The tail end of a story

21 February 2011



Born on a gold mine in South Africa, Jack Caldwell grew up around the slimes dams. He was educated as a civil engineer on a mining scholarship at the University of the Witwatersrand, Johannesburg. Since then he has consulted with mine owners the world over on the design, operation and closure of tailings impoundments. Here he gives a personal and straight-talking account of how to build successful tailings dams


The Teton dam failed in Idaho in 1976. It had been designed by the US Bureau of Reclamation and a group of experienced dam engineers. Eight years later, and I had just been retained to help design and construct the Cannon Mine tailings impoundment in Wenatchee, Washington state. Our site had the same rock strata as those that had piped and reportedly caused the failure of the Teton dam. Preliminary analyses showed that we needed an embankment at least 91m high to retain the tailings. The embankment eventually rose to a height of 103m, making it the highest privately owned dam in Washington.

Then, as now, tailings dams failed at a far greater rate than conventional water-retaining dams; and their failure caused great damage. We were nervous. Syd Hillis came to our rescue. He had been the chief geotechnical engineer on Tarbela, the largest earthern dam in the world. He had also been the chief geotechnical engineer for the Revelstoke dam, then the highest earth dam in the world. Now he was a private consultant and a peer reviewer of dams funded by the Asian Development Bank.

We met; we argued; we fought; and we became firm friends. I moved to Wenatchee to design the dam as we went. There was no money in the mine’s budget for site investigation, we did it as we stripped the soils. Syd came down once a month. He walked the site with me. He felt and tasted the soil; who is allowed to do that these days? And he insisted on conservative measures: filters placed every place piping was conceivable; filters upgradient and downgradient of the core, just in case cracking occurred; compacted rockfill; and large drains. The embankment was built above some 20,000 homes and stands today, reclaimed and the home of the Dry Gulch Riding Stables.

Proud as I am of this embankment that was built to the highest standards of any water dam, I have to admit that long before this I had designed a tailings impoundment that failed some years after initial construction. That was for the De Beer Diamond Mine in Kimberley, South Africa. Nobody was hurt; no environmental damage was done; it is not even listed in those long lists of tailings dam failures you find on the web.

I had started a life-long career in tailings dams as a child in South Africa, doing the forbidden riding around slimes dams of the mine where I grew up. My first job out of university was supervising the pouring of concrete for construction of the Hendrik Verwoed dam on the Orange River. That dam is now called the Gariep dam in honour of vast political change in South Africa.

As a masters student at university, I helped Professor Jennings investigate the failure of the Bafokeng slimes dam that killed 13. I got to talk so fast, they gave me the job of designing the new one; it is still in operation and easily seen on Google. But the difference between the attention paid to the design, construction, and operation of the Hendrik Verwoed dam and the Bafokeng dams, old and new, readily established why tailings dams fail more frequently than water dams.

Now nearly 65 years of age, I still consult on tailings dams. I am currently involved with tailings dams in Guatemala, Alberta, the Northwest Territories, and a small country in Africa that prefers to remain unnamed. I am obsessed with potential failure of the tailings facilities that I touch, and hence I write to explore reasons why tailings dams fail so often. And I seek ways to prevent the terrible frequency of tailings dam failure. Here are few ideas.

Failure of a system

I am no fan of attempts to ascribe failure of earth embankments, whether for tailings or water, to single causes. You know the typical list: foundation sliding; differential settlement; too much water; overtopping; piping; bad design; and so on. On the basis of real-life experience and much reading and thinking, I must conclude that the failure of dams, for tailings and water, is a failure of the system. That is the system by which the dam is regulated, permitted, designed, constructed, operated, monitored, and closed.

I believe that we need the following to reduce dam failure: good laws and regulations; the best designers and engineers; plus consistent and regular peer review.

In the absence of good laws and clear regulations, the designer is adrift – subject to the whims and whiles of owners who seek to reduce costs, cut corners, and do less than is necessary. Right now I am fretting over a water-retaining structure that was designed where there is no law, and the least possible was done. The dikes are failing and I am faced with telling the owners to spend a lot to upgrade or abandon the facility and build a new one. I am not popular and may be replaced by other consultants. So be it.

If the regulator is inexperienced, uneducated, or inattentive, they will permit any bad old design presented to them. They fail their public trust and duty. In too many countries where mines are developed, it is a simple matter to beguile the regulator, or worse to bribe them. The resulting dams are at high risk of failure.

Sad to say, but there are good consultants and there are bad consultants. There are too few good consultants to do all the work. Bad consultants do design dams and these often fail. Such facts are not listed in causes of failure, but I challenge investigators and historians to test my thesis that bad designers are a significant contributor to dam failure.

Then we have failure to peer review. Syd Hillis taught me the value of peer review. Without him, the Cannon Mine dam would not be what it is: safe and secure and likely to last for as long as I can conceive – a new geomorphic form in the landscape. I have produced designs that have been reviewed by other peers. When you assemble a team, at least three strong, of honest engineers, you cannot but succeed. As an owner you are assured things are properly done; as the design engineer, you know you are up to par; and as a member of the public, you can be assured you are safe. Thus I list these minimum desiderata for a safe dam, for water and/or for tailings:

• Appropriate laws and regulations.

• Trained and conscientious regulators and permit granters.

• The best design engineers you can assemble and afford.

• Focus on site selection, site characterisation, testing, and analyses, more analyses, and profound judgment.

• Peer review at every stage of design, construction, operation and closure.

• Informed and interested operators, who measure, monitor and report every observation.

• A maintenance and monitoring plan that can be communicated, understood and implemented.

• Transparency and regular reporting to a public body that posts the reports of design and operation on a readily accessible web site.

My theory is that no dam fails for a single reason. In every case of failure that I have been associated with, there are at least ten things that went wrong before failure occurred. I know from personal experience, that at all dams there are always ten things wrong. But not all dams fail. It is only when the stars malignantly align, when the ten wrong things line up in a negative way, that failure occurs and people die. (Nothing new or insightful in this observation/conclusion.) Standard accident avoidance theory and practice is that for every 100 incidents, there is one accident. For every 30 accidents, there is one death (the ratios vary depending on whom you believe – but the idea is sound.) It is the obvious pyramid of events: control the incidents, and you eliminate the accidents and deaths.

If an incident is defined as a small, irritating occurrence with no particular consequence, you may readily and graciously investigate the incident and put in place practices to prevent recurrence. I learnt this as chief geotechnical engineer on closure of the Operating Industries Landfill just to the east of Los Angeles. This 526,000m2 landfill is the largest hazardous waste landfill on the US’ Environmental Protection Agency’s superfund list. The slopes rise at 1.4: 1.0 – that is steep. And the slopes rise some 91m above I60, a six-lane freeway exiting the city. The landfill is underlain by an active fault. I felt an earthquake one day, and we swayed like jelly on the soft waste beneath.

I was digging and profiling test pits on the steep slopes. A clod broke free of the pile of excavated soil. The clod rolled down the slope, jumped onto the freeway and hit the side of a passing car. It indented the car door, leaving a dirty brown patch. This incident was carefully investigated. The result: a US$100,000 plywood barricade between the landfill and the freeway. We went on to complete US$100M of work safely. We owe our success to Jill Saminago who, as our health and safety officer, insisted on incident investigation and control as the way to save lives.

The lesson learnt from that story is that at every dam, at every stage of construction and operation, have and implement an Incident Control Plan.

According to my agreement with the editor of this magazine, I have but five-hundred words left to make my case. Let me use these remaining words to acknowledge Terzaghi and Peck. Terzaghi is the father of soil mechanics; it is he who first enunciated the basic principles of the control of soil and the flow of fluids through soils – something that happens at all dams. Starting as a professor in Vienna in the 1930s, he went on to a career as a soils mechanics and dam design consultant. He is coauthor with Peck, his student, of what is still the best book on the topic today, namely Soil Mechanics in Engineering Practice. Recently updated by Mesri, it is still on my shelf; I read sections every so often to remind me of good practice; I make all young engineers who come my way read it so that we have a common basis of understanding.

Terzaghi taught the professor who taught me geotechnical engineering. In Albuquerque, my office overlooked the suburb in to which Peck retired; he often came to the local engineers meetings and chatted with all of us as locals, as we all were. From him and his writings, alone and with Terzaghi, I learnt the dangers of the overlooked geological discontinuity, and the immense value of the observational method.

The observational method is easy to state and understand. It is terribly difficult, however, to compile and implement an Observational Method Plan. I have worked on two: one done informally; one done most carefully and formally. Most other attempts I have observed have used the words, but failed miserably to catch the intent or apply the power of the method.

Nevertheless, I must end by insisting that if you want to preclude failure of your dam, for tailings or water, you must struggle, strive, and fight to write and implement an Observational Method Plan. The reality is that dams are geotechnical structures. We cannot establish perfection in geotechnical conditions; we can make only faint attempts to model and predict performance; we must therefore monitor and adjust in accordance with a pre-existing plan to the actual conditions and behaviour of our dam as it is built, filled, operated, and taken safely to closure.

Jack Caldwell’s blog on mining and tailings dams can be found at www.ithinkmining.com. Email: [email protected]


Author Author
Containment Containment
Cover Cover


Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.