Injection moulders: scrap per shot and cooling set for safety

Heater bands and chillers work against each other

What happens. The barrel heater bands and the water chillers of the same machine run at the same time, each on its own controller. After a start-up or a change of setpoint, one heats while the other cools.

What it costs. Energy for heating and for cooling the same heat, and scrap while the temperatures settle after the start.

Why the usual fix fails. Each loop works on its own and meets its own target. Retuning one loop shifts the problem to the other.

How a fixed-price remote diagnosis checks it. From temperature and power logs of barrel, hopper and chiller circuits I check how long they work against each other and how much energy it takes. The fixed-price remote diagnosis tells you whether coordinated thermal control based on a model is worth building.

Background: Predictive control of process plants

Part weight and dimensions vary from shot to shot

What happens. From shot to shot the part weight and the dimensions change a little, with material, temperature and machine wear. The cavity pressure curves look slightly different each time.

What it costs. Scrap and sorting, and the safety margin that every customer drawing makes necessary.

Why the usual fix fails. A check of every n-th part is a sample, not a measurement of each one. Machine settings are corrected after the trend has been seen, not before.

How a fixed-price remote diagnosis checks it. From the process curves of many shots (pressures, positions, temperatures) and the measured part values I check how well weight and dimensions can be computed for each shot. The fixed-price remote diagnosis tells you if this supports correction during the run.

Background: Computed measurements

The switchover point is fixed while the viscosity changes

What happens. The machine switches from injection to holding pressure at a fixed screw position or time. When the viscosity of the material changes, the cavity is filled differently at the moment of switchover.

What it costs. Short shots when the melt is stiff, flash when it is thin.

Why the usual fix fails. Operators change the switchover point when scrap appears, and change it back when the next lot comes. Fixed settings cannot follow a material that varies.

How a fixed-price remote diagnosis checks it. From pressure and position curves I check whether the viscosity can be estimated early in the shot and the switchover adapted. The fixed-price remote diagnosis tells you what to expect.

Background: Adaptive predictive control

Cooling time is set conservatively

What happens. The cooling time is set so that every part is certainly cool enough to be ejected. Nobody measures the real temperature in the part.

What it costs. Cycle time, which means output per machine and cost per part.

Why the usual fix fails. Shortening the time by trial risks warpage and ejection marks, so the time returns to the old value. Surface sensors do not see the core of the part.

How a fixed-price remote diagnosis checks it. From mould temperatures, cycle data and part quality records I estimate the part temperature at ejection and how much cooling time is spare. The fixed-price remote diagnosis tells you if the cycle could be cut safely.

Background: Computed measurements

Several mould temperature circuits interact

What happens. A mould has several temperature circuits. Changing one shifts the temperature in the area of the next.

What it costs. Warpage, dimensions that vary and long settling after each start.

Why the usual fix fails. Each circuit is controlled on its own, so interactions are left to the setter. Tuning by hand takes time for each mould.

How a fixed-price remote diagnosis checks it. From circuit temperatures, flows and part measurements I identify the interaction and check whether coordinated control would settle faster and hold the temperature better. The fixed-price remote diagnosis tells you how much warpage it could remove.

Hot runner zones oscillate or overshoot

What happens. The zones of a hot runner swing around their setpoints or overshoot after a change. Neighbouring zones heat each other.

What it costs. Degraded material, black specks and variation in the filling.

Why the usual fix fails. Standard settings from the controller supplier are a compromise for all runners. Changing them without a model often makes it worse in another zone.

How a fixed-price remote diagnosis checks it. From zone temperatures and heater power I check the response of each zone and its neighbours and whether model-based zone control would be calmer. The fixed-price remote diagnosis tells you what to expect.

Every new recyclate lot needs new settings

What happens. A new lot of recyclate comes with a different melt behaviour. The setter finds new settings by trial, and the next lot starts again.

What it costs. Scrap and set-up time at every lot change.

Why the usual fix fails. A table of settings per lot grows without end and does not cover the next. Waiting for the lab on every lot takes time.

How a fixed-price remote diagnosis checks it. From machine curves at the start of each lot I check whether the melt properties can be estimated and the settings adapted automatically. The fixed-price remote diagnosis tells you how much of the set-up could be saved.

Background: Adaptive predictive control

Cavities of a multi-cavity mould fill unevenly

What happens. In a mould with many cavities, some fill earlier than others. The parts are not all the same.

What it costs. Rejects from the cavities that fill worst, and sorting by cavity.

Why the usual fix fails. Adjusting the runner balance by trial is slow and holds only for one material and one temperature.

How a fixed-price remote diagnosis checks it. From pressure signals and part results I check whether the balance can be estimated and how it changes with the conditions. The fixed-price remote diagnosis tells you what could be corrected from the process side.

Granulate is dried to be safe

What happens. The dryer is set for a fixed time and temperature that is long enough for the dampest delivery. The actual moisture of the material is not measured at the hopper.

What it costs. Energy for drying that was not needed, and material that is heated for too long.

Why the usual fix fails. A shorter time risks splay and weak parts when a damp lot comes. A moisture meter on every lot adds work.

How a fixed-price remote diagnosis checks it. From dryer air temperature, dew point, throughput and the moisture checks you have I check whether the end point can be computed. The fixed-price remote diagnosis tells you how much drying energy could be saved.

Background: Computed measurements

Barrels are heated while chillers cool

What happens. Across the hall, barrels are heated electrically while the chillers run to cool mould and oil. The two sides know nothing about each other.

What it costs. Electricity for heat that is removed again by cooling, and chiller energy.

Why the usual fix fails. Sub-meters show the numbers but not what to change. Heat recovery needs a plan and an investment.

How a fixed-price remote diagnosis checks it. From the power logs of heating and cooling and the production plan I check how the two interact and where coordination would save energy. The fixed-price remote diagnosis tells you what it is worth.

Background: Economic predictive control

Setters leave and the settings live in their heads

What happens. The experienced setters know which settings a mould and a material need. When they leave, the settings leave too.

What it costs. Long set-ups, variation between shifts and a dependency on a few people.

Why the usual fix fails. Setting sheets list the numbers, not the reasoning. New setters take years to learn.

How a fixed-price remote diagnosis checks it. From the history of past set-ups I check which settings were found and why, and build a documented model for setting assistance. The fixed-price remote diagnosis tells you whether it can carry the knowledge to new staff.

A bigger machine runs with the old settings

What happens. A new, larger machine is bought, and the old mould settings are taken over. The process behaves differently: more shot volume, other heating, other dynamics.

What it costs. An unstable process, scrap in the start-up and months until the new machine runs as well as the old.

Why the usual fix fails. Copying settings ignores the changed dynamics. Retuning by trial on the new machine takes production time.

How a fixed-price remote diagnosis checks it. From data of the old and the new machine I compare the response and check how the control must be adapted. The fixed-price remote diagnosis tells you how to bring the new machine up faster.

There is no quality value per part

What happens. Quality is checked on samples. For each single part there is no value that says whether it is good.

What it costs. Rejects that go undetected until the customer finds them, and a large sampling effort.

Why the usual fix fails. More sampling is more effort and still misses parts. Alarm limits on single signals produce too many false alarms.

How a fixed-price remote diagnosis checks it. From process curves per shot and the inspection results I check how well a quality value per part can be computed. The fixed-price remote diagnosis tells you if it is good enough to sort parts or stop a bad run.

Background: Computed measurements

Mould changes are planned by hand

What happens. The planner decides by hand in which order mould changes follow. Colour, material and tool changes all take time.

What it costs. Downtime for changes that could have been avoided or shortened, and orders that are late.

Why the usual fix fails. A spreadsheet or a board does not weigh the changeover times against due dates. A change in the orders breaks the plan.

How a fixed-price remote diagnosis checks it. From orders, changeover times and machine data I check how much downtime the plan has and whether a better sequence would reduce it. The fixed-price remote diagnosis tells you what to gain.

Background: Trajectory optimisation and setpoint generation

Every recyclate batch changes the melt flow and costs scrap

What happens. Batch variations and moisture in post-consumer material change the melt rheology. Part quality drifts with each batch, and the pressure in the cavity shows it.

What it costs. Scrap with every new batch, and a limit on how much recyclate the mould can take.

Why the usual fix fails. Learned systems that were trained for one mould do not transfer to other moulds without adaptation.

How a fixed-price remote diagnosis checks it. From cavity pressure curves, material lots and part results I check whether predictive control of the cavity pressure that adapts to each mould would keep part quality constant. The fixed-price remote diagnosis tells you what it needs.

Background: Adaptive predictive control