Part 1: Q&A on reliability-centered maintenance

Doug Plucknette, president of Reliability Solutions, Inc., discussed in a recent webinar how focusing on asset life cycle can bring about long-term change and improve reliability at a plant.

Several attendees submitted questions to Plucknette during the webinar, and his answers can be found below.

What are some effective predictive maintenance procedures that you recommend to improve the reliability of equipment?

Plucknette: The most important aspect of any predictive maintenance (PdM) program is to get people to understand that once point P has been detected, the item will fail, and replacing the item sooner rather than later reduces secondary damage and cost.

Do you have a preference between using a third party to perform PdM/condition-based maintenance or training people in-house?

Plucknette: Yes, I am a person who believes every company should train their own people, and the people who make up their condition monitoring group should be selected from their most respected tradespeople. This is simply my opinion. I believe those trained from within are more likely to be taken seriously than a third-party person whose only stake in the business is continued work.

What is your view of prescriptive maintenance?

Plucknette: We like to invent new terms in our community, and nearly every time someone introduces a new term, it comes from their lack of understanding of what maintenance is and how it should be performed. Prescriptive maintenance is one of these terms.

The reality is that “prescriptive maintenance” has been around for decades. It’s the maintenance plan or strategy that results from performing a thorough RCM (reliability-centered maintenance) Analysis on your equipment and is based on the failure modes that result from the context and environment in which you operate your equipment.

With the introduction of smart machines, machine learning, the Industrial Internet of Things (IIoT), I hope we can expect the introduction of more process verification techniques that will allow us to detect potential failures earlier and use that information to better understand the causes or failure modes that result in function failure of our assets.

Do you have any tips to get buy-in from plant leadership who think that reactive maintenance is the norm in the industry?

Plucknette: The first tip I would give a manager who thinks reactive maintenance is the norm is to find a new career. The business case for reducing reactive maintenance is simple. Use condition monitoring to detect failures early, plan, and schedule to replace these items before secondary damage occurs. Use RCM to develop a maintenance strategy that eliminates failure modes that don’t need to occur.

What is your view on machine learning and artificial intelligence and how they will change things with RCM?

Plucknette: This is a very popular question as of late, but it shows a lack of understanding of why RCM works. I recently spent three full days with a company who touts themselves as a leader in this field. We looked at a pilot they had set up where almost everything that rotated or moved had a sensor to collect data (vibration, temperature, amp draw, noise, timing, counts, run time, idle time, and downtime). The claim was, “with all the data we are collecting, we will soon know the failure rate for every component and part associated with this system. One will no longer need to use RCM to develop a maintenance strategy, the machine will tell us what it needs to be.”

My reply was simple and to the point:

Will the machine tell you when the pump bearing failed in the middle of the night and the correct bearing wasn’t available and that a substitute that fit was installed and that the new bearing may not last 1/10 of the time the original bearing lasted? Will the machine tell you that when the vibration data shows the bearing to be in alarm, you only have less than a day to replace it?

Will the machine tell you when someone adds the incorrect lubricant by mistake?

Will the machine shut itself down to prevent secondary damage when something goes into alarm? Or will it know when someone installs the cover on a junction box incorrectly, that the wiring is about to short? Or that you’re not maintaining your compressed air properly and your valves and instruments fail at a rate that is three times quicker than they should be?

With the smart machine crunching all this data and spitting out alarms and work orders at a rate that by far exceeds the capability of your workforce to complete them, will it fix the problems for you?

The idea of the IoT and smart machines excites me, but for some reason, very few of us can put into perspective the reality of a system that could quickly overwhelm an organization with work orders. It again shows how little most people know about RCM and how the tool works. RCM addresses and mitigates the failure modes that result from the context in which you operate your equipment and the environment in which it operates.

All the other failure modes—the engineered failure modes that smart machines were designed to address—have been known for years. They are available in failure mode libraries, and people including myself have mapped them to known technologies to detect them and mitigate secondary damage. It still comes down to: when we detect the potential failure, do you have the discipline to plan/schedule and replace the item before it fails? Or will you ignore the data and run the item to failure?

The truth is, as far as RCM is concerned, nothing has changed. We simply have more data to consider and with which we can build a business case for making better decisions.


Be sure to sign up for our next webinar! Register for free here.

Similar Posts