Managers often find themselves caught between a rock and a hard place: Do you replace your asset early then spend more on cost of capital, or do you replace your asset later and risk it failing? Mainstream Conference speaker Carl Kirstein presents a case study showing how you can make this decision easier by expressing risk as a cost.
Three production rope shovels at a mine were approaching the end of their expected lives and management wanted to know whether it would be worthwhile to stick to the replacement plan. The shovels were already running beyond their normal lifespan and stress cracks on structural members were becoming excessive. These cracks however did not clearly say whether it is worthwhile to replace the shovels earlier, first they had to be translated into risk.
Replace earlier or later?
We needed to determine the risk associated with replacement ages. The well known risk matrix would only provide a risk index or risk score, but these are not comparable to the cost of replacing a shovel earlier or later. For them to be comparable risk has to be expressed as a cost, or an expected loss. The best replacement ages of the shovels are determined by comparing their expected losses to the cost of capital for various replacement ages.
We needed to do reliability modeling of all the end of life components to determine the probability of failure in any year as well as the impact of their failure. The impact included the effect on production, maintenance costs and contractors. Combining all these impacts determines how much risk each shovel is carrying in any particular year.
We used a parametric-stochastic risk analysis method that we adopted and adapted for our company. The parametric modelling was done through curve-fits of actual data and from the educated guesses of technical experts. The stochastic modelling was done with a Monte Carlo method, using 10,000 simulations per component impact- and probability of failure. With 31 end-of-life components each with an average of 5 impacts per component on 3 shovels (plus a new one) for 5 years gives 62 million simulations. With today’s computing power on a normal laptop this remarkably took a few seconds each time the simulation was done.
How does this compare to other methods?
This is not a new concept, but it is certainly not pervasive, perhaps because it is too intimidating. However none of the other risk analysis methods that I have encountered quantifies risk into something that can be directly compared to costs. Besides, once the simulation tool has been developed, it is quite intuitive and easy to use. The maintenance personnel involved in the process thought that this was much better than selecting from a risk matrix, because everything they said was quantified and included. And, crucially, the method accommodated uncertainties, i.e. it relishes in ranges and distributions, instead of demanding exact values.
After doing the analysis we determined that the risk of continuing with one of the shovels was too high and it needed to be replaced as soon as possible, the other two shovels could still be nursed for another year or so.
The overall process made the strategy a lot clearer, giving a clear understanding of the cost drivers and allowed us to compare the cost of capital against the risk of replacing the shovels.
We built the model with 31 components to be replaced near the end of the shovels’ lives, but this turned out to be unnecessary for components that are replaced frequently. Going forward we have decided to base the simulations on only the components that are meant to last the life of the machine.
This method is a lot easier than it sounds. It allows the engineers and Management to speak the same language: money. This in turn allows the decision process to be much less painful.
Carl Kirstein is a popular Mainstream Conference speaker and the Principal Engineer of Mobile Equipment & Technology at Exxaro Resources Ltd.