On February 27, Roussos Dimitrakopoulos received the Leo Derikx Award, one of the annual Synergy Awards for Innovation presented by the Natural Sciences and Engineering Research Council (NSERC).
Whereas technology was once used to evaluate what happened in the past, mathematical modelling and computation power is increasingly being focused on gaining foresight. In other words, miners are now driven to determine the smartest way to plan operations, while taking into consideration all possible outcomes.
The name for this is “stochastic” mine planning – a process based on complicated mathematical models that challenge the very foundation of how operations have been conceived. Through this new lens, instead of one answer to the question ‘what is the most profitable way to mine?’ miners can see several options. And, though all paths remain uncertain, the use of quantified uncertainty is very profitable.
Be sure of uncertainty
“Conventional mine scheduling wisdom is to mine the highest economically valued parts of an ore body first, and to base your actions on the average of the possible places where that high economic value lies,” says Roussos Dimitrakopoulos, Canada’s research chair in sustainable mineral resource development and optimization under uncertainty at McGill University. “Stochastic mine planning, in contrast, evaluates multiple possibilities to minimize deviation from production targets. The financial implication is that you are led to higher economic values along with more metal production expectations.”
In brief, to incorporate stochastic theory into mine planning is to take into account that mines operate in uncertain, complex and high-risk global environments. The variables affecting a mine can combine in any number of ways, and their complex relationships can change dramatically over the life of a mine.
A history of the concept
When Paul Anderson took over BHP in the late 1990s, he would receive reports from oil divisions probabilistically, meaning that figures for estimated cash flows and other key indicators were accompanied by statistical reference points on how likely those targets were to be met. For the mining divisions, that range of probabilities came down to just one number.
“The mining industry doesn’t embrace statistical analysis the way the petroleum industry does,” says Anderson, who retired from BHP Billiton as managing director and CEO in 2002. “For instance, the theory we were pursuing at BHP was: could you have a mining company that had a sustainable dividend rather than being subject to the cyclicality of most mining companies? A lot of others in the industry said you couldn’t do it, that it is a cyclical industry and that stochastic techniques are just a bunch of fluff to make it sound like you are sophisticated.”
Along with BHP’s CFO at the time, Chip Goodyear, Anderson undertook an enterprise-wide project to apply probabilistic outcomes of mining projects, taking into account oil prices and the various minerals and currencies they were dealing with. By weighing out each of the projects, they came up with a cash flow at risk curve for the entire enterprise, allowing them to make decisions about whether to invest in, for example, a copper mine in Chile, an iron ore mine in Australia, or an oil well in the Gulf of Mexico, while minimizing volatility of earnings.