WM'02 Conference, February 24-28,2002, Tucson, AZ
A BENCHMARKING ANALYSIS FOR FIVE RADIONUCLIDE VADOSE ZONE
MODELS (CHAIN, MULTIMED_DP, FECTUZ, HYDRUS, AND CHAIN 2D)
IN SOIL SCREENING LEVEL CALCULATIONS
Jin-Song Chen and Ron Drake, Dynamac Corporation,
3601 Oakridge Boulevard, Ada, OK 74820
Zhixun Lin, ManTech Environmental Research Services Corporation
919 Kerr Research Drive, P.O. Box 1198, Ada, OK 74820
David G. Jewett, U.S. EPA, Office of Research and Development
Subsurface Protection and Remediation Division
National Risk Management Research Laboratory
919 Kerr Research Drive, P.O. Box 1198, Ada, OK 74820
ABSTRACT
Five radionuclide vadose zone models with different degrees of complexity (CHAIN,
MULTIMED_DP, FECTUZ, HYDRUS, and CHAIN 2D) were selected for use in soil screening
level (SSL) calculations. A benchmarking analysis between the models was conducted for a
radionuclide ("Tc) release scenario at the Las Graces Trench Site in New Mexico. Sensitivity of
three model outputs to the input parameters were evaluated and compared among the models.
The three outputs were peak contaminant concentrations, time to peak concentrations at the water
table, as well as those of time to exceed the contaminant's maximum critical level at a
representative receptor well. Model parameters investigated include soil properties such as bulk
density, water content, soil water retention parameters and hydraulic conductivity. Chemical
properties examined include distribution coefficient, radionuclide half-life, dispersion
coefficient, and molecular diffusion. Other soil characteristics such as recharge rate, was also
examined. Model sensitivity was quantified in the form of sensitivity and relative sensitivity
coefficients. Relative sensitivities were used to compare the sensitivities of different parameters.
The analysis indicates that soil water content, recharge rate, saturated soil water content, and soil
retention parameter, p, have a great influence on model outputs. In general, the results of
sensitivities and relative sensitivities using five models are similar for a specific scenario. Slight
differences were observed in predicted peak contaminant concentrations due to different
mathematical treatment among models. The results of benchmarking and sensitivity analysis
would facilitate the model selection and application of the model in SSL calculations.
-------
-------
INTRODUCTION
To standardize and accelerate the evaluation and cleanup of soils contaminated with radioactive
materials at sites on the National Priorities List (NPL) with anticipated future residential land use
scenarios, the U.S. Environmental Protection Agency (EPA) has developed the Soil Screening
Guidance for Radionuclides. The Guidance is aimed to provide a methodology to calculate risk-
based, site-specific, soil screening levels (SSLs), for radioactive contaminants in soil. In general,
when the radionuclide concentrations equal or exceed SSLs, further study or investigation is
needed.
Although SSL equations under certain assumptions and limitation can be used to calculate the
site-specific SSLs for surface and subsurface soils, they are simple and are used for preliminary
screening purpose. When the site conditions are more complex than the underlined assumptions
in the simple SSL equations, a detailed modeling approached is needed. Therefore, selecting a
proper model and recognizing the performance of the model are required in the process of SSL
calculation. In such an effort to provide background information for SSL calculation, EPA
conducted an evaluation of five unsaturated zone fate and transport models for radionucludcs.
The models evaluated are CHAIN, MULTIMED_DP 1.0, FECTUZ, HYDRUS, and CHAIN 2D
models. These five models are a subset of the potential models to the public, and other models
may be applicable for the SSL calculation. This study presents part of the evaluation
(benchmarking) results.
In general, different strategies can be used for model evaluation depending on the specific
objectives/goals of model evaluation. For example, to calculate the soil cleanup criteria,
Sanders (1) compared the predicted leachings of chemicals in the unsaturated soil zone under a
hypothetical environmental scenario using four unsaturated zone models — PRZM, SESOIL,
SLMI, and IMPACT. In an evaluation of three multimedia models (MEPAS, MMSOILS, and
PRESTO-EPA-CPG) used to support cleanup decision-making at hazardous, mixed, and-
radioactive waste sites, a review of process modules was conducted based on documentation, on
published reviews, and personal interviews by Moskowitz et al. (2). In a series of multimedia
benchmarking analysis for three risk assessment models — RESRAD, MMSOILS, and MEPAS,
Lanilak et al. (3) and others (4,5.6) examined mathematical constructs and assumptions,
similarities, differences of the models, and model performance for a given environmental
scenario. In addition to examining the performance of flow and transport processes, Nofziger et
al. (7) included sensitivity analysis in their evaluation of Superfund site vadose zone models.
Sensitivity analyses are considered as part of the model evaluation for the SSL calculation
because sensitivity analyses serve two purposes: (1) to evaluate the model's response to changes
in the input parameters, and (2) to quantify the likely uncertainties of the calibrated model
resulting from uncertainties associated with the input parameters, the environmental stresses, and
the boundary conditions (8,9). Thus, sensitivity analyses can provide an understanding of the
sensitive, important, and non-sensitive nature of model input parameters. Sensitive parameters
are those input parameters which produce significant changes in the model outputs of concern,
1
-------
even though the input changes may be relatively small. Non-sensitive input parameters under
certain model scenarios should not be considered as unimportant to the simulation, but merely
parameters that contribute little to the model's uncertainty. That is, sensitivity analyses provide
the user with an understanding of those input parameters that enhance the robustness of the
model and those input parameters that may lead to model uncertainties if care is not sufficiently
exercised (10).
Therefore, the objectives of this study are to evaluate the technical formulations and performance
characteristics of the five unsaturated zone models for radionuclides and to provide the
background information for the use of models in the process of the SSL calculation. The flow
and transport processes, and inputs/outputs of the five models will be reviewed and evaluated for
the capacity for simulating migration of the radionuclide. Base case simulations and sensitivity
analyses will be used to address the model performance. Comparison of the results of the base
case simulation and sensitivity analysis will be made for identifying the similarities and
difference of the model performance.
MODEL DESCRIPTION
The five models selected for this analysis address four essential processes that predominately
control the migration of radionuclides in the unsaturated zone. These four processes are
advection (derived from infiltration), dispersion, sorption, and radionuclide decay. Table 1
provides a summary of model components (processes considered) and the similarities and
differences among the models. Table 2 gives a summary of the use of the model including the
model outputs, applicability and limitations. Except for dimensionality, the HYDRUS model is
the most comprehensive code for the flow and transport in the unsaturated zone. It considers
hysteresis of soil water retention. The MULTIMEDJ3P model was initially developed as a
multimedia fate and transport model. The FECTUZ model is the unsaturated module of U.S.
EPA's composite model for leachate migration with transformation products (EPACMPT, 115
12) In general, any of the five models can be used in simulating fate and transport of
radionuclides in the unsaturated zone and can provide the time-varying concentrations of
radionuclides in the leachate entering ground water. These concentrations are used for the SSL
calculation. The detailed procedures for the SSL calculation can be found in the user's guide and
the technical background document of the U.S. EPA's Soil Screening Guidance for
Radionuclides (13, 14). The reader who is interested in detailed assumptions and mathematical
formulations of the models is referred to the specific model theory and user's manuals for
CHAIN (15), MULTIMED_DP 1.0 (16, 17,18), FECTUZ (11, 12), HYDRUS (19), and
CHAIN 2D (20).
Table 1. Summary Comparisons of the Vadose Zone Model for Radionuclides in the SSL Process
Model component HYDRUS MULTIMED-DP FECTUZ CHAIN CHAIN
2D
Contaminants
Organics • • • • •
Metals • • • • •
2
-------
Radionuclides (parent) • • • • •
Radionuclides (progeny) • • • • •
Sources types
Contaminated soil • • • • •
Landfill • • • -- •
Surface impoundment — -- •
Waste piles - -- •
Source term characteristics
Mass balance • • • • •
Multimedia partitioning * * • -- •
Source decay — *
Multiple contaminants per simulations • • • • •
Source release mechanisms
Leaching • • • • •
Direct release to:
Vadose zone • • • • •
Groundwater • •
Surface water -- •
Air - •
Medium-specific flow
Surface Hydrology
Precipitation -- •
Runoff - •
Infiltration • • • • •
ET • • •
Surface Water (Stream discharge) -- •
Vadose Zone
Vadose zone (Steady-state infiltration —>soiI • • • • •
Vadose zone (n-D dynamic) • - - - •
Groundwater -- — •
Medium-specific contaminant transport
Atmosphere (emission through diffusion) — •
Surface water (stream interception and mixing) — • ....
Vadose zone (1-D advection and dispersion) • • • • •
Vadose zone (2-D advection and dispersion) -- -- - - •
Groundwater
Homogeneous aquifer (1-D advection and - • •
Homogeneous aquifer (2-D advection and • •
Homogeneous aquifer (3-D advection and — * •
Medium-specific heat transport • — „ •
Contaminant transformations and fate processes
1st order decay (not decay products) • • • • •
1st order decay (with chained daughter and
granddaughter decay products) -straight chain • • • • •
1 st order deeay — branch chain • — • •
Non-1st order decay • .. •
Linear sorption (partitioning between water and • • ¦ • •
3
-------
Nonlinear sorption (partitioning between water and
•
„
•
•
Nonequilibrium sorption
•
„
"
«
Hydrolysis
-
•
•
~
—
Chemical reactions/speciation
—
•
•
—
—
Intermedia contaminant fluxes
Surface soil Air (volatilization)
•
•
»
•
•
Surface soil ~> Vadose zone (leaching)
•
•
•
•
•
Surface soil Overland (erosion, runoff)
"
•
-
-
-
Surface water ---Sediment (sedimentation)
-
Vadose zone --> groundwater (percolation)
•
•
•
t
•
Vadose zone —> Air (volatilization)
•
•
—
•
• denotes component is included in model; — denotes component is not included in model,
Table 2. Summary of the Use of the Unsaturated Zone Model for Radionuclides in the SSL Process
Model Processes, components, outputs
HYDRUS - Provides the leachate radionuclide concentrations entering the ground water so whether the
resulting concentration of the radionuclide at the receptor well would exceed the acceptable
level or not, can be examined
- calculates infiltration which can be used as inputs in the SSL calculation
- considers soil heterogeneity, nonlinear/nonequilibrium sorption, time-varing infiltration and
evapotranspiration
- considers hysterisis of soil water retension
- outputs radionuclide concentration in soil, cumulative flux across water table
- grid discretization for HYDRUS version 6,0 requires extra effort
MULTIMED DP - provides the leachate radionuclide concentrations entering the ground water so whether the
resulting concentration of the radionuclide at the receptor well would exceed the acceptable
level or not, can be examined
- uncertainty of model outputs can be examined
- considers runoff, evapotranspiration
- linked with a saturated flow and transport model
- requires a great amount of input data, expertise because of model complexity
FECTUZ - provides the leachate radionuclide concentrations entering the ground water so whether the
resulting concentration of the radionuclide at the receptor well would exceed the acceptable
level or not, can be examined
- uncertainty of model outputs can be examined
- linked with a saturated flow and transport model
- uses mixed units for the input data
CHAIN - provides the leachate radionuclide concentrations entering the ground water so whether the
resulting concentration of the radionuclide at the receptor well would exceed the acceptable
level or not, can be examined
- used for simplified radionuclide-contaminated site scenario
- simple-to-ues, less input data requirement
- as a preliminary assessment tool in SLL estimation
4
-------
CHAIN 2D - provides the leachate radionuclide concentrations entering the ground water so whether the
resulting concentration of the radionuclide at the receptor well would exceed the acceptable
level or not, can be examined
- calculates infiltration which can be used as inputs in the SSL calculation
- considers soil heterogeneity, nonlinsar/nonequilibnum sorption, time-varing infiltration and
evapotranspiration
- outputs radionuclide concentration in soil, cumulative flux across water table
- considers two-dimensional soil heterogeneity
METHODOLOGY
Conceptual Site Model for Radionuclide Leaching
To apply the five models for radionuclides evaluated herein in the SSL calculation, a conceptual
site model was developed at the Las Graces Trench Site, New Mexico, USA, The Site is in the
Chihuahuan Desert and on a basin slope of Mount Surnmerford. Climate in the region is
characterized by low relative humidity and an average class A pan evaporation of 239 cm per
year. Average annual precipitation is 23 cm. The site has been subject to extensive testing of
physical and chemical soil properties and water movement in the unsaturated zone. In addition,
results of tracer tests (chloride, bromide, and tritium) are available at the Site (21). It is assumed
that the Site had been used as a waste disposal/storage facility where radionuclides from tank
leaks or improper waste disposal were released to the soil surface for 1000 days with a total
amount for 3x104 mg /cm "Tc ("Tc concentration from the waste source is 1.25 x 10'2 mg/L).
Rainfall infiltration (with a net annual recharge rate of 87 mm/y) is the driving force for the
downward migration of radionuclide to the water table beneath. Base values of the input
parameter are given in Table 3. Uniform soil properties and solute transport parameters in Table
3 are obtained from the layered data of Wierenga et al. (21) and Porro and Wierenga (22),
respectively. It is assumed that steady-state uniform water flow occurs at the site. At the time
that source release ceases, the soil in the top 150 cm depth approximately contains 1.18 x 10~3
mg/kg of "Tc. The decay coefficients and the distribution coefficient for "Tc and its daughter
"Ru are taken from U.S. EPA (14).
The dispersion coefficient D in the unsaturated zone is characterized by a molecular diffusion
and a mechanical dispersion and is given by Hills et al. (23) as
where 8 is the initial water content of 0.16 cmVcm3, q is the infiltration rate of 8.7 cm/y, DL is
D=WD4 CEq. 1)
the dispersivity of 4,53 cm, tw is the tortuosity factor, and Dw is the diffusion coefficient in free
water of 1.73 cm2/d. The tortuosity factor of 0.19 is calculated from 0 as Tomasko et al, (24).
Consequently, the dispersion coefficient, D, is given as 0.33 cm2/d.
5
-------
Table 3. Values of Model Input Parameters
Parameters
Values
Source-Specific Parameters
Area of disposal facility (m2)
400
Width of disposal facility ( m)
20
Length of disposal facility (m)
20
Mass release of Radionuclide "TC (mg/cm2)
3xl0"4
Concentration of "Tc in recharge water from waste source (rrig/L)
l.25xl0"2
Duration of waste source being completely released (days)
1000
Potential recharge rate (mm/d)
0.024
Initial water content (em3/cm3)
0.16
Soil Properties in Unsaturated Zone
Saturated hydraulic conductivity, Ks, (cm/d)
270.1
Porosity («)
0.358
Saturated water content (cm3/cm3)
0.321
Residual water content (cm3/cm3)
0.083
Bulk density (g/cm3)
1.70
van Genuchten alpha coefficient, a, (cm*1)
0.055
van Genuchten beta coefficient, (3 (--)
1.509
Depth to water table (m)
6
Solute Transport Parameters
Decay coefficient for parent product "Tc (1/d)
9x10-9
Decay coefficient for daughter product "Ru (1/d)
stable
Distribution coefficient for parent product "Tc (cm3/g)
0.007
Distribution coefficient for daughter product "Ru (cm3/g)
5
Dispersion coefficient (cnr/d )
1.01
Dispersivity (cm)
4.53
Diffusion coefficient in free water (cm2/d)
1.73
Apparent molecular dispersion coefficient (cm2/d )
0.33
6
-------
Model Outputs of Interest
Using the base values of the input parameters (Table 3), the time evolution of the radionuclide
"Tc concentration in the leachate at the bottom of the unsaturated zone (i.e., at the entry point to
the ground water table, 6 m below the ground surface) is obtained and as shown in Figure 1.
The solid line in Figure 1 represents the typical breakthrough curve (BTC) for the base scenario
of the conceptual model using the CHAIN model.
In the SSL process, the concentration at a receptor well is assumed to exceed the Maximum
Contaminant Level (MCL) whenever the concentration at the ground water entry point exceeds
the MCL times a dilution attenuation factor (DAF). A value of 20 for DAF is proposed in U.S.
EPA (14). When this occurs, the concentrations of radionuclide in soil in mass units of mg/Kg
exceed SSL. The MCL for "Tc is 5.3 x 105 mg/L (14). Therefore, three characteristics of the
BTCs are of interest in this study:
1. The peak concentration of the radionuclide, Cpeak,
2. The time to reach peak concentration, Tpeak, and
3. The time when the concentration of radionuclide is high enough so that the
resulting concentration at a receptor well will exceed the Maximum Contaminant
Level (MCL). The time to reach MCL is denoted by TMCL.
In Figure 1, the arrow sign indicates a time point where "Tc concentration / DAF exceeds the
MC, i.e. TMCL. Similarly, the peak concentration of wTc, C^, and the time to reach peak
concentration can be read from the y-axis (concentration axis) and x-axis (time axis)
respectively.
Sensitivity Analysis
The sensitivity of a model is a measure of the change in a selected model output resulting
from a specified change in an input parameter. Mathematically, the sensitivity coefficient
Sy of a model's output y^ to a model's input parameter xj is the partial derivative of yj with
respect to Xj, while holding other pertinent input parameters fixed, and it is expressed as:
dy-
s« = (Eq'2)
Here, the model's output, yj, could be one of the three output quantities: Cp^, Tpeak and TMCL
The model's input parameter, xj, could be any one of the model input parameters given in Table
3. In the current analyses, only 12 parameters are included in the sensitivity analysis. They are
the distribution coefficient for parent product "Tc, recharge rate, initial water content, bulk
density, dispersion coefficient, dispersivity, diffusion coefficient in free water, saturated
hydraulic conductivity, saturated water content, residual water content, van Genuchten alpha
7
-------
coefficient, a, and van Gcnuchtcn beta coefficient, p.
The dimensions of Sy are those of y^ divided by those of Xj. Since Sjj> in general, is dimensional,
it may be difficult and confusing to compare sensitivities for different input parameters. These
problems are overcome by introducing a normalized form of Sy, called the relative sensitivity
coefficient, defined by:
By definition, Sfy is dimensionless, and comparison of the sensitivity coefficients between two
input parameters can be made. For example, Sr for Tpeak to recharge rate is -1.0, and Sr for Tp;ak
to initial water content is 0.80. Then Tpeak is more sensitive to the recharge rate than to initial
water content. Note that the comparison is made on the absolute values of two relative
sensitivity coefficients. Negative Sr value for recharge rate indicates that Tpeak decreases with an
increase in recharge rate.
Procedures Followed in the Sensitivity Analyses
The procedures for conducting a sensitivity analysis are as follows;
1. Select a model for analysis;
2. Select the model outputs of interest, namely Cpeak, T^, and TMCL;
3. Select a particular variable input parameter and select a range for this parameter;
4. Run model simulations using the base parameter values except for the one of the
variable input parameter;
5. Calculate the sensitivity (S) and relative sensitivity (Sr) of a model output to the various
input parameters following the approach (Equations 3 and 4) described above.
The procedures given above represent a scenario, and they are repeated for each input parameter
considered in each model.
RESULTS AND DISCUSSION
Impact of Input Parameters on BTC
Figure 1 gives the breakthrough curves of "Tc at the hypothetical water table located at 6 m
below the ground surface. The solid curve represents the breakthrough curve of "Tc using the
base value of 0.024 cm/d for recharge rate using the CHAIN model. In the sensitivity analysis
for recharge rate, a range of recharge rates around the base values (0.024 cm/d) were used to
obtain the BTCs. For example, the dotted curve in Figure 1 represents the breakthrough curve of
"Tc using the base values in Table 3 except using a lower bound of 0.014 cm/d for recharge rate.
In this manner, the impacts of changing recharge rate on the BTCs can be evaluated. As shown in
Figure 1, a decrease of recharge rate results in an increase of travel time for "Tc to first reach the
entering point (water table) and an increase of the residence time of "Tc in the unsaturated zone.
8
-------
0.010
Recharge Rate (cm/d)
0.014
0.024
o>
£
W
0.008
o 0.006
w*mm
o
§ 0.004
c
o
u
u 0.002
I—
m
m
5000
10000
15000
0
Time (days)
Fig. 1. Sensitivity of "TC breakthrough curve (through the unsaturated zone) to
the recharge rate using the CHAIN model.
The residence time is the time lag between the time where the source was released and the time
where "Tc is completely leached out of the unsaturated zone. For the three outputs of interest,
Cpeak increases with an increase in recharge rate; Tpeak and TMCL decrease with an increase in
recharge rate.
When Cpeak was plotted in correspondence to recharge rate, a curve as given in Figure 2(a) was
obtained. Figure 2(a) indicates an increase of Cpeak with an increase in recharge rate. The
sensitivity of Cpeak to recharge rate, as shown in Figure 2(b), can be obtained using equation 2.
Similarly, the relative sensitivity of Cpeak to recharge rate as shown in Figure 2(c) can be obtained
using equation 3. Central finite difference approximation was used for the calculations of
sensitivity and relative sensitivity. According to equations 2 and 3, relative sensitivity has the
same signs as sensitivity. In this manner, the sensitivities and relative sensitivities of Cpeak, Tpeak
and TMCL to twelve input parameters using the five models were obtained.
Table 4 summarizes the relative sensitivities of Cpeak, Tpeak and TMCL at the base scenario (i.e.,
using the base values as given in Table 3). For all of the models, the sensitivities with respect to
distribution coefficient for parent product "Tc, recharge rate, initial water content, bulk density,
dispersion coefficient, dispersivity, and diffusion coefficient in free water were obtained under
the constant, uniform water content (recharge rate) conditions. The sensitivities with respect to
saturated hydraulic conductivity, saturated water content, residual water content, van Genuchtcn
alpha coefficient, a, and van Genuchten beta coefficient, p, were obtained under the nonuniform
9
-------
00 0.20
tr 0.10
0.15
w
c
w 0.05
0.010
^ 0.008
W 0.006
£ 0.004
o
o
v 0.002
0.01 0.02 0.03 0.04
Recharge Rate (cm/d)
0.05
Fig. 2. Sensitivity and relative sensitivity of time to reach peak concentrations at the depth of 6
m to the recharge rate using the CHAIN model.
10
-------
moisture content condition. The main reason is because uniform water content and recharge rate
are the assumption in the CHAIN model. In doing so, model comparison on sensitivity results
can be made. In the latter case, such restriction does not exist.
The results given in Table 4 indicates that
1. Similar magnitudes of relative sensitivity were observed using five models;
2. Initial water content, van Genuchten beta coefficient, p, and saturated water content are
the most sensitive input parameters to predicted Cpeaio
3. Recharge rate, initial water content, van Genuchten beta coefficient, P, and saturated
water content are the most sensitive input parameters to predicted Tpeak and TMCL;
In other words, the advection process is the predominant process for the radionuclide "To
transport in the hypothesized scenario. Note that the input parameters — recharge rate, initial
water content, van Genuchten beta coefficient, p, and saturated water content are the controlling
parameters in water flow. As the result, they significantly influence the transport of "Tc.
Conversely, sorption (a magnitude of distribution coefficient of 0.007 cm3/g is small) and
dispersion are less predominant processes.
Table 4. Summary of Relative Sensitivities for All Models, Outputs, and Input Parameters
Model Input Parameter
Z
<
0
MULTIMED-DP
1.0
s
U.
Q
z
<
5
us
3
ad
a
>
X
Z
i
u
a,
q
a
w -
2 -
5
D
Z
B
u.
8
2
HYDRUS
z
<
X
u
*
2
P
—J
D
s
FECTUZ
CHAIN 2D
HYDRUS
Distribution coefficient
-0.06
-OQ6
-005
-0 06
•0 06
+0.06
+0 06
+0.07
+0.06
+0.06
+0.C7
+0 07
+0 08
+0 07
-t-0 07
Recharge rate
+0 00
+0.00
+0.11
+0.12
• 1.00
-1.00
-0.98
-1 00
-LOO
-0 89
• I 00
-1 00
-0.98
-0 93
Initial water content
-1.16
-0.80
-0 68
-1 03
-1.10
+0.80
+0.80
+0 81
+0 78
+0 18
+0.83
+0.88
+0.92
+0,8-1
—0.95
Bulk density
-0.17
-0 06
-0.05
-0 05
-0 OS
+0 06
+0 06
+0.08
+0 0?
+0.05
+0.08
+0.06
+0 07
+0 07
+0 07
Dispersion coefficient
-0.3«
~
_
-
-
-0 02
_
-
-
-0 10
-
_
_
Dispersivity
-
-0.17
-0.36
-0.28
-0.28
-
•0,01
-0 Q2
.002
-o m
-
-0,05
-009
-008
-0 07
Diffusion coef. in water
-
—1
-
¦0 10
-0 10
-
-
-OlOI
-0.01
-
-
-
-0.03
-0 03
Saturated conductivity
-
+0 04
+0.00
*O.0S
+0 05
_
-0.05
-0 00
-0 04
-O.CM
-
-0.05
-0.00
*0 04
-0 04
Saturated water content
-
-0.60
-031
-0.66
-0.66
-
+0.58
+0.58
+0 57
+057
-
+0 64
+0.68
+0 64
+0 64
Restsual water content
-
<123
.0 20
-OA!
-027
-
+0 23
+0.22
+0 24
+0.23
-
+0.24
+0,26
+0.25
+0 24
van Genuchten alpha
-
+0 05
+0.06
+O.II
+0.14
-0.05
-0 09
-0.05
-0.05
• -
-0-05
-0 09
-0.06
-0 05
van Genuchten beta
-
-KJ9S
+O.S6
+1.30
+1.38
-
-0.96
-1 04
-LOG
-i 00
-
-1 06
-1 13
-1.04
-t 03
11
-------
Comparison of Flow and Transport in the Base Scenario
The five models, except the CHAIN model, have a module for water flow in the unsaturated zone
and have the same van Genuchten model for soil water retention. For the base scenario, a
uniform soil layer is assumed with a constant flux boundary at the top and a water table at the
depth of 6 m. The soil moisture distributions using the HYDRUS, CHAIN 2D, FECTUZ and
MULTIMED_DP models are shown in Figure 3. An error in the originally distributed
MULTIMED_DP 1.0 code was detected (solid square points). The error arose from the incorrect
use of the residual water content in the van Genuchten model for the soil water retention. When
this error was corrected, the new water content results (solid triangle) give reasonably consistent
results with the other three models.
Parallel to the comparison of simulating water flow in the unsaturated zone, the predicted BTCs
were obtained using the base values for the CHAIN, HYDRUS, CHAIN 2D, FECTUZ, and
MULTIMED_DP models. A constant and uniform soil moisture profile was assumed. The
results were presented in Figure 4. It was found that
1. All five models give very similar BTCs except that the MULTIMED_DP and FECTUZ
models predict higher peak concentrations of "Tc then those by the CHAIN, HYDRUS
and CHAIN 2D models.
2. The differences of the predicted BTCs are mainly attributed by neglecting the molecular
diffusion in the MULTIMED DP and FECTUZ model. Both models assume molecular
diffusion (the first right-hand term of equation 1) is not significant compared to
mechanical dispersion (the second right-hand term of equation 2). Using the base
values in Table 3, a resulting dispersion coefficient of 1.01 cm2/d is obtained for the
CHAIN, HYDRUS and CHAIN 2D models and 0.68 cm2/d for the MULTIMED DP
and FECTUZ models. When the dispersivity, DL, of 4.53 cm (base value) was replaced
by 6.53 cm, a value for dispersion coefficient was obtained for the FECTUZ model (the
first right-hand term is zero in equation 1). The resulting BTC using the FECTUZ model
is reasonable close to those BTCs given by the CHAIN, HYDRUS and CHAIN 2D
models. The same is true for the MULTIMED_DP model (not shown).
3. The selection of a proper inverse Laplace Transform algorithm for solving the transport
equation is important in using MULTIMED_DP model. In the MULTIMED_DP
model, both the Stehfest and DeHoog algorithm are available. A stable solution can be
obtained using the DeHoog algorithm but not Stehfest. The DeHoog algorithm is also
used in the FECTUZ model.
The results of the above model comparison also depict that similar predicted BTCs are expected
to be obtained when the models were constructed for the same essential flow and transport
processes. Minor differences may come from different numerical techniques or grid sizes.
The above analyses are given toward the benchmarking of the five models for the SSL
calculation at a hypothetical scenario. Each model has more other functions/capacities that are
12
-------
not addressed in this study. A detailed evaluation of these five models can be seen in U.S. EPA
(15) and Chen et al. (25)
0
1.0
^ 2.0
£ 3.0
Q.
©
Q 4.0
5.0
6.0
0 0.10 0.20 0.30 0.40 0.50
Water Content (cm5/cm5)
Fig. 3. Water content distributions predicted by the HYDRUS, CHAIN
2D, FECTUZ, and MULTIMED DP 1.0 models. Note that the water
contents (¦) obtained from the originally distributed MULTIMED DP
1,0 code are in an error. The corrected code gives a consistent water
content distribution (A) with the other three models.
T
¦ ii
HYDRUS
CHAIN 2D
• FECTUZ
¦ MULTIMED DP w/error
* MULTIMED DP Correction
0.015
oo
c
o
c
©
o
c
o
o
u
0.010
0.005
HYDRUS
CHAIN 2D
FECTUZ (D =6.53 cm)
CHAIN L
FECTUZ ( Dl=4 . 53 cm)
MULTIMED DP
2500 5000 7500
Time (days)
10000
Fig. 4. Comparison of the BTCs for HYDRUS, CHAIN 2D, CHAIN,
FECTUZ, and MULTIMED_DP models for the base values of the input
parameters with the BTC for the FECTJjJ£ model with the base value of
Dl = 4.53 cm replaced by the value DL = 6.53 cm.
-------
based on model capacity, sensitivity analysis and inter-model comparison for a hypothetical
scenario at the Las Craces Trench Site in New Mexico, indicate that:
Any of the five models is capable in simulating fate and transport of radionuclides in
the unsaturated zone and can provide the time-varying concentrations of radionuclide in
the leachate for the purpose of the SSL calculation;
Recharge rate, initial water content, van Genuchten beta coefficient, p, and saturated
water content are the most sensitive input parameters to predicted C^, and TMCL;
The HYDRUS, CHAIN 2D, FECTUZ and MULTIMED_DP models, in general,
provide consistent results in water flow simulation assuming the detected error in the
MULTIPED_DP was corrected;
The CHAIN, HYDRUS, CHAIN 2D, FECTUZ and MULTIMED_DP models predicted
very similar BTCs; The differences in the predicted peak concentrations of "Tc using
different models are due to the differences in the implementation of the dispersion term.
Neglecting a molecular diffusion in the FECTUZ and MULTIMED_DP model results
in a lower dispersion coefficient and higher C^s. When adjustment for dispersion
coefficient is made, all five models can predict similar BTCs for the SSL calculation.
Disclaimer
The U.S. Environmental Protection Agency through its Office of Research and Development
funded and managed the research described here under U.S. EPA Contract No. 68-C-99-256 to
Dynamac. It has been subjected to Agency review and approved for publication.
Acknowledgments
The work was performed under U.S. EPA Contract No. 68-C-99-256 to Dynamac Corporation.
The support of Dr. David Burden, the Project Officer for the Contract, is acknowledged.
REFERENCES
1. SANDERS, P.F., "Calculation of soil cleanup criteria for volatile organic compounds as
controlled by the soil-to-groundwater pathway: comparison of four unsaturated soil zone
leaching models," J. Soil Contamination, 4(1): 1-24 (1995).
2. MOSKOWITZ, P.D., R. PARDI, V.M. FTHENAKIS, S. HOLTZMAN, L.C. SUN, and B.
IRLA, "Evaluation of three representative multimedia models used to support cleanup
decision-making at hazardous, mixed, and radioactive waste sites," Risk Analysis,
16(2):279-287 (1996).
3. LANIAK, G.F., J.G. DROPPO, E.R. FAILLACE, E.K. GNANAPRAGASAM, W.B.
MILLS, S.L. STRENGE, G. WHELAN, and C. YU, "An overview of a multimedia
benchmarking analysis for three risk assessment models: RESRAD, MMSOILS, and
MEPAS," Risk Analysis, 17(2):203-214 (1999).
4. WHELAN, G., J.P. MCDONALD, E.K. GNANAPRAGASAM, G.F. LANIAK, C.S.
LEW,W.B. MILLS, and C. YU, "Benchmarking of the saturated-zone module associated
1.
2.
3.
4.
14
-------
with three risk assessment models: RESRAD, MMSOILS, and MEPAS," Environ.
Engineering Science, 16(l):67-80 (1999).
5. WHELAN, G., J.P. MCDONALD, E.K. GNANAPR AG AS AM, G.F. LANIAK, C.S. LEW,
W.B. MILLS, and C. YU, "Benchmarking of the vadose-zone module associated with three
risk assessment models: RESRAD, MMSOILS, and MEPAS," Environ. Engineering
Science, 16(1):81-91 (1999).
6. WHELAN, G„ J.P. MCDONALD, R.Y. TAIRA, E.K. GNANAPRAGASAM, C. YU, C.S.
LEW, W.B. MILLS, and B. WILLIAM, "Source-term development for a contaminant plume
for use by multimedia risk assessment models," J. Contaminant Hydrology, 41(3):2Q5-223
(2000).
7. NOFZIGER, D.L., J.-S. CHEN, and C.T. HAAN, "Evaluation of unsaturated/vadose zone
models for superfund sites," U.S. Environmental Protection Agency. EPA/600/R-93/184
(1994).
8. ANDERSON, M.P., and W.W. WOESSNER,. "Applied Groundwater Modeling -
Simulation of Flow and Advective Transport," Academic Press, Inc., New York, NY
(1992).
9. ZHENG, C., and G.D. BENNETT, "Applied Contaminant Transport Modeling: Theory and
Practice." Van Nostrand Reinhold. New York, NY. (1995).
10. WILLIAMS, J.R., Y. OUYANG, J.-S. CHEN, and V. RAVI, "Estimation of Infiltration
Rate in Vadose Zone: Application of Selected Mathematical Models, Vol. II," U.S.
Environmental Protection Agency, Washington, DC. EPA/600/R-97/128b ( 1998).
11. U.S. EPA, "EPA's Composite Model for Leachate Migration with Transformation
Products, EPACMTP," Background Document. Office of Solid Waste, U.S. EPA,
Washington, DC (1995a),
12. U.S. EPA. "EPA's Composite Model for Leachate Migration with Transformation Products,
EPACMTP,". User's Guide. Office of Solid Waste, U.S. EPA, Washington, DC (1995b).
13. U.S. EPA, "Soil Screening Guidance for Radionuclides: User's Guide," Office of Radiation
and Indoor Air and Office of Solid Waste and Emergency Response, Washington, DC
EPA/540/R00/006 (2000a).
14. U.S. EPA, "Soil Screening Guidance for Radionuclides: Technical Background Document,"
Office of Radiation and Indoor Air and Office of Solid Waste and Emergency Response,
Washington, DC EPA/540/R00/007 (2000b).
15. VAN GENUCHTEN, M. TH, "Convective-dispersive transport of solutes involved in
sequential first-order decay reactions," Computers & Geosciences, 11 (2): 129-147 (1985),
16. LIU, S., W.B. MILLS, and B. BAEN, "Multimedia Exposure Assessment Modeling
Including Fate and Transformation Products (MULTIMED_DP 1.0): Implementation, Tests
and User's Manual," Technical Report prepared by U.S. EPA, Athens, GA (1995).
17. SHARP-HANSEN, S., C. TRAVERS, P. HUMMEL, T. ALLISON, R. JOHNS, and W.
MILLS, "A Subtitle D Landfill Application Manual for the Multimedia Exposure
Assessment Model (MULTIMED 2.0)," Technical Report prepared for U.S. EPA. Athens,
GA (1995).
18. SALHOTRA, A.M., P. MINEART, S. SHARP-HANSEN, T. ALLISON, R. JOHNS, and
W.B. MILLS, "Multimedia Exposure Assessment Model (MULTIMED 2.0) for Evaluating
the Land Disposal of Wastes-Model Theory," Final Report prepared for U.S. EPA (1995),
15
-------
19. SlMUNEK, J., K, HUANG, and M.TH, VAN GENUCHTEN, "The HYDRUS Code for
Simulating the One-Dimensional Movement of Water, Heat, and Multiple Solutes in
Variably-Saturated Media," Research Report No. 144. U.S. Salinity Laboratory. USDA,
ARS. Riverside, CA (1998).
20. SIMUNEK, J., and M.TH. VAN GENUCHTEN, 'The CHAIN_2D Code for Simulating
Two-Dimensional Movement of Water, Heat, and Multiple Solutes," Research Report No.
136. U.S. Salinity Laboratory. USDA, ARS. Riverside, CA (1994).
21. WIERENGA, P.J., R.G. HILLS, and D.B. HUDSON, 'The Las Cruces Trench Site:
Experimental results and one dimensional flow predictions," Water Resour. Res.
27:2695-2705 (1991).
22. PORRO, I., and P J. WIERENGA, 'Transient and steady state solute transport through a
large unsaturated soil column," Ground Water* 31(2), 193-200 ( 1993).
23. HILLS, R.G., P.J. WIERENGA, D.B. HUDSON, and M.R. KIRKLAND, "The second Las
Cruces trench experiment: experimental results and two-dimensional flow predictions,"
Water Resour. Res., 27(10):2707-2718 (1991).
24. TOMASKI, D. M. REEVES, B.A. KELLEY, and J.F. PICKENS, "Parameter sensitivity
and importance for radionuclide transport in double-porosity systems," In: B.E. Buxton
(ed.) Proceedings of the Conference on Geostatistical. Sensitivity, and Uncertainty Methods
for Ground-Water Plow and Radionuclide Transport Modeling. San Francisco, CA.
September 15-17, 1987. Battelle Press. Columbus, OH. pp. 297-321 (1989).
25. CHEN, J-S, R.L., DRAKE, Z. LIN, and D. JEWETT, "Evaluation of Computer Models for
Simulating Radionuclide Transport in the Unsaturated Zone," EPA Report under peer
review, National Risk Management Laboratory, Subsurface Protection And Remediation
Division, Ada, OK (2001).
16
------- |