g (9-2)
⎝ μg ⎠ ⎝ 5.6146 × 144g c ⎠
o
∂ ( φB w S w ) ⎧ ⎛ kk rw ⎞ ⎛ ρw Bw Mw g ⎞ ⎫
- = ∇ ⋅ ⎨ ⎜ ---------- B w⎟ ∇ ⎜ P o – P cw – ------------------------------------- D⎟ ⎬ + q w
------------------------
∂t ⎩ ⎝ μw ⎠ ⎝ 5.6146 × 144g c ⎠ ⎭
(9-3)
f oi ( P o, x i, …, x Nc ) = f gi ( P o, y i, …, y Nc ) (9-4)
The fugacities of component i in phase j (j = o, g), foi, and fgi are calculated from
an equation of state that is discussed next. For the black oil option, the phase
equilibrium is described by equilibrium K-values (Ki):
yi
K i ( P o ) = ----, i = 1, 2 (9-5)
xi
z i = ( 1 – L )y i + Lx i, i = 1, ... , N c (9-6)
Here, L is the mole fraction of oil phase in the total hydrocarbon mixture (i.e., one
mole of hydrocarbon mixture separates into L moles of oil phase and 1 - L moles
of gas phase).
Nc
∑ xi = 1 (9-7)
i=1
Nc
∑ zi = 1 (9-8)
i=1
Note that Equations 9-6, 9-7, and 9-8 suggest that the summation of the mole
fractions of all hydrocarbon components in the gas phase also equals one.
z i – Lx i
y i = -----------------, i = 1, ... , N c (9-10)
1–L
Nc – 1
xNc = 1 – ∑ xi (9-11)
i=1
and
Nc – 1
zNc = 1 – ∑ zi (9-12)
i=1
Equations 9-10, 9-11, and 9-12 are then substituted into Equations 9-1, 9-2, and 9-
3, 9-4 (or 9-5), and 9-9 to form 2Nc + 2 independent equations with the same
number of unknowns: xi (i = 1 to Nc - 1), L, zi (i = 1 to Nc - 1), F, Sw, and Po.
Note that the above choice of primary unknowns results in the L-x iteration. In the
VIP-EXECUTIVE simulator, these primary unknowns are chosen for a gridblock
whenever the mole fraction of the liquid phase in the gridblock, L, is less than 0.5.
Otherwise, the first Nc primary unknowns (xi and L) are replaced by yi (i = 1, Nc -
1) and V (which equals 1 - L) to avoid ill-conditioning in the resulting Jacobian
matrix. Numerically, this can be accomplished easily by swapping xi and yi, as
well as L and V in Equations 9-6, 9-7, 9-10, and 9-11. This alternate set of
primary unknowns results in the so-called V-y iteration.
φF = φ ( S o ρ o + S g ρ g ) (9-14)
The water mass balance equation and fugacity equality equations are the same as
Equations 9-3 and 9-4 or 9-5, and zi in Equation 9-8 is replaced by yi. With this
formulation, Equation 9-6 is not needed. Finally, the saturation constraint
equation, Equation 9-9, is replaced by
1 - S o - Sg - S w = 0 (9-15)
Unknowns xNc, yNc, and So are eliminated using the two mole fraction constraint
equations and the saturation constraint equation. The remaining primary
unknowns for the simultaneous (Nc) fugacity equations and (Nc + 1) hydrocarbon
and water mass balance equations are xi (i = 1, ..., Nc-1), yi (i = 1, ..., Nc-1), Sg, Sw,
and P. Before each Newton iteration is performed, the fugacity equality equations
are used to eliminate unknowns xi (i = 1, ..., Nc-1) and y1, or yi (i = 1, ..., Nc-1) and
x1. The choice of the unknowns to be eliminated depends on the relative amounts
of oil and gas phases within each gridblock. Unknowns xi and y1 are eliminated if
ρo So < ρg Sg. Otherwise, yi and x1 are eliminated. See Chapter 18 for a discussion
of the numerical solution of the differential equations and of the iteration
procedures.
The relaxed volume balance concept proposed by Coats, et al. [Reference 73], on
the contrary, always keeps material balances exactly while the saturation
constraint is not satisfied. In effect, this concept amounts to iterating out
saturation constraint error (or volume balance error) rather than mass balances.
Coats, et al.'s results show that iterations and CPU time are significantly less as
compared with the standard version of the fully implicit formulation.
Since the saturation constraint is not satisfied for the relaxed volume balance
option, the saturation error, 1-(Sw+So+Sg), must be taken into account when the
unknown variable So is eliminated during setting up the Jacobian. This modified
Jacobian and residuals are then solved by the linear solver. To keep the material
balances exactly, the unknown updating procedure is modified. The solution
vector for the primary unknowns from the solver is used to calculate interblock
flow terms and well rates using the linearized equations. The mass of each
component in the grid block is then calculated as the mass at the previous time
step plus the net interblock flow minus the production. The composition and
density of each phase and pore volume are then calculated using the updated
pressure. Water saturation is then calculated as the mass of water present divided
by water density and pore volume. For a three-phase grid block, the total
hydrocarbon moles are flashed to obtain oil and gas compositions, moles and
densities. The oil and gas saturations are calculated as their phase moles divided
by density and pore volume. The three saturations do not add to one and thus there
is a saturation constraint error. However, the material balance is exact for all
components.
For the relaxed volume balance option, convergence of the Newton iteration is
controlled by either the errors in saturation constraint (TOLSCN) and well rate
calculation (TOLWCN ), or the maximum change of primary unknows over one
iteration (TOLD). The TOLD criteria for convergence are the same as the
standard version of the fully implicit formulation. The TOLSCN criterion is used
to check the maximum saturation error of all grid blocks. The saturation error is
computed as 1-(Sw+So+Sg). To ensure a good material balance, the well rate
calculated from the well model must agree with the updated well rate using the
solution from linear solver. To converge, the ratio of the difference between these
two rates to the sum of the two rates must be less than TOLWCN.
10
Grid Coarsening
10.1 Introduction
The Grid Coarsening option is a feature which enables the user to combine
adjacent grid blocks into coarser grid blocks, thereby reducing the number of
blocks used for simulation and reduce run time.
∑ ( A i ⋅ PV i )
i
A = -------------------------------
-
∑ PV i
i
The following properties are exceptions:
For DX,DY,DZ,HNET a bulk volume weighted average is used, and the result
multiplied by the number of fine blocks in the appropriate direction.
∑ ( DX i ⋅ BV i )
i
DX = nx ⋅ ------------------------------------
-
∑
BV i
i
∑ ( DY i ⋅ BV i )
i
DY = ny ⋅ ------------------------------------
-
∑
BV i
i
∑ ( DZ i ⋅ BV i )
i
DZ = nz ⋅ ------------------------------------
-
∑ BV i
i
∑ ( HNET i ⋅ BV i )
i
HNET = nz ⋅ --------------------------------------------
-
∑
BV i
i
PV DZ
POR = ------- ⋅ ----------------
BV HNET
The half transmissibilities of the coarse block are calculated from those of the
constituent fine blocks using the tubes in parallel method.
+ - 2
TX = TX = ∑ -------------------------------------------
⎛ 1 1 ⎞
-
j, k
∑ ⎜⎝ -----------
TX
+
- + -----------⎟
TX ⎠
-
i i i
+ - 2
TY = TY = ∑
-------------------------------------------
⎛ 1 1 ⎞
i, k
∑
⎜ ------------ + ----------⎟
⎝ + -⎠
j TY j TY j
+ - 2
TZ = TZ = ∑
---------------------------------------------
⎛ 1 1 ⎞
i, j
∑
⎜ ------------- + -----------⎟
⎝ TZ + TZ k⎠
-
k k
The fine block transmissibility values will include changes applied with the
TOVER cards, but will not include changes applied with the OVER, VOVER
cards. Transmissibility values at coarse block boundaries may be modified
through the non-standard transmissibility multipliers.
Average coarse block permeabilities KX, KY and KZ that are used for
calculating well, influx and polymer properties, and for other uses by VIP-EXEC
are calculated from the average coarse block properties as follows:
-
TX DX
KX = ----------------------- ⋅ --------
2 ⋅ HNET DY
-
TY DY
KY = ----------------------- ⋅ --------
2 ⋅ HNET DX
-
TZ DZ
KZ = --------------- ⋅ --------
2 ⋅ DX DY
The face areas of the coarse block are calculated from the sum of the face
areas of the fine blocks that make a coarse block face.
11
Horizontal and Inclined Well Model
11.1 Introduction
VIP can model horizontal or inclined wells which intersect the grid mesh at
oblique angles areally or in cross section.
This chapter provides input examples and suggestions in using the horizontal or
inclined well option and a technical description of how VIP handles them.
A horizontal or inclined well has a larger contact area with the reservoir than a
vertical well, as illustrated in Figure 11-1. Therefore, production or injection rate
per unit well length is less than that of a vertical well for the same specified total
rate. This implies the following immediate advantages in practice:
■ It can reduce coning or cusping growth because of the lower fluid velocity in
the reservoir.
■ They allow drilling into many parts of a reservoir from an off-shore platform.
Because of these potential advantages, horizontal and inclined wells have been
drilled around the world for many different types of reservoirs: Bakken Shale
(thin reservoir), Austin Chalk, Bakken Shale (naturally fractured reservoir),
Prudhoe Bay, Bima Field, Helder Field (formation with cusping and coning),
Zuidwald Field (gas reservoirs), Cold Lake, and Mid-Continental (EOR
application). However, not all reservoirs show improved performances with
horizontal wells.
V2
Δ pk
Δpf
Δph
V1
Flow Direction
ΔP loss = ΔP f + ΔP k + ΔP h (11-1)
where
2
1 fLρv
Δp f = --------- -------------- (11-2)
144 2g c D
1 1 2
Δp k = --------- --- ρΔ ( v ) (11-3)
144 2
1
ΔP h = --------- ρg c ΔZ (11-4)
144
where f is the Moody friction factor, ρ is fluid mass density in lb/ft3, v is fluid
velocity in ft/s, gc is gravity acceleration in ft/s2; ΔZ is perforation depth
difference in ft, and L is well perforation length in ft.
R e = vρD
----------- (11-5)
μ
where μ is the viscosity of fluid in lb-s/ft2. The friction factor for a single-phase
flow, i.e., in gas or water injectors, is computed by using a pipe flow equation for
a single phase. For flow with Re < 2000, it is expressed by:
f = 64
------ (11-6)
Re
( 10 ⁄ 9 )
1 6.9 ⎛ e ⎞
----- = – 1.8 log ------- + ⎜ ------------⎟ (11-7)
f R e ⎝ 3.7D⎠
where e is the absolute roughness of the wellbore in feet and can represent the
effect of turbulence created by inflow through perforations. Flow is in the critical
zone from Re = 2000 to 4000. In this zone, the weighted average friction factor
from Equations 11-6 and 11-7 is used.
In general, producing wells have multiphase flow, but Equations 11-6 and 11-7
also may be used where a uniform mixture and no slipping among phases are
assumed. Beattie and Whalley7 examined experimental data of multiphase flow
friction loss and concluded that Equation 11-7 gave better matches in all of the
ranges of Reynold’s number by defining the effective density ρ and viscosity μ as
S S
1 ⁄ ρ = ----g- + -----L- (11-8)
ρg ρL
μ = μ L ( 1 – β ) ( 1 + 2.5β ) + μ g β (11-9)
where
ρL Sg
β = -----------------------------
- (11-10)
ρL Sg + ρg SL
VIP-EXECUTIVE uses Equation 11-7 with Equations 11-8, 11-9, and 11-10 for
producers as a default, while injectors use Equations 11-6 and 11-7.
adjusted to get a proper friction loss if field data or sound experimental data are
available.
The kinetic energy loss, ΔPk, is due to the velocity change along the well. The
flow velocity changes because of pipe diameter changes and influx through
perforations. The flux through perforations is assumed to enter a pipe with zero
longitudinal velocity, then be pushed to the pipe flow velocity in a pipe segment.
However, in most cases the kinetic energy loss is much smaller than the friction
pressure loss in well pipe flow.
On the other hand, pressure loss due to gravity can be significant if a well is
inclined. Therefore, the perforation depth should be computed accurately and
specified in VIP-EXECUTIVE input data.
The total pressure loss in a pipe segment is computed using Equations 11-2, 11-3,
and 11-4 for each perforation segment, then is added to the well pressure in cases
of producers.
P w ( i ) = P w ( i – 1 ) + ΔP loss ( i ) (11-11)
where Pw(i) is the well pressure of the i’th perforation of a well. For injectors,
pressure loss is subtracted from well pressure.
VIP-EXECUTIVE treats these pressure loss terms explicitly, and the wellbore
fluid gradient is computed using material balance equations. Furthermore, the
gradients computation is done only at the first inner iteration. (Refer to Section
39.6 for information on wellbore gradient calculations.)
Total well production can be computed from Equation 11-12 for an oil producer,
for example:
L
k ro, i
Q o = 0.001127 ∑ WI i K i H i ------------------- [ P grid, i – P w, i ] (11-12)
B o, i μ o, i
i=1
where the summation is performed over perforations and subscript i indicates the
perforation number. Here, WI is the well index; KH is the permeability-thickness
of the gridblock; kro and μo are oil relative permeability and viscosity,
respectively; Bo is the oil formation volume factor; and Pgrid is gridblock
pressure. In the next section, we discuss how VIP-EXECUTIVE computes
permeability-thickness KH and well index WI for horizontal or inclined wells.
where kx, ky, and kz are rock permeabilities in x, y, and z directions, respectively.
VIP-EXECUTIVE extends the theory to computation of a perforation length
inclined arbitrarily in space. Projected perforation lengths Lx, Ly, and Lz onto each
coordinate axis are transformed, then perforation length Lp in the transformed
plane is represented by the following:
2 2 2
Lp = Lx + Ly + Lz (11-14)
where Lx, Ly , and Lz are x, y, and z components, respectively. These, in turn, are
transformed into the isotropic medium:
k
L x = l ----e sin θ cos ω (11-15)
kx
k
L y = l ----e sin θ sin ω (11-16)
ky
k
L z = l ----e cos θ (11-17)
kz
where l is the physical well perforation length, θ is the angle of the well segment
with respect to vertical direction, and ω is the angle with respect to the x-axis as
shown in Figure 11-3.
Lx
Ly ω
Lz
x
θ
l
2π -
WI = ------------------------ (11-18)
rb
ln ⎛ -----⎞ + s
⎝ r w⎠
where rb is the equivalent radius, rw is the well radius, and s is the skin factor. In
computing rb, an analogy to Peaceman’s equation9 for a vertical well is adopted.
However, the equivalent radius and well radius are obtained in the transformed
plane similarly to the permeability-thickness. Peaceman’s equivalent radius for a
well parallel to the y-axis is expressed as
1⁄2 2 1⁄2 2
( k z ⁄ k x ) Δx + ( k x ⁄ k z ) Δz
R by = 0.14 --------------------------------------------------------------------------------
1⁄4 1⁄4
- (11-19)
0.5 [ ( k z ⁄ k x ) + ( kx ⁄ kz ) ]
where notation Rby indicates a well parallel to the y-axis. The numerator can be
simplified using properties in the transformed isotropic medium. Transferring the
denominator of Equation 11-19 to Rwy , the following can be concluded.
k ey k ey
R ey = 0.14 ⎛⎝ -------⎞⎠ Δx + ⎛⎝ -------⎞⎠ Δz
2 2
(11-20)
kx kz
1---
where: 2 (11-21)
k ey = ( k x k z )
k ex k ex
R ex = 0.14 ⎛ -------⎞ Δy + ⎛ -------⎞ Δz
2 2
(11-22)
⎝ ky ⎠ ⎝ kz ⎠
1---
where, 2 (11-23)
k ex = ( k y k z )
k ez k ez
R ez = 0.14 ⎛ ------⎞ Δx + ⎛ ------⎞ Δy
2 2
(11-24)
⎝ kx ⎠ ⎝ ky ⎠
1---
where, 2 (11-25)
k ez = ( k x k y )
For an arbitrarily inclined well, i.e., for any (θ,ω) in Figure 11-3, VIP-
EXECUTIVE approximates the equivalent radius as follows.
2 2 2
R be = r ex + r ey + r ez (11-26)
where
Note that Equation 11-26 yields Equations 11-20, 11-22, and 11-24 as special
cases and any equivalent radii in Equation 11-26 stay in an ellipsoid created by
them.
The wellbore radius in the transformed plane is expressed in a similar manner. For
example, a horizontal well that is parallel to the y-axis is approximated by
Rw
R wy = ------ ( k ey ⁄ k x + k ey ⁄ k z ) (11-28)
2
where Rw is the true pipe radius and Rw(key/kx)0.5 and Rw(key/kz)0.5 are the well
radii in major and minor axes. Similarly,
Rw
R wx = ------ ( k ex ⁄ k y + k ex ⁄ k z ) (11-29)
2
Rw
R wz = ------ ( k ez ⁄ k x + k ez ⁄ k y )
2
For arbitrary values of (θ,ω), the transformed well bore radius is approximated by
2 2 2
R we = r wx + r wy + r wz (11-30)
where
R be
R b = --------- R w (11-32)
R we
The permeability-thickness and the well index computed using Equations 11-13
through 11-32 are implemented in VIP; but users can override them if correct
values are known. Radius Rwe also forms an ellipsoid for arbitrary angles (θ,ω), as
shown in Figure 11-4..
y’
Rwy
x’
Rwx
Rwz
z’
The equivalent radius as computed above may not always be accurate because of
the assumptions on which Peaceman’s equation is based. Therefore, calibrate the
perforation WI’s in proportion to the computed values in such a way that WI from
a well test or production data is honored.
FPERF
WELL L IW JW LENGTH PWDEP DIAM ROUGH ANGLV ANGLA
ANGLA Angle of the well in areal plane with respect to the x-axis
in degrees (ω in Figure 11-3).
■ The perforations must be ordered on the FPERF card starting with the
perforation closest to the well head and ending with the farthest perforation.
Furthermore ANGLV values must follow the convention shown in Figure 11-
7. This is necessary for proper computation of friction coefficients by Beggs
and Brill. The angle is measured by drawing an arc from the “tail” of the
segment to the +z axis as shown in Figure 11-7.
■ When there are other vertical wells, users should repeat the FPERF card as
shown in Figure 11-8.
■ In general, the friction and kinetic pressure losses are small, but the
hydrostatic pressure loss can be significant in an inclined well. If the well
block depth changes significantly, users should input accurate values of pipe
segment depths in PWDEP rather than letting the simulator compute them.
VIP computes, as a default, the pipe segment depth using LENGTH and
ANGLV starting from the grid block center of the first perforated grid block
in FPERF card. If the first perforated well segment does not start at the center,
VIP would be using depths of well segments.
FPERF
WELL L IW JW LENGTH PWDEP DIAM ROUGH ANGLV ANGLA SKIN
FPERF
WELL L IW JW K LENGTH PWDEP DIAM ROUGH ANGLV ANGLA
BEGGS ON σ
where σ is the value of liquid surface tension in dyne/cm. The Beggs and Brill
method takes into account various flow regimes depending on the angle
(ANGLV) and it is often considered more accurate than other correlations for
multiphase flow in an inclined pipe. However, be aware that all correlations
have a range of applicable conditions which are only rough estimates.
The Beggs and Brill correlation can be turned off by deleting the keywords or
using the keyword entry:
BEGGS OFF σ
■ If you want to disregard the friction and kinetic pressure losses from the
computation,add the keyword:
NOFRICTION
■ Do not use DTOP or DBOT, the top and bottom perforation depths, since they
are meaningless in horizontal/inclined wells.
■ Friction loss is computed only for the perforated intervals that appear in the
FPERF card. Therefore any pressure loss between the wellhead and the first
perforation must be modeled using hydraulic tables.
■ The well index and permeability-thickness for an inclined well are computed
by transforming an original anisotropic diffusion equation into an equivalent
isotropic equation as discussed in Section 11.5. If users know these values
through field data such as well tests or other production data, it is suggested
you distribute the field values in proportion to the simulator-computed KH or
WI. In the case of vertical wells, the areal permeabilities KX and KY are
multiplied by DZNET/DZ. It is less clear what should be done for inclined or
horizontal wells, but the same multiplication is performed on KX and KY.
11.6 Examples
Consider an inclined well that is perforated in two layers which are separated by a
layer of grid blocks as shown in Figure 11-9. The portion of the well which is not
perforated still has the friction of gravity pressure loss. Therefore users are
required to provide input for the entire portion of the well and set STAT to OFF as
shown in Figure 11-9.
FPERF
WELL L IW JW STAT LENGTH DIAM ROUGH ANGLV ANGLA
1 1 1 1 ON 50 0.25 1.E-4 35 0.
X 2 1 1 OFF 25 0.25 1.E-4 50 0.
X 2 2 1 OFF 60 0.25 1.E-4 70 0.
X 3 3 1 OFF 52 0.25 1.E-4 85 0.
X 3 4 1 ON 70 0.25 1.E-4 60 0.
Figure 11-9: Well With Noncontiguous Perforations and Sample VIP Input
The inclined wells are assumed to be straight in this example and both ends of the
wells are also assumed to be located at the center of the grids. The well section
lengths are computed based on these assumptions. The portion of the reservoir
that has a well is shown in Figure 11-11 with the section length in each grid. Since
both wells run in parallel, only one cross section is shown.
9 15 25.02 18 17 50.03
10 15 50.03 19 17 50.03
11 15 50.03 20 17 50.03
12 15 16.66 21 17 50.03
12 16 33.81 22 17 50.03
13 16 50.03 23 17 33.81
14 16 50.03 23 18 16.67
15 16 50.03 24 18 50.03
16 16 50.03 25 18 50.03
17 16 50.03 26 18 25.02
FPERF
WELL IW JW L LENGTH DIAM ROUGH ANGLV ANGLA
C FT FT FT DEG DEG
7 14 9 15 25.02 0.25 1.E-5 87.9786 90.
7 14 10 15 50.03 0.25 1.E-5 87.9786 90.
7 14 11 15 50.03 0.25 1.E-5 87.9786 90.
7 14 12 15 16.66 0.25 1.E-5 87.9786 90.
C
7 14 12 16 33.81 0.25 1.E-5 87.9786 90.
7 14 13 16 50.03 0.25 1.E-5 87.9786 90
7 14 14 16 50.03 0.25 1.E-5 87.9786 90.
7 14 15 16 50.03 0.25 1.E-5 87.9786 90.
7 14 16 16 50.03 0.25 1.E-5 87.9786 90.
7 14 17 16 50.03 0.25 1.E-5 87.9786 90.
C
7 14 18 17 50.03 0.25 1.E-5 87.9786 90.
7 14 19 17 50.03 0.25 1.E-5 87.9786 90.
7 14 20 17 50.03 0.25 1.E-5 87.9786 90.
7 14 21 17 50.03 0.25 1.E-5 87.9786 90.
7 14 22 17 50.03 0.25 1.E-5 87.9786 90.
PRINT WLLYR
This is useful as it indicates how much production each well segment produces.
12
Hydraulically Fractured Well Option
12.1 Introduction
VIP-EXECUTIVE provides special features for the simulation of the detailed
performance of a single hydraulically fractured well. The fracture is assumed to
be a vertical fracture, and the simulation involves modelling only one-fourth of
the fracture and one-fourth of the drainage area of the well, as shown in Figure 12-
1. Flow within the packed fracture is controlled by two major components:
fracture closure and non-Darcy flow. The type and concentration of proppant may
be varied for each fracture gridblock.
R5000.0.1 12-173
VIP-EXECUTIVE® Technical Reference Guide Landmark
Figure 12-2: Areal Finite-difference Grid of One Quadrant of the Drainage Area.
12-174 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Cooke70 presented experimental results on the effects of closure stress, fluids, and
temperature on the permeability of several different proppants. The following
example is Cooke’s figure 3 for 10-20 mesh sand.
R5000.0.1 12-175
VIP-EXECUTIVE® Technical Reference Guide Landmark
ΔP μν 2
------- = ------ + αν (12-1)
ΔL kD
If the constant α or the velocity ν approaches zero, then the second term goes to
zero and the remaining expression is equal to Darcy’s law. Cornell and Katz72
reformulated the above equation as follows:
ΔP
------- = μν
2
------ + βρν (12-2)
ΔL kD
where the constant α is replaced by the product of the fluid density and the β
turbulence factor, which is a function of the porous media. This can be rewritten
as follows:
ΔP μν βρνk D⎞
------- = ------ ⋅ ⎛⎝ 1 + ---------------- (12-3)
ΔL kD μ ⎠
kD ΔP
ν = ----------------------------------- ⋅ -------
βρνk D⎞ ΔL
μ ⎛ 1 + ----------------
⎝ μ ⎠ (12-4)
kD ΔP
= ------ ⋅ F ND ⋅ -------
μ ΔL
1
F ND = -------------------------
-
βρνk D (12-5)
1 + ----------------
μ
1
F ND = ------------------------------------------------------------- (12-6)
– 9 βρνk D
1 + 5.6555 ×10 ⋅ ----------------
μ
12-176 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
b-
β = ----- (12-7)
a
kD
where kD is the effective permeability at a particular level of stress and a and b are
constants which are a function of the proppant type. These are shown graphically
in Cooke’s figure 13 and in tabular form in his table 2, which is shown in
Table 12-1 below.
Sand Size a b
1. Calculate the flow velocities using the last iterate values, including the effects
of fracture closure stress.
3. Repeat steps 1 and 2 for each flow face along the fracture until a minimum
relative FND change is achieved.
The initial timesteps will most likely take a few more outer iterations to achieve
convergence, due to the possibly large changes in the initial flow velocities. The
initial timestep size should be quite small, similar to that which would be used for
a single-well radial grid simulation.
R5000.0.1 12-177
VIP-EXECUTIVE® Technical Reference Guide Landmark
12-178 R5000.0.1
Chapter
13
Hysteresis
13.1 Introduction
The hysteresis options allow the use of relative permeabilities and capillary
pressures that are a function of both phase saturation and the history of the phase
saturation. The option is useful for the rigorous simulation of fluid flow in
reservoirs where the saturation of a fluid phase within a gridblock does not change
monotonically.
Figure 13-1 is a set of drainage and imbibition-drainage curves, where krnD is the
user-specified drainage curve, krnID is the bounding imbibition-drainage curve,
and krnID’ is a generated intermediate imbibition-drainage curve. The end points
of the bounding imbibition-drainage curve are Snu and Sntr.
krn
D
krn
ID
krn
kID ’
rn
Snr XSNR Snhmx Sntr Snu
Sn
Carlson’s Method
S nhmx
XSNR = --------------------------
- (13-1)
1 + CS nhmx
S nu – S ntr
C = ----------------------
- (13-2)
S nu S ntr
S nhmx – S nr
XSNR = S nr + --------------------------------------------- (13-3)
1 + C ( S nhmx – S nr )
S nu – S ntr
C = -----------------------------------------------------
- (13-4)
( S nu – S nr ) ( S ntr – S nr )
Snu is the maximum possible nonwetting saturation for the gridblock. For oil
relative permeability hysteresis, Snu = 1 - connate water saturation. Snr is the
critical nonwetting phase saturation for the drainage curve.
ID′ D
k rn ( S n ) = k rn ( S npd ) (13-5)
2 4X sn
S npd = S nr + 0.5 X sn + X sn + ----------
- (13-6)
C
and
X sn ≡ S n – XSNR (13-7)
Linear Interpolation
D
ID′ ( S n – XSNR )k rn ( S nhmx )
k rn ( S n ) = ---------------------------------------------------------
- (13-8)
( S nhmx – XSNR )
ID′ D *
k rn ( S n ) = k rn ( S n ) (13-9)
where
* S n – XSNR
S n = S nr + ----------------------------------- ( S nhmx – S nr ) (13-10)
S nhmx – XSNR
An additional option is included to allow the user to directly enter the bounding
imbibition-drainage curve, krnID, along with the bounding drainage curve, krnD.
krn
D
krn
ID
krn
ΔSn
kID ’
rn l
Snr XSNR Snhmx Sntr SnID Snu
Sn
ID
ΔS n = S n – S nhmx (13-11)
where
ID ID D
k rn ( S n ) = k rn ( S nhmx ) (13-12)
The value of SnID is determined by reverse table look-up from the user-specified
bounding imbibition-drainage curve. SnID is the saturation where the bounding
imbibition-drainage curve has relative permeability equal to the drainage relative
permeability at the reversal saturation, Snhmx. Therefore, the relative permeability
for any saturation on the scanning curve can be found from the bounding curve
by:
ID′ ID
k rn ( S n ) = k rn ( S n + ΔS n ) (13-13)
Utility Data
Tabular Data
The RPHYSO and RPHYSG data cards define the specific method of hysteresis to
be used for the oil or gas phase, respectively. If either card is entered without a
method specified, the default method is “LINEAR.” If either the RPHYSO or
RPHYSG data card is entered, then the user must also supply additional saturation
function data, depending upon which method is being used. For user-specified
imbibition curves, the ISAT array continues to refer to the drainage saturation
functions. An additional required array, ISATI, refers to the saturation functions to
be used for the imbibition functions. Otherwise, for either “CARLSON”,
“LINEAR”, or DRAINAGE”, the SGTR keyword for gas phase hysteresis and the
SOTR keyword for oil phase hysteresis must be included in the drainage table
referred to by ISAT.
These three curves are obtained when displacement is carried out completely to
connate saturation during either the drainage or imbibition process.
Pc Scanning Curves
b c
Snc
Swl Sw
If the drainage process is reversed before the wetting phase saturation reaches the
end point, Swl, the capillary pressure follows a curve of type(d) as shown in
Figure 13-4. Similarly, if the imbibition process is reversed before the nonwetting
phase saturation reaches the end point, Snc, then the capillary pressure follows a
curve of type(e) as shown in Figure 13-4. The behavior described here is valid
only when the system previously experienced a complete drainage displacement.
In such a case, all subsequent drainage cycles follow the secondary drainage
curve, type(c), as shown in Figures 13-3 and 13-4.
Killough provides the following equations for scanning capillary pressure curves
as shown in Figure 13-4.
Pc A
•
e
b c
d
•
B
Snc
d c c b
Pc ( Sw ) = Pc ( Sw ) – F ( Sw ) [ Pc ( Sw ) – Pc ( Sw ) ] (13-14)
where
1 - – 1---
--------------------------
Sw – Sw + ε ε
A
F ( S w ) = -----------------------------------------
- (13-15)
1 1
-------------------------------- – ---
–S +ε ε
max A
S w w
e b c b
Pc ( Sw ) = Pc ( Sw ) + F ( Sw ) [ Pc ( Sw ) – Pc ( Sw ) ] (13-16)
where
1 - – 1---
--------------------------
Sw – Sw + ε ε
B
F ( S w ) = -------------------------------------
- (13-17)
1
----------------------------- – ---1
S –S +ε ε
B
w wl
In the above equations, ε is a parameter affecting the shape of the scanning curves,
with a recommended value between 0.05 and 0.1.
The problem of saturation reversal during the primary drainage process requires a
different treatment. In this case, rather than follow a scanning curve back to the
residual nonwetting phase saturation, Snc, a new value of residual nonwetting
phase saturation, S*nc , is determined. The value of S*nc depends on the maximum
nonwetting phase saturation, SAn . Figure 13-5 demonstrates this process, as
described by Aziz and Settari.14
Pc A
Snc
Snc*
A
* Sn
S nc = ---------------------------------------------------
- (13-18)
1 1 A
1 + ------- – ---------------- S n
S nc 1 – S wl
The equations given above for curves (d) and (e) are used for this case too;
however, the bounding curves are normalized so that they cover the interval (Swl,
1-Snc*) rather than (Swl , 1-Snc).
Another situation occurs if a scanning imbibition process that begins at SAw (Figure
11-6) experiences a second reversal. In such a situation, a new drainage scanning
curve (f) is created that intersects the first scanning curve at reversal saturation S*w
and reaches the bounding drainage curve at starting point SAw . The new imbibition-
to-drainage scanning curve at the second reversal saturation, S *w , has the equation:
* b * * * c * b *
Pc ( Sw ) = Pc ( Sw ) + F ( Sw ) [ Pc ( Sw ) – Pc ( Sw ) ] (13-19)
where
1
--------------------------- – 1---
Sw – Sw + ε ε
c *
* *
F ( S w ) = ------------------------------------ (13-20)
1
--------------------------- – 1---
S –S +ε ε
c A
w w
Pc A
•
d f
b c
•
•
C
Snc
Quantity Scw is the unknown and can be determined by setting the equations for the
two scanning curves, Equations 13-14 and 13-19, equal to one another at reversal
point S*w .
Analysis of the gas-oil system is analogous to the water-oil system and is not
presented here.
Utility Data
Tabular Data
The capillary pressure hysteresis option requires additional data only in VIP-
CORE. The user must enter the PCHYSW data card to activate the water-oil
capillary pressure hysteresis option. The PCHYSG data card is entered to activate
the gas-oil capillary pressure hysteresis option. These two data cards contain
control parameters for the scanning curve shape parameter, the maximum allowed
level of scanning curves, the tolerance for defining saturation reversals, the
maximum or minimum saturation value for which hysteresis will be allowed, and
the selection of initializing on imbibition curves.
If either the PCHYSW or the PCHYSG data card is entered, the user must also
supply additional saturation function data. The ISAT array continues to refer to
the drainage saturation functions. The table indicated by ISAT must be modified
to include the secondary capillary pressure drainage curve, curve (c) in Figure 13-
3, in addition to the primary drainage curve. An additional required array, ISATI,
refers to the saturation functions for use for imbibition.
14
IMPES Stability
14.1 Introduction
When the IMPES option is chosen, the size of the timestep is limited by stability
considerations. In this chapter, we derive the maximum stable timestep, first for
the cases of two- and three-phase immiscible flow, and then for the case of two-
phase compositional flow.
∂( φρ o S o ) kk ro ρ o
---------------------- = ∇ ⋅ ⎛⎝ ----------------⎞⎠ ∇P o + q o (14-1)
∂t μo
∂( φρ g S g ) kk rg ρ g
---------------------- = ∇ ⋅ ⎛ ----------------⎞ ∇P g + q g (14-2)
∂t ⎝ μg ⎠
∂( φB w S w ) kk rw B w
------------------------ = ∇ ⋅ ⎛ ------------------⎞ ∇P w + q w (14-3)
∂t ⎝ μw ⎠
Assume one-dimensional flow and focus on a region not containing any wells.
Also, assume constant density and zero capillary pressure. Define the phase
mobilities as
k ro k rg k rw
M o = ------
- , M g = ------
- , M w = ------- (14-4)
μo μg μw
∂( φS o ) ∂⎛ ∂P
---------------- = kM o ------⎞⎠ (14-5)
∂t ∂x ⎝ ∂x
∂( φS g ) ∂P
---------------- = ∂ ⎛ kM g ------⎞ (14-6)
∂t ∂x ⎝ ∂x ⎠
∂( φS w ) ∂P
----------------- = ∂ ⎛ kM w ------⎞ (14-7)
∂t ∂x ⎝ ∂x ⎠
PVj ( SO jn + 1 – SO jn )
------------------------------------------------ = TX j – 1 ⁄ 2 MO jn– 1 ( P j – 1 – P j ) – TX j + 1 ⁄ 2 MO jn ( P j – P j + 1 ) (14-8)
Δt
PVj ( SG jn + 1 – SG jn )
------------------------------------------------ = TX j – 1 ⁄ 2 MG jn– 1 ( P j – 1 – P j ) – TX j + 1 ⁄ 2 MG jn ( P j – P j + 1 ) (14-9)
Δt
PVj ( SW jn + 1 – SW jn )
-------------------------------------------------- = TX j – 1 ⁄ 2 MW jn– 1 ( P j – 1 – P j ) – TX j + 1 ⁄ 2 MW jn ( P j – P j + 1 ) (14-10)
Δt
The subscript j is the block index, while the superscript refers to the timestep. The
acronym IMPES means implicit pressure and explicit saturation. That means that
the mobilities are evaluated from saturations at the “old” timestep level, and the
saturations at the “new” time level, n + 1, can then be calculated explicitly.
TX j – 1 ⁄ 2 MT jn– 1 ( P j – 1 – P j ) – TX j + 1 ⁄ 2 MT jn ( P j – P j + 1 ) = 0 (14-11)
MT jn = MO jn + MG jn + MW jn
VT = TX j + 1 ⁄ 2 MT jn ( P j – P j + 1 ) (14-12)
MG n MW n
FG jn = -----------j- , FW jn = ------------j- (14-13)
MT jn MT jn
PV j ( SW jn + 1 – SW jn )
------------------------------------------------- = VT ( FW jn– 1 – FW jn ) (14-14)
Δt
PV j ( SG jn + 1 – SG jn )
------------------------------------------------ = VT ( FG jn– 1 – FG jn ) (14-15)
Δt
There are various definitions for the stability of a difference equation. Angel and
Bellman62 give the following useful definition: “We will take stability to mean
that the effect of an error made in one stage of the computation is not propagated
into larger errors in later stages of computation. In other words, local errors are
not magnified by further computation.”
PV
------- [ SW jn + 1 – SW jn ] = VT [ f w ( SW jn– 1 ) – f w ( SW jn ) ] (14-16)
Δt
PV
------- [ ( SW jn + 1 + EW jn + 1 ) – ( SW jn + EW jn ) ] = VT [ f w ( SW jn– 1 + EW jn– 1 ) – f w ( SW jn + EW jn ) ] (14-17)
Δt
Subtraction of Equation 14-16 from Equation 14-17 yields the error equation
PV
------- ( EW jn + 1 – EW jn ) = VT [ f w ( SW jn– 1 + EW jn– 1 ) – f w ( SW jn– 1 ) – f w ( SW jn + EW jn ) + f w ( SW jn ) ] (14-18)
Δt
∂f w EW 2- ∂---------
2f
w
f w ( SW + EW ) = f w ( SW ) + EW --------
- ( SW ) + ---------- - ( SW ) + …
∂S w 2! ∂S w 2
For the purpose of stability analysis, errors are assumed sufficiently small that
higher powers of EW can be ignored. Then Equation 14-18 can be written
PV
------- ( EW jn + 1 – EW jn ) = VT ( f' w, j – 1 EW jn– 1 – f' w, j EW jn ) (14-19)
Δt
where
∂f w
f' w, j = --------- ( SW jn )
∂S w
At this point, we need another assumption in order to proceed, that f' w is “locally”
constant, i.e., that it doesn’t change much from block to block. Then
EW jn + 1 – EW jn = λ ( EW jn– 1 – EW jn ) (14-20)
where
VT ⋅ f' w Δt
λ = ----------------------- (14-21)
PV
The modulus of the amplification factors of all the components must be less than
or equal to one for stability.
Substitution of Equation 14-22 into Equation 14-20 gives, for each component,
A pn + 1
------------- – 1 = λ [ exp ( – iα p Δx) – 1 ]
A pn
or
A pn + 1
------------- = 1 + λ [ cos ( α p Δx ) – i sin ( α p Δx ) – 1 ]
A pn
A pn + 1 2
------------- = { 1 + λ [ cos ( α p Δx ) – 1 ] } 2 + λ 2 sin 2 ( α p Δx )
A pn
or
A pn + 1 2
------------- = 1 + 2λ [ 1 – cos ( α p Δx ) ] ( λ – 1 )
A pn
VT ⋅ f' w Δt
-≤1
------------------------ (14-24)
PV
PV
------- [ SW jn + 1 – SW jn ] = VT [ f w ( SW jn– 1 , SG jn– 1 ) – f w ( SW jn, SG jn ) ] (14-25)
Δt
PV
------- [ SG jn + 1 – SG jn ] = VT [ f g ( SW jn– 1 , SG jn– 1 ) – f g ( SW jn, SG jn ) ] (14-26)
Δt
PV
------- [ ( SW jn + 1 + EW jn + 1 ) – ( SW jn + EW jn ) ] = VT
Δt
[ fw ( SWjn– 1 + EWjn– 1 , SGjn– 1 + EGjn– 1 )
(14-27)
– f w ( SW jn + EW jn, SG jn + EG jn ) ]
PV
------- [ ( SG jn + 1 + EG jn + 1 ) – ( SG jn + EG jn ) ] = VT
Δt
[ fg ( SWjn– 1 + EWjn– 1 , SGjn– 1 + EGjn– 1 )
(14-28)
– f g ( SW jn + EW jn, SG jn + EG jn ) ]
Note that a truncated Taylor series for a function of two variables yields
where
∂f w ∂f w
f' ww = --------
- f' wg = --------
∂S w ∂S g
∂f ∂f
f' gw = --------g- f' ww = --------g
∂S w ∂S g
Thus, if we subtract Equation 14-25 from Equation 14-27, and Equation 14-26
from Equation 14-28, and substitute Equations 14-29 and 14-30, we get the
simultaneous error equations
PV
------- ( EW jn + 1 – EWjn ) = VT ( f' ww, j – 1 EW jn– 1 + f' wg, j – 1 EG jn– 1 – f' ww, j EWjn – f' wg EG jn ) (14-31)
Δt
PV
------- ( EG jn + 1 – EG jn ) = VT ( f' gw, j – 1 EW jn– 1 + f' gg, j – 1 EG jn– 1 – f' gw, j EW jn – f' gg EG jn ) (14-32)
Δt
Again, we assume that the f' are locally constant so that they don’t change much
from block to block. Also, we expand the errors in the Fourier series
NX
EW jn = ∑ Apn exp (iαp xj ) (14-33)
p=1
NX
EG jn = ∑ Bpn exp (iαp xj ) (14-34)
p=1
Substitution of Equations 14-33 and 14-34 into Equations 14-31 and 14-32 gives,
for each component, after division by exp (iα p x j ) ,
PV
------- ( A pn + 1 – A pn ) = VT ( f' ww A pn + f' wg B pn ) [ exp ( – iα p Δx) – 1 ]
Δt
PV
------- ( B pn + 1 – B pn ) = VT ( f' gw A pn + f' gg B pn ) [ exp ( – iα p Δx) – 1 ]
Δt
where
For stability, we need to show that v pn + 1 is, in some sense, not larger than v pn . We
choose as the measure of a vector the 2-norm, also called the Eucledian norm,
defined by
v pn 2 = (A pn ) 2 + (B pn ) 2
Since the norm of a product is less than or equal to the product of the norms, then
from Equation 14-35,
The 2-norm of the matrix F' is its maximum absolute eigenvalue, which is
defined below.
v pn + 1 2
A p = -------------------
v pn 2
and try to show that it is less than one, in magnitude, for all components p.
or
or
VT ⋅ Δt-
---------------- F' 2 ≤1 (14-37)
PV
This is the Buckley-Leverett throughput condition for three-phase flow.
Eigenvalues of F'
As stated above, the 2-norm of a matrix is the maximum absolute value of its
eigenvalues. The eigenvalues of F' are the values of λ that satisfy
f' ww – λ f' wg
det = 0 (14-38)
f' gw f' gg – λ
The eigenvalues of F' are the roots of Equation 14-39, and the maximum absolute
value of those roots is the 2-norm of that matrix.
The values of the f' are obtained from the derivatives of mobility (see Equation
14-4), which are substituted into the following.
∂M w ∂M t
M t ----------- – M w ---------
∂S w ∂S w
f' ww = ------------------------------------------
-
M t2
∂M w ∂M t
M t ----------- – M w ---------
∂S g ∂S g
f' wg = ------------------------------------------
-
M t2
∂M g ∂M t
M t ---------- – M g ---------
∂S w ∂S w
f' gw = ----------------------------------------
-
M t2
∂M g ∂M t
M t ---------- – M g ---------
∂S g ∂S g
f' gg = ----------------------------------------
-
M t2
where
Mt = Mw + Mo + Mg
and
∂M ∂M ∂M ∂M
-----------t = ----------w- + ----------o- + ----------g- , ph = w or g
∂S ph ∂S ph ∂S ph ∂S ph
If we ignore gravity and capillary pressure, Equation 8-1 becomes, for one-
dimensional flow,
∂ ( φHz c ) ∂ kk ro ρ o x c kk rg ρ g y c⎞ ∂P
⎛ --------------------
-------------------- = -⎠ ------ + q c
∂t ∂x ⎝ μ - + --------------------
μ ∂x
(14-40)
o g
where H is the total hydrocarbon moles per unit pore volume, xc is mole fraction
of component c in the liquid (oil) phase, yc is mole fraction in the vapor (gas)
phase, and zc is overall mole fraction. (Note that we use H instead of the F in
Equation 8-1 to avoid confusion with fractional mobilities used below.
Equilibrium Constraint
y
K c = ----c (14-41)
xc
NC
∑ ( yc – xc ) = 0 (14-42)
c=1
Saturation Constraint
H ( 1 – L ) + -------
HL- = 1
--------------------- (14-43)
ρg ρo
where L is the volumetric fraction of the hydrocarbon which is liquid (oil). Note
that
(1 – L)
Sg = H
--------------------- , S o = HL
-------- (14-44)
ρg ρo
∂ ( φH ) ∂ ⎛ kk ro ρ o kk rg ρ g⎞ ∂P
--------------- = - ------ + ∑ q c
⎝ μ - + ---------------
--------------- (14-45)
∂t ∂x o μ ⎠ ∂xg c
Let
kk ro kk rg
λ o = ---------- λ g = ---------- λt = λo + λg
μo μg
λ λ
f o = ----o- f g = ----g-
λt λt
∂P ∂P ∂P
v o = – λ o ------ v g = – λ g ------ v t = – λ t ------
∂x ∂x ∂x
( φH ) = – ∂ [ ( ρ f + ρ f )v ] + q
∂---------------
∂t ∂x o o g g t ∑c c (14-46)
PV j
--------- ( H jn + 1 – H jn ) = ρ o ( VT j – 1 ⁄ 2 FO jn– 1 – VT j + 1 ⁄ 2 FO jn )
Δt (14-47)
+ ρ g ( VT j – 1 ⁄ 2 FG jn– 1 – VT j + 1 ⁄ 2 FG jn )
The difference analogues of the constraint equations 14-42 and 14-43 are
NC
∑ ( Ycn, j – Xcn, j ) = 0 (14-48)
c=1
H jn ( 1 – L jn ) H jn L jn
--------------------------- + ------------- = 1 (14-49)
ρg ρo
Equations 14-47, 14-48, and 14-49 must also be satisfied by perturbed values of
the basic dependent variables, P, H, L, and z1, z2, ..., zNC-1, as well as by perturbed
values of any derived dependent variables, such as So = HL/ρo. Let EH be the
perturbation in H, EFO be the perturbation in FO, etc. Then the perturbation of
difference equation 14-47 is
PVj
--------- ( H jn + 1 + EH jn + 1 – H jn – EH jn ) = ρ o ( VT j – 1 ⁄ 2 + EVT j – 1 ⁄ 2 ) ( FO jn– 1 + EFO jn– 1 )
Δt
– ρ o ( VT j + 1 ⁄ 2 + EVT j + 1 ⁄ 2 ) ( FO jn + EFO jn ) + ρ g ( VT j – 1 ⁄ 2 + EVT j – 1 ⁄ 2 ) ( FG jn– 1 + EFG jn– 1 )
(14-50)
– ρ g ( VT j + 1 ⁄ 2 + EVT j + 1 ⁄ 2 ) ( FG jn + EFG jn )
PV
---------j ( EH jn + 1 – EH jn ) = ρ o ( VT j – 1 ⁄ 2 EFO jn– 1 – VT j + 1 ⁄ 2 EFO jn ) + ρ g ( VT j – 1 ⁄ 2 EFG jn– 1 – VT j + 1 ⁄ 2 EFG jn ) (14-51)
Δt
But, by neglecting higher powers of the error terms, we have the following
truncated Taylor series:
where
∂f o ∂f g
f' o, j = -------- ( SG jn ) = – f' g, j = – -------- ( SG jn )
∂S g ∂S g
Similarly,
PV j
--------- ( EH jn + 1 – EH jn ) = ( ρ g – ρ o ) ( VT j – 1 ⁄ 2 f' g, j – 1 ESG jn– 1 – VT j + 1 ⁄ 2 f' g, j ESG jn ) (14-52)
Δt
Consider now the saturation constraint equation 14-49. Perturb it, to yield
( H jn + EH jn ) ( 1 – L jn – EL jn ) ( H jn + EH jn ) ( L jn + EL jn )
----------------------------------------------------------------- + --------------------------------------------------------- = 1 (14-53)
ρg ρo
Subtract Equation 14-49 from Equation 14-53, and ignore higher powers of the
errors, to yield the saturation constraint error equation:
EH jn ( 1 – L jn ) – H jn EL jn H jn EL jn + L jn EH jn
------------------------------------------------------- + ----------------------------------------- = 0
ρg ρo
or
ρ o ( 1 – L jn ) + ρ g L jn
EL jn - EH jn
= ------------------------------------------- (14-54)
( ρ o – ρ g )H jn
H jn ( 1 – L jn )
SG jn = --------------------------
- (14-55)
ρg
( H jn + EH jn ) ( 1 – L jn – EL jn )
SG jn + ESG jn = ----------------------------------------------------------------- (14-56)
ρg
Subtracting Equation 14-55 from Equation 14-56, ignoring higher powers of the
error terms, yields
( 1 – L jn )EH jn – H jn EL jn
ESG jn = ------------------------------------------------------
-
ρg
ρ o ( 1 – L jn ) + ρ g L jn EH jn
ESG jn = ( 1 – L jn ) – -------------------------------------------
- -----------
( ρo – ρg ) ρg
or
EH jn = ( ρ g – ρ o )ESG jn (14-57)
PV j
--------- ( ESG jn + 1 – ESG jn ) = VT j – 1 ⁄ 2 f' g, j – 1 ESG jn– 1 – VT j + 1 ⁄ 2 f' g, j ESG jn (14-58)
Δt
PV j
--------- ( ESG jn + 1 – ESG jn ) = VT ⋅ f' g ( ESG jn– 1 – ESG jn ) (14-59)
Δt
NX
ESG jn = ∑ Apn exp (iαp xj ) (14-60)
p=1
A pn + 1 2 2 VT ⋅ f' g Δt VT ⋅ f' g Δt ⎞
------------- = 1 + ---------------------------- [ 1 – cos ( α p Δx ) ] ⎛⎝ ----------------------
- – 1⎠ (14-61)
A pn PV PV
VT ⋅ f' g Δt
----------------------- ≤ 1 (14-62)
PV
For simplicity, the water equation was omitted from the preceding analysis. It
seems reasonable that if it had been included, then the three-phase Buckley-
Leverett throughput condition of Equation 14-37 would be the appropriate
condition for stability.
NX
EH jn = ( ρ g – ρ o ) ∑ Apn exp (iαp xj )
p=1
If we assume that
NX
EH jn = ∑ Bpn exp (iβp xj )
p=1
B pn = ( ρ g – ρ o )A pn
and
βp = αp
B n+1 n+1
------------- = A
------------- ≤ 1 (14-63)
Bn An
ρ o ( 1 – L jn ) + ρ g L jn NX
EL jn = -------------------------------------------
H jn
- ∑ Apn exp (iαp xj )
p=1
Again, we assume a “locally constant” coefficient; that is, we assume that the term
multiplying the summation does not change much from block to block or from
one time level to the next time level. If we assume that
NX
EL jn = ∑ Cpn exp (iγp xj ) (14-64)
p=1
ρo ( 1 – L ) + ρg L
C pn = --------------------------------------- A pn (14-65)
H
and
γp = αp (14-66)
C n+1 n+1
------------- = A
------------- ≤ 1 (14-67)
Cn An
The situation for the error growth of the Zc is considerably more complicated.
Consider the equilibrium constraint equation 14-48. Its perturbed form is
NC
∑ ( Ycn, j + EYcn, j – Xcn, j – EXcn, j ) = 0 (14-68)
c=1
NC
∑ ( EYcn, j – EXcn, j ) = 0
c=1
Y cn, j = K c X cn, j
EY cn, j = K c EX cn, j
so
NC
∑ ( Kc – 1 )EXcn, j = 0 (14-69)
c=1
Since
z c = Lx c + ( 1 – L ) y c = Lx c + ( 1 – L )K c x c
Then
Z cn, j
X cn, j = --------------------------------------
- (14-70)
L jn + ( 1 – L jn )K c
Z cn, j + EZ cn, j
X cn, j + EX cn, j = ----------------------------------------------------------------------
- (14-71)
L jn + EL jn + ( 1 – L jn – EL jn )K c
NC Kc – 1 NC – ( K – 1 ) 2 Z n
c, j
∑ [ L n + ( 1 – L n )K ] c, j j ∑ ---------------------------------------------
n = EL n c
------------------------------------------- EZ - (14-73)
c=1 j j c c=1
[ L n
j + ( 1 – L n
j )K c ] 2
Since the Zc sum to one, the EZc sum to zero, so they are not all independent. Thus
Equation 14-73 should be written
NC – 1 Kc – 1 K NC – 1 NC – ( K – 1 ) 2 Z n
c, j
∑ - EZ cn, j = EL jn ∑ ---------------------------------------------
c
--------------------------------------
- – ------------------------------------------ - (14-74)
c=1
L jn + ( 1 – L jn )K c L jn + ( 1 – L jn )K NC c=1
[ L n + ( 1 – L n )K ] 2
j j c
If we again consider the coefficients of the errors constant for the purpose of the
stability analysis, then we can write Equation 14-74 in the form
NC – 1
∑ G c EZ cn, j = EL jn (14-75)
c=1
where
Kc – 1 K NC – 1
--------------------------------- – ------------------------------------ -
L + ( 1 – L )K c L + ( 1 – L )K NC
G c = -----------------------------------------------------------------------------
NC –( Kc – 1 ) 2 Z
∑ [ L + ( 1 – L )K ] 2
----------------------------------------
c=1 c
If we assume
NX
EZ cn, j = ∑ Dcn, p exp (iδc, p xj ) (14-76)
p=1
δ c, p = γ p = α p
and
NC – 1
∑ G c D cn, p = C pn (14-77)
c=1
NC – 1
∑ G c D cn, +p 1 = C pn + 1 (14-78)
c=1
NC – 1
∑ G c D cn, +p 2 = C pn + 2 (14-79)
c=1
on up to the n + NC - 1 timestep:
NC – 1
∑ G c D cn, +p NC – 1 = C pn + NC – 1 (14-80)
c=1
Let
C pn + 1 C pn + 2
χ p = ------------- = ------------- = … (14-81)
C pn C pn + 1
D cn, +p 1 D cn, +p 2
η c, p = ------------- - = …
- = ------------- (14-82)
D cn, p D cn, +p 1
and
Γ c, p = G c D cn, p (14-83)
NC – 1
∑ Γ c, p = C pn (14-84)
c=1
NC – 1
∑ Γ c, p η cr, p = C pn χ pr , r = 1, 2, …, NC – 1 (14-85)
c=1
Note that the superscript n refers to timestep level, while the superscript r is an
exponent. Equations 14-84 and 14-85 can be rearranged to
NC – 1
∑ Γ c, p ( η cr, p – χ pr ) = 0 , r = 1, 2, …, NC – 1 (14-86)
c=1
η c, p = χ p
It can be shown63 (albeit, with some difficulty) that this is a multiple root, and that
Equation 14-86 is equivalent to
NC – 1
( η c, p – χ p ) = 0
D cn, +p 1
-------------- = η c, p ≤ 1 (14-87)
D cn, p
Thus, we have shown that the errors in Zc do not grow if Equation 14-62 is
satisfied. Thus, if Equation 14-62 is satisfied, the system of difference equations
14-47, 14-48, and 14-49 is stable.
As pointed out repeatedly, the analysis presented here involves many assumptions
that are not strictly satisfied. It can be shown64 that if capillary pressure is taken
into account, the stable timestep is reduced somewhat; some 1D experiments
show a reduction of only a few percent. Accordingly, it is recommended that the
user monitor the solution carefully when the IMPES option is chosen.
To provide a margin of safety, the parameter STSLIM on the IMPSTAB card can
be set to something less than 1.0. That parameter is the largest fraction of the
maximum stable timestep (as computed by Equation 14-37) at which the
simulator is permitted to run.
15
Local Grid Refinement1
15.1 Introduction
The Local Grid Refinement (LGR) option is a special feature which enables the
user to improve the resolution and detail in a particular area of a reservoir study,
without increasing the resolution everywhere. The LGR option minimizes the
number of gridblocks, and therefore the CPU time required by allowing the grid to
be selectively refined in areas where more grid definition is required. It has
applications in the modeling of:
■ Horizontal wells
■ Delineation of faults
This chapter explains some of the general concepts with regard to LGR, along
with technical details on different LGR techniques and examples of how to set up
LGR models.
■ Grid refinements (and base grid) can use either the IMPES or fully implicit
formulation.
■ Central to the LGR option is a new linear solver, CBLITZ68, that enables the
efficient linear solution of the composite grid system. This implementation
allows composite grids, whether IMPES, implicit or mixed IMPES/implicit
grids, to be solved in one, two, three or four (dual permeability) dimensions.
■ Black Oil
■ Compositional EOS
■ Dual Porosity/Permeability
■ Polymer
■ Crossflow
■ Hydrocarbon/Water Tracking
■ Injection Regions
Except as noted below, datasets that were valid with previous versions of VIP-
EXECUTIVE should still be accepted as is. The following features, available in
previous versions, are no longer available either with LGR or single grid systems:
The implicit well option, previously available with IMPES, was removed because
it can be duplicated with the LGR option by declaring the well as being contained
in a line of one-for-one refined gridblocks, designated as an implicit grid (see
example in Section 15.10.2).
■ Grid name
Radial refinements are normally oriented in the Z direction for vertical or near-
vertical wells, but may also be oriented in the X or Y direction for horizontal
wells. The Z-axis for the radial refinement is always oriented along the wellbore.
When creating a radial refinement, the following properties must be defined:
■ Grid name
■ Portion of coarse grid to be refined (if more than one coarse gridblock along
the well, the same radial refinement is implemented through each gridblock).
■ Number of refined gridblocks in the radial (r) and angular (theta) directions.
Figure 15-3 shows the gridblock numbering convention for a radial refinement in
the z-direction.
z(k)
x(i)
y(j)
3
2
1
12 11 10 7 4 5 6
8
9
Figure 15-4 shows the gridblock numbering convention for a radial refinement in
the x-direction. Note that the z-direction of the radial grid is along the x-direction
of its parent grid.
x(i)
y(j)
z(k)
2
1
8 7 3 4
5
6
Figure 15-5 shows the gridblock numbering convention for a radial refinement in
the y-direction. Note that the z-direction of the radial grid is along the y-direction
of its parent grid.
y(j)
x(i)
z(k)
2
1
8 7 3 4
5
6
S
•
C •
A1 A2
( A 2 – A 1 ) ⋅ CS 1
T = ----------------------------------
2
- ---------------------------
-
CS ⎛ A 2 ⋅ CS ⎞
ln ⎜ ------------------⎟
⎝ A 1 ⋅ CS⎠
The gridblock transmissibilities in the θ(j) direction are calculated from a
harmonic average of the elemental skeleton block values. The gridblock
transmissibilities in the r(i) and z(k) directions are calculated by summing the
elemental skeleton block values.
1 2 3 4 5 6 7 8
32 9
31 10
30 11
29 12
28 13
27 14
26 15
25 16
24 23 22 21 20 19 18 17
skeleton blocks
gridblocks
1
T θ- = -----------------------------------------
-
1 - + -------
⎛ ------ 1 -⎞
∑ ⎝ tjθ- tjθ+⎠
j=1 4
1
T θ+ = -----------------------------------------
-
1 - + -------
⎛ ------ 1 -⎞
∑ ⎝ tjθ- tjθ+⎠
j=5 8
Tr ± = ∑ t jr ±
j = 1, 8
T rz ± = ∑ t jz ±
j 1 8
t = skeleton block transmissiblity
T = gridblock transmissiblity
■ New Models. When building a new simulation model from scratch, in most
cases it will be easier to use GRIDGENR to specify the grid structure and all
local grid refinements. The necessary keyword parameters and data will be
produced in files ready to be included in the VIP-CORE data set.
15.5.1 GRIDGENR
The GRIDGENR program is a graphical interface used to design the grid structure
for a reservoir simulation model and calculate the initial properties to be assigned
to each gridblock, such as depth, thickness, porosity, etc. GRIDGENR provides
three basic alternatives for defining the grid structure. These are (1) interactively
draw the reservoir grid, (2) digitize the grid from various sources, or (3) import
data that can be used to specify the grid structure. Following any of these three
options, the drawing features can be used to define local grid refinements, with
the resulting shape and size of the refinements shown on the surface of the grid
(see illustration below).
The Array Calculate and Output steps in GRIDGENR produce the appropriate
keyword parameters for input to the VIP-CORE initialization, as explained below.
Using GRIDGENR alone does not provide access to all the refinement features
available with the keyword parameters. However, it does provide access to a large
majority of them. For more information about the grid refinement features in
GRIDGENR, see the GRIDGENR User’s Guide.
The keyword parameters below are used to describe this structure for a VIP-
CORE initialization (i.e., for calculating the initial model state).
NX NY NZ NCOMP
5 5 1 2
LGR BASEGRID
CARTREF REF1
2 4 2 4 1 1
2 3 2
2 2 2
1
OMIT 4 4 4 4 1 1
RADZREF RAD
3 3 1 1
3 4 .25
1
2*0.5
2*0.5
ENDREF
ENDREF
ENDLGR
For more information about LGR keyword parameters, refer to the VIP-CORE
Reference Manual.
If VIP-CORE keywords are used to apply LGR, the user has the option to specify
all of the properties for the coarse and refined gridblocks or only to specify
properties for the coarse gridblocks. Where properties are specified only for the
coarse gridblocks, the simulator will calculate unique depths for each refined
gridblock, based on variations in reservoir dip and model layer thickness, but it
will not calculate unique initial rock properties for each refined gridblock. Instead,
each refined gridblock will have the same properties as its parent coarse
gridblock. This is not always a disadvantage, as explained in the following
discussion.
For a quick investigation of detailed flow behavior around a well, it may not be
necessary to use GRIDGENR to define the slight variation in rock properties for
the radial refinement around the well. In this case, only the keywords to define the
radial refinement need to be added to the VIP-CORE input file, and the rock
properties of the parent gridblocks will be propagated down to the refined
gridblocks.
Conversely, for a detailed study of infill drilling in a particular area, it would most
likely be better to use GRIDGENR to define the LGR area, in order to get a better
definition of the variation of rock properties within the entire refined area.
For the situation where the initialization data has already been prepared and it is
not feasible or desirable to go back to GRIDGENR (i.e., a reservoir model that has
already been history matched), the grid refinements can be specified manually
through the keyword parameters. The properties from the coarse grid would be
propagated down to any refinements, and a few additional history match runs may
be required to fine tune the match based on a better definition of fluid fingering
and/or coning. See Section 15.8.2 for the rules and explanations of the
propagation process.
Once the grid definition and properties are defined appropriately in the VIP-
CORE data deck, the initialization process is not significantly different from
normal.
Level 4
Grid
Level 2
Grid Level 1
Grid
Level 2
Grid
Level 4
Grid
Level 3
Grid
A coarse grid (or level 1 or root grid) is imposed on the region of interest. This is
further refined in areas where greater resolution is required (a refinement of a
level N grid is called a level N+1 grid). The number of levels of refinement is
arbitrary. Cartesian refinements may be further refined but radial refinements may
not.
■ Grid lines for refined cells terminate only on coarse cell boundaries.
Valid: Invalid:
Refinements must fill at least an entire coarse cell and the refined grid lines must
be continuous across the entire coarse cell.
Compatible Refinement:
Incompatible Refinement:
Refinements that span more than one coarse cell must have refined grid lines that
conform across the coarse cells.
Legal Refinement
Illegal Refinement
■ All grids of the same level must be independent, i.e., not touching.
Legal Refinement
Illegal Refinements
Independent refinements of the same level must be isolated from each other.
Only the pore volumes and transmissibilities at initial conditions are preserved
and reinstated as necessary, as grid refinements are deactivated and reactivated.
When a grid refinement that has been active for one or more timesteps is
deactivated, new values for the unknowns (pressure, composition, saturations,
etc.) are calculated for each of the gridblocks that had been refined in the parent
grid, such that precise material balance is maintained. For each parent gridblock,
the total moles of each component and the total water volume is computed from
the sum of its refined gridblocks, respectively. An iterative procedure is then
employed to compute the new values for the unknowns which will exactly
preserve the material balance of each of the components and water.
When a grid is activated after having been inactive for one or more timesteps, the
unknowns of the parent gridblocks are propagated directly to the refined
gridblocks, again preserving material balance, but with the possible instability
resulting from each of the fine gridblocks having the same pressures, saturations,
compositions, etc. as their parent gridblock. Considerable care should be taken to
ensure that the area around the refinement is approaching steady-state conditions
before deactivating or activating the refinement, in order to reduce or eliminate
the need for any timestep cuts. The activation/deactivation procedure is performed
after all input for the time interval has been read.
CBLITZ has a robust treatment of 3D radial grids, always allowing for complete
circular flow. CBLITZ allows grids to be active or inactive, and can handle
wellbore crossflow and faults within a grid or between grids.
The LGR keyword generally indicates that the grid is to be refined. This line is
followed by the data defining the refinements. Once all the refinements have been
defined, a line with the keyword ENDLGR ends the grid refinement data mode.
The entering of the LGR keyword has several other effects, even for a single-grid
model in which the LGR line is immediately followed by the ENDLGR line:
1. It causes the ROOT grid (or single grid) that is defined with DX, DY, DZ or
DR, DTHETA, DZ arrays to be converted to corner-point grids internally for
the calculation of pore volumes and transmissibilities.
2. It causes the default map type to be the new map file (Fortran Unit 27).
RMIN
The RMIN keyword allows the user to specify, for radial refinements, the
minimum outer radius allowed for the innermost ring of blocks. This is applied
globally to all radial refinements, but can be set on a grid-by-grid basis within the
LGR data.
Array data is introduced grid-by-grid. All array data must be grouped together by
grid and grid order is important. The following are rules and explanations of the
cascading process:
2. Array data should be input for a parent grid before inputing the data for a
child.
3. When array data is read in, it is immediately propagated to its lower level
grids.
5. Permeability arrays at gridblock faces are not allowed with the LGR option.
6. Any array data for a child grid can be input separately to replace any cascaded
values.
7. Except for the endpoint scaling arrays, an array must be input for the ROOT
grid before it can be input for a lower level grid.
8. At any level, including ROOT, any endpoint data not entered is set to that of
the appropriate rock type.
Array data of children grids (that has been assigned through propagation) can be
modified using the MOD or VMOD options. The NONE option for a child array
indicates that no data will be input and any MOD or VMOD options will apply to
the inherited data. The modified data is then propagated to the grid’s children. 00
Example: 00
ARRAYS
POR CON
0.3
.
.
.
ARRAYS CHILD1
POR NONE
VMOD
2 4 2 3 1 1 EQ
0.25 0.26 2*0.27 0.29 0.3
The CORP array consists of eight (x, y, z) coordinate values for each gridblock.
Formulation Options
In a multi-grid run all grids may use the IMPES formulation, all grids may use the
implicit formulation, or some grids may use IMPES and the remaining grids may
use implicit. By default all grids are assigned the formulation specified in, or
implied by, the Utility Data. The IMPGRID ... ENDIMPGRID data is used to set
the formulation for individual grids. If entered, this data must immediately follow
the RESTART card and precede any TITLE cards.
Timestep Controls
For multi-grid systems which may involve mixed levels of implicitness, it may be
necessary to have separate sets of maximum change parameters for both the
iteration and the timestep. To accomplish this, two new optional keywords have
been implemented: DTMPL and ITNMPL. When a DT keyword is read, the
maximum change parameters for both the IMPES grids and the IMPLICIT grids
are set to the values specified on the DT card. Also when an ITNLIM keyword is
read, the maximum change parameters for both the IMPES grids and the
IMPLICIT grids are set to the values specified on the ITNLIM card. If different
values are desired for the IMPLICIT grids, the DTMPL and ITNMPL keywords
must be used to specify the maximum change parameters for the IMPLICIT grids
only.
The new matrix solver CBLITZ is an iterative solver, using the Preconditioned
Generalized Minimum Residual (GMRES) method. It must be used for all multi-
grid systems and for three-dimensional radial single-grid systems using the
FLOW360 option. CBLITZ, as well as BLITZ, EXCEL, or GAUSS, may also be
used for other single-grid systems.
The block of data CBLGRID ... ENDCBLGRID may be used to specify some of
the CBLITZ parameters on a grid-by-grid basis. If entered, this data must
immediately follow the CBLITZ data.
Well Data
For multi-grid systems, it may be necessary to specify in which grid(s) that the
well resides, in addition to the standard IW and JW well location. If no additional
data is entered, the well is assumed to be in the ROOT grid, and the IW and JW
are relative to that grid. However, if the well perforations are in one or more of the
grid refinements, then the grid name must be specified on either the WELL card
or the FPERF card. If the well perforations are all within the same grid, then the
grid may be specified on the WELL card. If the well perforations are in two or
more grids, then the grid must be specified for each of the perforations, along with
the proper IW and JW relative to the grid.
If the well is in a radial grid and (IW, JW) have not been specified in the FPERF
data or in any previous WELL name and location data, then perforations will be
automatically generated in IW=1, for JW=1,2,...,NTHETA for the radial grid.
Other Data
For more detailed information regarding LGR-related data, see the VIP-CORE
Reference Manual and the VIP-EXECUTIVE Reference Manual.
A reservoir study was conducted without grid refinement (Figure 15-13), with two
levels of refinement with a Cartesian 3x3 grid around the wells (Figure 15-14) and
with two levels of refinements and 4x4 radial (Figure 15-15) refinements around
each well. The grid without local refinements (Case 1) was carefully constructed
so that the level of resolution around the wells is the same as in Case 2 with
Cartesian refinements around the well.
The cases and results are summarized in Table 15-1. Case 1 was run with both
IMPES and implicit. Both Case 2 and Case 3 were run with IMPES in the coarse
and level 2 refined grids, and implicit around the wells. The results show that Case
2 ran more than 4 times faster than implicit Case 1. Case 3 was almost three times
faster than implicit Case 1 even though the resolution around the wells in Case 3
was much better.
The results of production plots show almost identical results in the three cases for
most of the wells, but increased gas and water coning for some of the wells in case
3 as shown in Figures 15-16 and 15-17.
15.10 Examples
The following examples, in increasing complexity, illustrate the additional data
needed to set up models using local grid refinement. The basic assumption is that
the user is creating the data; GRIDGENR/ARRAY is not needed to create the
locally refined grid(s). The figures accompanying the examples depict only a
portion of the grid; e.g. the area around well G3.
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ! No LGRs defined
ENDLGR ! But uses new Transmissibility calculations
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
Simulator:
IMPLICIT
RESTART
BLITZ
C
C DT1 DTmin DTmax DP DS DV DZ
DT -1. 1. 100. 500. .15 .9 .9
C
ITNLIM 1 10 .01 .10 .14
WELL N NAME IW JW
1 G3 18 3
2 H2 20 13
3 A1 14 16
FPERF
WELL L
1 1
X 2
2 2
X 3
3 7
X 8
X 9
.
.
.
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ROOT
CARTREF G3 ! Separate Grid for G3
18 18 3 3 1 2
1
1
2*1
ENDREF
CARTREF H2 ! Separate Grid for H2
20 20 13 13 2 3
1
1
2*1
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
Simulator:
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ROOT
CARTREF G3 ! Cartesian refinement around Well G3
17 19 2 4 1 4
3*3
3*3
4*3
ENDREF
CARTREF
. . . etc
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
Simulator:
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ROOT
RADZREF G3 ! Radial refinement around Well G3
18 3 1 4 ! IW JW K1 K2
5 4 .6 RMIN 20. ! Nrad NTheta Rinner R1
4*3 ! Z splits
5*.5 ! X locations of center (relative to DX)
5*.5 ! Y locations of center (relative to DY)
ENDREF
CARTREF
. . . etc
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
Simulator:
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ROOT
RADZREF G3 ! Radial refinement around Well G3
18 3 4 9 ! IW JW K1 K2
5 4 .6 RMIN 20. ! Nrad NTheta Rinner R1
6*3 ! Z splits
7*.5 ! X locations of center (relative to DX)
7*.5 ! Y locations of center (relative to DY)
ENDREF
CARTREF
. . . etc
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
Simulator:
IMPLICIT
RESTART
IMPGRID ! Formulation by Grid
ROOT IMPES
ENDIMPGRID
START
CBLITZ
C
C DT1 DTmin DTmax DP DS DV DZ
DT -1. 1. 100. 200. .05 .9 .9 ! for IMPES Grids
DTMPL 500. .15 .9 .9 ! for IMPLICIT Grids
C
ITNLIM 1 10 .03 .10 .04 ! for IMPES Grids
ITNMPL .10 .10 .10 ! For IMPLICIT Grids
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ROOT
RADXREF G3 ! Refinement for Horizontal Well G3
17 19 4 6 ! i1 i2 JW KW
5 4 .6 RMIN 10. ! Nrad NTheta Rinner R1
2 3 4 ! X splits
4*.5 ! Y locations of center (relative to DY)
4*.5 ! Z locations of center (relative to DZ)
ENDREF
CARTREF
. . . etc
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
Simulator:
IMPLICIT
RESTART
IMPGRID ! Formulation by Grid
ROOT IMPES
ENDIMPGRID
START
CBLITZ
C
C DT1 DTmin DTmax DP DS DV DZ
DT -1. 1. 100. 200. .05 .9 .9 ! for IMPES Grids
DTMPL 500. .15 .9 .9 ! for IMPLICIT Grids
C
ITNLIM 1 10 .03 .10 .04 ! for IMPES Grids
ITNMPL .10 .10 .10 ! For IMPLICIT Grids
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
27 28 9 2
C
LGR ROOT
CARTREF G3 ! Cartesian refinement around Well G3
17 19 2 4 2 4
2 3 2 ! I splits
2 3 2 ! J splits
1 2 3 ! K splits
ENDREF
CARTREF
... etc.
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
ARRAYS
.
.
.
ARRAYS G3
DX XVAR
.5 .5 .4 .2 .4 .5 .5 ! I = 17, 18, 19
DY YVAR
.5 .5 .4 .2 .4 .5 .5 ! J = 2, 3, 4
DZ ZVAR
1. .5 .5 .5 .3 .2 ! K = 2, 3, 4
.
.
.
VIP-CORE:
.
.
.
C GRID DIMENSIONS
NX NY NZ NCOMP
5 1 1 2
C
LGR ROOT ! The Whole World
CARTREF Area1
1 1 1 1 1 1
25 ! I splits
25 ! J splits
9 ! K splits
RADZREF G3 ! Radial refinement round Well G3 in Area1
18 3 4 9 ! IW JW K1 K2 in Area1 indices
5 4 .6 RMIN 20. ! Nrad NTheta Rinner R1
6*3 ! Z splits
7*.5 ! X locations of center (relative to DX)
7*.5 ! Y locations of center (relative to DY)
ENDREF
ENDREF
CARTREF Area2
5 5 1 1 1 1
35 ! I splits
25 ! J splits
12 ! K splits
ENDREF
ENDLGR
C
DWB BWI VW CW CR TRES TS PS
1.00 1.005 0.9 2.8E-6 30.0E-6 127. 60 15
C
.
.
.
16
Miscible Options
16.1 Introduction
The VIP-EXECUTIVE miscible option can be used in a three- or four-component
mode. The three-component mode, which consists of reservoir oil, injected
solvent, and water, is an extended version of the Todd-Longstaff15 model. The
four-component mode consists of a heavy-oil component, a light-oil component
(solution gas), injected solvent, and water. The option has been a widely adopted
alternative to fully compositional models for the simulation of miscible or near
miscible processes. It introduces a mixing parameter to account for viscous
fingering and incomplete mixing in numerical gridblocks.
The miscible option accounts for the transformation between miscible and
immiscible conditions. The development and loss of miscibility are controlled by
a miscibility pressure that can be treated as a constant or as composition-
dependent. A damping function also is introduced for a smooth transition between
miscible and immiscible conditions.
∂ ( φFz i ) kk roe ρ o ρ oe g
------------------- = ∇ ⋅ ⎛ ------------ ------- x i⎞ ∇ ⎛ P o – ------------------------------------- D⎞ (16-1)
∂t ⎝ μ oe M o ⎠ ⎝ 5.6146 × 144g c ⎠
kk rge ρ g ρ ge g
+ ∇ ⋅ ⎛ ------------ ------- y i⎞ ∇ ⎛ P o + P cgoe – ------------------------------------- D⎞ + q o x i + q q y i
⎝ μ ge M g ⎠ ⎝ 5.6146 × 144g c ⎠
yi = Ki ( P ) ⋅ xi (16-2)
S o – S orm – S twb
k roe = αk rh -------------------------------------
- + ( 1 – α )k ro (16-3)
1 – S w – S twb
Sg
- + ( 1 – α )k rg
k rge = αk rh ------------------------------ (16-4)
1 – S w – S twb
S orw
S twb = ----------------------
- (16-5)
k row
1 + β ----------
k rw
Parameter Sorw is the residual oil saturation in the water-oil system, and β is a
parameter that controls the severity of water blocking. A small value of β (e.g., β
= 1) represents a strong water blocking behavior. The first terms in Equations 16-
3 and 16-4 represent the relative permeabilities to the oil and gas phases when the
hydrocarbons are miscible, while the second terms represent the respective
relative permeabilities when the hydrocarbons are immiscible.
k rh = k row (16-6)
Sg
f o ≡ 1 – ------------------------------
- (16-7)
1 – S w – S twb
Here, Sorw and Sgr denote the residual oil saturation in the water-oil system and
the residual gas saturation, respectively. Next, the normalized hydrocarbon
saturation for the krh curve is defined as
* 1 – S w – S hr
S h = ------------------------------
- (16-9)
1 – S wc – S hr
where Swc is the irreducible water saturation. Finally, the relative permeability to
miscible hydrocarbon is interpolated as
k rh = f o k row ( S we ) + ( 1 – f o )k rg ( S ge ) (16-10)
where the effective water and gas saturations, Swe and Sge, are
*
S we ≡ S wc + ( 1 – S h ) ( 1 – S wc – S orw ) (16-11)
*
S ge ≡ S gr + S h ( 1 – S wc – S gr ) (16-12)
The above interpolation scheme provides a smooth transition for krh from the krow
curve to the krg curve as the hydrocarbon mixture moves from pure oil to pure
solvent during miscible displacement. Note that Equations 16-7 to 16-12 are
purely empirical. They imply that the relative permeability to a fluid phase is
strongly composition-dependent. The validity of this scheme has yet to be
determined.
The relative permeabilities krow and krg in Equation 16-10 ignore relative
permeability hysteresis; i.e., the drainage krg and imbibition krow curves are used
for the calculation. Relative permeability hysteresis, if invoked by the user, is
applied only to the immiscible portion; i.e., the second terms on the right sides of
Equations 16-3 and 16-4 of the effective relative permeability calculations.
1.0
α
Sg Y3 ≥ ssminŠ
Sg Y3 < ssminŠ
0.0
pmis1 pmis2
PRESSURE
fraction in the gas pseudo-phase. In the former case, the miscibility pressure is
represented as a function of the total molar fractions of the gas (z1) and the solvent
(z3). In the latter case, the miscibility pressure is assumed to be a function of the
gas molar fraction in the gas-solvent mixture (y1). This is especially useful for
simulation of the loss of miscibility that is caused by gas injection following
solvent injection. For the three-component miscible model, miscibility pressure is
a function of the solvent total molar fraction. To invoke the miscibility-pressure
table option, the user must enter the miscibility pressure function in a tabular
form. Damping function α is also available with this option. Here, the damping
function is represented only by the solid line in Figure 16-1, and parameter pmis2
(which is a variable in this option) in Figure 16-1 is the miscibility pressure
calculated from the user-input table. In addition to the miscibility pressure table,
the user must enter the size of the transition zone (i. e., the value of pmis2 -
pmis1).
1 – αωμ αωμ
μ oe = μ o μm (16-13)
1 – αωμ αωμ
μ ge = μ g μm (16-14)
where the mixture viscosity, μm, is calculated from the 1/4-power mixing rule,
* S o – S twb – S orm
S o ≡ -------------------------------------- (16-16)
1 – S w – S twb
The mixing parameter, ωμ, is used to control the degree of fluid mixing. A value
of zero corresponds to the case of a negligible dispersion rate (piston-like
displacement). On the other hand, a value of one corresponds to complete mixing.
Chase and Todd16 suggest that the effective mass densities used in the gravity
terms be approximated by
⎧ ⎛ S o⎞ ⎛ S o⎞ ⎫
ρ oe = α ⎨ ρ o ⎜ -----⎟ + ρ g 1 – ⎜ -----⎟ ⎬ + ( 1 – α )ρ o (16-17)
⎩ ⎝ S n⎠ oe ⎝ S n⎠ oe ⎭
⎧ ⎛ S o⎞ ⎛ S o⎞ ⎫
ρ ge = α ⎨ ρ o ⎜ -----⎟ + ρ g 1 – ⎜ -----⎟ ⎬ + ( 1 – α ) ρ g (16-18)
⎩ ⎝ S n⎠ ge ⎝ S n⎠ ge ⎭
where
1⁄4 1⁄4
⎛ S o⎞ M – ( μ o ⁄ μ ge )
⎜ -----⎟ = -----------------------------------------------
1⁄4
(16-19)
⎝ S n⎠ ge M –1
1⁄4 1⁄4
⎛ S o⎞ M – ( μ o ⁄ μ oe )
⎜ -----⎟ = -----------------------------------------------
1⁄4
(16-20)
⎝ S n⎠ oe M –1
and
μ
M = ----o- (16-21)
μg
Note that Equations 16-19 and 16-20 are indeterminate when the mobility ratio
(M) is equal to one. For this case, the following alternate mixing model is used:
ρ oe = α [ ( 1 – ω ρ ) ρ o + ω ρ ρ m ] + ( 1 – α ) ρ o (16-22)
ρ ge = α [ ( 1 – ω ρ ) ρ g + ω ρ ρ m ] + ( 1 – α ) ρ g (16-23)
where
⎛ S o⎞ ⎛ S g⎞
ρ m = ρ o ⎜ -----⎟ + ρ g ⎜ -----⎟ (16-24)
⎝ S n⎠ ⎝ S n⎠
The viscosity mixing parameter, ωμ, and the density mixing parameter, ωρ, are
generally treated as constants throughout the entire reservoir. The recent advance
in simulation technology suggests that characterization of the mixing parameters
for each model region or gridblock may become possible. To broaden the
capability of the miscible option, a new feature to specify either one or both
mixing parameters for each model gridblock has been added to the simulator. This
feature is invoked by entering array data for the mixing parameters in the
ARRAYS section of the VIP-EXECUTIVE initialization module. The keywords
(array names) for the viscosity and density mixing parameters are OMGV and
OMGD, respectively.
must enter the solvent K-value [K3(P)], the compressibility factor of solvent in the
oil phase [Zso(P)], the viscosity of solvent in the oil phase [μso(P)] as a function of
pressure, and the two properties [Bs(P) or Zs(P), and μs(P)] specified in the first
option. The oil and gas phase viscosities, μo and μg in Equations 16-13 and 16-14,
then are calculated as:
* * ( 1 – x3 ) x3
μ o = [ μ o ( P, x 1 ) ] ⋅ [ μ so ( P ) ] (16-26)
( 1 – y3 ) y3
*
μg = [ μg ( P ) ] ⋅ [ μs ( P ) ] (16-27)
where
* x1
x 1 ≡ ------------- (16-28)
1 – x3
The oil and gas phase compressibility factors, Zo and Zg, are calculated by
* *
Z o = Z o ( P,x 1 ) ( 1 – x 3 ) + Z so ( P ) x 3 (16-29)
* *
Z g = Z g ( P,x 1 ) ( 1 – y 3 ) + Z s ( P ) y 3 (16-30)
Note that when the first option is invoked, the mole fraction of the solvent in the
oil phase, x3, in Equations 16-26, 16-28, and 16-29 is zero.
17
Non-Darcy Gas Flow
17.1 Introduction
Two features modelling non-Darcy gas flow near well are included in the well
model:
For the first option (pressure-dependent gas density and viscosity), the Russell
Goodrich17 and pseudo-pressure18 equations are implemented. For the second
option (rate-dependent skin factor), the user can specify a rate-dependent skin
factor for each well or for each perforation. If the rate-dependent skin factor for a
well is specified, then the user can choose either the invert thickness, invert
permeability-thickness, or constant option to allocate the rate-dependent skin to
its perforations. The pressure-dependent gas property option and rate-dependent
skin option can be invoked separately.
where:
Pl gridblock pressure
γ pressure gradient
Dl subsea depth
Without the non-Darcy gas flow option, gridblock pressure is used to calculate
wellbore viscosity and density. The bottom-hole pressure is not included in the
calculation.
17.2.2 Recommendations
The Russell Goodrich method and the pseudo-pressure method usually give the
same results except in the region where the density-to-viscosity ratio is not a
linear function of pressure and the pressure difference between the gridblock and
bottom-hole pressures is large. In low- and high-pressure regions, the density-to-
viscosity curve is usually a straight line. Both methods yield the same results in
those regions. In the middle transition region, the results could be different.
However, if the pressure difference between the gridblock pressure and bottom-
hole pressure is small, then both methods give approximately the same answers.
2π ⎧ k rgl ⎫
Q gl = ------------------------------------------------- ⎨ k l h l --------------- [ P l – P bh – γ ( D l – D ref ) ] ⎬
r bl μ gl B gl
ln ------- + S l + D sl Q gl ⎩ ⎭
r wl
(17-2)
where
Pl gridblock pressure
γ pressure gradient
Sl skin factor
* 2π ⎧ k rgl ⎫
Q gl = ----------------------- ⎨ k l h l --------------- [ P l – P bh – γ ( D l – D ref ) ] ⎬ (17-3)
r bl μ gl B gl
ln ------- + S l ⎩ ⎭
r wl
The flow rate with rate-dependent skin is expressed as a fraction of the flow rate
without rate-dependent skin as shown in the following equation:
*
Q gl = f l Q gl (17-4)
where
r bl
ln ------- + S l
r wl
f l = ---------------------------------------------------
- (17-5)
r bl *
ln ------- + S l + f l D sl Q gl
r wl
2
f l = –-----------------------------------
a + a + 4a-
(17-6)
2
where
r bl
ln ------- + S l
r wl
a = ----------------------
- (17-7)
*
D sl Q gl
■ constant option
Ds × ∑ hi
i=1 -
D sl = --------------------------- (17-8)
hl
where
hl perforation thickness
Constant Option
If the user does not want to scale the rate-dependent skin factor for each
perforation, then the constant option can be chosen. The program assigns each
perforation the same rate-dependant skin factor as the input data
D sl = D s (17-9)
nperf
Ds × ∑ ki hi
i=1
D sl = -------------------------------
- (17-10)
kl hl
To use this option, the user needs to pay attention to the following two conditions:
■ When the rate-dependent skin factor option is specified, the well index cannot
be zero. Well index set to zero is a special case in VIP-EXECUTIVE that
means the well index is adjusted to honor both the rate and bottom-hole
pressure constraints. In such a case, the rate-dependent skin factor does not
have any meaning. Thus, an error message prints in the output and the
simulation run terminates.
■ If the perforation skin factors are calculated from the input well skin factor,
each time new perforation information is specified in the FPERF card, the
program recalculates the skin factor for each perforation.
18
Numerical Solution
18.1 Introduction
VIP contains options for both the fully implicit and IMPES formulation. For the
same timestep size, the IMPES method is much faster than the fully implicit
method. The IMPES method solves the component conservation equations
explicitly; the timestep size is therefore constrained by the stability limit. The
fully implicit method is unconditionally stable and allows large timesteps to be
taken. For problems with small gridblocks, such as coning problems, the fully
implicit method is overall much faster. The procedure of numerical solutions for
the IMPES method is discussed in the following sections.
where Vm is the gridblock volume and superscripts n and n+1 denote the timestep
level. The second, third, and fourth terms in Equation 18-1 represent the regular
(non-fault) inter-grid flow term (Kn being the total number of regular neighbors),
the inter-grid flow term for fault connections (Kf being the total number of
neighbors through fault connection), and the injection/production term (Kw being
the total number of wells existed in the gridblock), respectively. Equation 18-1 is
in units of lb-mol/day. For gridblock M, the inter-grid transmissibility terms are:
k = k or kf ; I = 1 or 2 (18-2)
k = k or kf ; I = 1 or 2 (18-3)
n
n+1 n+1
0.5 [ ( ρ o M o ) M + ( ρ o M o ) MM ] g
ΔP ok = P o ( M ) – P o ( MM ) – -------------------------------------------------------------------------- ( D M – D MM )
5.16146 ⋅ 144g c
k = k or kf (18-4)
n
n+1 n+1 0.5 [ ( ρ g M g ) M + ( ρ g M g ) MM ] g
ΔP gk = P o ( M ) – P o ( MM ) – -------------------------------------------------------------------------- ( D M – D MM )
5.16146 ⋅ 144g c
n
+ [ P cg ( M ) – P cg ( MM ) ] , k = k or k f (18-5)
n
⎧ K ro ρ o x i ⎫
T k w oi ≡ ⎨ WI K h ------------------- ⎬ (18-6)
⎩ μo ⎭kw
n
⎧ K rg ρ g y i ⎫
T k w gi ≡ ⎨ WI K h ------------------- ⎬ (18-7)
⎩ μg ⎭kw
ΔP kw j ≡ P o ( M ) – P W – γ ( D M – D ref ), j = o, g (18-8)
ρ g ( M1 )M g g
γ = ----------------------------------
- (18-9)
5.6146 ⋅ 144g c
⎛K rw K ro K rg⎞
n
∑ ⎝ μw μo μg ⎠ k
Kh --------
- + -------
- + --------
k 1
where
ρj Mj g
γ j = -----------------------------------, j = o, g (18-11)
5.6146 ⋅ 144g c
o
ρw Bw Mw g
γw = ----------------------------------
- (18-12)
5.6146 ⋅ 144g c
In Equation 18-10, the summations are taken over all (Kp) perforated layers.
Variables γj in Equations 18-11 and 18-12 are in units of psi/ft.
where Tmko, Tmkg (k = k of kf ), Tko, and Tk are summations of Equations 18-2, 18-
3, 18-6, and 18-7, respectively, over all hydrocarbon components.
where
k = k or kf ; I = 1 or 2 (18-15)
n
⎧ K rw B w ⎫
T wkw ≡ ⎨ WI K h ---------------- ⎬ (18-16)
⎩ μw ⎭kw
o
n+1 n+1 ρw Mw g n
ΔP wk = P o ( M ) – P o ( MM ) – ----------------------------------- [ B w ( MN ) ] ( D M – D MM )
5.6146 ⋅ 144g c
n
– [ P cw ( M ) – P cw ( MM ) ] , k = k or k f ; MN = Max ( M, MM ) (18-17)
where
ΔP kw j ≡ P o ( M ) – P W – γ ( D M – D ref ) (18-18)
o
ρ w M w B w ( M1 )g
γw = --------------------------------------
- (18-19)
5.6146 ⋅ 144g c
Bw n
] } ⎛ ------⎞
n+1 n+1 n+1 n n+1
Q Na = { A Na + B Na [ P o ( M ) – P o ( M ) (18-20)
⎝ μ w⎠
Here, QNa denotes the amount (in STB/day) of water influx into the gridblock.
ANa and BNa are parameters defined in the Carter-Tracy Method.
B s T w G w ΔS W R M7
C x1 C x2 C L C z1 C z2 C f C s C p ΔP o R M8
H D W ΔP W RW
The first Nc equations are the fugacity equality equations followed by (Nc + 1)
mass conservation equations. The penultimate equation is the saturation constraint
equation, while the last equation is the implicit bottom-hole pressure constraint
equation.
R Mi ≡ ( x i Φ oi – y i Φ gi ) i = 1, ...,N c (18-22)
∂R Mi
g xim ≡ – -----------
- m = 1, ...,N c – 1 (18-23)
∂x m
∂R Mi
g Li ≡ – -----------
- (18-24)
∂L
∂R Mi
g zim ≡ – -----------
- (18-25)
∂z m
∂R Mi
g Pi ≡ – -----------
- (18-26)
∂P O
∂R M ( Nc + i )
B fi ≡ – ------------------------
- (18-28)
∂z i
∂R MNc + i
- i = 1, ...,N c – 1
B i ≡ – -------------------- (18-29)
∂F
∂R MNc + i
T i ≡ – --------------------
- i = 1, ...,N c – 1 (18-30)
∂P o
∂R MNc + i
G i ≡ – --------------------
- i = 1, ...,N c – 1 (18-31)
∂P W
∂R M ( 2Nc )
B ≡ – --------------------- (18-33)
∂F
∂R M ( 2Nc )
T f ≡ – --------------------- (18-34)
∂P o
∂R M ( 2Nc )
G f ≡ – --------------------- (18-35)
∂P W
∂R M ( 2Nc + 1 )
B s ≡ – ---------------------------- (18-37)
∂F
∂R M ( 2Nc + 1 )
T w ≡ – ---------------------------- (18-38)
∂P o
∂R M ( 2Nc + 1 )
G w ≡ – ---------------------------- (18-39)
∂P W
⎧ RT ⎫
R M ( 2Nc + 2 ) ≡ – ⎨ 1 – F [ ( 1 – L )Z g + LZ o ] --------- – S w ⎬ (18-40)
⎩ Po ⎭
∂R M ( 2Nc + 2 ) ∂R M ( 2Nc + 2 )
C xm ≡ – ---------------------------- – ---------------------------- m = 1, ..., N c – 1 (18-41)
∂x m ∂x Nc
∂R M ( 2Nc + 2 )
C L ≡ – ---------------------------- (18-42)
∂L
∂R M ( 2Nc + 2 ) ∂R M ( 2Nc + 2 )
C zm ≡ – ---------------------------- – ---------------------------- m = 1, ..., N c – 1 (18-43)
∂z m ∂z Nc
∂R M ( 2Nc + 2 )
C f ≡ – ---------------------------- (18-44)
∂F
Cs ≡ –1 (18-45)
∂R M ( 2Nc + 2 )
C p ≡ – ---------------------------- (18-46)
∂P o
∂R W
H ≡ ---------- (18-47)
∂P o
∂R W
D W ≡ ---------- (18-48)
∂P W
1. For gridblocks adjacent to aquifers, parameters ANa and BNa are calculated. In
addition, explicit properties such as inter-grid mobilities, inter-grid gravity
gradients, capillary pressures, and well-block mobilities are calculated.
*
* * *
I g x12 g L1 g z11 g z12
* *
g p1 R M1
Δx 1
*
I
*
g L2
*
g z21
*
g z22
*
g p2 Δx 2 R M2
*
*
I g z31 g z32
* *
g p3 ΔL R M3
Δz 1 *
B f1 B1 T1 G1 R M4
B f2 B 2 T2 G2 Δz 2 = R* (18-49)
M5
* * ΔF
I Tf Gf *
R M6
* * ΔS w
I Tw Gw *
R M7
* * ΔP o
Cp GT *
R M8
H D W ΔP W RW
* * ΔP o *
Cp GT R M8
AX = = = b (18-50)
H D W ΔP W RW
4. Residuals in the hydrocarbon balance and the water balance equations (RM6
and RM7 in Equation 18-21) are compared with the material balance error
tolerances (TOLHC in lb-mole/day for hydrocarbons and TOLRW in STB/
day for water). If the residuals do not exceed TOLHC and TOLRW, then the
Newton iteration is assumed to have converged. In this case, the computation
jumps to Step 15. If the maximum number of iterations have been performed,
the computation jumps to Step 13.
6. For the first iteration, the IMPES stability is tested. If the ratio of the total
phase outflow to the current phase amount in any gridblock for any phase,
FACIMP, exceeds a pre-specified limit (STSLIM; default value 1.0), the
timestep size factor is set equal to the user-specified IMPES stability target
throughput (STSTAR; default value 0.9) divided by FACIMP and the timestep
is repeated by moving to Step 14.
7. The maximum changes in Po, Sw , L, and zi over the iteration are calculated. If
none of the maximum changes exceeds the user-supplied maximum allowable
change (DPLIM, DSWLIM, DVLIM, and DZLIM), then the weighting factor
U1 is assigned a value of one. Otherwise, U1 is set equal to the minimum
value among the ratios of the maximum allowable change for each unknown
to its predicted maximum change over the iteration.
8. If none of the maximum changes over the iteration exceeds the user-specified
convergence tolerance (TOLP, TOLS, TOLV, and TOLZ), then the Newton
iteration is considered to have converged.
9. All primary unknowns are updated. The new values (at iteration level m+1)
are set equal to the most recent values (level m) plus U1 multiplied by the
calculated changes over the iteration. If either the new L or V in any two-
phase gridblock becomes negative, the variable is assigned a new value of
zero and the gridblock is redesignated as a single-phase block.
10. If the Newton iteration has converged and residuals need to be recomputed for
iteration summary output, the computation is returned to Step 2. If the Newton
iterations have converged and residuals are not needed, the computation
jumps to Step 15.
11. If the black oil option is used, simple analytical expressions are used to detect
the appearance of the second phase and the phase compositions. For the
compositional model, saturation pressures for those single-phase gridblocks
showing a noticeable change in the total mole fraction of any component over
the timestep (i.e., |Δzi| greater than 10-6 for any component i) are calculated to
detect the appearance of the second hydrocarbon phase. If the calculated
saturation pressure is larger than the gridblock pressure, the second phase is
assumed to have formed during the iteration and a flash calculation is
performed to determine xi, yi, and L. However, if the saturation pressure
calculation in the previous or current Newton iteration fails to converge, the
13. If the Newton iteration failed to converge, the timestep size is cut by one-half
(smaller factors for subsequent failures) and the timestep is repeated by
returning to Step 1.
14. If the IMPES stability check failed, then the new timestep size is set equal to
the old timestep size multiplied by the timestep size factor determined in Step
6 and the timestep is repeated by returning to Step 1.
15. If the Newton iteration converged, the maximum changes in Po, Sw , Sg, zi,
and V and their ratios to the maximum allowable change over a timestep
(DPMAX, DSMAX, DZMAX, DVMAX) are calculated. If any of the ratios
exceeds the value MAXOVR (default 1.5), the new timestep size is set equal
to the old timestep size divided by the maximum ratio, and the timestep is
repeated by returning to Step 1.
16. The converged solution is accepted. The material balance errors for the total
hydrocarbon and water over the timestep are calculated. A new timestep size
is selected, and a new timestep is started by returning to Step 1. The material
balance error for water or total hydrocarbon is defined as
F ip – F iip + C pd – C in
Err = ---------------------------------------------------
- (18-51)
C pd + C in
where Fip, Fiip, Cpd, and Cin (in STB for water or lb-mole for total
hydrocarbon) are fluid in place, initial fluid in place, cumulative production,
and cumulative injection of water or total hydrocarbon, respectively.
18.5.1 Gauss
This is a direct solver that uses Gaussian elimination with the gridblocks ordered
by D4 ordering (alternate diagonal). It can solve a system of equations arising
from the 5-point (2-D) and 7-point (3-D) finite-difference approximations. The
implicit well constraint equations, fully implicit, and IMPES formulations can all
be handled by GAUSS. However, it is not applicable to problems with faults.
18.5.2 BLITZ
BLITZ is based on the preconditioned, generalized minimal residual method
(GMRES).49 It can handle the following problems:
There are many parameters that the user can adjust to improve the convergence
rate of the solver. Some of the more important parameters are described below.
Parameter
18.5.3 EXCEL
EXCEL also is based on preconditioned ORTHOMIN. The preconditioners are
based on incomplete LU factorization with natural ordering and any level of infills
may be specified. EXCEL can handle both IMPES and fully implicit
formulations. In addition, it can handle problems with arbitrary nonstandard
connections that are not logically vertical. The data structure of EXCEL is based
on sparse matrix storage; therefore, it is not very efficient on a vector computer.
For some problems, it is an order of magnitude slower than BLITZ. The
parameters for controlling the EXCEL solver are described in the VIP-
EXECUTIVE Reference Manual, Chapter 7.
19
Optimal Material Balance Option
00000
19.1 Introduction
The Optimal Material Balance Option is the combination of three special features
implemented in the IMPES version of VIP-EXECUTIVE to improve
computational efficiency through code restructuring. These features are
Two techniques originally developed for the Optimal Material Balance Option are
now available also with the Standard Iterated IMPES formulation:
Results of test runs indicate that when using the Optimal Material Balance Option,
VIP-EXECUTIVE runs 3.4 to 4.7 times faster (with approximately the same
number of timesteps) than the standard iterated IMPES formulation for the SPE
5th Comparative Solution Problems24 (Scenarios 1 to 3). Similar speedups also
are obtained for pattern or partial-field-type systems with a limited number of
wells and no iterative well management calculations. For full-field black-oil and
compositional models with extensive well management calculations, CPU time
improvements of 43 to 55% were observed.
Because of the difference in the unknown update procedure described above, the
convergence criteria in VIP-EXECUTIVE are also different from those in
References 5 and 25. In VIP-EXECUTIVE, the Newton iteration is considered
converged if the residuals in Equations 9-2 and 9-3 or the changes in all primary
unknowns during a Newton iteration drop below the user-specified tolerances. On
the other hand, with preservation of the material balance for each component, the
simulators presented in References 5 and 25 determine the convergence of the
Newton iteration through the residual in saturation constraint Equation 9-9. With
this procedure, most timesteps converge after only one outer (Newton) iteration,
as compared with the two or more iterations required for VIP-EXECUTIVE.
Kn
VM n + 1 n + 1 n
------- [ φ
Δt
F – ( φF ) ] + ∑ [ TMko ΔPko + TMkg ΔPkg ]
k=1
Kv
Kn Kw
VM n + 1 n + 1 n + 1 n
-------- [ φ
Δt
B w S w – ( φB w S w ) ] + ∑ [ TMkw ΔPkw ] + ∑ [ T kkw ΔP kww ] = 0 (19-2)
k=1 kw = 1
Here, VM is the gridblock volume, superscripts n+1 and n denote new and old
timestep levels, and the second and third terms in both equations are the inter-grid
(including faults) flow and source terms, respectively. Using the IMPES
formulation, the flow coefficients (TMkj and Tkwj, j = o, g, w) are evaluated
explicitly.
After solving the simultaneous pressure and well constraint equations, the new
pressure of each gridblock and the new wellbore pressures are determined. Since
the porosity and water formation volume factor are only a function of gridblock
pressure, φn+1 and Bwn+1 can be updated readily from the new pressures.
Unknowns Fn+1 and Swn+1 then can be calculated directly from Equations 19-1
and 19-2. With this new procedure, the total hydrocarbon and water material
balances are satisfied exactly.
Kp g
K rj ρ j
q = WI I o BSEP ∑ Kh ∑ -----------
- [ P l – P w – γ ( D l – D ref ) ]
μj
l=1 j=0
Kp
K rw ρ w
+ WI I w ∑ Kh ---------------- [ P l – P w – γ ( D l – D ref ) ] (19-3)
μw
l=1
Here, q is the user-specified well rate; Wl is well index; Io and Iw are either 0 or 1,
depending on well specifications; Pl and PW are gridblock and wellbore pressures
at the reference depth; γ is the wellbore pressure gradient; and Dl and Dref are
gridblock and reference depths, respectively. The variable BSEP is 1.0, except for
surface production rate specifications where BSEP is the ratio of the specified
surface rate (in STB/day or MSCF/day) to the total hydrocarbon molar rate (in lb-
mol/day).
Using the IMPES formulation, the mobility terms in Equation 19-3 are treated
explicitly. The equation is linear with respect to both Pl and PW if BSEP is equal
to one. Under this condition, the new wellbore pressure calculated from the linear
solver and from the solution of the well constraint equation will be identical,
provided there is no change in the injecting or producing perforations over the
iteration because of crossflow. In this case, the material balance will be satisfied
exactly. This situation applies to all injection and production wells with reservoir
rate specifications.
For production wells with an oil or gas surface rate specification, the variable
BSEP is a function of the total hydrocarbon composition that depends on both the
gridblock and bottom-hole pressures. In this case, Equation 19-3 is no longer
linear. The new wellbore pressure, which is calculated from the linear solver
(used to update zi, F, and Sw) and the well constraint equation (used to update well
rates and calculate the material balance), are not identical. Thus, the material
balance will not be satisfied exactly for gridblocks containing producing
perforations. Another potential problem with the VIP-EXECUTIVE IMPES well
formulation is that when Equation 19-3 is linearized, the derivatives of BSEP with
respect to Pl and PW are neglected, resulting in a non-rigorous coupling of the
reservoir and well constraint equations. This also affects the material balance for
hydrocarbons and, possibly, the number of Newton iterations required for
convergence.
m+1 m
PW = PW + Δ Pw (19-4)
where ΔPw is the change in bottom-hole pressure calculated from the solver. The
new bottom-hole pressure then is used to calculate the reservoir rates and the new
BSEP from which the total well reservoir rate is converted to the surface rate.
This surface rate is then compared with the user-specified rate, q. If the difference
between the two rates is within the convergence tolerance (0.01% of the specified
rate), then no further iterations for PW and BSEP will be performed, and the
timestep is considered converged. Here, the material balance is satisfied exactly
because the same bottom-hole pressure, Equation 19-4, is used for zi, F, Sw, and
the production rate calculations. Second, well constraint Equation 19-3 is
rigorously linearized. Without this rigorous treatment, the Newton iteration will
not converge within one iteration for most problems. The total hydrocarbon molar
production rate for component i, Qi, and the total hydrocarbon molar production
rate, Qt, are defined as
Kp
⎛ K ro ρ o K rg ρ g ⎞
Q i ≡ WI ∑ Kh ⎜ -------------- x i + -------------- y i⎟ [ P l – P W – γ ( D l – D ref ) ] (19-5)
⎝ μo μg ⎠
l=1
Kp
⎛ K ro ρ o K rg ρ g⎞
Q t ≡ WI ∑ Kh ⎜ -------------
- + --------------⎟ [ P l – P W – γ ( D l – D ref ) ] (19-6)
⎝ μo μg ⎠
l=1
Qi
Z i ≡ ----- (19-7)
Qt
BSEP Q t – q = 0 (19-8)
where
⎛ K ro ρ o K rg ρ g⎞
H l = WI BSEP Kh l ⎜ -------------
- + --------------⎟
⎝ μo μg ⎠ l
Nc
⎧ ∂BSEP K ro ρ o K rg ρ g ⎫
+ WI Kh l ∑ ⎨ ------------------ -------------- ( x i – Z i ) + -------------- ( y i – Z i ) ⎬ (19-10)
⎩ ∂Z i μo μg ⎭
i=1
and
Kp
Dw = – ∑ Hl (19-11)
l=1
In the standard version of VIP-EXECUTIVE, the second term on the right side of
Equation 19-10 is neglected. In the current implementation of the optimal
material balance option, both terms on the right side of Equation 19-10 are used to
build the Jacobian coefficients Hl and Dw. Note that the partial derivatives of
BSEP with respect to Zi are provided by a new wellbore flash routine.26
Using the above procedure, the Newton iteration of the simultaneous reservoir
and well constraint equations usually converges in one iteration because BSEP
varies only slightly from one iteration to another. The well rate calculation will
not converge if the BSEP value changes significantly over one iteration, or if the
active injecting/producing perforations change over one iteration because of cross
flow. Under the latter condition, the functional form of the well constraint
equation is changed between two Newton iterations and an additional iteration
must be performed.
2. After each Newton iteration, the new phase saturations are calculated and the
residuals of the saturation constraint equation are then compared with the
maximum tolerance. If the convergence criterion is not satisfied, another
Newton iteration must be performed, and the standard VIP-EXECUTIVE
procedure for calculating the new well rates and bottom-hole pressure is
followed. Otherwise, Equation 19-4 is used first in the well routines to check
for convergence of the well constraint equations for producers. If the well
constraint equations fail to converge, the standard procedure in the wellbore
flash routine (WFLSHP) is adopted to determine the new wellbore pressure
and new well rates. If the well constraint equation is converged, no further
iterations for PW and BSEP are performed.
3. After the new well rates and bottom-hole pressure are determined, the total
hydrocarbon and water material balances for gridblocks containing wells are
calculated to check for convergence of the well rate calculation. To ensure a
good material balance, these well rates must agree with the rates obtained
from the linear solver and used for calculation of zi, F, and Sw. The material
balance is considered converged when the ratio of the difference between the
new well rate (as obtained from the linear solver) and the new well rate (as
calculated from the well constraint equation) to the sum of the two rates is less
than TOLWCN (default 10-5). Otherwise, a new iteration will be performed.
The unknown update and convergence control procedure generally reduces the
average number of Newton iterations per timestep. For problems in which the
iterative well management calculations do not apply and the active producing/
injecting perforations do not change between two iterations, the Newton iterations
normally converge after only one iteration, as compared with the three or four
iterations required for the standard VIP-EXECUTIVE procedure. This
improvement could account for a more than 50% reduction in CPU time. For
simulation models that involve intensive well management calculations (e.g., full-
field simulation models), a significant reduction in the average number of
iterations per timestep also is possible, although the new procedure normally
requires more than one iteration per timestep for these problems.
The terms one-phase and two-phase refer to the number of hydrocarbon phases.
Water is treated as an independent phase with no interactions (except for the flow
calculation) with the hydrocarbons. In the fluid-type reordering scheme, all Type
1 gridblocks are grouped together, followed by all Type 2 gridblocks, and so on.
For simplicity, most major arrays are stored permanently in the original, natural
order, and one additional NB (the total number of gridblocks) array is created to
relate fluid-type order to natural order. Rearrangement of the fluid-type order is
performed whenever any part of the simulation causes a fluid-type change for any
number of gridblocks. The rearrangement can be accomplished efficiently by
swapping a small number of array elements at a few gridblocks. This scheme is
illustrated by considering the simple case in Figure 19-1. In Figure 19-1A, 15
two-phase gridblocks (N2PHS = 15) and 5 single-phase gridblocks are at the
beginning of a Newton iteration. After the iteration, five of the two-phase
gridblocks become single phase (Figure 19-1B). Thus, the number of two-phase
gridblocks, N2PHS, reduces to 10. Rearrangement of the grid order can be done
easily by moving only those gridblocks that changed from two to single phase
and originally were located to the left side of new pointer N2PHS (10). In this
case, reordering is required only for three (Grid Numbers 6, 8, and 9 in Figure 19-
1B) of the five blocks switching from two to single phase. These three gridblocks
will be swapped with the three two-phase gridblocks that are to the right side of
new pointer N2PHS (Figure 19-1B). The gridblocks to be swapped are identified
by the arrow signs between Figures 19-1B and 19-1C; the new grid-type order is
shown in Figure 19-1C. After the gridblocks to be swapped are identified,
rearrangement of fluid properties for subsequent phase behavior calculations can
A |XXXXXXXXXXXXXXX|OOOOO|
B |XXXXXOXOOXOXOXX|OOOOO|
↓ ↓ ↓ ↓ ↓ ↓
C |XXXXXXXXXX|OOOOOOOOOO|
X = Two-phase grids
O = Single-phase grids
A Initial reordered gridblock distribution
B Distribution after unknown updating
C Distribution after grid reordering.
In the coefficient generation routine, calculations for the Type 1 and Type 2
gridblocks are done separately. For the Type 2 (full Jacobian) gridblocks, the
normal full Jacobian calculation is performed. In addition, the lower triangular
portion of the Jacobian must be stored for subsequent timestep use (or iteration)
because the standard version of VIP-EXECUTIVE stores only the factored and
normalized upper triangular portion of the Jacobian (Array C). The lower
triangular portion is needed to factor the residual arrays during the subsequent
partial Jacobian calculations. This procedure requires an extra storage of
Nc(Nc+1)NB/2 words. For the Type 1 (partial Jacobian) gridblocks, no derivative
calculations (for Array C) are performed for the fugacity equations. The lower
triangular portion of the Jacobian that is stored in the most recent full Jacobian
calculation is used to factor the residual array, Array C, which was stored
previously, is used to eliminate unknowns xi and L (or yi and V) in the saturation
constraint equation.
This option applies only to those systems without fault connections, since the
selection procedure is based solely on regular connections. Note that this option
should be applied only to those cases in which the phase change is initiated from
gridblocks containing well perforations. One illustration of this limitation is a
simulation model that invokes the boundary flux option in which an influx of gas
phase or depletion caused by an outflux of fluids could result in a single oil-phase
boundary gridblock to flash into two phases first. This possibility is not
accounted for in the current implementation. For this reason, this option should
not be used with the boundary flux option.
This procedure reduces the number of gridblocks that need the stability test by up
to a factor of 4 for the problems tested. This translates into a four-fold reduction in
the stability test and flash calculation portion of the CPU time.
With the grid reordering procedure in place, efficient vectorizations of the above
calculations become practical. In the current vectorization effort, only the
standard VIP-EXECUTIVE saturation pressure and flash calculation routines are
vectorized. To simplify the procedure and ensure a high efficiency, the bubble-
point pressure calculation (BUBPT) and dew-point pressure calculation (DEWPT)
routines are combined first to a saturation pressure calculation routine. This can
be done easily without any loss of the vectorization efficiency because the two
routines perform identical calculations (except for the selection of the correct real
root for the compressibility factor of the second phase after the solution of the
cubic equation). By combining the two routines and vectorizing the resulting
saturation pressure routine, the vector length will be equal to the total number of
single-phase gridblocks requiring either the bubble- or dew-point pressure
calculation. The combined saturation pressure calculation routine (SATPRS) is
called up to two times. In the first call, the saturation pressures are calculated for
all single-phase gridblocks that require such calculations. In the second call,
saturation pressures are calculated for only those single-phase gridblocks that fail
the first saturation pressure calculation. This is done by switching from the
bubble-point calculation to the dew-point calculation, and vice-versa. Note that
without the combined saturation pressure routine, the vectorized BUBPT and
DEWPT routines would have to be called twice each.
simulation problems, vector lengths in the flash calculation often are short
because few gridblocks change phases during a given timestep. In this situation,
the vectorized flash routines (FLASHV or FLSHSV) may not be as efficient as the
original routines (FLASH or FLSHSS) because the original flash routines have a
large number of vectorized loops with a small vector length of Nc. To ensure that
no efficiency is lost, a vectorized flash routine is called only if the number of
gridblocks requiring the flash calculation is greater than Nc/2. Results of several
test runs indicate that the vectorized flash routines are slower than the original
routines if their vector length is shorter than Nc/2.
20
Parallel Computing
00000
20.1 Summary
The following section discusses the details of the implementation of PARALLEL-
VIP. The primary development goals of this feature were to allow the faster
simulation of existing models and/or the simulation of larger models ranging to
millions of grid blocks. Parallelization is achieved through the use of local grid
refinement to subdivide a model into different domains for each of the parallel
CPU’s. Test problems described below for the simulator range from large IMPES,
black-oil cases with severe faulting to a fully-implicit seven-component
compositional simulation. Model sizes range to more than one million finite
difference cells for both black-oil and compositional cases. The following section
describes the implementation of message passing interface - a standard for
parallelization which has been implemented on almost all parallel computer
hardware as well as clusters of workstations. Results are shown for the IBM SP2,
the Silicon Graphics Origin 2000 and Power Challenge, and a cluster of IBM
workstations. Excellent parallel efficiencies were observed for all cases.
20.2 Introduction
The attraction of parallel computing to achieve high computing efficiencies has
existed for decades. What was generally lacking was the ability to easily port
software to the parallel environment. Because of the diverse nature of reservoir
simulation, simple parallelization schemes have not been viable for a usable,
general-purpose, approach to reservoir modeling. Profiles of computing work
load for a typical model often show tens, if not hundreds, of subroutines which are
involved in a substantial portion of the calculations. Because of this, major
reprogramming of reservoir simulation models is required to achieve high parallel
efficiencies. Several papers in the literature74-79 discuss techniques which have
been used to bring about efficient parallel reservoir simulations. The unique
feature of the simulator of this work is its reliance on local grid refinement for
parallelization. With local grid refinement the same simulation program can be
used to perform simulations either serially on a single processor or in parallel on
multiple processors simply through data manipulation.
The following section discusses the use of local grid refinement for
parallelization. Basically, each grid is assigned to a processor. Examples are
given in which variation of the processor assignment and the grid refinement can
dramatically effect parallel performance. Different preconditioners for the
parallel linear equation solvers are also tested. Finally, a load-balancing example
utilizes the flexibility of local grid refinement and processor assignment to
achieve improved parallel efficiency.
For grid to grid connections between two cells, contributions to the residuals and
to the Jacobian terms arising from intercellular flow are computed separately for
each cell. This results in a minor inefficiency since certain terms are computed
twice. However, as demonstrated later in the examples, the resultant overhead
does not appear to be significant when the number of boundary cells is small
relative to the total number of cells in a given grid.
When flow is computed between a cell in any given grid (i.e., Grid 1) and a cell in
a neighboring grid (i.e., Grid 2), the data for Grid 2 is referenced to a buffer in
which all pertinent quantities have been placed. The use of the buffer instead of
the actual grid allows parallel message-passing to be easily implemented as
described below. In addition, since data in the buffer is more compact than grid
data, fewer cache misses and greater computational efficiency may result even in
serial calculations.
For parallel domain decomposition the base grid may be totally refined into
subgrids so that all gridblocks in the base grid become inactive. Each grid is
For example, consider the simple case shown in Figure 20-2 in which the base
grid consists of four blocks. Each of these blocks is then refined to the level
desired for a given simulation. Grid 1 corresponds to the refinement of the upper
left block in the base grid and is assigned to processor A. Similarly, each
refinement of the base grid is assigned to a different (or the same) processor. At
the end of the decomposition the complete base grid is refined and all refined
blocks have been assigned to a processor. Multiple levels of grid refinements can
be accommodated as well. If further refinement of a cell is required to a radial or
Cartesian grid, the further refinement is assigned to a processor in the same
manner as the level 1 refinements.
DECOMPOSE
SQUARE 3 2 1
TALL 1 2 1
ENDDECOMPOSE
Since highly deviated and horizontal wells are so prevalent in modern oil field
operations, it would be excessively restrictive to require that each individual well
be contained entirely within a single parallel computing domain. Therefore, it
must be assumed that messages will have to be passed between domains in order
to gather all of the information necessary to allocate production and injection rates
amoung multiple perforated intervals. Furthermore, use of production or injection
constraints that involve groups of wells will, in general, lead to interdomain
coupling of well rates.
Allocation of well flow rates requires such grid cell dependent data as pressures,
saturations, and fluid composition. (This list is not all inclusive.) The base
simulator was originally developed for high efficiency on vector processing
computers. In order to allow unit strides in addressing data in the well
management code, all grid cell dependent data required for well calculations has
always been moved into arrays dimensioned by the total number of perforated
intervals at the beginning of each timestep or each nonlinear iteration, as
appropriate. Thus, the well management code has never addressed data stored by
grid cells.
At the time of this writing no parallelization of the well management code has
been attempted, so all processors simultaneously compute rates for all wells. The
run time reported reflect this sequential step in the simulation process.
Parallelization using the data flow described above is still possible; however,
additional efficiency can be achieved by replacing the broadcast operations with
more selective (but more complex) message passing.
Other techniques relying on scattering well information among all processors and
gathering resulting rates could lead to high overhead and thus negate any
improved parallel performance at least for simulations involving historical
production. On the other hand, tubing hydraulics calculations for predictive well
management may be computationally intensive enough to support gather/scatter
type operations. This latter technique will be investigated in the future.
A1
A2
A3
L = A4
Q1 R2 B1
Q2 R3 B2
Q3 R4 B3
The diagonals of the B submatrices are modified from the original coefficients to
maintain rowsum constraints. This approximate factorization leads to
independent solutions on each processor for the large Ai domains and the
boundary plane matrices, Bi. The back substitution portion of the preconditioning
simply involves the updating of residuals from boundary solutions, and a second
solve on the large Ai domains.
possible decomposition which can be solved with the composite grid technique
(See Reference 83 for additional details.) First, the larger domains of level 3 are
solved. Information from these solutions is then communicated across boundaries
where necessary. Next the equations are solved for level 2, and finally, level 1 is
solved. Each of the lower level solutions involve the use of rowsum constraints.
Back substitutions are then performed in the reverse order. The flexibility of this
solver allows the easy implementation of block red-black algorithms, or as a
special case, the linear decomposition of the strip decomposition solver described
above.
Figures 20-7 and 20-8 show results for the composite grid and a solution method
based on a linear (or strip) decomposition method for a 95x95x35 (315,875 cells)
hypothetical IMPES case. The efficiency of the solvers was compared to both a
serial version of the same technique and to the proprietary “serial solver” of the
original reservoir simulator used in this work. The serial solver utilized a single-
grid, nested factorization preconditioning. As shown in Figure 20-7, performance
of the strip decomposition solver is superlinear for almost all cases ranging from
3-16 processors. Compared to the serial solver, superlinear performance is also
observed. Superlinear performance was also observed for the composite grid
solver compared to the serial version of the same algorithm (See Figure 20-8).
Compared to the “serial solver”, the composite grid technique achieved about
75% of ideal speedup, i. e., a factor of twelve on sixteen processors. The speed
advantage for the strip method results from the two-way ordering discussed
above. Other studies have shown similar superlinear speedups on up to 71
the original grid was decomposed into both one-dimensional strips as well as 2-
dimensional “checker-board” grids. The number of grids was varied from 4 to 64
with the number of processors varying from 4 to 32. For the initial tests, the strip
parallel solver was applied to the one-dimensional decompositions. For the one-
dimensional, 4 processor case, each grid contained 24 yz-planes except the last
which had only 23 yz-planes. As the number of domains was increased, the
number of planes per domain diminished from 12/11 xy-planes for the 8 domain
case continuing to 3/2 yz-planes per domain for the 32 domain case. For the
composite-grid solver, the “checker-board” decomposition consisted of 2-
dimensional areal decompositions varying from 4 by 2 domains for the 4
processor case up to 8 by 8 domains for the 32 processor case. The number of
areal cells in each grid varied from about 24 by 48 cells for 8 domains down to 12
by 12 for the 64 domain case. The number of cells was variable since the
95x95x3 grid did not divide evenly. Since a red-black scheme is used for the
composite solver, the number of domains must be twice the number of processors.
Figure 20-9 shows a comparison of results for the strip and composite grid cases
for 4 through 32 processors. Two conclusions can be drawn from this test case.
First, the strip decomposition solver shows a significant degradation as the
number of domains/processors is increased. The number of iterations per Newton
iteration jumps from about 4 to 6.7 in going from four processors to thirty-two for
a seventy percent decrease in efficiency. The other conclusion is that the
composite grid solver, at least for this example, showed little variation in the
number of iterations over the range of processors. Maximum variation was from
4.5 to almost 5 iterations per Newton iteration or about 10%.
Figure 20-9: Effect of Decomposition on Strip and Composite Grid Linear Solvers
Because the flexibility of the composite grid solution technique, it was chose for
use in PARALLEL-VIP for all applications.
The IBM SP2 is a parallel computer built with the MIMD (multiple instruction,
multiple data) architecture and distributed memory. The number of processors for
the SP2 can range to 512. For the SP2 with “Thin-3” nodes used in this study,
memory can be expanded up to 1024 Mbytes. The SGI Origin 2000 machine is a
shared-memory device with maximum memory of 256 billion bytes and up to 128
processors. Peak single-processor performance for the two machines is in the
range of 400 double precision MFLOPS (Million Floating point Operations Per
Second). Actual performance varies considerably for the individual processors.
Typically for reservoir simulation, the SP2 processor delivers about 40-80
MFLOPS compared to about 20-60 MFLOPS for the SGI Origin 2000. That is,
for the problems tested the SP2 was an average of about 1.5 times faster. These
results are preliminary for the Origin 2000; further optimization should improve
performance.
The primary consideration for performance is strongly tied to memory access and
cache misses. Because of communication bandwidth, access to memory may be
relatively slow for all processors relative to peak performance, the ability of the
compiler/system to anticipate memory references and provide data to the cache
can uniquely determine processor performance, regardless of the raw speed of the
processor itself. That is, the data must be available for the processor to use. Since
the SP2 has a delivered performance of about 1.5 times that of the SGI Origin
2000, the overhead of message-passing also becomes dramatically more
important on the IBM machine.
There are generally three forms of message passing for a distributed memory
system: synchronous, asynchronous, and global operations. Synchronous
message passing requires that all messages be completed before computations can
proceed. Asynchronous, on the other hand, allows some overlap of processing
and communications. Global message passing operations distribute data from
single processors to one processor or visa versa. Often some form of arithmetic
function such as summation is performed as part of the global concatenation
operation. These functions are particularly useful for evaluation of reduction
functions such as dot products or for distribution of data from a single node to all
nodes in the parallel partition. Often the global communications have been
optimized for a particular architecture and are much more efficient than a simple
set of send-receive operations for achieving the same results. Because of the
greater raw processor speed and lower communications performance, efficiency
of message-passing is of great importance on the SP2 to achieve good parallel
performance. This implies that utilization of asynchronous message-passing must
be emphasized to overlap compute work with communication overhead.
Asynchronous communications were utilized wherever feasible in the reservoir
simulator. For example, in the coefficient setup portion of the model, intragrid
calculations were overlapped with communications of quantities required to
compute residuals and Jacobian terms for intergrid flow. Synchronous and global
operations were also used for this study in cases where efficiency was not a major
concern because of the length and number of messages involved.
The first test problem is based on the data from the SPE First Comparative
solution problem. Instead of the original dimensions; however, the problem size
was increased to 95x27x3 - 7695 cells. Fourteen wells were used instead of two.
Figure 20-10 depicts the structure of the model and the well locations. This
problem was used to test the simulator for three parallel machines used in this
study (T3D, SP2, and SGI Power Challenge). The model was decomposed along
the x-direction.
Figure 20-10: Grid and Structure for Small Homogeneous Example Problem
almost perfectly load-balanced case for the strip decomposition parallel solver,
i.e., one boundary yz-plane existed between equal large domains.
The second test problem was based on real data from a reservoir associated with a
salt dome. Three-dimensional seismic data were used to condition a geostatictical
model to generate porosity fields. These data were then placed in a geological
model to generate descriptions which were consistent with known geological
horizons and fault data. Finally, permeability fields were generated as functions
of the porosity data yielding variations of several orders of magnitude throughout
the model. A reservoir model was generated with dimensions of 35x35x15 for the
base grid. Two refined cases were then generated from these data resulting in
models of 105x105x15 (3x3 areal refinement) and 245x280x15 (7x8 areal
refinement) for 165,375 and 1,029,000 cells respectively. Black-oil,
compositional, IMPES, and fully-implicit cases were tested in parallel.
Figure 20-11: Grid and Structure for 3D Salt Dome Heterogeneous Model
Sloping Fault Heterogeneous Model. The third test problem upon which the
examples were based also utilized input from seismic, geological, and
geostatistical data. The model was developed by first utilizing a two-dimensional
geostatistical mapping package for defining heterogeneities in forty geological
horizons. An initial grid of 2.4 million cells formed the description. Although it
would be desirable to simulate this entire dataset, upscaling and layer combining
were used to reduce the eventual model to a 64x25x12 - 19200 cells. These cells
were then refined by two in each dimension to increase the overall problem size to
128x50x24 or 153,600 cells (a moderately large model - at least by today’s
standards).
Figure 20-12: Grid and Structure for Sloping Fault Heterogeneous Model
Typical PVT and relative permeability data, black-oil and compositional, for a
middle eastern reservoir were chosen for this problem as well as the Salt Dome
Example.
20.2.8 Results
The following figures show first the results for the total model speedup compared
to the serial case. Next is the speedup of the parallel solver compared to the serial
solver. Finally, the last bar in each series of the graphs represents the ideal
situation, a speedup of n on n processors. The serial model used for comparison
was the current release version of the base serial VIP simulator. The linear solver
used was the best available for the particular situation, generally nested
factorization for a single grid. All comparisons were based on elapsed times, and
not CPU times, to account for any overhead due to message passing.
Homogeneous Black-Oil
Simulations for the homogeneous black-oil case were run on three parallel
platforms: the Cray T3D, the SGI Power Challenge, and the IBM SP2. Results for
the Cray T3D are shown in Figure 20-14 for the cases of 4 to 32 processors. The
speedups used for these comparisons were based on the original serial model
using an unrefined grid. That is, a regular 95x27x3 grid with the same properties
and well data was used for the comparison. These speedups then represent actual
improvement in performance over current serial simulator capabilities. As shown
in Figure 20-14 simulator performance for the T3D was superlinear for four and
eight processors. As the number of processors increased to 16 and 32, the
speedups showed some degradation due to an increase by a factor of three in the
number of linear equation solution iterations. These encouraging results were not
duplicated on the SGI computer as shown in Figure 20-15. Above four processors
there appears to be a significant falloff in performance. Some of this can be
attributed to linear solver degradation similar to the T3D results; however, the
decline in performance for the case of 16 processors indicated that a significant
bottleneck existed due to limited bandwidth to memory. This lack of performance
of the Power Challenge is well known. Fortunately, based on results presented
later in this study, the SGI Origin 2000 appears to have overcome this difficulty.
Results in Figure 20-16 for the SP2 show good speedups with a factor of twelve
for the 16 processor case. Considering that the individual node on the SP2 is
substantially faster than that of the T3D, these results again indicate excellent
The 3D Salt Dome case represents a much more realistic example than the
homogeneous black-oil model. Two test cases were performed with this model
which demonstrated parallel performance for both IMPES and Fully-Implicit
model formulations. Figure 20-17 presents results for the IMPES simulations of a
twenty year case for the IBM SP2. Parallel speeds were superlinear for many of
the cases shown up to 13 processors. The best IMPES result shown in the figure
is a factor of 21 on 25 processors or about 84% parallel efficiency. This represents
about 2.5 microseconds per block step for the total simulation. The imbalance of
the static decomposition for the cases of 8,16, and 32 processors make these
results somewhat less than ideal as anticipated. Figure 20-18 presents the IMPES
results for the SGI Origin 2000 computer for 5-25 processors. Excellent results
were obtained up through 13 processors. The falloff at 25 processors for the
Origin 2000 results is not due to message-passing overhead but a unique feature of
the machine architecture. In an attempt to understand these results, two test cases
using four processors were performed on an eight processor Origin 2000. In the
first case each of the parallel tasks was assigned to a different node board so that
one processor of the two on each node remained idle. In the second case
processor affinity was made such that all four processors on two node boards were
allocated to the simulation. The difference in CPU times was about 30% between
the two cases. Elapsed times were virtually identical to CPU times for both cases.
These results indicate that at least for these cases the bandwidth to memory can
become saturated when more that one task is assigned to a given node board. For
all subsequent cases, similar results are observed as the number of processors
exceeds one-half of the total number of processors available (The total available
for this study was 32 processors.)
Figure 20-17: Results for SP2 3D Salt Dome 105x105x15 IMPES Black-Oil Model
Figure 20-18: Results for Origin 2000 3D Salt Dome 105x105x15 IMPES Black-Oil
Model
The final example for the 105x105x15 3D Salt Dome model used the Fully-
Implicit formulation. As shown in Figure 20-19 speedups were excellent on the
SP2 exceeding ideal for all cases up through 25 processors. For the implicit case
on 25 processors, performance was about 30 microseconds per block step for the
SP2.
Figure 20-19: Results for SP2 3D Salt Dome 105x105x15 Implicit Black-Oil Model
Results for the Sloping Fault 128x50x24 model were also obtained for both
IMPES and Fully-Implicit simulations. Figure 20-20 shows results for the IBM
SP2 for the IMPES case. Performance was excellent on 5 and 10 processors with
some degradation for the 16 and 20 processor cases. The best result was a factor
of 15 on 20 processors for a 75% parallel efficiency. The increase to 32
processors shows a significant reduction in performance primarily due to
message-passing overhead for the solver as indicated on the right of the figure.
Results for the Fully-Implicit cases were substantially better as shown in
Figure 20-21. Superlinear speedups were observed for all cases up to 32
processors with overall speedups approaching 70 on 32 processors. This
somewhat anomalous behavior was also observed on the Origin 2000 as shown in
Figure 20-22. Here the speedups were near 40 on 20 processors. This extreme
parallel performance appeared to be due to poor performance of the serial
simulator as opposed to high parallel efficiency. The poor serial performance was
related to the fact that the first dimension in the problem was a power of two.
References to memory which stride by a power of two can cause significant cache
thrashing. In natural ordering the first dimension, a power of two, is often
referenced as an offset to a given cell. This results in significant performance
degradation for both computers which were investigated. To alleviate the problem
a slightly modified simulation was performed with dimensions of 127x50x24. As
shown in Figure 20-22, the serial performance relative to the parallel performance
improved substantially so that for 20 processors the speedup was about a factor of
18 - a much more realistic figure. Speedups for the Origin 2000 on 32 processors
show the degradation in performance due to memory bandwidth as discussed
above.
Figure 20-20: Results for SP2 Sloping Fault 128x50x24 IMPES Black-Oil Model
Figure 20-21: Results for SP2 Sloping Fault 128x50x24 Implicit Black-Oil Model
Figure 20-22: Results for Origin 2000 Sloping Fault 128x50x24 Implicit Black-Oil
Model
Figure 20-23: Results for Origin 2000 Sloping Fault 127x50x24 Implicit Black-Oil
Model - Cache Effect
The overall dimensions of the previous 3D Salt Dome model were increased by
about a factor of 5 to 245x280x15 or 1,02,900 cells. Figure 20-24 presents results
for 5, 10, and 25 processors. As expected, the larger model size improves parallel
performance such that a factor of 23 on 25 processors was observed. This
corresponded to about 3 microseconds per block step for the IBM SP2.
Figure 20-24: Results for SP2 Large 3D Salt Dome 245x280x15 IMPES Black-Oil
Model
Figure 20-25: Results for SP2 3D Salt Dome 105x105x15 IMPES Compositional
Model
In implicit mode the compositional results are even more dramatic with
superlinear speedups for all cases except 32 processors as shown in Figure 20-26.
The load-imbalance due to static decomposition of the grid again affects the
performance of the 32 processor case.
Figure 20-26: Results for 3D Salt Dome 05x105x15 Implicit Compositional Model
The compositional case discussed in the previous section was increased in size to
1,02,900 cells. Since the serial simulation could not be performed an estimated
speedup of 4 on 4 processors was used as a conservative estimate upon which to
compare the other results. The actual speedups will likely far exceed this estimate
since the smaller compositional case showed a speedup of 5.5 on 4 processors.
Nonetheless, as shown in Figure 20-27 results were again excellent with estimated
speedups of 23 on 25 processors.
Figure 20-27: Results for SP2 Large 3D Salt Dome 245x280x15 IMPES
Compositional Model
The sloping fault model was also run in compositional mode. Speedups for the
IMPES case were much better than the corresponding black-oil IMPES model
since the solver did not dominate the performance. As shown in Figure 20-28
speedups were better than ideal for all cases up through 32 processors with a
factor of 35 on 32 processors. Clearly, the effect of cache thrashing is also
observed in this case but to a lesser extent than in the fully-implicit IMPES
example discussed earlier.
Figure 20-28: Results for SP2 Sloping Fault 27x50x24 IMPES Compositional Model
The realistic compositional case has been shown in previous studies77 to exhibit
significant load imbalance for the PVT calculations. The original grid was
subdivided in the y-direction into four domains with similar numbers of xz-
planes: 8,7,7,7. This resulted in a 200% variation in elapsed times for the PVT
calculations between processors 1-3, and 4. Overall elapsed times varied by a
factor of at least two. (See Figure 20-29 - TOTAL1.) With the linear
decomposition used for this problem static load balancing can be easily
accomplished. Three load balancing steps were taken in which the number of
planes respectively for CPUs 1-4 were changed to 7,7,7,8 (TOTAL2); to 7,7,5,10
(TOTAL3) ; and finally to 7,7,6,9 (TOTAL4). This balancing resulted in a
reduction in the elapsed times for the PVT calculations from 91 to 71 seconds or
about 22%. Overall elapsed times decreased by only about 18% since other
calculations in the simulation were now unequal among processors. Speedups
were improved from 2.91 for the original case to 3.44 for the final balanced
simulation.
Figure 20-29: Grid and Structure for Composition Static Load Balancing Example
The IBM SP2 used for most of the results provided a unique feature to test
performance of the model on a cluster of workstations. Simply by changing a
UNIX environmental variable, the communications of the SP2 are transformed
from the highspeed switch to ethernet speed. This results in an increase in latency
to about 250 microseconds and a decrease in bandwidth to about 1 million bytes
per second. These are comparable to an idealized cluster of workstations
communicating over ethernet.
Figure 20-30 shows results for the Sloping Fault example. (SP2-IP refers to
“internet protocol” or ethernet mode while SP2-US refers to “user space” or
highspeed switch mode on the SP2.) For four processors, the results are excellent
with speedups approaching ideal. As the number of processors are increased to
eight and eleven, the performance degrades significantly probably due to
saturation of the network and the significant latency with increased numbers of
messages. Performance for these latter cases was actually worse than that for four
processors. As shown in Figure 20-30, message-passing overhead significantly
affects solver performance for greater than four processors.
20.3 Conclusions
This work has presented a general-purpose parallel reservoir simulator which
utilizes local grid refinement as a natural and flexible framework. The parallel
model is based on the use of the portable MPI communications package. The
message-passing scheme in the model utilizes send-receive buffers and
asynchronous communication for minimization of overhead. Two parallel linear
equation solvers were developed using strip and composite grid decompositions.
Standalone results for the solvers showed excellent parallel speedups on an
idealized basis and compared to an efficient serial solver for greater than four
million nodes and up to seventy-one processors. The use of local grid refinement
allows for a variety of preconditioners to be utilized in the parallel linear equation
solution. Results for the overall simulator were good to excellent for a variety of
realistic simulations based on geophysical and geological data. A variety of test
problems were simulated in parallel ranging from black-oil to large compositional
cases with more than one million cells using both IMPES and fully-implicit
formulations. A static load balancing scheme for a compositional simulation was
21
Phase Equilibrium Calculations
00000
21.1 Introduction
VIP-EXECUTIVE is based on an assumption of local phase equilibrium. It is a
two-hydrocarbon phase simulator and allows no more than two hydrocarbon
phases to be present in a gridblock. Phase equilibrium between oil and gas is
described by the following system of equations:
x i φ oi = y i φ gi , i = 1, 2, …, n c (21-1)
z i = Lx i + ( 1 – L )y i , i = 1, 2, …, n c (21-2)
nc
∑ xi = 1 (21-3)
i=1
nc
∑ yi = 1 (21-4)
i=1
Here, φoi, φgi are fugacity coefficients of oil and gas, respectively. The fugacity
coefficients are functions of pressure, temperature, and composition of the
corresponding phase. For an expression of the fugacity coefficients, refer to
Chapter 23, PVT Representation.
more robust method be required. It is always more efficient to solve Equations 21-
1 through 21-4 in one iteration loop with the other simulator equations.
Equations 21-1 through 21-4 become a part of the simulator governing equations
only if the hydrocarbon mixture in a gridblock splits into two phases. If the
mixture exists as one phase, there are no phase equilibrium constraints for this
block, and Equations 21-1 through 21-4 are not part of the governing equations.
Therefore, it is important to know whether the mixture in a gridblock is a one- or
two-phase system. Determination of the number of phases in a block must be
performed on each outer iteration because it affects the governing equations and
the form of the Jacobian matrix.
Equations 21-1 through 21-4 are a set of nonlinear algebraic equations. For some
hydrocarbon mixtures, especially near the mixture critical point, the radius of
convergence of the Newton-Raphson iterative procedure for Equations 21-1
through 21-4 may be small. In that case, for the iterative procedure to converge,
one must provide an initial guess, which should be close to the solution. This may
not be feasible, and the simulator will have a problem with convergence in the
outer iteration loop. In this case, even when Equations 21-1 through 21-4 are
solved with the other governing equations in one iteration loop, convergence of
the outer iterations may be improved if at the end of an outer iteration — after
block pressure p and block mixture composition zi are updated — the system of
Equations 21-1 through 21-4 is solved separately and the new oil and gas
compositions, xi, yi, and the liquid fraction, L, are determined.
VIP-EXECUTIVE provides two options. The first option can be used for simple
compositional problems that do not develop conditions close to the mixture
critical point. The second option, the Gibbs option, is designed for more difficult
problems with near-critical point phase behavior. This is recommended for
miscible displacement problems. The second option is based on a combination of
a phase stability test and Gibbs energy minimization. The Gibbs option is
computationally intensive. Benchmarking shows that for simple compositional
problems when both options perform well, the second option results in a penalty
of about 30% in CPU time as compared to the first option. For miscible
displacement problems, however, the Gibbs option is more robust and could be
faster if it reduces the number of iterations.
VIP-EXECUTIVE provides two ways to solve Equations 21-1 through 21-4 (flash
calculations). The two methods are:
Despite the fact that the accelerated successive substitution method is the default,
we recommend use of the Newton-Raphson method because it is faster.
z i φ oi ( p, z ) = y i φ gi ( p, y ), i = 1, 2, …, n c (21-5)
nc
∑ yi = 1 (21-6)
i=1
for the gas composition yi, i = 1, ... , nc, and the bubble pressure, p. A similar
system of equations is solved in the dew-point routine:
z i φ gi ( p, z ) = x i φ oi ( p, x ), i = 1, 2, …, n (21-7)
nc
∑ xi = 1 (21-8)
i=1
nc
G1 = ∑ Ni μi ( p, T, z ) (21-9)
i=1
where Ni and μi are the number of moles and chemical potential of component i in
the mixture, respectively. The chemical potential and fugacity coefficient are
related as follows:
μ i = RT ln ( z i φ i ) + λ ( P, T ) (21-10)
That distribution of molecules between oil and gas phases which results in the
lowest value of the Gibbs energy (G2) gives the equilibrium state of the two-phase
system. To find the equilibrium state, the program minimizes the expression
(Equation 21-11) with respect to Noi and Ngi. However, to do that it must know
that the mixture exists as a two-phase system.
The Gibbs energy of the system formed by splitting off some small amount of
material from the original single-phase system to form the second phase can be
approximated as follows:
nc
G 2 ≈ G 1 + N ε ∑ x i [ μ i ( P, T, x ) – μ i ( P, T, z ) ] (21-12)
i=1
Here, Nε is the number of moles and x is the composition vector of the split-off
phase, while z is the composition vector of the original mixture.
If the function
nc
F(x) = ∑ xi [ μi ( P, T, x ) – μi ( P, T, z ) ] (21-13)
i=1
is positive for any composition vector x, then according to Equation 21-12, the
original mixture is stable and exists as a single-phase fluid. However, if there is a
composition vector x, for which F < 0 , then the original phase is unstable, and a
second phase is formed. Hence, the problem of phase stability also is formulated
as a minimization problem for the function defined by Equation 21-13. This
function is called the phase stability function.
∑ ( xi – yi )xi
i=1
M = -----------------------------------
- (21-14)
n c
2
∑ xi
1
If the value of M satisfies
then the block is considered near-critical and is sent through the phase equilibrium
calculations. If not, the calculations for this block are bypassed. The value of tskip
for the test is defined by the user on the GIBBS card. The larger the value of tskip
that is specified by the user, the more two-phase blocks are sent through phase
equilibrium calculations.
Also, if the mixture in a single-phase block is far from the mixture two-phase
region, then the block most likely remains single phase even if block pressure
decreases slightly or if the mixture composition changes even more than is
allowed in the previous check. To identify these blocks, the simulator relies on the
values of tolerances provided by the user on the GIBBS card in the recurrent data
set.
22
Polymer Modeling Option1
22.1 Introduction
The Polymer option extends VIP-EXECUTIVE to simulate the injection of
polymer thickened aqueous solutions into reservoirs. VIP-POLYMER™ models
all the important physical properties attributed to the flow of polymer solutions
through porous media. A list of the physical properties included in the model
follows:
■ Cation exchange between the clay (rock) and the aqueous phase.
∂ ϕ Q
( ϕC̃ p ) + ------ ∇ ( C pw U w ) = ------p (22-1)
∂t ϕp Vb
U w = kλ rw ( ∇P w – γ w ∇D ) (22-2)
C̃ p = C pw S w + Ĉ p (22-3)
∂ Q
( ϕC̃ e ) + ∇ ( C ew U w ) = ------e (22-4)
∂t Vb
C̃ e = C ew S w + Ĉ e (22-5)
subscript e stands for either the divalent cations or the monovalent anions where
∇ = Gradient operator
ϕ = Porosity
μo – μ∞
μ p = μ ∞ + ------------------------------------
· Pα – 1- (22-6)
γ
1 + ⎛⎝ ---------
· -
⎞
γ 1 ⁄ 2⎠
where:
·
μo Is the viscosity at zero shear rate ( γ = 0 ) and is a
function of polymer concentration and solution salinity
μo = μo(Cpw, CSEP).
·
μ∞ Is the viscosity at infinite shear rate ( γ → ∞ ) and is
approximated by μ∞ = μw.
Pα Is a parameter.
·
The equivalent shear rate, γ , is approximated by Hong66 as follows:
·
· 5γ c U w
γ = ------------------------
- (22-7)
kk rw ϕS w
where:
·
γ = Shear rate (sec-1)
n -
-----------
· ⎛ 1 + 3n ⎞ –1
n
·
γc = ⎜ ------------------⎟ where n is the power law index. γ c is
⎝ 4n ⎠
bounded between 0.779 and 1 as n varies between 1 and
0.
The polymer viscosity at zero shear rate, μo, is a function of polymer and
electrolyte concentrations. VIP-POLYMER has two options for calculating μo.
One option involves interpolation of μo vs. Cpw tables at various salinities. The
other option uses the following function:
2 3 S
μ o = μ w 1 + ( A p1 C pw + A p2 C pw + A p3 C pw )C SEP
p
(22-8)
where:
+ ++
C + βp C
C SEP = ---------------------------
- (22-9)
Sw
where:
Ĉ p ( lb ⁄ ac – ft )ρst
Ĉ p ( ppm ) = -------------------------------------------- (22-10)
ϕρ res 2.7194
a p – C pw
Ĉ p = ------------------------ (22-11)
1 + b p C pw
a p = a p1 + a p2 C SEP (22-12)
where:
⎛ lb ⁄ ac – ft⎞
ap1 = Parameter ⎜ ------------------------⎟
⎝ ppm ⎠
⎛ lb ⁄ ac – ft ⎞
ap2 = Parameter ⎜ --------------------------------------⎟
⎝ ppm – meq ⁄ ml⎠
bp = Parameter (1/ppm)
VIP-POLYMER has two options for calculating Rk. One option uses the following
functions:
( R kmax – 1 )b rk C pw
R k = 1 + --------------------------------------------
- (22-13)
1 + b rk C pw
1⁄3 –4
C rk [ μ ]
R kmax = 1 – -----------------------------------------
- (22-14)
1⁄2 1⁄2
(k) ⁄ ( ϕS w )
where:
[μ] =
lim ⎛μ o – μ w⎞
-----------------
- (22-15)
C pw → 0 ⎝ μ w C pw ⎠
The method used by VIP-POLYMER to calculate [μ] depends on the option used
for calculating μo. If μo is calculated using Equation 22-8, then
S
[ μ ] = A p1 C SEP
p
(22-16)
( Δμ o ⁄ ΔC pw )
[ μ ] = -------------------------------
- C pw = 0 (22-17)
μw
+ 2 + 2
(C ) C Qv ( C )
-------------- = β ------ -------------
- (22-18)
++ S w C ++
C
where
+ ++ -
C +C = C (22-19)
+ ++
C +C = Qv (22-20)
M
μ p = μ w + -------------------- (22-22)
S x
1 + ⎛⎝ ---⎞⎠
A
where:
· qw
5γ c
S = --------------------------- ---------
· - (22-23)
Kk rw ϕS w γ 1 ⁄ 2
x = pα – 1
M = μo – μw
Kk rw A dp
q w = ---------------- ------ (22-24)
μ dr
qw M
dp = – ------------------------- μ w + -----------------------------
- dr (22-25)
2πrhKk rw x
⎛ S ⎞
1 + ⎜ ------------⎟
⎝ 2πhr⎠
x
qw M ⎛S ⎞ x
p = – ------------------------- u w ln ( r ) + ----- ln ⎜ ----------⎟ + r (22-26)
2πrhKk rw x ⎝ 2πh⎠
Define:
–qw μ re
P e – P w = ---------------------- ln ⎛⎝ -----⎞⎠ (22-27)
2πhKk rw r w
x
⎛S ⎞ x
⎜ ----------⎟ + r e
–qw re M ⎝ 2πh⎠
Pe – pw = ---------------------- μ w ln ⎛ -----⎞ + ----- ln ----------------------------- (22-28)
2πhKk rw ⎝ r w⎠ x x
⎛S ⎞ x
⎜ ----------⎟ + r w
⎝ 2πh⎠
For these two fluids to give the same pressure drop, Pe - Pw , then
x
⎛S ⎞ x
⎜ ----------⎟ + r e
⎝ 2πh ⎠
ln -----------------------------
x
⎛S ⎞ x
⎜ ----------⎟ + r w
M ⎝ 2πh ⎠
μ = μ w + ----- ---------------------------------------- (22-29)
x re
ln ⎛ -----⎞
⎝ r w⎠
Thus all well bore calculations for a non-Newtonian viscosity fluid can be made
equivalent to a Newtonian fluid by use of an equivalent viscosity μ as defined by
Equation 22-29.
The instantaneous gel model uses the polymer model to inject and place the
polymer in the formation. The GEL keyword at the desired date in the recurrent
data will compute a gel saturation assuming an instantaneous reaction as follows:
Sw ( if cpw ≥ cpgel ) -
Sgel = --------------------------------------------------------
0.0 ( if ( cpw < cpgel ) )
where
The gel phase has the same compressibility as the aqueous phase. The water
relative permeability is calculated as follows:
The instantaneous gel model is very general and independent of the polymer
gelation mechanism.
23
PVT Representation
23.1 Introduction
In this chapter, the behavior of the fluid-rock system when pressure and
temperature are varied is discussed. Attention is mainly focused on the behavior
of the hydrocarbon properties in both the compositional formulation and black oil
formulation. Reference is made to the miscible formulation.
φ = M V φ o [ 1 + C r ( P o – P ref ) ] (23-1)
B w = B wo [ 1 + C w ( P o – P ref ) ] (23-2)
where Bwo is the reciprocal formation volume factor of water (in STB/RB) at
reference pressure Pref, and Cw is water phase compressibility.
For the general compositional model, an equation of state is used to determine the
oil densities and fugacities. The user may select one of four equations-of-state:
Peng-Robinson (PR), Redlich-Kwong (RK), Soave-Redlich-Kwong (SRK), or
Zudkevitch-Joffe-Redlich-Kwong (ZJRK).
All four equations of state obtain molar densities by solving the following cubic
equation:
3 2 2 2
Z j + ( E f1 B – 1 ) Z j + ( A – E f2 B + E f3 B )Z j – [ AB + E f4 B ( B + 1 ) ] = 0
(23-3)
where
Po
Z j = -----------
- (23-4)
ρ j RT
is the compressibility factor for the RK, ZJRK, SRK, and “unshifted” PR
equations, and
Nc
⎛ P o ⎞ ⎛⎜ 1 ⎞
Z j = ⎜ ---------⎟ ---- + ∑ x i b i s i⎟ (23-5)
⎝ RT ⎠ ⎜⎝ ρ j ⎟
⎠
i=1
RT ci
b i = Ω bi ----------- (23-9)
P ci
The symbol dik denotes the binary interaction coefficients between components i
and k. Definitions for those parameters not mentioned here are given later in this
section.
The cubic equation (Equation 23-3) may yield three real roots. The largest root is
always taken as the compressibility factor for the gas phase, while the smallest
positive root is the compressibility factor for the oil phase. The molar densities of
the hydrocarbon phases, ρj, then are calculated from Equation 23-4 or
Equation 23-5.
f ji
φ ji = ---------
- (23-10)
xi Po
where fji is fugacity of component i in phase j. This represents the tendency of the
component to escape from the phase. In an ideal solution it is equal to the vapor
pressure.
i o bP j Z +E B A
m2
ln φ ji = – ln ( Z j – B ) + ----------- ( Z j – 1 ) + ln ------------------------ -----------
BRT Z j + E m1 B E f5 B
Nc
2P o bi Po
× ---------------
AR T
2 2
- ∑ ( x k ( 1 – d ik ) a i a k ) – ----------
BRT
- (23-11)
k=1
The universal constants Ef1 to Ef5, Em1, Em2, and the default values of Ωai and Ωbi
are summarized in the following:
PR 1 2 -3 -1 2 2 1 + 2 1 – 2 0.457235529 0.077796074
RK 0 1 -1 0 -1 0 1 0.427480200 0.086640350
Peng-Robinson Equation-of-State
2
α i = [ 1 + χ i ( 1 – T ri ) ] (23-12)
or
The symbol ωi is the user-input acentric factor of component i, and Tri is the
reduced temperature (T/Tci) of component i.
Redlich-Kwong Equation-of-State
T ci
αi = ------- (23-16)
T
Soave-Redlich-Kwong Equation-of-State
Zudkevitch-Joffee-Redlich-Kwong Equation-of-State
αi = 1 (23-18)
The Ωai and Ωbi in the table above are initial guesses. The actual values of Ωai and
Ωbi for a given component i at a given temperature T below critical are determined
so that component vapor pressure and saturated liquid density at T are exactly
matched.
Lohrenz-Bray-Clark Correlation
The viscosities of the oil and gas phases are calculated using the Lohrenz-Bray-
Clark (LBC) correlation.28
3 4 0.16667
⎛ M j P cj⎞
μ j = 0.18383 ⎜ ---------------⎟ [ ( 0.1023 + ρ r {0.023364 + ρ r [ 0.058533
⎝ T cj ⎠
4 –4 U3
+ ρ r ( – 0.040758 + 0.0093324ρ r ) ] } ) – 10 ] + ----------------------------
N
-
c
∑ ( xi Mi )
1 (23-19)
where
ρ r = ρ j V cj (23-20)
Nc
Aj = ∑ xi Ai , A = M, P c, T c, V c (23-21)
i=1
10.736 Z ci T ci
V ci = --------------------------------
- (23-22)
5.6146 P ci
Nc 3 4 0.16667
⎛ M i P ci⎞
U 3 = 0.18383 ∑ x i M i ⎜ ---------------⎟ U 5i (23-23)
⎝ T ci ⎠
i=1
and
– 4 0.94
U 5i = 3.4 × 10 T ri if T ri < 1.5 (23-24)
or
–4 0.625
U 5i = 1.778 × 10 ( 4.58T ri – 1.67 ) if T ri ≥ 1.5 (23-25)
Pedersen Correlation
The Pedersen correlation29,37 is useful for heavy oils where the LBC correlation
fails to give a proper viscosity prediction for some cases.
To invoke the Pedersen viscosity option, enter keyword VISPE between the PVT
data and the surface separation data input. You may specify binary interacting
coefficients Xij in Equation 23-27 to calculate the mixture critical temperature and
k1 to k7 in Equation 23-39 to calculate the viscosity of the reference component
(methane).
In the input data, VISK (kj) and VISKJ (Xij) cards are optional. Default values are
used if they are not entered. The default values are zero for all of the interacting
coefficients Xij and are given after Equation 23-39 for coefficients kj. Only those
data overwriting the default values need to be entered. Sample input data are
shown below:
Sample:
VISPE
VISK
1 42.5
2 0.0002279
3 11770
4 600.3
5 16.49
6 1614
7 0.05552
VISKJ
3 5 -0.36409
3 6 -0.38517
3 7 -0.8
3 8 -0.62763
ENDVIS
such a case, only comprehensive viscosity data for one component (the reference
component) in the group are needed. Other data can be calculated from the
reduced viscosity. The Pedersen correlation uses methane as a reference
substance.
(23-26)
where Tcm is the mixture critical temperature and Pcm is the mixture critical
pressure. The subscripts o and m refer to the reference substance and mixture,
respectively. They are calculated from the mixing rules:
n n 1⁄3 1⁄3 3
⎛ T ci⎞ ⎛ T cj⎞ 1⁄2
∑ ∑ x i x j ⎜ ------
⎝ P ci⎠
- ⎟ + ⎜ -------⎟
⎝ P cj⎠
( 1 – X ij ) ( T ci T cj )
T cm = i-------------------------------------------------------------------------------------------------------------------------------------
=1 j=1
n n
- (23-27)
1⁄3 1⁄3 3
⎛ T ci⎞ ⎛ T cj⎞
∑ ∑ x i x j ⎜ ------
⎝ P ci⎠
- ⎟ + ⎜ -------⎟
⎝ P cj⎠
i=1 j=1
n n 1⁄3 1⁄3 3
⎛ T ci⎞ ⎛ T cj⎞ 1⁄2
8 ∑ ∑ x i x j ⎜ -------⎟ + ⎜ -------⎟ ( T ci T cj )
⎝ P ci⎠ ⎝ P cj⎠
P cm i=1 j=1
= -------------------------------------------------------------------------------------------------------------------- (23-28)
2
⎛ n n
⎛ T ⎞
1⁄3
⎛ T ⎞
1 ⁄ 3 3⎞
⎜ ci cj ⎟
⎜ ∑ ∑ x i x j ⎜⎝ ⎟⎠
------- + ⎜ -------⎟
⎝ P cj⎠ ⎟
⎝i = 1 j = 1 P ci ⎠
The coefficients Xij are introduced in Equation 23-27 to account for the binary
component interaction.
Here, MWm is the mixture molecular weight and is calculated by the following
equation:
–4 2.303 2.303
MW m = 1.304 × 10 ( MW w – MW n ) (23-29)
where MWw and MW n are the weighted average and the molar average molecular
weights:
n
2
∑ xi MWi
MW w = i--------------------------
=1
n
(23-30)
∑ xi MWi
MW n = ∑ xi MWi (23-31)
i=1
Parameters αm and αo in Equation 23-26 account for the density and molecular
weight effects:
– 3 1.847 0.5173
α m = 1 + 7.378 × 10 ρ r MW m (23-32)
– 3 1.847 0.5173
α o = 1 + 7.378 × 10 ρ r MW o (23-33)
The symbol ρr denotes the reduced density for the reference component:
⎛ P ⋅ P co T ⋅ T co⎞
ρ o ⎜ ----------------, ---------------⎟
⎝ P cm T cm ⎠
ρ r = ----------------------------------------------
- (23-34)
ρ co
The density of the reference component (methane) ρo is calculated from the three-
parameter Peng-Robinson equation of state.
The viscosity of the reference component is based on the model of Hanley et al.30
where
–1 –2 ⁄ 3 –1 ⁄ 3 1⁄3 2⁄3
ηo ( T ) = b1 T + b2 T + b3 T + b4 + b5 T + b6 T
4⁄3 5⁄3
+ b7 T + b8 T + b9 T (23-36)
and
b1 = -0.2090975x105
b2 = 2.647269x105
b3 = -1.472818x105
b4 = 4.716740x104
b5 = -9.491872x103
b6 = 1.219979x103
b7 = -9.627993x10
b8 = 4.274152
b9 = -8.14153x10-2
T 2
η 1 ( T ) = 1.6969859 – 0.13337235 ⎛ 1.4 – log e ---------⎞ (23-37)
⎝ 168⎠
j
Δη′ ( ρ o, T ) = exp ⎛ – j 1 + ---4-⎞
⎝ T⎠
⎧ ⎫
⎪ 0.1 ⎛ j3 ⎞ 0.5 ⎛ j j ⎞ ⎪
-⎟ + θρ o ⎜ j 5 + ---6- + ---7-⎟
× ⎨ exp ρ o ⎜ j 2 – ---------- – 1.0 ⎬ (23-38)
3⁄2 T T⎠
⎪ ⎝ T ⎠ ⎝ ⎪
⎩ ⎭
j1 = 1.035060586x10
j2 = 1.7571599671x10
j3 = 3.0193918656x103
j4 = 1.8873011594x102
j5 = 4.2903609488x10-2
j6 = 1.4529023444x102
j7 = 6.1276818706x103
k
Δη″ ( ρ o, T ) = exp ⎛ – k 1 + ----4-⎞
⎝ T⎠
⎧ 0.1 ⎛ k3 ⎞ 0.5 ⎛ k k ⎞ ⎫
-⎟ + θρ o ⎜ j 5 + ----6- + -----7 ⎟
× ⎨ exp ρ o ⎜ k 2 – ---------- – 1.0 ⎬ (23-39)
3⁄2 T T 2⎠
⎩ ⎝ T ⎠ ⎝ ⎭
k1 = 9.74602
k2 = 18.0834
k3 = 4126.66
k4 = 44.6055
k5 = 0.976544
k6 = 81.8134
k7 = 15649.9
ρ o – ρ co
θ = ------------------- (23-40)
ρ co
F 1 = 1-------------------------
+ HTAN (23-41)
2
F 2 = 1------------------------
– HTAN-
(23-42)
2
exp ( ΔT ) – exp ( – ΔT )
HTAN = ---------------------------------------------------- (23-43)
exp ( ΔT ) + exp ( – ΔT )
ΔT = T – T F (23-44)
where TF is the freezing point of methane. The units in Equations 23-35 through
23-43 are gm/cc for density, °K for temperature, and micropoise for viscosity.
Since the simulator uses a compositional formulation internally, these black oil
data are converted into compositional parameters (e.g., fugacity coefficients, Zo,
and μo are as functions of pressure and gas component mole fraction in the oil
phase). The compositional parameters then are reconstructed into equal-spaced
tables. For pressures beyond the range of the tabular data, VIP-EXECUTIVE uses
constant slope extrapolation to derive values at those conditions. The conversion
procedure is described in the following sections.
Gas Properties
The gas phase viscosity and compressibility factor are assumed to be a function of
pressure only (no compositional dependency). An equal-space viscosity versus
pressure table is constructed from the input viscosity versus saturation pressure
table. A Zg versus pressure table is likewise constructed. If Bg instead of Zg is
entered as a function of pressure, then the following equation is used to convert Bg
to Zg:
0.0056146 B g P o T s
Z g = ------------------------------------------------
- (23-45)
Ps T
In deriving this equation, gas compressibility at the surface temperature (Ts) and
surface pressure (Ps) is assumed to be one.
VIP-EXECUTIVE allows the user to input more than one set of PVT data for
different equilibrium regions. These data are converted to equal-space tables of K-
values (needed for fugacity equality equations), Zo, and μo.
Data conversion requires oil and gas component molecular weights, which are set
internally by the simulator. To ensure non-zero component K-values, the
simulator assumes that the oil component molecular weight, M2, equals the
maximum among all residual oil and stock tank oil molecular weights plus two
(2.0). Additionally, the gas component molecular weight, M1, is assumed to equal
the minimum among all gas phase molecular weights (29Gr) minus one (1.0).
Consequently, the gas phase has at least a trace of the oil component at any
pressure level (including the surface pressure) and vice versa.
VIP-EXECUTIVE also provides options for multiple API fluid regions and gas
solubilities. These options are not documented here.
The residual oil molecular weight is either directly entered by the user or derived
by the simulator. Normally the user is required to enter either the density and
molecular weight of residual oil (ρor , Mwor) or those of bubble-point oil (ρos,
Mwos) following the BOTAB card. These properties can be derived from each
other based on the differential liberation data. The procedure to derive ρor and
Mwor from ρos and Mwos is explained below.
ρ or = ρ os × B os – G g (23-47)
where ρor and ρos are given in gm/cc, Ns is the entry number of the saturation
pressure of the bubble-point oil, and Bos is the Bo (RB/STB of residual oil) for the
bubble-point oil. Again, based on 1 cc of the residual oil, the overall gas liberated
in gmole at the end of the differential liberation for the bubble-point oil is
1 Ps
G m = ------------------------------------------ --------- [ R s ( N s ) – R s ( N p ) ] (23-48)
5.6146 × 652.428 RT s
B os × ρ os
z o = ---------------------
- (23-49)
M wos
ρ or
M wor = -----------------
-
zo – Gm
Following a similar procedure, the density and molecular weight of the stock tank
oil can be derived from ρos and Mwos if the Rs, Bo (RB/STB of stock tank oil), and
Gr were known from the separator flash test. This is the case when the SEPTEST
card is entered in the VIP-EXECUTIVE input deck.
The second option is the use of the SEPARATOR card, where K-values for the
separator are entered. The stock tank oil molecular weight can be defined from the
separator K-values. However, the user may enter the stock tank oil molecular
weight (MWL) which, if entered, produces an overly defined fluid system. In this
case, VIP-EXECUTIVE honors the user-entered MWL and disregards the
molecular weight derived from the separator K-values in deriving the stock tank
oil compressibility (ZLSEP).
The K-value and oil compressibility tables can be derived from the user-entered
differential liberation data. First, the molecular weight of the liberated gas, Mwg,
at each liberation stage (pressure entry No. I-1 to I) is
M wg ( I ) = 29 ⋅ G r ( I ) (23-50)
The gas component mole fraction in the liberated gas phase is then
M 2 – M wg ( I )
y 1 ( I ) = ------------------------------
- (23-51)
M2 – M1
Based on 1 cc of residual oil, the amounts of the overall gas liberated from stage I
to the standard conditions in gm is
Np – 1
29. Ps
G g ( I ) = --------------------------------------- ---------
5.6146 × 62.428 RT s ∑ ( R s ( i ) – R s ( i + 1 ) )G r ( i + 1 )
i=1
(23-52)
which is similar to Equation 23-46. Thus, the density of the oil at stage I is
ρ or + G g ( I )
ρ o ( I ) = ---------------------------
- (23-53)
Bo ( I )
Since the amounts of total gas liberated from stage I to the standard conditions are
given by
1 Ps
G m ( I ) = --------------------------------------- --------- [ R s ( I ) – R s ( N p ) ] (23-54)
5.6146 × 62.428 RT s
ρo ( I ) × Bo ( I )
M wo ( I ) = --------------------------------------------- (23-55)
ρ or ⁄ M wor + G m ( I )
M 2 – M wo ( I )
x 1 ( I ) = ------------------------------
- (23-56)
M2 – M1
The equilibrium K-values of the gas component and the oil component at stage I
are calculated according to
y1 ( I )
K 1 ( I ) = -----------
- (23-57)
x1 ( I )
and
1 – y1 ( I )
K 2 ( I ) = --------------------
- (23-58)
1 – x1 ( I )
P ( I ) × M wo ( I )
Z o ( I ) = -----------------------------------------------------------------------
- (23-59)
5.6146 × 62.428 × ρ o ( I ) × RT
For the undersaturated oil phase, the user must enter ratios of the oil formation
volume factor at pressures above the saturation pressure to the oil formation
volume factor at the saturation pressure. Ratios of oil viscosity at pressures above
the saturation pressure to viscosity at the saturation pressure also are required. To
convert the undersaturated Bo values to undersaturated Zo values, the following
relationship is recognized from Equation 23-59:
⎛ Z o⎞
⎜ ------⎟
⎝ P o⎠ Bo ( Po )
----------- = ----------------
- (23-60)
s s
⎛ Z o⎞ Bo ( Po )
⎜ -----s-⎟
⎝ P o⎠
where the superscripts denote saturated conditions. Note that the right side of
Equation 23-60 represents the input ratios of formation volume factors. This
equation suggests that the same ratios can be used directly if the undersaturated
Zo/Po table is constructed. The independent variables in the two-dimensional table
are x1 and Po. For a given x1, the saturation pressure is given by the x1-pressure
relationship established earlier in this section. For any Po which is greater than the
saturation pressure, the right side of Equation 23-60 is determined from the input
undersaturated table through interpolation. The resulting value is then multiplied
s s
by Z o ⁄ P o to yield Zo/Po for x1 and Po. The two-dimensional oil viscosity table
μo(x1, Po) is constructed likewise.
Finally, the K-values are used to establish the following fugactiy equality
equations:
φ oi = K i ( P o ), i = 1, 2 (23-61)
φ gi = 1, i = 1, 2 (23-62)
After the k-values have been calculated for each of the data entries in the
differential liberation data, an equal-spaced table is constructed through a linear
interpolation process involving ln(KijPi) versus ln(Pi).
In the second option, the solvent is treated numerically as partially soluble in the
oil phase. The user must enter the solvent K-value, the compressibility factor of
the solvent in the oil phase, the viscosity of solvent in the oil phase as a function
of pressure, and the two properties specified in the first option. The oil phase and
gas properties are then calculated through mixing rules discussed in Chapter 16.
23.5.1 Introduction
The equation of state interpolation option is developed in VIP for a reduction of
CPU time required for compositional reservoir simulations. The new option
reduces the CPU time by factors of 1.7-8 and it matches results of fully
compositional simulations in tested compositional models. The effectiveness of
the new option has been confirmed in the Fifth SPE Comparative Model and
several full field compositional models.
The EOS interpolation option can be used for the phase-equilibrium calculations
in
■ separators, and
It also can be applied in any combination of these three parts of the reservoir
simulation. For example, it can be used for the phase equilibrium calculations
only in separators and well tubing strings.
The new EOSINT option significantly reduces the CPU time required for the
phase-equilibrium calculations and linearization of the equation of state. It is
implemented only in the IMPES finite-difference formulation.
Phase-Equilibrium Calculations
assuming that
■ pressure,
■ temperature, and
are known.
General Description
■ pressure,
■ temperature, and
■ fluid compositions.
The equation of state is applied for the generation of these tables in VIP-CORE.
The user should define pressure, temperature, composition entries of the EOS
interpolation tables. The recovery and compressibility factors are calculated for
all combinations of pressure, temperature, and composition entries.
The generated lookup tables are used instead of EOS for the phase-equilibrium
calculation in VIP-CORE and VIP-EXECUTIVE. A multidimensional
interpolation is applied for the determination of the recovery and compressibility
factors between the table entries.
Parameters of the equation of state must be formulated because the EOS is applied
for the generation of the EOS interpolation tables.
The EOSINT card should be included in VIP-CORE input deck to invoke the EOS
interpolation option. The NORESERVOIR, NOSEPARATOR, or/and
NONETWORK keywords can be input on the EOSINT card to exclude the
application of the EOS interpolation procedure for reservoir grid cells, separators,
or/and surface pipeline network system, respectively.
Interpolation Tables
Temperature entries (T1, T2, ..., Tnt) in the EOS interpolation tables can be input
on the TEMPERATURE card, where nt is the number of the temperature entries.
Composition entries of the EOS interpolation tables are defined from three
sources:
■ user input,
Composition entries (Z1, Z2,..., Znz) of the EOS interpolation tables can be input
after the CMP card, where Zj=(z1j,z2j,...zncj), j= 1, 2,...,nz are the composition
vectors; zij is the molar fraction of the i-th component in the j-th composition
vector; nz is the number of the composition entries; nc is the number of the
hydrocarbon components.
Initial compositions of the reservoir fluids defined after the OILMF, GASMF, or
COMPOSITION cards are automatically included in a list of the composition
entries.
The user can divide the composition entries into different paths. The simulator
interpolates the recovery and compressibility factors between paths. The paths can
be defined in the CMP, SWELLTEST, DIFEXPTEST, and MULCONTEST cards.
A pressure range in the EOS interpolation tables must include calculated values of
pressure in reservoir grid blocks, well tubing strings, and the surface pipeline
network system. If this condition is violated, the simulator outputs a warning
message or stops the run (if it is requested by the user using the STOP keyword on
the EOSINT card).
The pressure interval from the saturation pressure PSATjk to the maximum
pressure PMAX is divided into NPMAX-1 equal subintervals. End points of
these subintervals are considered as the pressure entries of the EOS
interpolation tables.
The number of the pressure entries NPMAX can be defined by the user in the
DIM card.
3. A much smaller pressure increment is used in the saturated region near the
saturation pressure than the pressure increment in a saturated region far below
the saturation pressure. The pressure increments in these two saturated
regions with liquid and vapor hydrocarbon phases are selected as follows:
The pressure interval from the minimum pressure PMIN to PSATjk is divided
into two intervals. The length of the first interval (near the saturation pressure)
is selected as 0.02 * (PMIN - PSATjk). Both intervals are divided into
NPMAX / 2 subintervals. End points of these subintervals are considered as
the pressure entries of the EOS interpolation tables.
Therefore, the simulator determines the 2 * NPMAX pressure entries for each
combination of the temperature and composition entries. Different sets of the
pressure entries are selected for the different composition and temperature entries.
Simulations of
■ swelling,
■ differential expansion,
PVT tests can be applied in VIP-CORE for the automatic generation of the
composition entries of the EOS interpolation tables.
The PVT test simulations can be invoked using the SWELLTEST, DIFEXPTEST,
and/or MULCONTEST cards.
Swelling Test
For the simulation of a swelling PVT test, the user must specify
The number of gas fractions ns can be specified on the NGASFR card. In this
case, the nf gas fractions are automatically calculated by dividing the interval 0 to
1 into (nf+1) equal subintervals.
Z j = ( 1 – f j ) × X + f j × Y, j = 1, 2, …, nf. (23-63)
These compositions are included in the list of the composition entries and are
output to a file (Fortran Unit 77).
For the simulation of a differential expansion test, the user must specify
The number of stages np in the differential expansion test can be specified on the
NPRES card. In this case, the np pressure values in these stages are automatically
calculated by dividing the interval from PMIN to PSAT into np equal subintervals,
where PSAT is the saturation pressure (bubble point or dew point pressure),
calculated by the simulator for the specified composition and temperature.
In the first stage, the composition is defined on the COMP card. In subsequent
stages, the fluid composition is set to the liquid phase composition in the
previous stage if the OIL or BOTH keyword is included on the DIFEXPTEST
card. The fluid composition is set to the vapor phase composition in the
previous stage if the GAS keyword is included on the DIFEXPTEST card.
3. Go to the next stage if the stage pressure is not smaller than the saturation
pressure.
5. Include the fluid composition in the stage in a list of the composition entries
of the EOS interpolation tables (if the stage number is larger than two).
7. Calculate densities and viscosities of oil and gas and output them to a file
(Fortran Unit 77) .
The simulations of the differential expansion test can be repeated several times for
different temperature values specified on the TEMP card.
For the simulation of multiple contact tests, the user must specify
In each stage of the multiple contact test, the following procedure is executed:
Zj = ( 1 – f ) × Xj + f × Yj , (23-64)
In the first stage, the liquid and vapor phase compositions X1 and Y1 are
defined on the OILCM and GASCM cards. In subsequent stages of the
backward multiple contact test, the liquid phase composition Xj is set to the
liquid phase composition in the previous stage and the vapor phase
composition Yj is not changed. In subsequent stages of the forward multiple
contact test, the vapor phase composition Yj is set to the vapor phase
composition in the previous stage and the liquid phase composition Xj is not
changed.
3. Stop the test simulation if the specified pressure is not smaller than the
saturation pressure.
5. Include the fluid composition in the stage in a list of the composition entries
of the EOS interpolation tables.
7. Calculate densities and viscosities of oil and gas and output them to a file
(Fortran Unit 77) .
The simulations of the multiple contact test can be repeated several times for
different pressure, gas fraction, and temperature values specified on the PRES,
GASFRAC, and TEMP cards.
The simulator calculates and stores saturation pressure values PSATjk (bubble
point or dew point pressure) for each combination of temperature and composition
entries (Tk, Zj), k = 1, 2,..., nt, j = 1,2,...,nz. The equation of state is applied for the
saturation pressure determination52.
At the same time, the simulator determines and stores the K-values at the pressure
equal to the saturation pressure:
jk jk jk
K i = y i ⁄ x i , i = 1, 2, …, nc, (23-65)
jk jk
where x i , y i are molar fractions of the i-th hydrocarbon component in liquid
and vapor phases at the temperature Tk, composition Zj, and pressure PSATjk.
The derivatives of the recovery factors with respect to pressure are also
calculated, because the cubic spline interpolation is applied.
The simulator calculates and stores the liquid and vapor compressibility factors
and their derivatives with respect to pressure for all combinations of temperature,
composition, and pressure entries of the EOS interpolation tables.
The liquid and vapor compressibility factors are determined as maximum and
minimum roots of the third order equation of state52.
The derivatives of the compressibility factors with respect to pressure are also
calculated, because the cubic spline interpolation is applied.
The simulator calculates and stores recovery factors and liquid compressibility
factors for all stages of separator batteries. The equation of state is applied. The
calculations are repeated for all composition entries of the EOS interpolation
tables. Values of pressure and temperature specified by the user for different
stages of separator batteries are used in the calculations.
The phase equilibrium procedure in the EOS interpolation option consists of the
following steps:
5. calculation of the liquid and vapor phase compositions using the recovery
factors;
6. calculation of densities of the liquid and vapor hydrocarbon phases using the
compressibility factors.
All temperature entries of the EOS interpolation table are sorted in increasing
order.
If temperature T is not larger than the first temperature entry, only the first
temperature entry is used in the interpolation procedure. If the temperature T is
not smaller than the last temperature entry, only the last temperature entry is used
in the interpolation procedure. In these cases, the simulator issues a warning
message or stops the run if it is requested by the user. Otherwise, two closest
temperature entries Tk and Tk+1 are selected as
Tk ≤ T < Tk + 1 . (23-68)
Two procedures are implemented for the selection of the closest composition
entries and multidimensional interpolation between the selected entries:
This method is applied if the MDI keyword is included on the EOSINT card.
The distance between two composition vectors Zk and Z is determined as
nc
j 2
D ( Z j, Z ) = ∑ ( zi – zi ) × ci . (23-69)
i=1
The non-negative coefficients c 1, c 2, …, c nc can be input by the user on the
COEFFICIENT card. As default, they are set to one.
Two composition entries Z j and Z j + 1 with the smallest distance from the
composition vector Z are selected.
F(Z) = ∑ ( ci × zi ) . (23-70)
i=1
The coefficients c 1, c 2, …, c nc can be input by the user on the COEFFICIENT
card. As default, they are set to molecular weights of hydrocarbon components;
therefore, the molecular weight of the fluid is used as the interpolation function.
All compositional entries of the EOS interpolation table are sorted in increasing
order of the interpolation function.
F ( Z j ) ≤ F ( Z ) < F ( Z j + 1 ). (23-72)
One or two pressure entries are selected for each combination of the selected
temperature and composition entries.
If pressure P is not larger than the minimum pressure PMIN, only the first
pressure entry is used in the interpolation procedure. If pressure P is not smaller
than the maximum pressure PMAX, only the last pressure entry is used in the
interpolation procedure. In these cases, the simulator issues a warning message or
stops the run if it is requested by the user. Otherwise, two closest pressure entries
jk jk
P m and P m + 1 are selected as
jk jk
Pm ≤ P < Pm + 1 . (23-73)
If all saturation pressures in the selected temperature and composition entries are
the bubble point pressures, the bubble point pressure at current temperature T and
composition Z is approximately defined as the maximum value of these saturation
pressures. In this case, two hydrocarbon phases are present if pressure P is less
than the bubble point pressure. Otherwise, only liquid phase is present.
If all saturation pressures in the selected temperature and composition entries are
the dew point pressures, the dew point pressure at temperature T and composition
Z is approximately defined as the maximum value of these saturation pressures. In
this case, two hydrocarbon phases are present if pressure P is less than the dew
point pressure. Otherwise, only vapor phase is present.
Consider, for example, this interpolation procedure for the recovery factor ri of the
i-th hydrocarbon component. A similar procedure is applied for the other recovery
factors.
1. If the current pressure P is larger than the saturation pressure PSATjk, the
recovery factor is defined as:
jk 1, ifPSAT jk is a bubblepoint
ri = ⎛ ⎞ (23-74)
⎝ 0, ifPSAT jk is a dewpoint. ⎠
2. If the current pressure is smaller than the saturation pressure PSATjk and two
pressure entries are selected, the recovery factor is calculated as:
jk 2 3
ri = a0 + a1 × P + a2 × P + a3 × P , (23-75)
If only one pressure entry is selected the recovery factor is set to known value
of the recovery factor in this pressure entry
jk jkm
ri = ri . (23-76)
The following linear interpolation procedure is used for the interpolation between
the selected composition and temperature entries:
■ Interpolation in composition:
jk j + 1, k
bj × ri + ( 1 – bj ) × ri
k
ri = if two composition entries are selected (23-77)
jk
ri
if one composition entry is selected.
D ( Z j + 1, Z )
b j = ------------------------------------------------------
-. (23-78)
D ( Z j + 1, Z ) + D ( Z, Z J )
F ( Zj + 1 ) – F ( Z )
b j = ---------------------------------------- . (23-79)
F ( Zj + 1 ) – F ( Zj )
■ Interpolation in temperature.
k k+1
dk × ri + ( 1 – dk ) × ri
Tk + 1 – T
d k = ------------------------ . (23-81)
Tk + 1 – Tk
In this case, the described interpolation procedure is repeated one or two times
for closest paths. These paths are selected based on values of the following
path function:
nc
Fp ( Z ) = ∑ ( cpi × zi ) . (23-82)
i=1
The coefficients c p1, c p2, …, c pnc can be input by the user on the
COEFFICIENT PATH card.
The recovery and compressibility factors are calculated using the linear
interpolation between selected paths.
Compositions of the liquid and vapor hydrocarbon phases in the saturated region
are determined using recovery factors. (In the saturated region, the current
pressure is lower than the saturation pressure.) The following procedure is
applied:
■ Define a molar fraction of the liquid phase (see Equation 23-66) as follows:
L = ∑ ri × zi . (23-83)
i=1
■ Calculate molar fractions of hydrocarbon components in the liquid phase as
follows:
ri × zi
x i = -------------, i = 1, 2, …, nc. (23-84)
L
( 1 – ri ) × zi
y i = ---------------------------, i = 1, 2, …, nc. (23-85)
1–L
■ Define K-values at the saturation pressure using the linear spline interpolation
in composition and temperature (see “Determination of Saturation Pressure
and K-values” and “Linear Spline Interpolation in Composition and
Temperature”).
z i, ifonlyliquidphaseispresent
xi = ,i = 1, 2, …, nc. (23-86)
z i ⁄ K i, ifonlyvaporphaseispresent
K i × z i, ifonlyliquidphaseispresent
yi = ,i = 1, 2, …, nc. (23-87)
z i, ifonlyvaporphaseispresent
Determination of Densities
Molar densities of the liquid and vapor hydrocarbon phases are calculated using
the compressibility factors and the equation of state as follows:
P
ρ n = -------------------------- , n = l, v, (23-88)
R × Ta × zn
where ρ l, ρ v are the molar densities of the liquid and vapor hydrocarbon phases;
z l, z are the liquid and vapor compressibility factors; R is the universal
constant; T a is the absolute temperature.
23.5.6 Example: Input of the EOS Interpolation Option in the Fifth SPE
Comparative Solution Model
As an example, input cards required for the application of the EOS interpolation
option in the Fifth SPE Comparative Model (scenario two)24 are described in this
section. Simulation results of the EOS interpolation and EOS model are
compared.
The following cards need to be added to a VIP-CORE input deck in the TABLES
section to apply the EOS interpolation option (comment lines are started with C):
C
C The EOSINT card must be included to invoke the EOS interpolation option
C
EOSINT
C
C Temperature entry of the EOS interpolation table.
C
TEMPERATURE 160
C
C Maximum and minimum pressure entries.
C The simulator internally selects pressure entries for each composition entries.
C
PMAX 4600
PMIN 940
C
C Path function coefficients.
C
COEFFICIENT PATH
0 10 0 0 0 0
C
C Composition entries of the EOS interpolation table is divided into three paths:
C Path 1 - reservoir oil;
C Path 2 - mixture of 50% of the reservoir oil and 50% of the injected gas,
C Path 3 - mixture of 5% of the reservoir oil and 95% of the injected gas.
C
C Composition entries in each path are automatically generated by means of
C simulations of the differential expansion PVT tests.
C
C Composition entries in Path 1.
C
DIFEXPTEST 1 BOTH 1
C Number of stages. Pressure in each stage is calculated automatically.
NPRES 20
Simulation results of the EOS interpolation and EOS models in the Fifth SPE
Comparative Project are presented in Figure 23-1. Oil, water, gas production
profiles, gas-oil ratio, and average reservoir pressure are compared. This
comparison demonstrates that the results of the EOS interpolation model match
those of the fully compositional simulation. Besides, the new option reduces the
CPU time in 30%. Significantly larger reductions of the CPU time have been
achieved in full field reservoir simulations and pattern models with large numbers
of grid cells and hydrocarbon components.
23.5.7 Conclusions
■ A new equation of state interpolation option is developed in VIP for the CPU
time reduction in compositional models
Figure 23-1: Comparison of EOS and EOSINT Model Results - Fifth SPE
Comparative Reservoir Model
Figure 23-2: Comparison of EOS and EOSINT Model Results - Cupiagua Full Field
Reservoir Model
Figure 23-3: Comparison of EOS and EOSINT Model Results - Cusiana Full Field
Reservoir Model
Figure 23-4: Comparison of EOS and EOSINT Model Results - PBU History Full
Field Reservoir Model
Figure 23-5: Comparison of EOS and EOSINT Model Results - Ursa Full Field
Reservoir Model
Figure 23-6: Comparison of EOS and EOSINT Model Results - D13 PBU Pattern
Model
Figure 23-7: Comparison of EOS and EOSINT Model Results - EWE PBU Pattern
Model
24
00000 Relative Permeability and Capillary Pressure
Adjustments Near the
Critical Point
24.1 Introduction
In a miscible displacement simulation, some gridblocks of a model may go
through a compositional path that leads to the critical point of the hydrocarbon
mixture. In a gridblock near the critical point, when a hydrocarbon mixture splits
into two phases, the compositions of the phases are similar. For this reason,
interfacial tension and capillary pressure between the phases become small and
approach zero, the oil and gas residual saturations decrease, and the gas and oil
relative permeability curves approach straight lines near the critical point.
r S o – S org f ( γ )
k ro = f ( γ )k ro + [ 1 – f ( γ ) ]k m ( S w ) ------------------------------------------ (24-1)
1 – S w – S org f ( γ )
r S g – S gc f ( γ )
k rg = f ( γ )k rg + [ 1 – f ( γ ) ]k m ( S w ) ---------------------------------------- (24-2)
1 – S w – S gc f ( γ )
r γ
P cgo = P cgo -------- (24-3)
γ ref
where kro, krg, Pcgo are the adjusted relative permeabilities to oil, gas, and gas-oil
r r r
capillary pressure, respectively; k ro , k rg , P cgo are the nonadjusted respective
values as calculated by VIP-EXECUTIVE from the rock curves; and km(Sw) is the
relative permeability to miscible hydrocarbon fluid near the critical point. This
value is a function of water saturation only. Also, Sorg is the residual oil saturation
to gas, γ is the gas-oil interfacial tension, γref is the reference interfacial tension of
R5000.0.1 Relative Permeability and Capillary Pressure Adjustments Near the Critical Point 24-407
VIP-EXECUTIVE® Technical Reference Guide Landmark
r
the gas-oil system on which the capillary pressure rock curve P cgo has been
measured, and f(γ) is a function of the interfacial tension defined as:
e
⎛γ ⎞
f ( γ ) = ⎜ ----*⎟ . (24-4)
⎝γ ⎠
Here, γ* is the threshold interfacial tension below which the above relative
permeability adjustment is used. For interfacial tensions greater than the threshold
value, f(γ) is assigned a value of one. Exponent e is a positive number generally in
the range from 0.1 to 0.25.
1 r r
k m ( S w ) = --- [ k ro ( S 0 =1 – S w, S g =0 ) + k rg ( S g =1 – S w, S o =0 ) ] (24-5)
2
The gas-oil interfacial tension depends on the oil and gas compositions and is
calculated from the correlation:
i=n c
1⁄4
γ = ∑ Pchi ( ρo xi – ρg yi ) (24-6)
i=1
where Pchi is the parachor of component i, and ρo, ρg are molar densities in gram-
moles per cubic centimeter.
24-408 Relative Permeability and Capillary Pressure Adjustments Near the Critical Point R5000.0.1
Chapter
25
Saturation Function
25.1 Introduction
Here, SWT and SGT define saturation-dependent relative permeabilities and
capillary pressures:
The SWT and SGT tables must follow the equilibrium data. All SWT tables must
precede SGT tables.
The gas and water relative permeabilities (for both two and three phases) are
directly interpolated from the tables. The two-phase oil relative permeabilities are
also from the tables. The three-phase oil relative permeabilities are computed
from the two sets of two-phase values according to Stone’s Model I, Stone’s
Model II, and the Saturation Weighted Interpolation Model.
Example:
SWT 1
SW KRW KROW PCWO
The title card SW, KRW, KROW, PCWO must appear in the order shown.
The above example shows that connate water saturation (Swl) is 0.2, residual
water saturation (Swr) is 0.3, water saturation at residual oil (Swro) is 0.8, and
maximum water saturation (Swu) is 1.0.
For equilibrium initialization, water saturation above the water-oil transition zone
is the connate water saturation and that below the transition zone is maximum
water saturation.
Example:
SGT 1
SG KRG KROG PCGO
0.0 0.0 0.9 0.0
0.1 0.0 0.7
0.16 0.0 0.55
0.2 0.02 0.45
0.4 0.07 0.1
0.5 0.12 0.05
0.7 0.23 0.0
0.8 0.3 0.0 0.0
The title card SG, KRG, KROG, PCGO must appear in the order shown.
The above example shows that the connate gas saturation (Sgl) is 0, the residual
gas saturation (Sgr) is 0.16, the gas saturation at residual oil (Sgro) is 0.7, and the
maximum gas saturation (Sgu) is 0.8.
For equilibrium initialization, the gas saturation above the gas-oil transition zone
is the maximum gas saturation and that below the transition zone is the connate
gas saturation.
where
* S o – S or
S o ≡ ------------------------------
1 – S or – S wl
* Sg
S g ≡ ------------------------------
1 – S or – S wl
* S w – S wl
S w ≡ ------------------------------
1 – S or – S wl
⎛ S orw – S org ⎞
S or ≡ S orw – ⎜ --------------------------------⎟ S g
⎝ 1 – S org – S wl⎠
Here, krocw is the relative permeability to oil at connate water (Swl), krow is the
relative permeability to oil in the two-phase water-oil system (without gas
present), krog is the relative permeability to oil in a gas-oil system with connate
water, and Sorj is the residual oil saturation in the water-oil (j = w) or gas-oil (j =
g) system.
Sg A Sg ⎞ A
S or = S org ⎛ ----------⎞ + S orw ⎛⎝ 1 – ---------
2 2 3
-⎠ – ε ( S gro S g – S g ) + ν ( S gro S g – S g )
⎝ S gro⎠ S gro
This collapses to the first definition of Sor when A=1, ε = ν = 0. The values A, ε,
ν are input on the STONE1 card.
k row k rog
k ro = k rocw ⎛ -----------
- + k rw⎞ ⎛ -----------
- + k rg⎞ – k rg – k rw
⎝k ⎠ ⎝k ⎠
rocw rocw
It assumes that the oil phase is uniformly distributed in the gridblock, while the
gas and water phases are completely segregated. Water saturation in the gas zone
is assumed to be connate water saturation (Swl), while gas saturation in the water
zone is zero. For average oil, gas, and water saturations of So, Sg, and Sw , the full
breakdown of the saturation distribution is as follows.
In the gas zone, which occupies a fraction, Fg, of the pore volume:
Oil saturation = So
Gas saturation = Sg + Sw - Swl
Water saturation = Swl.
In the water zone, which occupies a fraction, 1 - Fg, of the pore volume:
Oil saturation = So
Gas saturation = 0
Water saturation = Sg + Sw.
The material balance requires that the fraction of the pore volume occupied by the
gas zone, Fg, be
Sg
F g = -------------------------------
S g + S w – S wl
* *
S g k rog ( S g ) + ( S w – S wl )k row ( S w )
k ro = ------------------------------------------------------------------------------
-
S g + S w – S wl
where Sg* is gas saturation in the gas zone and Sw* is water saturation in the water
zone:
*
S g ≡ S g + S w – S wl
*
Sw ≡ Sg + Sw
1. Maximum gas saturation should not exceed 1.0 minus the connate water
saturation:
S gu ≤ 1 – S wl
2. Connate gas saturation should not exceed 1.0 minus the maximum water
saturation:
S gl ≤ 1 – S wu
3. The oil relative permeability at the connate water saturation and connate gas
saturation must be the same.
Example:
ISAT CON
1
ISAT X+ CON
2
ISAT X- CON
2
ISAT Y+ CON
3
ISAT Y- CON
3
ISAT Z+ CON
4
ISAT Z- CON
5
In the above data, well relative permeability calculations use saturation Table 1
and interblock flow calculations in the X- direction use saturation Table 2, in the
Y- direction use saturation Table 3, in the Z+ direction use Table 4, and in the Z-
direction use Table 5.
The following guidelines for use of the directional relative permeability option
should be helpful:
■ Different saturation function table assignments may be made for each flow
direction for both the VE and non-VE options. Even though the program
internally generates the VE curves, the user-specified end points from tables,
or arrays, are honored.
■ Separate saturation function end-point arrays may be input for each flow
direction. The rules for entering directionally dependent end-point arrays
follow the standard rules. For example, if the SWRO array is entered for the
X+ direction, the SWR array for the X+ direction also must be entered.
■ The program computes separate mobility arrays for each phase only for each
flow direction with data that are directionally dependent. Because of this,
large amounts of storage are required as the number of mobility arrays
increases. The user should avoid entering default values for arrays for a flow
direction that otherwise would not need a separate mobility array.
■ Well mobilities are not directionally dependent. The ISAT array, without
direction specification, controls assignment of the saturation function table for
well calculations. Other end-point arrays, without direction specification, also
are used in the well mobility calculations.
25.7.1 Implementation
The gas remobilization option can be invoked by the user any time during the
simulation (usually after the waterflood of a gas reservoir or a gas cap region and
at the onset of the pressure blowdown period). When this option is invoked, all
reservoir gridblocks are internally divided into two types. Type 1 gridblocks are
those previously invaded by water and which have reached the trapped gas
saturation. The remobilization of gas in these gridblocks is considered to be
controlled by gas expansion and hence is subject to the gas remobilization
calculation. Gridblocks not considered as Type 1 are classified as Type 2 and their
relative permeabilities to gas will be calculated using the standard gas hysteresis
options. The secondary drainage curve for Type 1 gridblocks is assumed to be
reversible, i.e., the subsequent secondary imbibition curve retraces the secondary
drainage curve. The gas remobilization option is available only if the gas relative
permeability hysteresis option is also invoked.
Numerically, the Type 1 gridblocks are those that satisfy all of the following
conditions at the time the gas remobilization option is turned on:
1. The water saturation must be greater than the historical minimum water
saturation (Swmin) plus a user-controlled incremental saturation (grmdsw), i.e.,
Sw > Swmin + grmdsw. The default value for grmdsw is 0.02.
2. The gas saturation must be less than or equal to the trapped gas saturation
(Sgtr) corresponding to the gridblock historical maximum gas saturation
(Sgmax), i.e., Sg ≤ Sgtr(Sgmax).
3. The gridblock historical maximum gas saturation must be greater than its
critical gas saturation, i.e., Sgmax > Sgc.
The threshold gas saturation at which the trapped gas becomes mobile, Sgrm, and
the secondary drainage relative permeability to gas at the maximum gas
saturation, Krgmax, are assumed to be functions of the trapped gas saturation.
These functions are rock-type-dependent and must be specified by the user in a
tabular form.
The secondary drainage relative permeability curve for each rock type must be
supplied by the user using a tabular format (i.e., Krg2d,r(Sg)) or a prescribed
analytic function. If the tabular format is chosen, the saturation endpoints are
identified as Sgrm,r and Sgma,r and their corresponding relative permeability
endpoints are 0.0 and Krgmax,r . Here subscript r denotes the rock type value. For
consistency, it is stipulated that Sgma,r be the same as the maximum gas saturation
specified in the gas saturation table (SGT). During the pressure blowdown period,
the gridblock trapped gas saturation, Sgtr,m, is first used to determine the gridblock
Sgrm and Krgmax (identified as Sgrm,m and Krgmax,m, respectively) values using the
user-input Sgrm(Sgtr) and Krgmax(Sgtr) table, where subscripts m denote the
gridblock variables. The gridblock secondary drainage relative permeability is
then determined using the standard two-point endpoint scaling method:
K rgmax, m o
K rg ( S g ) = ---------------------- K rg2d, r ( S g ) (25-1)
K rgmax, r
where
o ( S gma, r – S grm, r )
S g = S grm, r + ( S g – S grm, m ) -------------------------------------------- (25-2)
( S gma, m – S grm, m )
C
( 1 + C 2 )S g13
K rg ( S g ) = K rgmax, m ----------------------------
C3
- (25-3)
1 + C 2 S g1
where
S g – S grm, m
S g1 ≡ --------------------------------------- (25-4)
S gma, m – S grm, m
Printout of the input and reconstructed gas remobilization tables may be requested
through the SGT and RSGT keywords in the PRINT TABLES card in VIP-CORE.
Finally, at the time the gas remobilization option is invoked, the number of Type 1
The option was tested using a small test model in which a primary depletion
process was followed by water injection into the gas cap area and pressure
blowdown.
C
C Gas Remobilization Option Input
C
GASRM 1
SGTR SGRM KRGMAX
TIME 3650.
C
C Turn on Gas Remobilization Option
C
GASRMON 0.05
26
Separators
26.1 Introduction
VIP-EXECUTIVE is a compositional simulator. The governing equations of the
simulator are molar conservation equations; they describe movement of
hydrocarbon fluids in terms of moles at reservoir conditions. However, well
production rates are customarily reported in volumetric units at standard
conditions. In a compositional simulator, conversion of well rates from reservoir
conditions to volumetric units at standard conditions requires a separator model.
The separator model describes separation of reservoir fluids in surface facilities.
Besides calculating well production rates at standard conditions, the separator
model also calculates in-place volumes of reservoir oil and gas at standard
conditions. These calculations are performed when a region report is requested.
Vapor
Stream
Feed
Stage Stage Stage Stage
1 2 3 Ns
Liquid
Stream
Here, Flmn and Fvmn are fractions of the liquid and vapor streams that are sent to
stage n from stage m. These fractions are defined by a user on the SEPARATOR
card. The sum of the vapor and liquid streams in moles for component i, Qvmi,
Qlmi, respectively, which leaves stage m, is equal to the feed stream to this stage:
v l
Q mi = Q mi + Q mi, i = 1, 2, …, n c, n = 1, 2, …, n s (26-2)
Moles that are sent to the oil and gas sales lines, Qo and Qg, are equal to the sum
of all streams sent to these lines from all the separator stages:
nc ns nc ns
l l v v
Qo = ∑ ∑ Fno Qni, Qg = ∑ ∑ Fng Qni (26-3)
i = 1n = 1 i = 1n = 1
Here, Flno and Fvng are fractions of the liquid and vapor streams, which are sent to
the sales lines. They are defined by a user on the SEPARATOR card.
The volume rates of oil and gas, Vos and Vgs, at stock tank conditions which leave
the separator battery are calculated as follows:
Ts
V os = 1.9122 Z os ----- Q o (26-4)
Ps
Ts
V gs = 0.010736 ----- Q g (26-5)
Ps
Here, Zos, Ts, and Ps are the liquid compressibility factor, temperature, and
pressure at the stock tank conditions.
The Standing-Katz density correlation can be used to calculate oil volumetric rate
at standard conditions instead of Equation 26-4.
In the first case, the simulator generates a set of K-values for each of the PVT
tables defined in a reservoir model. Each of the sets of K-values with the default
separator configuration defines a default separator battery.
In the second case, a user defines separator configuration as well as K-values with
the SEPARATOR card. The SEPTEST option provides the most accurate method
for defining a separator. Here the input is the actual data from the separator test;
the simulator calculates the corresponding K-values.
The K-values for each stage of the separator battery are functions of the fluid
composition in the input stream of the separator. If fluid composition is not
significantly changed, the K-values do not need to be recalculated. This approach
has been implemented in VIP-EXECUTIVE.
The separator flash procedure is applied in each outer iteration of a timestep for
each well. In the first outer iteration, the K-values are calculated for each stage of
the separator battery for each well. The fluid composition, K-values, and liquid
compressibility factor are saved. If the maximum change in the composition in the
next outer iteration is less than the specified tolerance, then the K-values and
liquid compressibility factor are not updated.
The user can control the tolerance of the composition changes with the BHPITN
card. If this tolerance is set to zero, the K-values are recalculated in each outer
iteration of each timestep.
27
Simulator Performance
27.1 Introduction
In the simulator, we advance the solution in time while honoring the timing of
changes to data input and obtaining output at selected points. As long as the
correct answer is produced, the scheme employed should minimize work
performed (as measured by the CPU time consumed). There are tradeoffs between
speed and solution quality and, while the default settings will give good
performance for most problems, it may be necessary for the user to adjust them in
some cases. Three levels of control exist; timestep control, non-linear iteration
parameters and iterative solver parameters (if appropriate). The monitoring and
adjustment of performance at each of these levels is described in this chapter.
The simulator can select its own timesteps. They are constrained only by the
maximum changes in reservoir variables specified on the DT card, or by gridblock
throughput limitations when the IMPES formulation is used (see IMPSTAB card).
Timestep size is altered automatically to hit exactly the times or dates on TIME or
DATE cards. Under automatic timestep control, maximum pressure, saturations,
vapor fraction, and total composition changes may be exceeded slightly to save
the work required to repeat the timestep.
Timesteps are controlled by three VIP cards; DT, TCUT, and IMPSTAB, which
are described in detail in the user manual and are discussed in the following
paragraphs.
Each of the solution variables (pressure, water saturation, [vapor fraction or gas
saturation] and mole fraction) can be used to control timestep size and/or limit
maximum change - the option are; BOTH, DTONLY, MAXONLY and NONE.
The default is BOTH for the IMPES formulation and DTONLY for the fully
implicit formulation.
Two distinct conditions can trigger a timestep cut; convergence failure of the non-
linear iterations and violation of a maximum allowable change (see above). After
each convergence failure the timestep size is cut by a factor of two and the
timestep is retried. A maximum change violation causes a new timestep size to be
calculated by interpolation, subject to the most restrictive of the variables flagged
for timestep control.
If it was not already printed, an iteration summary will be printed by default for a
timestep that has a convergence failure unless the CNVFLOFF keyword is
specified on the OUTPUT card.
When controlled by the user, the timestep size will be set equal to a target factor
times the maximum stable IMPES timestep, provided that this timestep would
satisfy the constraints imposed by the DT and TCUT cards and not exceed the
IMPSTAB card limit.
This option is derived from an updating procedure for solution unknowns that
satisfies the material balance for hydrocarbons and water during each outer
iteration. Also, the convergence of the outer iteration is controlled by the residual
in the saturation constraint equation. The convergence tolerance may be specified
through the TOLSCN card. The use of this option will reduce the number of outer
iterations and hence the CPU time. This option is only applicable to the IMPES
formulation. OPTMBL only functions for the run in which it is specified, i.e., it is
not passed on restarts and must be reentered. For example:
STBCHK and PJACO are optional keywords that control stability tests and partial
Jacobian updates. STBCHK indicates that an OPTMBL stability test will only be
performed for single-phase gridblocks either with two-phase neighbors or that
contain wells. Otherwise, a stability test is performed for all single-phase
gridblocks. For certain compositional problems, this feature could significantly
reduce the CPU time. It is only used for compositional problems.
PJACO indicates that a partial Jacobian update of the fugacity equations will be
performed, in conjunction with the OPTMBL option. Jacobian coefficients are not
recalculated for gridblocks that satisfy a preset convergence criteria. For certain
Failure to converge the outer iterations of a timestep within the maximum number
of outer iterations will normally cause the timestep to be repeated. Convergence of
the outer iteration over a timestep is achieved when none of the changes in
solution variables which occurred during the previous iteration are larger in
absolute value than the corresponding tolerances established on the TOLD card.
Additional tests for convergence of residuals are defined by the TOLR cards.
When the material balance error for each phase in each gridblock is less than the
tolerances specified on the TOLR card, the conservation equations are assumed to
have been solved and the timestep is concluded. The TOLD tests are independent
of the TOLR tests. Either set of convergence tests can signal the end of the
timestep even if the other has not yet been satisfied.
In the following example, the TOLD is satisfied after two iterations, whereas the
TOLR would be satisfied after three. The net result is that the convergence criteria
would be satisfied after two iterations.
Maximum Pressure
Maximum Residual
TOLD
TOLR
Change
0 1 2 3
Iterations
For most problems, timesteps should converge in less than seven iterations, which
is the preferred upper limit. Individual limits can be turned off by setting them to
0.99, but this should only be done in extreme cases - material balances may be
adversely affected. In any event, active iteration maximum change values should
always be less than or equal to their timestep counterparts.
The TOLD tests are independent of the TOLR tests (discussed below). If either set
of convergence tests is satisfied, no further outer iterations are performed and the
timestep is concluded.
A zero value for any of the tolerances forces non-convergence based on maximum
changes.
The TOLR tests are independent of the TOLD tests. If either set of convergence
test is satisfied, no further outer iterations are performed and the timestep is
concluded.
i.e. when the solution residuals have been reduced by four orders of magnitude,
the timestep is assumed to have converged, regardless of absolute values.
A brief discussion of run optimization is included at the end of this section and
provides the checklist for possible problem areas.
**************************************************************************************
* *
* VIP-EXEC TIMESTEP SUMMARY *
* *
* OFFSHORE EUROPE DEMONSTRATION MODEL *
* FINE GRID MODEL *
* *
* VIP-EXEC *
**************************************************************************************
TIMESTEP DAILY PRODUCTION CUMULATIVE PRODUCTION DAILY INJECTION CUM. INJECTION AVG TIME ITN
------------ -------------------------------------- -------------------------- ----------------- ----------------- PRES STEP ---
OIL GAS WATER GOR WATER OIL GAS WATER GAS WATER GAS WATER CUTS
SCF/ CUT
NO. DAYS MSTB/D MMSCF/D STB/D STB FRAC. MMSTB BSCF MSTB MMSCF/D STB/D BSCF MSTB PSIA
---- ------- -------- -------- -------- ----- ----- -------- -------- -------- -------- -------- -------- -------- ------ ---- ---
1 1.0000 1.90000 3.07223 0. 1616 0. .00190 .00307 0. 0. 0. 0. 0. 5308. 0 2
2 6.0000 1.90000 3.07223 .01834 1616 0. .01140 .01843 .00009 0. 0. 0. 0. 5307. 0 2
3 31.000 1.90000 3.07223 .02910 1616 0. .05890 .09524 .00082 0. 0. 0. 0. 5306. 0 3
4 91.000 1.90000 3.07223 .04225 1616 0. .17290 .27957 .00335 0. 0. 0. 0. 5303. 0 3
5 181.00 1.90000 3.07223 .05870 1616 0. .34390 .55607 .00864 0. 0. 0. 0. 5299. 0 3
6 183.64 57.00 92.17 1.72174 1616 0. .49455 .79967 .01319 55.30 31500. .14615 83.25250 5295. 1 5
7 186.26 57.00 92.17 8.66352 1616 0. .64389 1.04114 .03589 55.30 31500. .29104 165.78 5292. 0 3
8 189.09 57.00 92.17 11.49 1616 0. .80489 1.30147 .06835 55.30 31500. .44724 254.76 5289. 0 3
9 192.25 57.00 92.17 13.53 1616 0. .98514 1.59294 .11114 55.30 31500. .62212 354.37 5286. 0 3
10 196.01 57.00 92.17 15.18 1616 0. 1.19958 1.93967 .16825 55.30 31500. .83016 472.87 5284. 0 2
11 200.76 57.00 92.17 16.64 1616 0. 1.47030 2.37742 .24730 55.30 31500. 1.09280 622.48 5281. 0 3
12 204.91 57.00 92.17 18.04 1616 0. 1.70684 2.75990 .32215 55.30 31500. 1.32229 753.20 5278. 0 3
13 209.38 57.00 92.17 19.03 1616 0. 1.96147 3.17163 .40716 55.30 31500. 1.56933 893.92 5275. 0 3
14 214.42 57.00 92.17 19.93 1616 0. 2.24880 3.63622 .50761 55.30 31500. 1.84809 1052.71 5273. 0 3
15 218.70 57.00 92.17 20.79 1616 0. 2.49288 4.03090 .59665 55.30 31500. 2.08490 1187.60 5271. 0 3
16 222.57 57.00 92.17 21.45 1616 0. 2.71363 4.38784 .67971 55.30 31500. 2.29906 1309.59 5269. 0 3
17 226.21 57.00 92.17 21.99 1616 0. 2.92099 4.72314 .75969 55.30 31500. 2.50024 1424.18 5267. 0 3
18 229.71 57.00 92.17 22.45 1616 0. 3.12063 5.04595 .83833 55.30 31500. 2.69392 1534.51 5265. 0 3
19 233.18 57.00 92.17 22.87 1616 0. 3.31801 5.36510 .91753 55.30 31500. 2.88542 1643.59 5263. 0 2
20 236.62 57.00 92.17 23.26 1616 0. 3.51452 5.68285 .99771 55.30 31500. 3.07607 1752.19 5262. 0 2
21 240.59 57.00 92.17 23.62 1616 0. 3.74070 6.04858 1.09146 55.30 31500. 3.29550 1877.18 5260. 0 3
22 250.41 57.00 92.17 24.02 1616 0. 4.30000 6.95295 1.32716 55.30 31500. 3.83813 2186.27 5255. 0 3
23 262.40 57.00 92.17 24.89 1616 0. 4.98378 8.05859 1.62570 55.30 31500. 4.50151 2564.14 5250. 0 3
24 273.00 57.00 92.17 25.84 1616 0. 5.58790 9.03543 1.89953 55.30 31500. 5.08762 2898.00 5245. 0 3
25 278.00 90.25 145.93 36.97 1616 0. 6.03915 9.76509 2.08436 87.56 58500. 5.52541 3190.50 5239. 0 4
26 288.93 90.25 145.93 42.23 1616 0. 7.02549 11.35996 2.54586 87.56 58500. 6.48233 3829.84 5230. 0 4
27 298.70 90.25 145.93 46.30 1616 .001 7.90770 12.78647 2.99848 87.56 58500. 7.33824 4401.69 5222. 0 6
28 308.64 90.25 145.93 48.83 1616 .001 8.80466 14.23681 3.48377 87.56 58500. 8.20844 4983.10 5215. 0 2
29 319.21 90.25 145.93 50.91 1616 .001 9.75869 15.77945 4.02190 87.56 58500. 9.13403 5601.50 5209. 0 3
30 322.60 90.25 145.93 52.85 1616 .001 10.06444 16.27383 4.20094 87.56 58500. 9.43066 5799.69 5206. 1 2
31 325.94 90.25 145.93 53.45 1616 .001 10.36567 16.76092 4.37936 87.56 58500. 9.72291 5994.95 5204. 0 2
32 329.94 90.25 145.93 54.02 1616 .001 10.72643 17.34426 4.59529 87.56 58500. 10.07291 6228.79 5202. 0 2
33 335.16 90.25 145.93 54.66 1616 .001 11.19801 18.10677 4.88092 87.56 58500. 10.53042 6534.47 5198. 0 2
34 342.64 90.25 145.92 55.46 1616 .001 11.87320 19.19845 5.29583 87.55 58500. 11.18543 6972.13 5194. 0 2
35 353.17 90.25 145.91 56.53 1616 .001 12.82288 20.73384 5.89063 87.55 58500. 12.10666 7587.71 5188. 0 2
36 359.67 90.25 145.87 58.00 1616 .001 13.41029 21.68328 6.26813 87.52 58500. 12.67632 7968.47 5184. 0 2
37 365.00 90.25 145.84 58.90 1615 .001 13.89090 22.45990 6.58177 87.50 58500. 13.14230 8280.00 5181. 0 3
SIMULATOR VIP-ENCORE
VERSION v32r0
INITIALIZATION LAST UPDATE 95/02/17
LAST CHANGE SET NAME base_cse
REVISION NUMBER 0
SIMULATION LAST UPDATE 95/02/17
LAST CHANGE SET NAME bpfix01
REVISION NUMBER 0
FORMULATION IMPES
EQUATION SOLUTION METHOD BLITZ
PERFORMANCE STATISTICS
----------------------
CURRENT RUN CUMULATIVE
----------- ----------
TIME SIMULATED (DAYS) 365.000 365.000
NUMBER OF SUCCESSFUL TIMESTEPS 37 37
NUMBER OF OUTER ITERATIONS 114 114
NUMBER OF INNER ITERATIONS 503 503
NUMBER OF UNACCEPTED TIMESTEPS 2 2
NUMBER OF CONVERGENCE FAILURES 0 0
CPU SECONDS / ACTIVE GRIDBLOCK TIMESTEP .0002968 .0002968
CPU SECONDS / TOTAL GRIDBLOCK TIMESTEP .0002416 .0002416
The following iteration summary shows slow convergence for a run with an
iterative solver. The linear convergence is good, as evidenced by the small number
of inner iterations, but the timestep does not converge in the permitted five outer
(non-linear) iterations. The timestep size is cut by a factor of two and it is
repeated. Cuts can also occur for a converged timestep, if a timestep maximum
change value is exceeded by more than the allowed limit. A new timestep size is
calculated by linear interpolation, to honor the most restrictive change limit which
has been violated. Following the first convergence after a timestep cut, the new
timestep is limited to an increase of no more than 25%.
********************************************************************************************************************************
* DETAILED ITERATION SUMMARY, TIMESTEP 31 SIZE 11.7 FROM 320.909 DAYS REASON: MAX VAPOR/SG CHANGE *
********************************************************************************************************************************
OUTER RWMAX RHCMAX DPMAX DSWMAX DVMAX DZMAX RWSUM RHCSUM INNER DAMPING
ITN (I,J,K) (I,J,K) (I,J,K) (I,J,K) (I,J,K) (I,J,K) ITN FACTOR
1 1846. .1819E+05 13.25 .2008E-01 .2181 .5224E-01 .5845E+05 -.2783E+06 5 .22929 (DV)
( 30 20 7) ( 25 7 3) ( 32 27 5) ( 15 19 3) ( 27 11 2) ( 27 11 2)
2 1423. .1402E+05 10.19 .1547E-01 .1681 .4026E-01 .4505E+05 -.2145E+06 5 .29750 (DV)
( 30 20 7) ( 25 7 3) ( 32 27 5) ( 15 19 3) ( 27 11 2) ( 27 11 2)
3 999.4 9850. 7.165 .1086E-01 .1180 .2828E-01 .3164E+05 -.1507E+06 5 .42368 (DV)
( 30 20 7) ( 25 7 3) ( 32 27 5) ( 15 19 3) ( 27 11 2) ( 27 11 2)
4 575.9 5677. 4.135 .6260E-02 .6801E-01 .1629E-01 .1824E+05 -.8683E+05 5 .73519 (DV)
( 30 20 7) ( 25 7 3) ( 32 27 5) ( 15 19 3) ( 27 11 2) ( 27 11 2)
5 152.5 1503. 1.099 .1656E-02 .1800E-01 .4313E-02 4830. -.2299E+05 5 1.00000
( 30 20 7) ( 25 7 3) ( 32 27 5) ( 15 19 3) ( 27 11 2) ( 27 11 2)
6 .6186 1.655 .0000E+00 .0000E+00 .0000E+00 .0000E+00 .2104E-01 -.5707E-01 0 1.00000
( 14 17 3) ( 29 19 7) ( 1 1 1) ( 1 1 1) ( 1 1 1) ( 1 1 1)
*** TIMESTEP CUT DUE TO CONVERGENCE FAILURE:
CURRENT SIZE = 11.6728 DTMIN = 1.00000 CUT FACTOR = 2.00000
*** TIMESTEP CUT DUE TO EXCEEDING MAXIMUM CHANGES:
CURRENT SIZE = 5.83639 DVMAX = .111261 ( 27, 11, 2) CUT FACTOR = 2.22522
The following good iteration summary shows rapid convergence. The next
timestep size will be calculated from the current timestep changes by linear
interpolation, or extrapolation, to honor the most restrictive change limit which
has been imposed, subject to a maximum increase of a factor of 5.
********************************************************************************************************************************
* DETAILED ITERATION SUMMARY, TIMESTEP 24 SIZE 10.6 FROM 262.401 DAYS REASON: TIME/DATE CARD *
********************************************************************************************************************************
OUTER RWMAX RHCMAX DPMAX DSWMAX DVMAX DZMAX RWSUM RHCSUM INNER DAMPING
ITN (I,J,K) (I,J,K) (I,J,K) (I,J,K) (I,J,K) (I,J,K) ITN FACTOR
GOOD CONVERGENCE
A. Small number of BLITZ iterations
B. Rapid reduction in 2-norm of residual from initial value
C. Small number of constraint iterations
A B C
1 2-NORM(ABS)= 3.8536E+01 2-NORM(REL)= 6.5890E-02 CITNS= 4
2 2-NORM(ABS)= 2.9722E-01 2-NORM(REL)= 5.0819E-03 CITNS= 4
3 2-NORM(ABS)= 5.7713E-02 2-NORM(REL)= 9.8679E-04 CITNS= 3
POOR CONVERGENCE
A. Large number of BLITZ iterations
B. Slow reduction in 2-norm of the residual from its initial value
C. Large (Max) number of constraint iterations
A B C
1 2-NORM(ABS)= 3.4026E+01 2-NORM(REL)= 9.10673-01 CITNS= 10
2 2-NORM(ABS)= 3.3998E+01 2-NORM(REL)= 6.0993E-01 CITNS= 10
3 2-NORM(ABS)= 2.7182E+01 2-NORM(REL)= 7.2751E-01 CITNS= 10
4 2-NORM(ABS)= 1.8116E+01 2-NORM(REL)= 4.8486E-01 CITNS= 10
5 2-NORM(ABS)= 1.0029E+01 2-NORM(REL)= 2.6843E-01 CITNS= 10
6 2-NORM(ABS)= 3.8303E+00 2-NORM(REL)= 1.0251E-01 CITNS= 10
7 2-NORM(ABS)= 9.7611E-01 2-NORM(REL)= 2.6125E-02 CITNS= 10
8 2-NORM(ABS)= 7.9748E-01 2-NORM(REL)= 2.1344E-02 CITNS= 10
9 2-NORM(ABS)= 7.4734E-01 2-NORM(REL)= 2.0002E-02 CITNS= 10
10 2-NORM(ABS)= 2.0066E-01 2-NORM(REL)= 5.3704E-03 CITNS= 10
11 2-NORM(ABS)= 2.0064E-01 2-NORM(REL)= 5.3700E-03 CITNS= 10
12 2-NORM(ABS)= 2.0064E-01 2-NORM(REL)= 5.3700E-03 CITNS= 10
13 2-NORM(ABS)= 2.0064E-01 2-NORM(REL)= 5.3700E-03 CITNS= 10
14 2-NORM(ABS)= 2.0066E-01 2-NORM(REL)= 5.3699E-03 CITNS= 10
15 2-NORM(ABS)= 2.0064E-01 2-NORM(REL)= 5.3699E-03 CITNS= 10
16 2-NORM(ABS)= 1.9753E-01 2-NORM(REL)= 5.2867E-03 CITNS= 10
17 2-NORM(ABS)= 1.9747E-01 2-NORM(REL)= 5.2850E-03 CITNS= 10
Although the section is mainly for IMPES runs, many of the notes may be applied
to IMPLICIT cases. Generally, IMPES is more susceptible to errors and
anomalies in data than is IMPLICIT, and will require more “tuning” to achieve
maximum computational efficiency. However, the work per timestep is far less
with IMPES, although timestep size may be limited. In a run where timestep size
is limited by data, such as in a history match with monthly production rate
changes, IMPES is likely to be much more efficient than IMPLICIT.
C IMPES
DT -.1 .01 30 500 .05 .05 .05
ITNLIM 1 6 500 .05 .05 .05
TOLR .001 .0001 RELTOL
C IMPLICIT
DT -1 .1 92 500 .1 .1 .1
ITNLIM 1 6 500 .1 .1 .1
TOLR .0005 .0005 RELTOL
Notice that the timestep size for IMPLICIT (.1 to 92 days) is much larger than for
IMPES (.01 to 30 days) and that both sets have the same change values specified
for timesteps and iterations.
■ Large enough timesteps can be taken with IMPLICIT to reduce the CPU time
required for a run. A rule of thumb is that if it is possible to average timesteps
three times that of IMPES, then IMPLICIT may be a good option.
The choice of solver hinges on two items, speed and memory. Very small
problems, i.e. less than 100 gridblocks, may run faster with direct solution
(GAUSS with D4 ordering) than with an iterative scheme (BLITZ). In general all
problems should always be run using BLITZ. The EXCEL solver should only be
used on problems where BLITZ has trouble converging, perhaps because of an
excessive number of arbitrary non-neighbor grid block connections. The point at
which one solver will be preferred over the other depends on the problem and
formulation selected, and will be different for speed and storage requirements.
Make short runs with the various combinations to determine which is best for your
problem.
(k) ⎧ (k) (k – 1) ξ ⎫
ρ = min ⎨ ρ max, max [ γ , ( ρ ) ]⎬ (27-1)
⎩ ⎭
where:
(k) (k) (k – 1) (k – 1)
γ
(k) F – F +J δ
= -----------------------------------------------------------------------
(27-2)
-
(k – 1)
F
The internally computed parameter γ(k) represents a measure of how closely the
linearization approximates the nonlinear function, in each Newton iteration.
Two of the above keywords are used to modify the parameters in equation (1).
TOLMN specifies the value of ρmin, and TOLEX specifies the value of the
exponent ξ.
The values of the controlling parameters of the adjustable tolerance algorithm can
be defaulted by specifying the ADJTOL to the solver card. The default values are
TOLMNdefault= RTOL so that, e.g., if RTOL if defaulted too, then TOLMN
defaults to 5.E-03; TOLMXdefault= 0.5; TOLSTdefault= max(TOLMX,5*RTOL)
and TOLEXdefault= 3.0.
This method is adapted from published work on inexact Newton methods for
constrained optimization problems(94). Our choice of TOLEX=3.0 differs from
the preferred value in optimization problems and was found by empirical tuning
of a test set of simulation models.
NOTE: The default mode for all solvers, i.e., when no solver card is specified, is to
run with the same fixed linear tolerance for all Newton iterations. The
adjustable tolerance algorithm must be explicitly requested by specifying
either the ADJTOL option or any number of the four controlling parameter
options described above.
Some behavior can be attributed to the type of model being run, but more
commonly we must use the information provided by the simulation program to
locate the cause.
The first step in analyzing your run performance is to obtain an iteration summary,
using PRINT ITER 1.
Get a feel for your run - examine the run statistics (timesteps, iteration,
unaccepted timesteps, convergence failures), check material balance on the
timestep summary line, and look at layer allocation in your wells.
Find out what your timesteps are being limited by. It will either be a timestep size
limitation or a maximum change value. Can this constraint be relaxed?
If the same block (or blocks) are flagged repeatedly in iteration and/or error
messages, check these blocks for small PV, very high transmissibility or the like.
When constructing a new model, you should perform an equilibration run (with
all well rates zero) for thirty days from time zero and check run performance.
Subsequently, examine the run performance at intervals throughout the history
match and/or prediction runs - changing controls and fluid flow conditions can
induce problems which are not evident initially. You will start out by looking at
the run statistics, which will often indicate whether there is a problem to begin
with. These indications can be such things as; too many non-linear iterations/
timestep, consumption of an excessive amount of cpu time and a high incidence of
unaccepted timesteps due to cuts and convergence failures.
Check List
timestep problems
It is easy to understand the increase in CPU time with an increase in the number of
spatial dimensions (1D, 2D, or 3D), total grid blocks, or number of wells. But it is
difficult to estimate absolute CPU time when there are so many other factors to
consider, such as solution technique (IMPLICIT or IMPES), solver selection
(GAUSS, EXCEL, BLITZ), and special features (e.g., predictive well
management, fluid tracking, dual porosity). Although, estimates will be provided
for CPU times, wall clock times will not be discussed. The wall clock time
associated with a particular simulation is highly dependent upon the configuration
of the computer and the workload which exists on the computer when the job is
executing. Since it is impossible to predict the workload which will exist when
your job is submitted, we will not provide estimates for these numbers.
For an Equation of State compositional model one can also say that the CPU time
is roughly proportional to the number of components to the 3/2 power.
For compositional problems utilizing the IMPES formulation, the data space
required increases roughly linearly with the number of components. For the
IMPLICIT formulation, the data space required increases with the square of the
number of components.
First, the user may specify relevent information about his model and run the
simulator to obtain an estimate of the storage required.
The following example data requests the storage for a 3-dimensional, black-oil,
faulted problem using the BLITZ solver in the IMPES formulation.
STORAGE NOCORERST
DIM NWMAX NPRFMX NPRFTOT BLITZ
50 5 250
IMPES
RESTART 0
NX NY NZ NC
44 80 23 2
FAULT 3000
BLITZ NORTH
15
STOP
END
Second, if the initialization data for the model has already been prepared and run
through VIP-CORE, but the production strategy has not yet been determined, the
simulator will pick up the restart file from VIP-CORE and use any dimensions
specified (or defaulted), such as the number of wells and perforations, to estimate
the storage required.
Example:
STORAGE
DIM NWMAX NPRFMX NPRFTOT BLITZ
50 5 250
IMPES
RESTART 0
BLITZ NORTH
15
STOP
END
Note that when using the STORAGE option in either case, the BLITZ keyword
needs to be the last keyword specified on the DIM card.
28
Single-Well Gridded Wellbore Simulation
28.1 Introduction
This option was designed primarily to simulate single-well commingled
production performance from multiple reservoirs with hundreds of feet of
separation between them. The wellbore is modelled as the innermost ring of
gridblocks, which allows for counter-current phase wellbore flow and wellbore
storage. Collins, et al92 presented a technique for modelling of horizontal wells
wherein the wellbore was treated as a second porosity in the dual-porosity
approach. The well "permeability" and "relative permeabilities" were adjusted to
yield the pressure drop and phase slip predicted from multiphase flow
correlations. This option makes use of parts of their technique; however, instead
of using the dual porosity to model the wellbore, the innermost ring of the radial
grid system is used to model the vertical wellbore. The equations for wellbore
flow are transformed into a form similar to Darcy's flow, and then applied to the
vertical flow within the first ring of the grid system (wellbore).
Six two-phase flow correlations are available in the current simulator for vertical
or inclined pipe flow, including Hagedorn and Brown, Dunns and Ross, Beggs
and Brill, Aziz and Govier, Orkiszewski, and Griffith, Lau, Hon, and Pearson. In
these correlations, flow conditions are divided into patterns or flow regimes.
Using the user-specified flow correlation and the computed flow regime at each
interval, the simulator transforms the wellbore flow equations into Darcy-type
vertical flow coefficients for each interval at the start of each time step.
ΔP loss = ΔP f + ΔP h + P k (28-1)
where
2
fΔLρv
Δp f = – ------------------ (28-2)
2g c D
ΔP h = – ρ g c ΔZ (28-3)
1 2
Δp k = – --- ρΔ ( v ) (28-4)
2
where f is the Moody friction factor, ρ is the total fluid mass density, D is the
wellbore diameter, v is fluid velocity, gc is gravity acceleration, ΔZ is mid-
perforation depth difference, ΔL is mid-perforation length difference, and
ρ = ∑ ρj ⋅ Sj (28-5)
j = o, g, w
The kinetic energy effect is usually considered to be negligible for wellbore flow
conditions, and will therefore be neglected here. The resulting pressure drop for
wellbore pipe flow is then:
2
fΔLρv
ΔP loss = – ------------------ – ρg c ΔZ (28-6)
2g c D
2
ΔP ΔL fρv
------- + ρg c = ΔΦ
-------- = – ------- ⋅ ------------- (28-7)
ΔZ ΔZ ΔZ 2g c D
R e = vρD
----------- (28-9)
μ
where μ is the average viscosity of the fluid expressed as
μ = ∑ μj Sj (28-10)
j = o, g, w
For laminar flow (Re < 2000), the friction factor is expressed by:
f = 64
------ (28-11)
Re
Substituting Equation 28-11 and Equation 28-9 into Equation 28-7 gives
ΔΦ – ΔL 64μ ⋅ v
-------- = ---------- ⋅ ----------------2 (28-12)
ΔZ ΔZ 2g D
c
For turbulant flow, the friction factor, pressure drop, and also liquid holdup factor
are calculated via one of the six two-phase flow correlations. Following these
calculations, the equivalent Darcy permeability can be calculated as
1---
⎛ 2g D ⎞ 2
k = μ ⋅ ⎜ ------------------------⎟
c
(28-15)
⎜ dp ⎟
⎝ f ⋅ ρ ⋅ ------- ⎠
ΔL
– k ⋅ k rj ΔΦ
v j = ---------------- ⋅ -------- (28-17)
μ ΔZ
where
k rj = S j, j = o, g, w (28-18)
A different representation for liquid holdup is used here than was used in Collins,
et. al.92 The following form is assumed for kro and krw:
k ro = α ⋅ S o
k rw = α ⋅ S w
where
μg
----- ≤ α ≤ 1 (28-19)
μl
Since the pressure drop calculated from the two-phase flow correlations included
the effects of liquid holdup, the "effective" permeability must also be adjusted as
k
k = ---------------------------------------
- (28-20)
( Sg + α ( 1 – Sg ) )
n+1 ( 1 – Sg ) n 1
α = ------------------- ⋅ α , with α = 1.0 (28-21)
hl
For flow velocities less than the critical lift velocity, simple equations are used to
enable a smooth migration from the mixture density gradient to the respective
phase density gradients, in order to allow counter-current flow in the wellbore.
The migration equations are
vc – v
ρ ml = ρ m + ( ρ l – ρ m ) ⋅ ⎛ -------------------------------------------------⎞ (28-23)
⎝ v c + ( rdamp – 1 ) ⋅ v⎠
vc – v
ρ mg = ρ m + ( ρ g – ρ m ) ⋅ ⎛ -------------------------------------------------⎞ (28-24)
⎝ v c + ( rdamp – 1 ) ⋅ v⎠
where v is the calculated flow velocity, ρm is the mixture density gradient, ρl is the
liquid density gradient, ρg is the gas density gradient, rdamp is a user-supplied
damping factor, and ρml and ρmg are the migration liquid and vapor phase
gradients to be used in the respective phase potential calculations.
The RI input variable, which is normally the wellbore radius, should be set to 0.0
for the case of simulating the wellbore as the first column of gridblocks. If the first
column of gridblocks is to be used for the annulus, then RI should be the outside
diameter of the tubing, and the first radius should be the inside diameter of the
casing.
The first radius specified will be the radius of the wellbore, followed by the
normal progression of radii to define the areal extent of the drainage area for the
well.
A separate equilibrium region should be defined for the wellbore [first column of
gridblocks] using the PVT data from the initially producing horizon with the
highest common datum pressure, but lowering the water level to below the
deepest productive horizon. This ensures that the wellbore is in pressure
equilibrium with and contains the fluid from the initial producing horizon.
ORKISZ Orkiszewski
All vertical flow transport parameters for the first column of gridblocks (wellbore)
are calculated dynamically for each time step, with special treatment for the phase
mobilities. These changes are as follows :
■ Phase mobilities for flow in the wellbore are based on the mixture viscosity,
instead of the phase viscosity.
■ If the gas flow velocity is greater than the minimum gas lift velocity, the
gravity gradients in the wellbore are based on the mixture density, instead of
the individual phase densities. If the gas flow velocity is less than the
minimum gas lift velocity, the damped phase gradients are used instead of the
individual phase gradients.
■ The pressure drop in the wellbore can be described as the sum of the pressure
drops due to friction, gravity, and kinetic energy. The kinetic energy pressure
loss is usually quite small and is neglected. At the start of each time step, the
gas and liquid superficial velocities are calculated based on the converged
conditions at the end of the previous timestep. From this, the Reynold’s
number and friction factor are calculated for each segment from the specified
correlation, after which the average mixture velocity and then the equivalent
Darcy velocity are calculated. The effective wellbore vertical permeability
(Kwe) is then calculated, and held constant over the timestep.
■ Phase "relative permeabilities" in the wellbore are set equal to the phase
saturations, with the exception that krl = α * Sl, where α is the liquid holdup
factor less than or equal to 1.0, which allows the gas phase to flow at an equal
or greater velocity than the liquid phase. The liquid holdup is obtained from
the available correlations.
Actual well production is from a standard well with a single specified perforation,
normally at (1,1,1), using the derived phase mobilities for the wellbore segment.
The production from each reservoir is determined by the implicit solution of the
Darcy flow from the reservoir into the wellbore.
Three additional output reports may be printed when using the WBSIM option.
These are all printed at the same frequency as the PRINT WELLS report. These
include; [1] an output table at the start of the timestep containing the effective
wellbore permeability for the step, Moody friction factor, mixture density, delta-P
friction, phase velocities, critical velocities [minimum gas flow velocities to lift
insitu liquids], Reynold’s numbers, and alpha and beta parameters for all of the
wellbore segments, [2] an output table at the end of the timestep for the wellbore
perforations, showing the production/injection rates and gas-oil ratios and
watercuts, production and injection cumulatives, and reservoir and wellbore
pressures, and [3] an output table presenting a profile of all of the wellbore
segments, containing the insitu saturations, average densities and viscosities,
alpha and beta flow parameters, effective wellbore permeabilities, and wellbore
pressures.
29
Surface Pipeline Network Options1
29.1 Introduction
The implementation of a surface pipeline network option in VIP-EXECUTIVE
is described in this chapter (also see Reference 69). Major features of this option
and areas of its applications are summarized below:
■ Bottomhole pressures and rates of production wells are determined from the
simultaneous simulation of the multi-phase fluid flow in reservoir, well
tubing, and surface pipeline network system.
■ The capabilities have been implemented for modeling the fluid flow in well
tubing. Hydraulic tables, analytical correlations, look-up tables for pressure
gradient definition, or any combinations of these tools can be used to
determine pressure drop from the wellbore to wellhead. This part of the option
can be used independently from the surface pipeline network model.
■ The surface pipeline network model can have a general “tree-like” structure
with any number of node levels but no loops. The structure of the surface
pipeline network system can be dynamically changed at any time during the
simulation.
■ The number of the hydrocarbon phases and their compositions are determined
internally using phase-equilibrium computations. A procedure has been
developed for the phase-equilibrium computations in compositional models.
This procedure is robust even near the critical point and is computationally
efficient. These features are especially important for the phase-equilibrium
computations with wide ranges of pressure and temperature values in the well
tubing models.
29.1.1 Background
The new option is designed for accurate modeling of interactions between
reservoir, production wells, and surface pipeline network system. It calculates
bottomhole pressure and rates of production wells from the simultaneous
simulation of the multiphase fluid flow in the reservoir, well tubing, and surface
pipeline network system. Any changes in a gathering network will affect
individual rates of production wells. Therefore, the production rate of any well
can be accurately defined only from an intersection of an inflow performance
curve (which is determined by the reservoir model) and an outflow performance
curve (which is defined by the surface pipeline network model). These
intersections are determined in the new option.
Production limits and targets in nodes of the gathering system may be specified.
For example, the user can specify the minimum pressure and/or maximum rates in
any node (such as gathering center, flow station, etc.) of the surface pipeline
network system and then, the simulator automatically determines rates of all
production wells connected to the node according to pipeline pressure loss
relationships.
The new option also models physical processes in different elements of the
gathering system using hydraulic models of a multiphase fluid flow in surface
network devices (such as pipelines, valves, etc.).
New capabilities are implemented for modeling the fluid flow in well tubing.
Besides hydraulic tables (which are currently available in VIP-EXECUTIVE),
analytical correlations, look-up tables for pressure gradient definition, or any
combination of these tools can be used to determine pressure drop from the
wellbore to the wellhead. This part of the new option can be used independently
from the more general surface pipeline network modeling.
The surface pipeline network model can have a general “tree-like” structure with
any number of node levels. However, loops can not be included in the model. The
structure of the surface pipeline network system may be dynamically changed at
any time during the simulation.
The following steady-state models of the multi-phase fluid flow in the surface
pipeline network system are implemented:
The number of the hydrocarbon phases and their compositions are determined
internally using phase-equilibrium computations. A new procedure is developed
for the phase-equilibrium computations in compositional models. The procedure
is robust even near critical point and computationally efficient. These features are
especially important for the phase-equilibrium computations with wide ranges of
pressure and temperature values in well tubing models.
29.1.2 Applications
The new surface pipeline network option has a wide range of applications in oil
and gas field models. Several important applications are listed below:
The following models are included in the current release of the surface pipeline
network option:
4. Valve Model
Significantly less CPU time is required if the hydraulic tables are applied instead
of the analytical models. However, the applications of the hydraulic tables have
the following disadvantages:
2. A hydraulic table should be generated for each surface network device and
each production well. Therefore, a lot of CPU memory must be allocated for
the storage of the large number of five-dimensional hydraulic tables.
Model Description
These pipe hydraulic models determine the pressure distribution (p(L)) along the
pipe assuming that the following parameters are known:
■ pressure pout at the pipe outlet L = LENGTH (where LENGTH is the total
pipe length)
dp- g- ρv dv- f ρ v 2
----- ( L ) = – ρ sin θ –
---- -----
- ----- – -------------- (29-2)
dL gc g c dL 2g c d
where:
dp-⎞
⎛ ----- g
= ----- ρ sin θ component of the pressure gradient due to a
⎝ dL⎠ elev gc potential energy or elevation change,
g gravitational constant
ρ fluid density
v fluid velocity
dv-
----- fluid velocity gradient.
dL
2
dp-⎞
⎛ ----- fρv
⎝ dL⎠ frict = – -------------
2g c d
- component of the pressure gradient due to
friction losses,
d Flow P(L)
dL
dP
P ( L – dL ) ≈ P ( L ) – ------- dL
dL
Solution Algorithm
1. Start with known pressure p(L) = pout at the pipe outlet L = LENGTH.
3. Define densities and volumetric rates of the oil, gas, and water phases at the
current pressure, p(L), from the equation-of-state (see Chapter 23).
4. Determine viscosities of the oil, gas, and water phases using Lohrenz-Bray-
Clark28 or Pedersen37 correlations. The techniques are similar to those
applied in VIP-EXECUTIVE for the viscosity definitions at reservoir
conditions (see Chapter 23).
6. Determine velocities of the oil, gas, and water phases as respective volumetric
rates divided by a cross-sectional area of a pipe.
dp
7. Calculate a pressure gradient, ------ ( L ) , at the current pipe location, L , from
dL
Equation 29-2. The major problem in this part of the calculation is the
determination of the following properties of the fluid mixture which consists
of the oil, gas, and water:
❑ friction factor f,
❑ density ρ,
❑ viscosity,
❑ velocity v
These parameters are included in the right-hand side of Equation 29-2. The
following correlations are implemented in VIP-EXECUTIVE for the
determination of the specified properties of the fluid mixture:
❑ Orkiszewski,
1. Select an integration interval to assure that the pressure drop in this interval
is not larger than the user specified value PRESIN and that L1 = L - ΔL is not
smaller than zero:
dp-
----- ( L )ΔL ≤ PRESIN, L – ΔL ≥ 0
dL
2. Calculate pressure, p(L0.5), at the middle of the interval ΔL as
dP
p ( L 0.5 ) = p ( L ) – 0.5 ------- ( L )ΔL where L0.5 = L - 0.5 ΔL.
dL
dp
3. Repeat Steps 2 - 7 to calculate pressure gradient ------ ( L 0.5 ) at the location
L0.5. dL
Model Description
The hydraulic model described by the hydraulic table determines pressure at the
inlet of the flow device assuming that the following parameters are known:
■ pressure and temperature used for the determination of volumetric rates in the
hydraulic table,
Solution Algorithm
The solution procedure for determining the pressure drop in a flow device,
described by a hydraulic table, is as follows:
1. Specify values of pressure and temperature at the outlet of the flow device.
3. Define volumetric rates of oil, gas, and water phases from the equation-of-
state (refer to Chapter 23, PVT Representation).
4. Calculate gas-oil ratio and water cut. (Note: this step is required only if the
five-dimensional table is used.)
5. Determine pressure at the inlet of the flow device from the hydraulic table.
The phase-equilibrium computations and table look-up are done only once in this
procedure. Therefore, the hydraulic table option requires significantly less CPU
time than the analytical models described in Section 29.2.1.
■ reduce the CPU time required by the analytical models in determining the
pressure distribution in pipes,
In this case, the hydraulic table (BHPTAB) defines pressure (BHP) at the
beginning of a pipe interval with the specified length (dzw) as a tabular function
of pressure (THP) at the end of this interval, liquid rate (QLIQ), gas-oil ratio
(GOR), water cut (WCUT), and inclination angle (ALQ). Alternatively, oil or gas
rate (QO or QG) may be used instead of the liquid rate, and gas-liquid or oil-gas
ratios (GLR or OGR) may be used instead of the gas-oil ratio, and water-gas ratio
(WGR) may be used instead of the water cut.
dp-
( L ) = THP
----- – BHP-
-----------------------------
dL dzw
Model Description
The valve model determines pressure at the inlet of a valve assuming that the
following parameters are known:
■ valve control,
The model predicts subcritical pressure drop across the valve using the following
equation50,51:
Q tot Q tot
p out – p in = – CVX ( X ) ---------------------- (29-3)
ρ
where:
Qtot total mass rate of the fluid in lb. per sec., which is
calculated as a sum of the mass rates of the oil, gas, and
water phases,
The pressure control valve model automatically selects the valve coefficient CVX
from a specified range. This coefficient minimizes the absolute value of the
difference between valve upstream pressure and target pressure. The valve
coefficient range is defined by the valve coefficient profile and valve status. If the
valve status is “OPEN”, the whole range of the valve coefficient profile is used in
the calculations. Otherwise, if the valve status is “CLOSE”, the valve coefficient
range only consists of the last entry in the valve coefficient profile.
In the rate control valve model, the valve coefficient is defined as one of the end
points of the valve coefficient profile. If the valve status is “OPEN”, the valve
coefficient is determined as the first entry in the valve coefficient profile.
Otherwise, if the valve status is “CLOSE”, the valve coefficient is selected as the
last entry in the valve coefficient profile.
The oil, gas, or liquid mass rate can be used (instead of the total mass rate) for the
definition of the pressure drop in a valve.
Solution Algorithm
The solution procedure for determining the pressure drop in a valve is as follows:
1. Specify values for pressure and temperature at the outlet of the valve.
3. Determine mass rates and densities of oil, gas, and water phases from the
equation-of-state.
■ Wellbore Model, which describes the multi-phase fluid flow from a reservoir
to wellbores of production wells;
■ Well Tubing Model, which simulates the flow of multi-phase fluids from
wellbores to wellheads;
The wellbore model defines an inflow performance curve of each production well.
The well tubing and surface facility models determine an outflow performance
curve of a well. The bottomhole pressure and rates of each production well are
calculated as an intersection of the inflow and outflow performance curves.
The wellbore, well tubing, and surface facility models are considered below.
Wellbore Model
■ bottomhole pressure,
In the surface pipeline network option, the conditions that exist in reservoir blocks
at the beginning of a timestep are used for the determination of the well inflow
performance. Therefore, the molar rates of the hydrocarbon components and the
water rate at the stock tank conditions of each production well are considered as
functions of the bottomhole pressure only:
QWw = QWw(PBHw),
where:
Models of flow devices, links, and nodes are used for the simulation of the multi-
phase fluid flow in production well tubing and surface pipeline network system.
Each link has only one inlet and one outlet. A link can be constructed from
several flow device models sequentially connected with each other. For example,
the link shown in Figure 29-2 consists of the four following flow device models:
tubing, valve, pipe, and hydraulic table (BHPTAB).
A node is the junction of several links. Each node can have any number of input
links but only one output link. The production wells can be connected to any node.
tubing and in connections between nodes. For example, the multi-phase fluid flow
from a wellbore to a wellhead of Well 5 is simulated by Link 3. Link 5 represents
the flow between the wellhead of Well 5 and Node 4. The flow between Node 4
and Node 5 is simulated by Link 6.
Link 6 Node 5
Node 4
Link 5
Node 2 Node 3
Node 1
Wellhead Wellhead
Link 3
Mass conservation and pressure equations describe the fluid flow in the surface
pipeline network system. These equations are described below.
The mass conservation equations for the hydrocarbon components and water
phase are formulated for each node of the surface pipeline network system.
The mass conservation equation for a hydrocarbon component states that the
molar rate of this component in an output link of a node is equal to the sum of
molar rates of the component in the input links of the node:
where:
The mass conservation equations of the water phase can be formulated as follows:
QW n = ∑ QW j + ∑ QW w, n = 1, 2, …, nnodes (29-6)
j ⊂ Nn w ⊂ Wn
where:
QWw water rate at stock tank conditions in the w-th well (see
Equation 29-4).
The pressure equations are formulated for each link. They determine the
pressure at the inlet of the link as a function of the pressure at the outlet of the
link, molar rates of hydrocarbon components and water rate in the link.
For a link connecting the j-th and n-th nodes, the pressure equation can be
formulated as follows:
P j = F jn ( P n, QC j, QW j ) (29-7)
where:
For a link connecting a wellbore and a wellhead of a production well, the pressure
equation is as follows:
For a link connecting a wellhead of a production well and a node, the pressure
equation is as follows:
THP w = F wn ( P n, QC w, QW w ) (29-9)
where:
The models of the flow devices from which the link is constructed are used for the
derivations of the pressure Equations 29-7, 29-8, and 29-9. These models are
described in Section 2.
Pressure Constraints:
Rate Constraints:
where:
Qon, Qgn, Qwn, Qln oil, gas, water, and liquid volumetric rates at the n-th node
The user assigns each node of the surface pipeline network system to a separator
battery. The volumetric rates of oil, gas, and water at any node are calculated by
flashing the feedstream through the separator battery attached to the node.
■ bottomhole pressure,
■ tubinghead pressure,
■ oil, water, and gas production rates at the stock tank conditions
■ pressure distribution,
subject to pressure and rate constraints (Equation 29-10) and (Equation 29-11).
The procedure developed for the solution of this problem is described in the next
section.
At the first stage, the procedure honors pressure and rate constraints Equations 29-
10 and 29-11 at nodes of the highest level of the surface pipeline network system.
The pressure and rate constraints Equations 29-10 and 29-11 at nodes of the lower
levels are disregarded at this stage. Then, the node level is reduced by one and the
procedure is repeated.
The following steps are executed in the surface pipeline network option for the
determination of the bottomhole pressure and rates of the production wells as well
as the pressure distribution in the surface pipeline network system:
1. Set node level (NL) to the highest node level in the surface pipeline network
system.
3. Determine molar rates of the hydrocarbon components and water rate at the
stock tank conditions of each production well from the wellbore flow
(Equation 29-4).
4. Calculate molar rates of the hydrocarbon components and water rates at the
stock tank conditions from the material balance Equations 29-6 and 29-7 at
nodes of all levels smaller than or equal to NL.
5. Determine volumetric rates of the oil, gas, and water phases at all nodes of the
NL-th level using flash calculations in separator batteries attached to the
nodes.
6. If rate constraints (Equation 29-11) are not honored at some nodes of the NL-
th level, calculate bottomhole pressures of the production wells and pressures
at the nodes from the solution of Equations 29-4 through 29-9 and the
following constraints:
where NQNL is the subset of nodes from the NL-th level. Rate constraints
(Equation 29-11) with some index p are violated at these nodes.
7. If pressure constraints (Equation 29-10) are not honored at some nodes of the
NL-th level, calculate bottomhole pressures of the production wells and
pressures at the nodes from the solution of Equations 29-4 through 29-9 and
the following constraints:
P n = PMAXn, n ⊂ NP NL (29-13)
where NPNL is the subset of nodes from the NL-th level. Pressure constraints
(Equation 29-10) are violated at these nodes.
8. Reduce the node level NL by one and repeat Steps 3-7 if NL is not smaller
than one.
3. Apply these rates as the well maximum production rates in the other outer
iterations of the timestep.
The frequency of the surface pipeline network calculations can be controlled. The
user can specify a time increment between two sequential recalculations of the
production rates that use the surface pipeline network calculations. However, the
production rates are redefined after each TIME or DATE card.
30
Total Compressibility Checks
30.1 Introduction
VIP-CORE performs several checks on black-oil input data. Each BOTAB table is
checked after it is read. Compositional input data are not checked. Validation of
the BOTAB data is divided into three parts:
BOTAB 1
API WTRO
27.5 200.
PSAT RS BO ZG GR VO VG
5000 930 1.48 .88 .80 .72 .028
Saturated 4000 720 1.37 .85 .80 .90 .025
oil data 3000 541 1.30 .80 .80 1.07 .022
2000 378 1.24 .81 .80 1.32 .018
1000 212 1.17 .86 .80 1.81 .014
14.65 0 1.05 .96 .80 3.34 .012
PSAT 3000. 4000. . .
DP BOFAC VOFAC BOFAC VOFAC . .
Under-saturated 100. .990 1.01 .995 1.02 . .
oil data 200. .980 1.02 .994 1.04 . .
300. .970 1.03 .993 1.06 . .
. . . . . . .
1. The BOTAB table entries must be entered with decreasing pressure. That is,
the first entry is at the maximum pressure in the table and the last entry is at
“standard” pressure.
3. The oil formation volume factor must decrease with decreasing pressure.
( B g ΔR s – ΔB o ) > 0 (30-1)
where ΔRs and ΔBo are the differences between two consecutive entries in the
BOTAB table. Bg is evaluated at the lower pressure of the two and is multiplied by
0.001 to yield compatible units. If the user entered a gas deviation factor (Z)
instead of Bg, then (for field units) the Z-factors are converted to Bg using the
following equation:
This check is suppressed if the user includes a NOCHK card in the UTILITY
section of the input file. This is not a good idea because the simulator may have
trouble running with data that failed the check which implies that the total
hydocarbon system expands with increasing pressure.
B o ( P sat + DP ) μ o ( P sat + DP )
BOFAC ( DP ) = -----------------------------------
- VOFAC ( DP ) = -----------------------------------
- (30-3)
B o ( P sat ) μ o ( P sat )
1. All PSAT values must be greater than standard pressure and increasing.
3. For each PSAT value, BOFAC values must be less than one and must decrease
with increasing DP.
4. For each PSAT value, VOFAC values must be greater than one and must
increase with increasing DP.
31
Tracking Calculations In VIP-EXECUTIVE
31.1 Introduction
There are several possible methods to perform fluid tracking in compositional
simulation. The method that is implemented in VIP-EXECUTIVE is designed to
track so-called equity fluids. Equity fluid is a legal concept associated with
ownership. The fluid tracking algorithm in VIP-EXECUTIVE is, therefore,
designed to determine oil, gas, and condensate reserves of a reservoir if the
reservoir is owned by several companies.
The equilibrium state of a gas cap, before start of production, presents a good test
example of a system to help identify a set of requirements for a tracking algorithm
that is consistent with equity ownership considerations. Note the following
requirements:
2. At any time and at any location in the reservoir, an equity (tracked) fluid may
be present in one of the two phases (oil, gas) or in both of them.
3. The sum of moles of all tracked fluids should equal total moles as calculated
by the simulator for each of the gas and oil phases. This should be satisfied
locally in any point of the reservoir and for each hydrocarbon component.
5. Transfer of moles of an equity fluid from one phase to another occurs only in
the case of a non-zero net mass transfer of moles between phases.
These six requirements are the basis for formulation of the tracking algorithm.
The first three of these assumptions clarify how an equity fluid is characterized
within the tracking procedure. The equity fluid is described by the number of
moles of each hydrocarbon component associated with this fluid in both the oil
and gas phases at any location in a reservoir.
Assumption 3 requires that oil and gas phases be comprised of all tracked fluids.
In other words, a list of tracked fluids must be complete to include all
hydrocarbon material in a reservoir.
it is present. An equity fluid also may move from one phase to another. By
definition (Assumption 5) such transfer occurs only in the case of a non-zero net
mass transfer of hydrocarbon components between phases.
Do not confuse these variables with the mole fractions of component i of fluid j.
These are not mole fractions. For a given fluid j, the sum of xijT over all
components i is not equal to one. It is equal to the total moles of fluid j in one
mole of oil.
Using these variables, requirement 3 from Section 31.2 may be written as follows:
n fl n fl
T T
xi = ∑ x ij, yi = ∑ yij (31-1)
j=1 j=1
Let uo be molar flux; i.e., uo is equal to the number of oil moles that flows through
a unit area per unit time. Each of these moles contains xi moles of component i.
Therefore, the oil molar flux of component i, uoi, is defined as uoi = uo xi. Similar
considerations can be applied to define molar fluxes of tracked substances. The oil
molar flux of component i of fluid j, uoij, is defined as uoij = uo xijT, and the gas
molar flux of component i of fluid j, ugij, is defined as ugij = ug yijT. It is easy to
show that the following is valid:
n fl n fl
o o g g
ui = ∑ u ij, ui = ∑ uij (31-2)
j=1 j=1
∂ o g
---- { φF [ Vy i + ( 1 – V )x i ] } + ∇ ⋅ ( u x i ) + ∇ ⋅ ( u y i )
∂t
p p I I I I
–qo xi – qg yi + qo xi + qg yi = 0 (31-3)
∂
---- [ φF ( 1 – V )x i ] + ∇ ⋅ ( u o x i ) – q po x i + q Io x Ii = T i (31-4)
∂t
∂ g p I I
---- [ φFVy i ] + ∇ ⋅ ( u y i ) – q g x i + q g y i = – T i (31-5)
∂t
The first of these equations describes material balance of component i in the oil
phase, while the second describes the gas phase. Here, Ti is the mass transfer term.
It is defined by Equations 31-4 and 31-5. Ti has the physical meaning of the
number of moles of component i that are condensed from gas into oil in a unit
volume per unit time.
Material balance equations for equity fluids are written in the form similar to
Equations 31-4 and 31-5:
∂ T o T p T I TI
---- [ φF ( 1 – V )x ij ] + ∇ ⋅ ( u x ij ) – q o x ij + q o x ij = T ij (31-6)
∂t
∂ T g T p T I TI
---- [ φFVy ij ] + ∇ ⋅ ( u x ij ) – q g y ij + q g y ij = – T ij (31-7)
∂t
Here, xijTI and yijTI are values similar to xijT and yijT, but are for the injected
streams. Composition and makeup of injected fluids in terms of tracked
substances are assumed to be known. Tij is a mass transfer term for tracked fluid j,
and defines the number of moles of component i of tracked fluid j that condenses
from gas into oil in a unit volume per unit time.
Equations 31-6 and 31-7 must be consistent with the mass transfer Equations 31-4
and 31-5, which are used in the simulator. This means that the equations that result
from summation of Equations 31-6 and 31-7 over all tracked fluids must be
∑ Tij = Ti (31-8)
j=1
Equation 31-8 is the only condition imposed on mass transfer terms Tij. This
condition does not uniquely define the mass transfer terms. Some additional
assumptions must be made to define Tij and close Equations 31-6 and 31-7.
Both oil and gas consist of several tracked substances. For condensation, Ti moles
of component i move from the gas into the oil. When deciding what part of these
moles comes from fluid j, one may look at the relative presence of fluid j in the
gas phase. One mole of gas has yi moles of component i and yijT moles of
component i are associated with fluid j. The relative amount of fluid j moles of
component i in the gas then is given by the ratio yijT/yi. It is reasonable to assume
the following expression for the mass transfer term of component i of fluid j:
T
y ij
T ij = T i ----- for condensation. (31-9)
yi
This definition of the mass transfer term is based on the assumption of equal
opportunity for tracked substances to participate in a mass transfer process. It
satisfies the consistency condition of Equation 31-8.
Note that there is no unique way to define mass transfer terms, Tij. Equations 31-9
and 31-10 provide one of many possible assumptions. For example, the following
definition of the mass transfer terms also satisfies the consistency condition:
g T
w j y ij
-,
T ij = T i ----------------------
n
if T i > 0 (31-11)
fl
g T
∑ wk yik
k 1
o T
w j x ij
-,
T ij = T i ----------------------
n
if T i < 0 (31-12)
fl
o T
∑ wk xik
k 1
Here, wjo and wjg are weighting factors that must be provided by the user. For
equal weighting factors, this definition of the mass transfer terms is reduced to the
previous one. Therefore, Equations 31-11 and 31-12 may be viewed as a
generalization of Equations 31-9 and 31-10. By assigning unequal weighting
factors, we have preferential evaporation or condensation of certain fluids relative
to other fluids.
for the tracked fluid molar flow rates used in Equations 31-6 and 31-7. Those
expressions may be interpreted as an assumption of the uniform distribution of
equity fluids within the oil and gas phases. And if that is the case, then the
preferential vaporization or condensation contradicts this uniform distribution of
the tracked fluids. In this regard, note that Equations 31-6 and 31-7 and
expressions 31-11 and 31-12 are used on the scale of gridblocks that is many
orders of magnitude larger than the pore scale. The processes of mass transfer
between the phases and of fluid flow actually occur on the pore scale. What looks
like uniform tracked fluid distribution within the phases on the macroscopic scale
of gridblocks may not be true on the subgrid scale of individual pores. Therefore,
the possibility of preferential vaporization or condensation of tracked fluids and
Equations 31-6 and 31-7 are not necessarily inconsistent.
The definition Equations 31-11 and 31-12 of the mass transfer terms actually are
implemented in the tracking algorithm of the VIP-EXECUTIVE simulator. This
definition provides an option to account for possible preferential evaporation or
condensation of some equity fluids. However, it is the user’s responsibility to
justify any specific set of weighting factors used in a simulation.
calculations within a reservoir. Tracking Equations 31-6 and 31-7 are not stand-
alone equations. They are coupled with the simulator equations through the mass
transfer term Ti. Whether the tracking equations are solved or not has no effect on
the simulator equations. The opposite, however, is not true. To solve the tracking
equations one needs a mass transfer term Ti , which comes from the solution of the
simulator Equations 31-4 and 31-5.
32
Tracer Analysis Option1
32.1 Summary
This is a description of an option which has been developed to improve and
expand the simulation capabilities in the analysis and interpretation of tracer tests
and in the design and performance analysis of waterflood and gas injection
projects. The option is built around the particle tracking method which allows
accurate simulation of tracer flow associated with convection and physical
dispersion. The method is nearly numerical dispersion free and allows accurate
simulation of tracer flow in field scale simulation. The algorithm, implemented in
VIP-EXECUTIVE, allows simulation of tracer flow within the framework of
three-dimensional, multi-phase, non steady state reservoir simulation. In addition
to accurate simulation of tracer flow the software allows:
32.2 Background
Waterflooding is the most widely used method of secondary oil production. A
waterflood project usually involves a number of water injectors and producers
arranged in some pattern. Optimal recovery of oil in such an operation requires
coordinated operation of water injectors and producers. This insures that oil is
displaced by water uniformly throughout the reservoir and there is no early water
breakthrough in some wells while large portions of the reservoir are not swept by
water. Such operation of a waterflood project requires an extensive reservoir
surveillance program as well as some specialized reservoir simulation tools which
help reservoir engineers understand the performance of the waterflood.
Surprisingly, the state of the art reservoir simulation technology is very rarely
used in this area of reservoir engineering. There are no real technical reasons for
this. With existing commercial reservoir simulators and with the hardware
capabilities available, it is possible to accurately simulate oil displacement by
water and predict performance of a waterflood project by taking into account
actual reservoir geometry, fluid and rock properties and large scale heterogeneities
as well as actual operational conditions. Reservoir simulation performed with a
commercial reservoir simulator on a fine enough grid system provides more
accurate results than the estimates made by using simplified models.
Displacement of oil by water may be adequately simulated in most cases on a grid
with ten blocks between wells and such models will be able to represent the most
important large scale heterogeneities. There is, however, a process often
associated with waterflood surveillance programs which commercial reservoir
simulators are not able to simulate adequately. This process is a tracer test.
In a field tracer test, a small slug of tracer dissolved in water or gas is injected into
a reservoir. It is followed by water or gas, respectively. Production of tracer is
monitored in surrounding production wells.
∂φS p ρ p c
--------------------- + ∇ ( q p ρ p c ) = ∇ ( D∇c ) + R p ρ p c (32-1)
∂t
However, the same solution method can not be used for the tracer flow equation.
A slug of tracer may have a scale of a few feet (if not inches). If we solve the
tracer transport equation by a finite difference method, we have to use a grid with
grid blocks which are small compared to the scale of the tracer slug. It is not
feasible to use finite difference grids with foot-size grid blocks or even less for
field scale applications. Using a grid system with larger grid blocks will result in
large truncation errors associated with the finite difference approximation of the
tracer transport equation. This will manifest itself in a numerical phenomenon
called numerical dispersion.
The tracer option presented here is based on a different method of solving the
tracer transport equation. This is not a finite difference type method. We simulate
tracer flow by a method of particles. Particle methods are best suited for purely
convective problems. These methods have been used to solve convective
problems in fluid mechanics, kinetic theory, and plasma physics. It has been
demonstrated that the particle methods can produce accurate solutions with a very
small amount of numerical dispersion. This approach was later extended to
problems with dispersion. For example, it was used to solve the Navier-Stokes
equations at high Reynolds number.88 This problem combines both convection
and dispersion. Recently, a particle method has been used in oil industry for
During simulation, results from the tracer option of VIP are written into a separate
output file. The frequency and amount of information sent to this file are
controlled by the user. The simulator writes two types of records: map records and
tracer production records. A map record contains information on the locations of
individual tracer particles at specific moments in time. This information can be
used for displaying the distribution of the tracer within the reservoir. A production
record contains information on tracer production rates for individual wells. The
simulator reports production rates for a well and for individual perforations of the
well as a function of time.
This option has a wider area of application. For example, in addition to its main
application, accurate simulation of tracer flow, it can be used to track water fronts
in waterflood operations. To do this we place particles at a water front when a well
starts water injection and move the particles with the velocity of the front
movement. The water front moves according to the Buckley-Leverett theory. This
front tracking capability is also implemented in VIP-EXECUTIVE. Tracking of
water front allows one to calculate areal sweep.
1
u
v
4 2
w
3
5
6
8
In corner-point option, each cell is defined by the coordinates of its eight corners
r ( x, y, z ) i (i =1,...,8)
r = r 1 + ur 21 + vr 41 + wr 51 + uv ( r 32 – r 41 ) + vw ( r 84 – r 51 ) + uw ( r 62 – r 51 )
+ uvw [ ( r 21 – r 34 ) – ( r 65 – r 78 ) ]
where
r ij = r i – r j
In non-well cells, particle movement calculations are based on fluids flow from
each cell face
Vp = Vx nx + Vy ny + Vz nz .
Velocity quantities are linear interpolations of phase velocities across the cell face
V x = ( 1 – u )v x- + uv x+
V y = ( 1 – v )v y- + vv y+
V z = ( 1 – w )v z- + wv z+
⎧ n x1 = n 21 ⎧ n y1 = n 41 ⎧ n z1 = n 51
⎪ ⎪ ⎪
⎪ n x2 = n 65 ⎪ n y2 = n 85 ⎪ n z2 = n 84
⎨ , ⎨ , and ⎨
⎪ n x3 = n 78 ⎪ n y3 = n 76 ⎪ n z3 = n 73
⎪ ⎪ ⎪
⎩ n x4 = n 34 ⎩ n y4 = n 32 ⎩ n z4 = n 62
In well cells, particle movements are decided by fluids flow from both wells and
cell faces. For each cell, a reference circle is first defined, which is the largest
circle without crossing cell boundaries or corner nodes on the plane perpendicular
to the well perforation. The radius and the location of this circle is then used as a
reference to calculate the particle radial velocities.
⎧ 2
⎪ xr = xw ±Rr ⁄ 1 + k
⎨
⎪ y = y ±R k ⁄ 1 + k2
⎩ r w r
where
k = ( yp – yw ) ⁄ ( xp – xw )
The velocities contributed from cell faces at the reference points V b are
calculated in the same way as for particles in non-well cells with the r components
V br and theta components V bθ .
The velocities contributed from the wells are
2
q log ( 1 + 2qΔt ⁄ αHr )
V wr = ---------------------------------------------------------------------- .
2 2
2αH ( r + 2qΔt ⁄ αHr – r )
where q is the flow rate, α is angle open to the well, r is the particle distance to the
well center, H is the cell average height, and Δt is the timestep size for particle
movement.
r c = q ⁄ ( αHV br )
to measure the relative impact from the well and from cell boundaries.
When particles move close to well, well impact becomes significant while
boundary effect is diminishing, the radial velocities are calculated by
⎧
⎪ VV br R ⁄ r r ⁄ rc > 1
⎪
V r = ⎨ wV b R ⁄ r + ( 1 – w )V wr r ⁄ r c ∈ ( τ, 1 ]
⎪
⎪ V wr r ⁄ rc ≤ τ
⎩
where
w = ( r ⁄ rc – τ ) ⁄ ( 1 – τ )
The theta components of the velocities are linearly distributed along the r
direction and zero when particles are in the vicinity of the wells.
⎧
⎪ V ( r ⁄ rc – τ ) ⁄ ( R ⁄ r c – τ ) r ⁄ r c > τ
V θ = ⎨ bθ
⎪ 0 r ⁄ rc ≤ τ
⎩
Particle reflections
Particles reflect while they hit the non-flow cell faces. The transformation
function from physical space to unit cube space can be written in general as
r = x ( u, v, w )i + y ( u, v, w )j + z ( u, v, w )k
where i , j , and k are the unit vectors of x, y, and z coordinates in physical space.
Each cell face can be expressed as
r x- = x ( 0, v, w )i + y ( 0, v, w )j + z ( 0, v, w )k , r x + = x ( 1, v, w )i + y ( 1, v, w )j + z ( 1, v, w )k
r y- = x ( u, 0, w )i + y ( u, 0, w )j + z ( u, 0, w )k , r y+ = x ( u, 1, w )i + y ( u, 1, w )j + z ( u, 1, w )k
r z- = x ( u, v, 0 )i + y ( u, v, 0 )j + z ( u, v, 0 )k , r z+ = x ( u, v, 1 )i + y ( u, v, 1 )j + z ( u, v, 1 )k .
The unit normal vector for each face are then calculated by
∂r x- ∂r x- ∂r x - ∂r x- ∂r x+ ∂r x+ ∂r x+ ∂r x +
n x- = --------- × --------- ⁄ --------- × --------- , n x+ = --------
- × --------- ⁄ --------- × --------- ,
∂w ∂v ∂w ∂v ∂v ∂w ∂v ∂w
∂r y- ∂r y- ∂r y - ∂r y- ∂r y+ ∂r y+ ∂r y+ ∂r y +
n y- = --------- × --------- ⁄ --------- × --------- , n y+ = --------
- × --------- ⁄ --------- × --------- ,
∂u ∂w ∂u ∂w ∂w ∂u ∂w ∂u
∂r z- ∂r z- ∂r z- ∂r z- ∂r z+ ∂r z+ ∂r z+ ∂r z+
n z- = -------- × -------- ⁄ -------- × -------- , n z+ = --------
- × --------- ⁄ --------- × --------- .
∂v ∂u ∂v ∂u ∂u ∂v ∂u ∂v
v' = v – 2 ( n ⋅ v )n
for a particle that has a velocity v and hits a point where the face normal is n .
r = r 0 + Δtv
where v is the particle velocity, Δt is the current timestep, and r 0 is the particle
start point. This calculated position is accepted as the particle new location if r
and r 0 are in the same cells.
If the particle moves outside the current cell, a set of nonlinear equations is solved
for u, v, w , and Δt 0 for each face
x ( u, v, w )i + y ( u, v, w )j + z ( u, v, w )k – vΔt 0 = r 0
-
x : u = 0, v ∈ [ 0, 1 ], w ∈ [ 0, 1 ], Δt 0 > 0 ,
+
x : u = 1, v ∈ [ 0, 1 ], w ∈ [ 0, 1 ], Δt 0 > 0 ,
-
y : u ∈ [ 0, 1 ], v = 0, w ∈ [ 0, 1 ], Δt 0 > 0 ,
+
y : u ∈ [ 0, 1 ], v = 1, w ∈ [ 0, 1 ], Δt 0 > 0 ,
-
z : u ∈ [ 0, 1 ], v ∈ [ 0, 1 ], w = 0, Δt 0 > 0 ,
+
z : u ∈ [ 0, 1 ], v ∈ [ 0, 1 ], w = 1, Δt 0 > 0 .
The results tell the location on cell face that particle has crossed or reflected,
travel time for particle reaching the face, and the new cell that particle will move
into. The tracking calculation continues based on the new cell and the remaining
time left from the current timestep.
In the case of a non-zero dispersion, the original tracer spike disperses in space
while it moves through the model. This causes the tracer to be produced over
some period of time. Figure 32-1 shows a comparison of the tracer production rate
as calculated by the particle method using 10,000 particles with the analytical
solution which exists in this case. The particle solution reproduces the analytical
result reasonably well. Oscillations, which are present in the particle solution, are
the result of a discrete representation of the tracer slug by particles. An increase of
the number of particles reduces the level of noise in the tracer production curve.
This is illustrated by Figure 32-2 which shows the tracer response for two slugs of
tracer represented by 10,000 and 500 particles. The level of noise is much higher
for 500 particles.
Table 32-1: Two Dimensional, Single Phase, 5-Spot Model. Comparison With
Analytical Solution
The tracer production rate as a function of time for a 5-spot pattern is shown in
Figure 32-4. The tracer production curve has a characteristic shape of a sharp
spike followed by a long tail. This shape is the result of the 5-spot geometry.
Figure 32-4 also illustrates the effect of grid system on tracer production. It shows
that refinement of the grid slightly affects the breakthrough time.
Figure 32-5 shows the effect of the number of particles representing a tracer slug
on tracer production. The figure gives the tracer production curves from three
slugs of tracers which are injected with an interval of 30 days. The tracers arrive at
a producer with the same interval of 30 days. The slugs are represented by 500,
1,000, and 2,000 particles. Figure 32-5 shows that an increase in the number of
particles reduces the level of noise in tracer production, especially in the tail part
of the curve.
A tracer production curve for the same model with dispersion (dispersivity is 0.5
ft.) is given in Figure 32-6. Physical dispersion causes earlier tracer breakthrough.
Instead of the sharp spike of Figure 32-4 the tracer production curve in Figure 32-
6 shows a significantly lower and wider peak.
33
Transmissibility Calculations
33.1 Introduction
Several methods of calculating transmissibilities are used in VIP-EXECUTIVE.
The method is chosen based on the grid system defined by the user. This section
describes the techniques used to calculate the inter-block transmissibilities when
the grid system is either Cartesian or radial. Corner-point geometry
transmissibility calculations are described in Chapter 3 and Appendix B in this
manual.
2.0 × C Darcy × TM i, j, k
T z = ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
DZB ij, k – 1 DZB i, j, k
---------------------------------------------------------------------------------------------- + ------------------------------------------------------------------------ -
KZ i, j, k – 1 × DXB i, j, k – 1 × DYBi, j, k – 1 KZ i, j, k × DXB i, j, k × DYB i, j, k
Note that all of the above equations use bedding plane lengths; namely, DXB,
DYB, DZB, and DZBN. If the input data were not entered in bedding plane
lengths, then the bedding plane lengths are calculated from the horizontal and
vertical lengths (see Figure 33-1).
DXB
DZB
DZ θx
DX
DXB DX / COS(θx),
DYB DY / COS(θy),
where:
If cell center depths, MDEPTH, are entered instead of gridblock top depths,
DEPTH, then the MDEPTH values are used in the above calculations.
The simulator removes the existing standard connections that apply across the
fault.
An average gross thickness for each layer is calculated for the two columns
separated by the fault. Thus, it is assumed that DZ is constant at the gridblock
face.
The shared gross thickness is calculated for each pair of cells connected across the
fault. This is based on the average gross thickness calculated in Step 2 above and
the fault throw (or shift) entered on the FX or FY card. Figure 33-3 describes this
process.
2 × H shared × C Darcy × TM i, j, k
T x = -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-
DXBi, j, k DXB i – 1, j, k
------------------------------------------------------------------------------
1
- + ---------------------------------------------------------------------------------------------------
2
-
KX i, j, k × DYB i, j, k × NTG i, j, k KX i – 1, j, k × DYBi – 1, j, k × NTG i – 1, j, k
1 1 1 2 2 2
2 × H shared × C Darcy × TM i, j, k
T y = -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-
DYB i, j, k DYB i, j – 1, k
------------------------------------------------------------------------------- + ----------------------------------------------------------------------------------------------------
1 2
The fault option does not allow permeabilities to be specified at the gridblock
faces; i.e., KXF, KYF, and KZF. A general description of faults, including input
requirements, is found in Chapter 7.
Average
4-ft Block Thickness,
1 Pairs ft
Fault
1 and 4 3
2
2 and 5 3
4
Shared
3 5
Block Thickness,
Pairs ft
6 2 and 4 2
3 and 4 1
3 and 5 3
Nine point transmissibilities are calculated by the method of Coats and Modine33.
The method rigorously accounts for non-heterogeneity and non-uniform grids and
insures that all computed transmissibilities are non-negative. The method is
applied in the x-y plane only. The option is invoked by the NINEPT card in the
VIP-CORE utility data, and may be used only for non-corner point cartesian grids.
Nine point differencing is not used across faults. Fault connection
transmissibilities are calculated as described in Section 33.2.2.
3 7 4
5 6 9
8
1 2
For each four quarter block system in the x-y plane in the reservoir, the
transmissibilities T12, T13, T34, T24, T14, and T23 are calculated. The diagonal
transmissibilities T14 and T23 are the full values between the respective
gridblocks. The total axial direction transmissibilities are given by the sum of the
values for neighboring four quarter-block systems. For example, the total x-
direction transmissibility between blocks 1 and 2 (denoted Tx2 in the simulator) is
given by the sum of the values represented by the two arrows in Figure 33-3. If
blocks 1 and 2 are on the reservoir boundary, then the transmissibility represented
by the lower arrow is equal to one half of the conventional value computed as in
Section 33.2.1. The transmissibilities for each four quarter block system are given
by
t 16 ⋅ t 26 t 36 + t 46
T 12 = τ 12 – w ⋅ ⎛ -------------------⎞ ⋅ ⎛ --------------------------------------------⎞
⎝ t 16 + t 26⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
t 16 ⋅ t 36 t 26 + t 46
T 13 = τ 13 – w ⋅ ⎛⎝ -------------------⎞⎠ ⋅ ⎛⎝ --------------------------------------------⎞⎠
t 16 + t 36 t 16 + t 26 + t 36 + t 46
t 36 ⋅ t 46 t 16 + t 26
T 34 = τ 34 – w ⋅ ⎛ -------------------⎞ ⋅ ⎛ --------------------------------------------⎞
⎝ t 36 + t 46⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
t 26 ⋅ t 46 t 16 + t 36
T 24 = τ 24 – w ⋅ ⎛ -------------------⎞ ⋅ ⎛ --------------------------------------------⎞
⎝ t 26 + t 46⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
t 16 ⋅ t 46
T 14 = w ⋅ ⎛ --------------------------------------------⎞
⎝ t 16 + t 26 + t 36 + t 46⎠
t 26 ⋅ t 36
T 23 = w ⋅ ⎛ --------------------------------------------⎞
⎝ t 16 + t 26 + t 36 + t 46⎠
where:
w = MIN ( 4 ⁄ 3, w 1, w 2, w 3, w 4 )
2---
⋅ τ 12
3
w 1 = -------------------------------------------------------------------------------
t 16 ⋅ t 26 ⎞ ⎛
⎛ ------------------ t 36 + t 46
- ⋅ ------------------------------------------- -⎞
⎝ t 16 + t 26⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
2---
⋅ τ 13
3
w 2 = -------------------------------------------------------------------------------
t 16 ⋅ t 36 ⎞ ⎛
⎛ ------------------ t 26 + t 46
- ⋅ ------------------------------------------- -⎞
⎝ t 16 + t 36⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
2---
⋅ τ 34
3
w 3 = -------------------------------------------------------------------------------
t 36 ⋅ t 46 ⎞ ⎛
⎛ ------------------ t 16 + t 26
- ⋅ ------------------------------------------- -⎞
⎝ t 36 + t 46⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
2---
⋅ τ 24
3
w 4 = -------------------------------------------------------------------------------
t 26 ⋅ t 46 ⎞ ⎛
⎛ ------------------ t 16 + t 36
- ⋅ ------------------------------------------- -⎞
⎝ t 26 + t 46⎠ ⎝ t 16 + t 26 + t 36 + t 46⎠
T 18 ⋅ T 28
τ 12 = ---------------------
-
T 18 + T 28
T 37 ⋅ T 47
τ 34 = ---------------------
-
T 37 + T 47
T 15 ⋅ T 35
τ 13 = ---------------------
-
T 15 + T 35
T 29 ⋅ T 49
τ 24 = ---------------------
-
T 29 + T 49
Δy 1
T 18 = k 1x --------- Δz 1
Δx 1
Δy 2
T 28 = k 2x --------- Δz 2
Δx 2
Δy 3
T 37 = k 3x --------- Δz 3
Δx 3
Δy 4
T 47 = k 4x --------- Δz 4
Δx 4
Δx 1
T 15 = k 1y --------- Δz 1
Δy 1
Δx 3
T 35 = k 3y --------- Δz 3
Δy 3
Δx 2
T 29 = k 2y --------- Δz 2
Δy 2
Δx 4
T 49 = k 4y --------- Δz 4
Δy 4
⎛ Δx 1 ⋅ Δy 1 ⎞
t 16 = k xy1 ⎜ -----------------------
-⎟
⎝ Δx 21 + Δy 21⎠
⎛ Δx 2 ⋅ Δy 2 ⎞
t 26 = k xy2 ⎜ -----------------------
-⎟
⎝ Δx 22 + Δy 22⎠
⎛ Δx 3 ⋅ Δy 3 ⎞
t 36 = k xy3 ⎜ -----------------------
-⎟
⎝ Δx 23 + Δy 23⎠
⎛ Δx 4 ⋅ Δy 4 ⎞
t 46 = k xy4 ⎜ -----------------------
-⎟
⎝ Δx 24 + Δy 24⎠
( Δx i + Δy i )k xi ⋅ k yi
k xyi = --------------------------------------------- Δz 1
k xi ⋅ Δy i + k yi ⋅ Δx i
C Darcy × TMi, j, k
T r = ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
LOG ( R i – 1, j, k ⁄ RM i – 1, j, k ) LOG ( RM i, j, k ⁄ R i – 1, j, k )
--------------------------------------------------------------------------------------------------- + ----------------------------------------------------------------------------- -
KR i – 1, j, k × DTH i – 1, j, k × DZBN i – 1, j, k KR i, j, k × DTH i, j, k × DZBN i, j, k
2.0 × C Darcy × TM i, j, k
T th = ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
RM i, j – 1, k × DTH i, j – 1, k RM i, j, k × DTH i, j, k
--------------------------------------------------------------------------------------------------- + ----------------------------------------------------------------------------- -
KTH i, j – 1, k × DR i, j – 1, k × DZBN i, j – 1, k KTH i, j, k × DR i, j, k × DZBN i, j, k
C Darcy × TM i, j, k
T z = -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
DZB i, j, k – 1 DZB i, j, k
---------------------------------------------------------------------------------------------------------------------------- + ----------------------------------------------------------------------------------------------- -
2 2 2 2
KZ i, j, k – 1 × DTH i, j, k – 1 × ( R i, j, k – 1 – R i – 1, j, k – 1 ) KZ i, j, k × DTH i, j, k × ( R i, j, k – R i – 1, j, k )
If input data contain gridblock face permeabilities KRF, KTF, and KZF, then the
above equations are modified to:
KZF i, j, k × C Darcy × TM i, j, k
T z = ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-
DZ i, j, k – 1 DZ i, j, k
--------------------------------------------------------------------------------------------- + ------------------------------------------------------------------------
2 2 2 2
DTH i, j, k – 1 × ( R i, j, k – 1 – R i – 1, j, k – 1 ) DTH i, j, k × ( R i, j, k – R i – 1, j, k )
H shared × C Darcy × TM i, j, k
T r = --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
LOG ( RM i, j, k ⁄ R i – 1, j, k ) LOG ( R i – 1, j, k ⁄ RMi – 1, j, k )
-------------------------------------------------------------------------- + ---------------------------------------------------------------------------------------------- -
KR i, j, k × DTH i, j, k × NTG i, j, k KR i – 1, j, k × DTH i – 1, j, k × NTG i – 1, j, k
33.4 Nomenclature
CDarcy Darcy’s constant:
1.127 x 10-3 in field units
8.52605 x 10-5 in metric units, where pressure is in kPa
8.36106143 x 10-3 in metric units, where pressure is in
kg/cm2.
Ri – Ri – 1
RM i = -----------------------------
-
⎛ Ri ⎞
LOG ⎜ -----------⎟
⎝ R i – 1⎠
TM transmissibility multiplier.
34
Unit Conventions
34.1 Introduction
Three unit conventions are available in VIP-EXECUTIVE. The default unit
system is FIELD (also known as customary, or English units). The others are
METRIC and LAB, which can be selected by placing a METRIC or LAB card in
the UTILITY section of the VIP-CORE input file. The METRIC option also can
specify pressure as kPa (default), kg/cm2, or bar units. All references to kPa in
the METRIC column of the table below can be replaced by kg/cm2 or bar
depending on the pressure option selected.
Permeability md md md
Viscosity cp cp cp
Volume (reservoir) rb m3 cc
The following equations can be used to convert gas deviation factors (Z) to gas
formation volume factors.
35
Velocity-Dependent Relative Permeabilties
35.1 Introduction
The VIP reservoir simulator can simultaneously model two effects of velocity on
fluid mobility:
1. The improvement in the mobility of both the oil and gas phases as capillary
number increases, due to what is known as the coupling effect. Capillary
number increases with increasing velocity and decreasing interfacial tension.
It is believed that the coupled flow of gas and condensate causes the
improvement in relative permeability at high velocities.
2. The reduction in the mobility of the gas phase due to the Forchheimer effect
as velocity increases.
The formulation of the velocity dependent models have been developed in the
Department of Petroleum Engineering at Heriot-Watt University under the
directorship of Professor A. Danesh and Professor D. H. Tehrani in a research
project sponsored by the UK Department of Trade and Industry and 11 oil and gas
companies.
When activated, the velocity dependency can be used to modify mobilities around
production wells. The velocity dependency can also be applied to inter-block
flow calculations. It can also be applied to both production wells and inter-block
flow.
Default parameters are built in that determine the magnitude of the capillary
number and non-Darcy flow effects. However, the user may selectively alter
some or all of these parameters.
μg Vg
N c = -----------
- (35-1)
σ
where:
qg
V g = --------------------------------
- (35-2)
ρ g Aφ ( 1 – S w )
The magnitude of the capillary number affects the oil and gas relative
permeabilities as follows:
S o – S oc f 2o ( N c )
k ro = f 1o ( N c )k rob + ( 1 – f 1o ( N c ) )k rm ( S w ) ----------------------------------------------- (35-3)
1 – S w – S oc f 2o ( N c )
S g – S gr f 2g ( N c )
k rg = f 1g ( N c )k rgb + ( 1 – f 1g ( N c ) )k rm ( S w ) -----------------------------------------------
1 – S w – S gr f 2g ( N c ) (35-4)
where:
1
k rm ( S w ) = --- [ k ro ( S o = 1 – S w ,S g = 0 ) + k rg ( S g = 1 – S w ,S o = 0 ) ] (35-5)
2
N cbo n o (35-6)
f 1o ( N c ) = ⎛ -----------⎞
⎝ Nc ⎠
N cbg ng
f 1g ( N c ) = ⎛⎝ -----------⎞⎠ (35-7)
Nc
N cbo
f 2o ( N c ) = 1 – exp ⎛ – m o -----------⎞ (35-8)
⎝ Nc ⎠
For interblock flow, the capillary number in each direction is calculated at the
beginning of each timestep. The capillary number will be calculated at each
block interface from:
–5 μg qg
N c = 1.9807 × 10 ------------------------------------- (35-10)
ρ g σAφ ( 1 – S w )
where:
μg = gas viscosity in cp
q g, ij = k rg, ij γ g, j + γ g, i in lbmoles/day
ρ g T ij ------------ ⎛ p j – p i + P c, j – P c, i – ----------------------- ( d j – d i )⎞
μ g, ij ⎝ 2 ⎠
krg,ij = Upstream relative permeability at the end of previous timestep (unless it is time 0)
A = Area in ft2
2 3
dyne – sec ⁄ cm 5.6146ft day
cp × 0.01 ------------------------------------------- × bbl ⁄ day ----------------------- × --------------------------------
cp bbl 3600 × 24sec- –5
------------------------------------------------------------------------------------------------------------------------------------------------------------
2
× 30.48cm ⁄ ft = 1.9807 × 10
ft × dyne ⁄ cm
The oil and gas relative permeabilities are modified using the above correlations
based on these values of capillary number, which is assumed to be constant during
the timestep.
Default values of parameters are assumed, unless the user re-defines them.
By default, no and ng will be set to 0.35 for sandstones and 1.16 for carbonates.
By default, mg will be set to 35 for sandstones and 4 for carbonates. By, default,
mo will be set to 0 for both sandstones and carbonates, which corresponds to an
Sorg of 0.
dΦ g μg qg
---------- = ⎛ ----------⎞ --------- (35-11)
dx ⎝ kk rg⎠ ρ g A
μg
ΔΦ g, ij = ⎛ ----------------⎞ q g, ij (35-12)
⎝ Tk rg ρ g⎠ ij
dΦ g μg qg β qg 2
---------- = ---------------- ----- + ----- ⎛⎝ -----⎞⎠ (35-13)
dx kk rg ρ g A ρ g A
μg kβ μg kk rg βq g
ΔΦ g, ij = ⎛ ----------------⎞ q g, ij + C ⎛ -------------⎞ q g, ij = ⎛ ----------------⎞ q g, ij 1 + C -------------------
2
⎝ Tk rg ρ g⎠ ij ⎝ ρ g TA⎠ ⎝ Tk rg ρ g⎠ ij (35-14)
ij μg A ij
FTk rg ρ g
q g, ij = ⎛ ---------------------⎞ ΔΦ g, ij
⎝ μ g ⎠ ij (35-15)
where:
1
F = --------------------------------
-
kk rg βq g (35-16)
1 + C -------------------
μg A
where:
2
k rg βρ g TΔΦ g
B = C -------------------------------
2
-
(35-18)
μg A
2 (35-19)
BF + F – 1 = 0
and:
F = –---------------------------------
1 + 1 + 4B- (35-20)
2B
B can be split into a constant part and a part that changes with each timestep. The
coefficient β can also be split into a constant part and a non-constant part. β is
defined as:
β β β β – β5 σ
β = β 0 k 1 φ 2 k rg3 S g 4 (35-21)
The values of β, β1, β2, β3, β4, and β5 can be reset with the NDARCY keyword.
Thus,
(35-22)
B = B const × B var
where:
1+β β
β0 k φ 2T
1
B const = C ------------------------------
- (35-23)
A
kk rg βρ g q g
C ------------------------- (35-24)
μg A
k md,
βg ft-1
ρg lb/bbl,
qg bbl/day
μg cp
A ft2
– 12 2 2
9.86923 × 10 cm ft –1 1day
md × ------------------------------------------------- × ------------------------------ - × ft × lb ⁄ bbl × bbl ⁄ day × ---------------------------------
md ( 30.48 ) cm
2 2 3600 × 24 sec – 16
- = 1.829744 × 10
C = ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
–4
6.719689 × 10 lb ⁄ ft – sec 2
cp × ------------------------------------------------------------------ × ft
cp
2+β β –β σ
k rg 3 S g 4 5 ρ g ΔΦ g
B var = ---------------------------------------------------
2 (35-25)
μg
For interblock flow, the constant part of B is evaluated at the beginning of the
simulation, while the variable part is recalculated at the beginning of every
timestep. Then F is evaluated, and kept constant during the entire timestep.
Consider the differential form of the radial well flow equation for each component
i:
kk ro Aρ o kk rg Aρ g dP
q i = x i -------------------- + y i -------------------- ------- (35-26)
μo μ g dr
2πkh k ro ρ o k rg ρ g
q i = ------------------- x i ------------- + y i ------------- ( p b – p wf )
ln r e ⁄ r w μo μg (35-27)
where pb is the well block pressure and pwf is the well flowing pressure, both at
the beginning of the timestep.
k ro ρ o k rg ρ g k ro ρ o k rg ρ g
z i ⎛ ------------- + ------------- ⎞ = x i ------------- + y i ------------- (35-29)
⎝ μo μg ⎠ μo μg
In VIP, to apply the effect of this integral expression, Equation 35-30 is multiplied
by a factor, FCB so that:
F CB 2πkh k ro ρ o k rg ρ g
q i = ------------------------ x i ------------- + y i ------------- ( p b – p wf ) (35-31)
ln ( r e ⁄ r w ) μo μg
Two bits of information are required to determine Equation 35-32. The first is the
overall composition, since densities and viscosities are compositionally dependent
as well as pressure dependent. We must also assume that the composition in the
completion block is constant. The second thing is how relative permeabilities
depend on pressure.
The compositions are determined from the produced well stream composition:
k ro ρ o k rg ρ g
w i = x i ------------- + y i -------------
μo μg (35-33)
and
wi
z i = -------------
n
-
c
∑ wj (35-34)
i 1
O’Dell and Miller (1967) introduced the concept that the gas volume fraction
calculated from the flash calculation is equivalent to the flowing gas volume
fraction, Vrg:
qg k rg ⁄ μ g
V rg = ----------------
- = ---------------------------------------
- (35-35)
qg + qo k rg ⁄ μ g + k ro ⁄ μ o
Rearranging:
k rg V rg μ g V rg μ g (35-36)
------- = ----------------- ----- = -------- -----
k ro 1 – V rg μ o V ro μ o
To handle velocity dependency, two features are added: (1) the gas and oil relative
permeabilities are calculated with capillary number dependency (2) a
Forchheimer multiplication factor is applied to the gas relative permeability.
dp μg
------ = ----------------------------- q g (35-37)
dr 2πrkhk rg ρ g
dp μg Cβ g
- q 2g
------ = ----------------------------- q g + -------------------------
dr 2πrkhk rg ρ g ( 2πrh ) ρ
2 (35-38)
g
k rg dp
q g = 2πkhrρ g δ g ------- ------ (35-39)
μ g dr
where:
1
δ g = -----------------------------------------
-
kk rg ρ g u g
1 + Cβ g ---------------------- (35-40)
μg
For each grid block with a production well completion in which condensate
qg
u g = --------------------
-
2πr e hρ g
a. At each pressure for which the integrand is calculated, the density and
viscosity are calculated from the pressure and composition.
b. The value of δg is calculated. For the first pass, the value of ug from the
previous evaluation point is used.
c. For pressures less than pb, a steady state solution is assumed and a new
radius, r2, is calculated from:
p1
r 2πkhk rg ρ g δ g
ln ----1 = -------------------- ∫ ----------- dp
r2 qg μg
p2
qg
u g = --------------------
-
2πr 2 hρ g
e. Steps b, c, and d are repeated until the successive velocities are within a
certain tolerance.
4. If a two phase region exists, for the range of pressures from phdp to p2 = pldp
or p2 = pwf, the following equation is integrated by the trapezoidal rule.
p h dp
k ro ρ o k rg ρ g δ g
m2 ( p ) = ∫ ------------- + ------------------- dp
μo μg
p2
a. At each pressure for which the integrand is calculated, the density and
viscosity are calculated from the pressure and composition.
k rg V rg μ g V rg μ g δ o
------- = ----------------- ----------- = -------- -----------
k ro 1 – V rg μ o δ g V ro μ o δ g
c. The values of δg and δo are recalculated. For the first pass, the values of
ug and uo from the previous evaluation point are used.
d. For pressures less than pb, a steady state solution is assumed and a new
radius, r2, is calculated from
p1
r 2πkh k ro ρ o k rg ρ g δ g
ln ----1 = ---------------------- ∫ ------------
- + ------------------- dp
r2 ( qo + qg ) μo μg
p2
qg
u g = --------------------
-
2πr 2 hρ g
qo
u o = --------------------
-
2πr 2 hρ o
g. Steps b through f are repeated until the successive velocities are within a
certain tolerance.
6. If a lower dewpoint pressure region exists, for the range of pressures from pldp
to pwf, the following equation is integrated by the trapezoidal rule.
p l dp
ρg δg
m3 ( p ) = ∫ ----------- dp
μg
p wf
k rg ( S w )m 1 ( p ) + m 2 ( p ) + k rg ( S w )m 3 ( p )
F CB = ----------------------------------------------------------------------------------------------
-
k ro ρ o k rg ρ g
------------- + ------------- ( p b – p wf )
μo μg
8. The multiplication factor is used to multiply both oil and gas mobilities, and
derivatives of gas and oil mobilities. The multiplication factor is assumed
constant for the time step.
Nomenclature
A = cross-sectional area, ft2
C = constant for unit conversion,
d = depth, ft,
F = non-Darcy flow mobility modification factor, dimensionless,
f 1g ( N c ) = capillary number dependent weighting factor between miscible and immiscible
gas relative permeability curves, dimensionless,
f 1o ( N c ) =capillary number dependent reduction factor between miscible and immiscible
condensate relative permeability curves, dimensionless,
f 2g ( N c ) = capillary number dependent factor affecting residual gas saturation, dimensionless
f 2o ( N c ) = capillary number dependent factor affecting critical condensate saturation,
dimensionless,
h = thickness, feet,
k = permeability,md,
k rg = gas phase relative permeability, fraction,
k rgb = gas phase relative permeability unaffected by capillary number, fraction,
k rm = miscible relative permeability, fraction,
k ro = condensate phase relative permeability, fraction,
k rob = condensate phase relative permeability unaffected by capillary number, fraction,
N c = capillary number, dimensionless,
N cbg = threshold capillary number for condensate relative permeability, dimensionless,
Greek
β = non-Darcy flow coefficient, 1/ft
β 0, β 1, β 2, β 3, β 4, β 5 = parameters of the non-Darcy flow coefficient,
Subscripts
g = gas phase,
i = component number,
o = condensate phase,
w = water phase, or well
36
Vertical Equilibrium
36.1 Introduction
Two separate vertical equilibrium (VE) options are available in VIP-
EXECUTIVE. These are:
At a new timestep, new fluid levels are calculated from fluid saturations. The
VIP-EXECUTIVE VE option accounts for saturation hysteresis during fluid level
movement. As a result, the model tracks the historical minimum water levels
(Hwmin) and the historical maximum and minimum gas levels (Hgmax and Hgmin).
These fluid levels and current fluid levels determine the saturation distribution in
each gridblock from which two-phase pseudo-functions are calculated. The oil
phase relative permeability then is calculated from the two-phase pseudo-relative
permeabilities using Stone’s Model I or Model II.
36.2.1 Initialization
At the beginning of each simulation run, the fluid levels are initialized from the
user input gas-oil and water-oil contact depths. Above the gas-oil contact, the
water saturation is equal to the connate water saturation (SWL) and the gas
saturation takes the maximum values (SGU), which is the highest gas saturation
entry in the gas saturation table. In the oil zone, the gas and water saturations
have the connate values (SGL and SWL). Below the water-oil contact, the water
saturation takes the maximum values (SWU) and the gas saturation has the
connate value. Two VE variables for each gridblock also are initialized: the water
level (Hw) defined as the water zone thickness divided by gridblock thickness, and
the gas level (Hg) defined as the gas zone thickness divided by gridblock thickness
(Δz). Finally, the average water and gas saturations in each gridblock are
calculated.
The thickness array TH for rectangular geometries, is the bedding plane thickness,
see Figure 36-1. During initialization, the thickness array is used to calculate the
average grid block properties about grid block centers. Blocks that are split by
phase boundaries are particularly sensitive to the actual values of block thickness.
Simulations with vertical equilibrium use the thickness array to calculate the
pseudo capillary pressures. Grid blocks with a large dip angle represent a larger
vertical span of reservoir than that implied by the bedding plane thickness, see
Figure 36-1. For this reason, THVE, is used by default in calculations of average
grid block properties and vertical equilibrium pseudo capillary pressures in VIP-
CORE version 2.4R. In previous versions of VIP-CORE the bedding plane
thickness TH was used.
The key word THCNTR in VIP-CORE allows the use of TH in place of THVE
thus allowing reproduction of old results.
TH
THVE
Additionally, the method of calculating dip angles in VIP-CORE 2.4R has been
changed.
The old method of calculating dip angles for block (i,j) used the slope between
blocks (i-1,j) and (i,j) in the x-direction, and the slope between blocks (i, j-1) and
(i,j) in the y-direction. Figure 36-2 shows a cross section of the resulting grid
block geometry generated by the old method. Notice the poor representation of
the structure in those regions with changes in the sign of the dip angle.
Figure 36-2: Block Angles Produced by the Old Dip Angle Method
The new method of calculating dip angles for block (i,j) uses the slope between
blocks (i-1,j) and (i+1,j) in the x-direction, and the slope between blocks (i,j-1)
and (i,j+1) in the y-direction. Figure 36-3 shows the new representation of the
same cross-section modeled in Figure 36-2. Notice the improved representation of
the reservoir structure.
Figure 36-3: Block Angles Produced by the New Dip Angle Method
Gas Level Gas Saturation Gas Rel. Perm. Oil Rel. Perm.
0.0
SGU KRGU 0
Hgmin
Hg SGRO KRGRO 0
SGTR 0 KROGR
Hgmax
For a given average gas saturation, Sg, the new gas level is calculated by
Note that these pseudo-relative permeabilities are valid only for flow in areal
directions. Thus, the VE option should be used exclusively in one- or two-
dimensional areal systems, unless the directional relative permeability option,
DRELPM, is also used.
level (Hg) is P* (i.e., zero capillary pressure), then the projected gas and oil phase
pressures at the grid midpoint are:
ρg Mg g
P g = P* – ------------------------------------- ( H g – 0.5 )Δz (36-4)
5.6146 × 144g c
ρo Mo g
P o = P* – ------------------------------------- ( H g – 0.5 )Δz (36-5)
5.6146 × 144g c
( ρ o M o – ρ g M g )g
P cg = P g – P o = ------------------------------------------ ( H g – 0.5 )Δz (36-6)
5.6146 × 144g c
When the gas level is monotonically decreasing (or increasing), the current gas
level, Hg, is the same as Hgmin (or Hgmax). Under this condition, the symbols
Hgmin (or Hgmax) in Equations 36-1, 36-2, and 36-3 should be replaced by Hg.
Water Level Water Saturation Water Rel. Perm. Oil Rel. Perm.
1.0
SWL 0 KRORW
Hwmax
SWR 0 KROWR
Hw
1-SOTR KRWRO 0
Hwmin
SWU KRWU 0
0.0
For a given average water saturation, Sw, the new water level is calculated by
Following the same procedure described in the gas-oil system, the water-oil
pseudo-capillary pressure, Pcw , is
o
( ρ o M o – ρ w B w M w )g
P cw = --------------------------------------------------- ( H w – 0.5 ) Δz (36-10)
5.6146 × 144 g c
When the water level monotonically decreases, the current water level, Hw , is the
same as Hwmin. Under this condition, symbols Hwmin in Equations 36-7 to 36-8
should be replaced by Hw .
■ Fluids within a cell are completely segregated. This implies that capillary
pressure effects are disregarded and there are no transition zones between
reservoir fluid phases.
8. The VEWO (VEGO) and VEITS options are completely vectorized. The
VEITS option is slower than the VEWO (VEGO) option.
oil-water contact takes a connate value (SWL). Gas saturation below the lowest
gas-oil contact has a connate value (SGL). Gas saturation above the highest gas-
oil contact is equal to gas saturation at connate oil (SGU). Gas has displaced oil in
the region below the highest gas-oil contact and above the current gas-oil contact.
Consequently, gas saturation is equal to gas saturation at residual oil (SGRO) in
this region. Oil has displaced gas in the region below the current gas-oil contact
and above the lowest gas-oil contact. Therefore, gas saturation takes a trapped
value (SGTR) from the primary imbibition process in this region. Oil has
displaced water in the region below the highest water-oil contact and above the
current water-oil contact. Consequently, water saturation has residual value
(SWR) in this region. Water has displaced oil in the region below the current
water-oil contact and above the lowest water-oil contact. Therefore, oil saturation
is equal to trapped oil saturation (SOTR) in this region. Water saturation below
the lowest water-oil contact is equal to water saturation at connate oil (SWU).
Sw = SWL, Sg = SGU
GOCh
Gas Sw = SWL, Sg = SGRO
GOC
Sw = SWL, Sg = SGTR
GOCl
Sw = SWL, Sg = SGL
WOCh
Oil Sw = SWR, Sg = SGL
WOC
WOCl Sw = 1 - SOTR, Sg = SGL
The procedures described below are used in the initialization module (VIP-
CORE) and the simulation module (VIP-EXECUTIVE) to compute saturation
distributions, directional relative permeabilities, and capillary pressures in each
gridblock.
The three depth regions (gas, oil, and water) can exist at the initial moment of time
in a gridblock (Figure 25-2). The oil-water and gas-oil contacts WOC, GOC are
set to the initial values, as specified by the user on the IEQUIL card. The highest
and lowest contacts are selected as:
The fractional volumes of the gas, oil, and water regions (FVg, FVo,FVw) are
calculated as FVp = Vp / Vb, p = g, o, w; where Vg, Vo, Vw are volumes of the gas,
oil, and water regions and Vb is a gridblock bulk volume. An accurate numerical
technique is used in VIP-CORE to calculate volumes Vg, Vo, Vw , and Vb. These
volumes are defined with an approximation tolerance, as specified by the user on
the VAITS card.
o o
The initial average block saturations S wb and S gb are calculated using the
saturation distributions in the block (Figure 36-5):
o
S wb = SWL*FV g + SWL*FV o + SWU*FV w (36-11)
o
S gb = SGU*FV g + SGL*FV o + SGL*FV w (36-12)
where SWL, SGL are connate water and gas saturations, SWU is water saturation
at connate oil saturation in an oil-water system, and SGU is gas saturation at
connate oil saturation in an oil-gas system.
Each block is divided into N sublayers (Figure 36-5). The number of sublayers is
specified by the user on the VEWO or VEGO card. The accuracy of the
calculations and the CPU memory requirements are increased by increasing the
number of sub-layers. The default number of sub-layers is equal to ten.
Vg
Vb Vj GOCh = GOC
Aj1
Vo A1
WOCl = WOC
Vw
Figure 36-5: The VE Initialization Procedure
For each sublayer, the fractional volume FVj and the fractional areas FAji for each
gridblock face are calculated from:
Vj
FV j = ------ , j = 1, 2, ...,N (36-13)
Vb
A ji
FA ji = ------ , j = 1, 2, ...,N, i = 1, 2, ..., 6 (36-14)
Ai
where Vj is the volume of the j’th sublayer (j = 1,2,...,N), Aji is the area of the
intersection of the j’th sub-layer and the i’th gridblock face, and Ai is the area of
the i’th face (i = 1,2,...,6).
Water saturations in the sub-layers depend on the current position of the oil-water
contact (Figure 25-1). Therefore, the current position of the water-oil contact
(WOC) is calculated from a condition in which the average block water saturation
t
should be equal to S wb :
j=N
t
∑ Swj ( WOC ) × FVj = S wb (36-15)
j=1
where Swj is a function of the location of the WOC. If the WOC falls inside a sub-
layer, the sub-layer is further divided in order to accurately calculate Swj.
j=N
t
∑ Sgj ( GOC ) × FVj = S gb
(36-16)
j=1
The lowest and highest contacts are recalculated. The water and gas saturations in
each sub-layer are defined as shown in Figure 25-1.
The relative permeabilities for an intersection of each block face with each sub-
layer are calculated by:
where Kwi, Kgi are values of rock water and gas relative permeabilities for the i’th
face.
ve
P cw = g ( ρ w – ρ o ) ( WOC – D center ) (36-20)
ve
P cg = g ( ρ o – ρ g ) ( GOC – D center ) (36-21)
where ρ g, ρ o, and ρ w are densities of the gas, oil, and water phases, and D center
is the depth of the block center.
Rock relative permeabilities and rock capillary pressures are determined from
user input tables:
rock t
K pi = K pi ( S pb ), p = w, g, i = 1, 2, ...,6 (36-22)
rock t
P cp = P cp ( S pb ), p = w, g (36-23)
The mixed relative permeabilities for each gridblock face and the mixed capillary
pressure are calculated as follows. Parameters FVEWO and FVEGO are specified
by the user for each gridblock using the FVEWO and FVEGO arrays. If the
VEWO (or VEGO) option is not used, only rock water (or gas) relative
permeabilities are applied in a simulation.
mixed ve rock
K wi = FVEWO K wi + ( 1 – FVEWO )K wi , i = 1, 2, ...,6 (36-24)
mixed ve rock
K gi = FVEGO K wi + ( 1 – FVEGO )K gi , i = 1, 2, ..., 6 (36-25)
mixed ve rock
P cw = FVEWO P cw + ( 1 – FVEWO )P cw , (36-26)
mixed ve rock
P cg = FVEGO P cg + ( 1 – FVEGO )P cg (36-27)
reservoir fluids is not used. This implies that oil-water and gas-oil transition
zones exist. The distributions of the pressure and saturations versus depth in each
cell are calculated from the capillary-gravity equilibrium conditions.
The keyword VEITS should be included in an initialization data set to invoke this
option. The capillary pressure curves should not be flat to use the VEITS option.
The program internally checks the capillary pressure curves. If they are flat, then
the VEWO, VEGO option is used instead of the VEITS option.
The procedures described below are applied in the initialization (VIP-CORE) and
simulation (VIP-EXECUTIVE) modules to calculate pressure and saturation
distributions, directional relative permeabilities, and capillary pressures in each
gridblock.
The block is divided into N sub-layers, where N is specified by the user on the
VEITS card. For each sub-layer, fractional volumes FVj and fractional areas FAji
(for each block face) are calculated and stored. The initial pressures and
saturations in each sub-layer are calculated using the capillary gravity
initialization procedure (VAITS).
Water pressures Pwj, oil pressures Poj, and water saturations Swj in each sub-layer
are calculated from the solution of the capillary-gravity equilibrium equations:
dP
--------o- = ρ o ( P o )g (36-28)
dD
dP w
---------- = ρ w ( P w )g (36-29)
dD
P o – P w = P cw ( S w ) (36-30)
■ Water pressure at a block center should be equal to the average block water
t
pressure, P wb :
t
P w ( D center ) = P wb (36-31)
■ The oil pressure at some depth, Do, should be equal to the average block oil
t
pressure, P ob :
t
P o ( D o ) = P ob (36-32)
Gas pressure (Pgj) and saturation (Sgj) in each sub-layer (j = 1,2,...,N) are
calculated from solution of the capillary-gravity equilibrium equations:
dP
--------g- = ρ g ( P g )g (36-34)
dD
P g – P o = P cg ( S g ) (36-35)
■ Gas pressure at some depth Dg should be equal to the average block gas
t
pressure, P gb :
t
P g ( D g ) = P gb (36-36)
After calculating the water and gas saturations in different sub-layers, the relative
permeabilities are calculated using the same procedure as in the VEWO (VEGO)
option.
37
Water Tracking Option
37.1 Introduction
VIP-EXECUTIVE can track flow, accumulation, and production of up to six
different water types. Note that the actual reservoir performance is completely
unaffected by this option. The water movement is exactly the same; only the
splitting and tracking of the water phase into different water types is performed.
Water types may be specified for the in situ water, for each water injection well,
and for each numerical aquifer. The major assumption is that the water properties
(density, viscosity) are independent of the water type.
n n+1 n n P n I
Δ [ ( T w F Twj ) ( Δ P o – ΔP cwo – γ w ΔD ) ] – Q w ( F Twj ) + Q wj
V φS w φS w n
n+1
= ----- ⎛ ----------j⎞ – ⎛ ----------j⎞ (37-1)
Δt ⎝ B w ⎠ ⎝ Bw ⎠
where
and
∑ FTw j
= 1.0 for each grid block (37-2)
j
∑ Sw j
= Sw for each grid block (37-3)
j
The fractional flow terms for the water types can be adjusted by two user-
specified parameters, if performance other than complete mixing of the water
types is desired. The single parameter, ftwmix, determines the relative amount of
mixing with the connate water. A value of 1.0 (default) specifies complete mixing
with the connate water, while a value of 0.0 specifies complete bypass of the
connate water. Any value between these two specifies partial mixing between the
connate water and all of the other water types.
First, Equation 37-4 is used to calculate the normalized fractional saturations for
each water type.
⎛ dSw j ⎞ Sw j
S wn = ( 1 – ftwmix ) × ⎜ -----------------------⎟ + ftwmix × ⎛⎝ --------⎞⎠ (37-4)
j ⎝ Sw – Sw R⎠ Sw
where
= Sw j for j ≠ in situ
and
∑ Sw nj
= 1.0 (37-6)
where Swnj is the normalized saturation of mobile water type j. Thus, the Swnj are
the fractions of the mobile water phase for each of the water types.
The grid array parameter, tkexp, determines preferential flow between the in situ
water and all of the other water types. These grid-block values are used as
exponents in the fractional flow (Equations 37-7, 37-8, and 37-9), such that a
value of 1.0 (default) yields no preference between the in situ water and all of the
other water types. As shown in Figure 37-1, a value less than 1.0 gives a higher
preference to flow to the extraneous water types (i.e., all water types except in
situ), while a value greater than 1.0 gives a higher preference to flow to the in situ
water.
The fractional flow terms are split between the mobile in situ water and the mobile
extraneous water types using Equations 37-7 and 37-8.
tkexp
FW EXT = ( 1 – S wnin – situ ) (37-7)
= fractional flow for all water types except the in situ water
1.0
tkexp < 1
tkexp = 1
FWEXT
tkexp > 1
The fractional flow terms for each of the extraneous water types then are
calculated using Equation 37-9, based on the relative amount of mobile fluid for
each type.
⎛ Sw n j ⎞ (37-9)
-⎟ × FW EXT, j ≠ in situ
F Twj = ⎜ ----------------------------
⎝ 1 – Sw nin – situ⎠
38
Well Inflow Performance
38.1 Introduction
This chapter outlines the relationships that VIP-EXECUTIVE uses to describe
well inflow performance. The Inflow Performance Relationship (IPR) describes
reservoir performance and is the relationship between the reservoir pressure at the
wellbore and the surface flow rate. IPR takes into account pressure losses that are
attributable to both the formation (reservoir) and completion resistances at the
wellbore, which are included in the total drawdown. The natural flowing point
occurs at the point of intersection between the IPR and the Tubing Performance
Curve (TPC), also known as the lift curve or outflow curve. At this point,
reservoir pressure at the wellbore becomes equal to the bottom-hole flowing
pressure required by the producing string so that the well naturally flows to the
surface. This is illustrated in Figure 38-1.
Figure 38-1: Inflow and Outflow Curve for a Production Well Showing Intersection
Point
Oil mole rate qlo and gas mole rate qlg in the l’th perforation are calculated in VIP-
EXECUTIVE using the following expression:
⎧ ⎫
⎪ kr ρj ⎪
q lj = WI l ⎨ k h ------------
l l
j
δ l [ P l – P bh – γ l ( D l – D ref ) ] ⎬, l = 1, 2 ...,NPER, j = o, g (38-1)
⎪ μj ⎪
⎩ ⎭
where
krj relative permeability for the oil phase (j=o), gas phase
(j=g), or water phase (j=w), which depends on the
gridblock saturations
where
o
ρw water density at standard conditions.
After summing Equations 38-1 or 38-2 through all open perforations, the oil mole
rate Qo, gas mole rate Qq, and water rate in standard conditions Qsw can be
written:
NPER NPER
⎧ k rj ρ j ⎫
Qj = ∑ q lj = ∑ ⎨ WI l k l h l ------------ δ l [ UW4 l – P bh ] ⎬, j = o, g (38-3)
⎩ μj ⎭
l=1 l=1
NPER NPER o
⎧ k rw ρ w ⎫
Q sw = ∑ q lw = ∑ ⎨ WI l k l h l -------------- δ l [ UW4 l – P bh ] ⎬
⎩ μw ⎭
(38-4)
l=1 l=1
where
The mole rates of the fluid components can be expressed in the following form:
NPER
Qi = ∑ ( q ol xw il + q gl yw il )
l=1
NPER
⎛ k ro ρ o k rg ρ g ⎞
= ∑ WI l k l h l δ l ⎜ ------------- xw il + ------------- yw il⎟ ( UW4 l – P bh ), i = 1, 2, …, N c
⎝ μo μg ⎠
l=1
(38-5)
Options 1, 2, and 3 are mutually exclusive and result in setting the same WI value
for each perforation. In the fourth option, the well index is allowed to vary for
each perforation. The values of well index for each perforation (WIL) can be
entered or computed internally using the data specified on the FPERF card.
If the RFLOW card is used, as in the second option, the well index is calculated in
VIP-EXECUTIVE as:
2π
WI = --------------------------------
- (38-6)
⎛ rb⎞
ln ----- + skin
⎝ r w⎠
where
rW wellbore radius
skin total skin factor including the skin from the partial
penetration effect.
Here, rb is defined as the distance from the well at which the local pressure is
equal to the pressure of the block. Note that rb may be calculated by applying
Peaceman’s34 formula, a general expression that is given below for wells located
in the center of rectangular gridblocks.9
1⁄2 1⁄2
⎛ k----y⎞ Δx 2 + ⎛ k----x⎞ Δy 2
⎝ k x⎠ ⎝ k y⎠
r b = 0.28 -----------------------------------------------------------------
- p (38-7)
k y⎞ 1 ⁄ 4 ⎛ k x⎞ 1 ⁄ 4
⎛ ---- + ----
⎝ k x⎠ ⎝ k y⎠
where
For isotropic properties, the above equation transforms to the following simple
form:
2 2 1⁄2
r b = 0.14 ( Δx + Δy ) (38-8)
The well index in Equation 38-6 also can be calculated in VIP-EXECUTIVE from
the productivity/injectivity index. A steady-state productivity index is defined as
the production rate of a chosen phase divided by the pressure drawdown incurred
during production of the fluid into the wellbore. Drawdown is the difference
between reservoir pressure at the drainage radius and the bottom-hole flowing
pressure of the well. Therefore, the productivity index is defined as:
qo
PI = -------------------- (38-9)
P d – P bh
where
Equation 38-9 also is used for the injectivity index when the injection rates are
treated as negative. Well index WI for an oil producer can be converted from PI by
the following expression:
( PI ) ( gf )
WI = ----------------------------------------------
- (38-10)
⎧ k h k ro ⎫
0.001127 ⎨ ----------------- ⎬
⎩ μo Bo ⎭
where
Note that Equation 38-11 is not always applicable. See Appendex A, Section A.6.
Several new optional variables are included in the FPERF card that allow the well
index to vary with each perforation and allow optional calculation of the reduced
entry skin factor. The reduced entry skin factor is added to the optionally specified
skin (damage) factor to determine the total skin factor used to calculate the well
index for each perforation. When using the FPERF card, VIP-EXECUTIVE will
calculate a default rb value for each layer if the RADB keyword is not specified.
The default rb value is calculated from Equation 38-7.
where
δ o Q so + δ g Q sg
BSEP = ----------------------------------
- (38-13)
Qo + Qg
δo, δg, and δw parameters for each phase, oil, gas, and water, (equal to 1
if the rate for that phase is specified and equal to 0 in all
other cases)
The wellbore flow model and surface separator model26 define the oil, gas, and
water rates of a three-phase producer in standard conditions as functions of the
bottom-hole pressure:
Q so = Q so ( P bh ) (38-16)
Q sw ( P bh )
- = WCUT ( P bh )
WCUT = ---------------------- (38-17)
Q so ( P bh )
Q sg ( P bh )
- = GOR ( P bh )
GOR = --------------------- (38-18)
Q so ( P bh )
The oil and gas production rates in reservoir conditions can be defined from the
wellbore flow (Equation 38-3). The oil and gas production rates in standard
conditions can be found using the multistage separator model.26 The water rate in
standard conditions can be calculated from Equation 38-4.
If the tubinghead pressure, Pth is specified by the user, then the problem consists
of determining the bottom-hole pressure, Pbh, which honors both the outflow
(Equation 38-15) and the inflow (Equations 38-16, 38-17, and 38-18).
39
Well Management Features
39.1 Introduction
This chapter describes in detail the technical details of the VIP-EXECUTIVE well
management features.
❑ Well data, such as name, location, and gathering center (WELL card)
❑ Well data, such as name, location, and gathering center (WELL card)
R5000.0.1 39-579
VIP-EXECUTIVE® Technical Reference Guide Landmark
❑ Well data, such as name, location, and gathering center (WELL card)
Note that if no productivity index is defined by a WI, PI, or RFLOW card, well
productivity automatically is adjusted to cause the well to flow at the rate
specified on the QMAX card (i.e., the well defaults to rate constraint).
❑ Well data, such as name, location, and gathering center (WELL card)
❑ Well data, such as name, location, and gathering center (WELL card)
39-580 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
❑ Well data, such as name, location, and gathering center (WELL card)
❑ Tubing data required to calculate BHP from THP (TUBE and DIAM
cards).
Note that if no injectivity index is defined by a WI, PI, or RFLOW card, the
injectivity automatically is adjusted to cause the well to flow at the rate specified
on the QMAX card (i.e., the well defaults to rate constraint).
R5000.0.1 39-581
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ The total number of perforations for all wells cannot exceed NPRFTOT, as
specified on the DIM card. The default value for NPRFTOT is 25.
There are many ways to specify perforation data. Some of the commonly used
formats and their implications follow:
■ The user can specify the layers that are open to flow. In addition, the user can
specify permeability (K), thickness (H), or permeability-thickness product
(KH) for each perforation. The permeability-thickness product for the
perforation depends on the data specified. The following options are
available:
❑ If both K and H are specified, the KH used in the model is the product of
the user-specified K and the user-specified H.
■ The user can specify the top and bottom depth of each perforation. The model
then automatically determines the layers that are open to flow. In this case,
perforation thickness H is determined from the top and bottom depth of each
layer. Within the simulator a perforation will be defined for each layer open to
flow.
39-582 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ The user can specify a unit number for each perforation. All perforations
within a well with the same unit number are considered to represent a single
completion. Perforation production constraints and all workover options
affecting perforations then are applied to completion, or “perforation unit,”
rather than to individual perforations. Production rates in the perforation units
are obtained by summing the rates of the individual perforations. Perforation
unit numbers may be specified for injection wells, but are not used.
The following special features are available for perforation data with the VE
option:
■ The position of the top (HTOP) and bottom (HBOT) of each perforation in a
gridblock can be specified. Therefore, a well may have more than one
perforation in a given reservoir layer when the VE option is being used.
In addition, the saturation tables can be scaled to different end-point values. The
following end-point saturations can be specified for each perforation:
The end-point saturation value used for the perforation calculation depends on
data specified by the user. The following options are available:
R5000.0.1 39-583
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ If end-point saturations and the ISAT value for the perforation are not
specified, then the end-point saturations for the gridblock are used. The end-
point saturations for each gridblock are initialized to the table values for the
gridblock. However, the user can specify (overread) end-point saturations for
each gridblock.
■ If the ISAT value for the perforation is specified, but the end-point saturations
are not specified, then end points from the ISAT table are used.
■ If specified, the end-point saturations for the perforation are used and the table
values (from ISAT or gridblock table) are not used.
■ To specify the end-point saturations for perforation Swmn, Swro, Sgmn, and Sgro
must be specified together. Swl, Swmx, Sgl, and Sgmx are optional. The default
values are Swl = Swmn, Sgl = 0, and Sgmx = 1-Swl.
39-584 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ Total production rate for all fluids (at reservoir conditions or in moles)
■ Bottom-hole pressure
■ Tubinghead pressure
■ Maximum drawdown.
The production rate for the well is specified by the QMAX card. The well
produces at this rate unless it causes a violation of one of the other constraints
defined by the user. In this event, the constraint is observed, which causes a rate
reduction.
■ The user can define a maximum water cut, water rate, or liquid-gas ratio for a
production well (WLIMIT card). If the user-specified maximum for a well is
exceeded, one of the following actions takes place:
❑ The perforation with the highest water cut (or liquid-gas ratio) is plugged.
A different limiting water cut (or liquid-gas ratio) value can be specified
for the last open perforation. Once a perforation has been plugged, it
never produces again, unless the well is reperforated by an FPERF card.
R5000.0.1 39-585
VIP-EXECUTIVE® Technical Reference Guide Landmark
❑ The well is shut in. It then can be tested periodically according to the data
on the TEST card. If water cut (or liquid-gas ratio) is found to be less than
the maximum during the test, then the well is returned to production.
■ The user can define a maximum gas rate or gas-liquid ratio (GOR) for a
production well (GLIMIT card). If the user-specified maximum for a well is
exceeded, one of the following actions takes place:
❑ The well is shut in. It then can be tested periodically according to the data
on the TEST card. If the GOR is found to be less than the maximum
during the test, then the well is returned to production.
■ The user can specify a maximum water cut and GOR applicable to each
perforation in a well. When the water cut and/or GOR for any perforation in a
well exceeds this maximum, then the perforation is shut in. The test to
determine whether a violation has occurred is performed only if the time
interval between tests is specified using the TSTPRF card. Therefore, it is
possible that the perforation value may exceed the maximum between test
intervals. In addition, note that flashes to surface conditions are required for
each perforation during these tests. This could cause a significant increase in
computer time if frequently performed (especially in compositional
simulations). Once a perforation has been shut in, it never produces again,
unless the well is reperforated using an FPERF card.
■ The user can specify a minimum rate for each production well (QMIN card).
The phase to which this minimum rate applies is specified by the ECOLIM
card. When the well rate falls below the minimum, the well is shut in. The
well can be tested periodically to determine whether the rate is above
minimum, if the appropriate data on the TEST card are specified.
39-586 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Tubinghead pressure constraints also can be imposed on gas producers and on gas
and water injectors. For these cases, the algorithm is based on the “average
pressure and temperature method” as described by Beggs.35
The ITNTHP card can be used to set the number of outer iterations for each
timestep during which the tubinghead pressure algorithm is performed. During the
rest of the timestep, the well is treated as a bottom-hole pressure-constrained well,
with BHP equalling the calculated value from the tubinghead pressure algorithm.
The default for ITNTHP is 2.
Pressure constraints and the productivity index are described in detail in Chapter
38.
R5000.0.1 39-587
VIP-EXECUTIVE® Technical Reference Guide Landmark
The status of the well is indicated by a four-character word in the well report. The
possible production well status messages and their meanings follow:
39-588 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Two methods are available to determine the time at which wells are tested.
1. All wells of the appropriate shut-in type tested at the same time (default).
2. Each well is tested at the time interval after it was shut-in. This is invoked by
specifying a negative value on the TEST card.
For the first method, well tests are scheduled for the time at which the TEST card
is read, plus the appropriate increment. Until the simulation reaches that time,
shutin wells remain shut in. Timesteps are not adjusted to hit the test time exactly.
Once the wells are tested, new tests are scheduled for a time that is the appropriate
increment farther into the simulation.
For the second method, each shut-in well is tested after the appropriate time
interval (absolute value) has elapsed since it was shut-in. Timesteps are not
adjusted to hit any test time exactly. If a well does not return to production/
injection when tested, it will be retested at a time that is the appropriate increment
further into the simulation.
R5000.0.1 39-589
VIP-EXECUTIVE® Technical Reference Guide Landmark
4. This process is continued until the bottom-hole pressure of the last interval is
calculated; this is the flowing bottom-hole pressure of the well for the outflow
curve.
39-590 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
A special option is available for net voidage injection where the volume of the
second injected phase is determined to maintain voidage balance (or a fraction of
voidage balance) at the user-specified well management entity. This feature is
described in Section 39.13.
■ Bottom-hole pressure
■ Tubinghead pressure
■ Maximum buildup.
The injection rate for the well is specified by the QMAX card. The well injects at
this rate unless it causes a violation of one of the other constraints defined by the
user. In this event, the constraint is observed, which causes a rate reduction.
Injection wells also can be used to reinject the produced fluids. If used for
reinjection, the injectors could:
■ Reinject a fraction of the total surface production rate of the injected phase
within a specified level of well management hierarchy (called FSTD
reinjection), or
When either the FRES or FSTD reinjection option is used, values on the QMAX
card for the corresponding injectors are fractions of total reservoir volume
production rate and the fraction of the total surface production rate, respectively.
R5000.0.1 39-591
VIP-EXECUTIVE® Technical Reference Guide Landmark
The BHP for injectors on THP control is calculated by using either the “average
temperature and pressure” method or by using hydraulics tables (BHITAB data).
■ The outer iteration number after which water injection rates are not
recomputed can be specified using the ITNSTQ card.
■ The user can specify the method used to compute the mobility for water
injection wells. End-point mobilities are used by default, but it is possible to
use total gridblock mobilities (WINJMOB card).
■ The water density and viscosity values used in the THP calculation can be
specified using the WTRTHP card. The default is to use the values specified
in the initialization data.
■ The outer iteration number after which gas injection rates are not recomputed
can be specified using the ITNSTQ card.
■ The user can specify the method used to compute the mobility for gas
injection wells. End-point mobilities are used by default, but it is possible to
use total gridblock mobilities (GINJMOB card).
■ The composition of injected gas for gas injectors must be specified using the
YINJ card with the STD or RES option.
■ The RECFAC card can be used to specify the fraction of each component in
the gas stream that can be recovered as liquid at the gas plant for use with gas
injectors on the FSTD reinjection option.
39-592 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
The WAG card is used to specify which wells are WAG injectors. Three items
must be input on this card: cumulative water injection volume per cycle,
cumulative gas injection volume per cycle, and the total number of WAG cycles to
be performed. Two other items may also be input: whether WAG cycling starts
with gas or water injection (default is water injection), and what happens to the
well when WAG cycling ends (water injector, gas injector, shut-in).
The QMAXWG card is used to specify the maximum water and gas rates for each
WAG well. The BHPWAG card may be used to specify a different value of
bottomhole pressure limit for water injection versus gas injection.
In most fine-grid simulation models, it is necessary to limit the size of the first
timestep following each changeover to avoid potential convergence failures. This
can be controlled by entering a DTWAG card which contains two values:
maximum size of the first timestep following the changeover from gas injection to
water injection and the maximum size of the first timestep following the
changeover from water injection to gas injection.
The changeover will automatically take place whenever the cumulative injection
volume exceeds the user-specified injection volume. At the changeover, the size
of the previous step is not reduced and the step is not repeated so as to exactly
match the prescribed injection volume. Instead, the excess injection volume will
be subtracted from the prescribed injection volume for the next cycle to
compensate for the over-injection. This is done to avoid excessive timestep cuts in
models with a large number of WAG wells.
R5000.0.1 39-593
VIP-EXECUTIVE® Technical Reference Guide Landmark
PROP well is cut back using the PROPTN option within the
injection region option.
39-594 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
39.6.1 Introduction
By default, the pressure gradient in the wellbore due to the effect of gravity is
treated as a constant with respect to depth for each well. If the MBAWG = OFF
option is invoked or if crossflow (XFON) is specified for one or more wells, then
the simulator computes variations in wellbore gravity gradient as fluids enter or
leave the wellbore at each perforated interval for all wells. In this case, the
wellbore gradient is calculated on the first iteration of each timestep, but is held
constant for the remainder of the timestep to avoid slowing the convergence rate
of the outer iterations. In the event that a well is shut in, plugged, or reperforated
during a timestep, the gradient calculation is repeated to reflect the change in
conditions. Two methods for calculating wellbore gradient are described below.
This section discusses the method to calculate an average pressure gradient γl used
in all the different perforations l = 1,2,...,NPER of a production well. The average
pressure gradient γ1 = γ2 = ... = γNPER = γ , is applied in VIP-EXECUTIVE in the
wellbore flow Equations 38-1 through 38-5. It is calculated in the following form:
g gravitational acceleration
R5000.0.1 39-595
VIP-EXECUTIVE® Technical Reference Guide Landmark
Mathematical Formulation
The wellbore pressure gradient opposite any perforated interval “j” is computed as
the volumetric average of the densities of the individual phases. The densities are
measured within the reservoir layer containing the perforated interval. The weight
factors used to compute the average gradient are the volume fractions of each of
the three phases within the wellbore and opposite the perforated interval:
n n n
γ j = E wj γ wj + E oj γ oj + E gj γ gj (39-2)
It would be more correct to compute the densities of the three phases at wellbore
conditions instead of using densities computed within the reservoir, but this would
introduce troublesome nonlinearities into Equation 39-2, since the densities of oil
and gas depend on both pressure and composition. For the gradient calculation,
the densities of the reservoir fluids are evaluated at the old time level, so they can
be treated as known quantities.
To compute the volume fractions of the wellbore fluids, we divide the wellbore
into grid cells, one cell opposite each perforated interval. A total volume balance
for wellbore grid cell “j” can be written as follows:
q Tj + 1 ⁄ 2 – q Tj – 1 ⁄ 2 = WI ( kh ) j λ Tj ( p j – p wj ) (39-3)
Assuming that “no-slip” conditions exist in the wellbore, the volume fractions of
water and gas are related to the total volumetric flow as follows:
E wj + 1 ⁄ 2 q Tj + 1 ⁄ 2 – E wj – 1 ⁄ 2 q Tj – 1 ⁄ 2 = WI ( kh ) j λ wj ( p j – p wj ) (39-4)
E gj + 1 ⁄ 2 q Tj + 1 ⁄ 2 – E gj – 1 ⁄ 2 q Tj – 1 ⁄ 2 = WI ( kh ) j λ gj ( p j – p wj ) (39-5)
No-slip means that all phases travel at the same velocity. The volume fractions of
water and gas that apply at the interface between the wellbore grid cells are
evaluated at the upstream cell. Similarly, mobilities (λTj , λwj , and λgj ) are
evaluated in the reservoir or the wellbore, depending on whether production or
injection is occurring in the perforated interval.
If the perforated interval is producing, then the mobilities are defined as follows:
39-596 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
k rw k ro k rg⎞
λ Tj = ⎛⎝ ------- + ------- + ------- (39-6)
μw μo μg ⎠ j
k rw
λ wj = ⎛⎝ -------⎞⎠ (39-7)
μw j
k rg
λ gj = ⎛ -------⎞ (39-8)
⎝ μg ⎠ j
If the perforated interval is injecting, then total mobility (λTj) is still defined by
Equation 39-6, but the mobilities of water and gas are defined as:
λ wj = E wj λ Tj (39-9)
λ gj = E gj λ Tj (39-10)
The wellbore pressure, pwj, is computed from the pressure in the adjacent
wellbore grid cell:
If the datum is contained within wellbore grid cell “j,” then j = jD, and pwj is
related to the wellbore pressure at datum by the following expression:
o o
p wjD = p w + γ j D ( D j D – D ) (39-13)
If the datum lies above the top perforated interval, then the top perforated interval
is treated as the datum-containing perforated interval and Equation 39-13 applies.
Similarly, if the datum lies below the bottom perforated interval, then the bottom
perforation is treated as the datum-containing perforation and Equation 39-13
applies.
One of the wellbore grid cells is designated as the one from which fluid is
withdrawn to the surface in the case of a producer, or the point at which fluid first
reaches a perforated interval in the case of an injector. This might correspond to
the bottom of the tubing string in a well that produces through tubing. Although
the algorithm allows this cell to be anywhere in the well, the current
implementation assumes that the cell is the top-most perforated interval. The
volumetric rate of fluid withdrawal, or injection, may be written as the following
expression:
R5000.0.1 39-597
VIP-EXECUTIVE® Technical Reference Guide Landmark
Q = ( a + b E g + c E w )q T (39-14)
where Q is the maximum flow rate allowed in whatever units are used to constrain
the well, and qT is the volumetric flow rate at wellbore conditions.
Once the pressure of each wellbore grid cell has been determined, the difference
in wellbore pressure between each wellbore cell and the wellbore datum is
computed and saved for the remainder of the timestep for use in allocating layer
production and injection rates.
o
DPWB j = p wj – p w (39-15)
Calculation Procedure
Equations 39-2 through 39-14 define the relationship between wellbore pressures
and reservoir pressure and saturation distributions. All of these equations are
linear, except for Equations 39-3, 39-4, and 39-5. The system of equations is
solved by two different procedures.
In the second method, the full system of equations is solved simultaneously using
the Newton-Raphson iteration. This involves solving four equations
simultaneously for each wellbore grid cell. These are Equations 39-3, 39-4, 39-5,
and either 39-11 or 39-12, as appropriate. Convergence normally is rapid. In the
rare event of convergence failure, the results of the first method are accepted.
Timestep cutting is not performed to force convergence of this iteration.
39-598 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
39.6.4 Nomenclature
Eg volume fraction of the gas phase in the wellbore.
kr relative permeability.
WI well index.
λ fluid mobility.
μ viscosity, in cp.
With subscripts:
g gas.
o oil.
T total.
w water or well.
R5000.0.1 39-599
VIP-EXECUTIVE® Technical Reference Guide Landmark
39.7.1 Theory
Consider a production well producing at a rate of q RB/D, with layer productivity
indices Jk, layer mobilities λk (total), and layer pressures at datum, Pk. The total
layer flow rates are given by:
q k = J k λ k ( P k – P bh ) (39-16)
where Pbh is the flowing bottom-hole pressure at datum. The total well rate is:
q = ∑ qk (39-17)
k
∑ Jk λk Pk – q
k
P bh = --------------------------------
- (39-18)
∑ Jk λk
k
If Pbh > Pk, then layer k is backflowing. The individual phase layer rates are given
by:
q mk = J k λ̂ mk ( P k – P bh ), m = o, w, g (39-19)
where
⎧ wellblock
⎪ λ if P k > P bh
λ̂ mk = ⎨ mk (39-20)
⎪ λ wellblock S wellbore if P k < P bh
⎩ mk mk
where Smkwellbore are the layer phase saturations in the wellbore. These wellbore
saturations and wellbore phase compositions could be obtained by solving
material balances and equilibrium constraints on a gridded wellbore, but the
equations are highly nonlinear due to rapidly changing conditions in the wellbore
and would require very small timesteps to converge.
39-600 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
⎧ *
⎪ * if P k > P bh
= ⎨ J k λ k ( P k – P bh )
*
qk (39-21)
⎪ 0 *
if P k < P bh
⎩
+
* Σ ( Jk λk Pk ) – q
P bh = ------------------------------------
- (39-22)
+
Σ Jk λk
where Σ+ denotes summation over inflowing layers. Incremental layer rates are
defined as:
*
q̂ = q k – q k (39-23)
and that:
q̂ k = J k λ k ( P̂ k – P bh ) (39-25)
where
*
P̂ k = MIN ( P k, P bh ) (39-26)
∑ Jk λk P̂k
P bh = ------------------------ (39-27)
∑ Jk λk
Substituting this into Equation 39-25 gives:
∑ Jk λk Jj λj ( P̂k – P̂j )
q̂ k = ------------------------------------------------- (39-28)
∑ Ji λi
j
The interlayer rate from layer k to layer j then may be written as:
q̂ kj = τ kj λ k ( P̂ k – P j ) (39-29)
where
R5000.0.1 39-601
VIP-EXECUTIVE® Technical Reference Guide Landmark
Jk Jj λ
τ kj = ---------------j- (39-30)
∑ Ji λi
j
λ mk
q̂ mkj = --------- q̂ kj (39-31)
λk
39-602 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Parameters on the PRINT card control the output at well management levels. The
FIELD parameter can be used to print production and injection summaries at all
well management levels.
FIELD
AREA 1 AREA 2
FS 3
FS 1 FS 2
GC 5 GC 6 GC 7
GC 1 GC 2 GC 3 GC 4
Wells Wells Wells
Wells Wells Wells Wells
■ Scale — Scale the rates of all wells to exactly meet the target.
■ GOR sort — Reduce the production rate of the highest GOR wells to user-
specified minimum rate until the target is met.
■ Water-cut sort — Reduce the production rate of the highest water-cut wells to
user-specified minimum rate until the target is met.
■ Gas rate sort — Reduce the production rate of the highest gas rate wells to
user-specified minimum rate until the target is met.
R5000.0.1 39-603
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ Water rate sort — Reduce the production rate of the highest water rate wells to
user-specified minimum rate until the target is met.
■ Average GOR — Reduce the production rate of wells whose GOR exceeds a
"target" GOR (only applicable for gas targets).
For the targeting options involving sorting to produce an ordered list of wells
(GOR, WCUT, etc), a frequency can be specified to control how often the lists are
recomputed (PTGFRQ card).
By default, the rates for the perforations for each well are recalculated based on
the reduced rate of the targetted phase. The user can specify the LSCALE ON
option so that the cut-back factor calculated for the targetted phase is applied to all
other phases for each well. Also, the perforation rates are cut back by the same
factor.
In addition, the user can specify the order in which phase targets should be
checked. The order in which the phase targets are checked can influence the final
distribution of producing wells.
The minimum rate used in these algorithms can be specified separately (using the
TRGQMN card) from the minimum economic rate for the well (specified using
the ECOLIM and QMIN cards).
A tolerance can be specified for each phase using the TRGTOL card. For any
remedial action to take place, the target must be exceeded by a fraction greater
than the tolerance (default value is 0.05 for all phases at all well management
levels).
A special feature called predictive well management can be used in the prediction
phase of a study to determine individual well rates and well assignment. This
feature is described in detail in Section 39.18.
39-604 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
R5000.0.1 39-605
VIP-EXECUTIVE® Technical Reference Guide Landmark
Additionally, the sales gas rate can be specified as a fraction of the oil surface
production rate.
1. Shrinkage gas
39-606 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
2. Fuel gas
3. Gas-lift gas
4. Sales gas.
Shrinkage gas is the gas lost because of the change of temperature and pressure in
the separator. It is not controllable once the temperature and pressure are fixed;
therefore, it is assigned the first priority. Fuel gas is required for field operations
and is assigned to the second priority.
If the produced gas is less than the specified shrinkage gas, then the shrinkage gas
volume is set to the produced gas volume. The produced gas is the produced
formation gas plus the gas-lift gas (at previous timestep) if the gas-lift gas is
included in the gas handling loop. The gas left after shrinkage (produced
formation + gas-lift at previous timestep + makeup + makeup gas-lift - shrinkage)
then is computed and compared to the specified fuel gas. If it is smaller than the
specified fuel gas, the makeup fuel gas (fuel gas - gas left) is added. The gas-lift
gas, sales gas, and gas available for reinjection are set to zero. Otherwise, the fuel
gas is subtracted from the gas left and is compared to gas-lift gas. If it is smaller
than the gas-lift gas, the gas-lift gas is reduced to the gas available. The sales gas
and gas available for injection are set to zero. Similar procedures are repeated for
calculation of sales gas. The remaining gas volume after sales gas is the gas
available for reinjection.
Optionally, the user may specify the composition of the gas available for
reinjection through the YREINJ card. The composition can either be set to the
produced gas composition or a user-specified mole fraction for each component.
The composition input on a YREINJ card applies only to the well management
entity designated on the card. It does not apply to any lower levels.
R5000.0.1 39-607
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ The “NONE” option neglects the calculated gas available for reinjection and
checks the specified target for each level of well management.
■ The “FIELD” option checks the specified injection target for each gathering
center, flow station, and area, but checks the effective target at the field level.
The effective target is defined as the minimum of the specified target and the
calculated gas available for reinjection.
■ The “ALL” option checks the effective target for each level of well
management.
■ Gas injection wells must be specified as either FSTD or FRES (INJ card),
then assigned to the appropriate level of well management.
■ The definition of the QMAX cards is the maximum gas injection rate, even
though the well type is FSTD or FRES.
39-608 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
39.10.1 Summary
The major gas sales option modifies the current gas handling loop computation
and processes the gas from the separators through an NGL plant to remove natural
gas liquids, an LPG plant to remove liquified petroleum gas, and an MI plant to
remove miscible injectants. In the NGL, LPG, and MI plants, the liquid recovery
factor for each component is specified as a function of a key component mole
fraction or the mole fraction of a key component plus fraction. Also, the fuel gas
rates specified by the user may exclude any component that has no BTU content
(e.g., CO2). For each gas handling loop calculation, up to two fuel gas streams can
be removed: one from the produced stock tank gas stream and the other from the
outlet gas stream of the MI plant. A portion of the outlet gas stream from the MI
plant may then be directed to a gas conditioning plant to remove the CO2
component; the outlet gas stream from the gas conditioning plant is the sales gas.
The remaining gas from the MI plant is then recombined with the available
makeup gas (if any) to form the reinjected lean gas. The CO2 component removed
from the gas conditioning plant may be either merged with the reinjected lean gas,
vented, or directly injected into a number of user-specified gas injectors. The
reinjected lean gas is distributed among the lean gas injectors specified as FSTD
gas injectors. Finally, the composition of the miscible injectant is used as the
injection composition for MI injectors (STD or RES gas injectors without
injection compositions specified). The major gas sales option may be selectively
invoked for any member of any well management level.
39.10.2 Description
In the surface facility at a certain field, the gas from the separators is sent to the
Central Gas Facility (CGF) where the field fuel gas is first extracted from the inlet
stream and the remaining gas is processed in the CGF to remove natural gas
liquids (NGL), liquified petroleum gas (LPG), and miscible injectants (MI). The
sales gas is removed from the outlet lean gas stream and is passed through a Gas
Conditioning Plant to remove CO2. The CO2 stream is then recombined with the
excess lean gas to produce a new lean gas injectant, or is vented, or is directly
injected into user-specified injectors. The methodology implemented here
represents a simplified approach to this complex process. It was determined that
in order to effectively simulate the CGF and the gas conditioning plant, CO2
R5000.0.1 39-609
VIP-EXECUTIVE® Technical Reference Guide Landmark
The major gas sales option is implemented as a part of the gas handling loop
computation in which the separator gas for each member of a well management
level (gathering center, flow station, area, or field) is processed to determine the
reinjection lean gas rates and compositions. The entire process is schematically
shown in Figure 39-2 and described in detail in the following sections.
Figure 39-2: Schematic Diagram of a Gas Handling Loop with the Major Gas Sales
Option
39-610 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
For each member of a well management level (e.g., Gather Center 1), the
component rates of the separator gas for all wells attached to the level are first
summed up to form the total component rates from which the total gas surface rate
and its composition for the member of the well management level are calculated
(see Box 1 in Figure 39-2). If the gas-lift gas is part of the gas handling loop, the
gas-lift gas rate from the previous timestep (GLGSV) is merged with the produced
gas stream (Box 2). Next, the shrinkage gas rate (GASSKG) is extracted from the
resulting gas stream (Box 3) with the constraint that the shrinkage gas rate should
not be greater than the available produced gas rate. In Box 4, the makeup gas-lift
gas rate (GASGLM), if specified, is added to the remaining gas stream, and the
user-specified first fuel gas (GASFUL) is removed. If the fuel gas rate is less than
the available gas, the current step’s gas lift gas (GASGLG) is removed from the
remaining gas stream and the resulting gas stream will be the gas available for the
NGL plant. In this case, the user-specified makeup gas (GASMKP) is not used in
this step and is the same as the available makeup gas (AVLMKP1) exiting from
Box 4. On the other hand, if the fuel gas rate is larger than the available gas, a
portion of the makeup gas will be used for fuel. In this case, the available gas for
the NGL plant is zero and the gas-lift gas for the current timestep is removed from
the remaining makeup gas to form the available makeup gas (AVLMKP1).
However, if the makeup gas (GASMKP) is not sufficient to satisfy the fuel gas
requirement, an extra makeup gas rate (GSMPFL) is calculated and the gas-lift
gas rate for the current timestep (GASGLG) and the amount of gas available for
the NGL plant are set to zero.
For a member of a well management level, if the major gas sales calculation is
specified by the user and the available gas for the NGL plant is greater than zero,
the gas stream will be processed first in an NGL plant to remove the natural gas
liquid. The composition of the inlet gas stream, y1(i), is the same as the produced
separator gas composition. For the NGL plant, the liquid recovery factor for each
component is tabulated as a function of either a key component mole fraction or
the mole fraction of a key component plus fraction. The mole fraction of a key
component plus fraction is defined as the sum of the mole fractions for the key
component plus higher numbered components in the inlet stream. The rates and
the compositions of the NGL and the outlet gas stream are then calculated.
The outlet gas stream from the NGL plant is then sent through an LPG plant in
which the LPG and the outlet gas stream’s composition and rates are calculated.
The outlet gas stream from this plant is then sent through an MI plant in which the
MI and the outlet gas stream’s composition and rates are calculated. The LPG and
MI plants are modeled the same way as the NGL plant with component liquid
R5000.0.1 39-611
VIP-EXECUTIVE® Technical Reference Guide Landmark
The user may specify the capacity of the NGL plant and/or the LPG plant by
entering the maximum feed rate and/or the maximum liquid rate (in surface gas
units) of each plant. If these constraints are in effect, a portion (Stream G1 in
Figure 39-2) of the inlet gas stream will bypass the NGL plant. The rate of stream
G1 is determined by satisfying both the maximum NGL rate (Stream QNGLOT in
Figure 39-2) and the maximum NGL feed rate (Stream QNGLIN) constraints.
Similarly, Stream G2 bypasses the LPG plant and the G2 rate is determined by
honoring both the maximum LPG rate (Stream QLPGOT) and the maximum LPG
feed rate (Stream QLPGIN) constraints. The G1 and G2 streams are then
combined with the outlet gas stream of the MI plant for further processing (see
Figure 39-2).
Fuel Gas #2
The major gas sales option allows some fuel gas (Fuel Gas #2) to be removed
from the outlet gas stream of the MI plant. If this outlet gas stream is larger than
Fuel Gas #2, the remaining lean gas amount (Lean Gas1, see Box 5) is calculated
and the available makeup gas is unchanged (i.e., AVLMKP1 = AVLMKP2).
Otherwise, Lean Gas1 is set to zero and a portion of the available makeup gas
(AVLMKP1) is used for fuel. In this case, a new available makeup gas rate
(AVLMKP2) is calculated.
In Box 6 of Figure 39-2, a sales gas stream is removed from the Lean Gas1, and
from the AVLMKP2 if the Lean Gas1 rate (excluding the CO2 content, if
applicable) is less than the specified sales gas rate (GASSLS). If a gas
conditioning plant is present, this sales gas stream is processed in the gas
conditioning plant to remove the CO2 component. The CO2 stream is then either
vented, distributed to a number of user-specified gas injectors, or combined with
the remaining lean gas and the available makeup gas (i.e., Lean Gas2 and
AVLMKP3 from Box 6) to form the reinjected lean gas stream (see Box 7).
The reinjected lean gas rate and its composition (yreinj(i)) determined from any
member of a well management level is applied to all lean gas injectors (with an
FSTD specification) attached to the member of the well management level.
Similarly, for MI injectors (with an STD or RES specification and no injection
compositions specified) attached to the member of the well management level, the
39-612 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
39.10.3 Implementation
The major gas sales option calculation may be declared for any members of a well
management level using a PLANT card. Without the PLANT card, the standard
VIP gas handling loop calculation will be performed. The major gas sales option
may be invoked for some well management levels and not invoked for other
levels. The detailed description for the PLANT card is given in the VIP-
EXECUTIVE Reference Manual.
The major gas sales option involves four main items: an NGL plant, an LPG plant,
an MI plant, and gas conditioning. The user has the control of invoking any or all
of the four main items. The NGL plant calculation can be invoked by an
NGLPLANT card followed by an NGL plant table. Similarly, the LPG plant
calculation is invoked by an LPGPLANT card followed by an LPG plant table and
the MI plant calculation is invoked by an MIPLANT card followed by an MI plant
table. The gas conditioning calculation can be declared by a GASCOND card
which also contains a user-specified component number. This component (e.g.,
CO2) will be excluded from the input and the output fuel gas rates. It will also be
removed in the gas conditioning plant during the sales gas calculation and handled
according to the data on the GASCOND card, if the gas conditioning plant
calculation is specified by the user through keyword PLANT on the GASSLS
(sales gas) card. In addition, up to two fuel gas streams may be removed during
the gas handling calculation: one from the produced separator gas stream and the
other from the outlet gas stream of the MI plant. This enhancement requires
modification of the GASFUL (fuel gas) card. The detailed descriptions for all of
the new and modified data are given in the VIP-EXECUTIVE Reference Manual.
As in the standard gas handling loop calculation, the reinjected lean gas
composition calculated for a member of a well management level will be used for
FSTD reinjection wells attached to the member. However, if an injection
composition for any well is entered through a YREINJ card, the composition
specified on the YREINJ card will be used as the injection composition. For an
MI injector, the user may specify the well management level for which the
calculated MI composition is to be based by entering keyword GATHER,
FLOSTA, AREA, or FIELD on the INJ card. In this case, a YINJ card for the MI
injector must not be entered. Otherwise, the composition specified on the YINJ
card will be used and the calculated MI composition will be ignored. As
mentioned, a QMAX card for all MI injectors must be entered to specify the MI
injection rate for each well. The MI rates calculated from the gas handling loop
R5000.0.1 39-613
VIP-EXECUTIVE® Technical Reference Guide Landmark
are not used. This modification to the INJ card is also described in the VIP-
EXECUTIVE Reference Manual.
The format for the Field Gas Handling Summary output (which will be printed if
keyword FIELD is entered on a PRINT card) is unchanged. If two fuel gas
streams are specified, the fuel gas shown in the summary will be the sum of the
two rates. Also, the reported fuel gas rate excludes the component identified on
the GASCOND card.
The NGL plant, LPG plant, and MI plant feed, liquid, and vapor rates and
compositions are reported as a subset of the separator report (invoked by keyword
SEP on the PRINT card). In addition, the calculated reinjected lean gas
composition is shown in the last column of the MI plant report. Notice that this
information is printed only for the members of the well management levels where
the calculated MI compositions are to be used as the injection compositions of the
MI injectors.
It should be noted that the feed composition for the NGL plant may be different
from the stock tank vapor composition shown in the separator battery report. This
is the case for multiple producer cases with non-uniform production compositions
because the feed composition for the NGL plant is based on the sum of individual
producer’s stock tank gas rates, whereas the stock tank vapor composition in the
separator report is determined by flashing the total molar production rates (i.e.,
sum of all well molar rates) for the separator battery. These two compositions will
be identical only if all producers have the same overall production compositions.
The major gas sales option is allowed only for compositional models and is
available for both IMPES and implicit modes.
39-614 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ Enable both gas and water injection with net voidage control.
■ If the user assigns gridblocks to injection regions, but does not explicitly
assign injection wells to injection regions, the program assigns the injection
region for each injector according to the location of the top perforation. If any
injection wells are explicitly assigned to injection regions, then all injection
wells should be.
■ An injector that does not physically exist within a region can still be assigned
to it.
■ Injectors in an injection region may inject water or gas. They can be specified
as an STD, FSTD, RES, or FRES well using the INJ card.
■ For the FRES injector, the qmax value on the QMAX card is the maximum
injection rate at reservoir conditions.
■ For the FRES gas injector, the injection gas composition used is the same as
that used in the gas reinjection option at the specified well management level.
R5000.0.1 39-615
VIP-EXECUTIVE® Technical Reference Guide Landmark
A pressure term, PTERM, defined in the following equation, is used for pressure
maintenance.
exp
⎛ P ref – P avg ⎞
PTERM = ⎜ ---------------------------⎟ (39-32)
⎝ P ref – P targ⎠
where Pref, Ptarg, and exp are user-specified reference pressure, target pressure,
and exponent, respectively. The reference pressure, Pref, must be greater than the
target pressure, Ptarg, and the average pressure, Pavg. The average pressure can be
either the pore volume weighted average or the hydrocarbon pore volume
weighted average pressure.
Two options, PROPTN and UNIFORM, are available for distribution of the total
reservoir injection rate to the voidage controlled wells. In the UNIFORM option
(which is the default option), the injection fluid is distributed equally to the
39-616 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
voidage controlled wells that are not constrained by injectivity. For wells
constrained by injectivity, the maximum well injectivities are honored. The
remaining rate then is distributed equally to the rest of the wells. For the PROPTN
option, the injection fluid is distributed proportionally to each net voidage
controlled well according to injectivity. Injectivity is subjected to the maximum
injection rate and pressure constraints.
Source water to the field may be specified using the IRSRCW card. The
maximum water injection to the field is the minimum of the field injection target
and the sum of the field production and specified water source. The procedure to
determine the gas available for injection to the field is discussed in Section 39.9.4.
The maximum gas injection to the field is the minimum of the specified field
injection target and the field available gas. The maximum injection to a region is
the minimum of the injection target for the region and the sum of the well
injectivities. The injection target for each region can be specified as a percent of
the total field target.
The injection region targets are checked first, followed by the field target. If the
total injection rate exceeds the maximum injection, all injection rates in the region
or in the field are decreased proportionally. However, if the calculated well rate is
smaller than the specified well minimum, then the minimum rate is honored. The
remaining well rates are decreased proportionally according to the remaining
target.
A special option, REDIST, is available in the general injection region. When this
option is selected, the extra injection fluid for a region will be redistributed to
other regions if the available amount cannot be fully injected into that injection
region. The injection fluid available for each region is calculated as the maximum
injection rate for the field multiplied by the percentage specified on the IRPCTA
card. The actual fluid injected into the region can be equal to or smaller than the
available amount. The cut back is due to voidage replacement, pressure
maintenance, maximum injection rate for wells in the region, or regional injection
target limitations. In the REDIST option, the difference between the available
amount and the actual amount will be redistributed to the other regions. The
amount distributed into each region is proportional to its injectivity.
R5000.0.1 39-617
VIP-EXECUTIVE® Technical Reference Guide Landmark
The total injection rate for the wells in priority 1 projects is calculated first. This
rate is then compared to the field effective target that was calculated earlier
(Section 39.9.6). If it is larger than the effective target, the injection rate for
priority 1 wells is cut back proportionally and no further calculation is done for
the lower priority wells. Otherwise, the effective target is reduced by the total rate
allocated to priority 1 wells. If the new effective target is not zero, then the
process is repeated for the next lower priority.
The user cannot specify FRES gas injection wells in the prioritization option. In
the “general injection region” option, the FRES well type is defined as a net
voidage injector. To add the capability of injecting gas according to specified
priorities and injecting water according to net voidage, water injection is
calculated after gas injection when the prioritization option is chosen.
The UNIFORM gas injection option is not available when the general injection
region option is used. This restriction also applies to the prioritization option.
39-618 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ The user must determine the fractional value injected at each injection well.
If both water and gas injection are taking place at a well management level, net
voidage injection may be more applicable. Net voidage is defined as the
difference between the total reservoir production rate and the total reservoir
injection rate for the phase not under net voidage control. For example, if a fixed
quantity of water is being injected, the gas injection required to maintain voidage
balance is the difference between total production and water injection.
The user can specify the injection target for an area as a fraction of its voidage.
The program calculates the area injection rate from area voidage and input
fraction. The calculated area injection rate can be allocated to each flow station
and the fluids from the flow station can be allocated to the gathering center
according to the respective voidage. The injection rate at the gathering center then
can be allocated to the injectors according to the well injectivities. If both water
R5000.0.1 39-619
VIP-EXECUTIVE® Technical Reference Guide Landmark
and gas are injected at the same well management level, then the net voidage
option is better for balancing fluid injection.
■ Only one phase (water or gas) can be in net voidage control, not both. An
error message is printed and the simulation run stops if both phases are under
net voidage control.
■ When using the INJTAR cards, do not mix injection targets at reservoir
conditions with those at surface conditions for different well management
levels.
39-620 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ The injection target for the upper well levels is allocated to the lower levels. If
the user specifies the injection targets for several levels, then they must be
specified at the same conditions (reservoir or surface).
■ The RSTD and FSTD options in the INJTAR cards are for surface conditions.
The RRES, FRES, and FRESN options are for reservoir conditions.
■ For different injection fluids, the user is allowed to use different conditions in
the INJTAR card. For example, the gas injection targets can be specified at
surface conditions and the water injection targets can be specified at reservoir
conditions.
■ The units for guide rate are the same as those for injection target.
■ For the default well guide rate, well injectivity at surface conditions is used if
the well defined in the INJ card is at surface conditions. If the well is defined
at reservoir conditions, the well injectivity at reservoir conditions is used. The
GURT option in the INJGR card has the same units as the injection target. For
the VOID and VOIN options, the calculated total and net voidage are used
directly in allocation of injection targets.
R5000.0.1 39-621
VIP-EXECUTIVE® Technical Reference Guide Landmark
39.14 Gas-Lift
In VIP-EXECUTIVE, gas-lift is used with the tubinghead pressure (THP) option
for three-phase producers. The gas-lift gas allocated to a well is added to that
well’s produced gas to derive the effective gas-liquid or gas-oil ratio for the well.
This ratio, along with the well’s liquid or oil rate, water cut, and THP constraint, is
used to look up the flowing bottom-hole pressure in the wellbore hydraulics table
(BHPTAB data). Gas-lift is otherwise transparent to the user; i.e., the gas-lift rate
is not included in daily gas production. The gas-lift option can be utilized only for
wells for which THP data have been entered.
A well’s gas-lift gas rate can be entered with one of two options on the QLIFT
card. A positive gas-lift gas rate causes a constant allocation of that rate to the
well. A negative value (any negative value is sufficient) causes invocation of an
automatic allocation procedure for the well. In predictive well management, all
wells in a gathering center with gas-lift are automatically eligible for gas-lift.
Here, a negative value of gas-lift gas rate has no significance. To prevent a well
from becoming eligible for gas-lift, a zero gas-lift gas rate must be specified.
Note that the gas-lift gas rate is calculated only during the first iteration of each
timestep. This calculated rate then is used for subsequent iterations.
Variations on the OPTTAB option relate to handling of the gas-lift gas shortage.
The user specifies on the QLIFTA card the amount of gas-lift gas available for
automatic allocation. If the amount allocated for all the wells on gas-lift exceeds
the available gas, some mechanism must be employed to cut back the gas-lift gas
rates. If no special option is specified, the program simply stops allocating gas-lift
gas to subsequent wells. The TABGLE option indicates that gas-lift is to be
removed from wells based on gas-lift efficiency (oil rate/lift rate). The TABWC
option indicates that gas-lift is to be removed from wells based on water cut. The
TABSCL option indicates that gas-lift is to be scaled back from all wells.
39-622 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
The PFMCRV option enables the user to calculate gaslift gas rates based on the
lift efficiencies of the wells, where lift efficiency is defined as the incremental oil
produced per increment of gaslift gas. Usage of this option requires that the
BHPTAB wellbore hydraulics tables have been input with sufficient range to
cover all of the oil, water, and total gas (produced reservoir gas plus gaslift gas)
rates that may be encountered with the gaslift calculations. A word of caution
should be noted here that the wellbore hydraulics algorithm will extrapolate
outside of the table range, if necessary.
In order to evaluate lift efficiency, a set of operating points are first calculated for
the well. These operating points are for gaslift gas rates approximately log-
linearly spaced between .2 MMSCF/D and 15.0 MMSCF/D, possibly adjusted by
the maximum GOR/GLR value in the specified BHPTAB table. A maximum of 9
operating points will be attempted. These operating points are the intersections
between the PI line (oil rate versus bottomhole pressure) and the lift curves (oil
rate versus bottomhole pressure for the various gaslift gas rates). See Figures 39-3
and 39-4.
These operating points define the performance curve for the well, for this point in
time, and the derivative of the curve is the lift efficiency. In order to facilitate the
interpolation between points and the evaluation of the derivative there, a
logarithmic function is fit through a subset of the pointcs. The function is:
2
Q o = A + B × [ ln ( GL ) ] + C × [ ln ( GL ) ]
The procedure starts by fitting a curve through the first 3 operating points.
(Special cases of less than 3 operating points will be described later). If the
efficiency at the third operating point is higher than the desired efficiency and if
there are additional operating points at higher gaslift gas rates, the curve fit will be
repeated for points 2, 3, and 4, and the efficiency calculated at point 4 and tested
against the desired efficiency. This process will be repeated (points 3, 4, 5; then 4,
5, 6; etc.) until an efficiency less than the desired efficiency is found or until there
are no additional operating points. At this time the equation can be solved for the
gaslift gas rate which yields the desired efficiency. If necessary, this curve will be
extrapolated outside the range of the last set of points used.
R5000.0.1 39-623
VIP-EXECUTIVE® Technical Reference Guide Landmark
1. If a well can make its QMAX at one of the gaslift gas rates, that point and all
higher lift rate points are discarded. If the desired lift efficiency exists within
any of the remaining points, the required lift rate is calculated. Otherwise, its
lift rate is set to the lowest lift rate point at which it could make QMAX.
Figure 39-3: Intersection of the Inflow Performance Curve (-PI) and the Lift Curves
for the Various Gaslift Gas Rates.
39-624 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
2. If only two operating points are found, the following logarithmic function is
fit through them:
Q o = A + B × ln ( GL )
If the desired efficiency exists between these two points, the required lift gas
rate is calculated. However, this curve will not be extrapolated. If the desired
efficiency is at a lift rate less than the lower operating point, the lift rate at the
lower operating point will be used. Also, if the desired efficiency is at a lift
rate higher than the higher operating point, the lift rate at the higher operating
point will be used.
3. If only one operating point is found, the "total" efficiency at the point (total oil
rate divided by the lift gas rate) is compared to the desired efficiency. If the
total efficiency is greater, the lift gas rate at that point is used of the next
timestep. If the total efficiency is less, then the lift gas rate is set to zero.
This process is repeated for each well, resulting in the total lift gas which is
required in order that each well will operate at the desired efficiency. If this total
lift gas requirement is less than the maximum available, these gaslift gas rates will
be used for the next timestep. However, if the required gas exceeds what is
R5000.0.1 39-625
VIP-EXECUTIVE® Technical Reference Guide Landmark
available, then there are several options as to how to reduce the required gas to
less than or equal to what is available.
There are two basic paths for reducing the total gaslift gas volume requirements.
These paths are; 1) efficiency scaling, and 2) the combined sequence of "hit-list" -
uniform scale-back and shut-in of low "total" efficiency wells.
2. User-specified "Hit-List"
The user can specify an ordered list of wells which will have their lift gas
turned off, one by one, until either the gas availability is reached or the list of
wells is exhausted. When a well has its lift gas turned off, the model will
decide whether the well can flow or not (based on the solution GOR and the
lift curve data).
3. Uniform Scale-Back
After exhausting the hit list wells, if there is still a gas shortage then all of the
wells will have their lift gas scaled back by whatever percentage it takes to
reach the gas availability (subject to a user-input maximum allowable
scaleback). The model also honors an additional set of user-input minimum
and maximum allowable GLR’s for the scaleback step.
Additional information is packed into the Production Well Summary for wells that
are using the PFMCRV option for automatic gaslift gas rate allocation. Since the
print line is completely full, the single column between CUM LIFT GAS and
LIFT STATUS is used for an additional status code. (The lift status will be
PFMCRV). The codes are as follows:
39-626 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
Wells for which at least two potential operating points were found an for which a
non-zero gaslift gas rate was determined, a second line is printed containing the
incremental gaslift efficiency for the well.
The gas-lift rate is used by the program only to determine the flowing bottom-hole
pressure that corresponds to the user-specified tubinghead pressure limit. Gas-lift
is otherwise transparent to the user; that is, the gas- lift rate is not included in daily
gas production. Gas-lift can be used only on wells for which a THP card is
entered.
R5000.0.1 39-627
VIP-EXECUTIVE® Technical Reference Guide Landmark
The maximum number of pump flow rates and pump discharge pressure values is
specified using the NPMPV parameter on the DIM card. In addition, the
maximum number of pump tables is specified using the NPMPMX parameter on
the DIM card.
39-628 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
39.16.1 Introduction
The Surface Facility Model (TSFM) extends the calculation of hydrocarbon
recovery volumes beyond the standard multi-stage surface separator conditions.
The liquid from the last stage of the separator is further processed through an oil
stabilizer, and any additional liberated gas from the stabilizer is added to the gas
from the surface separators for input into a gas processing plant. Recovery factors
are input for each component in order to calculate liquid recovery volumes, and
subsequently the dry gas volume out of the plant.
gas
gas rate
gas
plant
gas
separator
battery gas
oil
oil
stabilizer
H/C/well stream
oil
stock
oil rate tank
R5000.0.1 39-629
VIP-EXECUTIVE® Technical Reference Guide Landmark
39-630 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
39.17.1 Objectives
The automatic workover feature in VIP-EXECUTIVE has the following
objectives:
■ All field workovers that are meaningful in the simulator are identified and
performed.
■ Only one oil perf, gas perf, or water perf can be performed.
R5000.0.1 39-631
VIP-EXECUTIVE® Technical Reference Guide Landmark
Workover Rigs
Workovers are performed by rigs. Rigs may be assigned to any member in a well
management level: gathering center (GC), flow station (FS), area, or field. A rig
may be used at any gathering center within that well management member. For
example, a rig assigned to the field may perform a workover at any gathering
center, while a rig assigned to a gathering center may perform workovers in that
gathering center only.
Defining rigs involves specifying the number of available rigs, the time required
to complete a workover, and the time required to move a rig from one gathering
center to another (when necessary). Additional input allows any number of rigs to
be removed from a member in a well management level or for a new set of rigs to
totally replace a previously specified set.
where CURTIM is current time, AVLTIM is time at which the rig is available,
WRKTIM is time required to complete a workover, and MVTIM is time needed to
move a rig, if necessary. The only exception to this rule is that a rig is immediately
available to perform its first workover.
Only wells defined as producers are considered for automatic workovers. The
initial criteria for a well to be eligible for any workover are that:
■ A rig is available at the well’s gathering center, flow station, area, or field.
■ Enough time has elapsed since the last workover was performed on the well.
39-632 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
ON) and satisfies limits discussed in the next section. Additionally, the user may
specify a gas-oil ratio minimum (water-cut minimum) that the well must satisfy to
be eligible for a gas-shutoff workover (water-shutoff workover).
Note that a shut-in well is eligible for opening and shutoff workovers. The well
must have produced at some point before being shut in.
The concept of having all perforation operations occur during the workover
process requires certain data restrictions to the existing well and perforation limit
options. The PRFLIM option of checking individual perforations for GOR and
water cut is not allowed. The PLUG option within the GLIMIT and WLIMIT data
is not allowed, since PLUG closes the worst-offending perforations. The only
GLIMIT/WLIMIT options allowed are SHUTIN and LIMIT.
The number of workovers performed during a timestep depends on rig and well
availability. As long as a candidate well exists and a rig is available to the
gathering center to which the well belongs, the workover will be performed. The
type of workover performed is a direct function of user input: user-supplied data
define the relative number of each type of workover to perform. For example, let
values A, B, and C define the number of openings, gas shutoffs, and water
shutoffs, respectively.
The first 5 openings will be performed, then 2 gas shutoffs, then 3 water shutoffs,
then 5 openings, etc.
The first 4 gas shutoffs will be performed, then 1 water shutoff, then 4 gas
shutoffs, etc.
The count for each workover type is not reset each time workovers are performed.
From the first example, if only three opening workovers are to be performed
during the first set of workovers, then the next time workovers are performed, two
opening workovers will be performed, followed by two gas shutoffs.
The choice of wells on which the workovers are performed is based on the benefit
functions. Each of the three lists of candidate wells is sorted based on the
appropriate benefit function values. (Note that a well may be in more than one
list.) The top well in the appropriate list then is selected. If a rig is available, all
workover types for which the well is a candidate are performed. All workovers are
R5000.0.1 39-633
VIP-EXECUTIVE® Technical Reference Guide Landmark
When the PWM option is not in use, workovers will be performed at user-
specified intervals.
39-634 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ Always produce from a certain area of the field or from a specified set of
wells.
■ Produce from the lowest to the highest gas-oil ratio (GOR) well in the gas cap.
■ Produce from the lowest to the highest water-cut well in the area near the oil-
water contact.
■ The governing conditions for the prediction phase of the simulation study,
such as minimum well rate, maximum GOR, etc.
■ Allow the simulator to calculate well production rates on the basis of well and
facility data.
R5000.0.1 39-635
VIP-EXECUTIVE® Technical Reference Guide Landmark
Pressure Systems
Well production rates are generally unknown for the production phase of a
simulation study. However, the probable range of tubinghead pressures for a well
can usually be estimated. A "pressure system" corresponds to the tubinghead
pressure for a well. In the NEW option, the user can specify any number of
pressure systems for each well. In the MGOR option, only two pressure systems
exist: the low pressure system (default name is SYS1) and the high pressure
system (default name is SYS2). The program determines to which pressure
system each well flows and at what rates.
In many fields, artificial lift techniques are used to enable low productivity wells
to flow. During the history match period the engineer has a priori knowledge of
wells on artificial lift. During the prediction phase of a study, however, the
following questions must be answered:
39-636 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
■ How does artificial lift impact facility size at various well management
levels?
Some of the commonly used artificial lift methods in the field are gaslift, sucker
rod pumps and downhole submersible pumps. VIP-EXECUTIVE can simulate
only gaslift at present. In the NEW option, gaslift may be available at one or more
pressure systems in a gathering center and it is not necessary that gaslift be
available at all gathering centers. The MGOR option restricts gaslift to the low
pressure system only. Wells on gaslift are automatically connected to the low
pressure system.
When the total production rate at any well management level exceeds the user-
specified target, there are two options available to reduce the production rate:
In the first case, the model requires a set of rules to determine which wells could
be shut-in. Benefit functions are a convenient way to determine the order in which
wells could be shut-in. Similarly, benefit functions can be used to determine the
order in which wells are first opened to flow.
■ It can be used to exclude (or include) wells with specific characteristics such
as:
The user will have the option of choosing one of the following equations to
calculate the benefit function:
BF = PM × PA × ( A + B × Q g + C × Q o + D × Q w + E × Q glg )
⁄ ( F + G × Q g + H × Q o + I × Q w + J × Q glg ) , or
R5000.0.1 39-637
VIP-EXECUTIVE® Technical Reference Guide Landmark
B D F
BF = PM × PA × ( Q o + A ) × ( GOR + C ) × ( WCUT + E )
H
× ( GLGOR + G )
where:
Qglg Gaslift gas production rate for the well, MSCF/D (SCM/
D),
A, B, C, D,
E, F, G, H,
I, and J User defined coefficients.
The total incremental gas-oil ratio (TIGOR) for a well at point i on its generated
production curve is defined as:
( Q g + Q glg ) – ( Q g + Q glg )
i i–1
TIGOR i = ----------------------------------------------------------------------------
- (39-33)
Qo – Qo
i i–1
where:
39-638 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
The marginal gas-oil ratio (MGOR) is the TIGOR value above which no
production is added. The MGOR is computed for each entity in each well
management level (gathering center, flow station, area and field). In addition,
values are computed for the high pressure system, the low pressure system (all
wells) and the low pressure system (gaslifted wells only).
The production rate of a well depends on its THP. The possible flow rates for a
well depend on the number of THPs (or pressure systems) to which it can flow. In
addition, a well may be able to use artificial lift methods. Therefore, a well may
have more than one set of possible flow rates under appropriate conditions. The
first step in the NEW option consists of determining all such possible flow rates
for a well.
The user-specified data are used to determine the eligible THP values and
artificial lift methods for each well. The inflow performance curve for the well is
determined using the perforation data. The outflow performance is determined
using the user-supplied tabular data relating bottomhole pressure (BHP) and THP.
The flow rate at a given set of conditions is determined from the intersection of
inflow and outflow performance curves. This calculation is repeated for each
eligible pressure system and artificial lift method for the well.
At the end of the above step, a well could flow to more than one pressure system
and may have rates using one or more lift methods. A set of rules is used to select
the most appropriate pressure system for each well. This process is termed
"hydraulic categorization". The categories represent "bins" into which wells
would be sorted based on
R5000.0.1 39-639
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ the inability to produce at the minimum level of the next higher "bin" and
The following rules are used to determine the hydraulic category of each well:
■ The category depends on the ability of a well to flow above a certain (user-
specified) minimum rate. This minimum rate could be either oil rate or gas
rate.
■ The category is equal to the highest pressure system to which a well (without
lift) can flow above the minimum rate.
■ The category for a well can be fixed by the user, but the well production rate
must be above the minimum rate.
It is expected that an artificial lift method would increase the production rate for a
well (at a fixed THP). Therefore, a well is considered ineligible for artificial lift if
the incremental oil production rate "(lifted - non-lifted)/non-lifted" is less than a
user-specified threshold. The default value of this threshold is five percent.
At the end of this step, each well is assigned a category number. It should be noted
that a well can flow into any pressure system at or below its category, i.e., a well
categorized to flow to a high pressure system is still eligible to flow to a low
pressure system. The actual pressure system to which a well would flow is
determined in the next step.
In this step, wells are assigned to a specific pressure system and lift method (if
eligible). The well assignment is based on the category of the well and the user-
specified data. Multiple passes can be made in this step to selectively assign wells
to a specific pressure system at a specific well management level. The following
data must be specified by the user for each pass:
39-640 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
❑ If its water-cut, gas-cut ratio and gas-liquid ratio are below the user-
specified values for this pass. These cutoff values are optional.
2. The benefit function for each eligible well is then calculated. The user can
specify a different benefit function for each pass.
4. Wells are brought on production until the well list is exhausted or any
appropriate target is met at the specified well management level. If a target is
met, one of the three available options specified by the user is used to reduce
the rate to be equal to the target. The three options are:
❑ Scale back all rates by the same factor to exactly meet the target.
❑ Shut-in wells (and scale the last well) to exactly meet the target.
❑ Use an averaging technique such that the rates of the highest rate wells are
reset to an average rate to meet the target.
In the "shut-in" and "average" options, the well is not truly shut-in (such as shut-in
due to minimum rate, maximum GOR constraints). It will be eligible for
assignment in all subsequent passes and timesteps.
At the end of this step, either the production targets at the specified well
management levels are exactly met or all eligible wells are assigned. Since targets
are not checked at any higher well management level, one or more of them may
have been violated. In addition, it is possible that some targets at the lower levels
are not met and eligible wells have not been assigned.
This step is used to reduce production at those well management levels where the
targets have been violated. Multiple passes can be made during this step to select a
specific phase at a desired well management level. The user must specify the
following data for each pass:
■ The pressure system entity (or the entire well management level) to be
checked.
R5000.0.1 39-641
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ Phase to be checked.
1. It is determined if the target has been exceeded for the specified phase for
each member at the specified well management level.
2. If the shut-in option is used, wells are shut-in in the benefit function order
until the rate is equal to the target.
3. If the scale-back option is used, all eligible wells are scaled back
proportionally to meet the target.
The target must be exceeded by a fraction greater than the tolerance (default value
is 0.05 for all phases at all well management levels) for any remedial action to
take place.
At the end of this step, all user specified targets are compared with the actual
production rates. If any target is being violated, a warning message is printed. The
user can specify the action if such an event occurs. The possible choices are to
stop the run (after writing a restart) or continue the run after printing the warning
message.
The inflow performance curve for each well is determined using the perforation
data. The outflow performance is determined using the user-supplied tabular data
relating bottomhole pressure (BHP) and THP. The flow rate at a given set of
conditions is calculated from the intersection of the inflow and outflow
performance curves.
1. Calculate the flow rates at its HP THP limit if the well is eligible to flow to the
HP pressure system. Calculate the flow rates at four additional points (20%,
40%, 60%, and 80% of HP THP limit oil rate). If the gas-oil ratio (GOR) of
the 20% point is within ten percent of the GOR of the HP THP limit point
39-642 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
(i.e., the well is not rate sensitive), discard these points: they are not needed in
the subsequent calculations. The well is considered incapable of flowing to
the HP system if its water-cut at the HP THP limit exceeds the user-specified
water-cut limit.
2. Calculate the flow rates at its LP THP limit if the well is eligible to flow to the
LP pressure system. If the well is rate sensitive (rate sensitive in the HP
pressure system or the GOR at the LP THP limit point is not within ten
percent of the GOR at the HP THP limit point), calculate the flow rates at four
additional points between the oil rates corresponding to the HP THP limit and
the LP THP limit. The well is considered incapable of flowing to the LP
system if its water-cut at the LP THP limit exceeds the user-specified water-
cut limit.
3. If the well is eligible for gaslift, calculate the flow rates at its LP THP limit
with gaslift. The amount of gaslift gas injected is computed using the
performance curve algorithm with the lift efficiency being either user-
supplied or table-derived. If the fixed gaslift gas rate option is used, no other
point is generated. The well is either on gaslift at this corresponding rate or
not on gaslift. If the variable gaslift gas rate option is used four additional
points are generated. These points correspond to successively smaller
amounts of gaslift gas.
Note that a well will be placed on gaslift only if its lifted oil rate satisfies the
user-specified minimum incremental benefit due to gaslift and the water-cut
does not exceed the user-specified water-cut limit. If these requirements are
not satisfied, the offending point will not be included in the well performance
curve.
4. The lift efficiency used in the gaslift gas calculations under the performance
curve option is determined in one of two ways:
❑ User-specified value, or
As defined previously, the total incremental gas-oil ratio (TIGOR) is the slope of
each segment of total gas rate (QG + QGLG) versus oil rate (QO) from the well
performance curve. The MGOR option utilizes the TIGOR versus QO curve for
each well to determine the well’s rates at increasing values of TIGOR. The
TIGOR versus QO curve is generated as follows:
■ Define the TIGOR value at the first point as the GOR of the first point on the
well performance curve.
R5000.0.1 39-643
VIP-EXECUTIVE® Technical Reference Guide Landmark
■ Compute the TIGOR value at each subsequent point as the slope of (QG +
QGLG) versus QO between this point and the previous point on the well
performance curve.
■ Do not allow the TIGOR values to decrease. That is, the following equation is
enforced for each point j on the curve after the first:
Well Assignment
2. Determine the flow rates for each well at the current FIGOR value.
4. Determine if any excess capacity remains and if there exist any wells whose
rates can be increased.
5. If so, increase the current FIGOR value by the user-specified increment and
return to Step 2.
8. When the flow rate calculations are completed compute the marginal gas-oil
ratio (MGOR) for each entity in each well management level.
In general, wells will flow at the oil, gas and water rates corresponding to the
current value of FIGOR. Complications arise, though, in getting to this point or in
discovering that the well cannot flow at this FIGOR value. The following steps
are used in attempting to flow a well at FIGOR:
1. The rates may not be increased from the rates at the previous value of FIGOR
if any of these conditions hold:
b. The current value of FIGOR is less than the TIGOR value of the first
point in this well’s TIGOR versus QO curve.
39-644 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
c. A low pressure system target applicable to this well has been reached and
the well has already reached its maximum high pressure system rate.
d. A gaslift gas target applicable to this well has been reached and the well
has already reached its maximum non-lifted rate.
2. Determine the point in the TIGOR versus QO curve in which the current
FIGOR value lies.
a. If the gaslift gas rate cannot be increased and the current FIGOR value
lies in the gaslift region of the curve, set the point to the maximum non-
lifted point.
b. If the low pressure system rate cannot be increased and the current
FIGOR value lies in the LP system region of the curve, set the point to the
maximum high pressure system point.
3. If appropriate, compute the gaslift gas rates corresponding to the point on the
TIGOR versus QO curve.
4. Compute the oil, gas and water rates corresponding to the point on the TIGOR
versus QO curve.
Once the well flow rates have been computed for the current FIGOR value, the
user-specified targets must be checked. Any gathering center targets are checked
first, then any flow station targets, then any area targets and then any field targets.
For each entity in each well management level, the various targets are checked
and satisfied in the following order:
1. Gaslift gas: If the gaslift gas rate violates the available gas decrease the
current FIGOR value by ten percent (10%) of the difference between the
current and the previous FIGOR values. Continue this process until the
violation is eliminated. A well’s production rates will not be decreased below
those corresponding to its maximum non-lifted point.
2. Low pressure system: If any of the low pressure system targets are still
violated by more than the user specified tolerance:
c. For those wells pointing to the low pressure system, compute the rates
based on the new FIGOR value. A well’s production rates will not be
decreased below those corresponding to its maximum high pressure
system point.
R5000.0.1 39-645
VIP-EXECUTIVE® Technical Reference Guide Landmark
d. If any low pressure system target remains violated by more than the
tolerance, reset FIGH to this FIGOR and return to Step b.
e. Once the targets are satisfied, check that the cuts were not too deep. That
is, at least one of the violating rates should be between the target rate and
the target tolerance rate. If not, increase the current FIGOR to accomplish
this; i.e., reset FIGL to the current FIGOR and return to Step b. Note that
by increasing FIGOR a target can again be violated. Thus FIGOR can be
decreased, increased, decreased, etc., to satisfy the targeting criteria.
f. The final FIGOR value becomes the current FIGOR for subsequent
targeting checks.
3. High pressure system with no excess capacity in the low pressure system: If
any of the high pressure system targets are still violated by more than the user-
specified tolerance but no wells can be moved to the low pressure system
because it is already full, perform the same algorithm as in Step 2. Obviously
in this case the wells under consideration are the ones pointing to the high
pressure system.
4. Total: If any of the total targets (the entity as a whole) are still violated by
more than the user-specified tolerance, perform the same algorithm as in Step
2 on wells pointing to either pressure system.
5. High pressure system with capacity in the low pressure system: If any of the
high pressure system targets are still violated by more than the user-specified
tolerance and there is excess capacity in the low pressure system, try moving
wells from the high pressure system to the low pressure system in the
following manner:
a. Wells eligible to be switched to the low pressure system must satisfy the
criteria:
❑ TIGOR value at the HP THP limit point (TIGOR*) is larger than the
current value of FIGOR.
c. Starting at the beginning of the list, reassign wells to the low pressure
system until the high pressure system targets are no longer violated. Do
not switch a well if its rates would cause a low pressure system target to
be violated by more than the tolerance.
d. If any high pressure system target remains violated (due to low pressure
system violations and/or not enough eligible wells), undo the switches
performed in this step. Go back to Step 3 to perform the other algorithm
for high pressure system violations.
39-646 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
As defined previously, the marginal gas-oil ratio (MGOR) is the TIGOR value
above which no production is added. The MGOR is computed for each entity in
each well management level. In addition, values are computed for the high
pressure system; the low pressure system (all wells); and the low pressure system
(gaslifted welsl only). Each of these values is computed as the maximum TIGOR
value over the set of appropriate wells. An MGOR report is available to view
these results.
Once one of the PWM options is initiated, PWM calculations are performed
automatically at every DATE/TIME card. The user can also control the frequency
of additional PWM calculations in any of three ways:
During the timesteps when PWM calculations are performed, the user can control
the number of outer iterations with PWM calculations. The keyword controlling
this number is WMITN and the number of iterations generally is set to 1. (During
history matching the value of WMITN is zero, causing no PWM calculations to be
performed.) After the first WMITN iterations of the timestep, the following
procedure is used:
1. The well rates are saved (after iteration number WMITN) for wells cut back
to meet well management targets. These rates are considered to be the
maximum flow rates (QMAX) for the wells. In addition, those wells that are
not cut back due to targets are considered to be bottomhole pressure
constrained at the flowing bottomhole pressure calculated in iteration number
WMITN.
2. The rates for cut-back wells are fixed at the temporary QMAX for the
remaining iterations of the PWM timestep. However, the rates for bottomhole
pressure constrained wells can float due to pressure changes during
subsequent iterations.
R5000.0.1 39-647
VIP-EXECUTIVE® Technical Reference Guide Landmark
During the timesteps when PWM calculations are not performed, well rates are
determined using the same procedure as that used after WMITN iterations in
timesteps with PWM calculations. However, rates calculated using this procedure
may deviate significantly from the user-specified targets. To prevent the total rates
from deviating substantially from the specified targets, the user can specify
tolerances for production targets at any level(s) of the well management hierarchy.
Two types of rate comparisons are done:
In case of a violation, the user has one of the following three options:
Note that the default option is not to perform rate comparisons during the non-
PWM timesteps. Invoking the rate comparison enables the users to closely
monitor run performance and avoid PWM calculations at every timestep.
39-648 R5000.0.1
Appendix
A
Well Models
A.1 Introduction
In reservoir simulation, the important boundary conditions that drive the flow
occur at the wells. In the exact solution of the differential equations, each well
would be represented by an interior circular boundary of radius rw. One of two
types of boundary conditions could be specified for each well, namely the flowing
well pressure, pwf, or the flow rate, q. It would not be possible, of course, to
specify both.
rw
While such a fine grid might be suitable for single-well models, it is not practical
for 2D areal or 3D models. Instead, each well is embedded in a block whose
horizontal dimensions are much larger than the diameter of the well. As a result,
the pressure calculated for a block containing a well is greatly different from the
flowing well pressure, pwf. To account for this difference, a well index, WI, must
be used. Most of this chapter is devoted to the subject of how to calculate WI for
various 2D and 3D geometries.
where
A similar equation holds for each component. Note that WI is dimensionless, and
is the same for the entire well.
The flowing well pressure is related to the bottomhole pressure at the datum
depth, Dref, by
o
p wf = p bh + γ ( D – D ref ) (A-2)
Equation A-1 can be used two ways. The rate qo can be set in the finite-difference
equation for the block. After the simulator calculates the block pressure, pbl,
Equation A-1 can be used to calculate pwf and then Equation A-2 used to calculate
the bottomhole pressure, p obh . Alternatively, pobh can be set and then pwf calculated.
Equation A-1 can be combined with the finite-difference equation by eliminating
qo. Then, after the simulator calculates the block pressure, the flow rate can be
obtained.
This WI has units of md-ft. It may be different for each vertical block.
q o = PI [ p e – p wf ] (A-4)
or
q o = PI [ p av – p wf ] (A-5)
where pe is pressure at the exterior of the well’s drainage area, and pav is the
average pressure within the drainage area. Thus, the PI is determined not only by
the geometry, but also by the reservoir permeability, the relative permeability, and
the fluid viscosity. Well index, as defined by Equation A-1, has the advantage of
being dependent only on the geometry.
Δx
1 5 Δz
Δy
Δx
2
kρΔyΔz
q = ---------------------- ( p 1 – p wf ) (A-6)
μ ( Δx ⁄ 2 )
( Δx ⁄ 2 -)
( p 1 – p wf ) = qμ
------------------------ (A-7)
kρΔyΔz
The single-phase analogy of Equation A-1, which defines the well index, is
kρΔz
q = WI ------------- ( p 1 – p wf ) (A-8)
μ
Δy -
WI = ----------------- (A-9)
( Δx ⁄ 2 )
The inner radius of the first block, r1/2, may be the same as the wellbore radius, rw
. However, in some cases that may not be convenient, and r1/2 may be
considerably larger, as shown in the figure. In any case, the pressure calculated for
the first block will not coincide with pwf, and it will be necessary to make a
correction for it, through the well index.
kρ Δθ Δz
q = ----------------------------- ( p 1 – p wf ) (A-10)
μ ln ( r 1 ⁄ r w )
Δθ -
WI = ----------------------- (A-11)
ln ( r 1 ⁄ r w )
Note: In VIP, if one has a radial grid and uses an RFLOW card, rw and rb can be
entered, where rb is set equal to r1. But VIP computes
2π -
WI = ----------------------- (A-12)
ln ( r b ⁄ r w )
qμ r
p = p wf + -------------------- ln ----- (A-13)
2πkρΔz r w
As discussed above, we don’t expect p0,0 to be equal to pwf, but how well do the
pressures at the other nodes satisfy Equation A-13? If we plot the difference
between pi,j and p0,0 vs the log of the radial distance from node (i,j) to node (0,0),
then a straight line with the expected slope of qμ/2πkρΔz is obtained34. By
plotting the dimensionless pressure drop:
(pij - p0,0)/(qμ/kρΔz)
against the log of the dimensionless radius ( r ⁄ Δx ) = i2 + j2 , the straight line has a
slope of 1/(2π), as shown in Figure A-5. The pressure at node (1,0) is slightly off
from the straight line, but all the other pressures lie extremely well on the line.
This figure may be considered to be a universal plot, applicable to any square grid
with a single well located far from any boundaries.
qμ r ij
p ij = p 0, 0 + --------------------- ln -------------- (A-14)
2πkρΔZ 0.2Δx
Comparison with Equation A-13 shows that the well block looks like a well with a
wellbore radius of 0.2Δx. We call this radius the equivalent radius of the well
block, rb.
To obtain the well index, we proceed as follows. From Equation A-13 we can
write:
qμ r ij
p ij = p wf + -------------------- ln ----- (A-15)
2πkρΔz r w
2πkρΔz
q = --------------------------------------- ( p 0, 0 – p wf ) (A-16)
μ ln ( 0.2Δx ⁄ r w )
Comparison of Equation A-16 (in which the wellblock pressure is p0,0) with
Equation A-8 (in which the wellblock pressure is p1) gives:
2π
WI = --------------------------------- (A-17)
ln ( 0.2Δx ⁄ r w )
qμ rb
p b – p wf = -------------------- ln ----- (A-18)
2πkρΔz r w
We shall take this as a general definition of rb, for any geometry of the well block.
2π
WI = -----------------------
- (A-19)
ln ( r b ⁄ r w )
We express the well index in this form, since rb in many cases is easy to compute.
Note that Equation A-19 does not take skin into account. We shall show how to
incorporate skin in Section A.5.
------------- Δy
kρΔz ------ ( p – 2p ij + p i – 1, j )
μ Δx i + 1, j
kρΔz Δx
+ ------------- ------ ( p i, j + 1 – 2p ij + p i, j – 1 ) – q ij = 0 (A-20)
μ Δy
For the case shown in Figure A-4, the pressures are symmetric about (0,0), so that
p0,1 = p-1,0 = p0,-1 = p1,0. Also Δx = Δy. Then:
qμ - = 0
4 ( p 1, 0 – p 0, 0 ) – ------------ (A-21)
kρΔz
Assuming that p1,0 satisfies the radial flow Equation A-13 exactly gives:
qμ Δx
p 1, 0 = p wf + -------------------- ln ------ (A-22)
2πkρΔz r w
since r1,0 = Δx. From the definition for rb, Equation A-18 yields:
qμ rb
p 0, 0 = p wf + -------------------- ln ----- (A-23)
2πkρΔz r w
qμ Δx
p 1, 0 – p 0, 0 = -------------------- ln ------
2πkρΔz r b
Δx 2
ln ------ = ---
rb π
or
r –π ⁄ 2
-----b- = e = 0.208 (A-24)
Δx
We shall show later that an exact value for rb for a well far from the boundaries or
from any other well (i.e., an “isolated” well) in a uniform square grid is 0.1985 Δx.
Thus, a good rule of thumb to use for a square well block is rb = 0.2 Δx.
2 2
α = Δy/Δx rb/Δx rb/Δy r b ⁄ Δx + Δy
1/16 0.1406 2.2502 0.140365
Neither rb/Δx nor rb/Δy are particularly constant with α, but it can be seen that rb/
Δx converges to a limit for small Δy, and rb/Δy converges to the same limit for
small Δx. This leads to the possibility that a better quantity to divide rb by is the
length of the diagonal of each gridblock, because:
rb
----------------------------
2 2
Δx + Δy
looks like rb/Δx as Δy goes to zero, and looks like rb/Δy as Δx goes to zero. As can
be seen from the last column of this table, that ratio is very constant with α. So
now we have the more general rule of thumb, that:
2 2
r b = 0.14 Δx + Δy (A-25)
where γ is Euler’s constant, one definition of which is the limit of the difference
between the finite harmonic series and the natural log, i.e.,
⎛ n ⎞
γ = lim ⎜⎜ ∑ 1--- – ln n⎟⎟ = 0.5772157...
n→∞ k
⎝ 1 ⎠
That gives, for the ratio of Equation A-26, the value 0.1403649.
Note that, for a square grid with Δx = Δy, rb/Δx is 2 times this constant, or
0.198506, so that gives us the complete mathematical derivation of the first rule of
thumb, that rb = 0.2Δx for a square grid.
k x ρΔz Δy
--------------- ------ ( p i + 1, j – 2p ij + p i – 1, j )
μ Δx
k y ρΔz Δx
+ --------------- ------ ( p i, j + 1 – 2p ij + p i, j – 1 ) – q ij = 0 (A-27)
μ Δy
k e ρΔz Δy e
--------------- -------- ( p i + 1, j – 2p ij + p i – 1, j )
μ Δx e
k e ρΔz Δx e
+ --------------- -------- ( p i, j + 1 – 2p ij + p i, j – 1 ) – q ij = 0 (A-28)
μ Δy e
where
1⁄2
ke = ( kx ky ) (A-29)
1⁄4
Δx e = ( k y ⁄ k x ) Δx (A-30)
1⁄4
Δy e = ( k x ⁄ k y ) Δy (A-31)
2 2
r be = 0.14 Δx e + Δy e (A-32)
or
1⁄2 2 1⁄2 2
r be = 0.14 ( k y ⁄ k x ) Δx + ( k x ⁄ k y ) Δy (A-33)
In this case rbe is the radius of the (almost) circular isobar in the transformed
problem (i.e., in the xe-ye plane) that has the same pressure as the well block.
However, in the transformed problem, the well bore is elliptical, not circular. It is
shown in Reference 9 that the pressure solution to the exact differential problem
in the xe-ye plane essentially satisfies the equation:
qμ re
p – p wf = ---------------------- ln ------- (A-34)
2πk e ρΔz r we
where
2 2
re = xe + ye (A-35)
and
1 1⁄4 1⁄4
r we = --- r w [ ( k y ⁄ k x ) + ( kx ⁄ ky ) ] (A-36)
2
We see that Equation A-36 is a correction to the wellbore radius that accounts for
the fact that the well bore is elliptical in the xe-ye plane.
qμ r be
p b – p wf = ---------------------- ln ------- (A-37)
2πk e ρΔz r we
qμ rb
p b – p wf = ---------------------- ln ----- (A-38)
2πk e ρΔz r w
r b = r be ( r w ⁄ r we ) (A-39)
Combining Equations A-39, A-33 and A-36 then yields the final result for rb, that
is:
1⁄2 2 1⁄2 2
( k y ⁄ k x ) Δx + ( k x ⁄ k y ) Δy
r b = 0.28 ---------------------------------------------------------------------------------
1⁄4 1⁄4
- (A-40)
( ky ⁄ kx ) + ( kx ⁄ ky )
which can be used in Equation A-19 for the well index.
r be
r b = ----------------------------------------------------------------------- (A-41)
1⁄4 1⁄4
0.5 [ ( k y ⁄ k x ) + ( kx ⁄ ky ) ]
This equation is obtained by combining Equations A-36 and A-39.
In Reference 39, it is shown that rbe is independent of the location of the well
within such an isolated well block, as in the above figure. Then Equation A-32
still holds:
2 2
r be = 0.14 Δx e + Δy e (A-42)
In Reference 39, it is shown that if two wells with the same rate are placed in the
same well block (which is otherwise isolated from any other well block), then:
2 2 2
( 0.14 ) ( Δx e + Δy e )
r be = ----------------------------------------------
- (A-43)
r eAB
where reAB is the distance in the xe-ye plane between the two wells. In terms of the
actual coordinates of the two wells, this distance is given by:
1⁄2 2 1⁄2 2
r eAB = ( ky ⁄ kx ) ( xA – xB ) + ( kx ⁄ ky ) ( yA – yB ) (A-44)
Note that rbe is independent of the actual location of the two wells; the only thing
that matters is the scaled distance between them (provided they are isolated from
other wells).
If the two wells in the same block have different rates, qA and qB, then Equation
A-43 must be modified as follows:
qA + qB
2 2
q [ 0.14 Δx e + Δy e ]
( r be ) AA = ---------------------------------------------------------
- (A-45)
q
( r eAB ) B
Note that qA and qB do not have to be absolute rates; relative rates will do. Also
note that if qA ≠ qB, then the two wells will have different values of rbe.
Similarly, it can be shown for three wells with the same rate (A, B, and C), that the
following equation applies:
3 2 2 3⁄2
( 0.14 ) ( Δx e + Δy e )
( r be ) A = ------------------------------------------------------
- (A-46)
r eAB r eAC
Note that when there are two wells and they have the same rate, they have the
same rbe. But when there are three wells, even with equal rates, in general they do
not have the same rbe (unless reAB = reBC = reAC).
The above results can be generalized to take into account any number of wells, A,
B, C, D, ... through the following equation:
q qk qt
2 2
( r be ) AA ∏ ( reAk ) = [ 0.14 Δx e + Δy e ] (A-47)
k
In Reference 40, it is shown that if two wells with the same rate are placed in
adjacent blocks in the same row, as in the above figure, then:
Δx e – 1 Δy e
r be r eAB = ( 0.14 ) ( Δx e + Δy e ) exp 2 -------- tan ⎛ --------⎞
2 2 2
(A-48)
Δy e ⎝ Δx e⎠
On the other hand, if the two adjacent blocks are in the same column, then Δx and
Δy should be interchanged, to give:
Δy e – 1 Δx e
r be r eAB = ( 0.14 ) ( Δx e + Δy e ) exp 2 -------- tan ⎛ --------⎞
2 2 2
Δx e ⎝ Δy e⎠
Note again the isolation requirement: the pair of wells should be sufficiently far
from any other wells or from the grid boundaries. Also note again that the scaled
distance between the two wells is the important quantity, rather than the actual
location of the two wells.
If r cbe is the value of rbe when the two wells are at the centers of their respective
blocks, and r ceAB is the distance between the block centers (in the xe-ye plane),
then, if the wells are not centered, we can write:
c c
r be = r be ( r eAB ⁄ r eAB ) (A-49)
If the medium is isotropic, the gridblocks are square and the wells are at the
centers of two adjacent blocks, then Equation A1-48 reduces to:
π
r b Δx = ( 0.14 ) ( 2Δx ) exp ⎛⎝ ---⎞⎠
2 2
2
or
r b = 0.190Δx
This is not far from the rule of thumb of 0.2 Δx for a single isolated well in a
square grid. If the two wells are at the centers of two blocks that are not adjacent,
then this rule of thumb is even better. If the wells are not at the centers of two
blocks that are not adjacent, then Equation A-49 can be used to find the correct rb.
But we still require that the pair of wells be isolated from other wells or from the
grid boundary.
If the edge block is at the top or bottom boundary, then we should take:
1⁄4
r eAB = 2 ( k x ⁄ k y ) d
Again we note that if the medium is isotropic, the grid is square, and the well is at
the center of the edge block, then:
r b = 0.190Δx
We have already looked at cases of a single well near the edge of the grid. It is
perhaps less likely that the well will be exactly on the edge of the grid, as shown
in Figure A-7. Equation A-48 is not applicable here. Instead, as Reference 40
shows, the applicable equation is:
Δx e – 1 Δy e
r be = 0.14 Δx e + Δy e exp -------- tan ⎛ --------⎞
2 2
(A-50)
Δy e ⎝ Δx e⎠
Note that rbe is independent of the vertical location of the well on the edge. If the
well is exactly on the top or bottom edge of the grid, then Δxe and Δye should be
interchanged, to give:
Δy e – 1 Δx e
r be = 0.14 Δx e + Δy e exp -------- tan ⎛ --------⎞
2 2
Δx e ⎝ Δy e⎠
π
r b = 0.14 2 Δx exp ⎛⎝ ---⎞⎠ = 0.43Δx
4
This is quite different from the value of 0.2 Δx that one might expect.
Figure A-8 shows the situation where a well is located exactly at the corner of the
grid. This is an important case, as it occurs frequently in five-spot calculations. In
Reference 40, it is shown that:
2 2
r be = 0.14 Δx e + Δy e exp ( E ) (A-51)
where
π Δy 1 Δx Δy – 1 Δy e
E = --- --------e + --- 1 + ⎛ --------e – --------e⎞ tan ⎛ --------⎞ (A-52)
4 Δx e 2 ⎝ Δy Δx ⎠ ⎝ Δx e⎠
e e
Since Equations A-51 and A-52 are rather unwieldy, it may be preferable to use an
empirical equation that fits them quite well:
2 2 0.2520
r be = Δx e + Δy e 0.3816 + ------------------------------------------------------------ (A-53)
Δx
⎛ --------e⎞
0.9401
⎛ Δy
0.9401
+ --------e⎞
⎝ Δy e⎠ ⎝ Δx e⎠
For the special case of isotropic medium and square grid, then Equations A-51
and A-52 become:
r b = 0.14 2 Δx exp ⎛ π
--- + 1---⎞ = 0.72Δx
⎝ 4 2⎠
This value of rb, combined with Equation A-19, should be used for the well index
in five-spot calculations, if kx = ky and Δx = Δy.
Figure A-9 shows a single well A located somewhere in the interior of a corner
block. This situation is perhaps not too likely, but is included here for
completeness. Images of the well and the corner block are shown. In Reference 40
it is derived that:
4 2 2 2
( 0.14 ) ( Δx e + Δy e ) exp ( 4 E )
r be = ----------------------------------------------------------------------
- (A-54)
r eAB r eAC r eAD
For the special case where kx = ky , Δx = Δy, and the well is at the center of the
corner block, then Equation A-54 reduces to
4 2
r b = (-------------------------------------------------------------
0.14 ) ( 2Δx ) exp ( π + 2 )-
Δx ( 2Δx )Δx
or
r b = 0.188Δx
This is not far from the rule of thumb of 0.2 Δx. But it does require that the well be
exactly at the center of the corner block, which is not too likely.
2π
WI = ------------------------
ln ( r b ⁄ r w )
This equation was derived without considering the effect of skin. In this section,
we show how skin may be taken into account.
Figure A-10 depicts radial flow from an outer radius, rd, into the wellbore, which
has radius rw . Let the pressure at rd be pd. The reservoir has permeability k, but an
inner zone of radius ra has an “altered” permeability, ka. Assuming steady state:
qμ ra
p a – p wf = ---------------------- ln -----
2πk a ρΔz r w
qμ rd
p d – p a = -------------------- ln ----
2πkρΔz r a
Adding and rearranging the above equations yields the following equation:
qμ r
p d – p wf = -------------------- ⎛ ln ----d- + s a⎞ (A-55)
2πkρΔz ⎝ r w ⎠
where
k – ka ra
s a = ------------- ln -----
ka rw
In this context, skin may be interpreted as causing additional pressure drop due to
decreased permeability in the altered zone. However, it may be either positive or
negative, depending on whether ka is less than or greater than k. The skin due to
altered permeability is also referred to as “mechanical” skin.
qμ rd
p d – p b = -------------------- ln ---- (A-56)
2πkρΔz r b
But the equation for the well model in terms of the well index, WI, is:
kρΔz
q = WI ------------- ( p b – p wf ) (A-57)
μ
qμ rd qμ
p d – p wf = -------------------- ln ---- + -------------------------
2πkρΔz r b kρΔz ( WI )
rd 2π rd
ln ---- + ------- = ln ----- + s a
rb WI rw
or
2π -
WI = --------------------- (A-58)
rb
ln ----- + s a
rw
The first method uses the graph of Figure A-11, due to Brons and Marting41,
which shows sr as a function of two parameters, hp/ht and hte/rw , where hp is the
length of the interval open to flow, ht is the total thickness of the producing zone,
h te = h t k H ⁄ k V , and kH/kV is the ratio of horizontal to vertical permeability.
Figure A-12 shows three ways in which hp and ht can be interpreted. The first is
simple partial penetration; the second is a centered open interval; the third is with
multiple entries. In the latter two cases, hp and ht apply to a symmetry element.
The second method for calculating sr , due to Odeh42, allows for a single open
interval anywhere in the producing zone. It is an empirical equation derived from
some numerical studies:
h 0.825
s r = 1.35 ⎛ ⎛ -----t – 1⎞ { ln ( h te + 7 ) – [ 0.49 + 0.1 ln ( h te ) ] ln ( r wc ) – 1.95 }⎞ (A-59)
⎝⎝h ⎠ ⎠
p
where
h te = h t k H ⁄ k V
0.2126 ( 2.753 + z m ⁄ h t )
r wc = r w e
and zm is the distance from the top of the sand to the middle of the open interval.
(See Figure A-13.)
The two methods are obviously not the same. However, where they can be
compared (e.g., for zm = hp/2), they give comparable values for sr .
zm
hp
ht
s a = s t – s or (A-60)
Secondly, there is what we might call “local” restricted entry due to partial
completion of the well within a layer of thickness Δzl. The skin due to this local
restricted entry must be added to the mechanical skin. Thus, layer by layer, we
have:
2π
WI l = ---------------------------------
- (A-61)
rb
ln ----- + s a + s lr
rw
For example, consider the 3D situation shown in Figure A-14. The well
completely penetrates the top two layers and partially penetrates the third layer.
The overall skin, st, is determined from a well test. The skin due to overall
restricted entry is determined from hp1 and ht1 and subtracted from st to get sa.
Because total penetration through each of the top two layers is assumed, slr = 0
there. In the third layer, slr is determined from ht2, hp2, and zm2, and added to sa, as
in Equation A-61.
NOTE: On the FPERF card, the variable DHTOP is the distance between
the top of the block and the top of the perforated interval. Thus the
interval calculation for zm is DHTOP + hp/2. Also note that hp is
called H, and ht is called HTOT.
In the PI option, the productivity index is used by VIP to calculate the well index
when the well is first put on production or injection. For an oil producer, for
example:
PI ⋅ GF
WI = ----------------------------------------------------------------- (A-62)
0.001127 ∑ [ khρ o k ro ⁄ μ o ] l
l
GF is a geometric factor that converts the PI from a drainage radius basis to a
gridblock basis. The VIP manual suggests that:
ln ( r e ⁄ r w )
GF = -----------------------
- (A-63)
ln ( r b ⁄ r w )
where
Use t Page43
Now, we want
2π
WI = ----------------------------------
- (A-65)
ln ( r b ⁄ r w ) + s a
2π ( 0.001127 ) ∑ [ khρ o k ro ⁄ μ o ] l
l
PI = ----------------------------------------------------------------------------
- (A-66)
[ ln ( r b ⁄ r w ) + s a ]GF
∑ [ khρo kro ⁄ μo ]l [ ln ( rb ⁄ rw ) + st – t ]
l
GF = -----------------------------------------------------------------------------------------
-
[ khρ ⁄ μ ] eff [ ln ( r b ⁄ r w ) + s a ]
If we assume:
then:
ln ( r e ⁄ r w ) + s t – t
GF = ----------------------------------------- (A-67)
ln ( r b ⁄ r w ) + s a
qg μg r
p = p wf + -------------------------- ln -----
2πkk rg hρ g r w
Differentiation yields
qg μg
dp = -------------------------- d ( ln r )
2πkk rg hρ g
qg ρ
- d ( ln r ) = -----g dp
-------------------
2πkk rg h μg
where
( pb )l
ρ
∫-----g dp
μg
( p wf ) l
( ρ g ⁄ μ g ) l = ------------------------- (A-69)
( p b – p wf ) l
VIP provides three options on the WNDGDV card. In the STD option (or if no
WNDGDV card is used), no attempt is made to adjust the gas properties in the
well model of Equation A-68, and (ρg/μg)l is evaluated at (pb)l. In the PP option,
the integration indicated by Equation A-69 is carried out, using the trapezoidal
rule. PP stands for pseudo pressure, referring to the definition of pseudo pressure
m(p) given by Al-Hussainy, Ramey and Crawford44:
p
ρ
m(p) = 2 ∫ -----g dp
μg
p ref
Finally, the RG option invokes the Russell-Goodrich17 method, which also adjusts
the gas properties in the well model but avoids the integration, by evaluating (ρg/
μg)l at the average of (pb)l and (pwf)l . This option assumes that ρg/μg is a linear
function of pressure. A typical plot of ρg/μg versus p is shown in Figure A-15, and
it can be seen that in both low- and high-pressure regions, the curve is a straight
line. Both PP and RG yield the same results in both regions. In the middle
transition region, however, the RG method will be incorrect, unless the difference
between the gridblock and wellbore pressures is small. The PP method is always
the most accurate, but does involve some modest amount of additional computing
time.
The term D qgl is a rate-dependent skin that accounts for the quadratic term in
Forchheimer’s equation, which is:
∂p μg 2
------ = ---------- v r + βρ g v r (A-71)
∂r kk rg
gl q
where vr is the fluid velocity ---------------------
-.
2πrh l ρ gl
In Equation A-71, the first term on the right-hand side is the Darcy or viscous
component, while the second term is the non-Darcy or inertial component. β is the
coefficient of inertial resistance, and has dimension (length)-1.
– 15 kk rg γ g β
D = 2.223 ×10 ------------------------
2
-
hl rw ( μg )w
10
×10 -
β = 2.73
-----------------------
kk rg
where k is in md.
1⁄2 2 1⁄2 2
( k z ⁄ k x ) Δx + ( k x ⁄ k z ) Δz
r b = 0.28 --------------------------------------------------------------------------------
1⁄4 1⁄4
- (A-72)
( kz ⁄ kx ) + ( kx ⁄ kz )
The assumption that the well is not near any grid boundary may be hard to satisfy
in the simulation of a horizontal well. The question arises: how far does the well
have to be from the top or bottom boundary in order to use Equation A-72?
Δx k 1⁄2 zw
------ ⎛ ----z⎞ ≤ 0.9 ------ (A-73)
Δz ⎝ k x⎠ Δz
where zw is the distance from the well to the nearer of the top or bottom boundary.
Since kz is usually much smaller than kx, this inequality should be easy to satisfy,
if Δx is not too much bigger than Δz.
2 2 2 2 2
l p ( k y k z ) sin θ cos ω + ( k x k z ) sin θ sin ω + ( k x k y ) cos θ
2 2 2 2 2 2
Δy Δz 2 2 Δx Δz 2 2 Δx Δy 2
k ex --------- + -------- sin θ cos ω + k ey --------- + -------- sin θ sin ω + k ez --------- + --------- cos θ
ky kz kx kz kx ky
r b = 0.28 -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
2 2 2
1 - + -------- 1 2 2 1 - + -------- 1 2 2 1 - + -------- 1 - cos 2θ
k ex -------- sin θ cos ω + k ey -------- sin θ sin ω + k ez --------
ky kz kx kz kx ky
(A-74)
where:
1---
(A-75)
2
k ex = ( k y k z )
1---
2 (A-76)
k ey = ( k x k z )
1---
2 (A-77)
k ez = ( k x k y )
For a vertical well, θ = 0 and Equation A-74 reduces to Equation A-40. For a
horizontal well parallel to the y-axis, ω = 90° and Equation A-74 reduces to
Equation A-72.
B
Corner-Point Geometry
Let
x = x 1 + ( x 2 – x 1 )u + ( x 4 – x 1 )v + ( x 1 + x 3 – x 2 – x 4 )uv (B-1)
y = y 1 + ( y 2 – y 1 )u + ( y 4 – y 1 )v + ( y 1 + y 3 – y 2 – y 4 )uv (B-2)
dx dy = J(u,v)du dv
where J(u,v) is the so-called Jacobian of the mapping of the unit square back onto
the quadrilateral. It may be considered to be a scaling, or stretching, factor of the
mapping. It is given by the determinant of a certain matrix of partial derivatives,
as follows:
∂x- ∂x
----- -----
∂ ( x, y ) ∂u ∂v
J ( u, v ) = ---------------- = (B-3)
∂ ( u, v ) ∂y- ∂y
----- -----
∂u ∂v
∂x- = ( x – x ) + ( x + x – x – x )v
----- 2 1 1 3 2 4
∂u
∂x
----- = ( x 4 – x 1 ) + ( x 1 + x 3 – x 2 – x 4 )u
∂v
∂y- = ( y – y ) + ( y + y – y – y )v
----- 2 1 1 3 2 4
∂u
∂y
----- = ( y 4 – y 1 ) + ( y 1 + y 3 – y 2 – y 4 )u
∂v
Thus, the area of the quadrilateral may be obtained by integrating the Jacobian
over the unit square:
1 1
A = ∫ ∫ J ( u, v )du dv (B-4)
u=0 v=0
x = p 1, x u + p 2, x v + p 3, x w + p 4, x uv + p 5, x vw + p 6, x uw + p 7, x uvw + p 8, x
(B-7)
y = p 1, y u + p 2, y v + p 3, y w + p 4, y uv + p 5, y vw + p 6, y uw + p 7, y uvw + p 8, y
(B-8)
z = p 1, z u + p 2, z v + p 3, z w + p 4, z uv + p 5, z vw + p 6, z uw + p 7, z uvw + p 8, z
(B-9)
where
p 1, x = x 2 – x 1 p 2, x = x 4 – x 1 p 3, x = x 5 – x 1
p 4, x = x 1 + x 3 – x 2 – x 4 p 5, x = x 1 + x 8 – x 4 – x 5 p 6, x = x 1 + x 6 – x 2 – x 5
p 7, x = x 2 + x 4 + x 5 + x 7 – x 1 – x 3 – x 6 – x 8 p 8, x = x 1
This works similarly for p1,y , ..., p8,y in terms of the y‘s, and p1,z , ..., p8,z in terms
of the z‘s. It can be seen that u = 0 corresponds to face 1-4-8-5, u = 1 to face 2-3-
7-6, v = 0 to face 1-2-6-5, v = 1 to face 4-3-7-8, etc.
dx dy dz = J(u,v,w) du dv dw
∂x-
----- ∂x
----- ∂x-
------
∂u ∂v ∂w
∂ ( x, y, z ) ∂y- ∂y ∂y-
J ( u, v, w ) = ----------------------- = ----- ----- ------ (B-10)
∂ ( u, v, w ) ∂u ∂v ∂w
∂z- ∂z ∂z-
----- ----- ------
∂u ∂v ∂w
∂x
------ = p 1, x + p 4, x v + p 6, x w + p 7, x vw (B-11)
∂u
∂x
----- = p 2, x + p 4, x u + p 5, x w + p 7, x uw (B-12)
∂v
∂x- = p + p v + p u + p uv
------ 3, x 5, x 6, x 7, x (B-13)
∂w
V = ∫ ∫ ∫ J ( u, v, w )du dv dw (B-14)
u=0 v=0 w=0
Finally, let us consider how we might obtain the “average” thickness of the block
in the vertical direction.
In Figure B-3, consider the elemental tube (shown by dashed lines) bounded by u,
u+du, v, and v+dv. The thickness of the tube, i.e., its vertical extent, is z(u,v,1) -
z(u,v,0).
∫ J ( u, v, w )dw du dv
w=0
(B-18)
Now we consider how we might obtain the average distance from the left face to
the right face of the block. In Figure B-4, consider the element tube bounded by v,
v+dv, w, and w+dw. The length of that tube from the left face to the right face is
2 2 2
{ [ x ( 1, v, w ) – x ( 0, v, w ) ] + [ y ( 1, v, w ) – y ( 0, v, w ) ] + [ z ( 1, v, w ) – z ( 0, v, 2 ) ] } (B-19)
Then the expression in Equation B-19 for the length of the tube can be written
D [ ( 1, v, w ), ( 0, v, w ) ]
Thus the volume averaged distance from the left face to the right face is
1 1 1
1
DX = ---
V ∫ ∫ D [ ( 1, v, w ), ( 0, v, w ) ] ∫ J ( u, v, w )dw dudv
v=0 w=0 w=0
Similarly,
1 1 1
1
DX = ---
V ∫ ∫ ∫ D [ ( u, 1, w ), ( u, 0, w ) ]J ( u, v, w )du dv dw (B-21)
u=0 v=0 w=0
∫ f ( x )dx ≈ ∑ wk f ( xk ) (B-22)
x=0 k=1
for properly chosen values of wk and xk. The wk are weights and the xk are Gauss
points, or quadrature points, at which the function f(x) is evaluated. For N=1, 2, or
3, the weights and Gauss points are given by:
N = 1: w 1 = 1 x 1 = 1---
2
5 1 1
N = 3: w 1 = w 3 = ------, w 2 = 4--- x 1 = --- ⎛ 1 – 3--- ⎞ , x 2 = 1---, x 3 = --- ⎛ 1 + 3--- ⎞
18 9 2⎝ 5 ⎠ 2 2⎝ 5 ⎠
The triple integrals of the previous section may be approximated by the triple
summations:
1 1 1 Ni Nj Nk
∫ ∫ ∫ f ( u, v, w )du dv dw ≈ ∑ ∑ ∑ wi wj wk f ( u i, v j, w k ) (B-23)
u=0 v=0 w=0 i=1 j=1 k=1
where Ni, Nj, and Nk are, respectively, the number of quadrature points in the x-, y-
, and z-directions. These numbers are entered on the CORNER card as iquads,
jquads, and kquads.
It turns out that the integrands within all the triple integrals of the preceding
section for volumetric calculations are composite polynomials of low enough
degree that iquads = jquads = kquads = 2 provides exact integration. This is true
no matter how arbitrarily the corner points are located.
However, the story is different for evaluating transmissibilities. We shall see later
that if the harmonic integration option is chosen, the integrands are no longer
polynomials. In that case, the user should specify iquads = jquads = kquads = 3 to
get the most accurate integration.
Consider the two rectangular blocks in Figure B-5, where the points A and B are
the centroids of their respective blocks. Each block has its own permeability, kA or
kB. As this may be considered to be flow through two resistances in series, the
transmissibility from A to B is given by:
1 -
T AB = ------------------ (B-24)
1- + -----
----- 1-
TA TB
where
T A = k A Δz W ⁄ L A
and
T B = k B Δz W ⁄ L B
In this simple 2D case, Δz, which is perpendicular to the plane of Figure B-5, is
considered constant. Note that TA may be considered to be the “half-block
transmissibility” from the point A to the right face of block A.
LAB
In Figure B-6, we consider the somewhat more general case where the gridblocks
are not orthogonal, but the layer boundaries are parallel. Equation B-24 still holds,
with TA again being given by:
T A = k A Δz W ⁄ L A
We note that LA is now specifically the distance from the centroid A to the point
C, where C is the center of the interface between blocks A and B. If LAB is the
length of that interface, and if γ is the angle between the interface and the parallel
sides, then clearly:
W = L AB sin γ
Alternatively, let n be the unit vector normal to the interface at point C, and let ψ
be the angle between line AC and the normal n. As γ and ψ are complementary
angles, then:
W = L AB cos ψ
and
T A = k A Δz L AB cos ψ ⁄ L A (B-25)
We shall see that it will be also useful to express this as the following resistance
from the point A to the right face:
LA
1- = ------------------------------------
----- - (B-26)
TA k A Δz L AB cos ψ
In Figure B-7, we show the more general case of a single gridblock in which none
of the sides are parallel. Point A is still the centroid, and again we wish to
calculate the half-block transmissibility from point A to the right face. For this
purpose, recognize that, in mapping the gridblock to a unit square, the variable u
represents a fractional distance from the left face to the right face. Let point 5 be a
point moving along the upper edge, such that
x 5 ( u ) = x 1 + u ( x 2 – x 1 ), y 5 ( u ) = y1 + u ( y2 – y 1 )
and let point 6 be a similar point moving along the lower edge, such that:
x 6 ( u ) = x 3 + u ( x 4 – x 3 ), y 6 ( u ) = y3 + u ( y4 – y 3 )
Then the line L56(u), connecting points 5 and 6, is a moving line that slices the
block as shown in Figure B-8.
We may conceptualize the half-block transmissibility from the slice going through
the centroid to the right face as being the sum of transmissibilities of a collection
of tubes formed by lines of constant v, as shown in Figure B-9.
By analogy with Equation B-26, the resistance of that typical chunk is given by:
Δs
--------------------------------------------------
-
k A Δz ΔL cos ψ ( u, v )
where Δs is the distance along the tube lying within the slice, ΔL is the distance
along the slice within the chunk, n(u,v) is the normal to the slice, and ψ(u,v) is the
angle between the tube and that normal. But the area of the chunk is given by ΔA
= Δz ΔL, so the resistance of the typical chunk can be expressed somewhat more
succinctly as:
Δs
------------------------------------------
k A ΔA cos ψ ( u, v )
In the limit, then, the resistance of the tube from the slice going through the
centroid to the right face is given by the integral:
ds-
u=1 ----- ( u, v )
1- = du
-----
TA ∫ ------------------------------------------
k A ΔA cos ψ ( u, v )
du
u=u c
Finally, the transmissibility of all the tubes in parallel is given, in the limit, by:
1
1
TA = ∫ --------------------------------------------------------------
u=1 ds
- dv (B-27)
v=0
------ ( u, v )
du
∫ dA -----------------------------------------
- du
u = u c k A dv cos ψ ( u, v )
-------
Consider the gridblock with nonparallel sides that is shown in Figure B-11. We
wish to calculate the half-block transmissibility between the centroid and the right
face of that block. Again, point A is the centroid; point C is the center of the right
face. The right face is the interface between block A and block B (which is not
shown). Let AC be the vector from A to C. Let CD be a vector normal to the right
edge, whose length is equal to the length of the right edge, that is:
CD = L AB
Now, while Equation B-25 was derived for a gridblock with parallel sides, as in
Figure B-6, it can also be applied to this more general situation. Thus we can write
L AB cos ψ L A L AB cos ψ
T A = k A Δz ---------------------- = k A Δz ----------------------------
-
LA (L )
2
A
AC CD cos ψ
T A = k A Δz -----------------------------------
2
-
AC
Now, the scalar (also known as inner or dot) product of two vectors a and b
having angle ψ between them can be expressed in the following two ways:
a ⋅ b = a b cos ψ
a ⋅ b = ax bx + ay by
where ax and ay are the x- and y-components of the vector a, and bx and by are the
x- and y-components of the vector b, respectively. Thus we see that the half-block
transmissibility can also be expressed as:
( xC – xA ) ( xC – xD ) + ( yC – yA ) ( yC – yD )
T A = k A Δz ------------------------------------------------------------------------------------------------
2 2
-
( xC – xA ) + ( yC – yA )
( xC – xD ) = ( yE – yF ) = Lx
( yC – yD ) = ( xF – xE ) = Ly
Then
Lx ( xC – xA ) + Ly ( yC – yA )
T A = k A Δz --------------------------------------------------------------
2 2
- (B-28)
( xC – xA ) + ( yC – yA )
In Figure B-2 on page B-685, let Equations B-7, B-8, and B-9 be the functions
that map the coordinates of real space (x,y,z) into the coordinates of the unit cube
(u,v,w). Tubes are formed by surfaces of constant v intersecting surfaces of
constant w. On the other hand, any surface of constant u over the range
0≤v≤1 , 0≤w≤1
defines a slice of the gridblock (in real space) that corresponds to the slice shown
in Figure B-8. In general, the slice will not be a planar surface but, rather, a
bilinear surface.
dA = D ( u, v, w )dv dw (B-30)
where
2 2 2
D ( u, v , w ) = J xy + J yz + J zx (B-31)
with
∂x ∂y ∂y ∂z ∂z ∂x
----- ----- ----- ----- ----- -----
J xy = ∂v ∂v ; J yz = ∂v ∂v ; J zx = ∂v ∂v (B-32)
∂x- ∂y- ∂y- ∂z- ∂z- ∂x-
------ ------ ------ ------ ------ ------
∂w ∂w ∂w ∂w ∂w ∂w
The partial derivatives are obtained from Equations B-7, B-8, and B-9 by:
∂x
----- = p 2, x + p 4, x u + p 5, x w + p 7, x uw
∂v
∂x- = p + p v + p u + p uv
------ 3, x 5, x 6, x 7, x
∂w
x c = x ( u c, v c, w c )
y c = y ( u c, v c, w c )
z c = z ( u c, v c, w c )
where x(u,v,w), y(u,v,w), and z(u,v,w) are the polynomials defined in Equations B-
7, B-8, and B-9.
Along each tube of constant v and w, the derivative of arc length is given by:
2 2 2
ds- ⎛ dx ⎞ ⎛ dy ⎞ ⎛ dz ⎞
----- ( u, v, w ) = ⎜ --------⎟ + ⎜ --------⎟ + ⎜ --------⎟ (B-33)
du ⎝ du ⎠ ⎝ du ⎠ ⎝ du ⎠
The partial derivatives in Equation B-33 are given above in Equation B-11.
The only quantity in the integrand of Equation B-29 now left to be evaluated is the
cosine of ψ, the angle between the tube and the normal to the slice. Let p be a
vector along the tube, and n be a vector normal to the slice. Then, from the two
ways of expressing the scalar (or dot) product of the two vectors in 3D:
p ⋅ n = p n cos ψ
p ⋅ n = px nx + py ny + pz nz
we have
px nx + py ny + pz nz
cos ψ = -------------------------------------------
-
p n
or
px nx + py ny + pz nz
cos ψ = -----------------------------------------------------------------------
- (B-34)
2 2 2 2 2 2
( px + py + pz ) ( nx + ny + nz )
The components of p are clearly equal to the derivatives of x, y, and z with respect
to u. That is:
∂x- ; p = -----
∂y- ; p = -----
∂z-
p x = ----- y z
∂u ∂u ∂u
The normal to the slice is obtained by first finding the equation for the plane in
real space tangent to the slice where the tube intersects it. To get this equation, we
find two lines (or vectors) in the slice through the point u, v, w (in the unit cube)
that correspond to the intersection point x, y, z (in real space). See Figure B-12.
u=constant
Let line a be at constant u and v; let line b be at constant u and w. Then the
components of a and b are:
∂x ∂y ∂z-
a x = ------- ; a y = ------- ; a z = ------
∂w ∂w ∂w
∂x ∂y ∂z
b x = ----- ; b y = ----- ; b z = -----
∂v ∂v ∂v
The equation of a plane through the arbitrary point x1, y1, z1, parallel to a and b
is48:
x – x1 y – y1 z – z1
ax ay az = 0
bx by bz
Ax + By + Cz = D
then A is the cofactor of (x-x1), B is the cofactor of (y-y1), and C is the cofactor of
(z-z1). Thus:
ay az ∂y ∂z ∂z ∂y
A = = ------- ----- – ------- -----
by bz ∂w ∂v ∂w ∂v
ax az ∂z ∂x ∂x ∂z
B = – = ------- ----- – ------- -----
bx bz ∂w ∂v ∂w ∂v
ax ay ∂x ∂y ∂y ∂x
C = = ------- ----- – ------- -----
bx by ∂w ∂v ∂w ∂v
Note that A, B, and C are equal, respectively, to -Jyz, -Jzx, and -Jxy of Equation B-
32. But the quantities A, B, and C are also equal to the components of the normal:
nx = A ; ny = B ; nz = C
All the above discussion of HARINT has dealt with the calculation of half-block
transmissibility from the slice through the centroid to the right face. To get the
half-block transmissibility from the slice through the centroid to the left face, it
suffices to change the limits of integration In Equation B-35 as follows:
1 1
1
TA = ∫ ∫ ------------------------------------------------------------------------------------ dv dw
u = uc ds
(B-36)
v=0 w=0
------ ( u, v, w )
du
∫ ----------------------------------------------------------------
k A D ( u, v, w ) cos ψ ( u, v, w )
- du
u=0
To get half-block transmissibilities from the slice through the centroid to the front
and back faces, it suffices to rotate the gridblock about the z-axis, and use the
program that calculates the integrals of Equations B-35 and B-36. Finally, to get
half-block transmissibilities from the slice through the centroid to the top and
bottom faces, it suffices to rotate the gridblock about the y-axis, and again use that
same program.
Ax ( xC – xA ) + Ay ( yC – yA ) + Az ( zC – zA )
T A = k A -------------------------------------------------------------------------------------------------
2 2 2
(B-37)
( xC – xA ) + ( yC – yA ) + ( zC – zA )
Figure B-13: Projections of Right Face onto the Three Coordinate Planes
While Equation B-37 was described as the equation for the half-block
transmissibility from the centroid to the right face, actually it applies to all six
faces. Point C is now interpreted as the center of the relevant face, while Ax, Ay ,
and Az are the areas of the projections of that face. The sign of TA may turn out to
be negative, so its absolute value is used.
Unfaulted Case
As pointed out in the previous sections, the half-block transmissibility from the
centroid to all six faces is computed for each gridblock. These must be combined
to yield the full transmissibility between adjacent gridblocks. For example, if
gridblock B is to the right of gridblock A, then the transmissibility between their
centroids is:
1 -
TX AB = ------------------ (B-38)
1- + ----- 1-
-----
TA TB
Faulted Case
In the case of the FAULT option, the adjacent blocks may share only part of the
common face, as illustrated in Figure B-14. In this figure, AA is the total area of
the right face of block A, and is used in the calculation of TA. Similarly, AB is the
total area of the left face of block B, and is used in the calculation of TB. The area
of their mutual intersection is AAB. Then Equation B-38 is modified as follows:
A AB
TX AB = -------------------
- (B-39)
AA AB
------ + ------
TA TB
x ij = i cos [ ( j – 1 )Δθ ]
y ij = 3 + i sin [ ( j – 1 )Δθ ]
Figure B-15 shows this model for the particular case of Δθ = 45°.
It can be shown that if the grid were truly radial, rather than approximated by this
corner-point geometry, then the angular transmissibility between Blocks 1 and 3
in Figure B-15 would be:
Δz ln ( 2 ⁄ 1 )-
Tθ 1 – 3 = ---------------------------
Δθ
Δz ln ( 3 ⁄ 2 )-
Tθ 2 – 4 = ---------------------------
Δθ
Δθ Δz Δθ Δz
TR = ------------------------------------------------------------
- = ----------------------
-
3 – 23
3---------------- 2 3 – 13
- – ln ----------------- 19 ⁄ 5
ln ln -------------
33 – 22 22 – 12 7⁄3
Results for the two options are shown in Figures B-16, B-17, and B-18. In
Figure B-16, the angular transmissibility between Blocks 1 and 3, multiplied by
Δθ/Δz, is plotted against Δθ,and we see that HARINT converges to the correct
value for small Δθ, and is only a few percent low for Δθ as large as 20-30 degrees.
NEWTRAN is considerably poorer, and converges to an incorrect value. Similar
results are shown in Figure B-17 for the angular transmissibility between Blocks 3
and 4.
C
References
1. Van Everdingen, A. F. and Hurst, W., “The Application of the Laplace
Transformation to Flow Problems in Reservoirs,” Trans. AIME, 186, 305.
2. Carter, R. D. and Tracy, G. W., “An Improved Method for Calculating Water
Influx,” Trans. AIME, 219, 415-417.
10. Muskat, M., The Flow of Homogeneous Fluids Through Porous Media,
McGraw-Hill Book Co., 1937.
13. Land, C. S., “Calculation of Imbibition Relative Permeability for Two- and
Three-Phase Flow From Rock Properties,” Soc. Pet. Eng. J., Trans. AIME,
243,149-156, June 1968.
14. Aziz, A. and Settari, A., Petroleum Reservoir Simulation, Elsevier Applied
Science Publishers (1979).
16. Chase, C. A., Jr., and Todd, M. R., “Numerical Simulation of CO2 Flood
Performance,” Soc. Pet. Eng. J., 597, Dec. 1984.
17. Russell, D.G., Goodrich, J.H., Perry, G.E., and Bruskotter, J.F., “Methods of
Predicting Gas Well Performance,” J. Pet. Tech., 99-108, Jan. 1966.
18. Al-Hussainy, R. and Ramey, H.J., Jr., “Application of Real Gas Flow Theory
to Well Testing and Deliverability Forecasting,” J. Pet. Tech., 624-636, May
1966.
19. Conte, S.D. and Boor, C., Elementary Numerical Analysis: An Algorithmic
Approach, Second Edition, 315-318, McGraw-Hill Book Co, New York City
(1972).
20. Swift, G.W. and Kiel, O.G., “The Prediction of Gas Well Performance
Including the Effect of Non-Darcy Flow,” J. Pet. Tech., 791-798, July 1962.
21. Katz, D.L., et al., Handbook of Natural Gas Engineering, p. 50, McGraw-Hill
Book Co., New York City (1959).
22. Wallis, J.R., Kendall, R.P., and Little, T.E., “Constrained Residual
Acceleration of Conjugate Residual Methods,” SPE Paper 13536 presented at
the Eighth SPE Symposium on Reservoir Simulation, Dallas, TX (1985).
26. Litvak, M., “New Procedure for Wellbore Flash Calculations,” SPE Paper
21216 presented at the 11th SPE Symposium on Reservoir Simulation,
Anaheim, CA, Feb. 1991.
27. More, J.J. and Sorensen, D.C., “On the Use of Directions of Negative
Curvature in a Modified Newton Method,” Mathematical Programming, 16,
pp. 1-20 (1979).
28. Lohrenz, J., Bray, B.G., and Clark, C.R., “Calculating Viscosities of Reservoir
Fluids from their Compositions,” J. Pet. Tech., pp. 1171-1176 (Oct. 1964).
29. Pedersen, K.S., Fredenslund, A., and Christensen, P.L., “Viscosity of Crude
Oils,” Chem. Eng. Sci., 39, pp. 1011-1016 (1984).
30. Hanley, H.J.M., McCarty, R.D., and Haynes, W.M., Cryogenics, 15, p. 413
(1975).
32. Fayers, F.J. and Matthews, J.D., “Evaluation of Normalized Stone’s Methods
for Estimating Three-Phase Relative Permeabilities,” Soc. Pet. Eng. J., pp.
225- 232 (Apr. 1984)
33. Coats, K.H and Modine, A.D., “A Consistent Method for Calculating
Transmissibilities in Nine-Point Difference Equations,” paper SPE 12248
presented at the 1983 Reservoir Simulation Symposium, San Francisco, CA,
Nov. 15-18, 1983.
36. Modine, A.D., Coats, K.H., and Wells, M.W., “A Superposition Method for
Representing Wellbore Crossflow in Reservoir Simulation,” SPE paper 20746
presented at the 65th Annual Technical Conference and Exhibition of the
Society of Petroleum Engineers, New Orleans, LA, Sep. 23-26, 1990.
37. Pedersen, K.S. and Fredenslund, A., “An Improved Corresponding States
Model for the Prediction of Oil and Gas Viscosities and Thermal
Conductivities,” Chem. Eng. Sci., 42, pp. 182-187 (1987).
38. Nghiem, L., Collins, D.A., and Sharma, R., “Seventh SPE Comparative
Solution Project: Modelling of Horizontal Wells in Reservoir Simulation,”
Paper SPE 21221 presented at the 11th SPE Symposium on Reservoir
Simulation held in Anaheim, CA, Feb. 17-20, 1991.
41. Brons, F. and Marting, V.E., “The Effect of Restricted Fluid Entry on Well
Productivity,” J. Petr. Tech., pp. 172-174 (Feb. 1961); Trans. AIME, vol. 222
42. Odeh, A.S., “An Equation for Calculating Skin Factor Due to Restricted
Entry,” J. Petr. Tech., pp. 964-965 (June 1980).
44. Al-Hussainy, R., Ramey, H.J., Jr., and Crawford, P.B., “The Flow of Real
Gases Through Porous Media” J. Petr. Tech., pp. 624-636 (May 1966); Trans.
AIME, vol. 237.
46. Babu, D.K., Odeh, A.S., Al-Khalifa, A.J., and McCann, R.C., “The Relation
Between Wellblock and Wellbore Pressures in Numerical Simulation of
Horizontal Wells,” SPE Res. Eng., pp. 324-328 (Aug. 1991).
47. Franklin, P., A Treatise on Advanced Calculus, John Wiley & Sons, New
York, 1940, Page 375, Equation 110.
51. Brill, J.P. and Beggs, H.D., Two-Phase Flow in Pipes, University of Tulsa,
Sixth Edition, January, 1991.
52. Litvak, M.L., “New Procedure for the Phase-Equilibrium Computations in the
Computational Reservoir Simulator,” SPE Paper 25252, Twelfth SPE
Symposium on Reservoir Simulation, New Orleans, Louisiana, February 28 -
March 3, 1993.
54. Firoozabaid, A. and Thomas, L.K.: “Sixth SPE Comparative Solution Project:
A Comparison of Dual-Porosity Simulators”, Paper SPE 18741 presented at
the 1989 Symposium on Reservoir Simulation, Houston, Texas, Feb. 6-8.
55. Chorin, A.J.: “Numerical Study of Slightly Viscous Flow,” Fluid Mech.
(1973), v. 57 part 4 785-796.
56. Araktingi, U.G., Orr, F.M. Jr.: “Viscous Fingering Gravity Segregation, and
Reservoir Heterogeneity in Miscible Dieplacement in Vertical Cross
Sections,” SPE 20176 Presented at the Seventh Symposium in Enhanced Oil
Recovery, Tulsa, Oklahoma, April 22-25, 1990.
57. Garder, A.O. Jr., Peaceman, D.W., Pozzi, A.L.: “Numerical Calculation of
Multidimensional Miscible Displacement by the Method of Characteristics,
“SPEJ (March 1964) 26-36.
62. Angel, E. and Bellman, R.: Dynamic Programming and Partial Differential
Equations, Academic Press, New York (1972), p. 44.
63. Peaceman, D.W.: “Final Progress Report on IMPES Stability,” Memo to J.S.
Nolen and Associates, Aug. 17, 1988.
65. Meter, D.M. and Bird, R.B.: “Tube Flow of Non-Newtonian Polymer
Solutions: Part I Laminar Flow and Rheological Models,” AICHE Journal,
Vol. 10, No. 6, Nov. 1964, p. 1143-1150.
67. Hirasaki, G.J.: “Ion Exchange with Clays in the Presence of Surfactant,” Soc.
Pet. Eng. J., April 1982, p. 181-192.
68. Wallis, J.R. and Nolen, J.S.: “Efficient Iterative Linear Solution of Locally
Refined Grids using Algebraic Multilevel Approximate Factorizations”, SPE
25239, presented at the 12th SPE Symposium on Reservoir Simulation, New
Orleans, February 1993.
69. Litvak, M.L. and Darlow, B.L.: “Surface Network and Well Tubinghead
Pressure Constraints in Compositional Simulator”, SPE 29125, presented at
the 13th SPE Symposium on Reservoir Simulation, San Antonio, Texas,
February 1995.
71. Katz, D.L., Cornell, D., Kobayashi, R., Poettmann, F.H., Vary, J.A., Elenbaas,
J.R., and Weinaug, C.F.: Handbook of Naturaal Gas Engineering, McGraw-
Hill Book Co., Inc., New York (1959).
72. Cornell, D. and Katz, D.L.: “Flow of Gases Through Porous Media,” Ind.
Eng. Chem. (1953) 45, 2145.
73. Coats, K.H., Thomas, L.K., and Pierson, R.G., "Compositional and Black Oil
Reservoir Simulation," SPE 29111, paper presented at the 13th SPE
Symposium on Reservoir Simulation held at San Antonio, TX, Feb. 12-15,
1996.
74. Killough, John E., Foster, John A., Nolen, James S., Wallis, John R., and
Xiao, Jason, “A General-Purpose Parallel Reservoir Simulator”, presented at
the 5th European Conference on the Mathematics of Oil Recovery, Leoben,
Austria, 3-6 Sept., 1996.
75. van Daalen, D. T., Hoogerbrugge, P. J., Meijerink, J. A., and Zeestraten, P. J.
A., "The Parallelization of BOSIM, Shell's Black/ Volatile Oil Reservoir
Simulator", Proceedings of the First IMA/SPE European Conference on the
Mathematics of Oil Recovery, Oxford University Press, 1990.
78. Rutledge, J.M., Jones, D. R., Chen, W. H., Chung, E. Y.,"The Use of a
Massively Parallel SIMD Computer for Reservoir Simulation", SPE 21213
presented at the eleventh SPE Symposium on Reservoir Simulation,
Anaheim, 1991.
80. Nolen, J. S., and Stanat, P. L., "Reservoir Simulation on Vector Processing
Computers", SPE 9644 presented at the SPE Middle East Oil Technical
Conference, Manama, Bahrain, March, 1981.
81. Wallis, J. R., Foster, J. A., and Kendall, R.P., “A New Parallel Iterative Linear
Solution Method for Large-Scale Reservoir Simulation”, SPE 21209
presented at the Eleventh SPE Symposium on Reservoir Simulation,
Anaheim, California, February 17-20, 1991.
82. Burrows, Richard, Ponting, Dave, and Wood, Lindsay, “Parallel Simulation
with Nested Factorisation”, presented at the 5th European Conference on the
Mathematics of Oil Recovery, Leoben, Austria, 3-6 Sept., 1996.
83. Killough, J. E., and Wheeler, M. F., "Parallel Iterative Linear Equation
Solvers: An Investigation of Domain Decomposition Algorithms for
Reservoir Simulation", SPE 16021 presented at the 9th SPE Symposium on
Reservoir Simulation, San Antonio, Texas, Feb. 1-4, 1987.
86. Kohav, Gil, and Killough, John E., “An Asynchronous Parallel Linear
Equation Solution Technique”, SPE 29142 presented at the Thirteenth SPE
Symposium on Reservoir Simulation, San Antonio, 12-15 February, 1995.
88. Levitan, M.M., Li, K. - M.G.: “Tracer Flow Simulation and Interpretation of
Tracer Tests,” JETEC Paper. (1991).
92. Collins, D., Nghiem, L., Sharma, R. Li, Yau-kun, and Jha, Kamal: “Field-
scale Simulation of Horizontal Wells,” The Journal of Canadian Petroleum
Technology, January 1992, Volume 31, No. 1.
R5000.0.1 Index-719
VIP-EXECUTIVE® Technical Reference Guide Landmark
Index-720 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
R5000.0.1 Index-721
VIP-EXECUTIVE® Technical Reference Guide Landmark
Index-722 R5000.0.1
Landmark VIP-EXECUTIVE® Technical Reference Guide
R5000.0.1 Index-723