Responses to the May “Management Measures” column raised a couple of questions along the following line: “In regards to the Schedule Compliance KPI, what is the acceptable standard deviation? Example, if a PM is scheduled every 30 days, how many days past due is still considered compliant? (3-days = 10%)”
This is a great question. In fact, there are statutory environments where this kind of thinking is practically a legal requirement. For getting the job of timely maintenance done, though, it is the wrong way to look at the issue. It is very easy to become a curator of overdue work and study different levels of overdue-ness. It’s even oddly satisfying because it generates numbers in neat rows that seem to shed some new light on the issue of overdue maintenance. At the end of the week, though, the neat rows don’t make the plant run better. A much more productive approach is to direct your time and effort to doing maintenance work.
Here is a simple approach to handling the overdue issue: Schedule work for a given week. If your CMMS insists on a day, give it Friday. Then, on Wednesday of each week, conduct a maintenance planning and scheduling meeting with representatives from production, reliability, maintenance and engineering in attendance. Deliver a week's worth of completely planned jobs to the maintenance department. The work orders must all include complete job instructions, parts kits and contractor schedules – everything that is needed to complete the work. It must have been planned with tradespeople availability in mind, so that it really can be completed in the coming week. The equipment must be available to be worked on in that week. If any of the jobs is not possible in the coming week, it must be flagged at the meeting and moved to another week. This means the attendees must be prepared and authorized to make the necessary decisions about the work at that meeting.
Once the review is complete, attendees can agree to schedule the whole bundle of jobs for the following week. Then the maintenance people have a couple of days to arrange it into five days' individual workloads. This won't be a total surprise, since the maintenance planners are the ones who made up the week's workload in the first place. They will have worked with the engineers, production department, contractors and others with whom it is necessary to coordinate the jobs.
Then, any work in the bundle that is not done in the following week will be overdue. There is no need to make a statistical curve of the overdue work. It's late. The parts are underfoot, the scaffolding is up and wasting money, the contractors were scheduled and they didn't get to do the job. It's late, period.
For each missed work order, record the schedule breaker, including the reason it didn't get done. Then, each week (at first) and then every month (after the number of schedule breakers drops), schedule a cross-functional meeting about what is causing maintenance to miss schedules. Is production reneging on commitments to make the equipment available? Is there an attendance problem in maintenance? Is emergency work forcing planned work to be rescheduled? Was the schedule just too big for the workforce? When the same problem is seen to cause multiple schedule breakers, it must be solved.
I urge you to forget the X bar and R. Find out why scheduled work isn't being done and fix the problems. At the risk of sounding like Yoda, there is no amber; there is do or do not. If you do not, buddy, you’re late. You’ve spent the time and most of the money and the work’s not done.