The Size of the Queue Content influences the results

Question

question

TalHzf asked Mar 15, '23 TalHzf commented Mar 15, '23

The Size of the Queue Content influences the results

Hi all,

In my Modell I need to know what is the capacity of my machines to produce a proportional mix of Items.

The problem is in the queue size, I don´t know why the size influences the results, even that I give the Interarrival Time =0 ( Supposing that I am having no out of stock).

I gived to the queue 1,2 and 3 the values 3 and 10 and the results were completely different.

Thank you for your help

Modell Th.fsm

Software Version:

FlexSim 22.0.0

queue queue size

modell-th.fsm (53.0 KiB)

· 2

Ralf Gruber ♦ commented · Mar 15, 2023 at 08:27 AM

TalHzf,

Your dashboard shows the exact same results for me, if I run with max content 10 vs. max content 3 for Queues 1, 2, 3....I ran until your stop time of 86400. Are you looking at other results, too?

0 ·

TalHzf Ralf Gruber ♦ commented · Mar 15, 2023 at 08:51 AM

Hi @Ralf Gruber ,

I thought that the results will not be influenced by the size but when I run the modell with 10 this is the result

and with 3 these are the results

0 ·

1678870270766.png (10.0 KiB)

1678870309449.png (9.5 KiB)

Answer 1 · 2023-03-15T09:32:40Z

Felix Möhlmann answered Mar 15, '23 TalHzf commented Mar 15, '23

This happens because you are using the Express Version. When using random numbers, objects are by default assigned their own "stream" of random numbers (this is what "getstream(current)" you might have seen in some statistical distributions is used for). Due to this, the random numbers that are generated in triggers and options for that object are independent of the rest of the model. It does not matter if Source1 generates 10 numbers or 3 numbers before Source3 starts to generate items. The random numbers (and thus label values) will stay the same for Source3 regardless.

This, however, is not the case in the Express Version.

Here, only a single stream of random numbers is used. So when Source1 "uses up" 10 of those, the numbers that are used by other sources will be different than if Source1 only uses 3. See the example below where I have two sources generate random numbers between 1 to 5 when creating an item. The order in which the numbers appear is the same regardless of which source they were generated for. As a result the first three numbers for Source2 are different between the cases.

Whereas in a licensed version, the numbers are specific to the source (if getstream(current) is used). The order of numbers for Source2 is the same, regardless of the amount of numbers that were generated by Source1.

Edit: In this case you could get around this by having all items be created by a single source. Send those to an initial buffer which then distributes them to Queue1-3.

As long as the buffer is large enough, no type should run out for a long time, even if the ratio in which the types are generated does not exactly match the ratio in which they are consumed.

modell-th-fm.fsm

1678872157939.png (24.0 KiB)

1678872639980.png (6.1 KiB)

1678872683468.png (5.8 KiB)

modell-th-fm.fsm (52.7 KiB)

· 3

TalHzf commented · Mar 15, 2023 at 10:42 AM

Hi @Felix Möhlmann ,

Thank you so much for the clarification.

I want to ask if there any possibility in your modell to set an interarrival time for the item 1 and 2.

Thank you so much

0 ·

Felix Möhlmann TalHzf commented · Mar 15, 2023 at 11:12 AM

You could manage the release of items in a Process Flow. That way you can delay certain types by varying amounts.

A zone can restrict how many tokens of a certain subset/partition can enter the relevant section of the Process Flow so that only one item is ever pending to be released to each downstream queue.

modell-th-fm-2.fsm

0 ·

modell-th-fm-2.fsm (60.1 KiB)

TalHzf Felix Möhlmann commented · Mar 15, 2023 at 11:20 AM

Thank you so much for your help @Felix Möhlmann.

0 ·

question