I think it’s fair to say that we’ve all grown accustomed to miners overpromising and underdelivering. Whenever a new project is coming online there is a palpable uneasiness that is detectable in analysts, management, and investors. Will the project hit the guidance throughput? Mining Rate? Recovery?
I don’t want to group all projects into the underdeliver bucket but there is certainly a large percentage that not only fail to deliver on ramp-up projections but also in ultimate capacity. The chart below shows actual vs. projected mining rates for a large project that was recently constructed. Now I get that management can be overly optimistic and ramp-ups can be challenging. You’ll notice that the year 1 tonnage is way off of the projection but, by year 2 and 3, the company has closed much of the gap. What I’m more interested in is the levelling off of production and the inability to ever achieve expectations regarding ultimate mining rates.
Management will often highlight specific variables that are performing worse than expected. Shovel downs are decreasing availability. Pioneering is impacting productivity. A large weather storm reduced operating time. These are all possibilities. What I’m interested in, however, is why production falls short of budget when all variables are as expected.
But how can this be? How can production be less than expected when all inputs are as expected? Well, I think that the companies contracted out to perform mining studies underestimate volatility’s role in determining production rates.
Let’s look at a simple example:
The capacity table shows one shovel that is paired with three trucks. You’ll notice that truck and shovel capacity is fairly balanced, both around 25K tons mined per shift. You could expand the time period to encompass a year and this is how most feasibility studies determine equipment requirements. Now they would use variable parameters for each of these factors based on the specific conditions in the mine during the period (haul distance) and planned maintenance.
Ok, nothing wrong here, right?
These assumptions grossly underestimate the influence randomness within each variable. Fluctuations in, say, availability have asymmetric impacts on the production of the mine. -10% one day and +10% the next does not average out to zero impact on the mine. It averages something less.
This histogram of shift-by-shift availability reflects the average that is shown in the table (77%). This histogram reflects the outcome of 730 simulations (1 year’s worth of shifts) of another probability density function that is based on actual data.
These availability values were used to calculate shift production for an entire year. The chart below shows the mine’s production as a function of shovel availability. You’ll notice that tons mined per shift increases linearly as shovel availability is increased -the mine is shovel limited-. Once availability exceeds 70%, however, the mine is limited by truck capacity.
This is the root of the issue. Mine production falls short when shovel availability is lower than average and cannot sufficiently increase production when availability is higher than average. This asymmetry makes it effectively impossible for feasibility parameters to be achieved.
Average simulated production is barely over 20K tons per shift. A far cry (-18.5%) from the 25K tons per shift that would be predicted by the study. Coincidentally, this is the same delta between actual and planned tonnage in Y5 of the earlier chart.
Mines are fragile. They are hurt by volatility and cannot make up production on the positive side because rates are capped by a new bottleneck. It’s my opinion that this static mindset, when it comes to production scheduling, is the root cause of lots of the industry’s issues.
So what do we take away from this very simplified example?
Well, it pays to have buffer (excess) capacity. Miners with small equipment fleets are the most susceptible to volatile operating parameters as they don’t have other units to average out the shift-by-shift outcomes. In general, a static approach to forecasting will overestimate long term production.