Equipment Failure Catalyzes Losses; FM Global Believes Predictive Analytics Can Help
Risk management can often be an uncomfortable box for risk managers and CFOs who need to aggressively manage serious business risk that can impact the bottom line, while at the same time being frugal in how they spend money to manage those risks.
When it comes to managing equipment-related risk, the challenge is even greater. Shutting down production for even a few hours for maintenance can have a tremendous impact on profits, and often plant managers are reluctant do so.
Weighing the likelihood and potential severity of equipment failure can be challenging. And it can be next to impossible to decide where among hazy, hard-to-measure risks to invest limited resources.
A recent survey by my company, FM Global, of business leaders with company-wide responsibility for overseeing equipment operations or equipment risk in Fortune 1000-size organizations shows the risk of equipment failure is rising and the consequences could be serious.
In fact, the survey found equipment risks have increased over the past five years in the judgement of 43% of respondents. The stakes, they believe, are even higher, with 75% indicating they would expect it would take months, at least, for their companies to recover financially from the failure of critical equipment.
Fortunately, predictive analytics combined with historic data can slice, dice, filter and clarify equipment risks to progressively give CFOs and risk managers more concrete, actionable information than ever.
I’m specifically referring to data on actual risks generated through evaluations by engineers, either for insurance carriers or equipment manufacturers, who have visited actual commercial and industrial sites. This data is being enhanced by data on actual catastrophes and business disruptions throughout recent history.
Based on this data, risk managers can get improved information they can trust on exactly where to invest in risk reduction.
Here’s a simple analogy: There’s ample data on the safety, mileage and reliability of different automobile models. But let’s say you have a commercial fleet of 50 cars all of the same model and an annual maintenance budget of $5,000. It sure would be nice to know which cars in the fleet are most likely to fail, how costly those projected failures would be, and what particular mechanisms within those cars are the shakiest. Then you could spend that $5,000 wisely. Without that information, your $5,000 would likely be wasted.
Understanding how, when and why equipment might fail, is the key to ensuring resilience in a manufacturing environment, for example.
Here are three ways to use such data to help prevent equipment failure:
1) Risk benchmarking
It is possible to benchmark a specific company’s overall property risks relative to each property in a portfolio and against industry peers.
For instance, our clients’ portfolios are sorted into risk quality quartiles based on their inherent risk (e.g., are they in a flood zone?) — and deficient risk (e.g., do they lack sprinklers in their warehouses?).
This benchmarking has given property owners a good, basic understanding of their aggregate property risk.
It can also be applied to equipment risk.
It’s equivalent to saying, “I’m sorting your cars into four groups based on how risky they are, at first glance, relative to one another and the industry.” Properties benchmarked to be in the highest-risk quartile have proven to be 7 times more likely to suffer a loss than those in the lowest-risk category, and the losses are 30 times costlier.
2) Predisposed locations
Based on the historical data I mentioned, property owners can now look across their facilities and identify a small number of their perhaps hundreds of pieces of equipment that have the highest predisposition to suffer a loss.
Here’s where predictive analytics come into play.
Risk calculation is based on how their current conditions align with actual historical loss experience. With this information, a risk manager can now start prioritizing their planned investments to certain pieces of equipment in their portfolio. They can combine this list with their own knowledge of which operations within their businesses contribute the most to their bottom line.
Our data shows that locations flagged as most predisposed to suffer a loss are 15 times more likely to sustain a significant loss.
3) Relative likelihood and severity
It’s good to know which critical pieces of equipment are most likely to suffer a loss.
It’s even better to know which exposures within that equipment are most likely to be associated with a loss.
Here, the analytics are looking for combinations of deficiencies historically associated with losses.
For example, being able to ascertain that this piece of equipment at this location has the highest predisposition to a loss based on the very high likelihood that, given the age of the equipment and level of historical maintenance, the equipment will fail causing an estimated loss.
Experience shows that hazards flagged as highly likely are twice as likely to produce a loss.
Understanding how your valuable equipment, such as generators, turbines and chemical vessels, will perform and hold up over time can help you better identify which pieces are most likely to fail, how likely they are to fail, and how severe the loss would be.
The seven equipment factors known to correlate with failure (or resilience) are maintenance quality, operating conditions, environment, history, operators, contingency planning and safety devices.
Equipment flagged as “at risk” is 10 times more likely to break down, and breakdowns are five times more severe than others.
Why This Matters
There have always been smart people to estimate your risk. But understanding actual loss experience based on data gleaned from current conditions on specific pieces of equipment at specific locations makes equipment risk identification more robust than ever.
In this economy, such information is critical.
In recent years, our clients’ losses from equipment failure were responsible for nearly a third of all losses combined, rivaling those from fire.
Predictive analytics are taking the guesswork out of risk management by essentially providing a crystal ball to risk managers. It also provides underwriters a more detailed view of the risk as it exists today, and how it can be improved in the future. This allows bigger bets on providing large, stable insurance capacity at more competitive pricing.
Although it’s hard to prove a negative (i.e., that a given disaster didn’t strike because of a given preventive action) a loss-prevention dollar is more likely than ever to make a positive impact. &