1. Case Study: Permanganate Applied to VOCs in Fractured Shale Beth L. Parker, Ph.D. Department of Earth Sciences University of Waterloo Presented at the EPA Symposium in Chicago December 12, 2002 1

2. 2 Research Collaborators l Dr. Tom Al –University of New Brunswick l Dr. Ulrich Mayer –University of British Columbia l Drs. Ramon Aravena and John Cherry –University of Waterloo l Kenneth Goldstein, CGWP –Malcolm Pirnie, Inc l Grant Anderson, P.G. –U.S. Army Corps of Engineers

3. 3 Watervliet Arsenal in New York State Watervliet Arsenal Troy Hudson River Shale

4. 4 Hudson River Watervliet Arsenal Building 40 Area

5. Flow Overburden Bedrock PCE adapted from Mackay and Cherry, 1989 DNAPL DNAPL Passed Through Overburden Into Shale 5

6. 6 The Problem l Chlorinated ethenes – as high as 150 mg/L l Contamination down to 150 ft. bgs l All VOC mass in fractured shale l AOC is 200 ft west of Hudson River

7. 7 Study Area AW-MW-59 AW-MW-68 N Building 40

8. 8 Identification of Major Transmissive Zones Using Hydro-geophysics USGS, 2001

9. 9 Major Transmissive Zone Identified USGS, 2001

10. 93,600 5,310 5,150 16 MW-71 Fractures, Transmissive Zones, and PACKER TESTING Total VOCs from PACKER TESTING Leakage  g/L USGS, 2001 10

11. PCE and Degradation Products in Shale South North From packer test intervals during drilling PCE DCE 11

12. 12 Interconnected Fracture Network with Two Major Transmissive Zones 1 2 1 2 Cross-section view

13. 13 DNAPL was Initially Distributed in Many Fractures

14. ILLUSTRATION OF MATRIX POROSITY A Microscopic view of rock matrix mineral particle DETAIL A 14

15. 15 DNAPL Phase Initially Resides within Fractures Fracture Aperture 2b Fracture Spacing H O 2  f  m DNAPL Matrix porosity is 1000 times greater than fracture porosity

16. 16 DNAPL Disappearance by Diffusion Parker et al. (1994) Fracture Aperture 2b Fracture Spacing  f  m H O 2 DNAPL  f  m Dissolved Phase  f  m Dissolved Phase Early IntermediateLater Time

17. 17 Snake Hill Shale Formation Watervliet Site

18. Core Hole In Source Zone B.L. Parker, 2000 18

19. Diffusion Into Rock Matrix 19

20. Core Sampling for Migration Pathway Identification B.L. Parker, 2000 20

21. Core Diameter 3.2 Inches Sample Length ~Two Inches Overview of the Rock Core Method Parker and Colleagues 1997 21

22. 22 Two Rock Crushers Rock Crusher Crushing Cell 22

23. MW-74 Dec. 2001 (preliminary data) 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 0.010.1110100100010000 0 1E+06 conc. VOC (ug/L) depth (ft) PCE TCE c-DCE PCE solubility (240 mg/L) Rock Core Profile Dec 2001 Pre-KMnO 4 Injection CMT Aqueous PORTS Concentrations (Feb. 2002) 1 2 3 4 5 6 PCE – 8 ug/L TCE – 5 ug/L c-DCE – 350 ug/L PCE – 4 ug/L TCE – ND c-DCE – 1088 ug/L PCE – 458 ug/L TCE – 123 ug/L c-DCE – 2802 ug/L PCE – 9183 ug/L TCE – 382 ug/L c-DCE – 524 ug/L PCE – 13,988 ug/L TCE – 3,699 ug/L c-DCE – 15,155 ug/L PCE – 172 ug/L TCE – 259ug/L c-DCE – 11,745 ug/L 23

24. 24 WESTBAY ® MP SYSTEM 2.5 ft. Tripod Cable Reel Pressure Probe Sample Bottle 24

25. 25 SOLINST CMT ® SYSTEM 25

26. 26 Site Conceptual Model l VOC migration occurs in a large number of interconnected fractures l Nearly all VOC mass resides in the rock matrix rather than in the fractures

27. It is well established that permanganate completely destroys chlorinated ethenes However, to do so, it must be delivered to the contaminant mass 27

28. Can permanganate be effective for remediating chlorinated ethenes in fractured sedimentary rock? Important factors: l Delivery throughout fracture network l Diffusion rates into rock matrix l Oxidant Demand of Shale 28

29. 29 KMnO 4 Injections at Watervliet

30. 71 74 Potassium Permanganate Injection 59 76 75 65 34 ~145 feet bgs Phase 1 injection Phase 2 injection 30

31. 31 PILOT STUDY RESULTS in 2002 31

32. 32 Treatment Approach Permanganate Permanganate oxidizes chlorinated ethenes Solvent + MnO 4 - MnO 2 (s) + Cl - + Acid 13 C / 12 C and Chloride used to confirm destruction Stable chemistry in subsurface allows time for diffusion into matrix

33. 33 Remediation in Fractured Porous Media Treatment zone Early Time Later Time KMnO 4 in fracture Contaminated clay/rock B.L. Parker, 1993

34. 34 In Situ Oxidation in Fractured Porous Media Diffusion of both reactants occurs in opposite directions Readily destroys sorbed phase contaminants Greatly reduces time scale for remediation B.L. Parker, 1993

35. 35 Analogy to Fractured Shale Results from Permanganate Field Tests in Marine Clay

36. Oxidized zone shows extent of diffusion invasion and treatment by KMnO 4 B.L. Parker, 1996 Invasion front Top of clay 36

37. 37 KMnO 4 Profile in Clay

38. 38 Combined Profiles in Clay Reaction interface

39. 39 Snake Hill Shale Formation Watervliet Site

40. 40 Permanganate Diffusion into Matrix from Fracture z x y diffusion 2b = fracture aperture mass x x + C/C o

41. Elemental Manganese Profiles in Shale Transects Normal to Fractures Propagating in from Surface of Rock Sample Shale sample in 10 g/L KMnO 4 solution for ~6 weeks 41

42. 42 How long will MnO 4 take to remediate the source zone? l Answer being sought using ~field data ~laboratory tests ~numerical models

43. Preliminary MIN3P Simulations Watervliet Arsenal 3D multicomponent reactive transport model Model developed by Dr. Ulrich Mayer (1999) Now being modified for permanganate oxidation 43

44. 44 MIN3P Simulation Simulate 1D MnO 4 - invasion into shale matrix where PCE has been diffusing in for 40 years to examine rates of matrix clean-up Parameters: Site-specific , D e, f oc MnO 4 - R = 1 PCE R = 220 (estimated using f oc =0.5%) Source [ PCE ] = 150 mg/L for 40 years Injection [ KMnO 4 ] = 5 g/L

45. Chloride Diffusion Test Cell for Rock Golder Associates, Toronto 45

46. 46 Initial Condition 40 years PCE Diffusion-In MIN3P Model – Snake Hill Shale

47. 47 Matrix Profiles after 1 year MnO 4 - Injection MIN3P Model – Snake Hill Shale

48. 48 Matrix Profiles after 2 years MnO 4 - Injection MIN3P Model – Snake Hill Shale

49. 49 Matrix Profiles after 5 years MnO 4 - Injection MIN3P Model – Snake Hill Shale

50. 50 Partial Mass Destruction Greatly diminishes VOC mass flux into fracture network after permanganate is gone

51. PCE flux to fracture after partial permanganate treatment Frac3DVS Modeling Log - Log Scale No MnO 4 Case 51

52. 52 Preliminary Conclusions Permanganate… Diffuses and reacts in low K matrix Prevents release of mass from matrix to flowing groundwater while present in fractures Greatly reduces magnitude of flux from matrix even after partial treatment

53. 53 How do we know that VOCs are being destroyed ? l Chloride increases at many locations l Change in carbon isotope ratio of PCE

54. 54 ON-GOING WORK l Rebound monitoring after pilot injections l Permanganate invasion tests –Laboratory samples –Field cores l Reactive transport modeling –Single fractures and fracture networks l Design of full-scale system and monitoring

55. 55 Acknowledgements l Project Contributors: ~Steven Chapman and Martin Guilbeault (UWaterloo) ~Daria Navon and Andrew Vitolins (Malcolm Pirnie) ~Stephen Wood (U.S. Army Corps of Engineers) ~JoAnn Kellogg (Watervliet Arsenal) ~John Williams and Fred Paillet (U.S.G.S) l Funding: ~U.S. Army Corps of Engineers ~Solvents-In-Groundwater Research Program

56. 56 The End