ML20206H593
| ML20206H593 | |
| Person / Time | |
|---|---|
| Issue date: | 02/05/1986 |
| From: | Dale Goode NRC |
| To: | |
| Shared Package | |
| ML20206H509 | List: |
| References | |
| NUDOCS 8606260210 | |
| Download: ML20206H593 (18) | |
Text
Selection of soils for wick effect covers Daniel A Goode U.S. Nuclear Regulatory Commission, Washington, DC 8th Annual Symposium on Geotechnical and Geohydrological Aspects of Waste Mgmt.,
2/5-7/86, Fort Collins, Colorado ABSTRACT l
At low capillary pressures (dry conditions), fine-grained soils retain more moisture than coarse-grained soils (the wick effect). Since unsatu-rated hydraulic conductivity is controlled by moisture content, this con-trast can be utilized to reduce infiltration through waste disposal unit covers by installing a coarse-grained soil beneath a fine-grained soil.
The ratio of saturated hydraulic conductivities of the coarse and fine-grained soils has been proposed as a criterion for selecting soils for a wick effect cover. This criterion does not necessarily result in selec-tion of optimum, or even functional soils. This is demonstrated by numerical simulation of two different wick covers using two coarse-grained soils having similar saturated hydraulic conductivities but dif-ferent moisture retention characteristics. The selection of the optimum soil depends on the hydrologic boundary conditions and the unsaturated hydraulic conductivity of the coarse-grained soil at the interface.
1 INTRODUCTION The U.S. Nuclear Regulatory Commission regulates disposal of commercial low-level radioactive waste to protect the public health and safety and the environment. One requirement for disposal unit design is that infil-tration be minimized (10 CFR Part 61). Recently, the technical community has discussed the possibility of utilizing layered soil covers for near-surface disposal units. Layered soil covers and liners are also applica-ble to storage and disposal units for uranium mill tailings and hazardous waste.
It has been observed for some time that a coarse layer in a soil pro-file will cause an overlying fined-grained soil to retain more moisture than it the coarse layer is not present (e.g., Miller and Gardner 1962).
This phenomenon is important for agriculture as a technique to increase irrigation efficiency. Several authors have demonstrated that the so-called wick effect (increased moisture retention in the overlying fine-grained layer) or capillary barrier can also be used to limit infiltra-tion, for example as a cover for radioactive waste disposal units (Corey and Horton 1969; Frind et al.1976; Rancon 1980; Johnson et al.1983).
860626021o 860619 PDR MISC B60626o147 PDR
1 Several authors have assumed that flow will not occur in the coarse-grained layer unless the overlying fine-grained soil becomes saturated at the interface. Others consider that the pressure head at the inter-face must be non-negative before flow will occur in the coarse layer.
These assumptions may be appropriate for gravels but they do not neces-sarily apply to sands and other media (Johnson, et al.1983). Geotech-nical construction problems associated with gravel, including prevention of migration of the fine-grained soil down into the gravel, and avail-ability of suitable material indicate that sands may often be used as the coarse material in layered covers.
The purpose of this paper is to model one-dimensional water movement dynamics in layered non-gravel soil covers and to demonstrate that: (1) the ratio of saturated hydraulic conductivities is not an appropriate soil selection criterion for the wick effect; (2) a layered cover can perform less well than a homogeneous cover even if the coarse layer is not saturated; (3) the extent of infiltration reduction for layered soil covers is dependent on the unsaturated properties of the soils, particu-larly the cross-over. in hydraulic conductivity curves, and the moisture conditions in the overlying materials and the waste disposal unit; and (4) infiltration reduction, evaluated below the coarse-grained layer, ap-pears to be an appropriate criterion for selecting soils for wick effect Covers.
2 CONCEPTUAL MODEL AND S0IL PROPERTIES The soil cover is placed on top of waste and backfill, or other engineer-ing materials such as concrete, which have been emplaced in a shallow trench (Figure 1) or mounded bunker, etc. The cover is installed in lifts with or without geotextiles between separate soil layers. The top fine-grained layer (Layer 1) is overlain by topsoil of sufficient thick-ness to support appropriate vegetation and resist erosion.
Infiltration through the disposal unit cover is conceptualized as a one-dimensional process. Precipitation falling on the topsoil percolates down or is removed by runof f or evapotranspiration. These processes de-termine the boundary condition for the soil cover. This boundary condi-tion can be in the form of a flux, the amount of water percolating down through the topsoil to the top of the soil cover. For this paper, the boundary condition is represented by a specified pressure head which is assumed to be controlled by processes in the surface layer. Likewise, the pressure head at the bottom of the soil cover (Layer 3) is assumed to be controlled by the moisture conditions in the waste and ancillary materials.
Moisture movement through the soil cover is governed by the one-dimen-sfonal form of the Richards equation written in terms of pressure head (Hillel 1980). This form of the unsaturated flow equation allows simu-lation of saturated and unsaturated conditions, and implicitly satisfies the pressure continuity requirement at the interface between different soils. Hysteresis is not considered and the specified soil properties are considered to represent the field scale behavior of each soil. The latter assumption implies that the soils' properties will be determined by in situ field testing. Each layer is homogeneous and the only flow dynamics considered are those represented by the Richards equation with non-hysteretic soil properties.
TOPSOIL r
, J4*'4I!<<6i yg@ . -
yJk
., : '
- r h4r 3 gf % ~" - rNd f L1 s4
- ggg ,__
cff) ABQ@( Soll i LNL ARGEMENT *tyV _ fy _
K g hYNPj COVER jg.- ;; '% 3;g%, @#j:ggf_ __g maw;g w-3 m
" susw a _ r- - -
+
p 5 M-IN-PLACE
.h- .;.-
pU -F
/
MATERIAL N((/' k Fg --g ghN mWh [T hgy ASTE
.4:5.sF. mx+ - '-
A N SF" J~ % Kh
'// $L aa . $ .
Y/44%4, hh A%\N i TO PS OIL 4. . . . . . . . . . . . . .
'( * :::
s
.- $ss?4* WPn 2 y g cm LAYER 1 g s4 e FIN E G R AIN E D
.' Th'2;p F1. 3Espgt.;;;;g . 55;4... .; .
l'^ @' _. .,
^3!$?7 i:::;2:n:H
!.$!:Ei.f'
^
LAYER 2
":I"*" SOIL
- i:i:i !$9
! C m,p ij COARSE.GR AINEDj;
) COVER StMULATED v
- i.i;T a..- s.
- 1. a:: g.g.g g.a.g g g..g..g sammx - -- xmam -
.pf2 '%gQy w # ~ ~ '?M:"_anf f* .? W 7i LAYER 3 $[E3; FINE GRAINED q if"w@T{;if;P YF g-. . > 9~ s(cmPb y,m.m.;e at:M :
ag; g+ey:2 s,
)
mm 9-M?WWDW&
nw1.-rw r%s,y-%=. ;f-
~<-Q WASTE M ATERI ALS;MJ
'M,;cygpx;LL AND BACKFI c-.g ? ,-
mvs*NE-w:w:
l l
Figure 1. Schematic of a soil cover for a near-surface disposal unit.
l
The FEMWATER computer code (Yeh 1982) is used to solve steady state and transient unsaturated flow in two dimensions using linear finite ele-ments. Only the vertical dimension, z, is relevant for this paper. The computational grid is a single row of 36 quadrilateral elements, each 5 cm thick yielding a total thickness of 180 cm. For Covers 2 and 3, the middle 60 cm of the grid represents the coarse-grained material. Unsatu-rated hydraulic conductivity, moisture content, and specific capacity are linearly interpolated from input tables. This interpolation scheme is responsible for many of the first order discontinuities in plots (below) on a logarithmic scale.
The FEMWATER code uses a special procedure to calculate darcy veloci-ties (fluxes) which provides for continuity in the flux field. When ap-plied to a soil system with sharply varying conductivities, this method may result in an unrealistic flux field. This calculation does not af-fect the calculation of pressure head. For this study, vertical darcy -
velocities were calculated explicitly for each element using the FEMWATER pressure heads and the geometric mean hydraulic conductivity. This mean has been recommended for unsaturated flow computations by Haverkamp et al. (1977) and Schnabel and Richie (1984). The pressure head gradient is assumed to be linear within each element for this calculation.
The fine-grained material is the Peoria Loess studied by Johnson et al.
(1983). This material is described as a clayey silt by Foster et al.
(1984). Properties for the medium sand are also derived from Johnson et al. (1983). The second coarse-grained material is a crushed tuff being used for large scale unsaturated transport experiments at Los Alamos Na-tional Laboratory for NRC (Polzer 1985).
The porosity values for the loess, sand, and crushed tuff are 0.38, 0.3, and 0.34, respectively. The moisture characteristic curves illus-trate that the sand moisture content decreases more than the other two soils as pressure is reduced (Figure 2a).
C ,,
x '9 ;
x x n
\ 33~
N
\ \
N 5
-e - N N Si N N t ' y
\
a c- \ [S 3
i N
\ Te; \
\ \
=
u 1
W
\
\
+
Loess j 6
f 2 @m m t .\
yhr- g2;
'I x -
~ " " '- "5 ' 2. o 2.s 32 Soistud#contest hs Figure 2. Unsaturated soil properties for loess, medium sand, and crushed tuff: (a) moisture characteristic curves; (b) unsaturated hydraulic con-ductivity (PF = log (-h), h = pressure head in cm; data from Johnson et al. 1983 and Polzer 1985).
l
The saturated hydraulic conductivities are: loess - 2.0 E-6 cm/s; sand
- 1.0 E-4 cm/s; crushed tuff - 5.8 E-4 cm/s. At saturation, the crushed tuff is the most permeable medium. If a high ratio of saturated conduc-tivities is to be used as the only criterion for selecting a coarse-grained soil for a wick effect cover, the crushed tuff would be the pre-ferred material. However, the unsaturated hydraulic conductivity of the crushed tuff does not fall below that of the loess as pressure decreases (Figure 2b). This is in contrast to the sand which exhibits a lower un-saturated conductivity than the loess at pressure heads less than about <
-50 cm. This cross over of hydraulic conductivity curves is a critical I factor in assessing the performance of wick effect covers. The presence of this cross over is illustrated with a plot of hydraulic conductivity versus pressure head, and not versus moisture content, because the mois-ture contents in the two soils at the interface are not necessarily equal. -
Three prototype soil covers are simulated below under two different climatic conditions. Cover 1, the control, is a single layer cover of loess. Cover 2 utilizes the medium sand as the coarse-grained layer in between two ?ayers of loess. Cover 3 is also a wick effect cover, with the crushed tutt as the coarse-grained middle layer.
3 RESULTS FOR DRY AND WET CONDITIONS Each of the three covers are simulated under relatively dry conditions.
Initially, the pressure head is -500 cm at all depths. At time zero, the pressure head at the top of the soil cover is increased to -100 cm. This pressure is assumed to represent average moisture conditions in the over-lying topsoil in response to precipitation, evaporation, and transpira-tion. This system is not in hydraulic equilibriu:s and moisture moves into the soil cover from the topsoil at a decreasing rate until a steady state condition is established.
3:
E COVER 1
,1 -]Q$ --
o , _ . - .. - - --
TE g r ....
%_ { . . -'
M / ,/
5 : /
s-Y?1 f
~/ ,.!
>_ E- .
!$ : /
2g
...,,, ' " - ' Figure 3. Vertical darcy E
2 - flux downward under dry con-
'o
~
ditions just below the o 560 taho 1s'00 20'00 coarse-grained layer (z=60 TIME (HR1 cm).
The vertical darcy velocity downward at the top of the bottom lay:r of each cover (z=60 cm from the bottom) is compared in Figure 3. Initially, the darcy velocity for each cover is 1.0 E-9 cm/s. Prior to about 30 days (=720. hours) the flux in Cover 2 is less than the control (Cover 1).
This occurs because the sand in Cover 2 has very little water in storage initially and that stored water will not support the initial flux rate.
However, once the front reaches the top of the sand layer, it quickly moves to the bottom of that layer, faster than the corresponding movement in the single layer cover (Figure 4). The specific capacity of the sand is much less than that of the loess at low pressures and equilibrium in the sand is achieved with smaller increases in moisture content. From about 30 to 65 days the flux rate in Cover 2 is higher than that in Cover
- 1. However, after 65 days, as steady state is approached, the wick cover using a sand layer is more effective at reducing infiltration. This is because the front in the single layer cover reaches z=60 cm after about -
60 days. The pressure head at this elevation is less than -200 cm and as shown above (see Figure 2b) the sand's hydraulic conductivity is less than that of the loess. For this simulation, the sand layer does improve the performance of the cover in limiting infiltration. The steady state flux for Cover 2 is about 55 percent of that for Cover 1.
PRESSURE HERO (Cm
-600 -500 -4,00 -3,00 -2,00 -100 0 0 j COVER 1
/. COVER 2 b 10 i 30EE E -
1 r
/ /[:./ .
['
l
(. \
g- 9 ,\. - f*
90/
g_ f / Figure 4. Pressure head pro- !
/ files under dry conditions
/ at time 0, 10.45, 30.45, and ,
o 90.45 days. '
Cover 3, utilizing the crushed tuff as the coarse-grained layer, does not reduce infiltration compared to Cover 1 (Figure 3). The flux rate for Cover 3 is about 55 percent greater than that for Cover 1. The front also moves more rapidly to the bottom of the soil cover (Figure 4). At the pressures simulated in the tuff layer, the tuff is much more conduc-tive to water than the loess. The hydraulic conductivity curves for the tuff and loess do not cross despite the fact that the saturated hydraulic conductivity of the tuff is about 6 times as high as that of the sand, which does cross over. It is clear that the entire unsaturated hydraulic l conductivity curve, and not just the saturated value, must be considered i
when designing an effective wick effect cover.
l
l l
. l l
The oscillations apparent in Figure 3 are due to several reasons.
Darcy velocity is a function of hydraulic conductivity, values of which are obtained from an input table by linear interpolation. When plotted on a logarithmic scale, this interpolation results in a discontinuous slope. Secondly, fluxes will decrease and increase in response to move-ment of the front as discussed above for Cover 2. Finally, the darcy ve-locity, as a derivative of the pressure head solution, is more sensitive to numerical errors. It is not known to wnat extent each of these ef-fects is illustrated in Figure 3.
Covers 1 and 2 are also simulated under relatively extreme wet condi- !
tions. The initial pressure head is -100 cm. The boundary condition on the top of the soil cover is increased to 0 cm. Physically, this corre-sponds to a film of saturation which is in equilibrium with the atmo-sphere. This condition is unlikely in actual cover systems, but could occur during periods of above average precipitation in a humid climate.
Pressures greater than 0 cm could occur at the top of a layered liner system, for example beneath a liquid storage impoundment. The boundary condition at the bottom of the cover is maintained at -100 cm.
Cover 1 is essentially in equilibrium at 90 days with a monotonic de-crease in pressure head with depth. The steady state darcy velocity is 2.3 E-6 cm/s downward. Note that this value is greater than the saturat-ed hydraulic conductivity of the loess because of the high hydraulic gra-dient imposed.
S.
0 - ,....... . . ......... .. ............ ..... .
7 ,
? - :
C
[S: '
l d:
- l 0 -
E -
COVER 1
'EC.0XLR.2..
Figure 5. Vertical darcy ve-
, locity downward under wet
'o , conditions just below the
~o 500 tcho .io; zico coarse-grainedlayer(z=60 TIME !"Pi cm).
Cover 2 with the sand as the coarse-grained layer also reaches equilib-rium by 90 days with a resulting steady state darcy velocity of 2.9 E-6 cm/s downward, this flux is about 25 percent higher than the flux for l the single layer cover (Figure 5). As shown in the hydraulic conductivi- I ty curves (see Figure 2b), the sand is more conductive than the loess i above about -50 cm pressure. As the front moves through the cover under ;
these wet conditions, pressures greater than -50 cm occur in the sand re- l l
l l
l
sulting in increased hydraulic conductivity (Figure 6) and flux. The conductivity at the top of the sand is greater than that of the loess (at the interface) after a few days. The conductivity of the sand at the bottom of the layer is higher than the value for the underlying loess by over 1.5 orders of magnitude after about 20 days. The pressure head dis-tribution also corresponds to the moisture distribution; the sand layer is wettest at the bottom. The moisture content at the bottom of Layer 1 in Cover 2 (0.355) is less than the corresponding value for Cover 1 (0.373). The wick effect is not operative under these cor.ditions. For situations where these wet conditions exist, introduction of the sand layer into the soil cover decreases its effectiveness at reducing infiltration.
'o E
s '
5e ~ [
- Er /
it : /
g ........(..........................................
m , o _- - - --
S S- A,A 9 /
LAYER 1 B0T
/ LAYER 2 TOP h3N ./
gi M - --
F Figure 6. Unsaturated hy-draulic conductivity under
- - wet conditions in Cover 2 at the bottom of the top loess
'g , , , , layer, the top and botom of a 500 1000 1500 2000 the sand, and the top of the TIME IHRi bottom loess layer.
4
SUMMARY
These relatively simple one-dimensional simulations indicate some of the complex processes which must be addressed in selecting soils for layered cover systems. Selecting the coarse-grained soil with the highest satu-rated hydraulic conductivity will not necessarily result in reduction of infiltration. The unsaturated hydraulic conductivity curve for each soil must be determined and incorporated into appropriate simulations, repre-sentative of expected field moisture conditions. Of particular impor-tance is determination of the pressure range for which the coarse-grained soil is less conductive than the fine-grained soil. Performance of the cover depends on whether pressures experienced in the field fall within that pressure range. The value of long term darcy velocity below the coarse-grained layer appears appropriate as a criterion for comparison and ranking of alternative cover designs. These conclusions are made within the limitations of the conceptual model which may not be appropri-ate for all actual soil cover systems.
l
.- r.
REFERENCES Corey, J.C. and J.H. Horton 1969. Influence of gravel layers on soil moisture content and flow. Savannah River Laboratory DP-1160.
Foster, J.B., J.R. Erickson, and R.W. Healy 1984. Hydrogeology of a low-level radioactive-waste disposal site near Sheffield, Illinois.
U.S. Geological Survey. Water-Resources Investigations Rep. 83-4125.
Frind, E.0., R.W. Gillham, and J.F. Pickens 1976. Application of unsaturated flow properties in the design of geologic environments for radioactive waste storage facilities. In W.G. Gray, G.F. Pinder, and C.A. Brebbia (eds.), Finite Elements in Water Resources, p.3.133-3.182. London: Pentech Press.
Haverkamp, R., M. Vauclin, J. Touma, P.J. Wierenga, and G. Vachaud 1977.
A comparison of numerical simulation models for one-dimensional infiltration. Soil Sci. Soc. Amer. J. 41:285-294. -
Hillel, D.1980. Introduction to soil physics. Orlando, Florida:
Academic Press.
Johnson, T.M., T.H. Larson, B.L. Herzog, K. Cartwright, C.J. Stohr, and S.J. Klein 1983. A study of trench covers to minimize infiltration at waste disposal sites. U.S. Nuclear Regulatory Connission. NUREG/CR-2478 Vol. 2.
Miller, D.E. and W.H. Gardner 1962. Water infiltration into stratified soil. Soil Sci. Soc. Amer. Proc. 26:115-118.
Polzer, W.L. 1985. Field studies and modeling chemical processes in the unsaturated zone FY 1984 annual report - DRAFT. Los Alamos National Laboratory.
Rancon, D.1980. Application de la technique des barrieres capillaires aux stockages en tranchees. In Underground Disposal of Radioactive Wastes, Vol. I, p.241-265. Vienna: International Atomic Energy Agency.
Schnabel, R.R. and E.B. Richie 1984. Calculation of internodal conductances for unsaturated flow simulations: A comparison. Soil Sci.
Soc. Amer. J. 48:1006-1010.
Yeh, G.T. 1982. The implementation of FEMWATER (0RNL-5567) computer program. U.S. Nuclear Regulatory Commission. NUREG/CR-2705.
l l
l
1 Selection of soils for wick effect covers i
Daniel J..Goode '
U.S. Nuclear Regulatory Comission, Washington, DC 8th Annual Symposium on Geotechnical and Geohydrological Aspects of Waste Mgmt.,
2/5-7/86, Fort Collins, Colorado ABSTRACT At low capillary pressures (dry conditions), fine-grained soils retain more moisture than coarse-grained soils (the wick effect). Since unsatu-
- rated hydraulic conductivity is controlled by moisture content, this con-trast can be utilized to reduce infiltration through waste disposal unit covers by installing a coarse-grained soil beneath a fine-grained soil.
The ratio of saturated hydraulic conductivities of the coarse and fine-grained soils has been proposed as a criterion for selecting soils for a wick effect cover. This criterion does not necessarily result in selec-tion of optimum, or even functional soils. This is demonstrated by numerical simulation of two different wick covers using two coarse-grained soils having similar saturated hydraulic conductivities but dif-ferent moisture retention characteristics. The selection of the optimum soil depends on the hydrologic boundary conditions and the unsaturated hydraulic conductivity of the coarse-grained soil at the interface.
1 INTRODUCTION The U.S. Nuclear Regulatory Comission regulates disposal of comercial low-level radioactive waste to protect the public health and safety and the environment. One requirement for disposal unit design is that infil-tration be minimized (10 CFR Part 61). Recently, the technical community has discussed the possibility of utilizing layered soil covers for near-surface disposal units. Layered soil covers and liners are also applica-ble to storage and disposal units for uranium mill tailings and hazardous waste.
It has been observed for some time that a coarse layer in a soil pro-file will cause an overlying fined-grained soil to retain more moisture than it the coarse layer is not present (e.g., Miller and Gardner 1962).
This phenomenon is important for agriculture as a technique to increase I irrigation efficiency. Several authors have demonstrated that the so- 1 called wick effect (increased moisture retention in the overlying fine-grained layer) or capillary barrier can also be used to limit infiltra-tion, for example as a cover for radioactive waste disposal units (Corey and Horton 1969; Frind et al.1976; Rancon 1980; Johnson et al.1983). i l
l l
l
i .
. Several authors have assumed that flow will not occur in the coarse-
- grained layer unless the overlying fine-grained soil becomes saturated i at the interface. Others consider that the pressure head at the inter-face must be non-negative before flow will occur in the coarse layer.
l_ These assumptions may be appropriate for gravels but they do not neces-
, sarily apply to sands and other media (Johnson, et al.1983). Geotech-i nical construction problems associated with gravel, including prevention i of migration of the fine-grained soil down into the gravel, and avail-l ability of suitable material indicate that sands may often be used as
- the coarse material in layered covers.
The purpose of this paper is to model one-dimensional water movement dynamics in layered non-gravel soil covers and to demonstrate that: (1) the ratio of saturated hydraulic conductivities is not an appropriate soil . selection criterion for the wick effect; (2) a layered cover can perfom less well than a homogeneous cover even if the coarse layer is -
not saturated; (3) the extent of infiltration reduction for layered soil covers is dependent on the unsaturated properties of the soils, particu-i larly the cross-over. in hydraulic conductivity curves, and the moisture conditions in the overlying materials and the waste disposal unit; and (4) infiltration reduction, evaluated below the coarse-grained layer, ap-pears to be an appropriate criterion for selecting soils for wick effect covers.
! 2 CONCEPTUAL MODEL AND S0IL PROPERTIES The soil cover is placed on top of waste and backfill, or other engineer-j ing materials such as concrete, which have been emplaced in a shallow trench (Figure 1) or mounded bunker, etc. The cover is installed in i lifts with or without geotextiles between separate soil layers. The top
- fine-grained layer (Layer 1) is overlain by topsoil of sufficient thick-ness to support appropriate vegetation and resist erosion.
Infiltration through the disposal unit cover is conceptualized as a one-dimensional process. Precipitation falling on the topsoil percolates down or is removed by runotf or evapotranspiration. These processes de-
~ temine the boundary condition for the soil cover. This boundary condi-tion can be in the form of a flux, the amount of water percolating down through the topsoil to the top of the soil cover. For this paper, the boundary condition is represented by a specified pressure head which is i
assumed to be controlled by processes in the surface layer. Likewise, the pressure head at the bottom of the soil cover (Layer 3).is assumed to be controlled by the moisture conditions in the waste and ancillary
- i materials.
l Moisture movement through the soil cover is governed by the one-dimen-sional fom of the Richards equation written in terms of pressure head
! (Hillel 1980). This form of the unsaturated flow equation allows simu-lation of saturated and unsaturated conditions, and implicitly satisfies
- the pressure continuity requirement at the interface between different
! soils. . Hysteresis is not considered and the specified soil properties i are considered to represent the field scale behavior of each soil. The l latter assumption implies that the soils' properties will be detemined 1
by in situ field testing. Each layer is homogeneous and the only flow dynamics considered are those represented by the Richards equation with non-hysteretic soil properties.
I e
c ywe-. - - > - - , , , w -
w- -- y ,,---._-,.,--.,,,-,--._ev., s.- .- -
.m.u - - - . . - - , . ,,,+,,-,,--,,a., ,.~,e w, ,
TOPSOIL ' ..v1 dif#Vev
,e ::. 9..
y J>
-p bi-: f 1 i ~ W4 4 r- s$f5 !N'5{' Sgg ? .
O h / $ L. .L.
SolL if ENLARGEMENT N , QQ _ ,
'~
- h,Ag$ COVER ' " -
jf M M gir ==
ifM[ . ..
g$d52h* -
- "^# V EN M $i M " ~ --
w ._# #_y _ . ;- c ._ s,-
A -t M!
IN. PLACE M*OMbv^* ET%
!///$ EQSgMSWASTE -N /
MATERIAL
%M?.df,.C_'illM.T .
w94-.
V _ n.n. ;- n-e .x+S*
b
.=
M NE
=;J n A._ .!,y
_ . ~~
~/,/ .$.L5.%^ -- c ..n ...~
_~ . _ .5$w4
\// fY Y\ N
% A AI A%
dTOPSOILF
- s
/ fM.:y:qyF:5 Ef2: T
.a u can .jS
'F 7 LAYER 1 teh de.~ eu i n,m i _ . FINE GR AINEDE ~
aw:va ?
W.:::. ::. .M.M.M.W.W.M... :.:.:.;.x. .;.;.;.3.... .. M. ....M....
..M.M. .
i:i:s si::
^ S Oll.
LAYER 2 I!h.g c,gjp; COARSE GR AINEDji > COVER
- T::,:::
.. Z. ..g
. ..g j . . . . . . . . . . . . . . . . . . . . . . . . .
g g.g g.:.g.:.g. g. g .g SIMULATED
=jg;GlAQ:;*?f% i*- W*^ x 'i
+
qshl LAYER 3 D.ad$
qM c% gS - N[g/ FINE GRAINED I# ?N
' " llih7 f-f44 ye .wsg, ;yur5C f WJ.;,c'J 9; .s. 4-A29..19hFA-@y.%.NI?
a rw. a x-a Au,
' BJWASTE MATERIALS? wit
-22T AND BACKFILL ;f '#='
A"-i!GQ=}W.i.:~C-?
~:a, s m w r app-Figure 1. Schematic of a soil cover for a near-surface disposal unit.
l
The FEMWATER computer code (Yeh 1982) is used to solve steady state and transient unsaturated flow in two dimensions using linear finite ele- l ments. Only the vertical dimension, z, is relevant for this paper. The '
computational grid is a single row of 36 quadrilateral elements, each 5 )
cm thick yielding a total thickness of 180 cm. For Covers 2 and 3, the '
middle 60 cm of the grid represents the coarse-grained material. Unsatu-rated hydraulic conductivity, moisture content, and specific capacity are linearly interpolated from input tables. This interpolation scheme is responsible for many of the first order discontinuities in plots (below) on a logarithmic scale.
The FEMWATER code uses a special procedure to calculate darcy veloci-ties (fluxes) which provides for continuity in the flux field. When ap-plied to a soil system with sharply varying conductivities, this method may result in an unrealistic flux field. This calculation does not af-fect the calculation of pressure head. For this study, vertical darcy -
velocities were calculated explicitly for each element using the FEMWATER pressure heads and the geometric mean hydraulic conductivity. This mean has been recommended for unsaturated flow computations by Haverkamp et al. (1977) and Schnabel and Richie (1984). The pressure head gradient is assumed to be linear within each element for this calculation.
The fine-grained material is the Peoria Loess studied by Johnson et al.
(1983). This material is described as a clayey silt by Foster et al.
(1984). Properties for the medium sand are also derived from Johnson et al. (1983). The second coarse-grained material is a crushed tuff being used for large scale unsaturcted transport experiments at Los Alamos Na-tional Laboratory for NRC (Polzer 1985).
The porosity values for the loess, sand, and crushed tuff are 0.38, 0.3, and 0.34, respectively. The moisture characteristic curves illus-trate that the sand moisture content decreases more than the other two soils as pressure is reduced (Figure 2a).
3 'o
.N q
[. \\N [o - . . . h %
N x 5, - N
=
- N > .Si '
N
\ 5, ; \
\ -o N
@ \ $ \
\ %' \
=
. \ 8~ \
o,
~2 x k
ci -
LOCSS 58ND f1 E'o -
L0ESS 58ND 3R_05ffD3E.". .\
. .C86fD_1E.._
l y -1 2 , .
! 'o '
O.0 0.1 0.2 0. 3 0.4 0.0 0.5 1.0 1.5 2.0 2. 5 3.J MOISTURE CONTCNI Pr Figure 2. Unsaturated soil properties for loess, medium sand, anc' crushed tuff: (a) moisture characteristic curves; (b) unsaturated hydraulic con- ;
ductivity (PF = log (-h), h = pressure head in cm; data from Johnson et l al. 1983 and Polzer 1985). !
l
The saturated hydraulic conductivities are: loess - 2.0 E-6 ci../s; sand ,
- 1.0 E-4 cm/s; crushed tuff - 5.8 E-4 cm/s. At saturation, the crushed !
tuff is the most penneable medium. If a high ratio of saturated conduc- 1 tivities is to be used as the only criterion for selecting a coarse- !
grained soil for a wick effect cover, the crushed tuff would be the pre-ferred material. However, the unsaturated hydraulic conductivity of the crushed tuff does not fall below that of the loess as pressure decreases (Figure 2b). This is in contrast to the sand which exhibits a lower un-saturated conductivity than the loess at pressure heads less than about
-50 cm. This cross over of hydraulic conductivity curves is a critical factor in assessing the performance of wick effect covers. The presence of this cross over is illustrated with a plot of hydraulic conductivity versus pressure head, and not versus moisture content, because the mois-ture contents in the two soils at the interface are not necessarily equal.
Three prototype soil covers are simulated below under two different climatic conditions. Cover 1, the control, is a single layer cover of loess. Cover 2 utilizes the medium sand as the coarse-grained layer in between two layers of loess. Cover 3 is also a wick effect cover, with the crushed tutt as the coarse-grained middle layer.
3 RESULTS FOR DRY AND WET CONDITIONS 1
Each of the three covers are simulated under relatively dry conditions.
Initially, the pressure head is -500 cm at all depths. At time zero, the pressure head at the top of the soil cover is increased to -100 cm. This pressure is assumed to represent average moisture conditions in the over-lying topsoil in response to precipitation, evaporation, and transpira-tion. This system is not in hydraulic equilibrium and moisture moves into the soil cover from the topsoil at a decreasing rate until a steady state condition is established.
vg 1: COVER 1
,- ] COVER 2QtCR.C C
_ g. ,_
Q? a .....-
g r ,...
"~
x l ..-
El / ./
8: / .'
[2](N ,
,/
t; ? .'
9 :
- .. .....' " "- '~ ~ Figure 3. Vertical darcy
. : flux downward under dry con-b ~
ditions just below the o saa 1000 15'00 aico coarse-grained layer (z=60 TIME (HR) cm).
The vertical darcy velocity downward at the top of the bottom layer of each cover (z=60 cm from the bottom) is compared in Figure 3. Initially, the darcy velocity for each cover is 1.0 E-9 cm/s. Prior to about 30 days (=720. hours) the flux in Cover 2 is less than the control (Cover 1).
This occurs because the sand in Cover 2 has very little water in storage initially and that stored water will not support the initial flux rate.
However, once the front reaches the top of the sand layer, it quickly moves to the bottom of that layer, faster than the corresponding movement in the single layer cover (Figure 4). The specific capacity of the sand is much less than that of the loess at low pressures and equilibrium in the sand is achieved with smaller increases in moisture content. From about 30 to 65 days the flux rate in Cover 2 is higher than that in Cover
- 1. However, after 65 days, as steady state is approached, the wick cover using a sand layer is more effective at reducing infiltration. This is because the front in the single layer cover reaches z=60 cm after about -
60 days. The pressure head at this elevation is less than -200 cm and as shown above (see Figure 2b) the sand's hydraulic conductivity is less than that of the loess. For this simulation, the sand layer does improve the performance of the cover in limiting infiltration. The steady state flux for Cover 2 is about 55 percent of that for Cover 1.
PRESSURE HEfl0 (CM)
-600 -500 -4,00 -3,00 -200 -100 0 0 f- COVER 1 g~ /i COVER 10
/ i '
3QEU. 2."
1
/
(
/ [./:
0- # / :
g \
'l f
~ I -
\
8- j. ' ,. }. ->... ,.
90/
g- [ / Figure 4. Pressure head pro-
/ files under dry conditions
~
/ at time 0, 10.45, 30.45, and o 90.45 days.
Cover 3, utilizing the crushed tuff as the coarse-grained layer, does not reduce infiltration compared to Cover 1 (Figure 3). The flux rate for Cover 3 is about 55 percent greater than that for Cover 1. The front also moves more rapidly to the bottcm of the soil cover (Figure 4). At i the pressures simulated in the tuff layer, the tuff is much more conduc-tive to water than the loess. The hydraulic conductivity curves for the tuff and loess do not cross despite the fact that the saturated hydraulic conductivity of the tuff is about 6 times as high as that of the sand, which does cross over. It is clear that the entire unsaturated hydraulic conductivity curve, and not just the saturated value, must be considered when designing an effective wick effect cover. ,
~
The oscillations apparent in Figure 3 are due to several reasons.
Darcy velocity is a function of hydraulic conductivity, values of which are obtained from an input table by linear interpolation. When plotted on a logarithmic scale, this interpolation results in a discontinuous slope. Secondly, fluxes will decrease and increase in response to move-ment of the front as discussed above for Cover 2. Finally, the darcy ve-locity. as a derivative of the pressure head solution, is more sensitive !
to numerical errors. It is not known to wnat extent each of these ef-fects is illustrated in Figure 3.
Covers 1 and 2 are also simulated under relatively extreme wet condi-tions. The initial pressure head is -100 cm. The boundary condition on the top of the soil cover is increased to 0 cm. Physically, this corre-sponds to a film of saturation which is in equilibrium with the atmo-sphere. This condition is unlikely in actual cover systems, but could occur during periods of above average precipitation in a humid climate.
Pressures greater than 0 cm could occur at the top of a layered liner system, for example beneath a liquid storage impoundment. The boundary condition at the bottom of the cover is maintained at -100 cm.
Cover 1 is essentially in equilibrium at 90 days with a monotonic de-crease in pressure head with depth. The steady state darcy velocity is 2.3 E-6 cm/s downward. Note that this value is greater than the saturat-ed hydraulic conductivity of the loess because of the high hydraulic gra-dient imposed.
N g
9 - '
/
$o.
o- :
g
' ~
U-E - ! COVER 1 T C.V M .d. .
Figure 5. Vertical darcy ve-
. locity downward under wet Mo , , , ,
conditions just below the 500 loco .50; 2000 coarse-grained layer (z=60 TIME HPi cm).
Cover 2 with the sand as the coarse-grained layer also reaches equilib-rium by 90 days with a resulting ste: Jy state darcy velocity of 2.9 E-6 cm/s downward, this flux is about 2 percent higher than the flux for the single layer cove (Figure 5). As shown in the hydraulic conductivi-ty curves (see Figure L), the sand is more conductive than the loess above about -50 cm pressure. As the front moves through the cover under these wet conditions, pressures greater than -50 cm occur in the sand re-
sulting in increased hydraulic conductivity (Figure 6) and flux. The conductivity at the top of the sand is greater than that of the loess (at the interface) after a few days. The conductivity of the sand at the I bottom of. the layer is higher than the value for the underlying loess by over 1.5 orders of magnitude after about 20 days. The pressure head dis- i tribution also corresponds to the moisture distribution; the sand layer is wettest at the bottom. The moisture content at the bottom of Layer 1 in Cover 2 (0.355) is less than the corresponding value for Cover 1 (0.373). The wick effect is not operative under these conditions. For situations where these wet conditions exist, introduction of the sand layer into the soil cover decreases its effectiveness at reducing infiltration.
'o E
l s
b? .!
- ch /
=! /
g- ........(...........................................
w, i__- - - --
SOA .n 9~& /
~
! LRYER 1 B0T
/ LAYER 2 TOP
[i g5 s./ R I Figure 6. Unsaturated hy-draulic conductivity under wet conditions in Cover 2 at the bottom of the top loess
'o , , layer, the top and botom of a 500 1000 1500 2000 the sand, and the top of the TIME (HRi bottom loess layer.
4
SUMMARY
These relatively simple one-dimensional simulations indicate some of the complex processes which must be addressed in selecting soils for layered cover systems. Selecting the coarse-grained soil with the highest satu-rated hydraulic conductivity will not necessarily result in reduction of infiltration. The unsaturated hydraulic conductivity curve for each soil must be determined and incorporated into appropriate simulations, repre-sentative of expected field moisture conditions. Of particular impor-tance is determination of the pressure range for which the coarse-grained soil is less conductive than the fine-grained soil. Perfonnance of the cover depends on whether pressures experienced in the field fall within that pressure range. The value of long term darcy velocity below the coarse-grained layer appears appropriate as a criterion for comparison and ranking of alternative cover designs. These conclusions are made within the limitations of the conceptual model which may not be appropri-ate for all actual soil cover systems.
REFERENCES Corey, J.C. and J.H. Horton 1969. Influence of gravel layers on soil moisture content and flow. Savannah River Laboratory DP-1160.
Foster, J.B., J.R. Erickson, and R.W. Healy 1984. Hydrogeology of a low-level radioactive-waste disposal site near Sheffield, Illinois.
U.S. Geological Survey. Water-Resources Investigations Rep. 83-4125.
Frind, E.O., R.W. Gillham, and J.F. Pickens 1976. Application of unsaturated flow properties in the design of geologic environments for radioactive waste storage facilities. In W.G. Gray, G.F. Pinder, and C.A. Brebbia (eds.), Finite Elements in Water Resources, p.3.133-3.182. London: Pentech Press.
Haverkamp, R., M. Vauclin, J. Touma, P.J. Wierenga, and G. Vachaud 1977.
A comparison of numerical simulation models for one-dimensional '
infiltration. Soil Sci. Soc. Amer. J. 41:285-294.
Hillel, D.1980. Introduction to soil physics. Orlando, Florida:
Academic Press.
Johnson, T.M., T.H. Larson, 8.L. Herzog, K. Cartwright, C.J. Stohr, and S.J. Klein 1983. A study of trench covers to minimize infiltration at waste disposal sites. U.S. Nuclear Regulatory Commission. NUREG/CR-2478 Vol. 2.
Miller, D.E. and W.H. Gardner 1962. Water infiltration into stratified soil. Soil Sci. Soc. Amer. Proc. 26:115-118.
Polzer, W.L.1985. Field studies and modeling chemical processes in the unsaturated zone FY 1984 annual report - DRAFT. Los Alamos National Laboratory.
Rancon, D. 1980. Application de la technique des barrieres capillaires aux stockages en tranchees. In Underground Disposal of Radioactive Wastes, Vol. I, p.241-265. Vienna: International Atomic Energy Agency.
Schnabel, R.R. and E.B. Richie 1984. Calculation of internodal conductances for unsaturated flow simulations: A comparison. Soil Sci.
Soc. Amer J. 48:1006-1010.
Yeh, G.T. 1982. The implementation of FEMWATER (ORNL-5567) computer program. U.S. Nuclear Regulatory Commission. NUREG/CR-2705.
{
,