Planning a Month of Preventive Maintenance and Scheduling Around It

Preventive maintenance costs nothing or costs three late jobs, and the difference is which day you put it on. This walkthrough plans one month of PM at Corbin Fabrication in EDGEBIC by User Solutions: list what is due, pick the days from load data, block the hours with the right instrument, reschedule, and record the service dates so the reminders roll forward. It is one of the worked examples we publish so the routine is visible rather than described.

It takes about 40 minutes on the last Friday of the month, and it is the cheapest 40 minutes in the maintenance budget.

Step 1: List What Is Due

The due list does not come from EDGEBIC. There is no maintenance interval on a work center and no due warning at the terminal, so the cadence stays wherever you already track it: a maintenance system, a shared calendar, or the spreadsheet the maintenance lead owns.

Pull that list and write down what falls in the coming month.

Four jobs this month, one next. Note the important property of that list before you go further: knowing a service is due reserves nothing. The cadence and the plan are entirely separate systems, which is exactly why this walkthrough exists. The single-machine version of the task is how to handle preventive maintenance for a machine.

Step 2: Pick the Days From Load, Not From the Calendar

The instinct is to put PM on the due date. Do the opposite: put it on the cheapest day inside the window the interval allows.

Open Reports then Work Center Utilization and read two of its tabs. The Daily Heatmap shows the next 14 days per work center, and the Per Work Center tab gives each station a utilization rating: critical above 100 percent, high 85 to 100, good 60 to 85, low 30 to 60, idle below 30.

Corbin's July picture:

• Week of the 14th: Mill-2 at 94 percent. Every mill hour that week is spoken for.
• Week of the 21st: Mill-2 at 71 percent. There is room.
• Press-1: 46 percent all month. Two machines, and one of them is idle most days.
• Oven-1: 88 percent, but with a genuinely quiet Friday the 19th.

A 90 day interval is not a 90 day deadline to the hour. Moving Mill-2's teardown from Tuesday the 14th to Tuesday the 21st is a week of drift on a quarterly cycle, and the load data says that week is worth having. The full report is covered in how to run the work center utilization report.

Step 3: Pick the Right Instrument for Each Block

Three different situations, three different tools, and using the wrong one is expensive.

Press-1, 4 hours on one of two machines. Use the per-day capacity override. Press-1 normally offers 16 hours a day, being two machines on an 8 hour shift. Type 12 in the Capacity (h) column for the 16th with the reason "PM, one press down 4 h".

One trap here is worth stating plainly: Capacity (h) is the total for all instances on that day, not the hours for one machine. Entering 4 would say the whole station has four hours, which removes 12 hours instead of 4.

Mill-2, a full day teardown. Same dialog, single machine, capacity 0 for Tuesday the 21st, reason "quarterly teardown". The reason field is what carries the explanation here, since there is no separate downtime record to attach it to. The procedure is in how to block a machine for a maintenance day.

The compressor, 6 hours affecting the whole site. This one is not a work center block at all. A plant holiday closes the day for every work center and every shift, and it can be partial: untick the whole-day option and give the window, say 07:00 to 13:00. The scheduler subtracts only the overlap from each shift, so a station on an afternoon shift loses nothing.

Using a plant holiday for a single machine's service would idle the entire plant for a day only one station loses. Using a work center override for a site-wide outage would let ten other machines plan straight through a power cut.

Step 4: Enter Them, With Reasons

Open the per-day capacity dialog from the work center's row. Pick the shift, set the date range, and type the values in the grid. Each change saves as you type, the Overridden column ticks, and a per-row Reset removes an override and reverts the day to the shift default.

Always type the reason. Six months from now, a day showing 12 hours instead of 16 with no explanation is a mystery somebody will resolve by deleting it.

Corbin's four entries:

Twenty-six hours removed from the month, on four days chosen deliberately.

Step 5: Reschedule and Read the Damage

Saving capacity changes never moves an existing plan. The bars on screen were computed against the old calendar and stay exactly where they are until the next run, which is the single most common reason a planner thinks a block "did not work".

Run the plant. Then read the result, and read it twice, once for each candidate Mill-2 date, because this is the comparison worth having.

Teardown on Tuesday the 14th: three jobs move. One customer order goes two days past its due date and appears on the Late Jobs report with a blocker hint.

Teardown on Tuesday the 21st: one job moves by half a day. Nothing goes late. Press-1's four hour window costs nothing at all, because a station at 46 percent absorbs it without noticing. The oven's six hours push one job into the following Monday, which its due date already allows.

Same eight hours of maintenance. One week apart. One version is free and one costs a customer date.

That comparison is the entire argument to have with maintenance, and it is a much better conversation than a preference against a preference.

Step 6: Verify Before You Commit the Dates

Two checks, two minutes.

Open the Resource Calendar and confirm the blocked days read the reduced capacity you intended. This is where an override entered against the wrong shift shows itself, because the day will still read full hours on the shift you meant to reduce.

Then reopen the utilization report and confirm no station moved into the critical band as a result. Pulling hours out of one machine pushes its work somewhere, and if that somewhere is already at 96 percent you have simply moved the problem.

Step 7: Close the Loop After the Work

The month is planned. One more habit makes next month automatic.

When each service is actually done, record the real service date in whatever holds your cadence. The next interval counts forward from that date, so a machine serviced on the 21st rather than the 14th is next due a week later, and the due list stays aligned with reality instead of drifting a week further out of step every quarter.

If a PM slips, remove the override with the row's Reset button rather than leaving a zero-capacity day in the calendar. A blocked day nobody used is capacity thrown away, and it is invisible unless somebody goes looking.

What This Walkthrough Proves

1. The cadence and the plan are separate systems. Your maintenance records tell you PM is due; a capacity entry is what takes the hours out.
2. Choose the day from the load data. An interval is a window, not a deadline to the hour, and the cheapest day inside it is often free.
3. Match the instrument to the scope. Override for one machine, plant holiday for the whole site, and never the other way around.
4. Enter the day total, not the per machine hours. On a two machine station that mistake triples the block.
5. Nothing moves until you rerun. Capacity edits are not reschedules.

For the monthly capacity rhythm this fits inside, see month-end capacity review. For the day a machine breaks instead of being serviced, see a machine breaks at 10 AM.