Condition Based Maintenance and Performance
Author : Manou Hosseini
Maintenance and Reliability Optimization Specialist
Global Management Science Services
Plant Maintenance Resource Center Home
In an article titled Condition Based Maintenance -- What is the condition of Condition Based Maintenance? , Chris Staller expresses the salient view that: "a Condition Monitoring Program cannot become a successful Condition Based Maintenance practice unless the links to the operation are made and the financial metrics determined." He also explains that the Condition Monitoring market has been relatively flat for the past 5 years, hence the current growth rate is expected to be as low as 2-4% per annum. He adds that maintenance programs that were once responsible for cost reductions and productivity improvements are now subject to the same cost cutting. He does not envision these growth rates changing "Unless CM becomes a value-added asset integral to the operation."
In the present article, some performance related issues in the development of CBM strategies and determination of CBM financial metrics have been discussed. That means special attention is paid to the prime question of "how much CM is justified for an operating system?" This includes the analysis and prediction of the impact of CM programs on tangible performance metrics at a global level rather than only on equipment parts.
It should come as no surprise that many CM programs are experiencing a difficulty as a result of cost cuttings. In a discussion of the issue, we may allow for the fact that some organizations- frustrated by many untimely interruptions in their operations, have rushed to some unjustified spending on condition monitoring programs. Yet some negative impacts of extrapolations of the slogans of "excellence era" to the maintenance area might be discussed, a common one being the perception that excellence means being a collector of all good (new) things! Even though in practice, we are now seeing the signs that a paradigm shift to "optimization" is occurring: a collection of all good things is not necessarily optimal!! Is it not interesting to see that some enterprises are now the pioneers in "transitioning from maintenance departments to maintenance systems"? This is the kind of vision that I believe will change maintenance practices to "performance-oriented" systems and in integration with "operation" and "support" systems.
Table 1. Maintenance Process Maturity Levels
The maintenance maturity process, like many other processes, may evolve from an ad hoc level up to an optimized level (Table 1). The maintenance process maturity does not necessarily indicate the sophistication of maintenance tactics and technologies that an organization may deploy. Some organizations may use various kinds of CM tools but seldom measure the performance of their process and associated costs. On the other hand, some organizations may not have any condition based maintenance system in place, but conduct very sophisticated performance measurements. When the performance is measured and tracked, one will concomitantly realize the need for optimization and indeed transition from a maintenance department to a maintenance system! In an optimized process, the optimal amount of CBM (on-line/off-line), planned maintenance, and run-to-failure maintenance will be identified taking into account various performance criteria. It wouldn't be a surprise, for instance, if in some cases the best (optimal) practice (solution) would be a 100% run-to-failure policy without any need for CM and planned maintenance!
It is beyond the scope of this article to discuss the development of performance and dependability models for analysis, assessment and justification of condition based maintenance strategies. Here we limit ourselves to some very basic issues that should be taken into consideration when developing a CBM strategy. The impact of any maintenance initiatives, including condition monitoring, should be predictable and measurable, and indeed somewhere around the performance and dependability of the operating unit. Among the most sensible measures of performance is the production rate or throughput of the unit. However, in performance and dependability analysis, the "random" nature of failures should not be forgotten. Also we should remember that condition monitoring systems, especially fully integrated technologies, themselves, are subject to faults and failures and require care (maintenance!).
It goes without saying that it would be ideal to provide the intended operating system with a condition monitoring system that could detect any abnormal occurrences. However, this not only requires addition of various sensors, which in turn increase the complexity of the entire system, but still many elements of the operating system with on-off failure mode may remain un-monitored- because the prediction (detection) of the most of the failures with on-off type nature is technically very difficult or even impossible. This type of failure occurs without a measurable degradation trend and it does not produce signals that degenerate with time. So we should consider the magnitude of the impact of detectable and non-detectable faults on the overall performance of the operating system. On the other hand, for the detectable faults there should also be a trade-off between on-line monitoring and off-line inspection policies.
Let us talk about the random nature of faults and failures and the concept of prognostics. Due to various factors such as variability in: material, manufacturing, installation, operators and maintainers' skills, etc., no two nominally identical items show exactly the same survival or time to failure. That is why these kinds of variances are being dealt with using probabilistic approaches. The concept is simple. Let us have a look at a more objective example: the "time to repair" of a set of identical items. Even a highly skilled technician can hardly predict an exact duration time to assemble/disassemble two identical pieces of equipment. Rather, a mean duration time and possibly a variance might be suggested. This is the way that probability distribution functions represent data with variances. Due to the random nature of failures, even with the deployment of most sophisticated condition monitoring technologies, there would still be a possibility of failure. Even the aviation industry -with safety as its top priority- has given up the aim of achieving zero crashes.
In a prognostic approach, we ideally desire that the knowledge of the physical behavior of each individual item - provided by condition monitoring - gives the user the power to draw a specific and deterministic (not random) line for the remaining life ("time to failure") of each individual item. In practice, even in this approach, there are various sources of uncertainties because the actual value of the condition being measured may show random fluctuations, the measurements themselves may not be without error, and yet condition-monitoring devices, themselves, are subject to fault and failure. Added to this, is the initial estimation of the fatal limit. As a result, the predicted remaining life may be an uncertain quantity that is best described by a probability function. By the same token, the issue of cost justification applies to prognostics.
It should be mentioned that the argument of the need for performance prediction and financial justification of condition-based maintenance initiatives in production systems does not mean that we are not appreciating tremendous prognostics and CM research & development efforts, substantial accomplishments in CM technologies, and technological developments in the integration of CM and operation. We centered our attention here to the need for performance-oriented maintenance strategies, including a global approach to the performance prediction of production units, before the implementation of condition-based maintenance initiatives.
Copy Right Global Management Science Services
All right reserved
 Chris Staller, Condition Based Maintenance - What is the condition of Condition Based Maintenance? Condition Monitoring Newsletter, January 1999.
 Armstrong World Industries, Inc.
Copyright 1996-2009, The Plant Maintenance Resource Center . All Rights Reserved.
Revised: Thursday, 08-Oct-2015 11:51:47 AEDT