l"nsaturated Performance Comparison of Compacted Clay Landfill Liners

Carol J. \liller 1• Saad \1erayyan ,

and \lazli Yesillec •

Ahsr,.uC!

rests were conducted to determine the variation in volumetric ,vater content and pore \\ater suction for a variety of compacted clay soils used in the construction of landfill liners. The fit of the experimental data to an existing parametric model was imestigated for two different fitting techniques. The first technique involvcs the use of the rl.?tention curve computer program (RETC) developed for the U.S. EPA. The second technique employs the Solver subroutine included in Microsoft Excel. The parametric models resulting from either technique correlated well to the experimental data, HO\\evcr. the individual curve fit parameters varied significantly. The effect of these variations on the unsaturated behavior of compacted clay liners \\as e\aluatl.?d using the Hydrologic Evaluation of Landfill Performance (HELP) model. The curve fit parameters resulting from both the RETC and the Solver techniques were used as input to the HELP routine for simulation of variably saturated now through a covcr liner. There were no significant differences in the volume of leakage or rate of leakage predicted using the input from the t\,;o curve-fitting techniques. However. there was significant soil-dependent variation in the HELP output. Examination of the HELP output provides information regarding variation in the volumetric water content of the cover liner soil. This information can be used to predict pore water suction variations and susceptibility to desiccation cracking.

Illlrtldllc

lil117

Ihe primal')

runction of a cover liner at a landlill is to

limit the m

igration of m

oisture to the

underlying w

aste. thereby

reducing the volum

e and rate o

f leachate production.

C'm

er liners arc

constructed at

volumetric

water contents less than the

saturated value, and during the liiC o

f the containment system

moisture is added to and

removed from

the cover liner. T

herefore, it is expected that for at least a portion o

f it's liiC

. the cover liner will be operating in an unsaturated m

ode. T

he governing equation for Ilovv through a variably saturated liner, assum

ing one-dimensional. vertical now

is:

(I)

vvhere t is the time (sec): z is the elevation assum

ed positive upward:

Iji is pore water

suction head (m):

I\. t;J

is unsaturated hydraulic conductivity (m sec· I ): and c(1JI)

is

specific V

\(lter capacity (m

· I ). ('(w

)relates pore

water suction

to volum

etric water

content and is equivalent to the slope of the soil-w

ater characteristic curve (SW

CC

). T

he data necessary to construct a sw

ee

is typically determined using pressure plate

apparatus (AS

TM

, ](94

). T

he resulting data can be fit to existing parametric m

odels o

f the S

\Ve

C (B

rooks and C

orey, 1

9M

: Fredlund

and X

ing, 1

9(4

). O

ne comm

on m

odel is the van Crenuchten (van C

renuchten. 19XO

) equation:

(2)

\\ here thl' optimized paran1l'ters arc 0,. 0,. a. n. and m

. T

he etrect or the param

eters a, n, and m

on the shape of the S

\Ve

e is provided by I.eong and R

ahardjo (] 9(7

). V

an (Jenuchten (] 9X

O)

suggested a simplirying relation

hetween the

parameters m

and n (i.e

.. J1l

I-lin).

llow

e\er. in the present investigation the param

eters J1l and n are

assumed to be independent for m

ore flexibility in the curve-fitting procedure.

The

Swce

can be

used to

provide insight

to the

potential for

desiccation crackin~ o

f a cover liner soil. Durin~ desiccation, the saturation of a liner is reduced,

L

and the remaining pore w

ater is helll at increasingly large suctions. If the volum

etric w

ater content is kncmn, or has been predicted, the Sw

ee can be used to determine the

pore \\ater suctions corresponding to the know

n volumetric w

ater ('()I1tenl. Previ~us

research (Miller et al.. 199R

) has shown that the onset and resultinl..'. am

ount of crackm

g is C

OlTl'lalL'd to a soil-speci lic critical suction level.

llence the J;aximum

pore water

suction expected in

the con'!' liner can

he com

pared to the critical suction level

to estim

ate suscl'ptibility to desiccation cracking.

rhe Iandlill co\er liner problem

is not the only environmental application t~at

requirl's modeling or unsaturated conditions. A

similar problem

il1\olves surfilce spdl s

that m

igrate throut'h the unsaturated zone towards an unconlined aquifer.

In general,

the solution

or pollutant

migration

through the

unsaturated lo

ne

is a

mueh

more

difficult prohlem than contam

inant transport through the aquifer. In the aquircr, the

pore water prcssure is linearly related to depth helow

the fi'ee surt~\ce and the hydraulic

conductivity is assumed constant.

Ilov..ever. in the vadose lone. the presence or the air

phase im

pacts the

relationship hetw

een pore

water

pressure and

volumetric

water

content. as well as hydraulic conductivity and volum

etric water content.

This

study w

as conducted

to il1\estigate

the im

pact o

f variations

in thl'

characteri/ation of the unsaturated hehavior o

f a CO

\l't" liner soil on the performance o

f that

liner. T

he liner performance \\as evaluated using w

ater halancc output fi'om

a landfill sim

ulation model.

Materia/.\ u

nd

;\!<.'I!lOds

Thc m

aterials used in this study consist or six clay soils used as landfill cover

liners at live m

unicipal solid waste Iandlills in the m

ctropolitan Detroit vicinity.

One

of the sites had two active excavation regions, yiclding tw

o distinct clay soils. S

amples

of the liner materials w

ere retrieved from

the excavation areas used as horrow soU

t"Ces for

liner construction at each site. T

ahle I provides a sum

mary o

r the geotechnical characterization o

fcach soil.

In suhsequent discussions. the six soils will he referred to

using the numerical designation provided in colum

n one of this tahlc. i.e. the clay liner

from A

uhurn 1fills \vill he designated Soil 1.

lahle 1 -C

ieotechnical Characteri/ation o

f the Clay s

Max

Dr:

Opt.

l.iquid D

ensity M

oisture Inde.\

l.imit

(kW

m')

Content

.."6.32

.)-

31 21.1

X.h 1()

" 2

3.40 43

20.4 10.2

10 2()

UH

J 41

20.2 9.4

14 30

4.R3

47 -20.1

10.3 13

35 .."

.."

6.40 .)..:.

20.3 9.1

lh

.)­

0.51 34

1h.7 22.5

60 R3

Note:

Soil designations

1-6 refer

to soil

from

the follow

ing locations

Auburn

Ililis. W

oodland M

eadows (brow

n clay). W

oodland M

eadows (gra: clay),

Lvergreen. C

anton. and Sm

ith's Creek (vvith

.:2 s"" bv w

eight benton ill' added l. respect ivel:

Although m

ineralogical analysis was not includcd in the scope o

r thi.s research. previous investigations (S

alim et al..

199h) suggest that the mineralogical m

ake·up of

clay soils in southern \:1ichigan arc L

lirly similar and that site-specific m

ineralogical analysis

may

hl' unw

arranted. A

typical m

ineralogical com

position show

s the cia)

li'action to

he dom

inated h)

three m

inerals \vith ()()(y;)

illite. 12°/i) K

aolinite. and

()Oo

chlorite.

Volum

etric prcssure

platc and

pressure m

l'mbrane

extractors w

ere used

to determ

ine the

soil-\vater characteristic

curve (S

\VC

C)

during the

drying cycle

for rem

olded samples o

f each of the six soils.

The m

ethodology is described in AST

M

Standards

02

32

5

(!\ST

M.

1993) and

03

15

2

(!\ST

M.

1(9

4)

for the

plate and

mem

brane apparatus. respectively. The sam

ples were rem

olded using modified Proctor

compactive

effort at

moisture

conditions approximately

2(/1) w

et o

f optimum

. T

he volum

etric water contents w

ere used together with the applied suctions to develop the

SW

CC

s. Data

collected w

as fit

to the

models

described by

rq.

2. !\n

optImIzation

routine was used to lit the param

etric models to the m

easured data by altering the fitted param

eters using an iterative process until the diJTerenee hct\veen the predicted SW

CC

and the m

easured data \\as minim

i/ed. T

he quantity to lx' minim

i/ed is the Sum o

f the S

quared Residuals (S

SR

):

/I

SSR=

'w(e

.-e.) (3)

~

I.

III C

I

where \\, is the w

eighting Llctor w

hich is equal to 1.0. 0\\1 is till' m

easured volumetric

\vater content.

vvhich is

delined as

0\\1 nS

(n=soil

porosity and

S=

saturation corresponding to

0\\1). and OCi

is the calculated

volumetric w

ater content from each

model.

The m

inimi/ation process

for S

SR

\vas perform

ed using tw

o techniques: a standard

statistical packagl:

included w

ith M

icrosoft L

xccl'

and an

optimization

package developed

specifically for

this purpose

hy van

Cienuchten

et al.

(1991). Ik

tailcd

information

re~ardinl!

the cu

ne

lit tl)],

each soil

and each

optimization

technique is presented il; Tabl~

2. In

all cases. the S

SR

is

less than I(r

l • which is

similar to

the S

SR

values

ohtained by

I.eong and

Rahardjo

(19

97

). E

ach o

f the optim

i/.ation approaches

provides an

acceptable lit

to the

experimental

data. w

ith sim

ilar values

of the

SS

R.

The

individual curve

lit param

eters. how

ever. vary

significantly between approaches.

rahlc 2 \an

(,enllchten Clin

e !'it Param

eters

...•~.~.~.~~~.~~

~.~~

..~.~~.~

.. ~~".-

Soil Iecl1lliquc

a 11

III U

r O

s SSR

(kPa x JO

") 9.60X

10·bSolver

0.0035 O

.713J 31.6122

0.1449 0.2419

1 0.2104

I0.22X10

':-R

ETe 0.2856

1.0050 1.5980

0.1408 23.54X

IO·()

Solver

0.5809 56.9445

0.0185 0.2424

OJ35J

23.77X 10'('

2 R

FT

C

0.5828 150.7430

0.0067 0.2408

OJ1

5!

121,59X 10. 0

Solver

0.1125 0.8311

0..1·165

0.0003 0.3106

1 1

0.4

7X

I0'

RU

C

o 1525 1.0050

04CJ97

O.JJ86

0.3065 3.12X

I(r('Sol vcr

0.0 I 02 O

.S711 I U

il46

0.1962

O.323!

3.1 I X 10'('

,~

RI·T

C

o 15(JO

1.0050 I.01ilJ6

0.1702 0

12

02

42.67X

I 0'('S

olver 0.0121

1.6752 I C

J6.8758 0.2700

0.1485 5

RIT

e 001 17

!(J768 1

78

n0

6

02

70

0

OQ

8:>

4()0

6X

10" ~~

~-----

So

h C

I 0

3 ()()8

20

71

7

05

17

8

O1cl88

0:>525 1

67

5X

l0"

---_

.

~._--

_.

~--

--:r­

() R

ITe

OJ·15()

208·18 o 52()3

() 1525 O

S525

,lfJ06

XIO

'

It was observed that the hest-fit saturation vo lum

ctrie \vater content (0:) w

as

relatively independent of the approach used. although there w

as significant variation in

The m

aximum

variation in 0, hctvveen optimization

the value between the six soils.

techniques w

as 5°'0 for

Soil

I. T

he maxim

um variation

in 0, among the

soils was

I-'or a

greater than

100%.

representing the

variation hetw

een Soil

1 and S

oil 6.

saturated soil. the saturated volumetric w

ater content. 8,. is equal to the soil porosity.

Thus. it was expected that there w

ould he significant variation in 0, among the soils.

lhe optim

ized value of the

irreducihle volumetric w

ater content. t)r. show

ed

greater variation hetw

een techniljues

than the

0, variation.

The

variation hetw

een

optimization m

ethods was on the order o

f 15% for all soils. w

ith the exception of Soil 3

which show

ed a much greater dependence on optim

ization technique and a variation on

the order of 100%

. T

he variation in Or hetween soils w

as on the order of 50 0/1}. except

for Soil 3 which exhihited a Or m

uch less than the others.

The other curve tit param

eters: a. n. and m represent the shape and ofT

'iets of

the SW

CC

. T

hese parameters. especially the

latter tvvo. show

significant variation

between soil types and hetw

een optimi/.ation approaches.

How

ever. the curve shapes

that result from

the two approaches appear \ery sim

ilar. as shown in F

igure 1.

The

fhe sim

ilarities observedparam

eter set defining the shape of the S

WC

C is not unique.

in the S

WC

Cs from

the

two

techniques suggests that the added eflrlrt in

using the

RE

Te program

is unwarranted.

Operation o

f the RI:T

C reljuires know

ledge in DO

S

and Fortran form

atting of input tiles.

On the other hand. the S

olver subroutine does not

require formatted input and has a W

indows

interj~lCe.

Modeling o

f now in the unsaturated lo

ne requires input o

f the SW

Cc. usually

in a form sim

ilar to that in Eq. 2. along w

ith the associated curve tit parameters (0,. l\.

a. n. and m. in this case).

Therefore. it is expected that alternative data sets for these

This hypothesis is tested

using theparam

eters will

result in variable model output.

HELP m

odel for prediction of landfill cover liner per!rm

nance.

Leakage !>redic!iol7s

The 1IE

LP m

odel (S

chroeder et al.. 1994) w

as used to sim

ulate the variably

area.saturated

conditions in

a hypothetical landfill

cover liner of one

acre sur!~lCe

HU

.p is

a com

puter program

developed

by the

U.S.

Arm

y E

ngineer W

aterways

The m

odel includesE

xperiment S

tation for the analysis of a landfill w

ater budget.

precipitation. surface runoff evapotranspiration. and snow m

elt in calculation of the

Sik water hudget.

Excess m

oisture acts as additions to soil m

oisture storage in the

Cover soil w

hik

moisture deficits reduce soil m

oisture storage in the cover soil. lhe

model sim

ulates four di Iferent types of soil laycrs. subdivided based on their functions:

Vertical percolation layers. lateral drainage layers, harrier soil liners. and com

bination

geomem

hrane/harrier soil

liners. T

he transport

process lr)r

each layer

is one­

dimensional.

The m

odel simulates quasi-steady state and quasi-tw

o-dimensional flow

(Khanbi Ivardi l't a!"

1(9

5).

The

landlill cover

system

\Vas

modeled

simplistically

as a

t\Vo-com

ponent system

, comprised o

f a 60 cm thick upper sandy layer above a 90 cm

thick compacted

clay layer. T

he upper sandy layer was m

odeled as a \'crtical percolation layer of 60 cm

thickness with good vegetative cover, a surL

lcc slope of ](1'0 and slope length o

f 305 m

eters. T

he run-ofT curve num

ber corresponding to these surf~Ke

layer conditions was

determined internally in IIL

LP

to be 67. T

he climatological conditions (precipitation,

temperature, and solar radiation) sim

ulated \vere those included in the IIEL

P database as 100 years synthetic data for D

etroit. 1\11.

It is typical to designate a compacted clay liner as a harri£'!' soil lay£'!' for IIE

LP

!l1odeling.

Ilowever.

the harrier

soil layer

module

assumes

the cover

liner to

he saturated for the entire sim

ulation. T

herefore, for all runs performed, the 90 cm

thick clay

layer w

as sim

ulated as

a vertical percolation

layer w

hich allow

s for

variably saturated conditions.

Ihis approach allowed the use o

f the SW

CC

parameters fur the

Cover

liner and flennitted

the sim

ulation or variahly saturated conditions \vithin the

Cover liner.

The saturated hydraulic conductivity and the volum

etric saturated water content

(or porosity) used for each soil was that provided in T

ables 1 and 2. respectively. T

he initial

moisture

condition for

all sim

ulations corresponded

to a

constant m

oisture profile throughout the clay liner. w

ith the volumetric w

ater content equal to the field capacity.

The field capacity is defined as the volum

etric water content at a pore w

ater suction o

f 0.33 8

ar while the w

ilting point is the volumetric w

ater content at 15 Bar

(Schroeder et al..

1994). T

he volumetric w

ater contents corresponding to these pore w

ater suctions were determ

ined using the RL

TC

and Solver best-fit techniques for the

van Genuchtcn form

ulation of the S

WC

C (E

q.1) for each of the six cover liner soils.

These values are provided in T

able 3. The unsaturated characteristics determ

ined using the tw

o curve fitting techniques were used to test the sensitivity o

f model results to the

curve-lit procedure adopted. In

each comparative run.

the only input

parameters

varied were the field capacity and w

ilting point. A

ll other quantities remained at the

base run values.

Table 3 ~

Input Param

eters for IILL

P Sim

ulation

Soil T

cchniquc Fe

WP

Solver 0.2203

0.1475 1

RETC

0.21 R5

0.1470

Solver 03351

0.2519 :2

RETC

03351

0.2514

Solver 0.3039

0.2248 3

RETC

0.3023

0.2259

4 Solvcr

0.3] 30 0.2116

RETC

0.3123

0.2104

Solver 0.3480

0.2700 5

RE

Te

0.3472 0.2700

Solver 0.5512

0.3810 6

RETC

0.5514

0.3855

Results (//7d

Discussio/7

Figure 2 show

s the cumulative leakage over a live year period for each o.f the

cover liners simulated using the IIE

LP

model w

ith the Solver param

eters. while Figure

3 shows sim

ilar information

for IIU

J) runs using the R

ET

e parameters.

Tabl e

. presents the cum

ulative output at the end of the five year sim

ulation for each of the SiX

cover liners and both curve-litting techniques.

4

07

0

So

il 1

06

0

o 50

E-;0

40

Ol

C'll..l<:

~

0 3

0

..J 02

0

o 10

o 00

o 1

01

52

0

25

3

03

54

04

5

50

5

56

0

Tim

e (m

on

ths)

Figure 2.

Leakage from

each Soil Liner U

sing the Solver P

arameters

o 70

-~SoiI1

---S

oil2

06

0

~~~-~.

.. ----"

05

0

-------S

oil 3

IQl

04

0

So

il4

,~--_.------'"

-----~--_/

Ol

C'll..l<:

So

il 5C'llQ

lo

30

..J

02

0

o 10

00

0

0 5

10

15

20

25

30

35

40

45

50

55

60

Tim

e (m

on

ths

)

hg

ure 3.

Leakage from

each Soil Liner U

sing the RF rc P

arameters

Variation due to soil type

Although the cum

ulative leakage varies between soils (T

able -+),

the monthly

For all soils, the leakage was greatest during

patterns arc similar (F

igures 2 and 3).

months 17, 19, and 28.

The leakage did not reach steady state conditions during the 60

months sim

ulated. I'or the entire period o

f the simulation, leakage through the cover

liner constructed with Soil 6 w

as significantly less than that of the other cover liners.

At the end o

f the simulation period. the leakage through Soil 6 w

as approximately 60

1y<)

less than any of the other cover liners.

The dim

inished leakage can be attributed to a varidy o

f factors. including the decreased value of saturated hydraulic conductivity for

Soil 6 and the shape of the S

WC

C (suctions increase significantly as the volum

etric w

ater content is reduced). T

he leakage through Soil 5 exceeded that of all other soils

considered in the simulation.

The variation in leakage predicted in these sim

ulations is sim

ilar to that reported by Khire et a!. (199S

).

Table 4

Water B

alance Com

ponents from IIE

LP

Sim

ulation

CU

lllulative Output (111)

So

il

)7]2

Leakage

20

97

ET

R

Ulloff

IA 71 I

L'l Storage

-0.0358

.)03

8

20

07

] IA

71

4

I 0.0379

---_._~.~-

RfT

C

.)811 2

.0)6

7

1.4706 -0.0016

So

lver

.5865 2

.05

67

IA

706 -0

00

75

RfT

C

.)941 2

06

93

1

47

89

-0.0387

So

lver

.5630 2

06

93

-0.0047 ._

--

RIT

e

A.l11

2.2078 IA

73

0

-0.0057

Solver

A2

38

2

.20

78

2(4

3)

2.0

43

)

2.3

90

0

6 2

.39

46

(he cumulative evapotranspiration and

run-off also arc soil-dependent.

The

evapotranspiration component o

f the water balance w

as greatest for Soil 6. Soil 6 w

as Inost

successful at

preventing m

oisture m

ovement

through the

cover liner.

which

increased the m

oisture available for

evapotranspiration. S

urface runoff was alm

ost identical for all soils.

This is expected due to the identical runoff curve num

bers and precipitation history that w

ere used in simulating all six conditions.

Variation dU

l' to optimization tt:chnique

Differences in the predicted leakage due to the optim

ization technique adopted for

the curve-fitting

routine w

ere not

significant. T

he m

aximum

variation

in the

predicted leakage

between

the tViO

techniques occurred

for S

oil 1

and w

as approxim

ately 1JIYtl.

The

evapotranspiration and

runoff components

of the

water

balance \\etT not affected hy the optim

i/.ation technique.

Moisture Iluctuations

The IlL

! Y

model does not provide direct output o

f the calculated volumetric

water

content changes in

the covcr liner soiL

except on

an annual

hasis. M

onthly changes in soil m

oisture storage were calculated using the w

ater halance equation:

f., Soil Moisture =

P-

RO

-F

T-

L

L'1S

W

(4)

where

the quantities

on the

right hand

side o

f the equation

reflect the

cumulative

monthly

values o

f precipitation, runofr,

evapotranspiration. leakage,

and change

in snovv

water,

respectively. T

he snow

w

ater term

accounts

for that

portion o

f the precipitation occurring w

hen local temperatures are helow

freezing. C

hanges in the

volumetric w

ater content of the clay cover liner w

ere calculated for Soil 1 and 6.

The

variation on a monthly basis is show

n in hg

ure 4. A

lthough hoth soils were subjected

to identical climatological conditions. the resulting m

oisture states were very dilT

erent. T

he volum

etric w

ater content spans signilicantly difk

rent ranges

for the tw

o soils.

The volum

etric water content for the co

\er liner using Soil I varies hetw

een 0.15 amI

0.24 w

hile this range for

Soil 6 is

hetween 0.50 and 0.55.

Soil 6 w

as much m

ore efT

ective at prevcnting leakage, and thus a signiticantly larger fraction of the available

moisture

was

stored above

the liner

to m

eet the

evapotranspiration dem

ands. T

herefore, the liner constructed with S

oil 6 maintained com

paratively higher values of

volumetric w

ater content than So

ilS. w

hich allowed greater leakage.

-----

~---

-----­

~-

­

0.55

06

0 f-=...·IV"..·..~/'·...r~ .....··.._~·· ..

05

01: (I)

04

51: 0

04

0U

L.

(I) 0

35

-.

Soil 1

---S

oil 6 ~

I ~

03

0

"0 0.25

>

fL\.~/I:.i'"I':.i'V·~·""J\j\r··_IVJ···.......-A.r-".·

02

0

01

5

I

0.10 f-----.---,---.

o 5

1015

2025

3035

4045

5055

60 T

ime (m

on

ths)

Figure 4.

Volum

etric Water C

ontent Fluctuations in the Soil C

over Liner

llsing the S~('C.

the volumetric w

ater contents of

Fig. 4 can he converted to values o

f pore water suction.

Previous research hy the authors (M

iller et al.. 1998) has show

n that the onset of desiccation cracking m

ay he related to a soil-specilic critical suction value.

If that value is known for the cover soil under investigation. it vm

uld he possihle to

usc the IIEL

P output to determ

ine the likelihood that the covcr liner will

experience suctions equal to or greater than the critical value. W

ith this information, it

\vould be

possible to predict the onset o

f desiccation cracking. For exam

ple, if the critical suction is determ

ined to be 200 kPa for S

oils 1 and 6, the H

EL

P simulation

suggests that Soils

I and 6 are subjected to desiccation cracking 2°;() and 12%

of the

time,respecti\ely.

Further details

regarding m

oisture conditions w

ithin the

cover liner soil are provided in T

ahle 5. A

lthough Soil I experienced significantly lower volum

etric water

contents than Soil 6, it did not exceed the critical suction as frequently as did Soil 6.

Therefore. it

is expected that Soil 6 is m

ore susceptihle to desiccation cracking than Soil 1.

IlowC

\er. this limited analysis provides no m

ethodology to predict the extent of cracking.

Tahle 5 -

Details R

egarding Moisture C

onditions h1countered in Clay C

over Liner

Soil M

inimum

M

aximum

Pore ';0 o

f Tim

e %

of T

ime C

ritical V

olumetric W

ater W

ater Suction S

aturated Suction E

xceeded

1.7

6 1

00

0

11.7

Table 5 indicates that the cover liners constructed w

ith Soil

I and Soil 6 are expected to

reach saturation during IRJ<Y<) and IO.O<Yo of the 5 year sim

ulation duration, respecti\ely.

Ihis prediction suggests that analyses of clay cover liners that rely on the

assumption o

f saturated conditions m

ay be

inappropriate. S

uch assum

ptions likely

result in over-estimates o

f the amount o

f leakage through the cmer liner. w

hile under­estim

ating the potential I'llI' desiccation cracking.

5;ulIl/Jwrl' a!7d ( 'o!7clusio!7s

The variably saturated perform

ance of six com

pacted cia) soils as cover liners

was

compared

using the

HE

LP

m

odel. T

he relationship

between

the unsaturated

volumetric w

ater content and pore water suction w

as determined in the laboratory for

each soil using the pressure plate apparatus.

The data w

as fit to the \an

(Jenuchten m

odel of the soil-w

ater characteristic curve (SW

CC

) using two different optim

ization techniques.

The shape o

f the two resulting S

WC

Cs \vere very sim

ilar for each of the

six soils. although the values of the individual curve-fit param

eters \aried significantly (exceeding I ()()O

;) in some cases).

The variation in the shape o

f thl' resulting SW

CC

s j'llr the six soils w

as much greater than the \'ariation due to optim

ization technique.

The cover liner \vas

modeled

as a

I'ertical jJercolatio!7 larer rather than

the typical approach o

f using the IIEL

I) option I'llI' a harrier soil la)'er. T

his was done to

o\erco

me the assum

ption of continuous saturated conditions inherent in the harrier soil

larer specification. T

he design of each cover lincr sim

ulated \vas identical. the only variable being the input data set llsed for specificC

ltion of the sC

lturated and unsaturated soil characteristics.

T\\o

simulations w

ere completed for each o

f the co\cr liner soils

that included

the p

arameter sets

resulting from

both

types o

f swe

e

curve-fitting routines. T

he results from

the H

EL

P sim

ulations sho

wed

that the mo

nth

ly patterns o

f leakage w

ere similar for all co

ver liner soils.

In addition, leakage thro

ug

h the co

ver

liner was relatively unaffected by the Sw

ee

curve-fitting routine adopted. H

owever,

the leakage varied significantly from o

ne soil to another.

Th

e cov

er liner constructed w

ith Soil 5 ex

hib

ited the greatest leakage, w

hile the cov

er liner constructed with S

oil 6 exhibited the least leakage.

Alth

ou

gh

it is w

ell kn

ow

n that the relationship b

etween

volum

etric water content and pore-w

ater pressure is hysteretic' only the drying limb o

f the soil w

ater characteristic curve was used in these sim

ulations. T

here m

ay be great

variation in the soil water characteristic b

ehav

ior on the w

etting and dry

ing

sides of the

hysteretic curve.

Therefore,

it m

ay

be unrealistic

to adopt

the drying

curv

e for

simulation o

f the infiltration process.

Evapotranspiration

and ru

no

ff w

ere m

uch less

variable than

the leakage

component.

Th

ere was essentially no variation attributed to the different o

ptim

ization

approaches, w

ith only m

ino

r variations due to cov

er liner soil specification.

Ch

ang

es in soil moisture storage w

ere calculated on a mo

nth

ly basis for S

oils 1 and

6. T

he

range in

volumetric

water contents

predicted for

each soil

was very

different. T

he volum

etric water co

nten

ts were converted to pore w

ater suctio

ns using

the swe

c and com

pared

to an assum

ed critical value o

f 20

0 kP

a. C

ritical con

ditio

ns

were ex

ceeded

1.7%

and 11.7%

of the tim

e for Soil 1 and S

oil 6, respectively. B

oth

soils

were

predicted to be unsaturated d

urin

g a

significant portion of the

five year

simulation

period (8

2%

and

90

%

of the

time

for S

oils 1

and 6,

respectively). S

imulation

of the

variably saturated

nature o

f the flow

th

rou

gh

the

cov

er liner

is im

portant to realistically predict leakage behavior.

Ackn

uw

ledg

emen

t

Partial funding for this study w

as provided by the National S

cience F

oundation under

Gran

t C

MS

-97

13

92

2.

Additionally,

the W

ayne S

tate U

niversity F

aculty G

raduate Research

Assistantship ~l\vard program

supported student inv

olv

emen

t on the project.

Relerences

AS

TM

(1993). "Stan

dard

Test M

etho

d for C

apillary-Moisture R

elation

ship

s for Co

arse -and M

ediu

m-T

extu

red S

oils by Pressure-P

late Ap

paratu

s:' Designation: 0

2325-68, A

merican S

ociety for Testing and M

aterials, Annual B

ook of A

ST

M S

tandards.

AS

TM

(1994). "Stan

dard

Test M

ethod for Capillary-M

oisture Relatio

nsh

ips lor C

oarse

-and M

ediu

m-T

extu

red

Soils

by P

orous-Plate A

pp

aratus,"

Designation:

D

3152-72, A

merican

Society for T

esting and Materials, A

nnual Book o

f AS

TM

Standards.

Brooks, R. H. and Corey, A. T. (1964). "Hydraulic Properties of Porous Medium," IIydrology Paper No.3, Civ. Engrg. Dept., Colorado State University, Fort Collins, CO.

Fredlund. D. Ci., and Xing, A. (1994). "Equations for the Soil-Water Characteristic Curve," Canadian Gcotcchnical Journal, Vol. 31, 521-532.

Khanbilvardi, R. M., S. Ahmed, and P. J. Gleason (1995). "Flow Investigation for Landfill Leachate (1'111)," J. Env. Fngrg.. ASCI-:, 121 (1),45-57.

Khire, T\L Benson, C. and BoscheL P. (1996). "Water Balance Modeling of Earthen Final Covers:" .Inl. Geotcch. And Cicoenv. Engrg. ,123 (X), 744-754.

Leong, E. C. and Rahardjo, II. (1997). "Review of Soil-Water Characteristic Curve Equations:' J (Jcotech. and Geocn\'. Ic'ngrg., ASCE, 123( 12), 1106-1117.

Miller, C. L Mi, ILand Yesiller, N. (1998). "Experimental Analysis of Desiccation Crack Propagation in Clay Liners:' .In! Amcr. TVater Res. Ass., AWRi\, 34(3), 677­686.

Schroeder, P. R., Dozier, T. S., Zappi, P. A., McEnroe, B. M., Sjostrom. J. W., and Peyton. R.L. (1994). "The Hydrologic Evaluation of Landfill Performance (IIELP) Model: Engineering Documentation for Version 3:' EPA/600/9-941168b, U.S. Environmental Protection Agency Risk Reduction Engineering Laboratory, Cincinnati, 011.

Salim, L Millcr, C. and IIowaI'd, .J. (1996). "Sorption Isotherm-Sequential Extraction Analysis of I!cavy Metal Retention in [.andfill Liners," Soil Scicnce Society ojAmcric(l Journal. ()O( 1), 107-114.

van Genuchten, M.T. (1980). "A Closed Form Equation for Predicting the Ilydraulic Conductivity of l Jnsaturated Soils," Soil Sci Soc. Am. .1.,44, 892-898.

van Genuchten, M., Leij, F., and Yates, S. (1991). "The RETC Code Quantifying the Hydraulic Functions of Unsaturated Soils:' Rep. No. EPA/600/2-91/065. U.S. EPA Ofc. Of Res. And Devel., Washington, D.C.