Modeling Subsurface transport of oil-field brine at the USGS OSPER A Site, Oklahoma William Herkelrath and Yousif Kharaka U.S. Geological Survey Skiatook Lake What is it? How did it get there? What will happen to it? How long will it last? 100 m Skiatook Lake Oil wells at Skiatook Lake Time line of events at Skiatook Lake A Site: Oil production begins 1973 Major oil production stopped; site was abandoned; about 100,000 barrels were produced Skiatook Lake filled 2000 USGS studies began Skiatook Lake A Site map By Jim Otton Skiatook Lake A Site well locations A A(c) c c Geoprobe (1) Rotary (2) Otton Fig 6 Otton Fig 5 Distance along traverse, in meters Altitude, in meters above sea level AE07 AA06 AE51 AA01 AA02 AE06 AA10AE55 Skiatook Lake DRY pits salt scar erosion low: m (2/2003) Skiatook Lake levels high: m (3/2004) AE13 AA61 AA ,10022, ,700 19,100 29,900 10,800 11,300 26,100 ?? 4810 TDS (mg/L) 10,000 2,000 20,000 15,000 10,000 15,000 10,000 A A Total dissolved solids concentration distribution Transect parallel to the stream (Thordsen, et al.) W-E traverse, TDS 2-05 data Distance along traverse, in meters Altitude, in meters above sea level AA13 AA08 AE53 AA06 AA07 AA05 salt scar erosion low: 215.8m (2/2003) c c'' c' AA09 AA61 AE13 Skiatook Lake levels high: 219.1m (3/2004) 1,330 1, ,000 16,000 11,800 26,100 29,900 2,460 14,900 2,480 5,690 6, ,900 9, ,000 19,100 TDS (mg/L) SO 4 > Cl 2,500 5,000 15,000 10,000 2,500 5,000 20,000 10,000 Total dissolved solids concentration distribution Transect perpendicular to the stream (Thordsen, et al.) Barometric pressure Depth to water Barometric response removed BETCO (Toll and Rasmussen) Raw data Skiatook Lake A Site well locations A A(c) c c Geoprobe (1) Rotary (2) Skiatook monitoring well results: There is a large (~200x200x30 m) salt water plume (TDS up to ~30,000 mg/L). Wells have high barometric efficiency (0.75 to 0.90). Deconvolution analysis of the barometric response suggests the water bearing units are confined. None of the wells respond rapidly to rainfall. Water levels in many of the wells slowly move up and down in an annual wave pattern. Monitoring well results (cont.): Only the well closest to the lake responds to lake level changes. Water head in the other wells is above lake level, indicating flow is toward the lake. Skiatook slug test results: Dimensionless head vs. time data fit well to Cooper et al. type curves for confined aquifers. Hydraulic conductivity ranged from 0.3 to 7.0 cm/day (mean = 2.0 cm/day). Storage coefficient varied from 0.01 to 1x10 -6. Evidence/arguments concerning recharge: Annual potential evapotranspiration (~180 cm) is about double average rainfall (~89 cm). Soils are clay-rich and salt-affected, which promotes surface sealing, low soil hydraulic conductivity, and runoff. Rainfall typically occurs in violent bursts, promoting runoff (supported by our stream flow hydrograph data). Evidence/arguments concerning recharge (cont.): Hydraulic gradient in multiport wells is generally positive downward, indicating there is recharge. The hydraulic head (H) in many wells moves up ~1 meter January to June each year; assuming storage coefficient (S) = 0.01, possibly implies recharge = S* 1 cm/year. Evidence/arguments concerning recharge (cont.): The fact that a high concentration (TDS~30,000 mg/L) salt plume exists 33 years after the end of production indicates recharge and ground water flow through the plume is limited. If recharge and permeability are so low, how did the brine get there, and why is it still there? During the oil production era, the recharge was increased because the pits and the creek were full of salt water (up to 150,000 mg/liter). Salt water infiltrated for ~60 years. After abandonment, recharge reverted to low levels and fresh water flushing was slow The filling of Skiatook Lake reduced the lateral hydraulic gradient and ground water flow velocity A primitive flow and brine transport model of the Skiatook Lake A Site Model used STOMP (Subsurface Transport Over Multiple Phases) by White and Oostrom, Pacific Northwest National Lab Modeled flow and transport along a two- dimensional vertical slice running parallel to the stream (A-A transect). Ignored density effects. Primitive model assumptions model set up and initial conditions Solution domain 30 meters thick by 300 meters long 5 rock types (soil and Units 1-4). Hydraulic conductivity ~ 1 to 7 cm/day. Porosity ~0.10 to Recharge is steady at 1 cm/year Water table parallel to sloping ground surface at a depth of 5 meters Lateral head gradient is ~ m/m Longitudinal dispersion = 1.0 m, lateral dispersion 10.0 cm Primitive model assumptions Boundary conditions During oil productions, ( ) brine infiltrated beneath the pits at 15 cm/year. Brine TDS concentration 50,000 mg/L Oil production stopped in 1973, recharge conditions returned to ~ 1 cm/year. Skiatook Lake was filled in 1987, which raised the water table on the down slope end of the domain. ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1918 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1923 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1937 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1948 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1973 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1987 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 1997 ,100 22, ,700 19,100 29,900 10, , TDS concentration (mg/liter) Pit Lake Year 2005 Final comments: If the rocks and soils at the Skiatook A site were more permeable, the salt probably would have been flushed out long ago. Because of the low permeability, site remediation by water pumping/flushing is probably impractical. Left as it is, the salt scar will last many years.