1. Transport.chem.dat File  3 waters –Background, O 2 (solution 1) –Sewage effluent, NH 4 + and Cl - (solution 2) –Rain O 2 (solution 3)  NH 4 + oxidation –SOLUTION 1 and 2 contain dissolved oxygen –SOLUTION 2 only NH 4 + –NO 3 - produced where NH 4 + and O 2 mix  N 2(aq) and NO 2 - are disabled in.chem.dat  Data to be written for Model Viewer 1

2. Transport.chem.dat 2

3. Changes for Solute Transport 1. Save As… 2. Set solute transport and transient flow 3. Check dispersivities 4. Check distribution of chemical conditions 5. Initial head distribution 6. Check associated solutions of BC 7. Specify flux of sewage effluent 8. Set time step and simulation length 3

4. Save As  Save model—File|Save As… transport.p4w 4

5. Enable Solute Transport 5

6. Enable Transient Flow 6

7. Transport Calculations Require Dispersivity Parameters MEDIA Retain default of 1 m longitudinal and transverse dispersivity 7

8. Numerical Dispersion: Upstream-in-space, backward-in-time Grid spacing is 500 m Velocity is 100 m/yr, 1 yr time step 8

9. IC: Set Chemical Initial Conditions  Default CHEMISTRY_IC should have SOLUTION 1 9

10. IC: Use Steady-State Head Condition: flow.head.dat or ex5.head.200.dat 10

11. Flux: Solution 3 for Precipitation 11

12. Flux: Sewage Beds  Draw an arbitrary box  Flux boundary  Flux schedule (next page)  Solution 2  Z face 12

13. Denis LeBlanc’s Flux Estimates (m/d) Beds 1-4Beds 5-8Beds 9-12Beds 13-16Beds 19-25 000000 1936-0.0370000 1941-0.114 1946-0.023 1956-0.126 00 1971-0.05 00 1978-0.1140000 19840-0.038 00 199600000 13

14. Flux: Copy Zone for 4 More Sewage Beds  X 279725 to 279825  Y 821500 to 821600  Z 0 to 20 Flux: Modify Box Zone for Sewage Beds 1-4  Click on zone in BC tree  Ctrl-c, Ctrl-v  Choose “specified flux” 14

15. Flux: Locations for Sewage Beds X minX maxY minY maxZ minZ max Beds 1-4279725279825821500821600020 Beds 5-8279625279725821425821525020 Beds 9-12279525279625821475821575020 Beds 13-16279400279500821575821675020 Beds 19-25279500279600821700821800020 15

16. Flux: Sewage Beds 16

17. Set Print Frequencies for HDF File  Double Click PRINT_FREQUENCY –Select years for HDF chemistry –Print every 2 years –Data to be written are selected in SELECTED_OUTPUT section of transport.chem.dat 17

18. TIME: Set Time Step and Simulation Length  End time 1946  Time step 1 yr  Start time 1936 Double Click TIME_CONTROL 18

19. Running the Model  Save  Run the model 19

20. Visualize Results of Conservative Transport –View P at 1946 –Show|Solid –Show|Color Bar –Toolbox|Geometry—Z 20 –Toolbox|Color Bar  1e-10 min  Logrithmic –Toolbox|Overlay  Model Viewer  File|New— c:\student\CapeCodPhast\transport.h5 20

21. View NH 4 +, NO 3 -, and O 2 –Set to 1946  Toolbox|Crop –X min 0.45  Ctrl-d changes view  Toolbox->Data N(5)  Toolbox->Animation N(5) = NO 3 - species 21

22. NH 4 +, NO 3 -, O 2  NH 4 + from effluent  NO 3 - from oxidation –Note faster flow at top  O 2 consumed 22

23. To Decrease Numerical Dispersion  Refine resolution – Horizontal 250 m – Vertical 2.5 m -20 to 20 m – Time step 0.5 yr 23

24. Refined Grid  Refined resolution – Horizontal 250 m – Vertical 2.5 m -20 to 20 m –0.5 yr time step  Original resolution – Horizontal 500 m – Vertical 2.5 m -20 to 20 m –1 yr time step 24

25.  NH 4 +  NO 3 -  O 2 25

26. Summary  Introduction PHAST for Windows  Must always test numerical accuracy –Numerical dispersion for B-I-T, U-I-S –Oscillations for C-I-T, C-I-S  Cape Cod –Heads look good –Plume direction problematic –Tiny dispersivity is a killer, working on implicit TVD method 26

27. My Model Doesn’t Work: What to do when things go wrong  Start simple! Add complexity sequentially  Check problem definition for completeness and correctness –Properly posed problem that is physically realistic –Shape of region –Spatial property distributions (Model Viewer, prefix.O.probdef) –Boundary condition distributions (Model Viewer, prefix.O.probdef) –Initial conditions (Model Viewer, prefix.O.head and prefix.O.comps) 27

28. My Model Doesn’t Work: What to do when things go wrong  Grid may be too coarse (GRID) –Numerical oscillation –Excessive numerical dispersion  Time step may be too long (TIME_CONTROL) –Numerical oscillation –Excessive numerical dispersion –Excessive operator splitting error 28

29. My Model Doesn’t Work: What to do when things go wrong  Check tolerance and controls on iterative solver (SOLUTION_METHOD)  Check controls on automatic time-step algorithm for steady-state flow (STEADY_FLOW)  Review the simulation history –Time steps (prefix.log) –Maximum changes in head (prefix.log) –Solver iterations (prefix.log) –Global balance tables (prefix.O.bal) 29

30. My Model Doesn’t Work: What to do when things go wrong  Use most robust numerics –Set backwards-in-time and upstream-in- space differencing –Set equal dispersivities in all directions –No cross dispersion  Simplify the problem –Flow only –Conservative transport –Simplify the region and boundary conditions 30

31. My Model Doesn’t Work: What to do when things go wrong  Look for errors in chemistry –Run batch chemistry in PHREEQC –Run 1D transport in PHREEQC  Most problems with convergence in chemistry –Redox –Small total concentration of element (~1e-14)  Lack of chemical convergence may be the result of a poor transport solution (oscillations) 31

32. My Model Doesn’t Work: What to do when things go wrong Contact me dlpark@usgs.gov dlpark@usgs.gov 32

33. The purpose of computing is insight, not numbers. --R.W. Hamming (1962) --R.W. Hamming (1962) The purpose of computing is insight, not pictures. --L.N. Trefethen (1998) --L.N. Trefethen (1998) 33