Thermoforming Machine Maintenance: Heat, Vacuum and the Parts You Wait For

Half the downtime is two systems

A forming machine is a simple device with a lot of subsystems: heaters, a clamp, a vacuum system, a compressed air system, a cooling circuit, drives and a control. In the downtime book, two of those account for about half of the unplanned stops.

Heater element failure is 28 percent and vacuum system problems are 22. Between them they are the majority, and both are predictable. A heater element does not fail suddenly; it drifts in resistance over hundreds of hours and then opens. A vacuum pump does not fail suddenly either; it loses ultimate vacuum as its seals wear and its oil degrades.

The rest is hydraulics at 18, chain and clamp at 16, controls at 9 and everything else at 7. Note that the two largest categories are also the two cheapest to inspect, which is the argument for doing it.

Heater elements fail predictably

A tubular or ceramic element has a resistance that rises slightly as it ages, and a zone that is drawing noticeably less current than its neighbours is on its way out. A clamp meter on each zone, read once a month and written down, catches it weeks before it opens.

The second check is the part. A cold zone produces a local thin patch or an unmelted area, and it is visible on the formed part before it is visible on the controller. An operator who knows what a cold zone looks like on the product is a better detector than any sensor.

The third is the thermocouple. A thermocouple that has drifted reads a temperature that is wrong, and the controller then holds the wrong temperature perfectly. Checking one zone against a contact thermometer once a quarter is enough to catch it, and it takes ten minutes.

The fourth is the reflector and the insulation. A dirty reflector radiates less and the controller compensates by heating longer, which shows up as a slower cycle rather than as a fault. Cleaning is a maintenance task that pays back in cycle time, which is how you get it approved.

The vacuum system is a leak hunt

Vacuum problems present as webbing, incomplete draw, or a part that forms correctly at the start of a shift and badly by the end. All three are usually a leak rather than a pump, and the leak is usually a seal, a hose or a valve.

The first measurement is ultimate vacuum with the tool isolated. A pump that pulls to its rated level and holds is a good pump; one that reaches the level slowly and drifts back has a leak or worn vanes. The second measurement is leak-down: isolate the tool, pull vacuum, and time how long it takes to lose a stated amount.

The third is the valves. A solenoid valve that does not fully seat leaks continuously, and the symptom is a pump that runs hot and a vacuum that is never quite good enough. Valves are cheap and are usually replaced rather than rebuilt, which is why holding one is worth more than holding a pump.

The fourth is the filter. A clogged vacuum filter reduces flow without reducing ultimate vacuum, so the gauge looks fine and the draw is slow. It is the cheapest item on the list and the one most often found blocked.

Table 1 — A schedule built around the two systems that cause half the stops.

Hydraulics, clamp and the slow failures

Hydraulic problems are 18 percent of downtime and they are the expensive ones, because a valve is not a cheap part and the oil is not either. The first indicator is oil temperature: a system running ten degrees hotter than it used to is losing efficiency somewhere, and the somewhere is usually a valve or a pump.

The second is the filter. A filter that is bypassing is not filtering, and the evidence is in the oil analysis rather than in the gauge. An annual oil test costs less than a valve block and tells you which is coming.

The clamp is the third. Platen parallelism drifts as the machine wears, and the symptom is flash that appears on one side of the tool and then the other. Checking parallelism hot, at operating temperature, once a year is the only way to catch it before it becomes a tool problem.

The chain and the clamp mechanism are 16 percent and they are mechanical, which means they give warning. A chain that is slack, a bush that is worn or a guide that is dry all announce themselves as noise or as a cycle that has become uneven. Somebody has to be listening.

The cheap parts are the slow ones

The right hand panel of the chart is the part of maintenance planning that costs the most and is discussed the least. A thermocouple arrives in a week. A heater element takes six. A pump seal kit takes eight, a chain ten and a hydraulic valve twelve.

So the spare that stops the line for a week is not the expensive one. It is a heater element, which costs less than a hundred euros and takes six weeks to arrive, and a machine with one failed zone out of twelve runs at reduced output for six weeks or stops.

The rule is to hold spares by lead time rather than by price. Everything with a lead time over four weeks and a failure mode that stops the line should be on the shelf, and the holding cost of a heater element is trivial against a week of lost output.

The second rule is to hold the whole set, not one. A machine with twelve zones and one spare element will need the spare for the zone that fails next, which is not the zone that failed last. Two elements and one thermocouple per machine is a reasonable floor.

Table 2 — What to hold, judged by lead time rather than by price.

A schedule that fits the shift

A maintenance plan that needs a stopped machine will not happen in a busy month. The plan has to be built from tasks that can be done at a changeover, at start-up, or in the ten minutes while the tool is being swapped.

So divide the tasks by what they need. A clamp-meter reading needs the heaters on and nothing else, which means start-up. A filter clean needs the vacuum off, which means a changeover. A thermocouple check needs a contact probe and ten minutes, which means a planned stop.

Then put the dates in the production plan rather than in a maintenance system nobody opens. A quarterly thermocouple check that appears as a two-hour block on the schedule happens; the same check in a maintenance database does not.

Finally, record what was found. A maintenance log that only records what was done is a list of activity. One that records the reading, the date and the trend is what turns a schedule into a prediction, and it is the difference between replacing a heater when it fails and replacing it the week before.