APPLICATION OF GROUNDWATER MODELING TECHNOLOGY FOR
EVALUATION OF REMEDIAL ACTION ALTERNATIVES
WESTERN PROCESSING SITE, KENT, WASHINGTON
by
F. W. Bond
C. M. Smith
J. M. Doesburg
C. J. English
Battelle Project Management Division
Office of Hazardous Waste Management
Richland, Washington 99352
September, 1984
Prepared for
Municipal Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
CONTENTS
Disclaimer ii
Abstract iii
Acknowledgement iv
1. Introduction. 1
2. Conclusions 3
3. Description of the Study Area 5
Climate 5
Geology 5
Hydrology 8
Waste Disposal History 8
4. Model Development 13
Model Selection 13
Regional Model 14
Groundwater Flow Model Development ... 14
Structure 14
Boundary Conditions 17
Hydraulic Conductivity 17
Groundwater Potential 20
Hydraulic Stress 21
Porosity 21
Contaminant Transport Model Development 21
Contaminant Selection ....... 21
Source Location 21
Source Area Concentration 22
Source Duration/Leach Rate . . 22
Sorption/Retardation 22
5. Model Calibration 23
Flow Model Calibration 23
Transport Model Calibration 27
Base Case Results. . 31
6. Assessment of Remedial Action Alternatives 34
Assessment Approach and Results 35
No-Action 35
Source Removal 38
Cap • * * _ 40
Source Removal Combined with a Cap 42
Slurry Wall Combined with a Cap 44
Pump and Treat 47
Discussion of Results 49
7. Simplified Analytical Approach to the Western
Processing Data Set 53
Pumping Analysis ••••;••*• I * 53
Analysis of the Analytical Program Results 56
-------
FIGURES
Number Page
1 Western Processing Location Map 6
2 Smoothed Kriged Potential Surface for April, 1984 9
3 Western Processing Well and Waste Pond Locations 11
4 Western Processing Model Area 15
5 Finite Element Grid of the Western Processing Site 16
6 Cross Section Depicting the Structural Layers
of the Study Area 18
7 Location of Surveyed Values for Mill Creek and
Drainage Ditch 19
8 Model-Predicted 1983 Top of Layer 1 Potential Surface
for the Base Case Simulation . 24
9 Model-Predicted 1983 Top of Layer 2 Potential Surface
for the Base Case Simulation 24
10 Model-Predicted 1983 Top of Layer 3 Potential Surface
for the Base Case Simulation 25
11 Model-Predicted 1983 Top of Layer 4 Potential Surface
for the Base Case Simulation 25
12 Model-Predicted 1983 Bottom of Layer 4 Potential Surface
for the Base Case Simulation 26
13 Kriged TCE Concentration Contours 28
14 Model-Predicted Top of Layer 1 TCE Concentration Contours
for the Base Case Simulation 28
15 Model-Predicted Top of Layer 2 TCE Concentration Contours
for the Base Case Simulation 29
-------
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
29
30
30
33
33
36
36
37
39
41
41
43
45
46
48
48
Model-Predicted Top of Layer 3 TCE Concentration Contours
for the Base Case Simulation
Model-Predicted Top of Layer 4 TCE Concentration Contours
for the Base Case Simulation
Model-Predicted Bottom of Layer 4 TCE Concentration Contours
for the Base Case Simulation. .
Model-Predicted 1968 Top of Layer 1 TCE Concentration
Contours for the Base Case Simulation
Model-Predicted 1978 Top of Layer 1 TCE Concentration
Contours for the Base Case Simulation
Model-Predicted 1988 Top of Layer 1 TCE Concentration
Contours for the No-Action Simulation
Model-Predicted 1998 Top of Layer 1 TCE Concentration
Contours for the No-Action Simulation
Model-Predicted 2008 Top of Layer 1 TCE Concentration
Contours for the No-Action Simulation
Model-Predicted 2008 Top of Layer 1 TCE Concentration
Contours for the Source Removal Remedial Action
Model-Predicted Top of Layer 1 Potential Surface for the
Cap Remedial Simulation
Model-Predicted 2008 Top of Layer 1 TCE Concentration
Contours for the Cap Remedial Action Simulation
Model-Predicted 2008 Top of Layer 1 TCE Concetration
Contours for the Source Removal Plus Cap Remedial
Action Simulation
Model-Predicted Top of Layer 1 Potential Surface for the
Slurry Wall Plus Cap Remedial Action Simulation
Model-Predicted 2008 Top of Layer 1 TCE Concentration
Contours for the Slurry Wall Plus Cap Remedial
Simulation - • • • •
Model-Predicted Top of Layer 1 Potential Surface for the
Pump and Treat Remedial Action Simulation
Model-Predicted Top of Layer 1 TCE Concetration Contours
for the Pump and Treat Remedial Action Simulation ....
-------
32 Comparision of the Total Mass of TCE Remaining in the
Groundwater Flow System for the Five Remedial Action
Cases 50
33 Location Map of Pumping and Injection Well Used in
the Anayltical Solution 55
B-l Locations of Regional Values Used to
Determine Boundary Conditions 67
D-l Model-Predicted Water Table Surface for
Kh * 28.4 ft/day 70
D-2 Model-Predicted Water Table Surface for
Kh » 0.3 ft/day 71
D-3 Model-Predicted Water Table Surface for
Kv/Kh = 1/100 73
-------
TABLES
Number Page
1 Average Monthly Precipitation, Potential Evapotranspiration
(PET), and Actual Evapotranspiration (AET) 7
2 Comparison of Shallow and Deep Well Potentials 10
3 Calibrated Hydraulic Conductivity Values 20
4 Comparision of the Observed to Model-Predicted Maximum TCE
Concentrations in the Groundwater at the Three
Source Locations 31
5 Distribution of TCE in the Model Base Case Simulation . . 32
6 Model-Predicted Maximum Concentration in the Groundwater
at the Three Source Areas in the Year 2008 for the
Remedial Action Simulations 35
7 Model-Predicted Distribution of TCE for the No-Action
Simulations 38
8 Model-Predicted Distribution of TCE for the Source
Removal Simulations 40
9 Model-Predicted Distribution of TCE for the Cap
Remedial Action 42
10 Model-Predicted Distribution of TCE for the Source
Removal Plus Cap Remedial Simulation 44
11 Model-Predicted Distribution of TCE for the Slurry
Wall Plus Cap Remedial Simulation 47
12 Model-Predicted Distribution of TCE for the Pump and
Treat Remedial Action 49
13 Input Values Used in the Analytical Calculations 57
-------
References 59
Appendices
A. Recharge Calculations 61
B. Stream and Leakance Boundary Condicions 63
C. Calculation of Retardation Factor 68
D. Model Calibration 69
-------
A-l Runoff Program Results 62
B-l Stream Boundary Option Data Used to Simulate
Flux to Mill Creek 64
B-2 Stream Boundary Option Data Used to Simulate
Flux to the Ditch 65
-------
SECTION 1
introduction
The goal of this project was to evaluate the use of models in predicting
contaminant flow from the Western Processing Hazardous Waste Site in Kent
... 4.orhniral assistance to the U.S. Environmental
Washington, while providing technical assis
n Y This included the development of
Protection Agency EPA) Region X. mis
Protection «gei».jr v / * ^an„nr.t models of the site to be used for
groundwater flow and contaminant transp
evaluation of proposed remedial action alternatives.
The specific tasks of the modeling portion of «» deluded:
. a review of available data and Identification of deficiencies
. development of groundwater flow and contaminant transport models of the
. calibration of the flow and transport models with existing data; and
. evaluation of radial action alternatives for the site with the
calibrated models. ^ ^ devel0ped based on the
avalllirhXiloTl'data. This conceptual forced the framed for
deverc:rr;r£— ^
The Finite t em aroundwater flow within an area around
»i iQ7Q^ was used to model the grounawa^c
et al., 1979) was ^ gri(J was developed and the
the Western Process ng i • . boundary conditions, hydraulic
necessary data on ^ ^ ^
conductivities, and hydrauic Energy, and Solute Transport
The three-dimensional Coupled ^ ^ ^
(CFEST) code (Gupta et a FP3DGW in that it uses the same
pccct is an extension o
transport. CFEST is a ^ Ifl addition> CFEST
hydrologic data struc ure COUDie contaminant transport with
includes the necessary parameters to coup
groundwater flow. t d to observed 1984 potentiometric and
The models ® l0n x. Once calibrated, they were used
contamination data provide
-------
to predict the effectiveness of six proposed remedial action scenarios: 1)
no-action, 2) source removal, 3) cap, 4) source removal combined with a cap,
5) upgradient slurry wall combined with a cap, and 6) pump and treat. The
flow model was used to predict alterations in the flow field and volume of
water removed, while the transport model was used to predict the mass of
contaminant removed and average concentrations up to 25 years into the
future.
A simplified analytical approach was also used to analyze remedial
actions for the site. This action was taken to demonstrate the applicability
of an analytical approach versus use of a fully three-dimensional numerical
modeling approach.
2
-------
SECTION 2
CONCLUSIONS
A groundwater flow and contaminant transport model of the Western
Processing Site has been developed and calibrated. An acceptable calibration
was achieved both In terns of matching model-predicted to observed hydraulic
potentials and trlchloroethylene (TCE) concentrations, as well as accurately
predicting the flow rate of Mill Creek and the concentration of TCE in the
creek. The model as it currently exists provides an excellent base on which
future calibration and validation can build as more data become available.
The model results show that Mill Creek has been and will continue to be,
the primary discharge point for TCE migrating from the Western Processing
Site By 1983, almost half of the TCE that was estimated to have entered the
flow system during site operation had exited to mil Creek. Over the next 25
years (1984 through 2006), the no-action predictions show about 60* of the
remaining TCE-will discharge to Mill Creek. , . .
Of the total mass originally disposed of at the site, 20* remaps m the
flow system 25 years after the source removal action was implemented.
oz v/pars of the pump and treat remedial action.
Similarly, 5% remains after 25 years ox tne w f
o niimn and treat remedial actions, the mass of
With both the source removal and pump ana treat
, . * -in rrepk will be reduced by about 50* over the next 25
TCE discharging to Mill Creek wi n De reu
u onnoi Nnnp of the actions simulated in the model will
years (1984 through 2008). None or ine
prevent TCE from discharging to the creek.
It was found that placing a cap over the site provides very Imie
benefit because the majority of the TCE has a,Iready enterea„d ,s ,
extern A slurry wall, as simulated in the
transported by the groundwater system. j
* u 4„flffortive for altering the groundwater flow patterns
model, was shown to be ineffective
and reducing the discharge of TCE to the creek.
Because the creek 1s a natural discharge point for contamination, a
~ cKnnld be considered is allowing the creek to act as a
remedial action that should be
3
-------
natural collection point for treatment. If an initial, short-terra solution
is desired, the pump and treat and source removal remedial actions will
remove the largest amount of contamination in the shortest period of time.
These actions also significantly reduce the contaminant load discharging to
Mill Creek and could lower contaminant concentrations in the creek to
acceptable levels.
The results presented in this report provide a preliminary assessment of
remedial actions proposed for the Western Processing Site. Only a single
simulation was performed for each action; sensitivity and optimization runs
were not performed. While more work can and should be done with the model,
this initial effort has provided valuable insight into the relative
performance of the remedial actions simulated.
As part of this project, an analytical solution was used in conjunction
with the CFEST model to analyze the pump and treat remedial action. The
intent was to determine if simple analytical solutions can be of value in
analyzing complex data sets. While the analytical solution proved to be of
some use in determining a reasonable pumping rate for wells at the Western
Processing Site, its overall usefulness was limited. Analytical solutions
are suitable for evaluating simple hydrologic systems, but they are of
limited value when evaluating complex data sets (i.e., multiple
hydrostratigraphic layers; variable hydraulic conductivities, porosities,
storage coefficients, recharge, and pumping depths; and stream/aquifer
interactions) such as exists at the Western Processing Site.
4
-------
SECTION 3
DESCRIPTION OF THE STUDY AREA
The Western Processing Site 1s looted within the City of Kent,
approximately four miles (6 km) north of the business district (Figure 1).
The facility occupies an area of about 13 acres (5 ha), and when in operation
consisted of a small laboratory, a solvent recycling plant, a fertilizer
plant, bulk storage tanks, drum storage areas, piles of flue dust,
construction debris, and large cement-block above-ground storage lagoons for
liquid wastes, cooling water, and process water (EPA, 1983). Hill Creek,
also known as King County Drainage Ditch No. 1. runs across the northwest
corner of the site from south to north. A drainage ditch, bicycle trail, and
railroad tracks run along the eastern boundary of the site.
The annual average rainfall at the Western Process,ng Site s 39 ,n.
(99 There is a well defined dry season In the sunder and a rainy season
in the winter Table 1 shows the monthly average of precipitation, potential
transpiration, "he -IngeTr,:
Ui^rTlO toVcl/yr). using , method described by Dunne and Leopold
u m./yr cm/yr) was obtained. Where water is
(1978), a recharge o ^ ^ assumed that very little runoff
ponded on the Western r ^ recharge is possible. A detailed
description of r« calculations 1s contained in Appendix A.
GEOLOGY
citP lies in the broad flood plain of the Green
The Western Pr°ccss average 20 ft (6 m) above mean sea level.
River. Elevations in this v y
5
-------
S 196TH STREET
Western
Processing
LAGOONS FILLED WITH
CONTAMINATED "WATER
STEEL HILL FLUE DUST
CONTAINING HEAVY METALS
NOT TO SCALE
POLLUTED POND
ASPHALT
EBRIS
o
<
o •
LANT
DRUMS OF ZINC CHLORIDE
Figure 1. Western Processing Location Hap.
6
-------
TABLE 1. AVERAGE MONTHLY PRECIPITATION, POTENTIAL EVAPOTRANSPIRATION
(PET), AND ACTUAL EVAPOTRANSPIRATION (AET) FOR THE SEATTLE AREA
Month
January
February
March
April
May
June
July
August
September
October
November
December
Annual
Precipitation.* in.
PET,** in.
AET,** in.
5.73
0.3
0.3
4.24
0.6
0.6
3.79
1.2
1.2
2.40
1.8
1.8
1.73
3.1
3.0
1.58
3.8
2.9
0.81
4.5
2.0
0.95
4.1
1.6
2.05
2.8
1.9
4.02
1.8
1.8
5.35
0.8
0.8
6.29
0.5
0.5
38.94
25.3
18.4
*(NOAA, 1»m)
**(Ellis, 1984)
The sediments include alluvial fan deposits of sand, silt, peaty silt, and
day more than 150 ft (45 .) thick, primarily denved fro* Ht. Ra,ner and
transported by the White River (Luzier, 1969).
The Western Processing Site is underlain by sand, silt, gravel, clay,
peat, and artificial fill. The fill 1. « 8 ft (2.4 -) «1« -*¦ has a
lower hydraulic conductivity than the surrounding «ter,a Well log,
, riau laver exists between 30 and 40 ft (9 to
indicate that an intermittent clay layer exisw
12 m) below the surface in the area around the site.
The soil underlying Western Processing 1s clashed as^ urban land"
(USDA 1973). Urban !and 1s soil that has been .difitd by disturbance of
... of fill material several feet thick to
thp natural lavers with additions ot
the natural layers Retaliations. I" the Green River Valley the
accomodate large industrial installations.
* * n ft 10 9 to 3.7 m) thick, and is gravelly sandy loam
fill ranges from 3 to 12 ft {0.9 to s.i
4-.lv>a The surrounding soils are in the Oridia-
to gravelly loam in texture.
Seattle-Woodenville Association.
7
-------
HYDROLOGY
The water table has been encountered at very shallow depths ranging from
3 to 12 ft (1 to 4 m) and averages 6 ft (2 m) below ground surface (EPA,
1983). A groundwater mound is present in the central portion of the site
(Figure 2) due to increased infiltration of ponded water at the surface and
the low permeability of the fill material. Groundwater flow directions are
shown in Figure 2. Localized flow is to Mill Creek and the drainage ditch,
while the regional flow is to the northwest toward the Green River.
Comparison of potential values of well pairs for March through July,
1984 (Table 2) indicates that the Western Processing Site itself is a
groundwater recharge area. The groundwater mound has created a downward
hydraulic gradient to at least to 30 ft (9 m) below the surface, and the area
surrounding the site is a discharge area (upward hydraulic gradient).
Transmissivities calculated by CH2M HILL from pumping and slug tests
range from 11.5 to 22,400 gpd/ft (1.4 x 10"1 to 278 m /day), and average
3,620 gpd/ft (45 m2/day). Conductivities (transmissivity divided by
thickness of gravel pack) range from 0.8 to 743 gpd/ft (3.3 x 10 to
30 m/day) and average 127 gpd/ft2 (5.2 m/day). Laboratory permeability tests
were performed by CH2M HILL on sediment samples from Wells 35 through 44
(Figure 3). These values range from 6.7 x 10"3 to 70 gpd/ft (3 x 10"4 to
2.9 m/day) and average 8.5 gpd/ft2 (0.35 m/day).
WASTE DISPOSAL HISTORY
Western Processing began operation in 1957 as an animal by-products and
brewer's yeast processor. Since that time the operation expanded to include
the handling of solvents, flue dust, battery chips, acids, cyanides, and a
wide variety of industrial wastes (EPA, 1983). In 1982, the EPA found 26
priority pollutants in the surface waters around the site, all of which were
subsequently found in on-site soil and groundwater samples. As a result of
these findings, the EPA issued an order to require the owner to conduct
monitoring to ascertain the nature and extent of the hazard that exists at
the site. After the owner declared himself unable to carry out the necessary
monitoring, a court order was obtained to allow the EPA to investigate the
site. As a result of this action, disposal at the site ceased in 1982. in
8
-------
Figure 2. Smoothed Kriged Potential Surface for April, 1984
9
-------
TABLE 2. COMPARISON OF SHALLOW AND DEEP WELL POTENTIALS
Well
Well
No.
Groundwater
Elevation (ft AMSL*)
Recharge/
Discharqe
Depth
(ft AMSL*)
Date
11A
17.16
Recharge
12
ft
3/1/84
11B
16.14
29
ft
3/1/84
17A
18.81
Recharge
15
ft
3/1/84
17B
15.62
30
ft
3/1/84
1A
15.02
Discharge
12
ft
4/3/84
IB
15.51
30
ft
4/3/84
11A
17.25
Recharge
12
ft
4/3/84
11B
16.14
29
ft
4/3/84
17A
19.73
Recharge
15
ft
4/3/84
17B
15.45
30
ft
4/3/84
31A
17.24
Discharge
150
ft
4/3/84
31B
16.07
55
ft
4/3/84
32A
17.49
Discharge
106
ft
4/3/84
32 B
15.49
28
ft
4/3/84
33A
18.67
Discharge
65
ft
4/3/84
33B
15.99
38
ft
4/3/84
34A
18.07
Discharge
134
ft
4/3/84
34B
16.13
62
ft
4/3/84
~AMSL = Above Mean Sea Level
10
-------
0 66 132 198 264 330
(1:1800)
• MONITORING WELL
l-lll WASTE PONDS
WELL 30
SCALE IN FEET
25A, B
22A, B
WESTERN
PROCESSING
OffiC*
South 196th ST.
-l! (? ~Vc'
o
Office
ENTRANCE
• •
1A,B 2
Processing
Office Butiding
£=~ •
17A, B
11 A, B
Fipure 3. Western Processing Well and Waste Pone! Locations.
11
-------
1982 and 1983, the EPA installed a series of monitoring wells and collected
soil and water samples. An initial remedial measure was implemented (limited
capping, excavation and removal, drum storage, etc.) to contain some of the
waste until a more permanent remedial action could be designed and
implemented.
A review of the records of the waste received by the site indicate that
TCE was one of the most common wastes received, and that it was received from
about 1960 to 1980. Sampling results confirm the widespread distribution of
TCE in on-site wells, although it has been detected (at low levels) in only
two of the off-site wells.
The maximum observed TCE concentration in groundwater is 210,000 ug/L,
while the maximum observed soil concentration is 558,000 ug/Kg. The EPA
priority pollutant human health criteria for TCE in water is 27 ug/L at 10 ®
cancer risk (EPA, 1980). An analysis of the monitoring data indicated that
Reaction Ponds I and III and an area near Well 21 (Figure 3) were the primary
disposal areas for TCE.
12
-------
SECTION 4
MODEL DEVELOPMENT
A numerical model was developed to simulate groundwater flow and
contaminant transport at the Western Processing Site. The model was
developed in two steps: 1) a flow model was developed to describe the
groundwater flow 1n the area around the Western Processing Site; and 2 the
flow model was used to fom the basis of a transport model which simulated
the movement of contaminants in the groundwater. Although the model was
developed in two stages, the final result is a single model which can be used
to simulate groundwater flow and contaminant transport at the s,te. Because
the model was developed 1n a staged approach, the flow and transport portions
will be discussed separately.
MODEL SELECTION
i was selected for the Western Processing Site
A three-dimensional model was seiecte
« able to: simulate variations in permeability with depth;
simulate the vertical flow within the study area; simulate localized
!« I Mill Creek and the drainage ditch; and accurately simulate
discharge . , proposed remedial actions,
slurrv wall and pumping depths P P
y . , rnrip, selected to model the Western Processing Site are
The numerica The FE3DGW code simuiates
the FE3DGW flow code an ^ ^ simuUtes contaminant
groundwater flow wh e cornp1etely compatible such that the simulation
transport. The two co proceeds directly from calibration of
of transport phenomena us n ^ benchmarked against
FE3DGW based on flow proper verified by solution of standard
other numerical codes and have been
analytical problems.
13
-------
REGIONAL MODEL
The original approach to modeling the Western Processing Site included
the development of a regional model to describe flow within the valley
surrounding Western Processing. This regional model, to include the area
within a 1.5 mile (2.4 km) radius of the site, would have established
boundary conditions for the local model. The regional model was also
intended to establish reasonable transmissivities for the study area since
the reported values ranged over three orders of magnitude.
In reviewing the data for the regional system, it was determined that
sufficient data to calibrate this model were not available and could not be
obtained within the time frame of this study; therefore, only a local model
was developed. Transmissivities and boundary conditions for the local model
were estimated from the available data and adjusted in the model calibration
process. Additional data collection efforts by EPA Region X aided in
verifying some of the estimated parameters.
GROUNDWATER FLOW MODEL DEVELOPMENT
The flow model of the local area around the Western Processing Site was
developed based on the available hydrogeologic data. The model area is
2,790 ft (850 m) wide and 4,020 ft (1,225 m) long. The Western Processing
Site is located near the center of the model region (Figure 4).
A finite element grid was developed for the local model region to
properly represent the areal extent, boundary conditions, and primary
features of the hydrologic system. The grid consists of 311 nodes and 283
elements. The two-dimensional surface representation of the grid is shown in
Figure 5.
Data files were developed for the aquifer thickness and extent,
vertical and horizontal hydraulic conductivity, and , hydraulic stress
(recharge and discharge) using data received from EPA Region X. The data
used in the final calibrated flow model are discussed below.
Structure
The top 100 ft (30 m) below the water table was simulated in the model.
The top 30 ft (9 m) was simulated as a silt and fine sand material, except
14
-------
15
-------
Figure 5. Finite Element Grid of the Western Processing Site.
16
-------
that directly below the Western Processing Site where the top 10 ft (3 m) was
simulated as a fill material with a lower permeability than the surrounding
surface material. An intermittent clay layer was simulated between 30 ft (9
m) and 40 ft (12 m) below the surface. The material between 40 ft (12 m) and
100 ft (30 m) was simulated as a gravelly sand in the model.
In order to properly represent the different material types in the
model, the 100 ft (30 m) thick layer was subdivided into 4 layers from top to
bottom (Figure 6), with their thicknesses being 10 ft (3 m), 20 ft (6 m), 10
ft (3 m), and 60 ft (18 m), respectively.
Boundary Conditions
The model boundaries were defined with the FE3DGW code's "leakance
boundary condition" option, which can be defined as a combination of flux and
constant head (held) boundary conditions. This option uses the distance from
the boundary to a known potential, the potential at that distance, and the
cross-sectional area associated with each node to calculate the flux and
potential at the boundary nodes based on the model conditions. In effect,
this option calculates the potential along the boundary based on regional and
local data. The regional data used to calculate the boundary conditions were
obtained from regional wells and Green River elevations. Elevations along
the Green River were used since It Is assumed to be hydraullcally
interconnected with the water table.
Mill Creek and the ditch to the east of the site were simulated using a
stream boundary option. Rather than holding the groundwater potential at the
elevation of the creek and the ditch, this option allows the model to
calculate the groundwater potential based on the surface water elevation;
stream bottom elevation, cross-sectional area, thickness, and permeability;
and minimum stream depth. Surface water elevations were interpolated and
extrapolated at each node along Hill Creek and the ditch, from surveyed
values obtained in April, 1984 (Figure 7). A more detailed description of
the boundary conditions is provided in Appendix B.
Hydraulic rnnductivity
Initially, horizontal (Kh) and vertical (Kv) hydraulic conductivities
^ thP four material types discussed above based on the best
were assigned to tne
17
-------
WESTERN PROCESSING SITE
10FT
t
20 FT
10FT
50FT
FILL
SILT AND SAND
CLAY
GRAVELLY SAND
Not to Scale
Figure 6. Cross Section Depicting the Structural Layers of the Study Area.
18
-------
Figure 7. Location of Surveyed Values for Mill Creek and Drainage Ditch.
19
-------
available data. These values were then adjusted in the calibration process
until a good match was achieved between model-predicted and observed
potentials.
The hydraulic conductivities used in the final calibrated model for each
material (layer) are shown in Table 3. The values were uniform throughout
each layer. For all material types, the final calibrated values were within
half an order of magnitude of the original values estimated from the data.
Groundwater Potential
Due to a lack of transient data, it was assumed that the groundwater
system around the Western Processing Site is in steady state. A review of
the transient potential data that have been collected suggests that, although
the potentials do change with time, they do not ch^ge «•»¦>< to
significantly alter the flow field or flow velocities. Therefore, the steady
state assumption, while not completely correct, is considered to be
acceptable.^^r ^ ^ Aprj)j 1984 potential data was used to
represent the initial potential conditions (Figure 2). This surface was
represent A+ontl-a1 data from 36 wells on and around the site, and
prepared by knging Pand ^ ditch. Krigin9 is a statistical
from measuremen s^ a ^ & surface frora spatially-distributed data. The
technique use surfaCe was compared to the kriged potential
model-predicted potential surtace
surface in the model calibration process.
TABLE 3. CALIBRATED HYDRAULIC CONDUCTIVITY VALUES
Layer Material Type
Kh Kv
(ft/day) (ft/dav)
1 Fill
2 Silt and Fine Sand
3 Clay
4 Gravelly Sand
0.6 0.06
2.5 0.13
0.3 0.03
25.0 1.25
20
-------
Hydraulic Stress
The only hydraulic stress considered within the model region was
recharge from precipitation. Recharge was assumed constant over the area at
10 in./yr (25 cm/yr). The only exceptions were in the asphalted (capped)
area and the area of ponded water on the site (Figure 1) where recharge was
set at 0 in./yr and 22 in./yr (56 cm/yr), respectively. A detailed
description of the recharge calculations is contained in Appendix A.
Porosity
A porosity of 15% was used in all layers of the model except the clay
layer where the porosity was assumed to be 20%.
CONTAMINANT TRANSPORT MODEL DEVELOPMENT
The contaminant transport model was developed using the calibrated flow
model, observed or estimated migration parameters, and estimates of source
loading on the groundwater system as a function of time. Data input files
were developed to define source concentrations, leaching rates, retardation
factors, and dispersivity. In most cases, these data were not specifically
known for the Western Processing Site. As a reasonable estimate, initial
values were selected from the literature and final values were derived in the
model calibration process. The data used in the final calibrated transport
model are discussed below.
Contaminant Selection
A review of the list of wastes received at the Western Processing Site
shows that TCE was accepted for disposal throughout the operating life of the
site. Also, high concentrations of TCE have been measured in many of the
wells on site and it is one of the ubiquitous contaminants. Therefore, TCE
was selected for use in calibrating the transport model, and for comparison
of the various remedial action alternatives simulated.
Source Location
A review of the sampling results for TCE in the on-site wells (EPA,
1983) reveals three probable source locations: 1) Reaction Pond I,
21
-------
2) Reaction Pond III, and 3) near Well 21 (Figure 3). The areas of these
three sources were established as 5,700 ft2 (530 m ), 4,170 ft (388 m ), and
2,190 ft2 (204 m2), respectively, in the model.
Source Area Concentrations
The model simulated leaching of TCE into the groundwater rather than
direct infiltration; therefore, the initial TCE concentration at all three
source areas was set at the solubility limit of TCE in water, 1.1 x 10 ug/L
(Verschueren, 1977). The loading rate at each site can be calculated as the
infiltration rate, times the surface area of the source, times the initial
TCE concentration at the source. The infiltration rates used in the
calibrated model were 6 in./yr (15 cm/yr) at Reaction Pond I, and 10 in./yr
(25 cm/yr) at Reaction Pond III and around Well 21. Using the areas and the
initial concentration discussed above, the loading rates were estimated at
75 Ib/yr (34 Kg/yr), 230 Ib/yr (104 Kg/yr), and 300 lb/yr (136 Kg/yr) at
Reaction Pond I, Reaction Pond III, and Well 21, respectively.
Source Duration/Leach Rate
The sources were assumed to be actively leaching TCE into the
groundwater for 20 years, from 2958 through 1978. After 1978, TCE was no
longer considered to be leaching into the groundwater flow system, however,
the TCE already introduced was considered to be available for transport.
Sorpt i on/R et ar d at i on
During transport through soils, TCE undergoes retardation caused by
adsorption. Based on available data, an adsorption coefficient (Kd) of
between 0 1 and 1.0 (Richter, 1981) appears reasonable for the site. A Kd of
O 2 was used in the final calibrated model, which corresponds to a
retardation factor of 4, for the Western Processing Site. Calculation of the
retardation factor is described in more detail in Appendix C.
22
-------
SECTION 5
MODEL CALIBRATION
The flow and transport models were calibrated by adjusting certain model
input parameters until a good match was achieved between model-predicted and
observed data. A brief description of the calibration process for both the
flow and transport models is provided below.
FLOW MODEL CALIBRATION
Once the data were Input into the FE3DGW code, the model was run in the
steady-state mode to predict groundwater potentials. The model was
4.u nredicted flow-field to measured potential
calibrated by comparing the model-prea c
data* , L -nredicted and measured hydraulic potentials
The difference between model-preait
. , fnllowinq flow model parameters: the vertical
was minimized by adjusting the following
. ..witv the parameters controlling the flow to
and horizontal hydraulic conductivi y, A u x
j . h fstream bottom permeability and thickness),
Mill Creek and the drainage ditch is
and the boundary conditions.
. . j «M-ontial surface for the water table (top of
The final model-predicted potential
, with the kriged potential data (Figure 2)
Layer 1) (Figure 8) regime within the study area (localized
a„d the conceptual .odel * ™ , flow t0 the northwest).
now to Mill Creek and the ditch d r« ^ ^ ^ ^ ^
Potentia! surfaces for the top rf LW ^ Jhe model.predicted
4 are shown in Figure , > » ^ within the study area is
groundwater flux to ^pares well with a gain of 0.5 cfs
0.3 cfs (734 m /day). d as measured in May,
(1,223 m /day) along Mill Creek within the
1982, by EPA Region X. changes made in the calibration
A more thorough description o ided in Appendix D.
process and their impact on model results P
23
-------
Simulation.
Potential Surface for the Base Case
Simulation.
-------
Figure 10. Model-Predicted 1983 Top of Layer 3
Potential Surface for the Base Case
Simulation.
Figure 11. Model-Predicted 1983 Top of Layer 4
Potential Surface for the Base Case
Simulation.
-------
-------
Regional groundwater flow as predicted by the model is to the northwest,
while localized flow is to Mill Creek. Discharge to Mill Creek dominates the
flow patterns to a depth of about 30 ft (9 m) and its influence can be seen
at 100 ft (30 m). Below 30 ft (9 m) the flow is primarily controlled by the
regional gradient.
TRANSPORT MODEL CALIBRATION
Once the data were input into the CFEST code, the model was run in the
transient mode with five-year time steps from 1958 to 1983. The model was
calibrated by comparing model-predicted TCE concentrations to measured TCE
concentrations for 1983.
The difference between model-predicted and measured TCE concentrations
was minimized by adjusting the following transport model parameters:
retardation factor, leach rates, and source strengths. Dispersivity was also
adjusted in the calibration process, however, this factor had little impact
on model results. Leach time was assumed to be the period of active disposal
(20 years). Because the unsaturated zone is very thin, it was assumed that
TCE entered the saturated zone quickly after disposal.
A kriged concentration contour plot of TCE within the model study area
is shown in Figure 13. The model-predicted TCE concentrations for the top of
Layer 1 (Figure 14) compared reasonably well with the kriged values. The
model-predicted TCE concentrations for the top of Layers 2, 3, 4, and the
bottom of Layer 4 are shown in Figures 15, 16, 17, and 18, respectively.
In addition to matching the observed location of the TCE plumes from the
three source areas, it was also of primary concern to match the maximum
observed TCE concentration at these areas. This match is achieved through
source calibration. The measured and model-predicted concentrations are
shown in Table 4.
27
-------
Figure 13. Kriged TCE Concentration Contours.
contours ir. ug/l
Concentrjt ion"levels •
100
10.000
50.000
J 00.000
ISO.000
Figure 14 Model-Predicted 1983 Top of Layer 1
TCE Concentration Contours for the
Base Case Simulation.
-------
Figure 15. Mode]-Predicted 1983 Top of Layer 2
TCE Concentration Contours for the
Base Case Simulation.
mill creek
Concentrjt'or
COfloj'-! fr uo'L
Concentrftion levels
IOC
J 0.000
sr.ooo
100.000
J SO.000
Figure 16. Model-Predicted 1983 Top of Layer 3
TCE Concentration Contours for the
Base Case Simulation.
-------
Base Case Simulation.
Figure 18. Model-Predicted 1983 Bottom of Layer 4
TCE Concentration Contours for the
Base Case Simulation.
-------
TABLE 4. COMPARISON OF OBSERVED TO MODEL-PREDICTED MAXIMUM
TCE CONCENTRATIONS IN THE GROUNDWATER AT THE THREE
SOURCE LOCATIONS
Measured Model-Predicted
Concentration Concentration
Source Location (uq/L) , (uq/L)
Reaction Pond I 210,000 213,000
Reaction Pond III 140,000 141,000
Well 21 170,000 166,000
BASE CASE MOOEL RESULTS
As an additional calibration of the transport model, the concentration
of TCE in Mill Creek was calculated based on model results and compared to
the measured concentration. The model calculated concentration of 10 ug/L
and 40 ug/L based on creek flows of 15 cfs (0.4 m /sec) and 3 cfs (0.08
m3/sec), respectively, compared well with the creek TCE concentration of 15
ug/L measured in May, 1982 by EPA Region X.
The base case was defined as the 25-year simulation period from 1958
through 1983. Over this 25 year period, the model predicted that a total of
11 900 lb (5,400 Kg) of TCE were disposed of at the site and entered the
groundwater flow system. Of this total, the model predicted that 5,790 lb
(2 630 Kg) discharged to Mill Creek and the ditch, and 6,110 lb (2,770 Kg)
remained in the flow system in the year 1983. Of the total mass leaving the
system, about 95% discharges to Mill Creek and the remaining 5% discharges to
the drainage ditch.
The distribution of TCE in the study area as predicted by the model at
5-year intervals is su«arized in Table 5. The concentration in 1963 and
1978 are shown in Figures 19 and 20, respectively.
31
-------
TABLE 5. DISTRIBUTION OF TCE IN THE MODEL BASE CASE SIMULATION
TCE Remaining
Year
TCE
Inflow (lb)
TCE
Outflow (lb)
in Groundwater
System (lb)
1963
2,975
525
2,455
1968
2,975
975
4,450
1973
2,975
1,335
6,095
1978
2,975
1,620
7,455
1983
0
1,335
6,110
Total
11,900
5,790
6,110
32
-------
n
Base Case Simulation.
Base Case Simulation.
-------
SECTION 6
ASSESSMENT OF REMEDIAL ACTION ALTERNATIVES
Six remedial action alternatives were identified by EPA for potential
application to the Western Processing Site. The alternatives are:
• no-action;
t source removal;
• cap;
• source removal combined with a cap;
• upgradient slurry wall combined with a cap; and
§ pump and treat.
Each alternative was Implemented in the model to predict its impact on
reducing contamination levels at the site and in Mill Creek.
The base-case simulation consisted of running the final calibrated model
for 25 years from 1958 through 1983. The simulations were run to the year
1983 so that the model could be calibrated to the most recent set of
chemistry data. In 1983 certain model parameters were adjusted to simulate
the various remedial actions, and the model was run for an additional 25
years to the year 2008. The model results at the end of the 25-year
predictive period were used to evaluate the various remedial actions and to
hp most effective in reducing the level of
determine which action would be most en en
contamination. . ^
It is important to note that programmatic constraints of this study only
allowed for a single model run for each of the remedial actions. To properly
interpret the model results for each case, and to properly compare the
different cases, sensitivity and optimization runs should be performed.
While the results of this study are helpful in making an initial comparison
of the various potential remedial actions, they should not be considered
conclusive.
34
-------
ASSESSMENT APPROACH AND RESULTS
The six remedial actions simulated and how they were implemented in the
model are discussed below. For each case, results are presented in the form
of contour plots, maximum concentrations in the system, and total mass of TCE
in the system and exiting the system.
No-Action
The first step in the remedial action analysis was to simulate the no-
action scenario to establish a benchmark against which all other actions
could be compared. The no-action scenario entailed running the final
calibrated model (base case) 25 years into the future (1984 through 2008)
without any changes. The model simulated the continued migration of the TCE
which entered the flow system in the base case simulation.
TCE concentration contours at the top of Layer 1 in the years 1988,
1998, and 2008 for the no-action case are shown in Figures 21, 22, and 23,
respectively. The maximum TCE concentrations in the groundwater at each of
the three source areas in the year 2008 are listed in Table 6. The total
mass of TCE in the flow system, and the total mass discharging to Mill Creek
and the ditch at 5-year time intervals are shown in Table 7. Table 7 shows
that of the 11,900 lb (5,400 Kg) that entered the flow system between 1958
and 1983, 20* (2,375 lb (1,075 Kg)) remains in the system in the year 2008.
Of the 6,110 lb (2,770 Kg) of TCE remaining in the flow system in 1983, about
60* (3,790 lb (1,720 Kg)) exited to Mill Creek and the ditch by the year
2008. As in the base case, of the amount exiting the groundwater flow system
about 95* entered Mill Creek and the remaining 5% entered the drainage ditch.
TABLE 6. MODEL PREDICTED MAXIMUM CONCENTRATION (ug/L) IN
THE GROUNDWATER AT THE THREE SOURCE AREAS IN THE
YEAR 2008 FOR THE REMEDIAL ACTION SIMULATIONS
No-
Act i on
Source
Slurry
Source
Removal
Wall
Removal
Cap
and Cap
and Cap
810
50,810
1,590
13,415
330
2,545
630
480
2,400
52,740
4,365
12,055
Reaction Pond I 21,000 810 50,810 1,590 13,415 260
Reaction Pond III 770 330 2,545 630 480 580
Well 21 17,000 2,400 52,740 4,365 12,055 1,990
35
-------
r
Figure 21. Model-Predicted 1988 Top of Layer 1
TCE Concentration Contours for the
No-Action Simulation.
Figure 22. Model-Predicted 1998 Top of Layer 1
TCE Concentration Contours for the
No-Action Simulation.
-------
OJ
Figure 23. Model-Predicted 2008 Top of Layer 1
TCE Concentration Contours for the
No-Action Simulation.
-------
TABLE 7. MODEL-PREDICTED DISTRIBUTION OF TCE FOR
THE NO-ACTION SIMULATIONS
TCE TCE Total Mass in
Year Inflow (lb) Outflow (lb) System (lb)
1988 0 1,075 5,055
igg3 0 880 4,180
1998 0 730 3,460
2003 0 605 2,865
2008 0 500 2,375
Total 0 3,790
The potential surfaces for each model layer in the no-action scenario
were identical to the steady state potential surfaces for the base case
(Figures 8 through 12).
Source Removal
The source removal remedial action assumed that 10 ft (3 m) of fill
material 1s excavated and removed from the site. The intent here was to
remove the most highly contaminated soils. Source removal was simulated In
the model by setting the concentration of TCE to zero in the top model layer
(10 ft (3 m)) over the entire area of the site. The concentrations in the
impacted area were zeroed in 1984 and the model was run for 25 years to
simulate the migration of TCE remaining in the flow system.
TCE concentration contours at the top of Layer 1 in the year 2008 are
shown 1n Figure 24. The maximum TCE concentrations In the groundwater at
each of the three source areas in the year 2008 are shown in Table 6. The
total mass of TCE in the flow system and the total mass discharging to Mill
Creek and the ditch at 5-year intervals are shown in Table 8. Table 8 shows
that a total of 2,425 lb U, 100 Kg) of TCE exite the flo*- system to Mil
iLa oc wo at* neriod, and that 1,315 lb (595 Kg)
Creek and the ditch over the 25 year penuu,
remains in the flow system in the year 2008.
38
-------
Figure 24.
Model-Predicted 2008 Top of Layer 1
TCE Concentration Contours for the
Source Removal Remedial Action
Simulation.
39
-------
Year
1988
1993
1998
2003
2008
Total
TCE
Outflow fib)
Total Mass of
TCE in System (lb)
770
575
445
355
280
2,935
2,370
1,930
1,585
1,315
2,425
Because it was assumed that the area nf
excavation would be backfilled
with a material of similar conductivity the i
. , , . "'vny, the potential surfaces for each
model layer in the source removal case werp i 4.
«. +. , - # L identical to the steady state
potential surfaces for the base case (Figures 8 through 12).
„,r».
thereby reduce the .nlgration of TCE 1n the groundwater ^3^^^ cap
—»- - ~ -«.
The effect of eliminating recharge was to Oinhtiu i
over most of the site, and to reduce the mound i tn 0Wer ^ Wat6P
about 1 ft (0.3 (Figure 25).
TCE concentration contours at the too nf i i .r
. c. Tu • Uyer 1 for the year 2008 are
shown in Figure 26. The maximum TCE concentration* in
• , .. ations in the groundwater at
each of the three source areas in the vear ?nno ,
„ „ year 2008 are shown in Table 6. The
«,.» concentration all areas actuary Increased over those predicted by
the no-act ion case, as a result of eliminating the dilution effect of the
recharge.
Table 9 shows that a total of 3 475 ih n nt ^ ^
c c* * M-n r L * ( ' 5 Kg) of TCE ex1'ted the flow
system to Mill Creek and the ditch over thp
o fiQn lh /1 onn v \ ¦ ¦ „u 25-year simulation, and that
2,690 lb (1,200 Kg) remains in the flow system in the year 2008.
Cap
40
-------
Figure 25. Model-Predicted Top of Layer 1
Potential Surface for the Cap
Remedial Action Simulation.
Figure 26. Model-Predicted 2008 Top of Layer 1
TCE Concentration Contours for the
Cap Remedial Action Simualtion.
-------
TABLE 9. MODEL-PREDICTED DISTRIBUTION OF TCE FOR
THE CAP REMEDIAL ACTION
Year
1988
1993
1998
2003
2008
Total
TCF
Outflow (lb)
Total Mass of
TCE in System fib)
980
795
665
560
475
5,150
4,360
3,705
3,155
2,690
3,475
Source Removal Combined with a Cap
The source removal and cap scenario is a combination of the two cases
discussed previously. This case assumed that 10 ft (3 m) of fill material is
excavated from the site, the site is filled in with a material of similar
permeability, and a cap is placed over the site. The intent of the action is
to remove the most contaminated soils, and eliminate recharge. This case was
simulated in the model as a combination of the previous two cases: reduce
the concentrations in the top layer on-site to zero, and eliminate recharge
at the site.
The potential surfaces for each model layer for this case were identical
to the case where only the cap was simulated. The potential surface for the
top of Layer 1 is shown in Figure 25.
TCE concentration contours at the top of Layer 1 for the year 2008 are
shown in Figure 27. The maximum TCE concentrations in the groundwater at
each of the three source areas in the year 2008 are shown in Table 6. These
concentrations are higher than the source removal remedial action as a result
of decreased dilution due to eliminating the recharge.
Table 10 shows that a total of 2,355 lb (1,070 Kg) of TCE exited the
flow system to Mill Creek and the ditch over the 25-year simulation, and that
1,385 lb (630 Kg) remains in the flow system in the year 2008.
42
-------
Source Removal Plus Cap Remedial
Action Simulation.
43
-------
TABLE 10. MODEL-PREDICTED DISTRIBUTION OF TCE FOR
THE SOURCE REMOVAL PLUS CAP REMEDIAL ACTION
TCE
Total Mass of
Year
Outflow (lb)
TCE in System (lb)
1988
735
2,975
1993
550
2,430
1998
435
2,000
2003
350
1,660
2008
285
1,385
Total
2,355
Slurry Wall Combined with a Cap
The slurry wall and cap remedial action assumed that a low permeability
barrier is placed along the eastern and southern boundaries of the site, and
that a low permeability cap is placed over the site. The intent of the
slurry wall was to divert groundwater around the site, and the cap would
reduce recharge.
The cap was simulated in the model by eliminating recharge as discussed
in the previous two cases. The slurry wall was simulated by introducing a
row of long narrow elements in the model along the eastern and southern
boundaries of the site and assigning these elements a low permeabilny. The
Lei elements representing the slurry wall were assigned «, widt of 5 ft
(1.5 m), and the permeability was set at 2.8 x 10 ft/day (10 cm/sec) to a
e^fe^t'of'the slurry wall and cap was to slightly lower the water
table on site. The potential surface with the slurry wall and cap is shown
" FTrconcentration contours at the top of Layer 1 for the year 2008 are
shown in Figure 29. The maximum TCE concentration in the groundwater at each
in thp vear 2008 are shown in Table 6.
of the three source areas the y ^
Table 11 shows that a total of 3,9^b id °
. ... - nuer the 25-year simulation, and that
flow system to Mill Creek and the ditch over y
2,240 lb (1,015 Kg) ranains in the flow system in the year 2008.
44
-------
r
Figure 28. Model-Predicted Top of Layer 1
Potential Surface for the Slurry
Wall Plus Cap Remedial Action
Simulation.
45
-------
Fiqure 29. Model-Predicted 2008 Top of Layer 1
ICE Concentration Contours for the
Slurry Wall Plus Cap Remedial Action
Simulation.
46
-------
TABLE 11. MODEL-PREDICTED DISTRIBUTION OF TCE FOR
THE SLURRY WALL PLUS CAP REMEDIAL ACTION
Year
1988
1993
1998
2003
2008
Total
TCE
Outflow fib)
Total Mass of
TCE in System (lb)
1,095
910
760
635
525
5,040
4,135
3,385
2,755
2,240
3,925
Pump and Treat
The pump and treat remedial action assumed that a network of pumping
wells was installed in the vicinity of the three source areas. The intent of
the wells was to remove contaminated water from the flow system.
The pumping wells were simulated as discharge from nodes in the model.
Three wells were placed to the north and west of Reaction Pond I, two wells
were placed to the east of Reaction Pond III, and two wells were placed to
the north of Well 21. Each well was assumed to be screened to a depth of
3 o .
30 ft (9 m), and the pumping rate was set at 577 ft /day (16.3 m /day).
The effect of the pumping was to slightly lower the water table in the
center of the site as shown in Figure 30.
TCE concentration contours at the top of Layer 1 for the year 2008 are
shown in Figure 31. The maximum TCE concentrations in the groundwater at
each of the three source areas in the year 2008 are shown in Table 6.
Table 12 shows that a total of 5,810 lb (2,635 Kg) of TCE exited the
flow system to Mill Creek, the ditch, and the pumping wells over the 25-year
simulation and that 330 lb (150 Kg) remains in the flow system in the year
2008.
47
-------
r~-
Figure 30. Model-Predicted Top of Layer 1
Potential Surface for the Pump
and Treat Remedial Action Simulation.
Figure 31. Model-Predicted 2008 Top of Layer 1
TCE Concentration Contours for the
Pump and Treat Remedial Action
Simulation.
-------
TABLE 12. MODEL-PREDICTED DISTRIBUTION OF TCE FOR
THE PUMP AND TREAT REMEDIAL ACTION
Year
1988
1993
1998
2003
2008
Total
TCE
Outflow (lb)
Total Mass of
TCE in System (lb)
3,550
1,260
565
285
150
2,540
1,295
740
465
330
5,810
DISCUSSION OF RESULTS
Uncertainty exists in the model input data and, as a consequence, in the
model output. The effect of these uncertainties is significant with regard
to the absolute values of the model results. However, the uncertainties
should not have a significant effect on the relative performance of the
various model simulations. Therefore, while the magnitude of the model
predictions may not be exact, the relative performance of the various
remedial actions considered should be accurate.
The results of the five remedial action cases are displayed in
Figure 32 Figure 32 shows that the pump and treat case, the only active
remedial measure, achieves the best results in terms of mass of TCE removed
from the system. Of the passive remedial measures, the source removal and
source removal plus cap remedial actions are most effective, and achieve
nearly identical results. The no-action, cap, and slurry wall plus cap
remedial actions achieve nearly identical results and are ineffective.
One of the most significant observations that can be made from the
modeling results is that for all remedial actions, even the no-action case,
the mass of TCE in the groundwater flow system is greatly reduced over the
50-year model simulation period. The reason for this is that Mill Creek acts
as a natural sink for most of the contamination at the Western Processing
Site The base case simulation (1958 through 1983) shows that the TCE plume
reaches Hill Creek and the ditch in the first 10 years of disposal
operations. However, over the next 40 years (15 years of the base case plus
49
-------
Year
Figure 32. Comparison of the Total Mass of TCE Remaininq in the Groundwater
System for the Five Remedial Action Cases.
-------
the no-action case), the plume advances only slightly and the majority of the
mass of TCE in the system discharges to Mill Creek. For the base case, the
model predicted that almost half of the TCE that entered the system has
already exited the system by 1983. In the no-action simulation (1984 through
2008), the model predicted that about 80% of the TCE that entered the system
has exited by the year 2008. Of the amount of TCE exiting the system, the
model predicted that the majority of it (95%) entered Mill Creek while only
5% entered the drainage ditch.
The cap and slurry wall plus cap remedial actions show a similar trend
to that for the no-action case. While these cases alter the distribution of
TCE slightly, about 80* of the TCE in the system still discharges to Mill
Creek by the year 2008.
The caps in both of these cases have virtually no effect because the
leaching of TCE into the flow system was not simulated in the model past
1978. Therefore, eliminating the recharge, as was done to simulate a cap,
had no effect on reducing the amount of TCE entering the system. The cap did
slightly alter the groundwater flow pattern and actually slightly increased
the mass of TCE exiting to Mill Creek.
The assumption that TCE is leaching into the groundwater only during the
period of active disposal is questionable. This assumption was based on the
fact that the water table is very near the surface (about 5 ft (2 m)), and
that initial remedial measures have already been performed at the site. This
assumption needs to be studied further before categorically ruling out the
benefits of a cap at the site.
The model predicted that the source removal action would remove about
3 000 lb (1,360 Kg) of TCE from the system. Once that initial mass is
removed, the' mass remaining in the system discharges to Mill Creek at a
similar rate as in the no-action case. By the year 2008, about 90* of the
total mass of TCE that entered the groundwater during disposal will have
exited the system. Of the 6,000 lb (2,720 Kg) remaining in the system in
1983, about half of it was removed by excavation, 25% discharged to Mill
Creek and 25* remains in the system by the year 2008.
The pump and treat action removed about 2,500 lb (1,135 Kg) of TCE in
the first 5 years (1983 through 1988) in addition to the 1,000 lb (450 Kg)
that discharges to Mill Creek. After the first 5 years the mass of TCE
51
-------
removed by the wells rapidly decreases because the concentration of TCE in
the withdrawn groundwater decreases. By the year 2008, 95% of the TCE
remaining in the flow system in 1983 will have exited.
52
-------
SECTION 7
SIMPLIFIED ANALYTICAL APPROACH TO THE WESTERN PROCESSING DATA SET
The objectives of this subtask were to: 1} reformat the data used in the
CFEST model of the Western Processing Site to a form that lends itself to
simple analytical solutions; and 2) use simple analytical techniques to
predict hydraulic response at the site to remedial action alternatives. The
various remedial action alternatives that are appropriate for the site are
no-action, pump and treat, slurry wall with cap, source removal, and capping.
Remedial Action Modeling Volume 2. Simplified Methods for Subsurface and
Waste Control Actions (Brown, 1984), provided a compendium of analytical
solutions that may be appropriate for various remedial actions. Most of the
analytical solutions discussed in the report are appropriate for simple
rather than complex groundwater systems. Solutions were listed for remedial
actions similar to those proposed for the Western Processing Site. Given the
complex layering and hydrology of the site, only the pump and treat option
was analyzed with the simplified analytical solution.
PUMPING ANALYSIS
The CFEST model requires values for hydraulic conductivity (vertical and
horizontal), aquifer thicknesses, gradients, recharge, porosity and storage
in order to predict groundwater flow direction and velocity. The effects of
anisotropy and other inhomogeneities can be included in the model. Simple
analytical solutions are limited to smaller, less complicated data sets. The
analytical program compared to the CFEST model could not calculate the
effects of multiple layers, inhomogeneities, recharge, or variable thickness.
The calculator solutions were designed to handle simple systems or systems
where limited data are available. The computer models like CFEST can handle
large data sets and very complex hydrologic relationshlps.
Several hand-held calculator programs for well hydraulKS were
identified in Brown (1984). The Theis condition well field program, NWELLS,
53
-------
which can be used to determine the effect of pumping and/or recharge on an
aquifer (van der Heidje, 1983) was chosen as the most suitable analytical
program to analyze the data set used for the Western Processing Site. The
NWELLS program uses the Theis equation to determine the drawdown at a well
for a given set of conditions and sums the results for an observation well.
All calculations were performed on a Hewlett-Packard 41-CV hand-held
programmable calculator.
Many factors influence the groundwater flow patterns at the Western
Processing Site. A shallow groundwater system intersects the surface along
the west side of the site at Mill Creek. A groundwater/surface water
interface also occurs at the drainage ditch along the east side of the site.
A groundwater mound has been identified near the center of the site, and a
portion of the site has already been covered by an impermeable cap.
The pump and treat remedial action tested in the CFEST model was used to
guide the analytical program. The seven dewatering wells simulated in the
CFEST model were set up on a grid, and X-Y coordinates were determined for
the NWELLS program (Figure 33). The first 5 analytical simulations used a
hydraulic conductivity which was initially (before calibration) thought to be
representative of the surface materials on site (28.3 ft/day (8.6 m/day)).
The depths of the 7 withdrawal wells were 30 ft (9.2 m), the same as in the
CFEST model. The drawdown at a single well (the observation well in Figure
33) as predicted by the analytical solution was compared to the drawdown at
the same location as predicted by the CFEST model. A sugary of the input
parameters for the analytical calculation are shown in Table 13.
The first analytical runs were used to help determine a reasonable
pumping rate for the CFEST model. A pumping rate of 30 gpm (113 L/min)
resulted in a drawdown greater than the thickness of the aquifer. A pumping
rate of 10 gpm (38 L/min) (Case 6, Table 13) predicted a composite drawdown
of 19.8 ft (6.0 m), a reasonable range for the Western Processing Site.
After calibration of the CFEST model it was determined that the most
reasonable hydraulic conductivity for the sand and gravel layer was
2 5 ft/day (0.8 m/day), which corresponds to a transmissivity of 75 ft /day
(7.0 mm/day). This transmissivity was used in all subsequent analytical
calculations (Cases 7 through 13).
54
-------
55
-------
TABLE 13. INPUT VALUES USED IN THE ANALYTICAL CALCULATION.
Case
Withdrawal
Rate
Injection
Rate
Drawdown
(ft)
T
(ft2/day)
Years
T ime
1
30
-
- 59.4
850
25
2
30
-
- 58.6
850
20
3
30
-
- 56.0
850
10
4
30
-
- 47.3
850
1
5
30
-
- 37.8
850
0.08*
6
10
-
- 19.8
850
25
7
10
-
-190.0
75
25
8
10
3.3
- 46.7
75
25
9
10
5.0
+ 27.3
75
25
10
10
0.5
-168.3
75
20
11
3
3.3
+ 86.0
75
25
12
3
1.0
- 13.4
75
25
13
3
0.05/0.4**
- 46.9
75
25
**30 days
*0.05 gpm injected at ditch
0.4 gpm injected at Mill Creek
56
-------
Case 7 in Table 13 shows that the drawdown calculated using the 10 gpm
(38 L/min) pumping rate with the lower transmissivity value was calculated to
be 190 ft (58 m), obviously too large a value. A review of the CFEST data
showed that the creek discharge decreased slightly once the remedial action
wells began pumping. Based on this observation, a series of 20 injection
wells were sited along the creek and ditch to the east and west of the site
to simulate recharge to the surface aquifer. Three runs were made varying
the rate of injection at each well from 5.0 to 0.5 gpm (19 to 2 L/min). Case
9 showed that the potentiometric surface rose above the ground surface if 5.0
gpm (19 L/min) were injected at each of the wells. The other injection rates
resulted in too much drawdown. Case 11 was run with a withdrawal rate of 3
gpm (11.3 L/min) per pumping well, and an injection rate of 3.3 gpm
(12.5 L/min) at each creek and ditch well. Case 12 was similar but had an
injection rate of 1.0 gpm (3.8 L/min) at the creek and ditch injection nodes.
The 3.3 gpm (12.5 L/min) injection rate raised the surface of the water table
above the land surface. The 1.0 gpm (3.8 L/min) injection rate created a
drawdown of 13.4 ft (4.1 m) which was the closest match to the CFEST
predicted drawdown of 10.5 ft (3.2 m) with a 3 gpm (11.3 L/min) pumping rate.
The CFEST model runs showed that about 4.5 gpm (17 L/min) would be lost
from the creek and ditch if the remedial action wells were pumped at 3 gpm
(11.3 L/min). The creek would account for about 95% of the loss and the
ditch would account for about 5%. Case 13 shows the analytical program
solution using a 3 gpm (11.3 L/min) pumping rate at each of the wells, 0.05
gpm (0.2 L/min) injected at each ditch well, and 0.4 gpm (1.5 L/min) injected
at each creek well. The predicted drawdown of 46.9 ft (14.3 m) is about five
times greater than the 10.5 ft (3.2 m) predicted by the CFEST model.
ANALYSIS OF THE ANALYTICAL PROGRAM RESULTS
The analytical solution used on the hand-held calculator proved to be a
useful tool when used in conjunction with the CFEST code for remedial action
analysis. The calculator could be used to predict the effects of various
pumping rates before model runs were made so that the modeler could avoid
some of the scaling runs that are often required prior to choosing a
57
-------
particular pumping rate. In this task, the hand-held calculator program
showed that both the 30 gpm (113 L/min) and the 10 gpm (38 L/min) withdrawal
rates caused excessive drawdowns before running CFEST.
For this study, the analytical program was used in conjunction with the
model. The input parameters used in the program were taken from the model
calibration runs rather than from the raw field data. The interaction
between the hand-held calculator and the CFEST model was complementary. The
model could take into account the effects of a multi-layered anisotropic
system, and the calculator could be used to make estimates of the model
response given the calibrated parameters from the model.
The hand-held calculator program was helpful when used in conjunction
with CFEST but it is not adequate for independent solutions of a problem such
as the one at the Western Processing Site. The effects of surface water
bodies could only be estimated and the effects of the multi-layered aquifers
could not be duplicated. The CFEST predicted drawdown from the remedial
action wells could be duplicated, but only with what is believed to be an
unrealistic injection rate at the creek and ditch. The final analytical
solution to the problem using the calibrated CFEST parameters resulted in
over 46 ft of drawdown, greater than the thickness of the aquifer being
pumped.
The analytical solutions are suitable for evaluating simple hydrologic
systems or systems where limited data require simplifying assumptions.
Complex models like CFEST are more appropriate for handling complex data sets
(i.e., multiple hydrostratigraphic layers with different hydraulic
conductivities and porosities, stream/aquifer interactions, variable depths
of pumping wells, etc.) such as exists at the Western Processing Site.
58
-------
REFERENCES
Brown, S. M. 1984. Remedial Action Modeling Volume 2. Simplified MpthnHc
for Subsurface and Waste Control Actions. Anderson-Nichols & Co., Inc., Palo
Alto, CA.
Dunne, T., and L. B. Leopold. 1978. Water in Environmental Planning, w. H.
Freeman & Co., San Francisco, CA.
Ellis, R. 1984. Personal Communication. Soil Conservation Service
Spokane, WA.
Environmental Protection Agency. 1984. Western Processing Alternativpc
Assessment Study, 1983 Data. Environmental Services Division, EPA Region X
Seattle, WA.
Environmental Protection Agency. 1983. Investigation of Soil and Uatav
Contamination at Western Processing, King County, Washington. Environmental
Services Division, EPA Region X, Seattle, WA.
Environmental Protection Agency. 1980. "Water Quality Criteria Documents-
Availability." Federal Register, pp. 45, 231, 79318-79379, Friday,
November 18.
Gupta, S. K., C. T. Kincaid, P. R. Meyer, C. A. Newbill and C. R. Cole.
1982. A Multi-Dimensional Finite Element Code for the Analysis of CmmioH
Fluid, Energy and Solute Transport (CFEST). PNL-4260. Pacific Northwest
Laboratory, Richland, WA.
Gupta, S. K., C. R. Cole and F. W. Bond. 1979. Methodology for Rpioacn
Consequence Analysis — Part III, Finite-Element Three-Dimensional Ground.
Water (FE3DGW) Flow Model. Formulation Program Listing and User's Manual.
PNL-2939, Pacific Northwest Laboratory, Richland, WA.
Luzier, J. E. 1969. "Geology and Ground-Water Resources of Southwestern
King County, Washington." Water-Supply Bulletin No. 28, Department of Water
Resources, Olympia, WA.
59
-------
National Oceanic and Atmospheric Administration. 1974. Climates of the
United States - Volume II - Western States. U.S. Department of Commerce,
Washington, D.C.
Richter, R. 0. 1981. Adsorption of Trichloroethylene by Soils from Dilute
Aqueous Systems. Contract No. F49620-79-C-0038, Air Force Engineering and
Services Center, Environics Division, Environmental Engineering Branch.
U.S. Department of Agriculture. 1973. Soil Survey of King County Area.
Washington. U.S. Government Printing Office, Washington, D.C.
van der Heijde, P. K. M. 1983. "Theis Condition Well Field." HP-41C
Program Package, Holcomb Research Institute, Butler University, Indianapolis,
IN.
Verschueren, K. 1977. Handbook of Environmental Data on Organic Chemicals.
Van Nostrand Reinhold Company, New York, NY.
60
-------
APPENDIX A
RECHARGE CALCULATIONS
Recharge due to precipitation was calculated using the water balance
formula:
Recharge = Precipitation - Actual Evaporation - Runoff (A-l)
Average annual precipitation and actual evapotranspiration for the study area
are about 39 in./yr (99 cm/yr) and 18 in./yr (46 cm/yr}, respectively (NOAA,
1974). Runoff was calculated using a method developed by the U.S. Soil
Conservation Service and modified by Dunne and Leopold (1978). The technique
is based on a simplified infiltration model of runoff, daily precipitation
events, and empirical approximations which consider such factors as soil
type, land use, vegetative cover, and storm separation interval to determine
the antecedent soil moisture conditions.
A program developed at Battelle and based on the Soil Conservation
Service method was used to calculate runoff for the Western Processing study
area. The calculations were made using daily precipitation data for 1982 and
1983. The output from the program is a list of runoff estimates for a range
of runoff curve numbers. A runoff curve number of 70 was selected for the
study area based on the soil type (Group B), land use (residential area with
one acre lots), and total impervious area (20%) (Dunne and Leopold, 1978).
The curve number 70 converts to a curve number of 85. for normally wet
antecedent moisture conditions which is the case for the area around Kent.
The results of the model for the two runoff curve numbers at several
storm separation intervals for the year 1982 and 1983 are shown in Table A-l.
Using equation A-l, averaging the runoff for two years and storm
separation intervals of one day and two days results in an estimated recharge
of about 8 in./yr (20 cm/yr).
61
-------
TABLE A-l. RUNOFF PROGRAM RESULTS
Storm Separation
Interval
(days)
Runoff (in,
./yr)
1982
1983
CN 70
CN 85
CN 70
CN 85
0
0.3
2.5
0.3
2.9
1
6.8
14.8
7.6
17.3
2
7.9
16.9
16.0
25.1
3
11.4
20.5
19.2
28.1
CN = Curve Number
In the final calibrated model a recharge value of 10 in./yr (25 cm/yr)
was applied uniformly over the local model region except for two areas on
the western Processing Site. In the area of the pond (Elements 90, 91,
101 102 103, 112, 113, and 114) the recharge was increased to 22 in./yr
(56*cm/yr), and where the site is asphalted (Elements 128, 129, 130, 143,
144, 145, 146, 152, 153, 154, 163, 164, and 175) no recharge was assumed (0
in./yr).
62
-------
APPENDIX B
STREAM AND LEAKANCE BOUNDARY CONDITIONS
The boundary conditions in the FE3DGW model were defined using the
stream boundary options to describe flux to Mill Creek and the ditch, and the
leakance boundary option to describe flux across the perimeter boundaries.
This appendix provides a more detailed discussion of the data used in the
model to implement these options.
STREAM BOUNDARY OPTION
Surface water bodies are often expressions of the water table and can be
treated as such by holding the groundwater elevation at the level of the
surface water in a groundwater model. This is not always the case, however,
and the stream option in the FE3DGW code allows the potential to fluctuate
above or below a stream, and calculates a flux (to or from the stream) based
on the potential difference between the elevation of the stream and that of
the qroundwater. The data required by the model to make this calculation
are- the stream surface elevation; the stream bottom elevation, cross-
sectional area, thickness, and permeability; and minimum stream depth
These data were entered into the model for each node along Mill Creek
and the drainage ditch. The model calculates the flux to (gaining) or from
(losing) each node using Darc/s Law. The data used to implement the stream
option in the final calibrated n»del for Mill Creek and the ditch east of the
site are provided in Tables B-l and B-2, respectively.
The surface water elevation at nodes along Mill Creek and the ditch were
interpolated and extrapolated from measurements at five locates along the
creek and two along the ditch (Figure 7). The measurements were made on April
10, 1984 by EPA Region X.
63
-------
table B-l. Stream Boundary Option Data Used to
NODE CREEK CRfEK CREEK
NUMBER FLEVATinM length wioth elevation
2 JJ.«6 246.0 5.0 10.86
\\ n.Ro 369.o s.o io.a
22 ll'll f79*0 S*° 10,75
11,70 Zlb.fi 5.0 10.70
35 W'Vx a'46*0 5,0 l0'67
11.63 2^6.0 5,0 10.63
JI \\'h\ *48.0 5.0 10.61
55 JJ'SJ 9B*° 5*° 10.60
' 11.59 tao.o S.o 1^.59
IP? W'lL *51.0 5.0 10.57
tl«565 100.0 5.0 10 56^
;w jj.hb mo.. 5;.
!Ia 11.55 74,0 5.0 10.55
\Vy \\'l? 50'° 5-° 10.52
2 '1*5 82.° S.o 10 5
r; ^.o s.o ll'to
204 tt.4« 113.0 5.0 10*48
V« »".« 5!o \l*M
3»o 1,*27 IA5.0 5.0 10.27
251 215.0 5.0 10.07
2"! 10.87 246.0 5.0 9.87
9.67
9.47
9.35
9.3
<>.27
9.1fl
9.1
9.05
9.21
9. 1 7
9.13
9.09
9.04
9.01
8.98
263 10.67 295,0 5*.0
275 10.47 295.0 5.0
287 10.35 3«0.0 5.0
288 10,3 277.0 5.0
289 10.27 295.0 5.0
290 10,1* 32#.0 5.0
JOl 10.1 560.0 5.0
J09 10,05 215.0 5.0
286 10.21 1*0.0 5.0
285 10.17 23:>.0 5.0
272 10.13 262.0 5.0
271 10.09 32^.0 5.0
27o 10.04 328.0 5.0
2«2 10.01 360.0 5.0
293 V.98 1^0,0 5.0
Simulate Flux to Mill Creek.
C«EEK BOTTOM ————— MIN CREEK
THICKNESS PERMEABILITY DC*™
0,1 0.142 0.25
0,1 0.142 0.25
0,1 0,142 0.25
0,1 0.142 0.Z5
0,1 0.142 0.25
0,1 0.142 0.25
0,1 0.142 0.25
0,1 0.142 0.25
0,1 0.142 0,25
0,1 0,142 0.25
0.1 0.142 0.25
0,1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0,142 0.25
0,1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0,1 0.142 0.25
0.1 0.142 0.25
0,1 0,142 0.25
0.1 0.142 0.25
0,1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
0.1 0.142 0.25
O.t 0.142 0.25
0,1 0.142 0.25
-------
TABLE 8-2. Stream Boundary Option Data Used to Simulate Flux to the Ditch.
NODE
CPEEK
CREEK
CREEK
• C«EEK BOTTOM
-mmmmmmmmmmm
MIN CREEK
NUMBER
EtEVATIOH
length
WIOTH
ELEVATION
thickness
PERMEABILITY
depth
54
13,57
66.0
2.0
12.57
o.i
0,142
0,25
65
13.52
1 25.0
2.0
12.52
0.1
0.142
0,25
62
13,49
10ft.o
2.0
12. 49
0,1
0,142
0,25
99
13,15
*2.0
2,0
12.45
0,1
0.142
0,25
111
13,42
57.0
2,0
12.42
0.1
0,142
0,25
115
13,a
57.0
2,0
12.4
o.i
0,142
0.25
129
1 3.39
66.0
2.0
12.39
0.1
0,142
0,25
101
13.37
70.0
2.0
12.37
0.1
0.142
0,25
157
13.3a
66.0
2.0
12.34
0.1
0.142
0,25
166
13.32
57.0
2.0
12.32
0.1
0.142
0,25
l«4
13.3
57.0
2.0
12.3
o.i
0,142
0,25
193
13.15
73.0
2.0
12.15
0.1
0.142
0.25
209
13,0
62.0
2.0
t2.0
o.i
0,142
0.25
219
12. A
125.0
2.0
11.6
0.1
0,142
0,25
215
12.6
73.0
2.0
11.6
0,1
0,142
0.25
-------
LEAKANCE BOUNDARY OPTION
The "leakance boundary condition" option of the FE3DGW code allows
flexibility in defining external boundaries of the model region. Rather than
specifying a constant flux or held potential at the boundary, the leakance
option combines the two and allows the potential and flux to vary depending
on the conditions which exist within the study area.
The data required by the model to make this calculation are: the
distance from the boundary to a known potential; the potential at that
distance; and the cross-sectional area of the boundary. These data are
entered into the model for each node along the boundary (both surface nodes
and nodes at depth). The model calculates a boundary flux at each node using
Darcy's Law, which is in turn used to calculate the potential at the
boundary.
A map depicting the regional wells used to calculate the groundwater
potential at certain distances from the boundaries is shown in Figure B-l.
This map shows the distances to the extended boundary and the gain or loss in
potential elevation out to these distances.
The Green River elevations were interpolated from three measurements
taken by EPA Region X in April, 1984 :
1. east of benchmark 32 (southwest of site) - 9.8 ft AMSL (3.0 m);
2. east of benchmark 22 (west of site) - 8.8 ft AMSL (2.7 m); and
3. Tukwilla Gauge (north of site) - 7.9 ft AMSL (2.4 m).
66
-------
. w
C
67
-------
APPENDIX C
CALCULATION OF RETARDATION FACTOR
The retardation factor, K, can be calculated by the formula
K = 1 + B Kd (c_1)
where B is defined as the bulk density divided by the porosity and Kd is the
adsorption coefficient. Kd's for TCE have been reported in the range of 0.1
to 1.0 cm3/gm depending on the soil type. A Kd of 0.2 cm /gm was used to
represent the silty sand material at the Western Processing Site.
Using a bulk density for silty sand of 2.4 gm/cm and a porosity of 15*.
yields a value of 16 for B. Substituting B and Kd into equation C-l yields a
K of 4.2, which means that the TCE travels about four times slower than the
groundwater.
The capability to implement areally distributed values for K and
different K values for different layers (material types) in the model would
certainly be realistic. Presently, only a single value of K can be used to
represent the entire model region in the CFEST code. Therefore, a value of
4 0 «as used as the K throughout the model of the Western Processing Site.
68
-------
APPENDIX 0
MODEL CALIBRATION
About 35 computer simulation runs were made in the groundwater flow
model calibration process. During this process the difference between the
model-predicted and measured hydraulic potentials were minimized by adjusting
the model parameters which were least well known. Because a range of
parameters was tested, the calibration process can be considered as a
sensitivity analysis. If certain input parameters are changed too severely,
the difference between the model-predicted and measured potentials increases
dramatically. This process provides a range of reasonable values for the
model input parameters.
A few of the model calibration runs are discussed in this appendix to
demonstrate the sensitivity of the model to the primary calibration
parameters of horizontal and vertical hydraulic conductivity and stream flux.
HYDRAULIC CONDUCTIVITY
A range of horizontal hydraulic conductivities (Kh) between 30 ft/day
(9 m/day) and 0.3 ft/day (0.1 m/day) were tested in the model during the
calibration process. When the Kh was large, the resulting potential surface
did not depict the groundwater mound that has been observed on-site
i Kh wa<; set at the low value, the potential
(Figure D-l). Conversely, when the Kh was set at
• 1C chnwn in Fiaure D-2. A Kh of 2.8 ft/day
surface increased significantly as shown in ngure u ' *
(0.9 m/day) over the model area and 0.3 ft/day (0. m ay in e op t
3 m) on site produced the best match with observe potentials (Figure 2).
Vertical hydraulic conductivities (Kv) in the range 1/10 to 1/100 he
value of the horizontal hydraulic conductivity were teste in the 0 l
(Kv/Kh = 1/10 or 1/100). As was the case for changes Kh, if the Kv is too
large it allows most of the water to infiltrate and the resulting potential
surface does not depict the groundwater mounding on site. Similarly, ,f the
69
-------
-------
-------
Kv is too low, the water cannot infiltrate and the potential values increase
significantly. A calibration run where the Kv/Kh = 1/100 is shown in
Figure D-3. It can be seen that the water cannot infiltrate and potential
values increase dramatically. A Kv/Kh = 1/20 over the entire local model
area provided the best results and was used in the final calibrated model.
Thus, the final Kv was set at 0.14 ft/day (0.04 m/day) over the model region
with the exception of the top 10 ft (3 m) on site, where a Kv of 0.014 ft/day
(0.004 m/day) was used.
GROUNDWATER FLUX TO THE STREAM
The amount of groundwater flux to Mill Creek and the ditch was primarily
controlled by the values set for the stream bottom thickness and
permeability. It was decided to set the stream bottom permeability equal to
the vertical hydraulic conductivity, thus the principal variable was the
stream bottom thickness. Values between 0.05 ft (0.015 m) and 5 ft (1.5 m)
were tested in the calibration process. Using the large stream bottom
thickness, the water was not able to discharge to the creek and ditch, and
the resulting potentials were much too high (similar to Figure D-2). It was
found that a thickness of 0.1 ft (0.03 m) provided the best results.
72
-------
Figure D-3. Model-Predicted Water Table Surface for Kv/Kh - 1/100
73
-------
DISCLAIMER
The information in this document has been funded wholly or in part by
the U. S. Environmental Protection Agency under Contract No. 68-03-3116, Work
Assignment No. 21 to Anderson-Nichols and Co., Inc. with Battelle Project
Management Division, Office of Hazardous Waste Management as the principal
subcontractor. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
-------
ABSTRACT
Western Processing Hazardous Waste Site consists of 13-acres near Kent,
Washington, which operated as an industrial waste recycling facility from
about 1960 through 1982. During 1982, the U.S. Environmental Protection
Agency (EPA) conducted surface water sampling around the site and found 26
priority pollutants; all of which were subsequently found on site. As a
result of these findings and subsequent studies, the EPA initiated several
studies to characterize the site and evaluate remedial action alternatives.
One of the efforts initiated by the EPA was to develop groundwater flow
and contaminant transport models of the site to be used in evaluating
proposed remedial actions. The development and calibration of these models
and their use in evaluating remedial actions is discussed in this report.
A conceptual model of the study area was formulated based on the
available hydrogeologic and contaminant data. The conceptual model formed
the framework for developing the groundwater flow and contaminant transport
models of the area around the Western Processing Site.
Once calibrated, the model was used to evaluate the effectiveness of six
proposed remedial actions: 1) no-action, 2) source removal, 3) cap, 4)
source removal combined with a cap, 5) upgradient slurry wall combined with a
cap, and 6) pump and treat. Of these potential actions, pump and treat
produced the most favorable results in the simulation. Considering only
passive remedial actions the simulated source removal case produced the best
results.
The results of the remedial action simulations are preliminary in nature
because the model was run only once for each scenario; optimization runs and
sensitivity analyses were not performed. It is recommended that as more data
become available, the models be further calibrated and validated and that a
more thorough modeling analysis of the remedial actions be performed.
-------
ACKNOWLEDGMENT
The authors wish to thank the staff at EPA Municipal Environmental
Research Laboratory, particularly the Technical Project Monitor, Douglas C.
Ammon, for supporting this project in the important area of computer modeling
technology demonstration and application. This work was performed under Work
Assignment No. 21, Contract Number 68-03-3116, Task U, with the assistance of
Project Officer Lee A. Mulkey at the EPA Environmental Research Laboratory in
Athens, Georgia. The authors wish to acknowledge the support they have
received from Fred Wolf, EPA Region X office in making available data and
reports regarding the Western Processing Site. The efforts of the EPA Region
X, Hart-Crowser and Associates, CH2M HILL (Seattle area offices), and Gaynor
Dawson (Battelle) in helping to formulate the initial conceptual model and
providing technical assistance and review throughout the project are greatly
appreciated. Finally, the authors would like to thank Peggy Monter, Nancy
Painter, Beth Eddy, and Dick Parkhurst of the Battelle Office of Hazardous
Waste Management who were instrumental in preparing this report. Our
successful completion of this project to a great extent has been possible
because of the efforts of the people acknowledged above.
------- |