Académique Documents
Professionnel Documents
Culture Documents
tainty. For these models, parameters describ- where Pf is defined as the probability that the convolution integral of Equation 2 may
ing the transport process are represented by contaminant concentrations exceed the safe be solved directly using any appropriate
distribution functions with prescribed dose. The subscript off indicates that this is numerical integration technique.
means and variances. Therefore, calculated a probability of failure, e.g., contaminant
contaminant concentrations at pumping concentrations in groundwater exceed safe Aquifer and Contaminant
wells are described probabilistically. exposure doses. Lettingf1(r) andf1(c) define Characteristics
the distribution of safe doses and the distrib- The contaminated aquifer used in this study
Second Moment Formulation ution of contaminant concentrations, respec- is designated as Operational Unit 3 (OU3)
The second moment formulation is com- tively, the probability that contaminant con- and is located at Hill Air Force Base (AFB)
monly used by structural engineers to eval- centrations exceed the safe dose is defined as near Ogden, Utah (8). There are four hydro-
uate the probability of failure of structural 00 geological components of the aquifer; how-
elements such as beams under variable P F=(C)fC(C)dC
| (2) ever, simulations were only required for the
loading conditions (10). In the analysis of one contaminated shallow water table aquifer
beam failure under variable loading condi- where Fr(C) is the cumulative distribution directly beneath Hill AFB. The contaminat-
tions, probability distribution functions are function of fr(r) evaluated at C Figure 1 ed area of the shallow aquifer consists of
required to describe the variability in beam illustrates Equation 2. 50% lacustrine clay and 50% lacustrine
load-carrying capacity and to describe the The method is simplified if the func- sands. The thickness of the shallow aquifer is
variability in beam loading. In this study, tions for fr(r) and f1(c) are normally distrib- between 100 to 200 feet. Hydraulic conduc-
the second moment formulation is used to uted. In this case, the probability that conta- tivities range from a low of 106 ft/day to a
determine if the distribution of concentra- minant concentrations exceed the safe dose high of 102 ft/day. The average storage coef-
tions of TCE in a contaminated aquifer can be stated in terms of the means, Pr and ficient is estimated to be 0.2 with a range of
exceeds the distribution of safe doses in a py and variances, (ar and (a, of the normally 0.01-0.3 and the average retardation factor
diverse human population (10-12). In this distributed random variables and takes for TCE is 11.8 with a range of 4.5-28. In
manner, carcinogenic health risks posed by advantage of the characteristics of the stan- this study, hydraulic conductivity is modeled
drinking groundwater from a contaminat- dard normal distribution. Using reduced as a spatially varying random variable. A plan
ed pumping well can be estimated. variates of the standard normal distribution, view of Hill AFB is shown in Figure 2,
In this analysis, the concentration of a a safety index (c) is derived. Geometrically, showing the north-south orientation of the
contaminant (C; e.g., TCE in pg/l) within , is defined as the shortest distance from the contaminant plume and primary direction of
the subsurface is modeled as a random vari- origin of the reduced variates to the limit flow in the contaminated aquifer.
able with a known mean, variance, and dis- state where the safe dose concentration is
tribution. Variable safe doses are also mod- equal to the pollutant concentration. Model Simulations
eled as a random variable (1) with known Mathematically, , is defined as To illustrate the methodology presented in
mean, variance, and distribution. At any Equations 1 through 4 for evaluating the
time, the safe dose level should be greater Rr R health risks of a contaminated aquifer,
than the concentration of contaminant a r +a TCE transport behavior was analyzed with
encountered. That is R> C. Similarly, if R c (3) a two-dimensional (2-D) finite difference
< C, the contaminant concentration contaminant transport model based on the
exceeds the safe dose. As a probability, this Using the safety index, the probability that method of characteristics (MOC) (12,13).
event is expressed as contaminant concentrations will exceed the This model has been modified to run suc-
safe dose reduces to cessive Monte Carlo simulations to account
Pf =R< C) (1) for subsurface heterogeneity (12,14,15). In
the development of model transport equa-
tions, the units given in the definitions for
S
C.
P = 1-<zD ]r_ variables and parameters in groundwater
flow and contaminant transport equations
Ca (4) have been written in the basic units of
U
length (L), time (t) and mass (M). The
U where D (.) represents the standard normal model is composed of a coupled groundwa-
a0.
S distribution function. ter flow and contaminant transport model.
I.0 The approach presented in Equations 1 The 2-D groundwater flow is given as
through 4 is often called the second moment
formulation (10,11). The name stems from dh
Concentration (micrograms per liter) the fact that the model requires only the dis- S +W(x,y,t)
tribution means and standard deviations that t dt
Figure 1. Graphical representation of the second are functions of the first and second d F dhl dF dhl
moment method. fc(C) represents the distribution moments of the distributions. Equations 3
function of contaminant concentrations in
groundwater from a pumping well located in the and 4 are useful only if both distributions are
dx [x dx dy aY [dy Y]
contaminant plume. FR(C) is the cumulative distri- normally distributed or are assumed to be
bution function for safe doses for this contami- normally distributed. If the data are found to where St is the storage coefficient or storativi-
nant for a diverse human population. The area of be other than normally distributed, an alter- ty of the aquifer (dimensionless), h is
overlap of these two distribution functions is the native solution strategy must be used. One hydraulic head (L), W(x,y,t) is the volume
probability that contaminant concentrations in method is to transform the true distribution flux per unit area of the source term (positive
groundwater exceed safe dose concentrations.
This probability, Pf, is computed by the convolu- into an equivalent normal distribution using for outflow and negative for inflow; L t-l),
tion integral of Equation 2. the Rosenblatt transform (10). Alternatively, and K xx and Kyy are the x:- and yy-diagonal
groundwater flow velocities using Darcy's media is given as where, for equilibrium sorption, the dimen-
law. Darcy's law describes groundwater sionless retardation factor, Rf is defined as
flow through the porous structure of the
d±(C +.P s)= D
d
f
V =K -h;and, V = K -h (9)
x xx dx y- U d'y (6)