Telematics in Predictive Maintenance

By Sean M. Gladieux | May 18, 2016

Reprinted with the permission of Equipment Manager magazine, the magazine of the Association of Equipment Management Professionals.

Condition monitoring is the proactive process of evaluating equipment health and application data with the goal of providing recommendations for maintenance, repair, component replacement, and application management that will lower owning and operating costs and improve availability.

Historically, most maintenance and repair strategies have been built around a fixed-interval approach. This includes a fixed interval for preventive maintenance tasks and fixed hour intervals for replacing or rebuilding power train components, such as pumps and motors. However, using the power of telematics data and powerful predictive analytical engines, a condition-monitoring program can be created where maintenance activities and component rebuilds are effectively determined based on exhibited life factors, risk of failure, and statistical confidence.

Specific data inputs for performing condition monitoring include:

  • Electronic machine data
  • Fluid analysis
  • Equipment inspections
  • Equipment history and component tracking
  • Site conditions and application as it applies to condition monitoring

Electronic machine data includes both data downloaded from the machine by technicians and data transmitted via wireless telematics systems. Of course fluids analysis includes engine oil, hydraulic oil, transmission oil, final drive oil, and coolant. Equipment inspections by trained personnel remain a cost-effective way to immediately identify problems and to provide insight for the condition-monitoring program. Equipment service reports and machine history can also provide insight to ongoing issues. The site assessment evaluates equipment application and the work environment including seasonal weather conditions.

Unscheduled Repairs Lead to Big Costs In Mining Operations

Maintenance and repair are a necessary part of every mining operation, and both activities result in machine downtime. Although a good, proactive maintenance strategy can keep equipment running cost-effectively, unscheduled repairs can quickly develop into significant costs and have major negative effects on efficiency of operations.

For starters, the lost production time during an unplanned maintenance event is considerably higher than a planned procedure. It can also cost more if you factor in rush parts orders and overtime paid to technicians just to make sure all the parts and people are on hand to solve the problem. Other costs include towing damaged machines, paying operators who no longer have machines to run, and contamination control activities and additional safety measures that are required during the repair. 

To illustrate the cost difference between planned and unplanned repairs, we reference a Caterpillar customer’s mine in Chile. Using Cat Equipment Care Advisor (ECA), Cat dealer Condition Monitoring personnel detected a potential high-pressure pump issue in a 795F mining truck with nominal payload capacity of 345 short tons. The team alerted the onsite dealer maintenance crew, which addressed the potential issue before a failure occurred—all within a single shift.

Even though the machine was down for only 12 hours, the issue amounted to $60,000 in lost productivity. The part itself cost less than $10,000, and the labor to make the repair cost less than $5,000. All told, this one issue cost the mine about $75,000.

Though the planned repair was costly, an unscheduled repair would have cost 13 times more. The mining operation validated that early detection of the issue saved them nearly $1 million by preventing the part failure.

If the part had failed during production, the impact would have been felt in a number of ways. Fuel would have become contaminated, and over time, the broken pieces of the pump would have shattered and affected other systems, resulting in a total engine failure. The dealer calculated 110 hours of downtime would have been necessary to make that repair, with around $550,000 in lost productivity. Add that to $375,000 for parts and more than $65,000 in labor, and the mine would have lost $990,000. Instead, these costs were avoided.

The next phase of condition monitoring, data analysis, is focused on using all the data sets listed previously and performing aggregation, correlation and analytics, interpretation, and, finally, making a recommendation for repair or intervention. Aggregating all pertinent equipment and application data is the first step. Next is correlating the data to identify meaningful interrelationships and application of powerful analytical engines to identify abnormal deviations. Then a final review is performed by a condition-monitoring analyst who creates a custom, value-based recommendation for repair, maintenance or operational adjustment.

The intent is to harness the power of automation and analytics to reduce the burden of manual reviews and correlation. Proper usage of analytics has been proven to be much more accurate and faster than any expert human analyst can be. A properly designed condition-monitoring program for component rebuilds should be centered around the factors that limit component life and should apply analytics to determine any abnormal signs as early as possible on the P-F curve (below).

The benefits that make these efforts worthwhile are significant: 

  • Improved availability 
  • Improvement in planned versus unplanned maintenance: a measurement of the quality of the detection routines and overall effectiveness of the maintenance strategy
  • Improvement in mean time between shutdown: a measurement of reliability
  • Improvement in mean time to repair: a measurement of repair efficiency
  • Extended planned component replacement intervals
  • Reduced risk and enhanced cost control 
  • Avoiding catastrophic component failures and the associated costs 

Most equipment managers have been involved in some aspects of condition monitoring for a long time. Fluids analysis, plug inspections, formalized machine inspections, site studies, and some electronic data have been available for decades. Now, telematics data and the power of big-data analytics result in more accurate and faster identification of issues.

Predicting optimized maintenance

The capabilities are still evolving that will enable us to use all of this data to move from fixed-interval component rebuilds and maintenance activities to predictive maintenance and rebuilds.

Fixed interval maintenance focuses on:

  • What are the proscribed intervals for maintenance activities such as fluid and filter change-outs?
  • What are the proscribed intervals for component change-outs such as SMUs for an engine or transmission?

Predictive maintenance assesses causes and looks further ahead:

  • Why is this happening?
  • What will happen next and when?
  • What if these trends continue?
  • What actions are needed?
  • What factors for component life can be measured to determine if there is acceptable risk to run a component beyond a fixed interval?
  • What factors can be measured to determine the appropriate interval for a PM?

Despite more complexity, the benefits of predictive capabilities—predictable availability and predictable productivity—promise that the efforts will pay off. Stated in equipment manager’s terms, a predictive system will yield fewer failures and less contingent damage, minimum cost, and planned downtime.

Computer applications can enable analysis of all on-board data and off-board data to identify deviations and anomalies that human analysts cannot detect. The exceptions drive recommendations for scheduled work that averts unscheduled downtime, improves mean time between shutdowns, avoids damage, and reduces rebuild costs. 

Condition monitoring is an essential element of a robust maintenance and repair strategy. Historically, the effort to monitor and respond to critical changes in trends or performance has been done manually and via analysis of multiple non-interrelated data sets. Many OEMs are working to arm their dealer networks with the analytics tools to improve monitoring using telematics data, which will result in predictive maintenance and repairs. We are quickly moving toward true optimization of component rebuild intervals.

--Gladieux is product manager for Cat MineStar Health and Cat Equipment Care Advisor at Caterpillar Global Mining.

x
expand_less