Neil your Flaffing...your research is wanting to say the least...just another for you to chew on..
Tracas and Modelling in Hydrogeology (Proceedings of the TraM '2000 Conference
held at Liège, Belgium, May 2000). IAHS Publ. no. 262, 2000. 347
Measurements and models of solute behaviour in
the dual-porosity Chalk aquifer of the UK: the
applicability of tracer tests
SALLY WATSON, WILLIAM BURGESS, JOHN BARKER
Groundwater Tracing Unit, Hydrogeology' Group, Department of Geological Sciences,
University College London, Gower Street, London WC1E 6BT, UK
e-mail:
sally.watson@ucl.ac.uk
JULIO CARNEIRO
Universidade de Evora, Dep. Geociências, Col. Luis Verney, Apartado 94,
7002-554 Evora Codex, Portugal
SARA HAZELL
Dames & Moore, Wimbledon, London SW19 4DR, UK
Abstract Tracer tests provide direct observations of solute transport in
groundwater, but how relevant are small-scale tests to the prediction of longterm
behaviour of contaminants in a dual-porosity aquifer? To address this
question for the Chalk aquifer of the UK, a catchment at Tilmanstone
(southeast England) is being investigated; disposal of coalfield brine directly
to the Chalk aquifer took place over many decades and a large plume of
contamination has developed. Transport properties of the Chalk are being
established across a range of space and time scales. Results of laboratory-scale
diffusion experiments, single-borehole dilution tests, and borehole-to-borehole
tracer tests in the Chalk are reviewed in the context of observations of chloride
flux through the Eastry valley at Tilmanstone over several decades. Modelling
of groundwater flow and chloride transport, using a dual-porosity framework,
provides both the context and the criteria for judging the applicability of the
field-scale tracer tests.
INTRODUCTION
Groundwater-mediated solute transport in fractured porous media is dominated by advection
in fractures and diffusive exchange between fracture water and the matrix pore water (e.g.
Barker, 1993), their relative significance depending on the scale and duration of transport.
Limited knowledge of the complexity of aquifer properties and flow geometry severely
restricts our ability to predict solute migration at the field scale. Tracer tests have the
advantage of revealing aquifer transport characteristics directly, but there is a discrepancy
between the scales and durations of practicable tracer tests and of contaminant plumes.
To address this issue, we present observations at various scales of solute transport
in the Chalk aquifer, the epitome of the dual-porosity aquifer. We concentrate on the
Tilmanstone catchment in Kent, southeast England, where disposal of coalfield brine
in open lagoons over many decades has resulted in a large plume of contamination of
over 25 km2 in area (Fig. 1). Results of historical reviews and field investigations in
the 1970s (Ffeadworth et al, 1980) and predictions of an early model (Bibby, 1981),
form the basis for our current research.
348 Sally Watson et al.
^ D i s c h a r g e lagoon ©Town o 4km
y- Stream
"150 Chloride concentration mg/l (1994)
Fig. 1 Site location , indicating the extent of the saline plume.
The Tilmanstone catchment
A number of deep coalmines were active in southeast Kent from the early 1900s until
final closure in 1986. The Tilmanstone mine is situated at the head of a valley in the
Upper Chalk, which dips to the northeast and drains to springs in the northeast (Fig. 1).
The hydraulic gradient along the valley is approximately 0.008.
Brine pumped from the deep coalfield was discharged to unlined lagoons on the
Chalk surface at the head of the valley between 1906 and 1974. Flow rates and
chloride concentration increased from 1.4 Ml day"1 and 45 mg l"1 in 1906 to
11.8 Ml day"1 and over 2000 mg l"1 in 1958; thereafter the flow remained approximately
steady as the concentration increased further to 5200 mg l"1. About 190 Mm3
of brine, close to 320 x 106 kg of chloride, entered the Chalk aquifer over a 68 year
period. The resultant contaminant plume provides an opportunity to test the value of
tracer tests at different scales in predicting the long term plume development.
OBSERVATIONS OF CHLORIDE MOVEMENT IN THE CHALK
Catchment-scale chloride transport
The background chloride content of the Chalk aquifer is low: 30-40 mg l"1. By 1929,
chloride had reached 200 mgl"1 immediately to the north of the mine. By 1949 the
maximum chloride concentration in the aquifer was 900 mg l"1. By 1977, groundwater
chloride concentration had increased to 5000 mg l"1 within 1000 m of the lagoons, and
1000 mg l"1 at Eastry (Fig. 2). The aquifer was considered polluted over an area of
27 Ion", the plume being 8 km long in the direction of the valley and 2 to 5 km wide.
Headworth et al. (1980) estimated that 98 x 106 kg of chloride (30% of the total load),
had been discharged at the springs by the early 1970s. A fairly steady concentration of
slightly above 200 mg l"1 corresponds to an annual flux of 2-3 x 106 kg of chloride to
the streams since then. Since 1977 the centre of the plume has moved 2 km down the
Measurements and models of solute behaviour in the dual-porosity Chalk aquifer of the UK 349
SW NE
Tilmanstone
settling lagoons
Lower Venson ^
Farm E a s t rV
\1000 f..._2I>be „_,.____, ^
~ I 1 ™ / ; ; / -»
d o o '• I / :
- - i ~ f- Y
» - —JQQ. -
V * - : —
Key
Lined borehole
Unlined borehole
Borehole ID numbers
Porewater chloride (mg/l)
Fissure water chloride (mg/l)
\ 5 0 0
••-,200
Scale
1
500
20
- 10
- 0 S
•150
metres
0 250 _
metres
Fig. 2 Extent of contamination (1974) with pore water and fissure water profiling (after
H e a d w o r d s a/., 1980).
valley and the maximum chloride content (in pumped groundwater) has declined to
1500 mgl"1 , yet the total extent of pollution, as defined by the 150 mg F1 contour
(Fig. 1), has expanded, (Peedell, 1994; Ffazell, 1998).
A chloride pore water profile determined in 1999 (Fig. 3) suggests that the active
fractures are quite closely spaced, since the profile is locally smooth. The low
CI concentration mg/l
0 500 1000 1500
•
1
•
1 *
* t
•
Fig. 3 Chloride pore water profile 1999. Data collected as part of EU FRACFLOW
project (www.fracflow.dk) by British Geological Survey.
350 Sally Watson et al.
concentrations between 45 and 58 m b.g.l. and below 80 m b.g.l, at which depths bulk
permeability is low (Fig. 4), indicate that less chloride was transported into these zones.
Permeability variation with elevation EastryBHA
Permeability m/d
0 20 40 60 80 100 120
E
TO
Q
-20
S -30
<
E
i X
MX.
f•• • i
x A A
A à •- *
• 1
A
A K A
A X
•f
°n
®BH3
• BH4
ABH5
*BHA
ABH8
OBH7
I
Ct-Cb/Co-Cb
0 1
-50
/ .7
/ ,•;
- i - . . . -'-.\
" - \ - n.
\\ vw
/ 4
\ tj
\ v
Minutes since start
• 45
-165
915
Fig. 4 Single borehole dilution test results.
Borehole-to-borehole tracer tests
Natural-gradient borehole-to-borehole tracer tests are currently underway at the site.
Such tests provide information most effectively at the intermediate scale (Ward et al.,
1998), where the significance of diffusion depends on the distances between the
boreholes and the characteristics of the rock. For tracer residence times of a few
minutes, diffusion is negligible. The tracer test reveals the kinematic porosity of the
fractures but nothing about the matrix. For tracer residence times of a few hours, there
will be time for diffusion of a few millimetres into the matrix, which is significant in
comparison to the fracture apertures. The test will reveal behaviour characterized by
the fracture porosity and the area for diffusive exchange which is only significant
during early plume development. Where channelling is important the effective
apertures may be of the order of 1 cm and diffusion will have less effect. It is unlikely
that borehole-to-borehole tracer tests will provide suitable information to assess
pollutant transport over distances of kilometres and time scales of years or decades.
Single-borehole dilution tests
Interval dilution tracer tests can yield vertical profiles of the Darcy velocity of
groundwater flow (e.g. Hazell, 1998). With knowledge of the local groundwater head
Measurements and models of solute behaviour in the dual-porosity Chalk aquifer of the UK 351
gradient these can be interpreted as profiles of hydraulic conductivity which are more
appropriate for solute transport modelling than values from pumping tests, as they
involve smaller volumes of aquifer. Dilution tests at six boreholes at Tilmanstone
(Fig. 4) show that the Chalk becomes more permeable down-valley, and a marked
reduction in hydraulic conductivity is evident at a depth of 40 m. Our results show
considerably more detail than previous pumping tests and are consistent with
geophysical logs, which confirm a greater degree of fracturing in the upper 40 m.
Laboratory scale diffusion experiments
Diffusive exchange occurs between matrix pore water and mobile groundwater in
fractures and fissures at the sub-metre scale: the kinematic porosity, fracture and
matrix porosities, and the diffusion coefficient of the saturated matrix are the key
parameters (Barker et al, 2000; Fretwell et al, 2000). Diffusion coefficients for
chloride in the Chalk of southern England, summarized in Fretwell (1999), range from
5.2 x 10"11 to 1.3 x 10 - 9mV, but none relate specifically to Tilmanstone. Tests are
being conducted on recently acquired core material.
MODELLING CHLORIDE FLUX IN THE TILMANSTONE VALLEY
A finite-element dual-porosity model, calibrated against the field observations up to
and including the surveys of the 1970s, was developed by Bibby (1981). Transport was
described according to a parallel plate representation, and diffusion within the matrix
water was described by Fick's second law. The model was two-dimensional, it ignored
density effects, and transmissivity and storativity were independent of saturated
thickness. It was probably the first catchment-scale dual-porosity model applied to a
groundwater contamination incident.
The Bibby model predicted the future development of the plume, the chloride
concentration at the Eastry borehole (Fig. 5) and of the springs, and the flux of chloride
from the springs up to the year 2008. However, the parameters of the calibrated
transport model were far from realistic (Carneiro, 1996), and the aquifer is
rehabilitating at a rate considerably slower than the model predicted. The effects of the
seasonal desaturation of fractures (Fretwell et al, 2000), accentuated by droughts, and
1600
1930 1940 1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060
Year;
[ Model prediction •» Field data |
Fig. 5 Comparison of 1970s numerical model predictions and recent field data.
352 Sally Watson et al.
the three-dimensionality of the plume due to density effects, aquifer heterogeneity and
vertical hydraulic gradients, are considered the most likely causes of discrepancy.
A new model of chloride transport in the Eastry valley is being developed that will
incorporate these features, and include recent pore water chloride profiles and tracer
test results in its calibration and validation. The model will be used to assess the value
both of the tracer tests and of the various observations of plume development, in
making predictions.
We anticipate a particular challenge will be to make predictions on time scales
which are intermediate between those of tracer tests (weeks) and plume development
(decades).
Acknowledgements A Research grant from the Natural Environment Research
Council (no. GST/2/1811) under the DAEC/Envirorrmental Diagnostics Programme is
gratefully acknowledged. The matrix chloride profile illustrated in Fig. 3 resulted from
collaboration with the British Geological Survey, working on the EU FRACFLOW
project (www.ffacflow.dk), which is also gratefully acknowledged.