To line or not to line - the effect of geometry on

seepage rates from tailings storage facilities

John Waterhouse1, Robin Friday2
1
2

Golder Associates, 441 Vincent Street West, Leederville, W A 6007, Australia

Golder Associates, 25 Burwood Road, Hawthorn, Victoria 3122, Australia

Abstract: Seepage from conventional tailings storage facilities (TSFs) is inevitable. This paper
discusses the control which geometric and hydraulic factors may exert on long term seepage.
For a large TSF overlying a relatively shallow and poorly transmissive groundwater flow
system, the rate of downward seepage may initially be high. However, when the zone beneath the
TSF has saturated, the longer term seepage rate may be limited by the capacity of the groundwater
system to transmit seepage laterally. In this situation, an ordinary liner may only slightly reduce
long term rates of seepage. In an extreme case, only a near-perfect synthetic liner may control the
seepage rate more effectively than the inherent properties of the groundwater flow system.
For a smaller TSF with a deeper and more transmissive groundwater system, the long term
seepage rate may not be limited by the system characteristics so much as by the hydraulic
conductivities of the tailings and the unsaturated materials immediately under the TSF. The
volume beneath the TSF may not saturate and seepage rates will not be limited by the capacity of
the groundwater system to transmit seepage laterally. In these situations, liners may be more
effective if control of seepage is required.

1 INTRODUCTION
Tailings storage facilities (TSFs) are used for the permanent disposal of wastes
from the treatment of a variety of ores, typically utilising a slurry deposition
method. Evaporation ponds for disposal of waste water such as mine dewatering
discharges and tailings liquours behave in a similar manner to TSFs and are
implicitly included in this discussion. Seepage from other tailings disposal
methods such as dry stacking and paste technology are not considered in this
paper.
Unless elaborate methods, such as double liner systems, are used, some
seepage from a TSF is an inevitable consequence of the deposition of a slurry
within bounding embankments.
The purpose of this paper is to explore some of the controls on seepage rates
which become significant when large storages overlie groundwater flow systems
of limited thickness and low hydraulic conductivity.
The paper is written from the perspective of the Western Australian mining
environment, where environmental management is evolving in the context of
numerous mines in remote, arid locations. Whilst a detailed description of the
hydrogeology of Western Australia is beyond the scope of this paper, a short
description of the most common hydrogeological setting for the facilities which
are discussed is appropriate.
482

Typical Western Australian base metal or gold mining and processing projects
are located in a large area of Archaean rocks with generally low topographic
relief with a generally similar geological profile. This profile comprises thin
soils or alluvial material underlain by a lateritic profile of variably ironcemented, extremely weathered rock of various lithologies. The weathered rocks
grade into fresh rock at depths that vary from zero (fresh rock at surface) to, more
commonly, several tens of metres. Whilst at some localities the rocks themselves
are permeable enough to form heterogeneous aquifers, it is more typical for
hydraulic conductivities, particularly in the extremely weathered zone and the
fresh rock, to be so low that the materials below the water table can barely yield
water to wells. Often, the most permeable material is found in the transitional
weathering zone immediately above fresh rock (Morgan, 1993)
The depth to the water table is commonly in the approximate range 10 m to
40 m, although of course this depth varies according to the local topographic
relief. If, as is common, the main transmissivity is in and near the base of the
weathered zone, the thickness of the aquifer may be small and even when the
saturated thickness is increased by seepage, the transmissivity may not increase
significantly.
Groundwater quality in the weathered to fresh rocks is variable, ranging from
less than 1 000 mg/L total dissolved solids to 5 000 10 000 mg/L (Allen, 1996)
or more in places.
The climate in most of this area is semi-arid, with annual rainfall typically less
than 300 mm and annual pan evaporation in excess of 2 000 mm. Soils are
usually low in moisture content, clayey and with typically negative pore
pressures.
Larger mining and processing projects have potential to be active for periods
of 20 30 years.
2 SEEPAGE MECHANISMS
We start with a consideration of the mechanisms by which seepage develops
progressively beneath a TSF. There are three stages in the development of
seepage flow from a TSF:
1. Initial seepage as the first layer of tailings is deposited, when the decant pond
does not have an underlying layer of tailings.
2. Seepage through a progressively thickening tailings deposit.
3. Continued seepage, at declining rates, by drainage of the tailings deposit after
cessation of operations.
In parallel, but not necessarily corresponding stage by stage, there are five
possible stages in the evolution of the saturated-unsaturated profile and hydraulic
gradients beneath the TSF:
1. Initial downward movement of a saturation front, mostly directly beneath the
decant pond.

483

2. Mounding of the underlying water table in response to an increasing flux

through the unsaturated zone.
3. Coalescing of the downward moving saturated seepage front with the upward
moving water table (unless the water table aquifer has high transmissivity).
4. Development of an overall gradient for lateral flow away from the TSF, with
fully saturated conditions beneath the decant, and a slowly rising water table.
5. Progressive decay of the groundwater mound after the cessation of
operations.
The profile evolution may not always develop exactly in this way, for
example, if horizontal layers of low vertical hydraulic conductivity cause the
development of a shallow, perched system, in turn possibly preventing full
saturation of the profile. It is important to recognise that the geological systems
which we are considering are often heterogeneous at the scale of a TSF and also
variably anisotropic.
The downward seepage through the unsaturated zone is likely to be as a
saturated front if there is no effective liner and the seepage source is relatively
localised, as is anticipated when a tailings decant pond is formed. Otherwise the
downward seepage may occur as an unsaturated wetting front.
It is important to the understanding of this paper to recognise the potential for
high rates of initial seepage due to a combination of void filling, the large area
over which saturated tailings may drain and the possibility of the hydraulic
gradient exceeding one as a consequence of soil suction. The factor which
opposes the high seepage tendency is the relationship between the degree of
saturation and the vertical hydraulic conductivity which impedes the downward
flux. Simple calculations of downward seepage rates using Darcys Law and
saturated hydraulic conductivity values obtained by field testing will tend to
predict high rates of seepage.
3 LINER PROPERTIES
Liners are perceived by some as providing a control over seepage rates by
sealing the base of facilities such as TSFs or evaporation ponds. Whilst the
hydraulic conductivity of a liner is by definition expected to be lower than that of
the material above or beneath it, the long term effect of the liner will depend
upon the amount of hydraulic resistance that it imparts to the flow system as a
whole.
The hydraulic conductivity of liners varies over many orders of magnitude.
Our experience in Australia suggests that in practice it is difficult to achieve
vertical hydraulic conductivities lower than 10-8 to 10-9 m/s for compacted clay.
Given practical difficulties in maintaining appropriate moisture content during
construction and before coverage with tailings is complete, these values may be
optimistically low.
Synthetic membranes such as high density polyethylene (HDPE) may, under
ideal conditions, have hydraulic conductivities as low as 2 x 10-15 m/s, (Giroud

484

and Bonaparte, 1989). However, even minor defects increase the effective
hydraulic conductivity by several orders of magnitude. The effective resistance
to vertical seepage of various liners is compared in Table 1.
The effective resistance of the two HDPE liners was calculated using the
HELP3 computer program produced by Schroeder et al, (1994). This program is
typically used for hydrologic evaluation of landfill performance and takes
account of the typical defects which have been observed in manufacturing and
placement of synthetic membranes. The three main aspects affecting synthetic
membrane liner performance are:
Manufacturing defects, usually small size holes (pinholes)
Larger holes left during installation due to damage, incomplete coverage or
faulty welding
Placement care affecting the uniformity of contact between the membrane
and adjacent low permeability soil
For the two HDPE installation standards we assumed:
Poor
2
10
Poor

Pinholes/hectare
Installation holes/hectare
Placement quality

Good
1
2
Good

Table 1 Comparison of liner resistances

Liner type

Effective
resistance b/K
(s)

Compacted clay K = 10-8 m/s, 200 mm thick

2 x 107
-9
Compacted clay K = 10 m/s, 600 mm thick
6 x 108
HDPE poor installation adjacent to
1.4 x 109
-8
material with K = 10 m/s
HDPE good installation adjacent to
1.4 x 1011
-9
material with K = 10 m/s
Tailings K = 10-8 m/s, 10 m thick (saturated)
1 x 109
-9
Compacted clay K = 10 m/s, 600 mm thick
1.6 x 109
under 10 m of tailings
HDPE poor installation under 10 m
2.4 x 109
tailings
HDPE good installation under 10 m
1.4 x 1011
tailings
b = liner thickness
K = effective vertical hydraulic conductivity of liner
HDPE = High density polyethylene

485

Leakage with
water head 10 m
above base
(m3/day/km2)
43000
1400
620
6
860
540
360
6

4 EFFECTS OF DIMENSIONS AND LINER PROPERTIES

The important difference between seepage from large and small TSFs is that the
long term seepage from a large storage, that is greater than about 1-2 km2 has
greater potential to be constrained by the geometry and properties of the
groundwater flow system. This is due to the potential for a large discrepancy
between the relatively small cross-sectional area for lateral flow under a low
hydraulic gradient and the relatively large area for downwards seepage under a
hydraulic gradient which may exceed one.
An estimate of the radial seepage capacity of the aquifer beneath the TSF may
be made to illustrate the focus of this paper using the Thiem steady-state formula.
Q = 2 T s / ln(R/r)
where:
Q = steady radial flow
T = transmissivity
s = groundwater mound height
r = radius to which the mound is applied
R = radius of fixed head boundary
Table 2 shows some results of lateral flow calculations for TSFs of different
radii for two different aquifer transmissivities assuming R = 3.5 km and s = 20 m.
Table 2 Aquifer radial flow estimates
TSF
area
(km2)

TSF radius
(km)

1
2
5
10

0.56
0.80
1.26
1.78

Radial flow rate (m3/day)

Transmissivity
1 m2/day

Transmissivity 10 m2/day

69
85
120
190

690
850
1200
1900

Comparing the results in Tables 1 and 2 it can be seen that for an aquifer
transmissivity of 1 m2/day the aquifer has less lateral flow capacity than the liner
for all cases except the HDPE with good installation characteristics. This
transmissivity value has been adopted from an actual case study in which the
hydraulic conductivity values calculated from field testing of the weathered zone
aquifer were in the range 10-6 to 10-7 m/s for an aquifer thickness of about 25 to
30 m. If a good quality of liner could be installed in practice, it would be
intended that most of the seepage water would be collected in a toe drain or some
form of blanket drain above the liner and that only a small amount would seep
down to groundwater.
No large tailings storage facility in Western Australia has ever been lined
effectively and drained in this way to our knowledge.

486

For all the other liner types it can be expected that seepage will cause the
groundwater to mound by up to 20 m or more above the pre-operational water
table. In this situation, excess water (that which cannot be transmitted by the
aquifer) may either appear as seepage around the toe of the TSF or be collected
by sub-surface drains around the toe of the TSF. Seepage around the toe of the
TSF is environmentally undesirable and can lead to deposition of salts and waterlogging of plant roots, causing conspicuous death of adjacent natural vegetation.
For an aquifer with transmissivity of approximately 10 m2/day (or greater) the
analysis shows that the seepage outcome can be influenced by different types of
liner beneath the tailings. This is because the liner will exert the main control
over the combined system of downward seepage and lateral flow in the aquifer.
As the tailings deposit accumulates and consolidates, further resistance to
seepage will develop.
Based on the results in Tables 1 and 2, Figure 1 compares the resistance to
flow through the liner and through the aquifer for different TSF areas. The
resistance is expressed as driving head divided by flow. For the liner this is equal
to H/Q = b/KA where H is the head of water over the liner and A is the vertical
leakage area. For the aquifer the resistance to steady flow is estimated as s/Q =
ln(R/r)/2T. The medium HDPE liner has b/K = 1.4 x 1010 s.
Clay 0.2 m, K = 1x10-8 m/s
Clay 0.6 m, K = 1x10-9 m/s
Poor HDPE liner
Medium HDPE liner
Good HDPE liner
Aquifer T = 1 m2/day
Aquifer T = 10 m2/day

Resistance to flow (H/Q day/m2)

10

0.1

0.01

0.001

0.0001

1E-005
0

Area (km2)

10

Figure 1 Resistance to seepage for different liners and TSF areas

5 MODELLING
We have carried out transient flow modelling for several cases including tailings
storages and evaporation ponds. For this work we have used the computer
program Seep/w. Figures 2 and 3 show results from an axisymmetric model for

487

the case of an evaporation pond 980 m radius (3 km2 area) with a poor HDPE
liner. The geology was assumed to be uniform across the site and typical of the
Western Australian area of interest, with 1 m of alluvial soil over 9 m of
ferricrete over 26 m of extremely weathered bedrock. The liner was modelled
using 0.5 m thick finite elements with a vertical hydraulic conductivity selected
to match the resistance listed for this case in Table 1.
The model extends to 5 km radius where the groundwater head was fixed at
RL 20 m. The initial condition for the model included surface infiltration at a
low rate of 0.025 mm/year which raised the groundwater level under the centre of
the pond to RL 19 m.
The depth of water in the pond was raised linearly with time from 0 to 1 m
over two years, held for 28 years and then reduced linearly with time from 1 to 0
m over two years. The infiltration rate was then reduced to the original value
over the whole model area.
Figure 2 shows the water table rising due to seepage from the pond. Figure 3
shows the water table subsiding after the pond is emptied.
Figure 4 shows the calculated cumulative flow of water from the evaporation
pond for four different cases:
No liner
Clay liner 0.6 m thick with hydraulic conductivity 1 x 10-9 m/s
Poor HDPE liner as used in Figures 2 and 3
Medium HDPE liner
With the medium liner there was a rise of only 2 m in the water table beneath the
centre of the pond. As the water table remained well below the liner, seepage
was controlled by the liner. In the other three cases the water table mounded up
to the liner within about 1 to 15 years and the aquifer had a significant effect on
the total seepage from the pond. The results in Figure 4 are consistent with those
expected from the general study leading to Figure 1.

Evaporation pond with

HDPE liner, poor

Alluvial Soil:
Kh,v=1x10-6 m/s

0
-5

10 years

Ferricrete: Kh=1x10-7 m/s Kv=5x10-8 m/s

-10
-15
-20
-25

30 years
20 years
0 years
Extremely Weathered Bedrock:
Kh=3x10-7 m/s Kv=1.5x10-7 m/s

-30
-35

Figure 2 Water tables at 0, 10, 20 and 30 years after start of filling pond

488

5
0
-5
-10

Alluvial soil:
Kh,v=1x10-6 m/s

Evaporation pond drained

10 years
50 years
100 years

Ferricrete: Kh=1x10-7 m/s Kv=5x10-8 m/s

-15
-20

Extremely weathered bedrock:

Kh=3x10-7 m/s Kv=1.5x10-7 m/s

-25
-30
-35
0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Figure 3 Water tables at 10, 50 and 100 years after draining pond

For the no-liner case the initial seepage rate was limited to no more than 7000
m3/day, taking account of the water supply to the pond.
Modelling of tailings storages is similar to evaporation ponds but slightly
more complicated as layers of tailings must be progressively added to the model
and, depending on the aquifer and the liner used (if any) the tailings themselves
can influence the seepage results.
No liner
Clay liner 0.6 m, K = 10-9 m/s
Poor HDPE liner
Medium HDPE liner

Cumulative seepage to ground (ML)

6000

5000

4000

3000

2000

1000

0
0

10

20

Time (years)

Figure 4 Calculated seepage from evaporation pond

489

30

6 DISCUSSION
The analyses and a consideration of the results of modelling various seepage
cases show that the effectiveness of liners to control seepage depends on more
than the hydraulic conductivity of the liner itself.
This outcome demonstrates a need to assess seepage in the context of the
overall groundwater flow system within which the seepage will migrate laterally
after it has passed through the base of the TSF or evaporation pond. In the case
of aquifer systems that are shallow and thin relative to the area of the TSF, the
geometry and properties of the groundwater flow system may exert considerable
resistance to lateral flow. This resistance may be greater than most liners would
provide to limit vertical seepage.
For most TSF situations, recognition of the nature of seepage control is an
important aspect of the design and regulatory approvals process. For example,
there is little benefit in spending large sums of money lining a TSF if the liner
will not significantly change the long term rate of seepage from the facility.
Similarly, there is little benefit to a regulator in requiring a sophisticated liner
unless it will be more effective than the natural groundwater flow system.
A related issue is the long term stability and rehabilitation of the area of the
facility. For a tailings facility, seepage greatly assists the process of consolidation
of the material which has been deposited. Whilst there may appear simplistically
to be advantages in preventing seepage for environmental reasons, a poorly
consolidated tailings deposit may be more difficult to manage in the long term,
since reshaping and revegetation may be difficult and the geotechnical stability
may be compromised.
In turn, this emphasises the need for good hydrogeological understanding to
be developed early in the investigation of any new site which may be used for
tailings deposition or for evaporation ponds.
ACKNOWLEDGEMENTS
The authors thank their Golder Associates colleagues, particularly Mr David
Williams and Dr Stephanie Watson, for many useful and stimulating discussions
on numerous seepage analysis studies in recent years.

REFERENCES
Allen A.D., 1996. The Hydrogeology of the Northeastern Goldfields, Western Australia.
Western Australia, Geological Survey, Record 1996/4.
Morgan K.H., 1993. Dewatering open pit mines, Yilgarn Craton, Western Australia. In
Robertson I. et al (ed) Proc. International Mining Geology Conf., 285-295.

490

Giroud J.P.& Bonaparte R. 1989. Leakage through liners constructed with geomembrane
liners parts I and II and technical note. Geotextiles and Geomembranes 8(1), 2767, 8(2), 71-111, 8(4), 337-340.
Schroeder J.R., Dozier T.S., Zappi P.A., McEnroe B.M., Sjostrom J.W., Peyton R.L.
1994., The hydrologic evaluation of landfill performance (HELP) model, U.S.
Army Corps of Engineers, Waterways Experiment Station, Vicksburg, MS

Stosowa drena czy nie wpyw czynnikw

geometrycznych i
hydraulicznych na stopie przesczania ze zbiornikw flotacyjnych
John Waterhouse, Robin Friday
Streszczenie: Przesczanie z tradycyjnych zbiornikw flotacyjnych (TSF) jest
nieuniknione. W artykule dyskutuje si jakie czynniki geometryczne i
hydrauliczne mog wpywa na dugotrwae przesczanie. Dla zbiornikw
poflotacyjnych pooonych w stosunkowo pytkim i sabo przepuszczalnym
rodowisku wd podziemnych stopie pionowego przesczania moe by
pocztkowo wysoki. Jednake, po nasyceniu strefy poniej TSF, dalsze
przesczanie moe zosta ograniczone przez zdolno systemu wd
podziemnych do przewodzenia poziomego. W takiej sytuacji, wykonany system
drenau moe jedynie nieznacznie ograniczy dugookresowy stopie
przesczania. W przypadkach ekstremalnych, tylko prawie idealnie wykonany
system sztucznego drenau moe kontrolowa wielko przesczania.

491