OSCILLATION IN MAINTENANCE PROGRAMS
Ali N. Mashayekhi
System Dynamics Group, E60-357 , MIT
30 Memorial Drive, Cambridge, Massachusetts 02139
Presented at 1996 International System Dynamics Conference
July 1996, Cambridge, Massachusetts, USA.
1. Problem Statement
Maintenance is one the major functions in production activities. It has a high direct cost
and a profound impact on overhead cost through availability of equipment. Maintenance programs
are either reactive or proactive. In reactive programs repairs are made when equipment fails.
Proactive maintenance is a form of preventive or predictive maintenance. Preventive maintenance
(PM) is the regularly scheduled process of performing certain types of maintenance, inspections,
adjustments, and lubrications on equipment prior to failure. While it is being recognized that “the
higher production uptime and product yields more than justify the expense of their preventive-'
maintenance programs”, many plants experience frequent wanes in their preventive maintenance
programs. Some plants have shown oscillatory behavior back and forth between preventive and
teactive maintenance without much leaning about the causes. For example, as one of the managers
of a refinery in Ohio explains:
“A plant example of oscillation is our approach to preventive maintenance. In 1985 Lima Refinery
had a pretty effective PM program. This worked to identify all upcoming failures early enough to
plan repairs, shutdown before a failure event, etc. It worked. But this success had the side effect of
lowering the amount of failures to the point where the inspectors weren’t finding anything much to
repair (this was good), such that management perceived them to “not be busy enough.” The people
responsible for lowering the failures (both salaried and hourly) were given other duties “to fill their
idle time” such that they got away from the work of preventive maintenance and onto more
immediate reactive repairs. This caused failures to increase again, and oscillation happened. Some
anecdotal comments are that long term employees have seen PM programs come and go 4 or 5 times
in their career. They wonder why we didn’t stick with such a good thing.” (Manus P., 1995, p.9)
Such oscillation in the maintenance system is not desirable. This paper presents a model to provide
an explanation for such oscillation.
2. Model
Figure 1 shows the stocks and flows structure of a model of maintenance system that are
relevant and sufficient for the purpose of this paper. Figure 2 shows the reactive decision making
on staff within a major negative feed back of third degree. Figure 3 shows two other negative
loops that operate in a reactive mode of decision making about desired equipment under preventive
maintenance. In the reactive mode, maintenance staff is determined in reaction to percentage of
broken equipment. When percentage of broken equipment is high, maintenance staff is increased to
fix the broken equipment and when the percentage is low, maintenance staff is decreased to save
cost. Also, as shown in Figure 3, in the reactive mode, equipment under preventive maintenance is
set in reaction to staff availability and percentage of broken equipment. When staff availability or
when the percentage of broken equipment is high, desired equipment under preventive maintenance
is increased and vice versa. In the reactive decision making, the three loops shown in Figures 3
and 4 are active and generate an oscillatory behavior discussed in the next Section.
3. Model Behavior and Policy
Figures 4 show the behavior of the system under reactive policies when the three negative
loops discussed before are active. The system oscillates with a periodicity of about eight years.
Ali N. Mashayekhi, Center for Organizational Learning and System Dynamics Group, MIT
Buy
Under proactive policy, desired maintenance staff is set to be able to have all the equipment
under preventive maintenance plan. As long as the total equipment is not changed, the desired
staff will not change either. Then, maintenance staff is adjusted to become equal the desired staff.
Management is not reacting to pressures to cut cost or decrease percentage of broken equipment.
Desired maintenance staff is driven by total equipment. Figure 5 show the behavior of the system
under proactive policy. The result of proactive policy is high uptime with stable and on average
lower maintenance staff. Figures 6 and 7 shows the accumulated staff time and accumulation of
non-operating equipment, as two performance indexes, under reactive and proactive policies
during 40 years of simulation. In terms of both performance indexes, proactive policy results a
better performance.
Preventive Maintenance Completed
PMC
Equipment Under Reactive Maintenance
Change in Equipment Under PM
(CEUPM
EURMP
Equipment Under PM Plan
Maintenance Staff
MSTAFF
Equipment Under Reactive Repair
EURR momen
CMS
‘Change in Maintenance Staff
ETPM
EQUPMR Equipment Taken for PM
Equipment Under Preventive Maintenance Repair
BRERM
Brake Down of Brake Rate of Equipmept Under Reactive Main.
fe) 0.
EQUA
Repair of Equipment Under PM Repair of Equipment Under Reactive Maintenance
Figure 1: Stocks and flows structure of the model.
Equipment Under Reactive Repair EURR,
1(-)
Repro Equip Under RY Maintenance REQURM
Percievd Perventagg of Equipment Broken
Maintenance Staif MAIN STAFF
Figure 2: Reactive decision rules to change maintenance staff based on perceived
broken equipment shapes the major negative loop that creates oscillation.
Ali N. Mashayekhi, Center for Organizational Learning and System Dynamics Group, MIT
Rives
Perceived Percentage of Equipment Broken
Desired Equip Under PM Pen
Figure 3: Reactive decisions on change in equipment under preventive maintenance based on
availability of maintenance staff and the percentage of broken equipment accentuates
oscillatory behavior of the reactive mode.
4. Conclusion
Lack of systems thinking combined with reactiveness causes the preventive maintenance
programs not to last. Under reactiveness, maintenance programs can oscillate between reactive and
preventive maintenance. Since reactive decision making usually lead to short run results, they
become self approving as the causal linkages to the long run consequences are ignored.
Fragmentation does not allow the full consequences of reactive actions to be appreciated. In order
to understand the flaws of reactive decisions, a system perspective is necessary. With systems
thinking, consequences of reactive actions that are far in time and location from the action point can
be better understood. Such understanding facilitates improvement of decision makers mental
models and learning.
Percentage of operating equipment
#8 +: cupMP 2; EURME 3: PEREO. / 4: MSTAFE
: | 120.
4
100.00 ”
40.00
ip
| <
JA
Eqhipment Under Reactive Maintenance
AN
pe Z
Es
0.00 F a
3 0.00 Maintenance Staff
4: 0.00, :
20.00 25100 30/00 35/00 40.00
NX @ ss Graph 2: Page 6 Years 7:30 PM 1/21/96
Figure 4: Behavior of the system under reactive policies.
Ali N. Mashayekhi, Center for Organizational Learning and System Dynamics Group, MIT
363
9 3: cupmP 2: EURMP 3: PEREO 4: MSTAFE
y 120.007
: ee 3 3
3: 100.00 8 AS
4: 40.00
(Under Preventive Maintenance
Lo] Men ener | Percentage of operating-equipment—~
a Equipment
60.00 | ;'
50.00 je ;
30.00 Equipment Under Reactive Maintenance
Maintenance Staff
Y= :
0.00
0.00
0.004 ee
0.00 10.00 20.00 30.00 40.00
XQ 8 Graph 2: Page 7 Years 7:41 PM 1/21/96
Figure 5: Behavior of the system under proactive policy.
al
9 vs808 2 ANoE
ata
Reactive Policies
Se¢
\
poe Proactive Policy
Tobe Be Eas) 8
i Years 1135 PM 1n0R6
Figure 6: Accumulated non-operating equipment under different policies.
s 000
ossine 2 ASTINE
PS
React}ve Policies a
— LO
Pa Proactive Policy
. 00 oer So
a mere van svar oun
Figure 7: Accumulated maintenance staff time under different policies.
References:
- Monus Paul, 1995, Proactive Manufacturing Innovation: attachment, BP Oil Lima Refinery,
Lima, Ohio, USA.
Ali N. Mashayekhi, Center for Organizational Learning and System Dynamics Group, MIT
se4
CC BY-NC-SA 4.0