Customers Are Focused on One Queue

Customers Are Focused on One Queue

NewwUser90
Not applicable
30 Views
3 Replies
Message 1 of 4

Customers Are Focused on One Queue

NewwUser90
Not applicable

[ FlexSim 24.0.2 ]

Hi everyone,

This is my model: Flexsim V7.fsm


90969-1742280631373.png

For some reason, customers are concentrated at queue 2 and significantly more so than the other queues.

I am not sure why there is such a huge disceprancy..


Any advice and help is much appreciated! 🙂

0 Likes
31 Views
3 Replies
Replies (3)
Message 2 of 4

moehlmann_fe
Explorer
Explorer

The model adheres to the distribution logic that you implemented in the "QueueJoining" processor. Each customer is send to the queue with the lowest average waiting time. The average staytime statistic only updates whenever an object leaves the queue and the impact of a single object on the average will decrease the longer the simulation runs. As a result it can take a long time until the lowest average is not the lowest anymore, leading to a large number of customers in that queue. This will increase the average a lot overtime, so it takes even longer for other queues to "overtake" again and so on.

I think it would make more sense to use the total wait time of customers currently in the queue to determine where to send new ones. For this you can use a kinetic tracked variable whose rate is increased/decreased when a customer enters/exits the queue. Additionally, when a customer exits the queue, the value of the tracked variable would also be decreased by the staytime of that customer.

0 Likes
Message 3 of 4

NewwUser90
Not applicable

Flexsim V9.fsmI have made some changes to the model. Now, it is using the total staytime but I am not sure if it is being used correctly. Could you help take a look? 🙂


0 Likes
Message 4 of 4

moehlmann_fe
Explorer
Explorer

What I had in mind was something like this: Only measuring the wait time of people currently in the queue, not the total over the entire simulation.

flexsim-v8-fm.fsm

Though I got another idea. If you multiply the average waiting time (your first approach) with the current number of people in the queue and conveyor and then sort by that result, you should get a good distribution of people across the queues.

0 Likes