Planning and Scheduling / Operational Excellence

How planners estimate labor hours

Trying to be super-accurate with each job estimate? Doc Palmer says that's not how this works.

By Doc Palmer, PE, CMRP, Richard Palmer and Associates

This month we will talk about how planners estimate labor hours for job plans. Time estimates are a major part of the planning process, yet surprisingly, planners do not need to create super-accurate estimates. Speed of estimating and the general accuracy are more important than the precise accuracy of each estimate. Properly executing this concept makes a huge difference in planning and scheduling success.

To begin with, planning is all about running a Deming cycle in maintenance. Planners give head starts and craftspersons give feedback. Planners are craft historians who save and use the feedback to make future jobs better. (Planners cannot provide perfect job plans.) However, two issues keep planners from planning all of the work. One is this issue of making time estimates, and the other is an issue of how much detail to put into a job plan (which we will discuss next month).

Some people advocate “engineered” time estimates when planning work. These time estimates might consider such items such as how long it should take to remove each bolt over a certain size and the number of bolts on the job. It might take into account that a person walking at a determined pace walks 3 miles per hour and the distance of the job from the shop. It might also consider adding a percentage of extra time for jobs per foot of elevation for high jobs or per degree of temperature for being unusually hot or cold. 

I think these type of estimates take too long for planners to calculate and might be more appropriate for assembly-line-type work where jobs are repeated time and time again each day. However, maintenance tasks usually have far fewer repetitions: usually once or twice per year on a specific asset, and often by a different craftsperson each time.

Some people advocate averaging historical times, but again, maintenance tasks usually have a low frequency of repetition. Viewing history also shows wide variations for specific jobs on the same asset by different persons over the years. A typical spread of actual hours might look like: 2 hours, 4 hours, 5 hours, 5 hours, 5 hours, 7 hours, 8 hours, and 16 hours.

How should a planner use this information? An average time would be 6.5 hours; a median would be 5 hours; and a mode would be 5 hours. An average throwing out the outliers of 2 and 16 would be 6 hours. Which number would make a good time estimate?

Computer programs might automatically make these calculations from past jobs, but considering that typical wrench time for good maintenance forces without proper scheduling is only 35% instead of best practice 55%, we might ask whether all of these times are too high.

Instead, planners need simply make “ballpark” estimates for labor hours for maintenance job plans. They can quickly make these “guesses” from their own experience of how long they think a job should take (or from looking at some history or talking with some craftspersons) to make reasonable estimates.

The planner makes a simple, quick judgment of how long the job should take without unusual delays and with a craftsperson who is generally qualified for that work. Being quick is a critical part of the planning process. Trying to make “perfect” time estimates that consume too much planner time hinders running the Deming cycle. It is better that planners plan all the jobs than to become bogged down on a few jobs. In addition, these ballpark estimates are often very accurate in general; although individual estimates have a wide variation of accuracy, they have a very normal distribution. That is, even though an individual time estimate might be off quite a bit, the average of all the estimates from their actuals is probably quite close.

An example might be that a planner estimates a job taking 5 hours and the actual time is 2 hours; another job estimated at 3 hours takes 6 hours; and a job estimated at 2 hours takes only 1 hour. Two of the jobs are off by 100%! Nevertheless, the average of all the work estimates at 10 hours compares very closely with the actual average of the actuals at 9 hours.

The lesson from this example is that although individual time estimates are not that accurate, they are very useful to help in generally assigning work and creating schedules. For a week, a 10-person crew receiving 400 hours of estimated work really does receive about 400 hours of work.

How planners should come up with time estimates and their accuracy is certainly a controversial area of planning. However, remember that the purpose of planning and scheduling is not necessarily to be creating perfect plans and perfect schedules. The purpose of planning and scheduling is to help us complete better work and more work.

Getting all of the work through the hands of a planner before execution helps us complete better work because we are running the Deming cycle to avoid past mistakes. It also supports scheduling that helps us complete more work.

Quickly estimating job labor hours supports running this improvement cycle and scheduling. Quick estimates based on planner expertise are often good enough to prevent planners from getting bogged down and not planning enough of the work.