OEE measurements usually look at the past to analyse performance trends. So what if you could create a model that predicted the future of some of the statistics?
Setup / Changeover Time
One of the most important aspects of Opcenter APS scheduling is the ability to calculate sequence dependent changeover or setup times.
We often ask our customers this question:
"When a job finishes on machine X, how do you decide what to run next?"
The answer often boils down to trying to reduce downtime - in this case the inevitable time that a machine cannot be producing goods while labels or bottles or tools are being changed in readiness for the next product to be made.
To generate a schedule that tries to minimise this, we have to build in calculations such as those in a paint factory, where changing over from white to red is expected to take 30 minutes, but going from red to white needs 4 hours.
Here the black bars represent this on our Gantt chart:
Therefore we can graph the predicted time, and perhaps try to do something about it before the factory starts work.
Because we are doing finite capacity scheduling, we have to model the available capacity in some detail, taking account of future holidays, planned maintenance etc. So we know, out of the 168 hours per week, how many hours each resource is "on shift" and therefore available for production to be scheduled.
If the resource is on a 24/5 pattern then it will be off shift only at the weekend, leaving 120 hours available.
The schedule is made and might be able to fill up the entire week with work to do. Or, for some of the hours, the flow of materials from suppliers or other manufacturing operations might result in a period when there is nothing for the machine to do. Which we call idle time.
If the idle time is 20 hours this must be because the remaining 100 hours correspond to time when the resource is either in setup or will be actually working and producing products.
Hence we have the numbers that we need to calculate a predicted utilisation.
Knowing that we have the details behind the numbers to help us look deeper and improve the prediction to get more output from the resources.