-1 r r n CONFIDENTIAL REVIEW AND EVALUATION r OF GROUND-WATER CONTAMINATION AND PROPOSED REMEDIATION AT THE REILLY TAR SITE, ST. LOUIS PARK, MINNESOTA r. c Prepared by L Dr. James W. Mercer GeoTrans, inc. 209 Elden Street Herndon, Virginia 22070
[ Report to L U.S. Environmental Protection Agency Region V, Remedial Response Branch (5HR-13) Chicago, Illinois 60604
December 1984 L ieoT L GEOTRANS, INC. lrran« s P.O. Box 2550 Reston.Virginia 22090 USA (703)435-4400
EPA Region 5 Recorcte Ctr. i inn minium iflBiniiiiiiiw L 234542 r.
CONFIDENTIAL
REVIEW AND EVALUATION OF GROUND-WATER CONTAMINATION AND PROPOSED REMEDIATION AT THE REILLY TAR SITE, ST. LOUIS PARK, MINNESOTA
Prepared By Dr. James W. Mercer GeoTrans, Inc. 209 El den Street Herndon, Virginia 22070
Report To U.S. Environmental Protection Agency Region V, Remedial Response Branch (5HR-13) Chicago, Illinois 60604
December 1984 TABLE OF CONTENTS Page LIST OF FIGURES v LIST OF TABLES vii 1.0 INTRODUCTION 1 1.1 PURPOSE AND SCOPE 1 1.2 SITE HISTORY 2 2.0 CONCLUSIONS AND RECOMMENDATIONS 4 2.1 CONCLUSIONS 4 2.2 RECOMMENDATIONS 6 3.0 SITE HYDROGEOLOGY 7 3.1 GEOLOGY 7 3.1.1 Stratigraphy 7 3.1.2 Geomorphic Features 14 3.2 GROUND-WATER HYDROLOGY 17 3.2.1 Flow Directions 17 3.2.1.1 Mount Simon-Hinckley Aquifer 17 3.2.1.2 Ironton-Galesville Aquifer 25 3.2.1.3 Prairie du Chien-Jordan Aquifer 25 3.2.1.4 St. Peter Aquifer 29 3.2.1.5 Drift-Platteville Aquifers 33 3.2.1.6 Vertical Gradients 35 3.2.2 Flow Properties 35 3.3 CHEMISTRY 39 4.0 GROUND-WATER MODELING 42 4.1 CODE SELECTED 42 4.2 GEOMETRY 43 4.2.1 Layering 43 4.2.2 Boundary Conditions 47 TABLE OF CONTENTS
Page 4.3 HISTORY MATCHING 47 4.3.1 Wells 48 4.3.2 Flow Parameters 49 4.3.3 Results 49 4.4 PREDICTIVE SIMULATION WITH REMEDIATION 57 5.0 REMEDIAL ACTIONS 60 5.1 MULTIAQUIFER WELLS 60 5.1.1 Investigation 60 5.1.2 Remedy 61 5.2 NEAR-SURFACE CONTAMINATION 61 5.2.1 Deed Restrictions 61 5.2.2 Louisiana Avenue/Highway 7 61 Intersection Construction 5.3 DRIFT AQUIFER 62 5.3.1 Source Control 62 5.3.2 Gradient Control 62 5.3.3 Monitoring 62 5.3.4 Contingencies 63 5.3.5 Mitigated Impacts 63 5.4 PLATTEVILLE AQUIFER 63 5.4.1 Source Control 63 5.4.2 Gradient Control 63 5.4.3 Monitoring 64 5.4.4 Contingencies 64 5.4.5 Mitigated Impacts 64 5.5 ST. PETER AQUIFER 64 5.5.1 Source Control 64 5.5.2 Gradient Control 64 5.5.3 Monitoring 65 5.5.4 Contingencies 65 5.5.5 Mitigated Impacts 65 TABLE OF CONTENTS
5.6 PRAIRIE DU CHIEN-JORDAN AQUIFER 65 5.6.1 Source Control 65 5.6.2 Gradient Control 66 5.6.3 Monitoring 66 5.6.4 Drinking Water 66 5.6.5 Contingencies 67 5.6.6 Mitigated Impacts 67 5.7 IRONTON-GALESVILLE AQUIFER 67 5.7.1 Source Control 67 5.7.2 Gradient Control 68 5.7.3 Monitoring 68 5.7.4 Contingencies 68 5.7.5 Mitigated Impacts 68 5.8 MOUNT SIMON-HINCKLEY AQUIFER 68 5.8.1 Source Control 68 5.8.2 Gradient Control 68 5.8.3 Monitoring 69 5.8.4 Contingencies 69 5.8.5 Mitigated Impacts 69 6.0 REFERENCES 70 7.0 GLOSSARY OF TERMS 72 LIST OF FIGURES
Number Legend Page 1.1 Location of former plant, St. Louis Park, Minnesota 3 (from Hult and Schoenberg, 1984). 3.1 Preliminary structural contours at the top of the 11 Mount Simon-Hinckley aquifer, where contour interval is 40 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984). 3.2 Preliminary structural contours at the top of the 12 Ironton-Galesville aquifer, where contour interval is 40 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984). 3.3 Preliminary structure contours at the top of the 13 Prairie du Chien-Jordan aquifer, where contour interval is 40 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984). 3.4 Preliminary structure contours at the top of bedrock, 15 where contour interval is 100 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984). 3.5 Preliminary bedrock geology, Minneapolis South and 16 Hopkin's quadrangles (from Hult and Schoenberg, 1984). 3.6 Hydrogeologic section through Minneapolis and St. Paul 18 (from Horn, 1983). Section trace not shown, but lies in an approximate east-west direction. 3.7 Potentiometric surface for the Mount Simon-Hinckley 19 aquifer based on water levels measured during January- March 1971 (from Schoenberg, 1984). 3.8 Potentiometric surface for the Mount Simon-Hinckley 20 aquifer based on water levels measured during January- February 1980 (from Schoenberg, 1984). 3.9 Potentiometric surface for the Mount Simon-Hinckley 21 aquifer based on water levels measured during August 1980 (from Schoenberg, 1984). 3.10 Hydrograph showing water level changes in the 24 Mount Simon-Hinckley aquifer in well number 117N21W32DAD01 in Edina, MN (from Schoenberg, 1984). Well location shown in Figures 3.7 to 3.9. LIST OF FIGURES I \ ' Number Legend Page f 3.11 Potentiometric surface for the Prairie du Chien- 26 i Jordan aquifer based on water levels measured during January-March 1971 (from Schoenberg, 1984). , 3.12 Potentiometric surface for the Prairie du Chien- 27 ! Jordan aquifer based on water levels measured during January-February 1980 (from Schoenberg, 1984). 1 3.13 Potentiometric surface for the Prairie du Chien- 28 Jordan aquifer based on water levels measured during j August 1980 (from Schoenberg, 1984). 3.14 Hydrograph showing water level changes in the Prairie 31 , du Chien-Jordan aquifer in well number 117N21W16CCA01 in east-central Hennepin County, MN (from Schoenberg, ' 1984). Well location shown in Figures 3.11 to 3.13. 3.15 Potentiometric surface for the St. Peter aquifer based 32 ; on water levels measured during the winter of 1970-1971 (from Norvitch et al., 1974). 3.16 Generalized potentiometric surface for the drift 34 aquifer based on water levels measured on June 5, 1979 (from Ehrlich et al., 1982). Contours are in meters. 4.1 Map showing boundaries used by ERT and U.S.G.S. 44 (after ERT, 1983). i 4.2 Comparison of calibrated ERT model results with observed 54 data for the Prairie du Chien-Jordan aquifer (from ERT, 1983). 4.3 Map showing the location of private and industrial wells 56 in the vicinity of the Reilly Tar site (from Hult and Schoenberg, 1984).
VI LIST OF TABLES
Number Legend Page 3.1 Stratigraphic Section Underlying Reilly Tar Site 8 (From Hult and Schoenberg, 1984) 3.2 Average Daily Pumpage From the Mount Simon-Hinckley 23 Aquifer, 1970-1979 (From Schoenberg, 1984) 3.3 Annual Daily Pumpage From the Prairie du Chien-Jordan 30 Aquifer, 1970-1979 (From Schoenberg, 1984) 3.4 Flow Properties of the Various Units Beneath the 36 Reilly Tar Site 4.1 Vertical Aquifer Geometry Used in the ERT Model 45 4.2 Vertical Aquifer Geometry Used in the U.S.G.S. Model 46 4.3 Values of U.S.G.S. Hydrologic Properties "50 (After Stark and Hult, 1984) 4.4 Values of ERT Model Hydrologic Properties 51 (After ERT, 1983) 4.5 Values Used by ERT to Compute Travel Times 52 4.6 Withdrawal Rates for Remedial Action Simulation 58 (From Stark and Hult, 1984)
VII 1 REVIEW AND EVALUATION OF GROUND-WATER CONTAMINATION AND PROPOSED REMEDIATION AT THE REILLY'TAR SITE, I ST. LOUIS PARK, MINNESOTA ii
| 1.0 INTRODUCTION r This section outlines the purpose and scope of work. In addition, ' for completeness and to help highlight the problem, a brief site history ! is provided.
\!. 1.1 PURPOSE AND SCOPE In November 1984, GeoTrans, Inc. began an investigation of the ' Reilly Tar site in St. Louis Park, Minnesota, for purposes of trial preparation. The work was funded by the U.S. Environmental Protection Agency through a subcontract with PRC Engineering. The work was ) performed in conjunction with the U.S. Department of Justice, Minnesota Pollution Control Agency, and the Minnesota Attorney General's Office. The objectives of this investigation were, in part, to (1) review site « and basin-wide hydrogeologic data, (2) develop a conceptualization of site hydrology, (3) evaluate ground-water modeling performed at the site \ using the conceptualization of site hydrology as a basis, and (4) use all of the above to evaluate proposed site remediation. | The purpose of this report is to concisely summarize findings relevant to the objectives stated above. The report includes conclusions and recommendations, a review of site hydrogeology, and a i review of ground-water modeling, including proposed site remediation. L Where possible, conclusions have been summarized with supporting facts/ assumptions and data/reference sources. Comments have been restricted to those that concern hydrogeology. 1.2 SITE HISTORY Between 1917 and 1972, a coal-tar distillation and wood-preserving plant was operated by Reilly Tar & Chemical Corporation (RT&CC) on an | 80-acre site in St. Louis Park, Minnesota, as shown in Figure 1.1. Note the location of the former plant site, as it will be used for reference ir 1 in later figures. Coal-tar derivatives from this operation have j contaminated glacial drift and bedrock aquifers in this area (Hult and Schoenberg, 1984). Contamination of the Prairie du Chien-Jordan ' aquifer, the region's major source of ground water, was first documented in 1932 at a location 3,500 feet from the plant (Hult and Schoenberg, ^ 1984). Because the Prairie du Chien-Jordan aquifer underlies the area ; at depths of 250-500 feet, the most likely explanation for contamination i at this depth is the introduction of coal tar directly into multiaquifer 5 wells on or near the former plant site. In 1972, RT&CC ceased operation of the St. Louis Park facility and removed all equipment and buildings from the site (ERT, 1983). Since 1983, at least three major construction projects have been initiated at the site in order to convert the site to non-industrial uses (ERT, '. 1983). Contamination problems, however, have continued. In 1978, the occurrence of polynuclear aromatic hydrocarbon (PAH) compounds was ! documented and four St. Louis Park municipal wells were shut down (Stark and Hult, 1984). Use of these wells was discontinued, and during 1979 through 1981, three additional wells had to be removed from active use. ) Currently, a proposed remediation plan for the Reilly Tar site i calls for a combination of contaminant source control and hydraulic gradient control wells. The analysis used to help design this system of wells incorporates a three-dimensional, ground-water flow model. ! HENNEPlVt j COI.— i r
PLYMOUTH 93»25'
GOLDEN VALLEY
22*30"
ST. LOL//S P/i/?K
MINNETONKA BOULEVARD
Site of former plant
44*55' Base from U. S. Geological Survey Hopkint and Minneapolis South; both maps 1:24.OOO. 1067 (PhotoravUad 1972) 1 KILOMETER
Figure 1.1. Location of former plant, St. Louis Park, Minnesota (from Hult and Schoenberg, 1984). 2.0 CONCLUSIONS AND RECOMMENDATIONS This section contains conclusions about the flow system and extent of contamination, and recommendatons on a remedial action p*lan.
2.1 CONCLUSIONS The conclusions based on information discussed in this report are as follows: (1) The ground-water system underlying the Reilly Tar site is complex owing to stratigraphy, geomorphic features, and extensive development. (2) Coal-tar derivatives have entered the ground-water system at the Reilly Tar site via three major paths (i) infiltration of spills and drippings on the site itself, (ii) recharge from ponds south of the site that received surface runoff and contaminated process water, and (iii) recharge through deep multiaquifer wells located on or near the site. 1 (3) Because of these mechanisms, every aquifer beneath the Reilly Tar site has been contaminated with coal-tar derivatives A J \ consisting primarily of PAH and phenolic compounds. (4) Contaminants in the Drift aquifer are moving laterally to the east and southeast and vertically downward into the Platteville aquifer. Contaminants in the Drift and Platteville aquifers / 1 t -N have moved a minimum of 4,000 feet to the east. ^{ oJlvc^ . (5) Toward the east and south of the plant site, the Glenwood confining bed has been eroded and replaced by drift material in a buried bedrock valley. Where the Glenwood confining bed is eroded, contaminated water moves downward entering the underlying St. Peter aquifer and valley-fill materials. (6) Contaminants entering the St. Peter aquifer via either multiaquifer wells or buried valleys, generally travel eastward and southward from the plant site. * (7) Contamination of the Prairie du Chien-Jordan aquifer, the region's major source of ground water, was first documented in 1932 at a location 3,500 feet southeast from the plant site. (8) Conservative contaminants can move in the Prairie du Cjvien-Jordaji_aqu,rFer at a iejjp_clty_of_3_ft/d. following the regional gradient toward the south-southeast. (9) Contaminants have been found in wells in the Prairie du Chien-Jordan aquifer that are north, west, and southwest of the plant site. The presence of contaminants in these wells may be explained by pumping from nearby municipal and industrial wells and the influence of multiaquifer wells. Whereas the pumping wells have caused water-level drawdown, the multiaquifer wells have produced a hydraulic head increase in the Prairie du Chien-Jordan aquifer. Both of these effects have contributed to altering the regional eastward hydraulic gradient. (10) Contamination of deeper aquifers—the Ironton-Galesville / aquifer and the Mount Simon-Hinckley aquifer—has also I i\ *''Yi^ occurred as a result of recharge through multiaquifer wells. The behavior and extent of contamination in these aquifers is less well understood because of data limitations. ( 2.2 RECOMMENDATIONS ' Based on the extent of contamination and the importance of the impacted ground-water supply to the area, remediation of th'e site to /" j contain and clean up the contamination is necessary. Details of the remediation are given in Section 5.0 and are not repeated here. In i general terms, the remediation should consist of pumping and treatment '" managed through monitoring such that the contamination within each aquifer is controlled and diminished with time. 3.0 SITE HYDROGEOLOGY [ In order to adequately evaluate the ground-water flow modeling, a brief review of the site hydrogeology is provided in this section. / ! Although emphasis is placed on the site, some basin-wide aspects are discussed as appropriate. This section has three subsections including (1) geology, (2) ground-water hydrology, and (3) chemistry. The geology subsection addresses stratigraphy and system geometry. The ground-water hydrology subsection includes flow directions and flow properties, and r~ the chemistry subsection presents the importance and extent of contamination. *i.
3.1 GEOLOGY The purpose of this section is to lay the framework for the ground water and chemistry sections. It is divided into stratigraphy and geomorphic features. i f 3.1.1 Stratigraphy The Reilly Tar site overlies a stratigraphic section that can best be described as layer-cake geology. The section consists of consolidated sedimentary rocks that were deposited in a north-south j s trending trough, known as the Twin Cities artesian basin. Glacial drift mantles the area and fills valleys and channels that dissect the section. The stratigraphic section is shown in Table 3.1. As may be seen, there are ten hydrogeologic units, including six aquifers. The lowermost layer shown in Table 3.1 is the Mount Simon-Hinckley / aquifer. It is underlain by Precambrian rock and is composed of silty , to coarse-grained sandstone. It is generally greater than 250 feet in Table 3.1. Stratigraphic Section Underlying Reilly Tar Site (From Hult and Schoenberg, 1984)
Approximate Hydrogeologic Maximum Thickness Geologic Water-Bearing Unit 1n Feet Characteristics Characteristics
Drift, 220 Till, outwash, lake clay, peat, and valley fill. Hydraulic characteristics highly visible. Till undlfferentlated has vertical hydraulic conductivity as low as 10*8ft/s. Burled outwash aquifer has transmlsslvlty as high as 10,000 ft2/d. Plattevllle 35 Dolomltlc limestone and dolomite, gray to buff, thin Hydraulic conductivity primarily from fractures, aquifer to medium-bedded, some shale partings. Solution open joints, and solution channels. Specific channels and fractures are concentrated 1n upper part capacities of wells are generally between 10 and and contain sand of glacial origin. 100 gal/min/ft of drawdown, If pumped at about 12 gal/m1n for one hour. Glenwood Shale and claystone, green to buff, plast1c<£cTsSjj Very low hydraulic conductivity. Vertical confining to slightly flssll; lower 3 to 5 feet grade from hydraulic conductivity 1s estimated to be about bed claystone with disseminated sand grains to sandstone 10"10ft/s based on laboratory measurements of core with clay matrix. samples. St. Peter 100 Sandstone, white to yellow, very well sorted, fine to Supplies about 10X of ground water pumped In the aquifer medium-grained, poorly cemented, quartzose. St. Louis Park area. Can yield more than CO 500 gal/m1n. Sandstone Is poorly cemented and wells tend to pump sand or fill In. Basal 65 Slltstone and claystone, red, green, and white; parts Hydraulic conductivity 1s highly anlsotroplc; St. Peter are plastic 1n texture and poorly indurated; Inter- siltstone and claystone restrict vertical flow confining bedded with fine-grained quartz sandstone. Individual but sandstone may yield as much as 100 gal/min bed beds are continuous in the vicinity of the site. to wells. Prairie du Chlen- 210 Prairie du Chlen: Dolomite, grayish-brown to buff, Supplies about 751 of ground water pumped In the Jordan aquifer generally thickly bedded In vicinity of site. St. Louis Park and Metropolitan area. Generally Jordan: Sandstone, white to pink, fine to coarse- yields more than 1,000 gal/min to high-capacity grained, moderately well cemented, quartzose to wells. Hydraulic conductivity of the Prairie du dolomltlc. Chlen part of the aquifer Is due to fractures, open joints, and solution channels. (Continued on next page) Table 3.1. Continued
Approximate Hydrogeologlc Maximum Thickness Geologic Water-Bearing Unit In Feet Characteristics Characteristics
St. Lawrence- 150 Slltstone and sandstone, gray to green, poorly sorted, Confining bed; hydraulic characteristics poorly Franconla glauconltlc and doloniitlc. known. confining bed Ironton- 50 Sandstone, white to light green, moderately well sorted, Regionally an aquifer, but no wells are known to Galesvllle fine to coarse-grained, quartzose. yield water only from this unit In the study area. aquifer Eau Claire 105 Slltstone and shale, green, glauconltlc. Confining bed; hydraulic characteristics poorly confining bed known. Mount Slmon- >250 Sandstone, grayish-white to pink, sllty to coarse- Supplies about 15X of ground water pumped 1n the Hlnckley grained, well cemented, quartzose; parts are medium St. Louis Park and seven-county metropolitan area. aquifer to coarse-grained, well sorted. Generally, yields more than 1,000 gal/mln to high-capacity wells. thickness and dips in a southeasterly direction as shown in Figure 3.1, which is a structural map showing elevations at the top of the Mount Simon-Hinckley aquifer. » The Mount Simon-Hinckley aquifer is overlain by the Eau Claire C • - J( confining bed, which is-cbmpr-'used.of siltstone and shale. It is approximately 105 feet in thickness. Overlying the Eau Claire confining bed is the Ironton-Galesville aquifer. It is also a sandstone, but is only approximately 50 feet thick. As with the Mount Simon-Hinckley aquifer, the Ironton-Galesville dips in a southeasterly direction as indicated in the structural map in Figure 3.2. Continuing up in the stratigraphic section, the next unit is the St. Lawrence-Franconia confining bed. This unit is comprised of siltstone and sandstone, and is approximately 150 feet thick. The next unit is the Prairie du Chien-Jordan aquifer, which is approximately 210 feet thick. This aquifer contains two rock types. The Jordan Sandstone is fine to medium-grained sandstone, whereas the Prairie du Chien consists of dolomite, shale, and sandy dolomite. A structural map at the top of this unit is given in Figure 3.3, which shows a relatively flat surface with a slight southeasterly dip. The Prairie du Chien-Jordan aquifer is overlain by the basal confining units of the St. Peter aquifer, or by glacial drift, where the St. Peter has been removed by erosion. This unit is composed of siltstone and claystone and is approximately 65 feet thick. The next unit is the St. Peter aquifer, which is a fine to medium- grained sandstone. It is approximately 100 feet thick.
10 93*15'
44*S2'30"
012 MILES I I iI Ii I 'Ii '—i_ 3 KILOMETERS
Figure 3.1. Preliminary structural contours at the top of the Mount Simon-Hinckley aquifer, where contour interval is 40 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984).
11 r ^.x>\ ?; ^ife).7\H-.-.f;Vv.'-wiV '"i )-\^%- J "i\f t
012 MILES i I i i i i i 0123 KILOMETERS
Figure 3.2. Preliminary structural contours at the top of the Ironton-Galesville aquifer, where contour interval is 40 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984).
12 17'30" »3'15'
44«62'30'
2 MILES
2 S KILOMETERS
Figure 3.3. Preliminary structure contours at the top of the Prairie du Chien-Jordan aquifer, where contour interval is 40 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984).
13 The St. Peter aquifer is overlain by the Glenwood confining bed. This is a thin unit (approximately 7 feet thick) that is comprised of shale and claystone. ' The Platteville aquifer is the next unit, which consists of dolomitic limestone and dolomite. It is approximately 35 feet thick. The uppermost unit is glacial drift. This is a thick deposit of till, outwash, and valley-train sand and gravel, lake deposits, and alluvium. Stark and Hult (1984) indicate at the bottom of this unit is a shale, called the Decorah Shale, that, where present, separates the glacial drift from the Platteville aquifer.
3.1.2 Geomorphic Features According to Hult and Schoenberg (1984), four major geomorphic processes may be responsible for the relief on the bedrock surface, where the Platteville Limestone is considered the uppermost bedrock unit. These processes are (1) stream erosion, (2) subglacial plucking and abrasion, (3) subglacial stream erosion, and (4) in situ chemical weathering. As indicated previously, in certain locations, erosion has removed the Platteville aquifer, the Glenwood confining bed, the St. Peter aquifer, and the basal St. Peter confining bed. In these locations, drift overlies the Prairie du Chien-Jordan. • To demonstrate the effect of the geomorphic processes on the bedrock surface, a structural map at the top of bedrock is shown in Figure 3.4. This was drawn displaying a dendritic pattern on the assumption that stream erosion is responsible for the major depressions in the bedrock surface. The bedrock geology corresponding to this surface is shown in Figure 3.5. As may be seen, below the site, the drift lies on top of the Platteville and Glenwood Formations. Southeast
14 P • 3'30' 27'30" 17'30 • 3'1S' 45*00' r
87'30"-
55'-
44*32'30"
2 9 KILOMETERS
Figure 3.4. Preliminary structure contours at the top of bedrock, where contour interval is 100 feet and datum is National Geodetic Vertical Datum of 1929 (from Hult and Schoenberg, 1984).
15 17*30* • 3*15*
67*30"-/--W
44«52'30-
* • KILOMETERS
CORRELATION OF MAP UNITS Platteville and Glenwood"^ . |0pg Formations, undivided : r1 . \ ; |_Qsp St. Peter Sandstone /; Ope Prairie du Chien Group J!
Figure 3.5. Preliminary bedrock geology, Minneapolis South and Hopkin's quadrangles (from Hult and Schoenberg, 1984)
16 of the site, approximately 0.5 miles, the drift overlies the St. Peter sandstone. The irregular bedrock surface with geomorphic features is illustrated further in Figure 3.6, which is a hydrogeologic*section through Minneapolis and St. Paul. Note, particularly, the region under the Mississippi River.
3.2 GROUND-WATER HYDROLOGY The purpose of this section is to describe the ground-water flow in each of the aquifers discussed in the preceding section. The flow patterns will help to interpret ground-water chemistry discussed in the next section. The topic of ground-water hydrology is divided into flow directions and flow properties.
3.2.1 Flow Directions Where possible, potentiometric surfaces and flow directions are discussed for each aquifer described in Section 3.1.1. In ascending order, these include: (1) Mount Simon-Hinckley aquifer, (2) Ironton- Galesville aquifer, (3) Prairie du Chien-Jordan aquifer, (4) St. Peter aquifer, (5) Platteville aquifer, and (6) Drift aquifer.
3.2.1.1 Mount Simon-Hinckley Aquifer According to Schoenberg (1984), 10% of the ground water pumped for water supply in the Twin Cities metropolitan area comes from the Mount Simon-Hinckley aquifer. Potentiometric surfaces for the Mount Simon-Hinckley aquifer are shown in Figures 3.7 to 3.9. As may be seen, the highest water levels (about 900 feet above sea level) occur toward the northwest. The lowest water levels (less than 600 feet above sea level) correspond to the location of major pumping centers in east-central Hennepin County. 17 200
SEA LEVEL FOND DU LAC FORMATION 200 200 Geology by M. A. Jursa Minnesota Geological Survey. 1980 2 MILES 0 1 2 KILOMETERS VERTICAL EXAGGERATION x 40 NATIONAL QEODETIC VERTICAL DATUM OF 1929
EXPLANATION St. Lawrence-Franconia Glacial drift aquifer confining bed
Decorah-Plattevllle- Ironton-Galesville aquifer Glenwood
St. Peter aquifer Eau Claire confining bed
Prairie du Chlen- Mount Slmon- Jordan aquifer Hinckley aquifer
Figure 3.6. Hydrogeologic section through Minneapolis and St. Paul (from Horn, 1983). Section trace not shown, but lies in an approximate east-west direction.
18 44* 10'-' EXPLANATION &t L 'iL£—I —too- WATER-LEVEL CONTOURS --Shows aDBroximate altitude It which <•»!*' l*v«i would have tlood in tightly cased wells Hatchurtd linct indicate • rea within a con* o* dCDr***ion Contour interval 60 feet National Geodetic Vertical Datum el 1029
Figure 3.7. Potentiometric surface for the Mount Simon-Hinckley aquifer based on water levels measured during January-March 1971 (from Schoenberg, 1934).
19 D .A K O -T A
<& lrrr^._±rrrr_^_V_ i Cjm* »v » [•.„.•... ••-••i
EXPLANATION t._.Jfi^r' —MO- WATER-LEVEL CONTOURS --Shows •DOroximait (Itrtud* *t which water level would have «tooa in lightly eased wells. Htlcrtured lines indicate area within a cone of depression Contour interval SO feet National fleodetic Vertical Datum el H28
Figure 3.8. Potentiometric surface for the Mount Simon-Hinckley aquifer based on water levels measured during January-February 1980 (from Schoenberg, 1984).
20 -Z')-~J
DAKOTA
EXPLANATION —«00~ WATER-LEVEL CONTOURS --Shows •oproiimat* altitude it which w«i«r level would have stood in lightly eased wells Hsichured tines indicate • rea within a cone o< depression Contour interval 80 Figure 3.9. Potentiometric surface for the Mount Simon-Hinckley aquifer based on water levels measured during August 1980 (from Schoenberg, 1984). 21 Natural ground-water flow in the Mount Simon-Hinckley aquifer is from the northwest toward the southeast and the Minnesota and Mississippi Rivers (Schoenberg, 1984). This natural flow hfes been altered by pumpage that has caused a large cone of depression in the water level surface that was largest in 1971 (see Figure 3.7). The cone of depression encircles major pumping centers in St. Louis Park, Edina, and St. Paul. Schoenberg (1984) points out that the Mount Simon- Hinckley aquifer is only slightly hydraulically connected to streams and is greatly influenced by pumping. As may be seen in Figure 3.7, the water levels are for January- March 1971. Schoenberg (1984) suggests that water levels measured during the winter represent "average" annual water levels because water levels are fairly stable during the winter months. By comparing water levels measured in the winters of 1971 and 1980 (Figures 3.7 and 3.8), it may be noted that average water levels in the Mount Simon-Hinckley aquifer rose by as much as 60 feet in the center of the cone of depression during this period. The most likely reason for this increase in hydraulic head is the decreased pumping indicated in Table 3.2. Superimposed on this 10-year trend are seasonal changes in water levels. Figure 3.9 shows water levels in the Mount Simon-Hinckley aquifer for the summer of 1980. By comparing Figures 3.8 and 3.9, it may be noted that water levels declined by more than 50 feet over a large area. This decline is also indicated in Figure 3.10, which shows a hydrograph for a well in Edina. The large decrease in water level in 1980 is primarily due to increased summer pumpage due to below-average precipitation during 1980 (Schoenberg, 1984). As indicated in Figure 3.10, a similar event occurred in 1976. 22 Table 3.2. Annual Daily Pumpage From the Mount Simon-Hinckley Aquifer, 1970-1979 (From Schoenberg, 1984) Year Pumpage (MG/D) 1970 24.9 1971 22.8 1972 21.8 1973 23.6 1974 22.3 1975 21.1 1976 21.0 1977 19.2 1978 18.6 1979 20.9 23 UJ 260 Well number 117N21W32DAD01 I i 19711972 1973,1874 1975 1976 1977 1978 1979 1980 Figure 3.10. Hydrograph showing water level changes in the Mount Simon- Hinckley aquifer in well 117N21W32DAD01 in Edina, MN (from Schoenberg, 1984). Well location shown in Figures 3.7 to 3.9. 24 3.2.1.2 Ironton-Galesville Aquifer Little information is available on the water levels in this aquifer. No published potentiometric surfaces were found. * 3.2.1.3 Prairie du Chien-Jordan Aquifer According to Schoenberg (1984), 80% of the ground water pumped for water supply in the Twin Cities metropolitan area comes from the Prairie du Chien-Jordan aquifer. Potentiometric surfaces for the Prairie du Chien-Jordan aquifer are shown in Figures 3.11 to 3.13. As may be seen, the highest water levels (more than 900 feet above sea level) occur in northern Washington County, central Hennepin County, and southern Scott and Dakota Counties; the lowest water levels (less than 700 feet above sea level) occur along the Mississippi River toward the southeast. In addition to these regional trends, a locally persistent mound in the potentiometric surface of the Prairie du Chien-Jordan aquifer occurs just east of the plant site (Stark and Hult, 1984). According to Schoenberg (1984), the Mississippi, Minnesota, and St. Croix Rivers are in hydraulic connection with and influence the pattern of flow in the Prairie du Chien-Jordan aquifer. Water generally flows toward these rivers from water level highs northeast, northwest, and south of the Twin Cities. Considerable recharge occurs in the Lake Minnetonka area (Sunde, 1974). In Hennepin County, water in the Prairie du Chien-Jordan aquifer generally flows from west to east under a regional hydraulic gradient of about 10 ft/mi, which increases near the Mississippi and Minnesota Rivers (Stark and Hult, 1984). Because of the transmissive nature of the Prairie du Chien-Jordan aquifer and its hydraulic connection to rivers, heavy pumping has caused only localized 25 -t> "•• *—.:.—r=-t!l Y A N* O K~ *A !••• ttoni U.S. O«ol«tlcil ltit< »••• ««D. i:soo,ooo. iaas EXPLANATION I —•00- WATER-LEVEL CONTOURS — Shows •pproiimite •itiludc «t which w»ier level would have stood in tightly cased wells Hatchured lines indicate • rea within a cone of depression Contour interval 25 feet National fleodetic Vertical Datum el 1829 Figure 3.11. Potentiometric surface for the Prairie du Chien-Jordan aquifer based on water levels measured during January-March 1971 (from Schoenberg, 1984). 26 ^72EP@£ri\;l •£ EXPLANATION WATER-LEVEL CONTOURS --Showi 100'ominiit altitude «l which • Her level would have (food in lighlly eased wellt Matchured linei indicate are* within a cone of depretnon Contour interval 25 leet National Oeodetle Vertical Datum e( 1*28 Figure 3.12. Potentiometric surface for the Prairie du Chien-Jordan aquifer based on water levels measured during January- February 1980 (from Schoenberg, 1984). 27 i4,y«.Vn«or)vv-'! 7&^*^r3&£&\1 _~~> /^* J. %••••• "^" JWtlMDIPOl.l35>',,. Wtf^lSfnWmfl! ' »«-'#.«T ^^*^ »i Bt3TJh"'vJJv*r'i\vv4 ,. ,Y-->>.?ur ^.ej'x^^i V \?^^^j cP/i r£~- \i?^^-^~>-£jtL«" \ \ ^^ EXPLANATION WATER-LEVEL CONTOURS — Sfcowt looroiimitc ittilude II which wtter level would h««» (food in tightly c«t«d wtlli Hatchurtd lints indictt* tr«t within t Cont Ol dtprcttion Contour interval It l«»t National Oeodttic Vertical Datum of 1*29 Figure 3.13. Potentiometn'c surface for the Prairie du Chien-Jordan aquifer based on water levels measured during August 1980 (from Schoenberg, 1984). 28 cones of depression in the potentiometric surface of this aquifer (see Figures 3.11 to 3.13). In Hennepin County, the greatest drawdown (approximately 50 feet) has occurred in the downtown Minneapolis and St. Louis Park areas (Stark and Hult, 1984). By comparing water levels measured in the winters of 1971 and 1980 (Figures 3.11 and 3.12), it may be noted that average water levels changed less than 5 feet over most of the area, but rose or declined as much as 25 feet locally in response to pumpage and recharge. During this period, annual pumpage from the Prairie du Chien-Jordan aquifer varied little as shown in Table 3.3. An estimate of seasonal variations in the Prairie du Chien-Jordan aquifer may be obtained by comparing 1980 winter water levels (Figure 3.12) with 1980 summer water levels (Figure 3.13) and by examination of the hydrograph in Figure 3.14. This figure shows the water level in a well located in east-central Hennepin County. During this period, seasonal declines of water levels from winter to summer lessened, and the area where these declines exceeded 10 feet decreased (Schoenberg, 1984). 3.2.1.4 St. Peter Aquifer According to Sunde (1974), the St. Peter aquifer is the source of water for many of the smaller and older wells in the St. Louis Park area. A potentiometric surface for the St. Peter aquifer is shown in Figure 3.15. As may be seen, part of the aquifer has been eroded away. The highest water levels (approximately 900 feet above sea level) occur toward the west and northeast, where recharge occurs. Recharge also occurs as leakage through the overlying Glenwood Shale. Flow is 29 Table 3.3. Annual Daily Pumpage From the Prairie du Chi en-Jordan Aquifer, 1970-1979 (From Schoenberg, 1984) Year Pumpage (MG/D) 1970 148.3 1971 145.2 1972 144.5 1973 158.4 1974 160.4 1975 153.2 1976 159.9 1977 156.6 1978 152.3 1979 148.5 30 UoJ < u. cc D CO 60 o z 70 80 IU 90 CD Lu too tu u. -r 110 Well number 117N21W16CCA01 UJ > J1971 1972 1973 1974 1975 1976 1977J1978 1979J1980 UJ -I cc UJ Figure 3.14. Hydrograph showing water level changes in the Prairie du Chien-Jordan aquifer in well 117N21W16CCA01 in east-central Hennepin County, MN (from Schoenberg, 1984). Well location shown in Figures 3.11 to 3.13. 31 RAMSEY COL, \ MINNESOTA wells MILES valley boundaries Figure 3.15. Potentiometric surface for the St. Peter aquifer based on water levels measured during the winter of 1970-1971 (from Norvitch et al., 1974). 32 generally toward the Mississippi River, with water levels that are below 800 feet above sea level. The west to east regional gradient varies from about 5 to 15 ft/mi. J Deviations from the regional flow pattern may occur where buried bedrock valleys cut through the St. Peter aquifer. Such conditions occur east of the site and at a buried tributary bedrock valley to the south of the site. Water-level measurements presented by Hult and Schoenberg (1981) indicate a northerly head gradient from the buried valley toward multiaquifer wells located near the plant site. Also contributing to the northerly flow is pumpage north of the site. 3.2.1.5 Drift-Platteville Aquifers According to Ehrlich et al. (1982), the Platteville Limestone and the glacial drift are hydrologically connectecL^EoWtwing Ehrlich et al., the detailed stratigraphy of the drift at 'St. Louis Park is Complex, but can be divided into three units: (1) a lower unit of till, outwash, valley fill deposits, and deeply weathered bedrock, (2) a middle unit of glacial sand, and (3) an upper unit of lake deposits and till. Water in the middle unit discharges laterally to the east and southeast and vertically into the lower unit (Ehrlich et al., 1982). Head differences indicate a downward vertical leakage from the middle unit to the Platteville Limestone. The regional gradient of the potentiometric surface of the Platteville aquifer is similar to that of the middle unit of the drift. A generalized water table configuration is shown in Figure 3.16. As may be seen, the hydraulic gradient is approximately 15 ft/mi tending in a southeasterly direction toward the Mississippi and Minnesota Rivers. 33 Contour Conversion m ft 274 900 271 890 268 880 265 870 Figure 3.16. Generalized potentiometric surface for the drift aquifer based on water levels measured on June 5, 1979 (from Ehrlich et al., 1982). Contours are in meters. 34 3.2.1.6 Vertical Gradients The primary focus of the sections on the various aquifers was horizontal flow and gradients. Vertical flow also occurs afrid is important in terms of solute migration between aquifers. In this section, vertical gradients for all the aquifers are very briefly discussed. The vertical gradient throughout much of the area, excluding discharge points such as major rivers, is downward. JJmjmxJuces downward leakage and the downward migration of any contaminants jthat enter the ground-water system. This is especially true in areas where confining beds are missing. Near the site, the difference in elevation between the water table and the potentiometric surface in the St. Peter aquifer is about 10 feet (Sunde, 1974). Larson-Higdem et. al. (1975) have estimated that downward leakage to the Prairie du Chien-Jordan aquifer in Hennepin County is about 3.5 in/yr. According to Sunde (1974), the hydraulic head difference between the Prairie du Chien- Jordan and the St. Peter aquifers is about 60 feet. 3.2.2 Flow Properties The flow properties for all the units discussed in Section 3.1.1 are summarized in Table 3.4. According to Stark and Hult (1984), the dis- tribution of aquifers and confining beds within the drift is poorly known outside the Reilly Tar site. Results of one aquifer test near the site are shown in Table 3.4, as well as the value used by Guswa et al. (1982) in their regional modeling study. No estimates of storage were found. Hydraulic conductivity in the Platteville aquifer is primarily the result of fractures, open joints, and solution channels. Results from one aquifer test near the site are presented in Table 3.4. The computed transmissivity was 9,000 ft2/d (Stark and Hult, 1984). 35 Table 3.4. Flow Properties of the Various Units Beneath the Reilly Tar Site Unit Transmissive Property Storage Property Reference / Comment Drift T = 5000 ft2/d Guswa et al. (1982) / Modeling Aquifer T = 9000 ft2/d Stark and Hult (9184) / Aquifer test Platteville T = 9000 ft2/d Stark and Hult (1984) / Aquifer test Aquifer Glenwood L = 5 x 10"6 (ft/d)/ft Guswa et al . (1982) / Modeling Confining Bed KV = lO'is ft/d Stark and Hult (1984) / Lab analysis CO T = 6000 ft2/d Guswa et al . (1982) / Modeling St. Peter T = 5000 ft2/d Norvitch et al . (1974) / Aquifer test Aquifer K = 12.5 ft/d 4 = 0.28 Norvitch et al. (1974) / Lab analysis K = 10.9 ft/d Basal St. Peter L = 2 x 10'5 (ft/d)/ft Guswa et al. (1982) / Modeling Confining Bed T = 12,000 ft2/d Guswa et al. (1982) / Modeling T = 11,000 ft2/d S = 4.0 x 10~" Norvitch et al. (1974)/ Aquifer tests Prairie du Chien- Prairie du Chien Jordan Aquifer T = 6800 ft2/d S 1.3 x 10~" Norvitch et al. (1974) / Aquifer tests 0.06 Norvitch et al. (1974) / Lab analysis Jordan T - 5900 ftz/d S 7.2 x ID'S Norvitch et al. (1974) / Aquifer tests * 0.32 Norvitch et al. (1974) / Lab analysis -- Continued on next page -- Table 3.4. (Continued) Unit Transmissive Property Storage Property Reference / Comment St. Lawrence- Franconia L = 1 x 10'5 (ft/d)/ft Guswa et al. (1982) / Modeling Confining Bed Ironton-Galesville T = 50 ftz/d Guswa et al. (1982) / Modeling Aquifer K = 0.35 ft/d = 0.25 Norvitch et al. (1974) / Lab analysis K = 0.24 ft/d Eau Claire L = 2 x 10"5 (ft/d)/ft Guswa et al. (1982) / Modeling Confining Bed C•-OJ T = 3000 ft2/d Guswa et al. (1982) / Modeling T = 2600 ftz/d S = 2.8 x 10~3 Norvitch et al. (1974)/ Aquifer tests Mount Simon- Mount Simon Hinckley K = 3.22 ft/d T = transmissivity K = hydraulic conductivity = vertical K IS = leakance S - storage coefficient <)> = porosity The Glenwood shale has a very low hydraulic conductivity. Stark and Hult (1984) provide a vertical hydraulic conductivity estimate of 10" ft/d based on laboratory measurements of core samples. Based on regional modeling, Guswa et al. (1982) determined a leakance value (vertical hydraulic conductivity divided by confining bed thickness) of 5 x 10"6 (ft/d)/ft. As may be seen in Table 3.4, transmissivity values for the St. Peter 2 aquifer range from 6,000 to 5,000 ft /d. In addition, Norvitch et al. (1974) provide results from laboratory analysis for median horizontal and vertical hydraulic conductivity, as well as a porosity estimate of p\ 0.28. For the basal St. Peter, the only value was a leakanice of 2 x 10" (ft/d)/ft provided from the regional modeling of Guswa et al. (1982). The Prairie du Chien-Jordan aquifer has several values of storage 2 and transmissivity. Regional modeling yielded a value of 12,000 ft /d, 2 whereas the average value from several aquifer tests was T = 11,000 ft /d and S = 4.0 x 10- 4. In addition, because of the different rock types, Norvitch et al. (1974) provide separate T and S values for the Prairie du Chien Group and the Jordan Sandstone. They also provide from laboratory analysis average porosity values of 0.06 and 0.32 for the Prairie du Chien Group and Jordan Sandstone, respectively. The only value for the St. Lawrence-Franconia confining bed was a leakance value of 1 x 10" (ft/d)/ft. This was determined through regional modeling (Guswa et al., 1982). For the Ironton-Galesville aquifer, Guswa et al. (1982) used a o transmissivity of 50 ft /d. Laboratory analysis yielded a porosity of 0.25 and estimates of median horizontal and vertical hydraulic conductivity (see Table 3.4). 38 The only value for the Eau Claire confining bed was a leakance value of 2 x 10" (ft/d)/ft. Again, this value was used in the regional modeling of Guswa et al. (1982). Finally, several values of transmissive and storage properties are provided for the Mount Simon-Hinckley aquifer. Average values from aquifer tests are provided by Norvitch et al. (1974). In addition, they also provide median values from laboratory analysis for hydraulic conductivity and porosity. All values are given in Table 3.4. In addition to the transmissive and storage properties of the aquifers, another parameter that is needed to characterize the system is recharge. Norvitch et al. (1974) give an estimate of average annual precipitation of 28.3 inches and an average annual evapotranspiration of 22.5 inches. This leaves a possible 5.8 inches for recharge. 3.3 CHEMISTRY The chemistry of the site is discussed by several authors. Most of the information in this section has been summarized from Ehrlich et al. (1982) and Hult and Schoenberg (1984). The major contaminant in ground water in the vicinity of the former plant site is creosote by-products, which is a complex mixture of chemical compounds. It is typically composed of about 85% PAH, such as naphthalene, anathracene, and phenanthrene, and 2-17% phenolics such as phenol and methylated phenols. The remainder consists of various nitrogen and sulfur-containing heterocyclic compounds. Phenolic compounds are generally more soluble in water than PAH. For example, the solubility of phenol is more than 10 g/1 at 25°C and pH 7.0, while the solubility of naphthalene under the same conditions is 39 only 0.032 g/1. When creosote is mixed with water, two phases generally emerge—a lighter aqueous phase enriched in phenolics and a denser hydrocarbon phase enriched in PAH (Ehrlich et al., 1982). According to Hult and Schoenberg (1984), coal-tar derivatives have entered the ground-water system through three major paths. Contamination of the Drift aquifer resulted from (1) infiltration of spills and drippings on the site itself, and (2) recharge from ponds south of the site that received surface runoff and contaminated process water. Contamination of the Prairie du Chien-Jordan and possibly deeper aquifers has resulted in part from (3) coal tar that entered a deep well on the site. Continuing from Hult and Schoenberg (1984), contaminants in the Drift aquifer are moving laterally to the east and southeast and vertically into the Platteville aquifer, which directly underlies the Drift aquifer in much of the affected area. 0_n__and immediately south of the site, a hydrocarbon fluid phase is moving vertically downward with respect to the aqueous phase. Vertical movement from the Platteville aquifer to the underlying St. Peter aquifer is restricted by the Glenwood confining bed, which separates the two aquifers. The Platteville aquifer and Glenwood confining bed have been removed by erosion to the south and east, and the St. Peter aquifer underlies the drift. Where the Glenwood confining bed is eroded, contaminated water has the potential to move from the Drift-Platteville aquifer system into the St. Peter aquifer. Multiaquifer wells and the buried valleys strongly influence the direction of ground-water flow in the Platteville and, therefore, the direction of contaminant transport. 40 Contaminants in the Drift and Platteville aquifers have moved a jninimum of 4,000 feet to the east. Immediately south of this area, the Platteville aquifer and Glenwood confining bed have been eroded. Fluid containing approximately 2 mg/1 of organic contaminants may be entering the underlying St. Peter aquifer and valley-fill materials. Contamination in the major bedrock aquifer of the Twin Cities area, the Prairie du Chien-Jordan aquifer, reached a well 3,500 feet from the plant as early as 1932. Contaminants can move fairly rapidly through the Prairie du Chien-Jordan aquifer because the upper part of this aquifer is a solution-channel carbonate rock of high transmissivity and low effective porosity. The regional gradient in the Prairie du Chien-Jordan aquifer is to the east, but locally the direction of ground-water flow is affected by pumping from municipal and industrial wells and multiaquifer wells through which water flows into the Prairie du Chien. Because the rate and location of pumping is continually changing, the concentration of contaminants reaching individual wells fluctuates. In 1978, coal-tar derivatives were found to the north of the site in four municipal wells completed in the Prairie du Chien-Jordan aquifer. The most remote of these wells is approximately 2 miles from the nearest probable source of contaminants to the aquifer, well W23 on the site. The northward direction of flow is in agreement with water levels measured in August 1977. Since November 1978, pumping patterns have been altered significantly by shutting down four of the wells that were found to be contaminanted. Consequently, the direction and rate of contaminant transport may have been altered. 41 4.0 GROUND-WATER MODELING Simulation of a ground-water system refers to the construction and operation of a model whose behavior assumes the appearance of the actual aquifer behavior. Models used at the Reilly Tar site include analytical and numerical models, both of which are based on mathematical equations that describe fluid flow through porous media. The models were used to reproduce flow conditions at the Reilly Tar site. After these conditions are adequately reproduced, the models may then be used to estimate future hydraulic gradients produced by the introduction of source and gradient control wells. That is, the models can be used, within the limits of the uncertainty' in estimated subsurface parameters and discharge rates, to evaluatethe effectiveness of proposed remediation. Several ground-water modeling studies have been performed that include the area affected by the Reilly Tar site. These studies include Barr (1977), Hickok (1981), Guswa et al. (1982), ERT (1983), and Stark and Hult (1984). All of these studies were limited to ground-water flow; that is, transport processes (other than convection) were not considered. This section concentrates on the modeling of ERT (1983) and Stark and Hult (1984) and emphasizes those aspects that relate to remediation. 4.1 CODE SELECTED The code used in both ERT (1983) and Stark and Hult (1984) was the U.S. Geological Survey three-dimensional flow model described in Trescott (1975), Trescott and Larson (1976), and Torak (1982). The model is based on a finite difference approximation and can be used in either a fully-three-dimensional mode or a quasi-three-dimensional mode. 42 Table 4.2. Vertical Aquifer Geometry Used in the U.S.G.S. Model Layer Approximate Number Unit Thickness (ft) 4 St. Peter Sandstone Aquifer 135 (with some Drift) 3 Basal St. Peter Confining Bed 30 2 Prairie du Chien Group 125 1 Jordan Sandstone 80 46 4.2.2 Boundary Conditions In order to obtain a unique solution of the equation describing ground-water flow, additional information about the flow process is required. This information includes boundary conditions, which consists of the geometry of the boundary and a description of hydraulic head on the boundary. In both models, the Minnesota and Mississippi Rivers are treated with constant hydraulic head boundary conditions, where the U.S.G.S. fixes their boundary heads at different values depending on the time period being simulated. For the ERT model, along the western and northern boundaries, no-flow conditions are employed for the Prairie du Chien-Jordan and Drift-Platteville aquifers, while flow boundaries are specified for the underlying layers (ERT, 1983). For the U.S.G.S. model, the north and west boundaries are treated as constant hydraulic head. The bottom boundaries in both models are treated as no-flow. Keep in mind that the bottom of the ERT model corresponds to the contact of the Mount Simon-Hinckley aquifer with Precambrian rock, and the bottom of the U.S.G.S. model corresponds to the contact of the Jordan Sandstone with the St. Lawrence-Franconia confining bed. Both models also have a specified flux as the boundary condition at the top boundary, which represents either recharge and/or leakage. The ERT value was 7.5 in/yr, whereas the U.S.G.S. used 5.5 in/yr. 4.3 HISTORY MATCHING Before using a model in a predictive mode (in this case, to simulate proposed remedial actions), it is a common practice to calibrate the model by reproducing observed data, that is, history 47 matching. The U.S.G.S. model was calibrated using two steady-state conditions and a transient condition. According to Stark and Hult (1984), the steady-state phases simulate (1) conditions prior to significant ground-water development (approximately 1885-1930), and (2) average winter conditions in the system during 1970 through 1977, a period of large annual ground-water withdrawal. A transient simulation of seasonally variable ground-water withdrawal from 1977 to 1980 for which changes in potentiometric surfaces with time were documented was used for transient calibration. The ERT jnode1_was calibrated using jane .jet of steady^sjtate^data. The water level data used are a composite, but emphasize the winter 1970-1971 data in Norvitch et al. (1974). 4.3.1 Wells The first steady-state simulation of the U.S.G.S. model had no wells. The second simulation used average annual ground-water pumpage for 1970 through 1977. The simulated pumpage was 22.7 MG/D from 121 wells, primarily from the Prairie du Chien-Jordan aquifer. Note that this is considerably less than the amount indicated in Table 3,3 because a smaller area is modeled. For the transient simulations, each year was divided into three pumping seasons: (1) Spring (January-April), (2) Summer (May-September), and (3) Fall (October-December). Summer pumpages averaged about 33 MG/D, and were about 1.7 times the average spring and fall rates. The simulated pumpage in the ERT model amounted to 32.3 MG/D. Data regarding local pumping centers was obtained from Hult and Schoenberg (1981), Guswa et al. (1982), and Minnesota Department of Natural Resources (1982). 48 4.3.2 Flow Parameters The flow parameters determined by the _U._S.G_.S^after model calibration are shown in Table 4.3. Comparison of these parameters to those in Table 3.4 yields the following observations: (1) the trans- mi ssivity of the St. Peter aquifer is about half that of previously determined estimates in Table 3.4, (2) the leakance of the basal St. Peter confining bed is 3 x 10 (ft/d)ft, which is very similar to that determined by Guswa et al. (1982), (3) the transmissivity and storage coefficient of the Prairie du Chien Group is similar to previously determined values, and (4) the storage coefficient of the Jordan Sandstone is similar to previously determined values, but the transmissivity is about one third of earlier estimates. The flow parameters determined by the ERT modej calibration are shownjinTab^e 4.4. Comparison of these values with those in Jable_3_.4_ \l shows that the ERT values of transmissivity are consistently less by about one half of earlier estimated values. The leakances are also less, whereas the recharge is higher than that estimated from Norvitch et al. (1974). In addition to the parameters in Table 4.4, ERT used the parameters in Table 4.5 to estimate travel times. Comparison of these values to the values in Table 3.4 shows similar values except for the Prairie du Chien Group, which has a low porosity of 0.06. The value_of_porqsUy used by ERT will tend to overestimate travel times for contaminants moving in the Prairie du Chien Group. 4.3.3 Results The results are divided into two sections. One section briefly presents how well the calibration process was conducted, and the other section presents what was determined about the flow system. 49 Table 4.3. Values of U.S.G.S. Hydrologic Properties (After Stark and Hult, 1984) Vertical Hydraulic Transmissivity Conductivity Leakage Storage Unit (ft2/d) (ft/d) (in/yr) Coefficient 4 Drift -- 0.007-0.3 io- 4 St. Peter 2,700 0.3 5.5 io- 5 en 4 o Basal St. Peter 600 9 x 10" io- Prairie du Chi en 5,900-7,000 65 4 x IO"4 Jordan 1,600-2,000 2.0 7 x IO"5 Table 4.4. Values of ERT Model Hydrologic Properties (After ERT, 1983) Transmissivity Recharge Leakance Unit (ft2/d) (in/yr) (ft/d)/ft Drift-Platteville K = 65 ft/d Glenwood 1.4 x 10 | St. Peter 2800 7.5 Basal St. Peter 1.0 x 10"5 i ! Prairie du Chien-Jordan 4500-7500* St. Lawrence-Franconia 6.0 x 10" ! Ironton-Galesville 20 Eau Claire 8.6 x 10"8 I 1 Mount Simon-Hinckley 870 * Beneath site T = 6000 ftz/d 51 Table 4.5. Values Used by ERT to Compute Travel Times (From ERT, 1983) Hydraulic Conductivity Unit Porosity (ft/d) Drift-Platteville 0.30 65 St. Peter 0.28 12 Prairie du Chien-Jordan 0.15 37 Ironton-Galesville 0.25 1 Mount Simon-Hinckley 0.22 8 52 The calibration process conducted by the U.S.G.S. yielded computed hydraulic heads that were generally within 10 feet of measured water levels for all three simulations. Figure 4.2 provides an example of how well the ERT calibration was performed. It shows a comparison of the calculated and observed potentiometric surface for the Prairie du Chien- Jordan aquifer; comparisons with other potentiometric surfaces are not reproduced here. It should be noted that both modeling studies were limited by available data, including detailed pumpage information. Assuming that the calibration for both models is sufficient, the following conclusions were made. Concerning the Drift-Platteville aquifer, with the exception of the area influenced by Lake Minnetonka, transport is predicted to be directed to the east and south at approximate velocities of 0.5 ft/d (ERT, 1983). The ultimate destinations of contaminants are (1) the rivers, (2) any wells in the vicinity, or (3) deeper aquifers that are hydraulically connected with the Drift-Platteville aquifer through buried bedrock valleys. According to ERT (1983), computer results indicate that a portion of the flow in the Drift-Platteville aquifer above the buried bedrock valley is intercepted and flows vertically downward in the valley to the St. Peter aquifer. In general, flow predictions in the St. Peter aquifer indicate travel to the eastern and southern river boundaries at velocities on the order of l_to 2 ft/d (ERT, 1983). From starting points _north_of_the bedrock valley, flow tends_to .be eastward, while south of the valley, flow is more to the southeast. Although muHlaqurFer wells in the St. Peter aquifer may serve as conduits of contaminants to the St. Peter, the effect that these wells have on the potentiometric surface of the St. Peter aquifer is unclear. 53 IB.UO 10 c. -J ro Ib.VO -h O o o !*.•• -l 3 •o «-»• o> 3" -J n o Ol !«.*• I Ul»- cr i«Nce IN o -i n> ro HlLfS en 3 CL I c_, m 0 » -j —I a. cu a 3« O Q. 01 ro £> —« c -•• -I -ti ro ro m -i c i3 -•*• 2.HI -H O • cr Minnesota River 1/1 H H H H M H H H h H It H-« tO VO I CO < .*0 co ro . , c\ .0 2.0U «.UO b.OO d.OO IU.OU 12.OU 14.00 I*.00 18.00 20.00 22.00 2*.00 o. tu OIS1ANCL FHUM UMlblH IN 1 OlhECTlUN, III HILtS c-f Oi EXPLANATION Obterved (Hull end Schoenbarg 1981 and Norviich •! •!. 1973) Numbera ahow computed hydraulic head Modal-Computed in 10'a of leet. n.g.v.d. River or Lake Boundary For the Prairie du Chien-Jordan aquifer, Stark and Hult (1984) show that near the plant site, summer water levels are about 20 to 25 feet lower than winter water levels. However, gradients along streamlines average about 0.0025 through the plant site and do not vary significantly from summer to winter. Ground water flowing through the plant site generally moves to the south-southeast. The velocity computed by Stark and Hult (1984) through the plant site area in the Prairie du Chien is about 3 ft/d, assuming a gradient of 0.0025, a hydraulic conductivity of 65 ft/d, and a porosity of 0.06. The presence of contaminants in the Prairie du Chien-Jordan in wells to the north, west, and southwest of the plant site may be explained by changes in pumping stresses. According to Stark and Hult (1984), local hydraulic gradients may have been altered during periods of heavy pumping, creating the potential for transport of contaminants from the plant site toward the north. Hult and Schoenberg (1981) present data showing that multiaquifer flow in wells significantly influenced the potentiometric surface of the Prairie du Chien-Jordan aquifer near the plant site. These data indicate that the water levels in well W23 (see Figure 4.3 for location) were higher than water levels in all nearby surrounding wells in the aquifer, and that this cone of impression was created by water moving into the Prairie du Chien-Jordan aquifer from the overlying St. Peter aquifer through the bore of the well. The presence of this cone of impression, coupled with variable pumping at surrounding wells, is thought to have been sufficient to induce flow in all directions from well W23. ^ ,' | f _ j ^-+^ \] 55 93*22'30 83»20' 44»57'30 J_ \- i '.. . - .. ..v-^ - ' ."r• B^i£''~--:-.-•• _*^*^ (.•••• • •j <-." '•o/*"--. :.•:-..: -i:-5=.-- .j. i-.-::.^'. ..^—' - . ' /•/ -.j, - -. •:•*- - -I* -~~ :-.^X ;:-;--;''"^yt7^"^B|'L^'-';:..=_± ;_ . ;.;-..- '^•-J;'" ' - '••-''», '.'."''.' ' \ '•• ; t;,- -i. :. """" Lwis PAR : .*".'f'". _ _/L ^^?2 iLwi?3» ••'••,„... '^v;.^ WTO i '•'• •-''••• ' '' '•"'' - •' ' •*! •>> "i * ^ • ''"" •. ; • i'\:* »;~.~-~* -'• W22 \ •-" -?y- .-... ; x • i^59 .^W34W4, ^l-. "" ^^^h^sl*^27 ^-w^xv*3 7 5*^* -»»ll°-«•••r-- ?^s^SS^Sfifev 2. - - - W5B W57 .tigfi^5^yV4VW4^" i!i ^2 >J^% \WA ' .*^ '/ 1^^Siw^KT!^ds-*^H- . -^ r~"~r^ -* •.;::--. .= , -s- '- W72 -r-Tr-'TTn1 44*55- i-^. PRIVATE AND INDUSTRIAL WELLS EXPLANATION LOCATION OF WELL OR PIEZOMETER "P" Prefix denotes small-diameter piezometer • "W" Prefix denotes 4-inch diameter or larger W5.P25 we|| Numbers indicate project well or piezometer number | I Site of former plant Figure 4.3. Map showing the location of private and industrial wells in the vicinity of the Reilly Tar site (from Hult and Schoenberg, 1934) 56 ERT (1983) uses an analytical model to simulate the effects of production and multiaquifer wells on the potentiometric surface of the Prairie du Chien-Jordan aquifer near the plant site. The potentiometric high created by these wells was found to be sufficient to induce flow from the plant site toward any of the nearby municipal wells if each of the wells in a particular area was pumped at full capacity to equilibrium. Computer simulations also were used to estimate contaminant pathways in the Mount Simon-Hinckley aquifer (ERT, 1983). Ground-water movement on a regional scale was found to be dominated by pumping centers, with typical velocities on the order of 0.1 ft/d toward the pumping wells. 4.4 PREDICTIVE SIMULATIONS WITH REMEDIATION ERT (1983) did not evaluate any remediation using their model. The U.^G.S^ simulated a plan that consisted of pumping well W23 (source control) and several other wells (gradient control) as shown in Table 4.6. Transient simulations were conducted using the spring and summer pumpages indicated in Table 4.6. According to Stark and Hult (1984), the simulated impact on the potentiometric surface of the Prairie du Chien-Jordan, resulting from a proposed gradient control plan using five discharge wells, suggests that the plan will be effective in limiting the expansion of the contaminated volume. The simulations also indicate that manipulation of withdrawal from nearby wells not controlled under the remedial action plan could alter the effectiveness of the gradient control measures. Evaluation of proposed remediation using the U.S.G.S. model is an on-going process being conducted by the Minnesota Pollution Control Agency (MPCA). Although the well extraction design has not been 57 Table 4.6. Withdrawal Rates for Remedial Action Simulations (From Stark and Hult, 1984) Pumping Rate (gal/min) Wei 1 Name Spring 1980 Summer 1980 SLP 5 167 320 SLP 6 1,100 660 SLP 7 40* 40* SLP 8 570 910 SLP 9 93* 93* SLP 10*** 188* 188* SLP 14 160 260 SLP 15*** 410* 410* SLP 16 400 650 Flame Industries 26 30 Hopkins 3 300 485 Methodist Hospital 162 785 Minnesota Rubber 146 155 McCourtney Plastics 256 194 Food Producers 20 30 W23** 50 50 Proposed well*** 750 750 SLP 4*** 750 750 Based on 1980 seasonal water use data unless otherwise noted * 1970-1977 mean annual rate ** Planned as source control well *** Planned as gradient control well 58 finalized and probably will not be until it has been field tested, work by the U.S.G.S. and MPCA indicate that it is feasible to use a few wells to control the hydraulic gradients in the aquifers beneath the Reilly Tar site. That is, the wells have the capacity and the aquifer properties are such that a small number of wells can control the hydraulic gradients. These extraction wells include some existing we11s. 59 5.0 REMEDIAL ACTIONS r I Remedial actions can be evaluated using results from earlier investigations, as previously reviewed. Final remedial action selection I and design will require further analysis, some of which as indicated is on-going. In this section, remedial actions are proposed for all ir aquifers, as well as some surface considerations. Subsections, in ; general, cover individual aquifers. As additional information is gained, these proposed remedial actions should be modified and refined. { 5.1 MULTIAQUIFER WELLS ^ Following Hult and Schoenberg (1984), a multiaquifer well is defined as any well that hydraulically connects more than one aquifer. The connection may be due to original open-hole construction or to deterioration of casing or grout seal. The significance of multiaquifer wells has been established by several studies including ERT (1983) to be i (1) a recharge mechanism providing a pathway for coal-tar derivatives to enter deep aquifers, and (2) a recharge mechanism for ground water contributing to a potentiometric high in the Prairie du Chien-Jordan aquifer that induces flow away from the plant site. i 5.1.1 Investigation Because of the significance of multiaquifer wells, further investigation similar to that described in Hult and Schoenberg (1984) ; should be conducted. This investigation should include but not be V . limited to (1) locating all multiaquifer wells, which may allow i contaminant transfer between aquifers, (2) logging the wells with downhole television camera and/or downhole geophysical tools, and 60 (3) additional monitoring. Such investigations should be performed on suspected multiaquifer wells within the area of contamination in aquifers that may discharge to the wells. 5.1.2 Remedy Multiaquifer wells that are no longer useful should be permanently sealed with grout; those that are useful should be cased appropriately. I 5.2 NEAR-SURFACE CONTAMINATION The full extent of near-surface contamination needs to be determined. This is especially important for the drainage area south of the plant site including the area which was formerly a tributary to Minnehaha Creek. 5.2.1 Deed Restrictions In order to protect future public health, owners and future owners of land impacted by runoff and waste discharge from the Reilly Tar site should be notified, and appropriate deed restrictions, such as limiting excavation, should be implemented. 5.2.2 Louisiana Avenue/Highway 7 Intersection Construction Any dewatering or soil removal from this area should be considered a potential hazardous waste removal action, and should be treated as such. This should include proper protection of construction workers and proper disposal of water and/or soil removed from this location. 61 5.3 DRIFT AQUIFER ! For the purpose of this section, the Drift aquifer is considered by i U.S. EPA and the state to be contaminated if the concentration of total I { PAH exceeds one microgram per liter or the concentration of phenolics r exceeds ten micrograms per liter. 5.3.1 Source Control Source control measures that should be considered include (1) soil removal, (2) capping, and/or (3) pumping and treating. The source areas in the Drift aquifer are the swamp area south of the site and the contaminated soils beneath the site. Pumping pertains primarily to the middle drift unit. 5.3.2 Gradient Control The extent of contamination in the Drift aquifer, as defined above, needs to be investigated further. In addition to source control, to limit further outward and downward migration, gradient control wells should be utilized. The exact location, design, pumping rates, and treatment should be specified after additional studies have been made. The location of these wells should be determined, in part, by the extent of contamination. The purpose of these wells is to maintain an inward hydraulic gradient toward the source. 5.3.3 Monitoring To insure that the source and gradient controls are functioning properly, monitoring of wells adjacent to the controls is required. 62 Water-quality monitoring and water-level measurement should begin for a ' period of time prior to initiation of the source/gradient controls in order to provide background data. i i 1 5.3.4 Contingencies ir In the event that monitoring indicates that the source/gradient controls are not working, that is, the extent of contamination is not contained with inward hydraulic gradients toward the source, then ,' additional source/gradient controls need to be implemented. (' ' 5.3.5 Mitigated Impacts The impacts that are mitigated by these measures are the further i spread of contamination through the Drift aquifer, further downward transport of contaminants to deeper aquifers via bedrock valleys or multiaquifer wells, and discharge to local surface water bodies. i i 5.4 PLATTEVILLE AQUIFER I Contamination in the Platteville aquifer is defined as in Section 5.3 for the Drift aquifer. i 5.4.1 Source Control The extent of contamination in the Platteville aquifer needs to be i investigated further. Source control measures that should be considered , are pumping and treating. i 5.4.2 Gradient Control Gradient control considerations for the Platteville aquifer should be the same as those described in Section 5.3.2. i 63 5.4.3 Monitoring Monitoring considerations for the Platteville aquifer should be the same as those described in Section 5.3.3. 5.4.4 Contingencies Contingency considerations for the Platteville aquifer should be the same as those described in Section 5.3.4. 5.4.5 Mitigated Impacts The impacts that are mitigated by these measures are the further spread of contamination within the Platteville aquifer and further migration of downward transport of contamination to deeper aquifers, especially at buried bedrock valleys and through multiaquifer wells. 5.5 ST. PETER AQUIFER Contamination in the St. Peter aquifer is defined as PAH concentrations that exceed the potable criteria defined in Section 5.6.4. 5.5.1 Source Control The extent of contamination in the St. Peter aquifer needs to be established and used to determine if source control is necessary. If it is necessary, then pumping and treatment will be required. 5.5.2 Gradient Control If gradient control is also deemed necessary, then considerations should be similar to those outlined in Section 5.3.2. 64 5.5.3 Monitoring New monitoring wells should be installed in the St. Peter aquifer. Water-quality monitoring and water-level measurements should be made in these wells and in other existing wells on the same dates. Mpjvrtprijig sjiould contjnue at SLP3 and_atJeast six other St. Peter wells semi annual ly jFor at least five years. This monitoring will be used for initial assessment of contamination in the St. Peter aquifer. Additional and more frequent monitoring will be required if gradient and/or source controls are deemed necessary. 5.5.4 Contingencies In the event monitoring detects PAH concentrations greater than drinking water critria, defined in Section 5.6.4, in the St. Peter aquifer, a source/gradient control system will have to be implemented. 5.5.5 Mitigated Impacts Assuming multiaquifer wells have been sealed, the impacts that are mitigated by these measures are the further spread of contamination within the St. Peter aquifer. 5.6 PRAIRIE DU CHIEN-JORDAN AQUIFER 5.6.1 Source Control Well W23 should be pumped as a source control well. Pumping should begin immediately to contain the high-level contamination observed in the aquifer in the vicinity of the former plant site. 65 5.6.2 Gradient Control Gradient control is required to prevent further migration of contaminants within the Prairie du Chien-Jordan aquifer. The gradient control needs to be flexible, depending on how much institutional controls there are over existing private and industrial pumpage. For that reason, several wells should be considered for use. A plan that appears to work has been described in Section 4.4. An inward hydraulic gradient toward the gradient control wells should be maintained in the aquifer. 5.6.3 Monitoring Because of the variability in hydraulic gradient within the Prairie du Chien-Jordan aquifer, monitoring and water-level measurement are very important. Several wells should be used as water level and water-quality monitoring points to assess contaminant distribution changes and gradient control system effectiveness. Monitoring frequency will depend on the use of the well (e.g., potable water supply, industrial, or monitoring). Additional monitoring wells may be needed to adequately monitor the aquifer. Water use must also be monitored within the area. This data and the water-quality data are essential information for assessing the gradient control system effectiveness. 5.6.4 Drinking Water For the purposes of this report, any water shall be considered to be below drinking water criteria for PAH if it contains total carcinogenic PAH at a concentration of less than 28 nanograms per liter, the sum of the concentrations of benzo(a)pyrene and dibenz(ah)anthracene 66 at less than 5.6 nanograms per liter, and total other PAH at a concentration of less than 280 nanograms per liter. The use of any supply water which exceeds any of these criteria shall be discontinued 1 until such time as the criteria are met by treatment or other means. It is assumed in this discussion of Prairie du Chien-Jordan aquifer remedial actions that St. Louis Park well number 15 will be returned to service through the use of granular activated carbon treatment. This measure is necessary for gradient control and for restoration of the City's water supply capacity. t - 5.6.5 Contingencies j Contingency considerations for the Prairie du Chien-Jordan aquifer should be the same as those described in Section 5.3.4. In addition, ' any drinking water supply wells that become contaminated (exceed drinking water standards) should be treated. I 5.6.6 Mitigated Impacts The impacts that are mitigated by these measures are the protection ! of public drinking water supplies and the further spread of contamination through the Prairie du Chien-Jordan aquifer. 5.7 IRONTON-GALESVILLE AQUIFER 5.7.1 Source Control The extent of contamination in the Ironton-Galesville aquifer needs to be investigated further. Source control measures that should be considered are pumping well W105 and treating the effluent. 67 5.7.2 Gradient Control Gradient control considerations for the Ironton-Galesville aquifer should be the same as those described in Section 5.3.2. 5.7.3 Monitoring Monitoring considerations for the Ironton-Galesville aquifer should be the same as those described in Section 5.3.3. 5.7.4 Contingencies Contingency considerations for the Ironton-Galesville aquifer should be the same as those described in Section 5.3.4. 5.7.5 Mitigated Impacts The impacts that are mitigated by these measures are limiting the further spread of contamination within the Ironton-Galesville aquifer and further downward migration of contamination to the Mount Simon- Hi nckley aquifer. 5.8 MOUNT SIMON-HINCKLEY AQUIFER 5.8.1 Source Control The extent of contamination in the Mount Simon-Hinckley aquifer needs to be investigated further. Source control measures that should be considered are pumping and treating the effluent. 5.8.2 Gradient Control Gradient control considerations for the Mount Simon-Hinckley aquifer should be the same as those described in Section 5.3.2. 68 Existing St. Louis Park municipal wells may serve this purpose depending on future water use trends. 5.8.3 Monitoring Monitoring considerations for the Mount Simon-Hinckley aquifer should be the same as those described in Section 5.3.3, and should include wells SLP11, SLP12, SLP13, and SLP17. These wells should be monitored at least annually.. 5.8.4 Contingencies Contingency considerations for the Mount Simon-Hinckley aquifer should be the same as those described in Section 5.6.5. 5.8.5 Mitigated Impacts The impacts that are mitigated by these measures are the protection of public drinking water supplies and the further spread of contamination through the Mount Simon-Hinckley aquifer. 69 6.0 REFERENCES Barr Engineering, 1977, "Soil and ground water investigation, coal tar distillation and wood preserving site, St. Louis Park, Minnesota, Phase II report." Ehrlich, G.G., D.F. Goerlitz, E.M. Godsky, and M.F. Hult, 1982, "Degradation of phenolic contaminants in ground water by anaerobic bacteria: St. Louis Park, Minnesota," Ground Water, Vol. 20, No. 6, pp. 703-710. ERT, Environmental Research & Technology, Inc., 1983, "Recommended plan for a comprehensive solution of the polynuclear aromatic hydrocarbon contamination problem in the St. Louis Park area, Volumes I-IV, Appendices A-L," Document P-B690-161. Guswa, J.H., D.I. Siegel, and D.C. Gillies, 1982, "Preliminary evaluation of the ground-water-flow system in the Twin Cities metropolitan area, Minnesota," U.S. Geological Survey Water-Resources Investigations Report 82-44. Hickok and Associates, 1981, "Final report: Study of groundwater contamination in St. Louis Park, Minnesota." Horn, M.A., 1983, "Ground-water-use trends in the Twin Cities metropolitan area, Minnesota, 1880-1980," U.S. Geological Survey Water-Resources Investigations Report 83-4033. Hult, M.F., and M.E. Schoenberg, 1981, "Preliminary evaluation of ground-water contamination by coal-tar derivatives, St. Louis Park area, Minnesota," U.S. Geological Survey Open-File Report 81-72. Hult, M.F., and M.E. Schoenberg, 1984, "Preliminary evaluation of ground-water contamination by coal-tar derivatives, St. Louis Park area, Minnesota," U.S. Geological Survey Water-Supply Paper 2211. Larson-Higdem, D., S.P. Larson, and R.F. Norvitch, 1975, "Configuration of the water table and distribution of downward leakage to the Prairie du Chien-Jordan aquifer in the Minneapolis-St. Paul metropolitan area, Minnesota," U.S. Geological Open-File Report 75-342. Minnesota Department of Natural Resources (MDNR), Division of Waters, 1982, "Monthly and annual pumpage of industrial and municipal water permit holders in Hennepin County, Minnesota, 1970-1981." Reeder, H.O., W.W. Wood, G.G. Ehrlich, and R.J. Sun, 1976, "Artificial recharge through a well in fissured carbonate rock, West St. Paul, Minnesota," U.S. Geological Survey Water-Supply Paper 2004. 70 Schoenberg, M.E., 1984, "Water levels and water-level changes in the Prairie du Chien-Jordan and Mount Simon-Hinckley Aquifers, Twin Cities metropolitan area, Minnesota, 1971-1980," U.S. Geological Survey Water-Resources Investigations Report 83-4237. Sunde, G.M., 1974, "Hydrogeologic study of the Republic Creosote site," report to City of St. Louis Park, Gerald M. Sunde, Consulting Engineer, Bloomington, Minnesota, July. Stark, J.R. and M.F. Hult, 1984, "Ground-water-flow model of the Prairie Du Chien-Jordan aquifer, St. Louis Park, Minnesota," U.S. Geological Survey Water-Resources Investigation Report 84- . Torak, L.J., 1982, "Modifications and corrections to the finite-difference model for simulation of three-dimensional ground-water flow," U.S. Geological Survey Hater Resources Investigations 82-4025. Trescott, P.C., 1975, "Documentation of finite-difference model for simulation of three-dimensional groundwater flow," U.S. Geological Survey Open-File Report 75-438. Trescott, P.C., and S.P. Larson, 1976, "Supplement to documentation of finite-difference model for simulation of three-dimensional groundwater flow," U.S. Geological Survey Open-File Report 76-591. 71 7.0 GLOSSARY OF TERMS Conservative contaminant - a solute that behaves as a perfect tracer, that is, it is not sorbed. Therefore, it is not retarded and is convected with the interstitial velocity of ground water. Hydraulic conductivity - a property of the porous medium that indicates how well the medium will transmit water. More specifically, it is defined as the volume of water that will move in unit time under a unit hydraulic gradient through a unit area measured at right angles to the direction of flow. Leakance - (also known as an interaquifer transfer coefficient, leakage coefficient, or hydraulic conductance per unit area) is equal to the hydraulic conductivity of a confining bed divided by the thickness of the bed. Porosity - a measure of interstitial space in rock or soil and is expressed as the percentage ratio of void space to the total volume of the rock or soil. Storage coefficient - the volume of water an aquifer releases from, or takes into, storage per unit surface area of aquifer per unit change in the component of head normal to that surface. Transmissivity - the rate of flow of water through a vertical strip of aquifer one unit wide, extending the full saturated thickness of the aquifer, under a unit hydraulic gradient. 72