Several simple, but essential key performance indicators (KPIs) help assess and guide proper planning and scheduling practices. Maintenance planning provides the productivity piece of maintenance and literally “increases your workforce without hiring” to complete work. Yet we need to know if planning is “working” and how to guide its proper deployment. This paper shows how to control the planning function. These maintenance planning and scheduling KPIs include scheduling, planning and storeroom support. This paper also discusses tricky KPIs such as mean time between failure (MTBF) and planned vs actual time estimates.
Controlling Planning and Scheduling Overall
Planning and scheduling have great potential to help maintenance. The essence of planning is not to provide perfect job plans. It is to provide a service of giving head starts to jobs and collecting actual job feedback to continually and forever improve plans for better head starts in the future. Planning also supports the scheduling function, which requires time and skill estimates. The essence of (weekly) scheduling is not to provide complex advance personnel assignments and time slots. It is to provide a mission of work each week for each crew.
It does this by researching the backlog to provide a simple list of the most appropriate work to fill the forecasted available labor hours. Surprisingly, imperfect but gradually improving plans and simple batches of work dramatically increase the productivity of maintenance crews. Note that the whole strategy is based on the allowance of imperfection. It is OK if plans are not perfect, and it is OK to break the schedule. With these allowances, plants see dramatic increases in work order completion rates.
Nevertheless, plants do not do planning and scheduling for the sake of doing them. Plants plan and schedule maintenance work to attain available and reliable production capacity. Increased work completion over normal levels increases available and reliable production capacity over normal levels in this sense: Plants “right-size” their maintenance workforces over the years essentially in response to operations complaints. The louder operators complain, the larger the backlog grows. Plants soon hire maintenance staff to control the complaints. While the maintenance staff generally does enough proactive work to be a “good” plant in this environment, sizing a maintenance workforce over the years due to complaints is in reality organizing around reactive maintenance.
No one ever calls for doing more proactive maintenance over the amount to keep the operators satisfied. Therefore, if planning and scheduling delivers a higher-than-normal rate of work completion, by definition, all extra work is proactive. The difference between a good plant and a great plant is usually that the great plant drives itself to do as much proactive maintenance as possible. Thus eradicating much more reactive maintenance.
Supposing that planning and scheduling could help improve plant production capacity through doing more proactive work, how does a plant know that it is doing planning and scheduling correctly? That issue is a matter of control. An effective maintenance KPI is often thought of when the term control comes to mind. But KPIs are not the only form of control. Hiring and training the proper persons for planning and scheduling roles is the primary control of planning and scheduling. Not everyone is capable of being a competent planner or scheduler, just as not everyone is capable of being a good electrician dentist, or nursery worker.
Don’t expect to hire the wrong persons for planners and somehow otherwise control them with maintenance planning and scheduling KPIs, procedures, continual meetings or direct supervision (all different forms of control) to do a good job. Competent planners and schedulers generally have decent craft skills, organizational skills and communication skills. (If a planner could be weak in an area, surprisingly it is craft skill because planners must be good craft historians, not experts at every maintenance task.)
After the plant has hired the right persons for planning and scheduling roles, then maintenance performance measurements are an appropriate method of control for a number of areas associated with good planning and scheduling. Management reviews such maintenance planning and scheduling KPIs and takes action as appropriate to control planning and scheduling.
An important consideration before discussing individual maintenance planning and scheduling KPIs is the notion of “Ys” and “Xs” (sometimes known as lagging and leading indicators). In Six Sigma terminology, a “Y” is the result of process, whereas an “X” is an input to the process. A “Y” is something one cares about, but really the process determines its value. On the other hand, one can determine what “Xs” are put into a process, although the “Xs” have no value in themselves. If one can determine a relationship between the “Ys” and “Xs,” the “Ys” can be influenced by controlling the “Xs”.
An example best illustrates Y=f(X), that “Y” is a function of “X.” I care about my daughter’s school grades, but grades are a “Y.” I know that simply fussing at my daughter to make good grades might affect the grades in undesirable ways. She might cheat and not really learn. But I know that her getting enough sleep on school nights and doing homework are two “Xs” that influence the “Y” of good grades. Therefore, I concentrate on making her go to bed on time and insist she do some homework at home. The grades take care of themselves.
Similarly, the ultimate “Ys” are plant availability and reliability that are affected by the “Xs” of planning and scheduling. We don’t care about planning and scheduling in and unto themselves, but we do them because we care about plant performance. Yet, even within looking at planning and scheduling, we see various “Ys” and “Xs.” It is important to know what we can affect (“Xs”) and why (effect on “Ys”) along the way.
The first maintenance planning and scheduling KPI in scheduling is schedule compliance. How did we do on the schedule? This is actually a “Y” resulting from all the things (“Xs”) that went on during the week, causing us to meet or break the schedule. An appropriate note at this point is the term “compliance.” Compliance makes the KPI sound like an “X” under the control of the supervisor. Instead, I like to use the term “success,” which clarifies that it is a resulting “Y” that has many inputs.
I used to give credit for schedule success in terms of hours, but my experience is that it becomes difficult to explain why someone only got credit for five hours of work on a job that took them 10 hours to complete (because the scheduled time was only five hours). Simply dividing the number of completed jobs (that were on the original schedule) by the number of jobs that were on the original schedule expressed as a percentage makes an easily explained KPI for a maintenance technician. I think overall that large jobs and small jobs wash out, and we are only looking for 40 to 90 percent success. Generally, we are looking for more of a bowling score (67 percent) rather than a school grade (>=90 percent).
“We gave you 10 jobs, and although you competed 20 jobs, only six of the 20 were on the original schedule, so our schedule success is 6/10 = 60 percent.” This is an awfully dangerous indicator, and although we would like wonderful schedule success, insisting on it causes gaming (cheating) because schedule compliance is a “Y.” We will discuss this KPI later in this paper as one of the tricky maintenance planning and scheduling KPIs.
The Red-Green Report developed by Howard Matthews and Pamela Pifko of SaskPower is a great encouragement to focus people on the schedule without mandating a score. After the week is over, list the completed work from the schedule in green, the uncompleted work as black, and any work done that was not scheduled as red. It gives a great measure of productivity and level of distractions.
The second KPI in scheduling is the primary “X” that drives the productivity. We must simply give a crew a 100 percent fully loaded schedule each week to defeat Parkinson’s Law. Parkinson’s Law (Cyril Parkinson 1955) says, “The amount of work assigned expands to fill the amount of time available.” We simply must give each crew enough work. See how this KPI is an “X?” We can make ourselves give each crew enough work. Without giving the crew enough work, schedule compliance is meaningless anyway. What we are trying to do with scheduling 100 percent of the labor capacity is to give each crew a “mission” to replace a falsely guided notion that maintenance is here simply to “take care of” operations (a less productive strategy for work order completion).
The third KPI in scheduling is the “Y” we prize. If we simply give the crew enough work, we should complete more work each. Typically, we see a step-change in work order completions per month (Y-KPI 3) when we fully load maintenance crews (X-KPI 2) and allow them to break the schedule (seen in Y-KPI 1 being between 40 and 90 percent). An associated Y-KPI 3b is that when we see work order completions rise, the backlog will probably drop (unless we are also stopping the use of blanket, standing work orders). We will also look at backlog management under tricky maintenance planning and scheduling KPIs.
The first KPI in planning is called planned coverage. Again, I use pure numbers of work orders rather than hours charged against planned jobs. In the earlier maintenance KPI example for schedule compliance, the crew completed 20 jobs, but only six were on the original schedule, so we obtained 60 percent schedule compliance. However, we are now interested in the 20 jobs completed. How many of those jobs went first through the hands of a planner before assignment to the crew? A planned job is more effective and efficient than an unplanned job because the planner gives a head start for proper skills, tools, parts and job steps, including all the useful information learned over the years. Let’s say that 16 of the 20 went through the hands of a planner first. Our score for planned coverage would be 16/20 = 80 percent. We are looking for a score of >=80 percent.
This KPI begs the question of “What do we consider a planned job?” Is it simply being seen by the planner first or is there some minimum level of information input required? I give credit for being a planned job if the plan uses a reusable job plan out of the CMMS and has any estimated hours. The reusable plan ensures the planners are running the improvement cycle and not just putting job steps on the work order directly (which do not constitute a current state-of-the-art plan).
The estimated hours ensures we can schedule. Other than that, the quality of the plan is set by the selection of the proper planners in the first place. Here, we just want to make sure they are running the Deming cycle of improvement and supporting scheduling. (This method of giving credit for planned work also makes gathering CMMS data easy.) This KPI is an “X.” We can make ourselves use reusable plans and provide estimates.
The second KPI in planning is also an “X” and is about the same as the first. The sheer number of reusable plans in the CMMS should rise and then eventually level off somewhat after about six months as we start to reuse the plans. Most companies only make reusable plans for PMs (a static amount), but we want them for most maintenance jobs. We repeat maintenance work. We should be using plans that planners can improve over the years.
Naturally, the third KPI in planning is how many work orders are returned to planning after job completion with helpful feedback that could be used to improve a plan? This KPI would be easy to track with a simple box for mechanics to check if they think the plan should be changed. A plant could also add a “failure” code for “plan needs updating” if multiple choices are allowed for work order completion checking. A very low score for this KPI does not mean we have great planners but that we are not a learning company. This KPI is an “X” in that we can make ourselves report feedback, but it is sort of a “Y” from management’s perspective: Do the mechanics understand their role?
The rate of emergency and urgent work order per month is the fourth KPI for planning. This is an unusual KPI for planning and actually is a “Y.” From a plant perspective, it is the ultimate measure of reliability, but from a planning perspective, we care because it affects planners. Emergency work requires planners to abandon planning to help jobs and progress. Furthermore, urgent work that planners can plan without making supervisors wait certainly means rushing.
Controlling Storeroom Support and Other Tricky Maintenance Planning and Scheduling KPIs
Wrench time is the first of the tricky KPIs. The measurement and improvement of wrench time shows why we are completing more work. Wrench time is the percentage of time an otherwise available craftsperson is actually working with their tools on a job. Increasing from the 35 percent typical of a “good” plant to the 55 percent of a plant doing planning and scheduling properly means 30 people become as productive as 47 people (30 x 55/35 = 47). Nevertheless, wrench time is a “Y” resulting from good planning and scheduling. Measuring it makes people nervous, and we do not have to measure it to see that work order completion rates are rising. Wrench time also does not give us a complete picture anyway. Someone could be working on the wrong job in a very inefficient manner and still have good wrench time.
Schedule compliance is the second of the tricky KPIs. I like to call this “Y” KPI schedule success instead. I do like to measure schedule success because simply measuring it means we must continue scheduling. It is this continuing of scheduling that is my primary “X” anyway. To fill a schedule, we must plan enough work. To plan enough work, planners must not get bogged down on one or two jobs at the expense of planning all the work. And to keep planners even planning means that management must protect them from too many other duties. So simply measuring schedule success encourages good behaviors through the process. Yet, schedule compliance/success is one of the most dangerous KPIs in maintenance. Plants that set high targets and even give bonuses practically beg their workforce not to schedule enough work and end up at only 35 percent wrench time.
Interestingly, I like to measure schedule success, but I don’t care about the score. On the other hand, the score of wrench time is important, but I don’t need to measure it.
Backlog management is the third tricky KPI. Most plants use this KPI to decide when to hire or not replace maintenance staff and to encourage the creation of work orders. Unfortunately, if the backlog is growing, any decision to hire staff might be a mistake if we are only at 35 percent wrench time. It would be better to properly schedule 100 percent of crew capacity before any hiring. Management frequently gets alarmed at growing backlogs and often sends a misguided message to operations not to create work requests as much. That can be foolish.
Any work out there needs to be identified. I agree with John Day (Alumax 1993) that the secret to a well-running plant is to generate a lot of proactive work requests. Most plants do not create enough proactive work requests, and so plant availability and reliability are not as good as they could be. This “Y” KPI, if used at all, should be to encourage the creation of proactive work orders.
Planned versus actual time estimates is the fourth tricky KPI. This is a subtly counter-productive KPI. The only way to make it more accurate is to consistently estimate too much time on all the job plans. Then mechanics can take their time and match the estimates. We end up not scheduling enough work and have overly high schedule success, but low wrench time and productivity. Planned versus actual is an unnecessary “Y.”
Mean time between failure (MTBF) is the fifth tricky KPI. The “Y” KPI of MTBF is very popular, but in practice becomes hard to measure because of uncertainty of the definition of “failure.” Obviously, every time a pump fails, the plant does not suffer because of backup pumps and systems that can be temporarily bypassed. Larry Bradley of St. Johns River Power Park in Jacksonville, Florida, suggests simply using any emergency and urgent work order as a “failure” because the system failed to provide a week’s notice. Thus, 600 emergency and urgent work orders per month becomes a MTBF of 1.2 hours.
[(30 days/month)(24 hours/day)/(600 E&U/month) = 1.2 hours]
Finally, the sixth tricky KPI is stockouts. This “Y” KPI is a measure of storeroom support. A stockout occurs during maintenance when an expected part in the storeroom is not available (and cripples the job). The industry rule of thumb might be about a 3 percent level of acceptable stockouts (also called service level), meaning that of all parts requests, the storeroom might be missing parts 3 percent of the time. Stockouts is a huge issue because stock is so expensive to purchase and carry. A $10 million value of a storeroom has an ongoing cost of 10 to 40 percent, meaning that the plant spends $1 million to $4 million each year to “carry” the stock whether it uses it or not. This carrying cost is due to the cost of operating the storeroom, taxes, insurance and the lost opportunity to invest or otherwise use that tied-up capital.
Consider the enormous pressure to reduce stock levels. One of the primary considerations in whether to decrease stock levels is the level of stockouts, yet many plants falsely think they have zero stockouts and unwisely decide to reduce stock. These plants do not even know they have stockouts because many times mechanics finish the job using an alternate part or method (perhaps inferior or costlier) when faced with an unavailable part. Instead, plants must somehow track such unknown stockouts perhaps with a checkbox or extra failure code.
Nevertheless, once a stockout has occurred, a job has been crippled. Stockouts are a reactive “Y” KPI. In addition to tracking stockouts, a plant would be wise to create proactive “X” KPIs to allow heading off stockouts to begin with. One such KPI is measuring the number of stock items that have a quantity on hand (QOH) below their minimum, yet no purchase order has been issued. Another proactive maintenance planning and scheduling KPI would be to measure the stock items where QOH equals zero, but there is supposed to be a minimum on the shelf. These KPIs allow dealing with the storeroom before stockouts occur and it is too late.
Be mindful of “Xs” and “Ys” for planning and scheduling. Putting pressure on “Ys” simply encourages gaming and does not improve maintenance productivity. Properly managing the “Xs” does increase productivity.
The “Xs” to push are starting crews with 100 percent loaded schedules, planning enough of the work to stay ahead of scheduling, planning using reusable plans and encouraging mechanics to provide feedback.
Use measures of schedule success very carefully to encourage good behavior and try not to mandate a score.
Managing these maintenance planning and scheduling KPIs should result in better work order completion rates through higher wrench times. You should also see reactive work fall as plant availably and reliability rise.
This article was previously published in the Reliable Plant 2017 Conference Proceedings.
By R.D. (Doc) Palmer, Richard Palmer & Associates Inc.