Reliability Centered Maintenance (RCM)…
… is the world’s leading method for identifying maintenance and other activities required to sustain reliable performance of physical assets. If you want a proactive maintenance program that really works, then Reliability Centered Maintenance is the most thorough approach you can take to get there.Since the 1970’s it has been responsible for huge improvements in airline flight safety – crash rates today are 1 / 120th of what they were before RCM was introduced. Elsewhere, particularly in critical applications like nuclear power, it has produced similarly impressive improvements in reliability. That leads to more uptime, in turn resulting in greater production and revenues, lower costs, improved safety and environmental compliance.
RCM was initially created by Stan Nowlan and Howard heap when they published their studies in a paper entitled, “Reliability Centered Maintenance” in 1978. Since then a number of variations have arisen, John Moubray’s RCM2 is perhaps the best known method. John “industrialized” Nowlan and Heap’s work. Other methods sprung up and the standard SAE JA-1011 was created to clear up confusion. Today, our own version of RCM is “Reliability Centered Maintenace – Re-Engineered“, RCM-R®.
RCM is a team-based method that combines the knowledge and expertise of maintainers, engineers and operators. The team (usually 3 to 5 people) analyses one system (not necessarily a piece of equipment) at a time. Depending on complexity of the system (not size) it can take anywhere from 10 to 15 meetings (each 3 hours long). In those meetings a trained facilitator asks questions about the system and gets answers from his team of experts. During that process they make decisions on future actions that are both technically sound and “worth it” from the perspectives of cost and risk.
The seven questions as required by the SAE standard are:
- What are the functions of the asset and their desired levels of performance?
- What are the failed states associated with each function (i.e.: functional failures)?
- What are the failure modes that lead to those failed states?
- What are the effects of those failure modes? (i.e.: what happens if we do nothing to prevent them from occurring?)
- How do those failure modes matter? (i.e.: safety, hidden, operational, non-operational?)
- Can we do anything proactive to predict or prevent the failure?
- If not, then what can we do? (i.e.: run-to-failure, failure finding, re-design or some other one-time change?)
On the surface, those questions appear easy enough to answer but there is a great deal behind each of them. For instance, the very first question asks about functions. The functions are the things the system does for us. They are NOT merely a description of what the system “is”. There is usually a primary function (i.e.: the main reason we have the system) as well as a number (sometimes many) secondary functions (e.g.: environmental, safety, control, containment, appearance, protective, economy, efficiency and sometimes even superfluous functions we really don’t want). All functions must be listed and then standards of performance associated with them. The standards define what we “want” from the system, not what it is capable of doing, so engineering specs are often not particularly helpful at this stage except perhaps as a sanity check to ensure we are not asking more of the system than it was designed for.
Many technical people are so used to dealing with “what” a system is that they forget what it does. Here we are interested in what it does and much less in what it is. For instance a single function like, “to contain the pumped liquid” implies an absolute standard (i.e.: it cannot leak at all) and it applies to all components that have a containment function (e.g.: housings, pipes, gaskets, valves, seals, etc.). It is not necessary to list all the components at all. Operations people grasp this right away because it fits so well with their perceptions, but technical people often struggle with this different perspective.
Likewise, each of the RCM questions is “loaded” – they cannot be answered easily without truly understanding what is behind the question. I’ve seen many attempts to read a book and then perform RCM that have turned out poorly.
The first four questions are the equivalent of performing a Failure Modes and Effects Analysis (FMEA), focused on functions, not components. The last three questions take the user through a decision logic (a diagram is used for this) to help arrive at logical and worth-while decisions about what actions can be taken. The diagram is very efficiently structured so that the most likely answers are reached early. It also ensures that not only does the team arrive at technically sound decisions (i.e.: prevent only what is preventable, etc.) but also decisions that are worth implementing. It requires a brief analysis of risk or costs, depending on the nature of the consequences from each failure mode. If risk is reduced, or costs are lowered, then the decision is “worth it” to implement. If not, then the team continues to look at technical options until they are exhausted and it must accept run-to-failure or some sort of redesign (or other one-time change) as a default solution.
Once the analysis is completed its decisions must be put into practice. The various tasks, task frequencies and assignments of who does the work must be entered into the maintenance management work order system for automated triggering and issue of work orders. For decisions about technical redesigns the recommendations must be sent to engineering for further development work. If the decisions require training, rewrites of procedures or some other one-time change, those too must be carried out by the appropriate responsible authority. If there is one aspect of RCM that leads to failed attempts, it is this – lack of follow up on the decisions made.
Doing an RCM analysis requires collaboration among maintenance, operations and engineering departments. Getting the outcomes implemented requires those three as a minimum and often involves training, human resources, finance and possibly other departments. It is a “whole organization” process, not, as the name suggests, a “maintenance only” initiative. In fact, if it is approached with only maintenance it will not work.
The diagram below shows a brief outline of a process for introducing RCM to a site:
Assuming a decision is made to continue after the concept is proven, here is what a typical implementation plan looks like:
Anyone participating in the analyses needs to have the basic analyst training (a 2 day course using RCM-R®). It is wise to carry out a few pilot projects early in the implementation to gain experience and confidence in the method. Initially you will want an experienced facilitator to lead those analyses. That will prove the concept, set the standard that should be expected of your own facilitators once they are trained, and deliver useful results very quickly.
Facilitators need additional training. We offer a 3-day facilitation skills course for that purpose. That qualifies your facilitators, but they still need experience. Mentoring follows – 3 analyses – one in which we actively co-facilitate, one in which we are there in support, and a final analysis where we are absent and check the results. It’s a bit like learning to fly a plane: ground school, initial flight training while instructor holds a dual control, more flying where the instructor has hands off the controls, and finally a solo.
RCM-R® includes some mathematics – most of it is fairly easy, but we find that many analysts are not comfortable with it. The Facilitator needs to know how to use the basic formulas and when to use the more complex ones. We also offer reliability training so that you can have someone on staff with formal exposure to the math who can help with those more complex situations.
Our consultants are highly trained and experienced RCM “experts” who teach the courses and mentors newly trained facilitators. Companies who have used our services have all demonstrated excellent results from their RCM-R® work.
I would caution you not to cheap out. Once you have had the training, carried out pilot projects and begun to analyze your own systems you are ready to go solo without the outside help. From that point onwards the only help you are likely to need is to deliver additional training. Using untrained analysts on your review teams is a mistake – it slows down the analysis, and if they have strong personalities and opinions, they can sway the analysis off course. Don’t expect a good facilitator who is not highly familiar with RCM will be able to lead your analysis meetings. While the basic facilitation skills are the same, they can be easily side-tracked if they are not familiar with RCM-R®’s nuances.