| Open-cut coal-mining operations typically start with the excavation
of a box cut to expose the coal seam. Mining then proceeds down
dip following the coal with overburden (spoil) being progressively
dumped in spoil piles onto the floor of the previously mined areas
(in-pit dumping). As the coal seam dips deeper, it ultimately becomes
no longer economical to mine any further. A final void is then left
at the limit of mining. The spoil piles are then reshaped and rehabilitated
to form the catchments which direct storm runoff into the final
void.
The characteristics of open-cut spoil heaps result in a distinctive
hydrology. These characteristics can be classified into two main
groups:
- Spoil materials. Spoil materials that contain
fresh sedimentary or basalt sequences will have a different hydrology
and quality compared to natural surface materials (soils).
- Spoil structure. The spoil structure is a
consequence of the methods used for placing and reshaping spoil
and topsoil.
Several major research studies were carried out in collaborative
programs funded by the coal mining industry to develop objective,
reliable and practical methods for simulating the hydrological behaviour
of spoil-void systems (Figure 1). Among them was the 3-year-long
ACARP (Australian Coal Association Research Projects) Project C7007
on the hydrology and water quality of final spoil-void systems.
Our role in this project included:
- Formulation of a data monitoring system
- Collection of field data
- Evaluation of a range of instrument types
- Review of current research
- Mathematical modelling of hydrologic processes
- Preparation of a hydrologic model parameter database
- Preparation of site specific hydrologic models
- Presentation of study findings to the mining industry and regulators.
The findings of this research are presented in this article along
with comments on aspects of the project that were done well and
those that could have been done better.
Figure 1: Spoil alongside void at Moura in the Bowen Basin,
central Queensland |
Overview of ACARP Project C7007
The key objectives of ACARP Project C7007 were to:
- Establish reliable procedures for predicting the hydrology
and water quality of final spoil-void systems based on computer
modelling and field monitoring
- Validate those procedures scientifically by refining an existing
prototype computer model and performing model calibration on a
network of eight monitored and sampled test sites1 that represented
a wide range of spoil and catchment types. The catchments varied
in size from 15 hectares to 250 hectares (38 acres to 625 acres).
- Use those procedures to predict the probability of discharge
from voids over long timeframes.
Monitoring of the eight sites included taking measurements of
rainfall and other meteorological data spoil catchment runoff, piezometric
heads at the base of spoil piles, and void and pond water levels.
We used mathematical modelling to acquire an understanding of some
of the hydrological processes and to develop simplified relationships
suitable for practical modelling. A prototype daily water-balance
model of spoil-void systems called the Spoil Hydrology Lumped Parameter
Model (SHLPM) was developed as an outcome of a special research
project funded by BHP Coal. BHP Coal has agreed to its licence-free
release.
Figure 2: Typical cross-section of void and spoil |
Figure 3: Cut-throat flume, South Blackwater in the Bowen
Basin, central Queensland |
The SHLPM was developed to assess the long-term hydrological responses
of spoil-void systems for open-cut coal mines (Figure 2). It was
designed for situations where the general objective is containment
within the void of all water reporting to the void.
SHLPM uses daily rainfalls to predict runoff and deep recharge from
surface subareas and to route them to the appropriate destinations,
including large surface ponds, the void, the natural drainage system
and the saturated region at the spoil base. Evaporation losses from
the spoil, ponds and the void are taken into account. Other exchanges,
such as flows between the void and the spoil or between the void
and the natural groundwater system, are included in the modelling.
In carrying out the ACARP project, we effectively validated SHLPM
as an appropriate tool for hydrologic modelling of final spoil-void
systems.
SHLPM Parameter Calibration
Simple spreadsheet models were developed to obtain estimates for
the catchment-specific parameters required in SHLPM. This was done
for each of the monitoring site catchments using the information
gathered during the project’s field data collection program.
While the general concept adopted for formulating the spreadsheet
models was essentially the same for all situations, minor variations
existed between the model used for the ponds and voids and the model
used for the flumes (Figure 3). The difference between models was
related to output in terms of whether:
- Runoff volumes were required as the only model output (flume
model)
- Pond/void water-level variations were also required as model
output (pond and void model).
The basic principles behind the spreadsheet models involved determining
a daily water balance for the catchment considering the rainfall
inputs, evaporation losses, soil moisture storage and amount of
runoff. We assumed initially that each monitored catchment was homogeneous
in nature and could be effectively modelled with a single soil-storage-capacity
parameter. This model assumed that the upper storage capacity of
the catchment’s soil was equivalent to the depth of a single
bucket. When rainfall inputs exceeded evaporation losses, the bucket
filled. When the bucket capacity was exceeded, catchment runoff
was deemed to commence.
The results we obtained indicated that the single-bucket approach
could not accurately reproduce the hydrologic responses that were
observed during both small runoff events and large runoff events.
It became apparent that a more refined approach using multiple buckets
to model the catchment could achieve a much better all-round calibration.
A spreadsheet configuration allowing the definition of up to three
upper storage zones (three buckets) was selected on the basis that
a very good match between observed and modelled results was obtained.
The relative proportions of the total catchment area assigned to
each bucket could be altered during the calibration if necessary
to achieve a better overall fit.
The calibration procedure involved selecting model parameters for
the smallest size bucket, first by matching the water-level rises
for the smaller events. Model parameters were then selected for
the middle-size bucket by matching the water-level rises for the
moderate events and for the largest bucket by matching the water-level
rises for the large events. The results obtained indicated that
a much better calibration could be achieved when compared to the
single-bucket model and that matches to both small and large hydrologic
responses were possible with only a small additional effort being
required in the calibration.
We prepared a database of calibrated SHLPM parameters that described
parameter variation due to different mine site location, methods
of spoil placement, level of disturbance, degree of vegetation and
rehabilitation status. Comparison of the monitoring sites showed
that the calibration parameters were not solely dependent on ground
type. Some variation will be related to the quality of the data
collected at each site, but it must be concluded that other catchment
characteristics, such as regional geology, soil type and catchment
slope, may also have a significant influence on the calibration
parameters.
Preparation of Site SHLPMs
For those spoil-void catchments that had reliable observed water-level
data and were previously modelled using spreadsheet hydrologic models,
a SHLPM of each catchment was constructed using parameters sourced
directly from the modelling. Predicted water-level responses from
the SHLPM were then compared to the observed water-level responses
to verify that the calibration parameters derived from the spreadsheet
modelling remained valid. In all such cases, the SHLPM achieved
an acceptable match to the observed water levelswith only minor
variations being required to the parameters derived from the spreadsheet
modelling as part of the final calibration process.
Calibrations of the SHLPM model to observed water-level data at
five mine sites indicated that the hydrologic concepts included
in the SHLPM are valid. The modelled flow transfer between the void
and the spoil in the SHLPM is in accordance with the observed hydrologic
response in the voids investigated during this project. The observed
immediate response in the void after rainfall, followed by a delayed
response and then a gradual decline that flattened over time, is
consistent with the SHLPM modelling concepts for flow exchange between
the void and the spoil. It can be concluded that the SHLPM can be
used to provide a reliable means of simulating and matching the
observed catchment behaviour for spoil-void catchments.
Assessment of the long-term hydrologic response in each of the modelled
voids was undertaken by running long-term rainfall records through
the SHLPM of each site. Results from these simulations indicated
that the time taken to fill a void is variable and depends on the
size of the void and the size of the catchment. Modelling of individual
spoil-void systems is required to determine the time to fill a specific
void. None of the voids that were modelled in this project were
overtopped during the long-term simulations.
Water Quality
The influences and history of final void water chemistry can be
explained by studying ratios of the major ions. The water chemistry
is dominated by salinity. Some water samples have evidence of acid
generation followed by neutralisation, with the end result being
water of high salinity and indicatively high sulfate.
We prepared a simple mass balance model as part of this project
that demonstrates the potential impacts of salt concentrations in
surface runoff and saturated base flows on long-term water quality
in a void. The model assumes that all losses from the void are due
to evaporation, the system is closed (all water is collected by
the void, no matter what the source), and the water chemistry is
derived by continuous increase in the salt content due to removal
of pure water (evaporation).
Evidence shows that continuous, steady declines in water quality
do not occur at mine sites. The model is therefore realistic, and
could be used if the system meets these simplifying assumptions.
Mass balance modelling requires extensive spoil characterisation
and leach testing. It parallels the methods used in other countries
to predict acid mine drainage. This approach is applicable to coal
mines in the Bowen Basin (Collinsville would be the exception) and
the Hunter Valley.
The mass balance model does not take different flow volumes into
account. Thus, a water balance model, of which SHLPM is an example,
may be used to account for solute inputs and outputs via the different
flows (buckets) in the system if combined with water chemistry.
The disadvantages of this model are that it is parameter-intensive,
and it requires some parameters that would be difficult to measure
and others that are unlikely to be available while a mine is in
a planning or start-up phase.
Equilibrium models are similar to the mass balance model, with the
information gained by leach testing being derived by chemical thermodynamic
constants and equilibrium chemistry. The approach has promise for
predicting final water quality, particularly at data-poor coal mine
sites where, during the planning stage, geological and resource
estimation data might be all that is available.
These models have been tested and verified and are widely used,
including in the coal industry, for prediction of water chemistry.
Van Berk and Wisotzky (1994) have used the equilibrium model, PHREEQE,
to predict sulfide oxidation potential from brown coal overburden
in the Rhineland.
Lessons Learned
Understanding of the hydrologic processes is a vital stage in the
prediction of long-term behaviour of final void-spoil systems. This
research has determined that accurate long-term predictions of void
behaviour can be made using data gathered from relatively short-term
monitoring.
It is recommended that every mine seeking final void closure should
have a SHLPM of the void-spoil catchment prepared to determine the
long-term response of that void. A field monitoring program should
be implemented to gather site-specific data for the catchment and
appropriate SHLPM parameters can be derived by calibrating the SHLPM
to the observed data. Alternatively, appropriate SHLPM parameters
can be sourced from the parameter database formulated during this
project. It is recognised, however, that SHLPM parameters derived
from observed field data at specific sites will provide more accurate
results than those obtained using generic parameters from the database.
A major component of ACARP Project C7007 was the collection of field
data necessary for the validation of the modelling. As an overall
observation, data management during the project was poorly executed
and contributed to the limited amount of useful data available for
model validation. Experience gained during this study has led to
the preparation of guidelines to better manage data collection.
A significant finding was that field data should be checked, processed
and evaluated as soon as possible after it is collected to ensure
that any data errors and inconsistencies are identified at an early
stage, providing a more reliable, continuous, long-term data record.
The study found that the reliability of water level monitoring equipment
was a major issue. Overall, experience gained in this study indicates
that more reliable measurements can be obtained for ponds and voids
when the sensor remains fully submerged throughout the entire period
of data collection. In addition, more reliable data can be obtained
from flumes when the flume design minimises sediment accumulation
and when the water level measurements are obtained without the sensor
being in contact with the water and sediment.
Further, it is recommended that SHLPM parameters determined for
individual mine sites using site-specific field data be incorporated
into the parameter database that has been initiated in this research
project and be made freely available to the industry.
Benefit to the Industry
The industry benefit of this research is one of licence to operate.
Community acceptance of allowing final voids to remain unfilled
at the completion of mining will be greatly influenced by the overtopping
potential and quality of accumulated water in the void. This project
has advanced the selection of model input parameters, improved the
validation of models and, therefore, improved the reliability of
model predictions. |
