Dam Monitoring Effects
Aug 5th, 2020
Dam monitoring effect is a case history. A client acquired a mining properties portfolio. It included inactive dams at closed mines. Some of these structures were relatively well documented, while others were not. As a result, the client was facing important decisions, one of them being the effort to enhance the knowledge on the structures. The question the client asked was: “What happens to a particular dam if the level of knowledge we have on an inactive dam goes from “no knowledge” to a future “enhanced knowledge”? Can you please characterize the knowledge-probability of failure relationship?

Ore2 Tailings and knowledge level and Dam Monitoring effects
We deployed ORE2_Tailings on the dam. At first, we derived the 30+ KPIs from the existing level of knowledge using the Factor of safety (FoS)=1.5 the engineers evaluated for the dam. Beside the dam being inactive, we made other assumptions and conditions we summarize below.
We then described levels of knowledge which dictate KPIs partial and progressive alterations as follows:
- No knowledge: encompasses only external observable symptoms,
- Poor knowledge: as above plus assumptions on “good behavior and engineering common practice” of prior owner.
- Today’s knowledge: based on all information available today. That includes reports, monitoring results and discussions to date and finally,
- enhanced future knowledge: considers FoS would be confirmed. This requires to further remove geotechnical uncertainties by additional studies
ORE2_Tailings dam ratings
ORE2_Tailings allows to study causality of the potential failure. In addition it also to expresses a “rating note”. The minimum note is 0, the maximum note is 10. The rating note qualifies the overall quality of the structure and its management, maintenance, monitoring etc. Of course the 0 and 10 are theoretical extremes. In general the statements below are true:
- A site visit and discussions on site generally allows to go to 1 or slightly above. That would have been the case for this dam if no information would have been available. Let’s note however that in this case thick knowledge can be paramount. Indeed we have used it many times in our practice.
- Of course aerial views (including Google earth and perhaps a historic InSAR analysis, satellite optical imagery and NDVI) also can bring the knowledge level up looking backward and/or forward in time.
- There is no perfect knowledge and a bad dam will remain bad even if we know all the details of how bad it is and why.
- Finally IoT, big data and AI may be interesting additions, although dam failures are not frequent enough to allow machine learning.
The graph below shows how the dam rating evolves when knowledge level increases for the specific inactive dam we studied.

Dams ratings change with knowledge
We can see from the graph that:
- If we had no knowledge on the dam it would have a very poor rating note,
- poor knowledge would lead to 4 to 5.
- Today the dam sports a note of 7.2 and finally,
- if the knowledge is further increased the rating note could go higher than 8. However it would certainly not reach 10 for the reasons explained above.
The drivers of the overall notes vary as follows, based on the assumptions we made for this specific inactive dam:
- No knowledge: geotechnical/geomechanical gaps
- Poor knowledge: construction/geomechanical gaps
- As per today: construction/geomechanical gaps and finally,
- future enhanced knowledge: construction
Let’s consider now various “possible states” of the Dam.
It is interesting to see the influence of knowledge on the dam per se, or the dam combined with Probable Maximum Flood (PMF) and/or seismic activities. We have selected a few “possible states” for this analysis, being clear that we may select others. Here they are:
- Dam with infinite capacity ancillary water management facilities (PWM): this is the case of a dam where the ancillary water facilities would be designed with a PMF with incredible return (say over 100,000 years).
- PMF=1/1,000 ancillary facilities protecting the dam and finally
- dam as above (first bullet) and 1/10,000 quake leading to FoS=1
Here are the results of the annual probability of failure (PoF/yr) per each “possible state”, as the knowledge level goes from “no knowledge” to “future optimum knowledge”:

We see from the graph that both the PMF (1/1,000) and the MCE (1/10,000) become preponderant with respect to the level of knowledge. As a result they reduce the effect of knowledge level on the structure itself. This reconfirms how important it is to gain confidence on the return and effects of these phenomena.
As per the estimated PoF/yr PWM here is it variation as knowledge level (Dam rating note) increases:

Investing to increase knowledge, e.g. monitoring, investigations, can, if we consider the extreme cases, generate reductions of PoF spanning over 3 orders of magnitude! The specific dam in this case history is evaluated today to stand just below the lower bound of the world-wide benchmark.
Dam Monitoring effects can be quantified
A cost-benefit analysis on further investigation based on quantitative risk analyses can lead to better mitigative roadmaps, especially when a dam portfolio is analyzed.
Cost-benefit analysis of this kind can support risk informed decisions. They avoid squandering CAPEX and launching in unsustainable long-term actions. In addition, they also help justify initiatives at Board level and in front of the public, finally fostering CSR and SLO.
Tagged with: dams, effects, effects of mitigation, factor of safety, Information, insar, monitoring, NDVI
Category: Hazard, Mitigations, ORE2_Tailings, Risk analysis, Risk management, Uncategorized
Leave a Reply