When leadership makes the decision to defer maintenance it is usually made on
the belief that the problem being deferred will follow the logical condition
path toward failure of:
Good > Not So Good > Worse > A
Problem > A Real Problem > Bad > Breakdown
linear path allows the decision-maker the time to define
the maintenance need, measure the seriousness of the problem,
analyze the options, choose an optimum solution of improvement,
initiate the maintenance needed, and control the machine in the
future to prevent a recurrence. There is the belief that the progression will
allow continued production and have adequate time to recognize the final stages
of failure and act to prevent a breakdown. This
deterministic and scientific method of management that works so well in
every other linear aspect of business will not work in the maintenance arena
The Function of Maintenance is Nonlinear
Nonlinear functions can but do not have
to follow the straight line failure progression mentioned above. Nonlinear
functions can flip from performing to nonperforming without following a logical
failure progression. Nonlinear functions can jump from Not So Good to
Breakdown without going through all the middle steps.
Scientific methods of management cannot measure and predict nonlinear
maintenance breakdown events because of the number of variables that cannot be
controlled by a maintenance manager are almost infinite. It is impossible to
quantify just ten of these variables such as (1) engineering design, (2)
manufacturing quality, (3) replacement parts quality and availability, (4)
unpredictable weather, (5) operational support for preventive maintenance, (6)
the quality of repair mechanics and technicians, (7) management commitment to
predictive and preemptive maintenance, (8) the quality of energy (Electricity or
fuel) used, (9) the operating conditions, and (10) the training level and
commitment of the asset operator .
Therefore, if it is impossible to predict any of
these variables then it is
impossible to predict the sum of these variables. This is Hard Science.
infinite number of possible initial conditions that will define the destiny of a
machine toward a breakdown event simply cannot be predicted no matter how many
ACCURATE data points are measured. This is what defines the problems associated
with dynamic and nonlinear systems. The hard science of this statement can be
verified in any major university by contacting their Department of Nonlinear
Studies. It is being said that the 20th Century will be remembered
for relativity, quantum mechanics, and the
is important to understand that the study of chaos (Dynamic and nonlinear
systems) is the study of finding hidden order in apparently chaotic systems such
as turbulence, weather, biology, and the performance of machines.
Dr. Edward Norton Lorenz of MIT is considered to be the father of the
Chaos Theory due to his discoveries in weather prediction. In the early days of
computing, his nonlinear weather model formulas with only 10 feed-back variables
proved that small changes in
initial conditions can amplify over time in such a way that long-range
weather prediction would be impossible.
work proved the sensitive dependency on initial conditions was so powerful that
the very smallest of errors in measurement (Thousandths of a unit) of the
initial starting condition of any of the variables could result in very large
differences in the later state. This sensitivity to initial conditions was later
The Butterfly Effect. The mathematical engine that drives his discovery
is a strange attractor called
The Lorenz Attractor.
The Lorenz Attractor describes a system that
never succeeds but never fails.
It flails along between an apparent upper and lower boundary
but the frequency and amplitude are unpredictable.
I have never heard a better definition of a
maintenance budget better than that!
The basic chart above is from
Chaos by James Gleick. The color annotations are by me.
With just the smallest of differences in initial conditions (By the flap of a
butterfly’s wings.) in a nonlinear feedback formula, a system can flip between
performing and nonperforming without having to go through the linear sequence of
Good > Not So Good > Worse > A Problem > A Real Problem > Bad > Breakdown.
One machine might offer advanced warning of a failure over a long time and there
is plenty of time to defer maintenance and choose an optimum solution for
competitive bids and continued production without a breakdown event. However, an
identical machine sitting beside it under the exact same operating conditions
might breakdown with little or no warning at all and when this happens the
True Risk/Reward Ratio for Deferred Maintenance may exceed 60:1 in
repair dollars and 15:1 in downtime compared to an early intervention
maintenance event. The cost associated with the breakdown event will be the
square of the primary failure part and will square again with each
cascading level of failure.
if you come to believe that the function of maintenance is nonlinear and
unpredictable, how can you as an leader use this information to better support
the operational needs of your organization?
When a machine is known to need repairs,
do not expect it to follow the logical and linear sequence to failure. There is
no guarantee of time to resolve the issue before failure.
When a machine is known to need repairs,
do not demand or expect a prediction of failure by your maintenance manager
because this is not possible. The variables are too infinite to calculate.
Approximations will be all that can offered as to time of failure, downtime, or
Remember, nothing you will spend during
an early intervention will be as much as a breakdown event. Offer your
maintenance manager all the direct support needed in funding and downtime to
create an early intervention event to repair the machine as soon as practical.
The less time between detection and early intervention the less likely to incur
the 60-times penalty in money and 15-times more in downtime.
Understand that machines cannot be
motivated by human standards. Money and downtime must be made available when the
machine demands it, not when the budget demands it if you want to avoid
exponentially escalating breakdown costs.
Just because a maintenance manager can’t predict future maintenance costs
doesn’t mean they are bad managers. The problem is simply unpredictable. If
allowed early intervention, they can produce the lowest maintenance cost per
unit of production possible even though they cannot tell you what that cost will
David Geaslin is a graduate of The University of Texas at Austin with degrees in
Industrial Management & Marketing; a former Marine Corps Aviator and Aircraft
Maintenance Officer (1968-1975); the CEO of his maintenance service company for
15 years; and has consulted offering coaching and seminars in the management of
maintenance since 1990. He lives in Gonzales, TX and travels offering his
services wherever needed.
Houston Mobile: (832)