Boris Karloff was the ultimate movie monster: tough, scary, the stuff of nightmares. He had just one weakness: he was scared of fires. Mines in many ways are like Dr. Frankenstein’s monster; rugged, imposing, but also scared of fires. Maybe a different kind of fire though, the statistical fire of an outlier.
Outliers are defined as “a data point that differs significantly from other observations”. Popularized by Malcolm Gladwell’s book of the same name, outliers present areas for unique insight. Uncovering why something is an outlier is great for developing new and innovative processes, revealing new insights and discovering step-change improvements.
Unfortunately, analysis and resolution of negative outliers can also lead to significant waste of capital and investment. For example, flooding is frequently categorized by the likelihood of an amount of rainfall over a period of time: a “100-year flood” or rainfall causing a flood that statistically should only happen once every 100 years. This doesn’t mean you won’t get two 100-year floods in two consecutive years. It just means that statistically, it’s unlikely. Mathematically, this is 1% x 1% = 0.01% likelihood. Fairly unlikely.
Mines tend to operate similarly, responding to major events, whether they are safety, production losses, major equipment or plant failure, or other catastrophic failures. A “fire” is run to, resolved, and great effort is made to help ensure that “fire” never happens again. Again, Fire = Bad.
A better response lies in a data-driven, two-stage approach to problem solving. First, quantify the likelihood of the negative event happening. Multiply that by the event’s impact from a financial or societal perspective.
Example 1: Equipment Failure
Equipment failure is a common example of a negative outlier, especially if it impacts production. One piece of equipment critical to most mines is the loading and hauling units. Let’s say in an example mine, there are two loading units (wheel loader, face shovel/excavator, etc.) and 10 hauling units. Generically speaking, you could estimate that each loading unit handles five hauling units. If a loading unit goes down, 50% of the production is lost while it is being repaired. Conversely, if one hauling unit goes down, only 10% of the production is lost.
Modern mines measure their equipment effectiveness using a Time Utilization Model (TUM). The Global Mining Guidelines (GMG) Group has developed a good example. This measures the amount of time the equipment is used, when it is not being used, and why it is not being used. Let’s say for example, a mine’s TUM calculations show that unplanned maintenance downtime of face shovels is 1% of the total available hours. Conversely, the unplanned maintenance downtime of haulage equipment is 10% of the available hours due to more potential failure modes with more subsystems that could fail, such as brakes, tires, suspension, hydraulics, payload, etc. Obviously, if a face shovel goes down, it will receive a lot of attention as 50% of the production is stopped. But which has the biggest impact over time?
Shovels = 1% unplanned maintenance downtime x 50% production loss = 0.5% total impact to production
Haulage Fleet = 10% unplanned maintenance downtime x 10% production impact = 1% total impact to production
Equipment failure is a common example of a negative outlier, especially if it impacts production.
Clearly, the shovel going into an unplanned downtime is a larger “Fire” and remember, “Fire Bad”. But the smaller fires occurring more frequently on the haulage fleet have the greater impact when you inspect the data. If you’re focusing efforts on a Reliability Centered Maintenance program, starting with the haulage fleet makes more sense.
Example 2: Safety
Accidents, especially fatalities, should always be treated with the greatest attention. Investigating why an accident happened and how to prevent it in the future is essential. Part of working with Hexagon and its award winning safety suite sometimes means I am working with new customers who are trying to prevent the recurrence of a recent safety incident. We frequently begin with a comprehensive review of the recent incident or accident. Utilizing frameworks like the ICMM Critical Control Management Framework, we analyze the details and determine what preventative measure could have prevented the accident.
Unfortunately, these sites have often missed common scenarios that are well known, well understood, and easily prevented through a combination of standard operating procedures (SOP) and safety technology when those procedures are missing. These are not outliers, but rather typical, well known and well understood. For example, parking behind the blind spot of a truck is usually a violation of site SOPs and a safety technology like Hexagon’s MineProtect Vehicle Intervention System would prevent the truck from reversing over a vehicle whose driver neglected those SOPs.
Hexagon’s MineProtect Vehicle Intervention System prevents a truck from colliding with a vehicle whose driver neglected standard operating procedures.
Inversely, sometimes the accident or incident comes from very rare circumstances that could never be predicted. These “corner cases” usually involve a series of poor decision-making coupled with physical unplanned breakdowns and use cases that today’s technology could not prevent. It equates to a series of unfortunate events – the safety version of a 100-year flood: a combination of events creating a unique outlier. This does not mean that good ideas cannot come from reviewing the breakdown, but it may mean that preventing the exact series of breakdowns from ever happening again is not the biggest scenario facing the wider industry.
In this case, we can look at the wider industry and see that globally, we still need to focus on solving those critical scenarios that are well known and still not covered with the controls and SOPs that could prevent them. The reality that we can prevent accidents and incidents with known practices but still are not is sobering.
Conclusion
Prioritization is critical to tackling problems in mining. Focusing on a data-driven approach allows us to separate the fires into those that are objectively the biggest based on total impact, versus those that only appear on the surface to be critical because of the immediacy of the problem. Yes, fire bad, but let’s focus on stopping the most impactful fires first.