Page 1 of 1

Tourpointfixation

Posted: Mon Jan 09, 2017 3:45 pm
by Lauterfeld
Hi all,

I have an input plan with only one tour.
Tour.tourPointFixation is NONE.
TourPoint.tourFixed is false and
TourPoint.remainPlanned is false.
Tour.vehicleFixed is true.

There are 3 stopps in the tour, it is a manually built tour. The 3 stopps are not drivable because one of the stopps is not rechable in the given time window.

2 questions:

1) Why doesn't xTour remove the stopp which is not reachable? Remainplanned is set to false. When does xTour remove orders out of the input tour? Only in case of 'extrem violations'?

2) Why does xTour always respect the sequence of the stopps in the input plan? It seems that he tries to make the best out of the given tour with its manually built sequence but he does not change the sequence of the stopps of the input plan.

In my case xTour reclaims the correct violation TOUR_OPENING_INTERVAL in the response. But remainPlanned is false, so the order is allowed to be removed. Also xTour would be allowed to change the sequence of the stopps.
My goal: plan manually a "lead" tour and ask xTour to make a "gold" tour out of the input plan :-)

Thanks in advance for any information about that!

Regards,
Volker.

Re: Tourpointfixation

Posted: Mon Jan 09, 2017 3:52 pm
by Joost
xTour cannot fix a non-valid inputPlan. If parts of the inputPlan are deemed violated these will be ignored during optimization.

Re: Tourpointfixation

Posted: Mon Jan 09, 2017 4:28 pm
by Bernd Welter
In fact - Joost is right. Tom be a bit more detailed:
  • A tour that violates at least one constraint is a VIOLATED TOUR.
  • A chain that contains at least one VIOLATED TOUR is a VIOLATED CHAIN.
  • A planning call with an empty input plan will never return VIOLATED CHAINS / TOURS.
  • A VIOLATED INPUT TOUR will be returned without any manipulation but the distance/period scheme is calculated based on the input sequences of the violated chains tours.
  • Automatic optimization will only be applied to VALID chains.
Imagine a scenario where chain1={tour1,tour2,tour3} are given with an invalid tour3.
Such an input plan will not be optimized even if tour1 and tour2 are valid.

Quite sure this answers the questions end enlightens you ;-)

Best regards Bernd

Re: Tourpointfixation

Posted: Mon Jan 09, 2017 6:56 pm
by Lauterfeld
Hi together,

now I'm enlighted. :idea: :shock:

But the sense of 'remainPlanned' becomes a bit more senseless. If my input plan is valid from A to Z, xTour has no reason to remove an order because all tours are drivable. Sure some orders could change the tour and some orders build a very bad tour but all remain planned. Or is it dependend of the goad ranks? If my goal rank für numbers of tours is very high and the rank of unscheduled orders is low?

Regards,
Volker.

Re: Tourpointfixation

Posted: Tue Jan 10, 2017 8:10 am
by Bernd Welter
Hello Volker,

it depends on the potential moves whether such a constraint is meaningful or not. Imagine moves such as
  • replace scheduled order 123 by unscheduled orders 456 and 789 to achieve higher number of orders but reduced distances
northern track: input tour. tour is too long to add one of the other two southern point.<br />southern track: output tour. By removing the northern point it is possible to add two southern points
northern track: input tour. tour is too long to add one of the other two southern point.
southern track: output tour. By removing the northern point it is possible to add two southern points
lauterfeld.gif (27.98 KiB) Viewed 9342 times
Another important statement: goals are applied only during the improvement phase.
Planning with construction and no further steps won't apply the goals.

Best regards Bernd

Re: Tourpointfixation

Posted: Tue Jan 10, 2017 9:22 am
by Lauterfeld
Hi Bernd,

thanks for the explanation. Yes that makes sense if there are an input plan and unscheduled orders.

Volker.