Network Analyst - VRP anomalies

[AlfonsoDodero]

Hello everyone,

I'm having some issuesusing the Network Analyst Vehicle Routing Problem solver and I really hope that someone can help me.

So, I have about 160 orders modelling trash bins to collect.

But I see some anomalies in the definition of their sequence: in my tests some orders (curb approach: No U-Turns) are skipped in a first step on the arc and then they are collected in the following. This results obviously determines a lack of accuracy.

Do you think there is a solution?

Limiting the route to collect by 50 orders solve this issue.

Could it be therefore  an inherent problem of the algorithm of VRP  dueto the complexity of the Network Dataset and the density of the points?

Thank you very much

Alfonso

[AlfonsoDodero]

This is the recap of my problem:

I would need to put the order "6" immediately after 2. On the contrary, the VRP solver assignes it at the end of the route when the truck goes back to the depot. Is it possible to get the result I want without including time variables?

Thank you!

[JamesHood]

I am experiencing the same problem.

In a test data set (70 orders) I increased the distance cost for the route and this fixed the problem.  (The default for time cost is 1 and the default for distance is null.  I set the distance cost to 1).

I theorize that by telling the solver that distance cost is important, it tries to reduce the amount of overlap.

However, when I increased my data set to 250, the same problem appears even with the distance cost set to 1.

I tried some tests by setting the distance cost much higher (2, 100, 1000, 10000) and this does not resolve the issue.

I am currently looking into this further and will post back if I find any additional information but I wonder if you were able to solve this?  I wonder if anybody has some advice?

I'm on 10.3 Advanced.

Thanks!