In previous Engineering Case History columns, I have
discussed checklists and their value in avoiding errors during
startups. While the rules cannot replace common sense or a
logical and methodical approach, they can help avoid
embarrassing situations. Here are 20 rules that are most
Never assume anything. The new bearings are in stores,
and they will be there if there is a failure is an
assumption. The bearings may not be in inventory; likewise,
they can be corroded, damaged or, worse, the wrong size.
Follow the data. The shaft failed due to a bending failure,
because the bearing failed, because the oil system failed,
because the maintenance schedule was extended.
This is following the data.
Dont jump to a cause. Most of us want to come up with
the most likely cause for a failure or other situation
immediately. Such causes are usually based on our past
experience, which may not be valid for this particular
Calculation is better than speculation. A simple analysis is
worth more than someone who tries to base the cause on past
Get input from others but realize they could be wrong. Most
individuals want to be helpful and provide input as to the
cause. However, such input may not be credible.
When you have conclusive data adhere to your principles.
Safety issues are a good example. Your position may not be
readily accepted by others because of budget, contract or time
constraints. Before taking a stand, have other senior technical
people agree with you because it could affect your career.
Management doesnt want to hear bad news. Do not just
discuss the failure and the problems it can cause to
management. Present good options that can also be used at other
plant locations to avoid similar failures.
Management doesnt like wish lists. Only present what
is needed, not what you would like to have. Adhering to company
standards or national codes is usually a good approach.
Management doesnt like confusing data. Keep technical
jargon to a minimum and present the information as clear as
possible with illustrations, photos, models and examples.
Management doesnt like expensive solutions. Only
present one or two cost-effective solutions with options, costs
Admit when you are wrong and obtain additional data.
Admitting to being wrong is one of the most difficult acts.
When other data contradicts yours, accepting the truth must be
done; otherwise, you will look foolish.
Understand what results you are seeking. The analysis should
be done to determine why the rotor crackednot to
redesign the machine. Too often, we get so involved in the
analysis that we forget to just solve the problem.
Look for the simplest explanation first. For example, a new
drive belt was installed too tight and then broke the
Look for the least cost and easiest solution. You need to
understand what caused the failure first. For example,
if a drive belt was too tight, then train the machinists on the
correct tightening procedure. Put a placard on the equipment
explaining the procedure clearly and include caution areas.
Analytical results, tests or metallurgical results should
agree. When the metallurgical analysis says it was a fatigue
failure and your analysis says it was a sudden impact, someone
is in error. They should both indicate the same failure
Trust your intuition. When you feel something is wrong but
cant prove it, its time to do an analysis and get
Utilize your trusted colleagues to confirm your approach.
Talking with my engineering and field colleagues has been the
most useful method in finding the true cause of a
Similar failures usually have happened before. It is your
job, as the reliability or maintenance engineer to survey your
company and the literature for the cause of similar failures
and to determine if it is useful data for troubleshooting this
Always have others involved when analyzing high-profile
failures. When safety, legal or major production issues are
involved, it is unwise to make critical decisions on your own.
This is the time for a team approach so that nothing is missed.
Also involve others to develop and implement the final
Someone usually knows the failure cause. It has been my
experience from interviewing engineers, operators, machinists
and technicians that some of the plant staff usually knew the
true cause of a failure. A good interviewing procedure
is an important part of the troubleshooting process.
Dr. Tony Sofronas, P.E., was
worldwide lead mechanical engineer for ExxonMobil
Chemicals before retiring. He now owns Engineered
Products, which provides consulting and engineering
seminars on machinery and pressure vessels. Dr.
Sofronas has authored two engineering books and
numerous technical articles on analytical methods.
Early in his career, he worked for General Electric
and Bendix, and has extensive knowledge of design and
failure analysis for various types of