Does anyone know if Plant Simulation can do combinatorics analysis?
Is it possible using Experiment manager ou GA for this?
In my case, I have 13 workers that are available to work in 20 workstations.
I need vary the settings and look for to throughputPerHour.
1 worker works in all workstation -> throughputPerHour= X
2 workers work in all workstation (settings 1) -> throughputPerHour= Y1
2 workers work in all workstation (settings 2) -> throughputPerHour= Y2
2 workers work in all workstation (settings 3) -> throughputPerHour= Y3
2 workers work in all workstation (settings n...) -> throughputPerHour= Yn...
Can someone help me?
***can you share with me examples?
Solved! Go to Solution.
It seems you want to simulate scenarios where you have different configurations for the WorkerPool. Of course this is possible.
First you create a table with the alternative creation tables of the Workerpool, as well as a variable that shows the desired setting (i.e. referring to which creation table applies, the ExperimentManager should change this variable). Then you configure the ExperimentManager to write the setting to the Workerpool (SimTalk: setCreationTable). You do this in the Configuration method: ExperimentManager>Tools>Advanced settings>Use configuration method and click on the button.
There comes an example model with your installation: Start page>Example models>Small examples>Tools and Optimization>ExperimentManager>Experimental design
I hope this is enough explanation, assuming that you know some SimTalk.
This means that I can create only variables and Plant Simulation does combinatorial analysis to me, without I set a scenario.
I observed the example of Plant Simulation and there are a scenario setting.
To you understand my doubt follow below my case...
you'll observe a table called "work plan". Initially, I defined many types of setting, varying number of workers at workstations. However, this isn't practicable, the ideal to my is that Plant Simulation does this to me.
Nothing wrong with your SimTalk-skills! That makes some suggestions a bit easier.
If I understand the problem correctly, you want to assign skills/services to workers, right? This is an interesting problem, with many possible solutions (well, in theory); you already mentioned this.
I think there are two approaches. One is a classical optimization problem (assign services to workers to achieve certain output with minimal workers). I think this could work with the Genetic Algorithm. The other is what I would call practice-oriented (add services for nearby stations to workers where they are needed the most, based on statistics from the model objects). This could work with Rule-based settings for the ExperimentManager (Advanced settings>Rules).
Maybe @Peter has some suggestions how to specify this optimization problem.
Hope this helps. I will be off-line for a couple of days, but others can help you further.