How to set die PM intervals by produced parts

Deciding that die PM should trigger on produced parts is the easy half, and the argument for it is settled: a die wears by the hits it makes, not the weeks that pass. The stall comes at the next question. What number? Fifty thousand? A quarter million? Set the interval too tight and the toolroom strips healthy dies all year. Set it too loose and the interval is just a slower way to run to failure. The right number is not a guess, though. The die has been telling you all along; the method is mostly about writing down what it said.

Start from what the die was built to do

The first anchor is the build itself. The toolmaker or die supplier who built it usually has a service expectation: a sharpening cycle for a stamping die, a cleaning cycle for the gates and vents of a mould, a life expectancy for the wear plates and springs. Those figures are a starting point, not a verdict, because they assume the material, press and running conditions the builder had in mind, and your more abrasive material or higher press speed can halve them. But an interval anchored to the build spec is a defensible first setting, and defensible matters when the interval has to be explained later.

Let the failure history argue with the spec

The second anchor is what has actually happened. Pull the last few failures on the die, or on its nearest siblings, and ask one question of each: how many parts had it run since the last service when it failed? If wear-driven failures are arriving before the spec interval says they should, the spec has lost the argument. If the die has run multiples of the spec interval with clean findings at every service, the spec was conservative for your conditions. Either way the failure counts, not the calendar dates, are the evidence. A failure log that never recorded the count at failure cannot join this argument, which is itself a reason to start recording it today.

Adjust on findings, not on faith

From the first count-triggered service onward, the findings drive the number. A strip-down that finds real wear at the interval says the interval is about right, or slightly late. A sequence of clean services says it can extend, in measured steps. A failure between services says tighten, and look at whether the wear mode is even count-driven. Corrosion on a die that sat idle for a month is a calendar problem, not a count problem, and it needs a time-based check alongside the count trigger rather than a tighter count.

Record the count at every service, or the interval cannot learn

None of this works if the produced-part count is not written down at every service and every failure. The count at service, against the interval it was set to, is the data the next adjustment stands on, and it is also the exact record an auditor or customer reads when they ask whether the die was maintained on the right basis. In VoraTool a PM plan set to a produced-part interval reads as every N produced parts, the die record carries its current count, and each completed PM event stores the count it was performed at beside the outcome. The interval, the evidence for it and the proof it was honoured all live on the tool record, which is what lets the number improve service by service instead of being re-guessed every year.