Russell Mills PhD,Rebecca Freeman, Malcolm Barker
The global mining industry lives with the risk of catastrophic events such as water storage or tailings dam failures as part of its daily operations, and has developed a number of approaches to enable mine management to understand the nature of the risks and the ways in which they are being managed. One such approach involves the use of bowties for the understanding of the hazards and risks. Building from bowties, the second approach involves the selection and management of controls critical to the prevention or mitigation of the catastrophic event. The Australian mining industry is a world leader in this regard and the purpose of this paper is to illustrate how bowties are constructed, how risks can be semi-quantitatively estimated, how critical controls are selected and managed, and how, if all this is done well, risks can be demonstrated to be as low as reasonably practicable (ALARP).
This paper sets out key themes and presents an example for a tailings dam failure to illustrate the role of bowties and critical controls in management of catastrophic events. It will also highlight the role of bowties in the anticipated introduction of a Safety Case approach to dam risk management. Bowties provide a useful tool for the transfer of risk management knowledge from the designer, to allow dam owner / operators to better understand their risks and to recognise the link between design and operational controls and how they are used to manage those risks to ALARP.
— OR —
This paper reviews methods used to estimate the MCE in Australia and New Zealand. In the ICOLD (2016), NZSOLD (2015) and proposed ANCOLD (2016) guidelines, the deterministic approach is applicable only to fault sources, whereas the probabilistic approach is applicable to both fault sources and distributed earthquake sources. Although ICOLD (2016) states that the use of a deterministic approach to develop the SEE “may be more appropriate in locations with relatively frequent earthquakes that occur on well- identified sources, for example near plate boundaries,” the proposed ANCOLD (2016) guidelines retain the use of the deterministic approach for critical active faults which show evidence of movements in Holocene time (i.e. in the last 11,000 years), or large faults which show evidence of movements in Latest Pleistocene time (i.e. between 11,000 and 35,000 years ago). In Australia, active faults make a significant contribution to the probabilistic MCE only at near-fault sites, and even in those cases most of the hazard comes from distributed earthquake sources. However, some sites may be close enough to nearby or even more distant identified active faults that a Deterministic Seismic Hazard Analysis (DSHA) produces MCE ground motions that are far larger than those obtained probabilistically even for very long return periods. Conversely, the deterministically defined MCE may be lower than the probabilistically defined MCE for very long return periods at near fault sites in New Zealand, requiring the probabilistic approach.
Extending the useful life of a dam to an extent well beyond what was envisaged by the original designer poses diverse challenges. In this paper, three case studies are described, one involving strengthening of two similar dams and two cases involving raising. In all three cases, the dams continue to provide a reliable source of supply in a water scarce country.
The Woodhead and Hely-Hutchinson Dams have substantial historical significance which guided the selection of restressable post-tensioned anchors as the preferred method of strengthening.
The Stettynskloof Dam was almost doubled in height by constructing a clay core rockfill embankment abutting the downstream face of the existing concrete gravity dam. The new structure was well instrumented to cover areas of concern but the dam was found to perform as largely predicted by the designers.
Keerom Dam faced both technical and regulatory challenges that were eventually overcome and the raising of the dam was able to proceed. A further raising will increase the utilisation of this valuable resource still further.
Peter Woodman, Andrew Northfield, Hench Wang
Current empirical approaches assume different fatality factors for the ‘fail’ and ‘no fail’ scenarios even when the same hazard is experienced by a property. This approach can lead to some inconsistencies particularly for small dams and retarding basins. This paper looks at the base data behind the current fatality factors and explores possible alternatives to the current approach. The paper will rely on a number of examples from a recent investigation undertaken by GHD for Melbourne Water on a number of their retarding basins.
Matthew Ind, Kate Brand and Mark Ferrier
The framework for undertaking a dam breach analysis for water dams is reasonably well established with a depth of information and software available to guide practitioners on a consistent approach to undertaking failure impact assessments. In contrast, dam breach modelling for tailings dams is currently a developing field with a wide range of modelling approaches taken and an inconsistency in the quality of the failure impact assessments undertaken. Recent tailings dam failures at the Mt Polley Mine in British Columbia, Canada and the Fundäo and Santarém dams at the Samarco iron ore operation in Minas Gerais, Brazil have provided a sobering reminder of the hazards presented by tailings dams and the clean-up challenges that are significantly more complex than a similar failure of a water dam.
Current guidelines and approaches to dam breach modelling are often done assuming the run-out material from the breach is just water without due consideration of the impact from tailings loss. There is limited analysis undertaken on credible failure modes of tailings dams with an assumption that the embankment just “breaks” at some random point without appreciation of the failure mechanism. The misunderstanding of failure modes leads onto inconsistencies with application on whether a ‘sunny-day’ or extreme flood event modelling should be applied, with one or the other selected without explanation.
This paper outlines a framework that can be applied when undertaking a dam breach study for tailings dams to enable a consistent and credible assessment of potential failure impacts. The following tasks are discussed in detail in support of this framework:
Robert Shelton, Jako Abrie, Matt Wansbone
The Mahinerangi dam – arguably the most valuable in Trustpower’s portfolio of 47 large dams – is over 80 years old and needs a plan of work to confirm it meets current design standards.
The dam was completed in 1931, subsequently raised in 1944-1946, and strengthened with steel tendon anchors in 1961.
A comprehensive safety review (CSR) in 2007 noted a potential deficiency in the fully grouted anchors and a program of work commenced to re-evaluate the overall stability of the dam.
A potential failure mode assessment revealed that the dam may need upgrading to meet the criteria for maximum design earthquake (MDE). Areas of uncertainty were identified and a significant programme of survey, geological mapping, concrete testing and site specific seismic assessments have been carried out to reduce risk and uncertainty in design.
The paper discusses the dam’s history, current condition, and describes the ongoing programme of work planned to extend the life of the dam for another 80+ years.