The KPI “Weeks of Maintenance Backlog” reflects the amount of work that is lined up to be planned, scheduled, and performed by the maintenance department.
Derivation: As usual, the concept is simple — how much work does the department have piled up? But the data still have to be developed and standardized.
To minimize random noise in the KPI, some guidelines need to be in place. For instance, PM and PdM work often hits the system in large chunks, some of which will not be performed for months. These chunks, like shutdown work, could create variations in the total workload that would obscure any progress or slippage generated by the maintenance department. One approach that works well is to limit the work reported in this KPI as follows:
- Only work to be performed by the maintenance group that is reporting is included. That is, work performed by outside contractors who are not usually on site, but who arrive for turnarounds or projects, is not included.
- PM and PdM work is included out to the normal scheduling horizon, usually five to 12 weeks. Assuming that the flow of PM/PdM work is fairly uniform, this moving line will include the work that is currently being planned and scheduled, but not a whole year’s worth of condition monitoring.
- Standing or permanent work orders, if they exist, should be broken into weekly chunks that are entered into the backlog in such a way that they support the logic of the point above.
The unit of measure for this KPI is work weeks. The amount of backlog is expressed as work weeks using the number of man hours of work available and comparing it to the weekly capacity of the maintenance department. While recognizing that the actual available capacity of the maintenance department fluctuates from week to week, the department should select an average capacity to use as a denominator in computing the backlog KPI. This will avoid the creation of additional noise in the data.
Finally, “backlog” is used as meaning available work. It does not matter whether the work is behind schedule or especially urgent. It must simply be part of the work load that the department is expected to plan, schedule and perform.
Maintenance backlog is computed as follows:
Maintenance backlog in weeks = Man Hours of Available Work / Number of Man Hours’ Weekly Capacity in the Maintenance Department.
Significance: More significant than the actual number of weeks’ backlog the department has on the books is the rate and direction of change in the backlog. As maintenance is brought under control, the backlog typically shrinks to the point where it stabilizes at the planning horizon plus a few weeks. Many plants will shoot for a three or four month backlog.
It is clear that a high percentage of emergency and urgent work will disrupt backlog management just as it disrupts any other attempts to manage maintenance. In this way the backlog number provides a supporting reading of how well unscheduled maintenance is being minimized. Backlog also provides a sense of whether maintenance capacity is adequate and whether broad-based programs like painting and light bulb changing should be performed by in-house or contracted help. This kind of long-horizon planning can help to keep the maintenance team well utilized and still large enough to maintain the needed expertise on site.
Getting started: The key data to be created are total maintenance backlog, then the standardized backlog to develop this KPI, and then the capacity of the maintenance department. If all maintenance work is performed on work orders, the gross or total backlog is a straightforward computation. Chances are this list already exists in your planning system. All work orders should have at least rough estimates from the validation process that converts a work request to a work order. It is necessary for determining whether the work should be performed and by whom. A rough financial estimate is needed for the same reasons.
Work requests should be converted into work orders in about a day. A pile of unreviewed work requests can hide potential disasters; it just can’t be allowed. If work order requests aren’t handled promptly, an unofficial review system always emerges, creating double work and shadow manipulation of priorities. The official review should happen at least once a day.
When the total workload is developed, the KPI workload can be developed by trimming off the long-horizon and outsourced work. The result will be the numerator of the KPI ratio.
Similarly, the actual capacity of the maintenance department for each week must be calculated for each trade before planning has any real significance. This calculation should be performed out to the planning horizon and updated weekly. This information is used for loading planned work into the schedules for future weeks. An average should be derived and used for the KPI calculation. The average should be updated no less than annually and no more than quarterly to preserve the quality of the calculation. Of course major changes in plant workload or staffing will require spot changes, but these should be very infrequent. With these two numbers in hand, the backlog KPI can be computed.
Assessment of today’s situation: The weeks of maintenance backlog and changes to the backlog are key indicators of whether the maintenance operation is controlling the maintenance workload or vice versa. Decisions about staffing, scheduling method, and use of contractors are all driven by this information. Backlog information, combined with the rate of completion of scheduled work and a healthy discussion of schedule breakers (issues that prevent scheduled work from happening) provide a quick health check of the maintenance program. Combine these with a discussion of whether emergency and urgent work is being held under 5% of the shop load, and whoever reads the KPIs will have a clear view of who is winning the battle of equipment availability.
Read Stanton McGroarty's monthly column, Management Measures.