TPM and Part Replacement Schedules

On the Lean Enterprise Institute website, a reader asked the following question:

“My management has hired a TPM consultant who makes us systematically replace certain parts in our equipment even though they’re working fine. This seems needlessly costly. What do you think?”

Over the years, “TPM” has become an umbrella term for all improvement activities in process industries, and not just maintenance. In this question, however, it is used in its original sense of “Total Productive Maintenance,” meaning involvement of all employees in the maintenance of facilities and equipment to support production. There is a body of knowledge associated with it, in which I don’t recall seeing anything about deciding when equipment parts should be replaced. Generally, TPM tells you how maintenance work should be done, not what it consists of.

TPM’s first step is Autonomous Maintenance, which delegates routine checks and small maintenance activities to production operators. There are many other, higher levels, but Autonomous Maintenance is the only one I have ever seen implemented, to the point that TPM is often equated with Autonomous Maintenance. Besides the scheduling of part replacements, there are many other aspects of Maintenance that I don’t believe TPM addresses, but that you have to in a Lean implementation, such as the role, structure, and size of the Maintenance department.

On these issues, I have found that you are more likely to find answers from industries where maintenance plays a more central role than in Manufacturing, such as commercial or military aviation, or nuclear power. On part replacement in particular, seminal work was conducted 45 years ago at United Airlines when the Boeing 747 was first released. United’s maintenance experts realized that the replacement schedules they had previously used on the 707 could not be economically carried over to the much larger 747, and they undertook a systematic analysis of the plane’s components that led to the development of a theory now known as “Reliability Centered Maintenance,” or RCM.

One discovery they made was that the “bathtub curve” of reliability theory textbooks only applied to 4% of the 747 components. According to that theory, a component is subject to “infant mortality” when new, wear-out when old, and have a “useful life” phase in-between, during which they have a low and constant failure rate.  It was observed on vacuum tubes in the 1950s, and assumed to apply to everything, with consequences on maintenance and part replacement policies. Obviously, you would want to monitor parts closely when new and replace them just before wear-out kicks-in.

What the United people found was the parts exhibited instead a variety of patterns and that some, in particular, never had a wear-out phase. As a consequence, there was no point in systematically replacing them after a fixed interval or use count.

The consequences of a component failure on an aircraft in flight also varied greatly depending on whether it is a passenger reading light, an avionic system, or the rudder. You don’t need the reading light to stay in the air and you can’t replace the rudder in flight, but you can have a standby avionic system take over. This  Failure Mode Effect Analysis (FMEA) served as the basis for targeted redundancies.

The FMEA concept is known in manufacturing, but I have never seen it applied to production equipment. Targeted redundancies are used, for example, in machining centers by placing the same frequently used cutting tools in two pockets, with the second tool automatically taking over when the first is worn out.

The equipment supplier can provide generic recommendations, but they may not match your specific application.  If you want to improve your equipment part replacement policies, you will need to collect and analyze technical data on the behavior of your machines, on your shop floor. With today’s sensors, data acquisition and control systems, it is technically feasible. If United Airlines could do it in 1969, you can in 2014. What is most missing is  analytical capability. Today’s Computerized Maintenance Management Systems (CMMS) are still focused work order administration, not the technical analysis of equipment behavior.

Once you have worked out appropriate part replacement policies, you need to work out the logistics of making spare parts available when needed, which is a whole other topic.