Different behavior between nodes and A*

Question

question

Emmanuel VS asked Jan 4, '23 Emmanuel VS commented Jan 5, '23

Different behavior between nodes and A*

Hello everyone.

I was testing the use of Nodes and A* and I noticed a certain difference in the behavior of some resources.

I made a demo replicating a simple task sequence executed by an operator. As you can see, Operator 16 was able to finish the total number of boxes earlier, causing the difference in utilization indicator presented in the dashboard.

What is the logic behind this difference? Thank you in advance

Demo_utilization-04012023.fsm

Software Version:

FlexSim 21.0.10

astar navigator utilization network node

demo-utilization-04012023.fsm (59.1 KiB)

Answer 1 · 2023-01-04T21:57:10Z

Kavika F answered Jan 4, '23 Emmanuel VS commented Jan 5, '23

Hey @Emmanuel VS, the reason for the disparity between speeds is the Navigator each Task Executer (TE) uses. The TE connected to the nodes uses the "DefaultNetworkNavigator" which takes into account Acceleration, Deceleration, and Flip Threshold.

The other TE uses the AStarNavigator, which doesn't take these things into account (only Max Speed).

So, what you're seeing is the difference that Acceleration/Deceleration can make with a TE. I cannot say as to why these things were implemented the way they were.

1672869298286.png (7.9 KiB)

1672869381024.png (8.0 KiB)

· 4

Joerg Vogel commented · Jan 05, 2023 at 07:37 AM

Please look at this question about another involved navigator. Travel times are eventually controlled by a different parameter value in a travel task.

If a travel on a network node path lasts shorter, then the end speed at last network node drops from max velocity to null in zero time. You have to adjust the mentioned parameter in travel task.

0 ·

Emmanuel VS Joerg Vogel commented · Jan 05, 2023 at 03:02 PM

@Kavika F @Joerg Vogel

Thank you very much for your responses. I imagined that the difference was because of the topics that you discuss.

I was observing the example sent by @Joerg Vogel and it works perfectly. I replicated it also in my model (attaching version) and the difference that I see is that, as in Astar there is no node/control point (AGV), that acceleration/deceleration does not occur. How could I generate that equivalence? Could it be by discarding these parameters as @Kavika F mentioned?

Demo_utilization-05012023.fsm

0 ·

demo-utilization-05012023.fsm (59.9 KiB)

Kavika F ♦ Emmanuel VS commented · Jan 05, 2023 at 05:23 PM

Yes, if on your Task Executer connected to a Node Network has Acceleration and Deceleration set to 0 m/s/s, then it will behave like the Task Executer from AStar. Another reason the AStar Task Executer seems to move slower than the other is because of the pathing he has to take to place the box. That difference can add up over time causing the difference you see.

0 ·

Show more comments

question