Balance total transit time for groups of vehicles

[harisbal]

Hello,
I am running a VRPTW problem with N vehicles. Each vehicle belongs to a certain group (e.g. A or B). I am trying to balance the total transit time between those two groups. In other words, I need that groups A and B to have travelled for similar durations (e.g. +- 10%). Since I am interested to balance the pure transit time (excluding slacks and waiting time) I cannot use: time_dimension.CumulVar(routing.End(index))
Any suggestions?

[jmarca]

Maybe create additional dimensions that are just transit time, not time?  Just don’t add slack, waiting, loading/unloading, etc. JamesOn Feb 26, 2024, at 17:32, harisbal ***@***.***> wrote: Hello, I am running a VRPTW problem with N vehicles. Each vehicle belongs to a certain group (e.g. A or B). I am trying to balance the total transit time between those two groups. In other words, I need that groups A and B to have travelled for similar durations (e.g. +- 10%). Since I am interested to balance the pure transit time (excluding slacks and waiting time) I cannot use: time_dimension.CumulVar(routing.End(index)) Any suggestions? —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[harisbal]

That's a very good idea actually! Thanks.
Out of curiosity though, I was wondering whether there is a built in way to track the transit time of a vehicle.

[jmarca]

Not to my knowledge, no. I mean, at a fundamendal level, the solver knows nothing aside from what you tell it. However, you tell it how to define (and therefore minimize) travel cost with the call to: ``` routing.SetArcCostEvaluatorOfAllVehicles(transit_callback_index) ``` (or similarly using the per-vehicle call `SetArcCostEvaluatorOfVehicle(...)`) So there may be some buried way to get at that value. But I don't know of it if there is one, and none of the "Get..." functions under RoutingModel here <https://developers.google.com/optimization/reference/python/constraint_solver/pywrapcp#routingmodel> look promising. I do agree that creating a new dimension to track something that is already being tracked internally seems a little bit duplicative, but I think it is the best way. James

…

On Tue, Feb 27, 2024 at 01:49:10AM -0800, harisbal wrote: That's a very good idea actually! Thanks. Out of curiosity though, I was wondering whether there is a built in way to track the transit time of a vehicle. -- Reply to this email directly or view it on GitHub: #4101 (comment) You are receiving this because you commented. Message ID: ***@***.***>

-- James E. Marca Activimetrics LLC

[harisbal]

Thank you for the insights @jmarca! Adding a new dimension is not such a burden.