If it ain’t broke don’t fix it. Instead, find the worn out component that may break before it causes a catastrophic failure to the whole machine. Mining’s traditional running-time-based approach to asset maintenance is gradually giving way to a more sophisticated predictive method that relies on algorithms and diagnostic data obtained from equipment sensors to produce real-time warnings about potential breakdowns.
The shift is already contributing to less downtime and improved reliability. But experts say the mining sector lags behind other industries in integrating its data collection systems, thereby rendering much of the data collected virtually useless, and stops companies from really reining in the dreaded downtime that costs them hundreds of millions of dollars each year.
Once upon a time, the concept of long-term reliability
really did not exist. But, unlike their operators, machines are not biological creatures. If they are well-maintained and their parts replaced when worn out, they can theoretically go on for years and years without diminished performance. This rather revolutionary notion
is actually not new, as it comes to us from the engineers who developed the process called reliability-centred maintenance (RCM) for the aviation
industry in the 1950s and 1960s.
After World War II, the commercial air travel industry was ripe for expansion. But crash rates were troublingly high and very often caused by equipment
failure. These were believed to be due to equipment age and so time- based maintenance was ramped up. To everyone’s dismay, crashes actually increased.
The industry was frantic to fix the problem. After carefully analyzing the crash data, aviation engineers quickly discovered that, in fact, only 11
per cent of airline equipment failures were due to age. Astonishingly, the highest proportion of failures actually occurred in new or recently
repaired equipment. They realized they needed to meticulously tease out each underlying factor behind the failures in order to prevent new ones. Thus RCM was born. Its first success was the development of the Boeing 747 in the early 70s.
Maintenance then and now
These principles inform most of the predictive maintenance programs used in today’s mining companies. But most operations are still using a combination of running- time-based and monitoring-based maintenance.
Traditionally, maintenance has been performed either when a breakdown occurs or scheduled after a certain
number of running hours. However, as the aviation industry so starkly demonstrated, time-based maintenance is not
necessarily the way to go. In addition, this requires having
replacement equipment for use when broken equipment is in the shop — a huge investment.
Predictive maintenance programs help reduce these problems. “You don’t necessarily have the truck come in for traditional standard maintenance,” explains
Malcolm Scoble, professor of mining engineering at the University of British Columbia. “You can call it in when the part is reliably observed to be
approaching failure, scheduled at a sensible and convenient time.” This is done using a combination of wear-and-tear data coming from sensors and
operator inspections. Predictive models based on that same data can tell you when to fix a subcomponent. This takes less time, fewer parts and costs less than pulling the machine out of service for a large component repair, such as an engine when it is not warranted.
There are a variety of sources of data for reliability mon- itoring. The principal of these are sensors located on the machines. Most heavy machinery used in mining now comes hardwired with a number of sensors to monitor things like vibrations and engine temperature. Some
have sensors that keep an eye on air filters; others monitor the oil and tires. “We’re offering vibration sensors as standard,” says Joe Helfrich, senior
vice-president of surface mining products at equipment manufacturer Bucyrus. He adds that the cost per sensor has gone down enormously in the last few
Operator inspections are also key. At many companies, operators still enter their inspection data on a paper check sheet, which is filed away after- ward.
There are higher tech alternatives, such as asset performance management company Ivara’s integrated system, which has operators using handheld devices
instead of paper sheets to enter their inspection data. This is then transmitted wirelessly to the centralized system where it is crunched with the rest of
the data to produce recommendations on when repairs should take place. But users of this sort of approach are still in the minority.