March 13, 2007
The Random Element in Reliability
Director of Training Services
SUCCESS by DESIGN
This is the first article of a two part series. The series is about the random element in reliability and the expectations we have in regards to reliability. I will discuss a situation I recently had to deal with, in regards to my car’s reliability. I know what a shock. The first part will talk about the mechanical reliability of the vehicle and the expectations of reliability. The second part of the series will cover the reliability of the maintenance/service people and their training.
I would first like to explain the random element in reliability. Reliability is basically a system’s ability to perform what you want, when you want, and to what level you expect. I used the word expect because as a user we have certain expectations of the item we are using which may differ from the original purpose of the item. In any case, the item will fail to meet expectations from time-to-time leaving users unhappy. No matter what maintenance we perform on an item we will still incur failures. Some of the failures are preventable and some are not. Random failures are, most of the time, in the not-preventable category or they cost so much to prevent it just is not feasible. The general method dealing with these kinds of failures is to let them run to failure.
I recently was on a belated Valentines trip to Disney World with my wonderful bride of almost eleven years. Well, on the way home my car decided to fail. And I mean fail in a huge way. The engine started to knock. Car trouble is never a good sign on a Sunday afternoon, over two hours from home, with a good family friend watching the kids. We ended up having to have the car towed and renting another car to get home. Needless to say the car did not meet our expectations.
Our car (with brand not being mentioned) is actually in the latest Consumer Reports April 2007 issue as one of the more reliable used vehicles in its price range. The engine model in the car has been around forever and most mechanics and other people with, good knowledge about cars, claim it is nearly bullet proof. As a matter of fact I have never seen so many jaws hitting the floor when I have told people about the failure. My car was six years old and had 62,000 miles on it and all the recommended maintenance has been done to date. The car was in excellent condition and ran smooth. The mechanic at the dealership confirmed that it was a rod knock but couldn’t give any extra information because they don’t open the engines anymore. I will discuss that in the next part of the series.
The failure seems to be a random failure due to the many sources of feedback I have received. Even though I was never able to get an answer on the specific cause of the failure there are some conclusions I can make about the situation. First, I can make the point that the maintenance should have been adequate because it is based on usage or time for average conditions. I was driving a mixture of types of traffic but mostly at cruising speeds with a smaller percentage of stop-and-go. I think I fit into the average driver category. Automobile maintenance today captures both sides by using mileage and a calendar base as a fall back to capture the age related issues for average conditions.
Related to automobile maintenance, plant equipment in many facilities is maintained on an operating hour or calendar basis. This is often based on average usage for a set period of time. Actual run time can be much more accurate because it compensates for slower and heavier use times. So we should expect some failures to occur that are unexpected. The random failures really are a product of some variation or groups of variations in products, people, or environment that are so far beyond the scope possible methods of feasible elimination that we just have to accept them. Perfection is not an option.
The second conclusion I can make is that there probably was some way of predicting the failure by using oil analysis, vibration analysis, or some other technology but probably not worth the cost. The use of one of these technologies would be beyond overkill in this application and stresses the point of feasibility in maintenance. The cost of the replacement of my engine after everything including rental is around $4000. The use of a technology in this situation might have cut my end bill in half but still would have cost the same in the life of the vehicle to date. I do understand that I have not incurred the cost of lost operation in this case but you can see the point that is being made. The cost of oil analysis or some other technology can sometimes outweigh the failure that you are trying to predict. Just because you can do the task does not mean you always should. I have seen a lot of money wasted on performing predictive maintenance arbitrarily without any real financial basis to the decision.
The third conclusion is that there is no relation to an age specific failure giving me no option to do an age based replacement of the engine. I could have replaced the engine at 50,000 miles and might have still incurred the problem or possibly inserted a whole new failure. The original engine might have lasted to 150,000 miles or more but I would never know that having removed it that early. Random failures occur at any point in the life of an item and therefore you get no benefit from age based replacements.
The final conclusion is that this failure might not have been predictable at all. The failure could have been caused by some manufacturing defect or a foreign object that might not have been there long enough to detect using a predictive technology. The problem could have been caused by a single piece of gasket that broke off and plugged an oil passage in the rod bearing. I am no expert on engines and am only making a point that something small and insignificant can sometimes cause huge problems unexpectedly.
The expectation in any maintenance program should be that no matter how much preventive, proactive, and predictive maintenance is performed that there will still be failures that occur randomly and that are just not feasible to address. The only real thing you can do is make sure you have a spare or some plan to avoid the consequences of the failures. Part 2 in the series will cover the reliability of the maintenance/service people and their training.
Robert Apelgren is Director of Training Services for SUCCESS by DESIGN. He received his BS in Industrial Technology from Roger Williams University and an MBA from the University of Phoenix. Mr. Apelgren is a Certified Maintenance and Reliability Professional (CMRP) and a member of the Society for Maintenance and Reliability Professionals (SMRP) Best Practices and Standards Committees. He is also certified NAVAIR RCM Level III and Instructor, and NAVSEA RCM Level II. In addition, he has 15 years of maintenance experience as a technician, supervisor, coordinator, consultant, and instructor. Mr. Apelgren has provided instruction in various RCM methodologies and has presented papers at various conferences. Mr. Apelgren can be contacted via email at . http://www.motordoc.net.
