Wrench time causes controversy for a number of reasons. The concept of wrench time explains why there is a significant opportunity for maintenance to increase its productivity. However, a plant does not need to measure it. In fact, I would argue that a plant should probably not measure it. Wrench time can be a terrible measure for a number of reasons. Nevertheless, if a plant does measure wrench time, there are proper and improper ways to do it.
In January we discussed the great value of maintenance planning in terms of wrench time (“Plan-it Fitness”). Wrench time is essentially the percentage of time crafts spend actually working without delays out of the total time available. Improving wrench time from a typical 35% to a best-practice 55% yields a 57% improvement in productivity, with all of the extra work being proactive. A plant completing 1,000 work orders per month could be completing 1,570 work orders per month with the same workforce, or 570 extra proactive work orders per month for free. Planning and scheduling answers the question: How can we complete more proactive work to head off failures when we have our hands full of reactive work?
Nevertheless, a plant does not need to measure its wrench time if it can accept that without proper planning and scheduling, maintenance productivity is probably at 35% and that proper planning and scheduling will improve it to 55%. Furthermore, simply measuring wrench time will not improve it. Proper planning and scheduling will improve it. A plant starting to plan and schedule properly should see its work-order completion rate rise – generally a step change of about 57% – with most of the new work that’s completed being proactive. So why measure wrench time when what you actually wanted was a higher rate of work-order completions anyway? The concept of wrench time simply explains why proper planning and scheduling helps increase work-order completions.
There are additional reasons not to measure wrench time. For one, the act of measuring can make people nervous and start unwarranted rumors about staffing levels. For another, wrench time can be a terribly misleading measure. A carpenter could show up at the wrong house and hammer slowly all day without taking a break or needing more nails. The carpenter might thus have 100% wrench time doing the wrong job. Yet we presume that for the time available to work, management generally directs crafts to the right work, and craftspersons generally work efficiently.
Wrench time has its merits as a metric, but it is not the perfect KPI.
Still, in spite of the risk of upsetting its workforce and the effort involved, a plant may decide it wants to measure its maintenance wrench time. A plant might want to convince itself that it is a typical plant, recording 35% wrench time, or gauge whether it has an effective planning and scheduling program. Note that wrench time really measures the effectiveness of planning and scheduling rather than the willingness of craftspersons to complete tasks efficiently.
Not all methods to measure wrench time are valid. One type of study, using work and delay times self-reported by the craftspersons, is not valid. Most such results will be in the 70% or higher range. It is difficult for craftspersons to recognize that a delay moment here or there is not actually “work.” In addition, if it is difficult for management to recognize that typical wrench time is only 35%, how can one expect craftspersons to understand that 35% is “OK” and report themselves accurately?
A second type of study, called DILO (“day in the life of”), in which observers follow specific persons around all day, also is generally not valid – most of the time, this type of study will produce wrench times of 50% or higher. Why are DILO studies not accurate indicators? For one thing, the persons selected for following might not be “typical” craftspersons on “typical” days. For another, persons being followed generally do not act as they normally would.
A third type of study, in which observers go at statistically set times into a shop area to see what visible persons are doing is – you guessed it – also not valid. This analysis, too, tends to produce wrench time results upward of 50%. These studies don’t take into account persons who are available but who aren’t in the shop area. These unseen persons might have been traveling, in the storeroom, or in some other delay area.
The best method for measuring wrench time is with a statistical method where each person in the workforce has an equal chance of being observed over a sufficient period to represent the actual workforce over time. Most plants can, over the course of about a month, conduct a reasonable study that might be representative of the ongoing plant using a single day each week for observations. For example, the observer could use Monday the first week, Tuesday the second week, etc., through Friday the fifth week. Going down through names on a set roster and making two observations each half-hour over the course of eight-hour days would provide 160 observations and margin of error of plus or minus 7%. Thus, a plant at 35% wrench time would know if it is between 28% and 42% and probably not at 55% (or 80%). A plant that improved to 50% wrench time would know if it is between 43% and 57% and probably not a typical plant at 35%.
The concept of wrench time is invaluable to unlocking great productivity through proper planning and scheduling, but you do not have to measure it. Be cautious if you do. Feel free to email me if you would like a copy of an actual wrench-time study with its methodology and results, a slide presentation on doing an in-house study yourself, and my current list of work and delay categories and definitions that I think are helpful.