Risk Assessment for Closure of Acid Generating Tailings Areas

Introduction
For this study, detailed characterization of the tailings areas was available from conventional ABA studies, various kinetic tests and historical data on acid generating tailings from other regional mines and monitoring data for the basin. The data showed that the tailings had minimal neutralization potential (NP), an acid potential (AP) of >100kg/t and minimal lag time between tailings exposure and the onset of oxidation. Closure strategies that were considered included various dry covers, elevated water tables, collection and treatment, and flooding. Conventional environmental pathways techniques were applied to assess the options and concluded that flooding was the preferred option at the three sites considered. The weakness in the conventional assessment was that it considered only routine operations and assumed the facility would perform as designed. This left both the public and regulators to ask what were the risks of failure over the long term, were these risks acceptable, and what were the long-term cost implications of failure and long-term maintenance of the sites?

In order to address these issues a risk assessment approach was taken. The risk assessment studies addressed a variety of factors including: failure modes and consequences from non-routine events and design/construction flaws; environmental effects that could occur as a result of failure; the frequency of failure; costs to clean-up and repair the site in the event of a failure; costs to care and maintain the site in perpetuity; potential consequences of failure and the need for design change; and the consequences of failure to maintain the site.

The primary risks to the facility were the failure of a retention structure which could spill acid tailings into the environment, and drought which could cause the surface to dewater and tailings to generate acid and dust. The basic framework for the model was the @RISK code (commercially available). The model involves a probabilistic framework that allows for inputs to be entered as distribution functions (e.g. lognormal, triangular, uniform). The model included four subroutines, the Maintenance model, Earthquake Response model, Flood Response model, and Drought model. The subroutines used actuarial statistics on earthquake frequency, meteorological data, dam failure frequency etc. for periods of 10,000 years. Where data were not available, expert opinion was used to fill in these gaps. The consequences of failure were established for each structure based upon type of failure, depth of breach, volume of water spilled etc. The outputs of these subroutines fed into the Integration model which was run for various time periods (200 and 1000 years in this case). The integration model was used to:

• Predict the number and severity of occurrences (failures);
• Predict the consequences (tons of tailings lost, months and areas of beach exposure);
• Determines the cost to repair the site and remediate downstream damage for each event;
• Calculate total and discounted costs for maintenance and site repair; and
• Provide summary statistics on the cumulative probability distributions for cost and failure consequences.

The worst case outputs were also subjected to detailed environmental assessment to address the potential impacts to human health and biota (health risk and eco-risk). Figure 1 illustrates the conceptual approach to the risk assessment from the problem formulation stage through a feedback loop which triggers consideration of modifications to the system in the event that some perceived identified attribute of risk fails to meet the previously established criteria. Figure 2 illustrates conceptually the integration of the facility risk model described above and the environmental consequence model used to assess the environmental and health consequences arising from the accidental release of tailings water and/or solids. A modified version of the Monte Carlo Assessment tool (UTAP) was used to assess the potential effects on future water quality, the environment and people. In the event that failure consequences were deemed unacceptable, the design of the facility was modified to reduce either the frequency or magnitude of failure.

Results
The probabilistic risk assessment was an extremely useful tool as it was able to provide key data on the long-term performance of the flooded basins containing reactive tailings which will require flooding in perpetuity. The model confirmed:

• Care and maintenance was essential to reduce the failure rate and consequence. Costs for care and maintenance were modest with a net present value in the range of $2 million for each of the three sites assessed.
• Some failures were more likely to occur and may result in unacceptable consequences. These results were used to modify designs.
• It was not necessary to implement all measures to reduce risk as it was demonstrated that the costs for many of these measures far exceeded benefits.
• The model results provided both regulators and the public with a level of comfort as the risks for all "what if" scenarios were quantified (both what can happen and what are the consequences).
• The model was the key tool used to assess long term costs and financial assurance needs.

Environmental Risks Associated with Historic Uses of Slag

Some 50 million tonnes have been used productively as road base, rail ballast or fill for other construction purposes. Continuing to use slag for construction fill would effectively increase the longevity of the slag dump.

The objective of this study was to develop a preliminary generic assessment of the environmental risks associated with historic uses of slag. The primary purpose was to understand the potential human health and environmental risks (liabilities) that might arise from use of slag as construction fill, particularly as might result from acid generation and metal mobilization. A secondary objective was to provide input with respect to management of future slag production. The risk assessment was performed following generally accepted practice. In brief, the major elements of the human health risk assessment were as follows.

An initial problem formulation/hazard assessment phased was used for identification and characterization of slag constituents. An initial screening on the basis of environmental mobility, toxicity and regulatory limits was performed to identify the species that were to be considered in the screening risk assessment. An extensive database was reviewed on the characteristics of slag, including chemical compositions, mineralogy, chemical reactivity (in terms of potential acid generation) and leachability. Ranges of characteristics for slag, and leachate from slag, were developed from the data and used as input to the risk assessment.

The next phase of the project focused on exposure assessment using environmental models to assess how the selected slag constituents move through the environment and result in exposure to people. The assessment considered potential exposures based on actual past usage of slag as construction fill, and available studies regarding the environmental implications of slag storage, and usage in construction applications. These studies were reviewed, and where appropriate, were used to examine, or provide a context for, the assumptions used in the risk assessment. Data were available on surface water, groundwater, soil, vegetation and sediment in the vicinity of sites where slag was used as construction fill.

Next a dose-response assessment, or toxicity assessment, using published consensus scientific literature was carried out to develop toxicity profiles for the various slag constituents. Lastly a risk characterization was carried out that integrated the results of the exposure and dose response assessments to describe the nature and (estimated) magnitudes of potential risks to people.

For situations where slag is used as construction fill, slag leachate may add to background levels in nearby streams. In this project, potential ecological risks to fish and other aquatic species were calculated using a similar approach to that used for the human health risk assessment.

Sources of uncertainty exist at each stage of the risk assessment. For the purposes of the screening risk assessment, intentionally conservative assumptions were made to ensure that the risks calculated to people or to the environment arising from use of slag as construction fill are very likely to be overestimated (i.e. conservative). Some of the uncertainties arise from variabilities in the leachability of slag, in the calculation of environmental concentrations in air or water; from the assessment of toxicity or from the characteristics assigned to the receptors (exposed individuals) themselves.

Based on the conservative screening calculations described in the report, no human health effects arising from use of slag as aggregate would be expected. The results of the risk assessment were input to the company's decision making and strategic planning process. The work provided guidance for dealing with these materials during the short term (operational) period, as well as providing the company a degree of confidence with respect to the long term behaviour of the material and any potential liabilities that might be associated with the re-use of the materials.


A Corporate Liability Risk Analysis of Tailings Decommissioning

The assessment of the potential financial liability associated with the decommissioning of the Atlas uranium tailings pile, located on the edge of the Colorado River, provides a classic illustration of the need to consider potential changes in risk acceptance, linked in this case with changing demographics, over time in the decision making process. This case illustrates the dramatic effects that may arise as a result of evolving legislation, changing social and demographic conditions and the variable nature of acceptable risk.

In this particular instance, the owner had in place an approved reclamation plan for its 10.5 million ton tailings stack based on on-site reclamation in accordance with regulatory requirements. A technical update (slope stability) in the regulatory guidance required minor revision and amendment in this aspect of the approved plan. On issuance of the notice of amendment acceptance by the regulator, public and political opponents of on-site plan used the comment period to raise new issues. These opponents successfully solicited enough political support to have the regulator take the unprecedented step of initiating a new EIS of the previously approved closure plan and an off-site alternative.

During this period, Atlas was assessing restructuring options including various merger options. In both cases, but particularly the latter, a key management decision element centered on the viability of the existing closure plan, that is, is it appropriately protective, and what is the likelihood that it would ultimately be found acceptable or would the off-site option be mandated. To address these questions, Atlas initiated and carried out a comprehensive corporate risk assessment program (see Figure 3) that considered the following factors in the assessment of the potential environmental liability associated with the site:

• Political issues;
• Local state and federal regulations;
• Regional and local concerns;
• Publicly perceived risks;
• Actual risks;
• Environmental impact potential;
• Engineering considerations and cost;
• Financial options; and
• Fatal flaws

The risk assessment work was carried out in an interactive manner by a multidisciplinary team that included owner representatives, national and regional legal counsel, public relations, engineering, environmental and risk managers. The work included analysis of existing and alternative reclamation scenarios; review of existing and uncertain engineering issues, designs, and concepts; assessment of industry practices and precedents; review of health and safety records; evaluation of the existing cost basis; assessment of potential health risks to workers and the public under varying scenarios; and consideration of legal, regulatory and political drivers that might affect the outcome of the decision-making process. A detailed probabilistic (Monte Carlo) model was developed that explicitly included the uncertainties associated with the decommissioning concepts, the engineering requirements and outstanding engineering issues, long-term performance considerations including various failure modes, and consequence scenarios.

The risk assessment supported the validity of the original plan. It illustrated that while the health risks associated with either option were small, the health risk for the off-site option were six times higher than those of the on-site option, while the costs of the off-site option were about ten times that of the on-site option. The corporate risk assessment also found that the area of most uncertainty with respect to the selection of a final plan lay in the political arena.