Avoid inefficient tours / minimum load per tour

[Bernd Welter]

Some partner just asked me how to tackle this requirement (sounds similar to viewtopic.php?t=1915 but is different) - so how do we recommend to approach this requirement?

The route planning tests fully met expectations, but the logistics department reports a problem with the car not filling up to its full capacity, I don't know if we can influence it, from what I saw, in the API we can only set the upper limit, e.g. max 20m^3. our idea is to make the cars fill the limit as high as possible

Code: Select all

"constraints": {
        "maximumLoads": [
          {
            "dimension": "VOLUME",
            "value": 2100
          }
        ],
        "route": {
          "maximumDistance": 1100000
        }
      }

currently often the routes in terms of their arrangement are very good, but have a volume of, for example, 1300, the logistics department communicated that they only send cars that have a minimum volume of 1700-1800, so we tried to set other costs for the car:

Code: Select all

"costs": {
        "perHour": 40,
        "perKilometer": 0.42,
        "fixed": 5000
      },

unfortunately we still have not been able to reduce the number of used cars in favor of more full cars.
The main goal is to fill each car to at least 1700-1800 volume.

[Bernd Welter]

Here's some feedback from my colleague Tom H. from the Competence Center - please see this in the context of API version 1.4.2 (subject to change):

Response 1 ( refers to the underlying case of the screenshot where the map is disabled )

Just had a quick look at the sample request and I think the main issue is the input in this case. The total volume of all orders is equal to 15552, spread out over 188 orders which are situated across the south of X-land. For an average load of 1700 per vehicle all orders would need to be executed by 9 vehicles or less. Given the distance constraint (1100 km max) and to a lesser extent the timing constraints this doesn’t really seem feasible in my view.

Response 2

With the features currently available in DAPI I’m afraid the only real way to try and prioritize individual vehicle utilization over trip efficiency is fixed costs and as you mention this will not work across multiple subtours.

Limiting the number of allowed subtours to 1 using the grouping workaround (assigning all orders to a single consecutive group) would be a way to avoid a case like the one you mention but of course this other far reaching effects on the planning.