Stochastic Simulations - Big Valley

For this exercise we will revisit the Big Valley model and perform a stochastic simulation.

Do the following:

  1. Download and unzip start.zip if necessary.  Open start.gpr
  2. Change the MODFLOW Run Option in the Global/Basic Package dialog to Stochastic Simulation
  3. Check the settings in the Stochastic Options dialog.
  4. Parameter Randomization
    1. Go to Parameters dialog
      1. Randomize K fields
      2. Change standard deviation for all fields to 0.3 log cycles.
      3. Change to random sampling
      4. 30 instances
    2. Save and run as montecarlo-1
    3. Browse through the folder of solutions. Click on individual solution folders to update contours.
    4. Right-click on folder of solutions and select Statistical Analysis.  View mean head, etc.
    5. Review Latin Hypercube distributions/options in the Parameters dialog.  No need to run another model.
  5. Gaussian Simulation
    1. Create a dummy, empty scatter point set. Rename it to gauss points.
    2. Select the Interpolation|Gaussian Simulation Options command.
      1. Log interpolation
      2. 30 realizations (this number is extremely low - for demo purposes only)
      3. Truncate 0 to 200
      4. Mean = 0.4
      5. Variogram
        • Range = 700-1000
        • Contribution = 0.3
        • Nugget = 0
    3. Select the Run Gaussian Simulation command.
    4. Turn on log contouring
      1. Contouring Options.
      2. Use the Populate Values button)
    5. Run MODFLOW
      1. Change all K polygons to -100 key value
      2. Map -> MODFLOW
      3. Delete all of the HK_XXX parameters except for HK_100
      4. Turn on Use multiplier arrays option.
      5. Change Multiplier option for HK_100 to Mult. Arrays.
      6. Use the Data Set(s) button to select gaussian folder. Note run configuration at bottom.
      7. Save and run as montecarlo-2.gpr.
      8. Reset contour options and browse through folder of solutions.
  6. Probabilistic Capture Zone Analysis
    1. Right-click on one of the stochastic solutions and select Risk Analysis.
      1. Probabilitistic capture zone analysis.
      2. Next.
      3. Stop in cells with weak sinks.
      4. Review placement options. Try four particles per cell.
      5. Weight results based on residual error.
      6. Finish.
    2. Select each of the three data sets generated to view. Each one represents probability of capture for a particular well.
    3. Change contour options
      1. Contour specified range.
      2. Change lower limit to 0.1 and turn off the Fill below option.

Note: There is a permissions problem when running the gaussian simulator on CAEDM. It tries to write a temp scratch file and it can't. To solve the problem, copy the FIELDGEN.EXE file to a local drive and change the path to the file using the Edit|Preferences dialog.