The title of this post may be a bit misleading, but hopefully you can understand.
I have a process where in reality, there is a stream of parts fed to a machine. This machine waits for the correct amount of parts to arrive, and when it has the necessary amount of parts, it processes them for 6 seconds. It then ships the finished (assembled) part away and waits for the next stream of parts.
My issue is in regards to the Working-statistics of the machine.
Although it may seem most reasonable to apply an Assembly-object for this purpose, I have used a SingleProc due to custom scaling of the number of MUs in the model because of performance issues. So 1 MU in the model actually represents for ex. 12 real parts. Consequently, the MUs are fed with an interval where the space between the model's MUs in reality represents the single MUs i have decided to scale down and which aren't shown in the model. Naturally I have adapted the MULength to reflect this (so a MU with 2m length represents 10 real parts each 20cm long that aren't shown).
When gathering Working statistics for this machine, the utilization in the simulation is (as expected) pretty low, since the process time is shorter than the "waiting time" for the stream of parts. However, showing these kind of statistics may be misleading to a customer where they expect to see a utilization of for example 80%, but see a utilization of 20% in the model.
Now to my question: How do I realistically model the utilization of such a machine with such a flow? Ideally, the utilization should somehow be a combination of the actual processing time (when it has received all parts), and the time it waits for the amount of parts it is expecting. But this seems complicated to model if the stream of parts to the machine becomes uneven. Also, is this the correct way to think about it, or is there another way?
Has anyone encountered such a modeling task?
Edit: I noticed the SingleProc has a statistics attribute "statRelativeOccupation" -- would this be a more reasonable measurement of the machine's "utilization" given my description of the situation above?
Solved! Go to Solution.
Could you give an example of what you mean?
Right now the process time of the SingleProc is set to process 1 MU (12 real parts), but only when the MU arrives (it doesn't take into account the 12 "unseen" parts).
As an example, if there are 10 "real parts" in 1 MU i am sending to the SingleProc, and the SingleProc has a maximum speed of 10 "real parts" per minute, the ProcTime of the Singleproc for that MU will be: 60seconds/10PartsPerMin * 10 RealParts = 60 seconds.
Is this what you mean?