PoweredByRx.com

You are viewing a resource from Life Cycle Engineering's Reliability Excellence™ Interactive Model.
If you would like to explore more about Rx, navigate the links at the top of this page.

Eliminate These Obstacles Before You Implement Predictive Maintenance

By R. Keith Mobley, CMRP, Principal SME, Life Cycle Engineering
As appeared in IMPACT e-newsletter

Implementing an effective predictive maintenance program should be relatively simple and straightforward. After all, it’s simply a matter of selecting the right technologies, procuring a suitable system, building an asset database, acquiring a baseline or benchmark data set, and then maintaining the program.  What’s the big deal? Unfortunately, creating an effective predictive maintenance program is a bit more complicated than it appears. This article addresses four of the more serious problems that are routinely faced when a plant attempts to add predictive maintenance to their maintenance program.

1. Is Predictive Maintenance Justified?

The first obstacle that one must face is to determine whether predictive maintenance technologies can add real value to an effective preventive maintenance program.  In most cases, the answer is yes, but predictive maintenance is not a panacea for all applications or for all plants. The first step in this process is to determine whether or not the critical assets that make up your plant lend themselves to predictive maintenance.  

This process requires two essential tasks: a comprehensive criticality analysis that clearly defines the relative importance of each asset to the functionality of the plant; and a simplified failure modes and effects analysis (SFMEA) that defines the specific failure modes of each of these assets.  

Once the specific failure modes of each asset are determined, the next step is to determine whether or not one or more of the predictive technologies can be used to identify these failure modes before they occur. If so, the addition of predictive maintenance may be justifiable. If not, it will provide little real value. Too many plants select one or more of the predictive technologies without any real idea if they can be used effectively on critical plant assets. As a result, these programs provide little value.

2. Selecting the Most Effective Technology(ies)

Selecting the most effective predictive technology or technologies for your plant is the second obstacle. In addition to the steps mentioned in the preceding paragraph, this decision must be based on a detailed cost-benefit analysis. For example, vibration monitoring might be the best technical choice, but the higher recurring cost of data acquisition and analysis might be prohibitive. If so, Ultrasonics or better visual inspection may be a more cost-effective selection. While the latter choices cannot provide the same level of analysis that vibration provides, they can provide the ability to prevent unexpected catastrophic failure of critical system components.

Once you select the technologies, the next hurdle is purchasing the best predictive maintenance system or systems for your application. This process can be difficult because a first-time user may not be able to determine the real strengths and weaknesses of the various systems that are available in the marketplace. All systems seem to provide all of the capabilities needed to support all applications and plants, but unfortunately this is simply not true.  

For example, the price range of fully functional infrared scanning systems is between $9,000 and $60,000 for the instrument and software. The least expensive system provides most of the capabilities of the most expensive and can effectively support most predictive maintenance programs. Most of the vibration monitoring systems range in price from $20,000 to $30,000 and appear to provide universal capabilities. However, the response characteristics, such as low-frequency capability, dynamic range, bandwidth, resolution, narrow-band filtration and conversion of data into useable format will vary greatly.

The predictive maintenance systems must match the unique requirements of each plant and its critical assets. Extreme care must be used in this selection process. Each plant must clearly understand the unique requirements of its critical assets before attempting to select predictive maintenance systems.

3. Lack of Training

The average predictive maintenance analyst receives between five and 15 days of formal training as part of the initial setup of a predictive maintenance program. The initial five days are normally provided by the vendor and are limited to a basic introduction to the specific technology, i.e. vibration, infrared, etc., and how to use the basic functions of the procured system. With this knowledge, the analyst is expected to set up an effective, comprehensive database that will form the basis for the predictive maintenance program. Based on my experience, this level of knowledge is simply too low to provide any hope of setting up an effective database. The majority of programs that have been established in this manner have failed to provide any real benefits and most are discontinued within the first two years.

A fortunate few analysts receive an additional two to three weeks of training in the specific technologies selected for their plant. Third-party companies that specialize in one or more technologies and provide intermediate and advanced training for new analysts generally provide these courses. While some of these courses are good and provide a better technical understanding of predictive technologies, most lack the practical knowledge that new analysts need to achieve maximum benefits in their plants.

4. Use of Predictive Maintenance Information

Perhaps the biggest obstacle to success is that most plants elect not to use the information generated by their predictive maintenance team. For some reason, many plants seem to rely more on the instincts of their maintenance technician than on the scientific data provided by predictive maintenance technologies. Not too long ago, a maintenance manager questioned why a contractor who provided vibration-monitoring services recommended changing more than 500 bearings in a 30-day period. When the reason for these bearing changes was fully investigated, the vibration contractor had recommended changing only five bearings; the rest had been changed because the in-house maintenance technicians felt that they didn’t sound or feel right.  

The other problem is that too many plants place the recommendations of the predictive maintenance program much lower on the priority list than routine preventive tasks, scheduled rebuilds, and other work requests. It seems that the predictive tasks are performed only when there is time left after all other tasks are complete. This doesn’t seem to make much sense, but old habits are difficult to break.

There are many other obstacles that limit the successful application of predictive maintenance, but these are almost universal throughout American industry. Unless they are anticipated and every effort is made to eliminate them, or at least mitigate their impact, there is little chance that real benefits will be derived from the inclusion of predictive maintenance in a plant or facility. The good news is that all of these obstacles can be eliminated with minimal effort and the added benefits will more than offset any incremental cost. 

 

Keith Mobley has earned an international reputation as one of the premier consultants in the fields of plant performance optimization, reliability engineering, predictive maintenance and effective management. He has more than 35 years of direct experience in corporate management, process design and troubleshooting. For the past 16 years, he has helped hundreds of clients worldwide achieve and sustain world-class performance. Keith can be reached at kmobley@LCE.com.

© Life Cycle Engineering, Inc.

Reliability Excellence (Rx) Logo

For More Information

843.744.7110 | info@LCE.com