Light on Tailings dams failures

Light on Tailings dams failures

Nov 13th, 2019

The Church of England Pensions Board and the Council of Ethics of the Swedish National Pension Funds published some recently collected numbers shedding some light on Tailings dams failures.


How many active dams in the world did we assume earlier?

In the literature we oftentimes see the number 3500 to characterize the total tailings dam inventory in the world. We used that number in both the following papers:

The origin of that number is Davies, M. P., T. E. Martin, et al. (2000). Mine Tailings Dams: When Things Go Wrong. Tailings Dams 2000, Association of State Dam Safety Officials, U.S. Committee on Large Dams, Las Vegas, Nevada: pp. 261-273. The paper states:

“There is a very poor database of the world’s tailings dam inventory. From an extensive literature review and discussions with regulatory officials worldwide, it is estimated that there are somewhat more than 3500 tailings dams worldwide. This total is made up of contributions that include the following where relatively good inventory lists exist: 350 in Western Australia, 65 in Quebec, 130 in British Columbia, 400 in South Africa and 500 in Zimbabwe”.

The tally of geographic areas above gets to 1445. More than doubling that was considered sufficient to include the US, Latin America and a few other major mining areas of the world.

The Church of England inquiry

The Church of England inquiry has led to a total number minus duplicates (companies reported the same tailings storage facility) of 1635 tailings dams from:

  • 37 out of the top 50 largest mining companies (74%) in the world,
  • 76% of the top 50 largest mining companies by market capitalization,
  • 69% of the industry by market capitalization, and
  • all 22 out of 22 publicly listed members of the International Council of Mining & Metals.

This list totals only 43% of the contacted companies. Indeed 416 companies did not respond, representing 57%.

The same source reveals a total of 667 active dams.

The paper defines the number of dams per continent as follows:

  • Australia 272
  • Asia 226
  • Africa 266
  • North America 498
  • South America 279, and finally
  • Europe 91

The total here is 1632, not 1635. Thus one could assume that the active vs. inactive dams is around 667/1635= 0.4 (one active dam per 2.5 inactive dams).

We note here that the article seems to use tailings dams and tailings storage facility as synonyms. Now, TSFs may include more than one dam, and the list also includes dry stacks and in pit depositions. Like in other occasions we note the paramount role that clarity in the language and glossary should have in this type of communication

A statistical graph showing one century of tailings dams failures around the world.

From a presentation we displayed in a paper  Space Observation, Quantitative Risk Assessment Synergy Deliver Value to Mining Operations & Restoration, (see the presentation) Rouyn-Noranda, 2018, Symposium on Mines and the Environment we can count the following geographic distribution of tailings dams failures (any size) in the decade around 2000 (last two column at the right):

  • Oceania 0
  • Asia 6+7
  • Africa 0
  • North America 5+10
  • South America 7+10, and finally
  • Europe 9+2

Total is 56. Of course, like usual, these numbers include some under-reporting.

How many tailings dams can we estimate in the world?

As we recently stated in this blog (facts about worldwide tailings dam performances) one can make statements about tailings dam reliability only once one knows the world inventory “anagraphic”.  

If we look at the world inventory from the “storage type” point of view, we have, based on the same communication of the Church of England:

  • downstream construction 523
  • upstream construction 635
  • centerline construction 121
  • dry stacking storage 66, and finally
  • in pit storage 53

Now if we focus only on tailings dam we can count 523+634+121=1278 downstream, upstream, and centerline dams. It is difficult to understand from reading the communication what causes the difference between 1632 (or 1635) and 1278. Again, probably, a glossary/language problem or form return interpretation difficulties? Indeed the difference is 1635-1278=375 which is still way larger than drystacks+in pit storage or 66+53=119.

Furthermore, if we assume that the 43% of responders are representative of the “entire population”, the world population of dams (active and inactive) would be either 1635/0.43 or 1278/0.43 thus, respectively 3802 or 2972. At this point we would be inclined to think that the good old estimate of 3500 dams in the world remains valid.

Additional light on Tailings dams failures

Now that we have confirmed the old estimate of 3500 dams in the world, we can say that 0.4*3500 are active, that is 1400.

If we now consider the  recent study (A Comprehensive Review on Reasons for Tailings Dam Failures Based on Case History) we commented on our blog post ( link) we notice that in the last 18 years we had the following recorded failures per type of dam construction:

  • 11 upstream
  • 2 centerline and finally
  • 4 downstream

N.B. the total from this paper (17) is different from the total (56) above likely because of “counting biases”, poor information regarding tailings dam failures, cause, dam construction method, and dam height.

Assuming the numbers above, failures occurrence rates from upstream and centerlines are roughly the same: 11/635~= 2/121 (or appx. 0.017, 0.016), which are more than double (2.35 times) the downstream dams failures (4/532=0.007 failures per dam).

We suspect that again approximate glossary and definitions may be influencing these number. For example we think that “centerline” may include the so-called “modified centerline”. This is in many cases more an upstream construction than a centerline one, significantly influencing the “fragility” of the structures.

If we now take the occurrence rates per dam type and convert them to yearly frequency we get 0.017/18=9.44*10-4, respectively 0.007/18=3.89*10-4.

The sum is 1.33*10-3 which multiplied by 3500 leads to 4.6 failures per year.

This number is still fraught by numerous assumptions and uncertainties. However, it compares very well with our first estimates made in 2013 (Oboni, C., Oboni F., Factual and Foreseeable Reliability of Tailings Dams and Nuclear Reactors -a Societal Acceptability Perspective, Tailings and Mine Waste 2013, Banff, AB , November 6 to 9, 2013) and in the aftermatch of Mount Polley’s dam break, for the decade around 1979, namely 1*10-3.


As these numbers are frequencies, which by definition are “averages”, it is too early to conclude that we are actually seeing an acceleration in the rate of failure of dams. That is, of course, once we filter:

  • sensationalist media and activism,
  • ludicrous correlations made with straight lines through a spread of points with a correlation coefficient far from significant.

What we think is undeniable, however, is the fact that at three to four major failures on average per year the situation is untenable both from the ethical standpoint and the Social License to Operate (SLO) one.

Our industry has to actively work to prioritize mitigative plans of actions, communicate them to public and governments using coherent and unified language to avoid any further blunder.

Tagged with: , ,

Category: Consequences, Mitigations, Optimum Risk Estimates, ORE2_Tailings, Probability Impact Graphs, Risk analysis, Risk management

3 responses to “Light on Tailings dams failures”

  1. john metzger says:

    Valuable contribution.

  2. Lindsay says:

    Actally the correct esrimate is 29,000-35,000 16,000 active……
    coe is 7% of all tailings depositions 4% of tailings facilities nt represeentative of world tailigs ortfolio and ull of errors

    • Cesar Oboni says:

      Thank you Lindsay for this comment.
      It is true that the number of tailings dams on the planet remains unknown despite the actuarial efforts to date.
      The very large numbers of dams you describe should be classed by size and other parameters if you want to be able to meaningfully talk about the “world portfolio” risk.
      You should also define failure, consequences and probabilities (or rates if you look backward) of those failures. Risk should be used to take better decisions, not to deliver sensationalist news.

      In prior papers you have focused your attention on very large failures, so you should be consistent in your statements and use large dams capable of generating those failures only.
      Indeed, if you claim that there are 35,000 dams in the world, and then we consider the long term average of about 3.5 very large catastrophic failures per year, we get a rate of 1/10,000 per annum.
      If that was the case, tailings storage facilities would not have earned the bad reputation they have.
      Their rate of failure would indeed be comparable to Class 5+ nuclear accidents as we showed in a paper back in 2013.
      If we look at the dams capable of generating those large catastrophic failures you have focused in the past, we bet that the 3,500 estimate may not be so wrong.
      Then the rate is around 1/1,000, as we published back in 2013.

      Severe and mortifeous accidents with a rate of 1/1,000 are known to scare the public and stir public opinion as it has occurred in the last decade.
      We showed in our 2013 paper that indeed the mining industry had reached a threshold of societal unacceptability, like the nuclear industry.
      Given that, as expected and after a couple more dramatic events, a number of initiatives burgeoned on the planet, culminating with the Global industry standard.

      To conclude: the worldwide risk is not proportional to the number of dams: you have to be a tad more sophisticated than that.

      Claiming the number is larger does not necessarily ensure the attention of governments and decision-makers or will lead to misplacement of mitigative capitals.

      We think the high rate of catastrophic failures in the last decade has triggered pertinent reactions world-wide: in order to maintain the motivation we need to be rational and avoid sensationalism.
      Unless the word risk is used properly, misinformation and confusion will arise instead of better dams performances.
      Now it is time to engineer a better future, and to communicate risk in the most rational and dispassionate way.

Leave a Reply

Your email address will not be published. Required fields are marked *

Riskope Blog latests posts

  • New achievements in risk assessment and management
  • 2-05-2023
  • New achievements in risk assessment and management will be attained thanks to SRK Consulting merging with Riskope. Indeed, we are…
  • Read More
  • Open letter to the organizer of the tailings dam round robin exercise
  • 29-03-2023
  • Dear Ryan, please receive this open letter to the organizer of the tailings dam round robin exercise. It explains our…
  • Read More
  • Landslides risk assessment and monitoring
  • 8-03-2023
  • During the first couple decades of our professional life we worked extensively with Landslides risk assessment and monitoring in the…
  • Read More
  • Get in Touch
  • Learn more about our services by contacting us today
  • t +1 604-341-4485
  • +39 347-700-7420

Hosted and powered by WR London.