Enhanced Data-Driven Assessments of Hydraulic Capture Zones

Transcription

1 Enhanced Data-Driven Assessments of Hydraulic Capture Zones David E. Dougherty, Subterranean Research, Inc. David A. Wilson, EPA Region 5, GEOS Federal Remediation Technologies Roundtable Optimization Conference, Dallas, June 2004

2 TAKE-HOME MESSAGE: Union of Physical and Statistical Approaches is Greater than Sum of the Parts

3 Outline of Talk Context Straightforward Extensions More Involved Extensions Examples Concluding Remarks

4 P&T Evaluation (per EPA 2002) Evaluate plume capture Perform and interpret treatment process monitoring Perform and interpret groundwater monitoring Evaluate extraction well performance Evaluate injection well performance

5 Multiple Lines of Evidence Compare actual pumping rates to plan Thermo Chem, Pump and Treat System 160 Design Rate = 150 gpm Initial Flow Rate = 72 gpm Observed Rate, gpm Total System Flow Rate, gpm /01/03 02/01/03 01/01/03 12/01/02 11/01/02 10/01/02 09/01/02 08/01/02 07/01/02 06/01/02 05/01/02 04/01/02 03/01/02 02/01/02 01/01/02 12/01/01 11/01/01 10/01/01 09/01/01 08/01/01 07/01/01 06/01/01 05/01/01 04/01/01 03/01/01 02/01/01 01/01/01 12/01/00 11/01/00 10/01/00 09/01/00 04/01/00 10/01/99 Compare actual heads to plan Compare gradients or fluxes Compare chemistry data for contradictions

6 Previous Talk Tonkin et al. Kriging with Trend (= Universal Kriging) modified to include log terms centered on each pumping or injection well Observed Head Data Kriging with Linear Trend Estimation Gradients Particle Tracking Observed Head Data Kriging with Linear Trend Superposed with Sum of Analytical Thiem Well Solutions* Estimation Gradients Particle Tracking *Thiem well equation: 2D, confined, steady, fully-penetrating, no leakage or storage effects or well losses

7 Extensions Straightforward Analytical forecast model More complicated regional head model Well losses Partial penetration Water table conditions More Involved Numerical forecast simulator Heterogeneous material properties Irregular lithology Variable source/sink contributions

8 Approach to Create Head Map For each monitoring event, have observed locations and observed heads h o (x o ) Two mapping methods: 1. Use a method that creates a head map from the observations G(h o (x o )) 2. A calibrated forecast model (e.g., Thiem or MODFLOW) can create the head map h f

9 Improved Map Blend Use a weighted average of the two maps h a = 1 W G h o W h f Weights may vary from point to point. Rearrange h a =G h o W h f G h o W=0!Observations only, W=1!Forecast only How do we select weights?

10 Ad hoc Selection of Weights Distance from observations Statistical Notice structure of update equation: h a =G h o W h f G h o Base value + (weight X residuals) Kalman Filter allows covariance info Under many assumptions, interpret weights as resulting from kriging the residuals.

11 Process Observe Sample, Analyze, Report, Buddy Check Compare and Combine Model Forecast Another topic Calibrate, Estimate parameters, Reconceptualize Data Assimilation Estimate/Map Business Logic Decide/Interact Implementation/Negotiation

12 More Details Compare and Combine Regression Model Residuals Bias Debiased Residuals Spatial Variability Model (e.g., variogram) Interpolated Debiased Residuals Estimate/Map

13 Examples Using Analytical Models Regression Only Kriging Only March2003: TC-10M,S March2003: Blended TC-10M,S March2003: TC-3S,M Northing, ft GMMW-7A,B,C,D GMMW-6A,B,C,D GMMW-4A,B,C,D 624 GMMW-5A,B,C,D GMMW-3A,B,C, TC-10M,S TC-3S,M TC TC-22 TC-11M,S 610 TC-14M EW-1 PZ-8 PZ-2 PZ-1 EW-6 EW-4PZ-9EW-5 PZ-4 PZ-5PZ-6PZ-7 PZ-3 TC-26 TC-28 TC-29 TC-25 EW-3TC-27 EW TC-30 OU2-MW1M,S,I 6 SG TC-3S,M628 TC-23 TC-13M GMMW-7A,B,C,D GMMW-6A,B,C,D 626 GMMW-4A,B,C,D 8 TC-16S,I TC-20S 614 TC-22 TC-11M,S 612 SG-02 TC-18M,S,I 612 TC-19M,S,I SG SG OU2-MW4M,S,I TC-33 TC-15M,S,I TC GMMW-5A,B,C,D GMMW-3A,B,C, 610 S OU2-MW1M,S,I GMMW-5A,B,C,D GMMW-3A,B,C, OU2-MW2M,S,I TC-17M,S,IR TC-21S,I TC GMMW-7A,B,C,D GMMW-6A,B,C,D GMMW-4A,B,C,D Northing, ft 62 Northing, ft TC-14M EW-1 PZ-8 TC-13M PZ-2 PZ-1 EW-6 EW-4PZ-9EW-5 PZ-4 PZ-5PZ-6PZ-7 PZ-3 TC-26 TC-28 TC-29 TC-25 EW-3TC-27 EW-2 TC-16S,I TC-17M,S,IR TC TC-33 TC-21S,I TC-32 TC-15M,S,I TC-20S SG-01 TC-23 TC E E E+07 OU2-MW2M,S,I SG3 606 Easting, ft 608 OU2-MW4M,S,I SG1 SG-02 TC-18M,S,I 612 S TC-19M,S,I SG TC-14M 60 EW-1 PZ-8 TC-13M 610 PZ-2 PZ-1 8 EW-5 EW-6 EW-4 PZ-9 PZ-4 PZ-5PZ-6PZ-7 PZ-3 TC-26 TC-28 TC-29 TC-25 EW-3TC-27 EW-2 TC-16S,I TC-17M,S,IR TC-33 TC-21S,I TC-32 TC-15M,S,I TC-20S E TC-11M,S E TC SG E TC-30 Easting, ft OU2-MW1M,S,I OU2-MW2M,S,I TC-31 SG3 606 SG1 608 OU2-MW4M,S,I SG-02 TC-18M,S,I S TC-19M,S,I SG SG E E E+07 Easting, ft Dougherty/Wilson Federal Remediation Technologies Roundtable Optimization Conference, Dallas, June 2004

14 Kriging Only Regression Only Blended with estimated in-well heads Blended

15 MW26B MW26W MW25W MW25B MW27W MW27B BP 118 MW15W MW24W MW24B MW15B MW15H MW7B MW7W MW7H BP 19L BP 21 BP 20 BP 2 MW17W MW17B MW16W MW16B MW16H MW8W MW3H MW10B MW3B MW13H MW13B MW13W MW8B MW3W PIEZO#5 PIEZO #6 MW10W MW5W EZ #1 MW5B PIEZO #7 PIEZO #1 MW1H MW1W MW1B MW6W MW11H MW11W MW2W MW11B MW6B MW2B MW14W MW18W MW18B MW14B MW19B MW19W MW9 MW22 MW21B MW21W MW20 Example Using Numerical Models Kriged Head: Kriged Head Data E E E E E+06 Northing 1.078E E MW117 MW25W MW25B MW104 ** MW102 MW101 ** MW119 MW105 MW103 ** ** MW26W MW26B MW118 EW15 EW18 EW13 MW111A MW108 MW112 EW14 MW110 EW16 MW107 MW113 EW12 MW27W MW120 MW27B MW106 ** MW109 MW15W MW15B MW24B MW24W EW9 EW11 MW7B MW7W W. E MW14W EW4 MW14B MW123 MW21B MW21W EW3B MW114MW115 EW1 MW20 MW17W MW17B EW10 MW3B MW3W EW7 MW11W MW11B MW18W MW18B MW2W MW2B MW22 MW121 MW19W MW19B EW5 MW MW23B MW23W MW E+06 MW1B MW1W E E E E E E E E+06 Easting 896.3

16 MW26B MW26W MW25W MW25B MW27W MW27B BP 118 MW15W MW24W MW24B MW15B MW15H MW7B MW7W MW7H BP 19L BP 21 BP 20 BP 2 MW17W MW17B MW16W MW16B MW16H MW8W MW3H MW10B MW3B MW13H MW13B MW13W MW8B MW3W PIEZO#5 PIEZO #6 MW10W MW5W EZ #1 MW5B PIEZO #7 PIEZO #1 MW1H MW1W MW1B MW6W MW11H MW11W MW2W MW11B MW6B MW2B MW14W MW18W MW18B MW14B MW19B MW19W MW9 MW22 MW21B MW21W MW20 Example Using Numerical Models Simulated Head, ft: Modflow Head Data E E E E E+06 Northing 1.078E E MW117 MW25W MW25B MW104 ** MW102 MW101 ** MW119 MW105 MW103 ** ** MW26W MW26B MW118 EW15 EW18 EW13 MW111A MW108 MW112 EW14 MW110 EW16 MW107 MW113 EW12 MW27W MW120 MW27B MW106 ** MW109 MW15W MW15B MW24B MW24W EW9 EW11 MW7B MW7W W. E MW14W EW4 MW14B MW123 MW21B MW21W EW3B MW114MW115 EW1 MW20 MW17W MW17B EW10 MW3B MW3W EW7 MW11W MW11B MW18W MW18B MW2W MW2B MW22 MW121 MW19W MW19B EW5 MW MW23B MW23W MW E+06 MW1B MW1W E E E E E E E E+06 Easting 896.3

17 Variography of Residuals 15 monitoring events filtered out events with too few observations 15 monitoring events summarized with current data

18 MW26B MW26W MW25W MW25B MW27W MW27B BP 118 MW15W MW24W MW24B MW15B MW15H MW7B MW7W MW7H BP 19L BP 21 BP 20 BP 2 MW17W MW17B MW16W MW16B MW16H MW8W MW3H MW10B MW3B MW13H MW13B MW13W MW8B MW3W PIEZO #5 PIEZO #6 MW10W MW5W EZ #1 MW5B PIEZO #7 PIEZO #1 MW1H MW1W MW1B MW6W MW11H MW11W MW2W MW11B MW6B MW2B MW14W MW18W MW18B MW14B MW19B MW19W MW9 MW22 MW21B MW21W MW20 Example Using Numerical Models Updated Head, ft: Blended Head Data E E E E E+06 Northing 1.078E E MW117 MW25W MW25B MW104 ** MW102 MW101 ** MW119 MW105 MW103 ** ** MW26W MW26B MW118 EW15 EW18 EW13 MW111A MW108 MW112 EW14 MW110 EW16 MW107 MW113 EW12 MW27W MW120 MW27B MW106 ** MW109 MW15W MW15B MW24B MW24W EW9 EW11 MW7B MW7W W. E MW14W EW4 MW14B MW123 MW21W MW21B EW3B MW114MW115 EW1 MW20 MW17W MW17B EW10 MW3B MW3W EW7 MW11W MW11B MW18W MW18B MW2W MW2B MW22 MW121 MW19W MW19B EW5 MW MW23B MW23W MW E+06 MW1B MW1W E E E E E E E E+06 Easting 896.3

19 MW26B MW26W MW27W MW27B BP 118 MW15W MW24W MW24B MW15B MW15H MW7B MW7W MW7H BP 19L BP 21 BP 20 BP 2 MW17W MW17B MW8W MW10B MW13H MW13B MW13W MW8B MW3H MW3W MW3B PIEZO #5 PIEZO #6 MW10W MW5W EZ #1 MW5B PIEZO #7 PIEZO #1 MW6W MW11H MW11W MW2W MW11B MW6B MW2B MW18W MW18B MW19B MW19W Example Using Numerical Models Blended Head Data Kriged Conc: MW15W MW15B 1.079E E+06 Northing 1.078E+06 MW117 MW102 MW104 ** MW101 ** MW118 MW119 MW26W MW26B EW18 MW120 MW105 ** MW103 ** EW15 EW13 MW111A MW108 EW14 MW110 MW107 MW112 EW16 MW113 MW27W MW27B EW12 MW106 ** MW109 MW24B MW24W MW7B MW7W MW17W MW17B MW114 EW1 MW115 EW9 EW11 W. E EW10 MW3B MW3W MW11B EW7 MW11W EW5 MW19W MW19B E+06 MW2W MW2B MW18W MW18B E E E E+06 Easting MW1B MW1W

20 Wrapup Union of physical and statistical approaches is greater than sum of the parts Data are perfection challenged. Models are perfection challenged. Interpretations are perfection challenged. Decisions are perfection challenged. " Validation, robustness, observability are key.