Chapter 5c

From GARDGuide
Revision as of 00:00, 29 May 2009 by Sbrionez (talk | contribs) (5.5.2 Geochemical Modeling)

5.0 Prediction

5.1 Introduction
5.2 Objectives of Prediction Program
5.3 The Acid Rock Drainage Prediction Approach
5.3.1 Acid Rock Drainage/Metal Leaching Characterization
5.3.2 Description of Phases
5.3.3 Water Quality Prediction
5.4 Prediction Tools
5.4.1 Introduction
5.4.2 Geological and Lithological Investigations
5.4.3 Hydrogeological/Hydrological Investigations
5.4.4 Geochemical Testing Methods
5.4.5 Data Management
5.4.6 Quality Assurance/Quality Control
5.4.7 Screening and Evaluation Criteria
5.4.8 Reporting
5.5 Modeling of Acid Rock Drainage, Neutral Mine Drainage, and Saline Drainage for Characterization and Remediation
5.5.1 Introduction
5.5.2 Geochemical Modeling
5.5.3 Hydrological Modeling
5.5.4 Hydrogeological Modeling
5.5.5 Gas Transport Modeling
5.5.6 Statistical Evaluation
5.6 Conclusions
5.7 References
List of Tables
List of Figures
First Page: Sections 5.1, 5.2, and 5.3
Second Page: Section 5.4 Prediction Tools
This is the Third Page: Sections 5.5, 5.6, and 5.7, Lists of Tables and Figures

Top of this page

5.5 Modeling of Acid Rock Drainage, Neutral Mine Drainage, and Saline Drainage for Characterization and Remediation

5.5.1 Introduction

Modeling and prediction have significant value as management tools and for gaining an understanding of the geochemical, physical, and biological systems at mine and process sites (Oreskes, 2000). The primary objective of mine and process water quality prediction is to evaluate the potential for geologic materials and mine and process wastes to generate acid and contaminants, and the potential to affect water resources. As an important corollary, the need for and nature of mitigation measures is determined through prediction.

In principle, modeling can be applied to all mine and process facilities, including mine portal effluent, subsurface waters (wells or underground workings), waste dumps, process tailings piles, surface waters, pit lakes, and open pits. The type of modeling used depends on both the objectives and the type of source or pathway. A wide variety of codes are available for these various environments, but the critical factors are the quality of their databases of these codes, the inherent assumptions, and, most importantly, the knowledge and experience of the modeler.

Figure 5-19 presents a generalized approach to the development, calibration, and use of a model. The modeling process starts with a strong conceptual model and the mathematical model can then be used to update the conceptual model as necessary. Calibration of the model is a critical part of the overall process.  

Figure 5-19: Generalised Model Process

The nature and sophistication of the prediction effort may vary depending on the desired outcome. A prediction exercise aimed at merely answering a “yes/no” question (for example: will the water quality criterion for arsenic be exceeded?) requires less up-front understanding of the system being evaluated, in which case the use of relatively simple modeling tools may suffice. In contrast, when a more quantitative answer is required (for example: what is the expected arsenic concentration), the complexity of the modeling effort may be quite significant, requiring both a detailed conceptualization of the system being modeled as well as use of advanced modeling codes. Care should therefore be exercised in selecting models so that they suit the need of the application and are compatible with the range and quality of the input data. The use of more sophisticated tools does not necessary equate to more accurate and precise modeling outcomes. According to Oreskes (2000) and Nordstrom (2004), the current computational abilities of codes and advanced computers far exceed the ability of hydrogeologists and geochemists to represent the physical, chemical, and biological properties of the system at hand or to verify the model results. In light of these considerations, the meaning of “accuracy” and “precision” in the context of mine and process water quality modeling must be reassessed on a case-by-case basis, and numeric analysis needs to be conducted to reflect the uncertainty inherent in predictive modeling. USEPA (2003) recommends the following should be submitted at a minimum to substantiate modeling used for regulatory purposes, regardless of the specific model/code being used:

  • Description of the model, its basis, and why it is appropriate for the particular use
  • Identification of all input parameters and assumptions, including discussion of parameter derivation (i.e., by measurement, calculation or assumption)
  • Discussion of uncertainty
  • Sensitivity analysis of important input parameters

The general understanding of geologic materials, mine and process wastes, and the hydrogeochemical factors that govern mine and process water quality continues to advance through the implementation of laboratory and field experiments. In particular those experiments that isolate one variable at a time to identify its effect on overall discharge water quality are of great value. Similarly, ongoing characterization and monitoring of mine and process facilities allows for development of improved scaling factors needed to extrapolate results from smaller scale tests to an operational level. Also, the tools required for geochemical, hydrological, and hydrogeological modeling already exist. Therefore, modeling can be a valuable component of mine water quality prediction and for evaluating management and mitigation options.

Additional detail on geochemical, hydrological, and hydrogeological modeling, including listings of commonly-used codes, can be found here [provide Wiki link].  

Top of this page

5.5.2 Geochemical Modeling

This Section 5.5.2 describes the conceptual, thermodynamic, and kinetic fundamentals of geochemical modeling and its application to prediction of mine water quality in support of mine site characterization and remediation. The emphasis in this section is on the basic processes that models attempt to represent with discussions of the usefulness and the limitations of modeling.

Three basic approaches have been used with geochemical data: forward geochemical modeling, inverse geochemical modeling, and geostatistical analyses.

Forward modeling is also known as simulating (i.e., potential reactions between rock and water are simulated from initial conditions of a known rock type and composition). Reactions are allowed to proceed in equilibrium or kinetic or combined modes. Changes in temperature and pressure can be invoked, changes in water flow rate can be assessed, and minerals can be allowed to precipitate as they reach equilibrium solubility or dissolve as they become undersaturated. Potential reactions can be simulated to see what the consequences are. This type of modeling is the least constrained. A great many assumptions are either invoked as input data or invoked as dictated by the program that may not apply to the specific system being simulated. This approach assumes the modeler has a significant amount of information on the ability of minerals to maintain equilibrium solubility or their rates of reaction.

Inverse modeling assumes a water flow path is known and that water samples have been analyzed along that flow path. Such data can then be converted into amounts of minerals dissolved or precipitated along that flow path. Several assumptions are still made regarding the choice of minerals and their relative proportions contributing to the water chemistry, but the calculations are constrained with actual data. Inverse modeling can also be done without any recourse to kinetic or thermodynamic data, in which case it represents a relatively simple mass balance calculation. When speciation and thermodynamic and kinetic properties are included for additional constraints, the possible reactions become quite limited and the modeling is much more meaningful.

Geostatistical modeling of geochemical data takes place as part of block model development, and is discussed in more detail in Chapter 4 and Section 5.4.5.

Modeling of any type does not lead to a unique solution but the possibilities are more limited with greater amounts of carefully collected field data. Martin et al. (2005) summarized the benefits and limitations of geochemical modeling as follows:


  • Provide insight into potential future conditions.
  • Determine which variables are most important in determining future conditions.
  • Assess the effects of alternative approaches to ARD management.
  • Assess potential effects of uncertain parameters
  • Establish objectives and test conditions for field and laboratory studies
  • Integrate available information.


  • Insufficient input data
  • Modeling can be challenging and results misinterpreted
  • Uncertain and variability of the results
  • Difference between modeled and actual field conditions.

Alpers and Nordstrom (1999) and Mayer et al. (2003) provide a review of geochemical models for use in mine water quality prediction.

Top of this page

5.5.3 Hydrological Modeling

Generally, a hydrological model is an analog of a natural or human-modified hydrological system. This generic definition encompasses models of surface-water and groundwater systems. Scientists and engineers commonly use the term hydrological model to refer to models of surface-water systems, and consider hydrogeological models for groundwater systems as a separate subject. This Section 5.4.3 follows the latter convention, describing hydrological models in the context of surface-water systems.

Hydrological models range from simple algebraic calculations to complex reactive-transport computer codes. Physical analogs, such as stream tables, can also be useful simulations of complex surface-water systems. Hydrological models can be used to predict the fate and transport of mine drainage through a surface-water system, providing important input to human-health or ecological risk assessments. Hydrological models can also be used to estimate the water-quality and water-quantity evolution of pit lakes over time. Hydrological models can be coupled with hydrogeological and geochemical models to incorporate the interaction between surface water and groundwater into the simulation and account for geochemical reactions.

Selection of an appropriate, quantitative hydrological model depends on the type of output that is required and, critically, on the conceptual model of the system being evaluated. A robust conceptual model will identify the important physical and geochemical characteristics of the field-scale system being evaluated. Based on that identification, an appropriate hydrological model can be selected that quantitatively represents those important processes. For complex systems or to assess a range of different types of processes, multiple hydrological models can be applied to predict the fate, transport, and potential impacts of mine discharges.

Top of this page

5.5.4 Hydrogeological Modeling

Hydrogeological models address water flow and contaminant transport below the land surface. As with hydrological models, approaches to hydrogeological simulations range from simple to complex. The universe of hydrogeological models includes physical and electrical analogs. With the advent of powerful personal computers and high-level programming languages, these approaches are rarely used in current practice.

Much literature exists regarding hydrogeological modeling, as do a number of computer programs. Zheng and Bennett (2002) provide an excellent introduction to the topic of contaminant-transport modeling. Maest and Kuipers (2005) provide a review of hydrogeological models more directly focused on ARD prediction.

Three following basic types of hydrogeological models are available, in order from simple to more complex:

1. Analytical models of flow and contaminant transport 2. Analytic element models 3. Numerical models

As a general rule, hydrogeological models should be as simple as possible while still representing the physical system with an adequate degree of precision and accuracy. More complex models should only be selected when project needs dictate, when simpler models are demonstrably not adequate, or when suitable data are available for model parameterization and calibration.

Hydrogeological models are useful tools for predicting the potential generation and resulting impacts of ARD. Models can be used to fill data gaps, either in space or in time. They can also be used to test alternative conceptual models in an iterative process designed to understand the complex natural or human-modified subsurface system.

Top of this page

5.5.5 Gas Transport Modeling

Gas transport, particularly the transport of oxygen into unsaturated waste-rock piles, can be an important process affecting the generation of ARD. Principal modes of oxygen transport include diffusion and advection. Wels et al. (2003) provide a comprehensive overview of the role of gas transport in ARD generation and methods that can be used to model gas transport.

Relatively few models have been developed specifically to address gas transport in the subsurface and the application to ARD-related problems. Modeling the complete set of physical and chemical processes operating within a waste-rock pile requires a multiphase code capable of simulating gas and water flow in the unsaturated zone, chemical interactions with the solid matrix, heat generation and transfer, and chemical mass transfer in the liquid and gas phases.

Top of this page

5.5.6 Statistical Evaluation

The use of statistics can be helpful in finding groupings and correlations among many parameters in a large data set. For instance, water quality results may be grouped into sets that may relate to hydrogeochemical processes. However, caution should always prevail. Statistics is a form of mathematics and supports and helps to understand science and engineering. Statistical results demonstrate correlations or the lack thereof but are nondeterministic. Parameters can correlate but not be deterministically related. Correlated parameters may indicate unknown relationships that were overlooked. Several types of multivariate correlative manipulations using regression techniques are in common use (Davis, 2003), including Principal Component Analysis (PCA), Cluster Analysis (CA), Probability Distributions (PD), and Factor Analysis (FA). These and other techniques often depend on assuming certain characteristics for the data set that are not necessarily correct (e.g., data follow a normal distribution, sufficient data are available to apply statistical tests, and levels of variance are comparable among parameters being correlated). Perhaps the best uses of statistical methods are for reasonable interpolation of spatial or temporal data and for identifying potentially causal parameters that had not previously been recognized.

Top of this page

5.6 Conclusions

Mine water quality prediction is an integral component of any study related to ARD. As described in this Chapter 5, a standard and structured methodology is used, particularly for new mine development. National regulatory frameworks and global guidelines frequently incorporate elements of this approach. Defensible ARD/ML and water quality predictions are being developed using state-of-the-art techniques by knowledgeable practitioners.

Top of this page

5.7 References

Alpers, C.N., and D.K. Nordstrom. (1999). Geochemical Modeling of Water-Rock Interactions in Mining Environments. In: The Environmental Geochemistry of Mineral Deposits, Part A: Processes, Techniques and Health Issues (Eds.: Plumlee, G.S. and M.J. Logsdon). Reviews in Economic Geology Vol 6A. Society of Economic Geologists, Inc.
AMIRA International. (2002). ARD Test Handbook. Project P387A Prediction & Kinetic Control of Acid Mine Drainage. Ian Wark Research Institute and Environmental Geochemistry International Pty Ltd.
British Columbia Acid Mine Drainage Task Force (BCAMDTF), (1989). Draft Acid Rock Drainage Technical Guide, Volume 1. August (1989).
Davis, J.C. (2003). Statistics and Data Analysis in Geology. Wiley International, New York.
Jambor, J.L. (2003). Mine-Waste Mineralogy and Mineralogical Perspectives of Acid-Base Accounting. In: Environmental Aspects of Mine Wastes (Eds.: Jambor, J.L., D.W. Blowes, and A.I.M. Ritchie). Short Course Series Volume 31. Mineralogical Association of Canada.
Lapakko, K.A. (2003). Developments in Humidity-Cell tests and Their Application. In: Environmental Aspects of Mine Wastes (Eds.: Jambor, J.L., D.W., Blowes, and A.I.M. Ritchie). Short Course Series Volume 31. Mineralogical Association of Canada.
Maest, A.S., and J.R. Kuipers. (2005). Predicting Water Quality at Hardrock Mines: Methods and Models, Uncertainties and State-of-the-Art. Washington DC: Earthworks.
Martin, J. G., Wiatzka, J., Scharer and B. Halbert. 2005. Case Studies that Illustrate the Benefits, Limitations and Information Requirements of Geochemical Modelling. In: Proceedings of the 12th Annual British Columbia – MEND ARD/ML Workshop. November 30 and December 1, 2005. Vancouver, BC.
Mayer, U., Blowes D.W., and E.O. Frind. (2003). Advances in Reactive-Transport Modeling of Contaminant Release and Attenuation from Mine-Waste Deposits. In: Environmental Aspects of Mine Wastes (Eds.: Jambor, J.L., Blowes, D.W. and Ritchie A.I.M.). Short Course Series Volume 31. Mineralogical Association of Canada.
Mills, C. (1999). Acid Rock Drainage at Environment.
Morin, K.A., and N.M. Hutt. (1997). Environmental Geochemistry of Minesite Drainage: Practical Theory and Case Studies. Vancouver: MDAG Publishing.
Nordstrom, D.K. (2004). Modeling Low-Temperature Geochemical Processes. In: Treatise on Geochemistry. (Eds.: Holland, H.D. and K.K. Turekian), Volume 5, Surface and Ground Water, Weathering and Soils. Elsevier Ltd.
Oreskes, N. (2000). Why predict? Historical perspectives on prediction in earth science. In: D. Sarewitz, R.A. Pielke, Jr., and Radford Byerly, Jr., Prediction: Science, decision making, and the future of nature, Island Press, Washington, D.C., 25-40.
Price, W.A. (1997). Draft Guidelines and Recommended Methods for the Prediction of Metal Leaching and Acid Rock Drainage at Mine sites in British Columbia. BC Ministry of Employment and Investment.
Price, W.A. (2009). In prep.
Shaw, S. (2005). Case Studies and Subsequent Guidelines for the Use of the Static NAG Procedure. In: Proceedings of the 12th Annual British Columbia – MEND ARD/ML Workshop. November 30 and December 1, (2005). Vancouver, BC.
Thompson, A., B. Price, K. Dunne and J. Jambor. (2005). Guidelines for the Determination of Mineralogy and Mineralogical Properties. In: Proceedings of the 12th Annual British Columbia – MEND ARD/ML Workshop. November 30 and December 1, (2005). Vancouver, BC.
United States Environmental Protection Agency (USEPA). (2003). EPA and Hard Rock Mining: A Source Book for Industry in the Northwest and Alaska. EPA 910-R-99-016.
Wels, C. Lefebre, R. and A.M. Robertson. (2003). An Overview of Prediction and Control of Air Flow in Acid-Generating Waste Rock Dumps. In: Proceedings of the 6th International Conference on Acid Rock Drainage (ICARD), Cairns.
White III, W.W., K.A. Lapakko, and R.L. Cox. (1999). Static-Test Methods Most Commonly Used to Predict Acid-Mine Drainage: Practical Guidelines for Use and Interpretation. In: The Environmental Geochemistry of Mineral Deposits, Part A: Processes, Techniques and Health Issues (Eds.: Plumlee, G.S. and M.J. Logsdon). Reviews in Economic Geology Vol 6A. Society of Economic Geologists, Inc.
Wolkersdorfer, Ch. (2008). Water Management at Abandoned Flooded Underground Mines. Springer, Heidelberg.
Younger, P.L. and D.J. Sapsford. (2006). Acid Drainage Prevention Guidelines for Scottish Opencast Coal Mining: the Primacy of the Conceptual Model. In: Proceedings of the 7th International Conference on Acid Rock Drainage (ICARD), St. Louis.
Zheng, and G.D. Bennett. (2002). Applied Contaminant Transport Modeling. Wiley-Interscience, New York.

Top of this page

List of Tables

Table 5-1: Methods for Geochemical Characterization
Table 5-2: Geologists Observations and Logging of Core for ARD Analysis
Table 5-3: Example Chemistry Table
Table 5-4: Example ABA

Top of this page

List of Figures

Figure 5-1: Generic Prediction Program Flowchart
Figure 5-2: Generalized Flowchart for the ARD Prediction Approach at Mine Sites (Maest and Kuipers, 2005)
Figure 5-3: Conceptual Model Showing Metal and Acid Source Regions at Iron Mountain and Downstream Transport Pathways to the Sacramento River
Figure 5-4: Flowchart for Metal and Acid Source Regions at Iron Mountain and Downstream Transport Pathways to the Sacramento River
Figure 5-5: Schematic Illustration of Geochemical Characterization Program (modified from Maest and Kuipers, 2005)
Figure 5-6: Example Plot of NP from Total Carbon vs. NP from Modified Sobek
Figure 5-7: Example Plot of Total Sulphur vs. Sulphide Sulphur
Figure 5-8: Example Plot of ABA vs. NAG Results
Figure 5-9: Example Plot of Metal Loadings vs. Sulphate Content
Figure 5-10: Humidity Cells
Figure 5-11: Example Plot of HCT Results
Figure 5-12: Wall Washing
Figure 5-13: Test Cells for Waste Rock
Figure 5-14: Test Plot for Paste Tailings – Somincor Neves Corvo Mine, Portugal
Figure 5-15: Example of Block Model Use: ARD Potential of Pit Highwall Above Final Pit lake
Figure 5-16: Example of Block Model Use: ARD Potential of Pit Wall after Cessation of Mining
Figure 5-17: Decision Tree for the Determination of Acid Generation Potential (AMIRA, 2002)
Figure 5-18: Example Plot of ABA Results and ARD Criteria
Figure 5-19: Generalised Model Process

Previous Page (Page 2) of Chapter 5

Beginning (Page 1) of Chapter 5

Top of this page