Modeling of discrete process time distribution

[michael_p2]

[ FlexSim HC 5.3.4 ]

Hello,

we want to model discrete process times in FlexSim Healthcare from historic data as best approximation (a continuous distribution just does not capture reality in our case).

Our data looks similar to this P(18min)=6.24%, P(27min)=6.25% etc...

P(6.25,18,12.50,27,6.25,28,12.50,32,6.25,33,6.25,34,6.25,36,6.25,39,6.25,40,6.25,45,6.25,52,6.25,55,6.25,57,6.25,60)

We tried to implement this by reading from a table but received an error message. Is FlexSim really limited to 9 variables as input to the process time?

Thanks,

Michael

[Matthew_Gillespie · Accepted solution]

The P command can only take 18 parameters, but the P command is a shortcut command intended to make it easier to write short probabilistic functions.

You should use the dempirical command. This command references a table with one column of percentages and a second column of values. Check out this sample model.

dempiricalexample.fsm

You can also use the empirical and cempirical commands to generate continuous values from a similar set of percentages and values.

[Matthew_Gillespie]

Here is that same model, but built in FlexSim 17.0 for standard FlexSim users.

dempiricalexample-standard.fsm