Home  Contact Us  Search  Site Map  Downloads  About Unipass

Mode Flow

Table of Contents

  1. Description

  2. Define the Random Variables

  3. Parametric Files

  4. Defining External Response

  5. External.In

  6. Limit State Function

  7. Failure Probability

  8. PDF/CDF of Limit-State Function

  9. Example of Sensitivity Analysis

  10. References


 

Description:

UNIPASS™ provides a complete generic interface example with full installation option.  By default, all the required files for this example are installed in the following directory:

 C:\Program Files\Unipass\Examples\External\ModeFlow

 The sample files would run as is unless the users have installed the example files in a different directory than the one specified above.  MODFLOW-2000, a public domain software maintained by the U.S. Geological Survey [ 3]  is installed in

 C:\Program Files\Unipass\Examples\External\ModeFlow\bin   

 All the source codes and documentation for MODFLOW-2000 is packed in the following self-extracting file:

 C:\Program Files\Unipass\Examples\External\ModeFlow\Modflow\mf2k1_0.exe  

 MODFLOW is a computer program that numerically solves the three dimensional ground-water flow equation for a porous medium by using a finite –difference method.  A new version of MODFLOW, called MODFLOW-2000, which is designed to accommodate the solution of equation in addition to the ground-water flow equation.

 

Figure 1 1 MODFLOW-2000 illustration depicting common components modeled

Figure 2 Finite difference discretization of MODFLOW example

Consider the under ground water flow problem illustrated in Figure 2.  Constant pressure is specified along the boundary of the reservoir, possible a river or a lake.  Water inflow in the form of rain and denoted as ‘recharge’ permeates the top layer.  Water outflow is divided between the draining ditch, various pumps (not shown), the reservoir, and the two layers of sand aquifers below.  Further details of the under ground water flow model can be found in Reference [ 4].  Listings of all the input files can be found in Reference [ 3 ] starting with page 89.  The random variables used to model the uncertainties of the parameters are listed in Table 1.  The purpose of this analysis is to predict the probability of water that accumulates in the system exceeding a prescribed allowable amount (i.e. a flood situation).

 Table 1 Uncertainties modeled in the MODFLOW-2000 example

Variable Name Description
TRPY1 Defines the anisotropy of the horizontal hydraulic conductivity, or permeability (depending which one was specified), in the top layer.  If TRPY1=1 the layer is isotropic.  In other words, hydraulic conductivity is the same in both the column and the row direction.  TRPY=2 implies that water the hydraulic conductivity, or permeability, in the column direction is twice the value in the row direction.
HY1 Hydraulic conductivity in the row direction.  Multiply this value with TRPY1 to obtain the hydraulic conductivity in the column direction.
VCONT1 Vertical conductivity divided by the thickness between the top layer and the layer below
VCONT2 Vertical conductivity divided by the thickness between the bottom layer and the layer above.
recharge Amount of rainfall (ft/sec).

Back to Top

  Defining A Distribution For Each Parameter

  Open the following sample file with UNIPASS™:

“C:\Program Files\Unipass\Examples\External\ModeFlow\modflow_example2.PAS”

 Open the Variable Definition Window and review the distributions defined for this model.  For instruction on how to modify these distributions please refer to Reference [ 1].  The variable window with all the distributions for this problem are shown in Figure 3.    

 random_var.jpg (82909 bytes)

Figure 3 Distribution defined for MODFLOW-2000 example

Back to Top

Defining The Parametric Files

Open the parametric file editor and load the file ‘twri.bc6’ into the editor as shown in Figure 4.

 twri_bc6.jpg (48440 bytes)

Figure 4  Loading the sample input file ‘twri.bc6’ into the parametric file editor

 Note that all the random variables defined are listed in the top left window.  Select ‘TRPPY1’ from the list and highlight the second numeric field in the third line of the parametric file editor (see Figure 5).

 twri_bc6a.jpg (47915 bytes)

Figure 5 Selecting a variable and an associated field in the parametric file editor

 Next, double click ‘TRPY1’ in the random variable list, the highlighted field in the parametric file editor show be replaced with ‘TRPY1’ enclosed by brackets (see Figure 6).   This creates a template for UNIPASS™ to insert a value into this field every time it prepares the input file ‘twri.bc6’.   Repeat the same procedure for HY1, VCONT1, and VCONT2.  A sample parametric file was saved as ‘backflow.par’.  Do not overwrite this file without verifying the contents.

 twri_bc6b.jpg (47460 bytes)

Figure 6 Creating a template for a single distribution in the parametric file editor

Backflow.par

         0  1.00E+30         0  0.00E+00         0         0  

 1 0 0

CONSTANT   <TRPY1     >  TRPY

CONSTANT   <HY1       >  HY layer   1

CONSTANT   <VCONT1    >  VCONT layer   1

CONSTANT   1.000000E-02  TRAN layer   2

CONSTANT   <VCONT2    >  VCONT layer   2

CONSTANT   2.000000E-02  TRAN layer   3

Recharge.par

         1         0    NRCHOP,IRCHBD

         1              INRECH

         0<recharge>                                  RECH-1

Back to Top 

  Defining External Response

Open the external response window.  All the parameters have already been defined.  Hover the mouse right above the command line and hold, a small window will pop up to show the entire content of this item (see Figure 7).   The batch file ‘twri.bat’  is just a batch file to start MODFLOW-2000.  Click once on the command line icon and the complete command line including the input and the output file name will appear in the tiny status window on the bottom left (see Figure 7).  The parametric file, the input file, and the output file have already been defined as ‘backflow.par’, ‘twri.bc6’ and ‘twri.lst’ respectively.  Right click on any item in the tree to see a menu of all the available operations.  

external_interface.jpg (39188 bytes)

Figure 7 Generic interface definition for MODFLOW-2000

Only one values will be extracted from the output file ‘twri.lst’.  Because the output file format actually changes with the number of iteration MODFLOW used, a fix format interface would fail.  Instead, three keyword search instructions have been specified.  Double click on the first keyword: ‘VOLUMETRIC BUDGET FOR ENTIRE MODEL’.  A window will ask to load the file ‘twri.lst’.  Click yes and the file will be loaded.  Immediately after, the first keyword string will be located and selected in the display to the right.

keywords.jpg (89841 bytes)

Figure 8 Locating the first keyword defined for external(1)

 Double click on all the search commands and verify the correct response value are highlighted.

Back to Top

External.in

twri

backflow.par, 1, twri.bc6, twri.lst

recharge.par, 1, twri.bcf, twri.lst

  Defining Limit State Function

Open the function definition window

limit_state.jpg (69139 bytes)

Figure9 Limit state function definition

 Only one limit state function has been defined using the external function.  The allowable value for this model is 300.  Only one cutest is required for this component problem.  For more details on function definition refer to [ 1 ].  Keep all the default parameters and run the solver.  The result is shown in Figure 10.

solver.jpg (86886 bytes)

Figure 10 Animation of PDF of the limit state function for the MODFLOW-2000 example (at the end of 100 iterations)

Back to Top

Failure Probability

By 100 sampling points in the Monte Carlo Simulation, 

no of g call

no of Sample Points

Pf

ß

COV of Pf

10

10

4.0000000E-01

2.5334718E-01

4.0824829E-01

20

20

6.0000000E-01

-2.5334718E-01

1.8731716E-01

30

30

6.0000000E-01

-2.5334718E-01

1.5161961E-01

40

40

5.5000000E-01

-1.2566139E-01

1.4484136E-01

50

50

5.8000000E-01

-2.0189355E-01

1.2156613E-01

60

60

6.0000000E-01

-2.5334718E-01

1.0629880E-01

70

70

6.2857143E-01

-3.2807221E-01

9.2541341E-02

80

80

6.5000000E-01

-3.8532059E-01

8.2558881E-02

90

90

6.0000000E-01

-2.5334718E-01

8.6548464E-02

100

100

5.8000000E-01

-2.0189355E-01

8.5524994E-02

Back to Top

PDF/CDF of Limit-State Function

pdfcdf.jpg (118663 bytes)

Back to Top

Example of Sensitivity

Normalized Factor:

max|(s)d(Pf)/d(m)|

= 2.6254327E-01

max|(s)d(Pf)/d(s)|

= 5.8000537E-01

Variable

Name

Normalized

(s)d(Pf)/d(m)

Normalized

(s)d(Pf)/d(s)

TRPY1

-1.8409765E-06

-9.9999308E-01

HY1

-1.0000000E+00

4.8325390E-02

VCONT1

7.9263458E-01

1.2216327E-01

VCONT2

2.8356208E-01

1.4070679E-01

recharge

-3.8264491E-06

-1.0000000E+00

Plot of (s)d(Pf)/d(m)

sensmean.jpg (67147 bytes)

Plot of (s)d(Pf)/d(s)

sensstd.jpg (62800 bytes)

Back to Top

References:

 [ 1 ] UNIPASS™Quick Start Guide Version 4.2”, 2002, PredictionProbe, Newport Beach, CA.

 [ 2 ] Mary C. Hill, Edward R. Banta, Arlen W. Harbaugh, and Evan R. Anderman, “MODFLOW-2000, The U.S. Geological Survey Modular Ground-Water Model —User Guide To The Observation, Sensitivity, And Parameter-Estimation Processes And Three Post-Processing Programs”, U.S. Geological Survey, Open-File Report 00-184.

 [ 3 ] Arlen W. Harbaugh, Edward R. Banta, Mary C. Hill, Michael G. McDonald, MODFLOW-2000, “The U.S. Geological Survey Modular Ground-Water Model—User Guide to Modularization Concepts And The Ground-Water Flow Process”, U.S. Geological Survey, Open-File Report 00-92.

 [ 4 ] Harbaugh, A.W., and McDonald, M.G., 1996a, “User's documentation for MODFLOW-96, an update to the U.S. Geological Survey modular finite-difference ground-water flow model: U.S. Geological Survey Open-File Report 96-485”, 56 p.

Back to Top

 

 

Last Updated 02/08/10

Send mail to info@predictionprobe.com with questions or comments about this web site.
Copyright © 2001-2004 PredictionProbe, Inc.