Assign a value to a state variable following a time series distribution

[PierreUL]

Hi!

I try to modelize a production process. All my entities are of the type. But I want to give at each entity parameters such as height and length. I want that Simio generate it randomly in a state variable as an entity exit the source.

My problem is the those parameters don't follow a standard distrubution as a normal ou triangular one. It follows time series (chronological series). Since it's a current pattern in production, I was wondering if there is a simple way to implement it in Simio?

Thank you in advance!

Pierre

(I hope it's clear enough? Don't hesitate to ask me precisions if it isn't.)

[GSanchez]

You can use a table to assign each value and increment the row value each time an entity leaves

[PierreUL]

Hi,

Thanks but my problem is not how to assign a value. I can already do it and it work. My problem is more about the statistics, how to assign the right distribution. So far I assigned a value following a normal distribution (random.normale(m, s)) but it doesn't match to the reality at all. I need to modelize time series. so is there special features in simio that could help me to do it?

Thanks in advance,

Pierre

[dsturrock]

Simio does not have a distribution for a time series, but there are other ways:

For example, let's say you are assigning length, but that average length increases by 1 inch each hour. You could use an expression like:

Random.Triangular(12+TimeNow, 16+TimeNow, 20+Timenow)

which at time 0 would give you a mode of 16, gradually increasing to a mode of 17 after 1 hour and to 26 after 10 hours.

If it is not a linear increase, you could put values in a table and at successive times read from a different row in the table:

Random.Normal(MyTable[CurrentIndex].CurrentMean,5)

Or you could use a lookup table and put time and value pairs in the lookup table:

Random.Normal(MyLookupTable[TimeNow]

[PierreUL]

Hi!

Actually it's still different. I have a random value that could come from a normal distribution and this value repeats randomly between 2 and 10 times.

But anyway I'll see with these hints.

Thank's a lot!!

[dsturrock]

If the sample itself is reused 2-10 times (highly unusual) than you could assign the sample to a state, then use that state value multiple times. You might also look into using the Math.SumOfSamples function - that might work for you, depending on what you are doing with it.