Scheduled maintenance involves routine repairs, inspections and overhauls that may require disassembly, cleaning
and repair of major sub-systems. It is a labor intensive task. It is expected that scheduled maintenance increases the
reliability and hence the life of an asset. Therefore, the cost of unscheduled maintenance is expected to drop
following the scheduled maintenance. This may not be necessarily the case in practice. Analytics can help determine
the optimal scheduled maintenance intervals and repair scopes during each visit based on analysis of historical repair
events and their effectiveness.
The analytical approaches to scheduled maintenance involve optimizing the time between these visits and the scope
of work during each visit. Due to operational constraints the scheduled maintenance activities are not always carried
out on time. From the field data perspective this presents a great opportunity to analyze the effectiveness of the
scheduled maintenance with respect to the time between scheduled maintenance events.
The idea is that if the longer time intervals do not result in statistically significant increase in field failures or
unscheduled maintenance then there is an opportunity to increase the time between the scheduled maintenance
events and the other way around. This approach also provide for improvements in the repair scope. For example if a
failure rate of a given component increase as you increase the interval, then you can mandate an inspection for that
component during the scheduled maintenance event and replace that part if necessary. Analytics techniques such as
recurrent event models are very useful in determining expected field failures. Also, many of the data mining
techniques including decision trees are useful in identifying the factors that contribute to field failures.
Real Life Examples
Many of the Rail car leasing companies and railroads generally do not spend a significant amount of their
maintenance budget on the scheduled maintenance. The industry as a whole prefers the break-fix model. A railcar is
not a complex asset, it does not have a power source or an engine and this mode of operations seem work.
However, there are opportunities to save by implementing a scheduled maintenance program for expensive
components such as wheels. This requires collecting the trip information to know the usage, historical failures and
application of Analytics to determine wheel life. Break-fix is expensive for wheels especially if the owner does not
have the control of the repair in the field where a 3rd party may attend to the repair and charge back to the owner.
However, this is not the case with locomotives. Locomotives are complex assets and field failures may result in
significant disruption of service and potential safety violations with larger consequences, such as a derailment while
carrying hazardous materials. Therefore the locomotives follow a rigorous scheduled maintenance program based on
the model and the age, etc. Trucks follow a similar scheduled maintenance and inspection program for the same
reasons as locomotives.
The most rigorous scheduled maintenance visits apply to aircrafts for obvious reasons. In general, aircrafts go through four
types of maintenance checks named as types A, B, C, and Heavy Maintenance Visit (HMV). Type A check is
generally conducted every 3 months and may take only few hours and can be performed at an airport hangar. Type B
checks are performed between 3 to 18 months apart and generally performed overnight. Type C checks are
performed every 18 months or so and can take about 3 days to perform. Every 5 years or so, the aircraft undergoes
what is called as a “Heavy Maintenance Visit” (HMV). This is a very elaborate maintenance activity and is mandatory.
For various constraints related with resources, the HVM may sometime get conducted before 5 year period. This
leads to a Maintenance Scheduling Optimization problem, which in airline parlance, is called as “Green Time Burn
Locomotive, Truck and Aircraft engines need to go through overhauls. Engine removal and maintenance is a time
consuming activity. Therefore, repair providers generally need to keep a pool of spare engines to replace the
removed engine. This may apply to other large components such as traction motors and wheel sets on locomotives.
In the field, optimal scheduled maintenance programs can significantly deviate from the ones recommended by the
OEMs. In our experience, significant financial benefits can be associated with optimizing schedule maintenance
intervals between visits and the repair scope during each visit for all three industries considered.
We hope to provide relevant insights and a fresh perspective on how analytics can make a profound impact on asset-intensive organizations. So stay tuned for our next post.