June 2015

A modular approach to modelling water flow and quality

open pit copper mine in Atacama desert, Chile
  • By Brent Usher, Manager, Geosciences, Senior Hydrochemist, Klohn Crippen Berger

An alternative method for constructing models that can be applied to a variety of situations – including modelling water flow

Resource project planners are increasingly called upon to assess the long-term environmental and economic impacts of feasibility studies and for seeking regulatory approval. These assessments require the use of models to inform predictions of water flow and water quality and to identify potential changes to the systems over time.

The traditional approach has been to develop complex numerical models specifically for each project. These models are generally time-consuming to construct and require large data sets to provide meaningful predictions; in many cases the predictions are not supported by the available data. The overall effect of these issues is the potential for time and cost problems that may lead to project extensions and over-budget project completion.

The modular approach to model construction

An alternative approach is to use modules to construct the required water model. These models can be used fruitfully in several situations, such as where the initial data is scarce, where the system’s complexity requires several simplifying assumptions, where more rapid results are required to advance a project or where off-the-shelf numerical water prediction models are not readily available.

This approach takes a number of pre-made standard modules and adds them together to provide theoverall model, much in the same way as a modern assembly line. Two examples can be provided to illustrate the approach: one related to prediction of water volumes associated with coal seam gas (CSG) abstraction across two basins, and the other for detailed geochemical assessment of a mine waste facility.

Conceptual model

The main elements required to simulate the system are first decided upon in a conceptual model (eg Figure 1 for a mine tailings storage facility (TSF)). For a TSF these may include the tailings pond, the source(s) of tailings, the climate etc. The elements chosen will be specific to the desired simulation.

diagram showing the possible conceptual elements within a tailings storage facility

Including elements within the model

Elements identified in the conceptual model are added together to provide the framework of the prediction model. These might include, for example, an element containing the calculations required for the water balance of the system, or an element containing the geochemical behaviour of the system. The tailings pond, identified above in the conceptual model, would form part of the water balance module in this example. The model modules can then be populated with data that are specific to the project, such as observed background water quality, measured volumes and flows and waste rock types etc. Well-constructed modules have the flexibility to allow a range of different data types to be incorporated. In the case of a TSF, these might include chemistry from static and kinetic testing, water quality monitoring and mine processing data. A well-designed model can include ‘user defined’ elements that are set up to be populated by project-specific data.

Module modification

The basic modular model allows more time to be spent on customisation to the needs of the project. Increasing levels of sophistication and higher levels of complexity in model calculations and concepts can be achieved because the basic framework of the model is known to operate correctly.

Examples of the modular approach

Tailings storage facilities and waste rock dumps

The modular approach to model building has been adopted successfully at Klohn Crippen Berger and has been used on a number of small- to large-scale projects. The TSF system described in the conceptual design stage of the model above is shown as an example.

Modelling TSFs can be a complex and time-consuming process that may be streamlined significantly by taking a modular approach. Figure 2 shows a TSF model build, using the GoldSim water balance / water quality platform. The container and element interface of GoldSim lends itself particularly well to modular modelling. In this example, the main elements of the TSF (such as the pond, phreatic zone, etc) are modules that contain the calculations required to simulate each element: the TSF pond for example, contains calculations to describe the inflows and outflows of water, the geochemical reaction products from the TSF beach and the behaviour of solute within the pond. Each of these modules has been subject to prior testing and calculation checks, so the generic calculations are familiar and consistent with current knowledge and quality control procedures.

graphs showing sulphate and chloride concentrations simulated in the GoldSim TSF model

Modular simulation of a TSF generally comprises a set of global modules that are populated with data provided by the TSF operators. These include the TSF physical parameters, waste rock (WR) characteristics, ambient climatic effects, geochemical data and so on. The modular approach is designed to read this information within the framework of the generic TSF model. Once the basic TSF model has been constructed, it can then be customised for conditions specific to the particular TSF, such as climatic conditions (eg high rainfall in tropical settings) or the physical layout of the facility.

Model outputs derived from this modular approach demonstrate the close calibration of the simulated output with observable data. Figure 2 shows the modelled TSF water quality (in terms of sulphate and chloride concentrations over time) compared to observable water quality sampled over a period of five years at a TSF. In the two diagrams, the concentrations are normalised to the observed data and expressed as percentages of the real water quality. Figure 2 shows that the simulated water quality is generally within 20  per cent of the observed water quality, an uncertainty that is comparable to the uncertainty that would reasonably be expected due to the analytical error of the real data.

Predicting the water associated with CSG extraction in the Surat and Bowen Basins

As a second example, a modular approach was used to construct a water production tool (WPT) to provide a rapid and flexible evaluation of the associated water production for different a variety of industry situations.

How the tool estimates CSG water production

A simple description of how the WPT estimates CSG water production is as follows:

  • Individual well pumping responses are assessed using the Theis equation, a non-equilibrium groundwater flow equation that accounts for the effect of pumping time on well yield in confined aquifers.
  • Pumping effects are then projected spatially to assess the effect of each pumping well on nearby wells. Due to interference effects between wells, less pumping is needed at an individual well to achieve the same target drawdown performance.
  • This water production is then upscaled to the basin-wide scale of the currently projected CSG industry. The geographic extent of the WPT is bounded by the lateral extent of the Queensland portion of the Surat Basin, and in the north includes southern areas of the Bowen Basin.
  • Two modifications are then applied to address:
    • changes to water production rates as gas flow begins to dominate over water as production wells mature (dual-phase effects)\
    • near-edge effects at the margins of the Surat Basin, where production zones are very shallow and local geological and hydrogeological conditions can have a greater influence on water production.

It is important to note that the WPT is not a groundwater flow model, and does not in any way predict effects other than the estimated water production amounts from CSG activities. The extent of the WPT is provided in Figure 3.

Figures 3 and 4 showing spatial extent of the water production tool modular model and a comparison of a standard groundwater model and water production tool

How does this differ from a groundwater model?

This is not a groundwater model. The tool:

  • Is constructed so that input is relatively simple and simulations can be run relatively quickly
  • You do not require advanced learning and experience in hydrogeology and modelling to run it – thus it is intended to become a modifiable tool for a wider range of planning uses.
  • With this practicality comes some limitations:
    • it does not predict groundwater levels or groundwater flows
    • it does not predict ‘impacts’ other than well-to-well interference on pumping rates
    • it does not predict the production of gas.

A graphical comparison of an industry standard groundwater model versus a modular-based water production tool is provided in Figure 4.

Like any other model, the WPT was verified, validated and calibrated by comparing historical (actual) values of water production for site-specific models with WPT outputs for periods of time consistent with the reported data (validation). Comparisons between this simpler, modular approach and several other CSG water estimates indicated that the results for water production were comparable, but scenario evaluation, ease of use and data requirements were far less onerous than for a basin-wide groundwater model, while spatial discretisation and robustness was better than could be achieved with more generic industry-forecasting methods.

Conclusions

Taking a modular approach to constructing predictive water flow and water quality models has the potentialto provide cost and time benefits that have obvious advantages to resource planners. Model element modules also have the advantage that particular elements can be standardised and rigorously assessed, which aids understanding and quality control procedures and allows regulatory bodies to evaluate simulations with a clear approach.

Here we have shown two greatly contrasting examples that demonstrate the use of the modular approach. This approach should be strongly considered as part of the modelling toolbox for predicting project flows and quality. 

References

Klohn Crippen Berger Ltd (2012). Forecasting coal seam gas water production in Queensland’s Surat and southern Bowen basins. Prepared for the Department of Natural Resources and Mines as part of the Healthy HeadWaters Coal Seam Gas Water Feasibility Study. www.dnrm.qld.gov.au/__data/assets/pdf_file/0020/106139/csg-water-forecasting-tech-report.pdf . September 2012.

Strand R, Usher B, Strachotta C, Jackson J, 2010. Integrated Water Balance and Water Quality
Modelling for Mine Closure Planning at Antamina. Mine Closure 2010: 5th International Conference on Mine Closure, Chile.

Usher B, Strand R, Strachotta C, Jackson J, 2010. ‘Linking fundamental geochemistry and empirical observations for water quality predictions using GoldSim’ International Mine Water Association Conference, IMWA 2010 – Mine Water and Innovative Thinking. Sydney Nova Scotia. Canada.

Share This Article