Effective Optmization Series: Cost for trucks and trailers
For Optimizer costs information is the basic driver for all planning steps. Internally it evaluates each transport plans by summing all the cost terms and selects the best with the lowest cost. Without proper costs it can produce you many surprises.
What are proper costs? The most important guidance is to orient the cost settings by the real cost rate. Then the optimizer can output solutions close to your expectation. Moreover, only if you start with cost rates near to the real cost, you will have the possibility to add new resources, tariffs or whatever without large challenges. If you don’t have them in detail or you have only some average achieved cost, then use this cost rates as starting point.
And you need to decide for a currency. Internally the optimizer is doing NO conversion of the currencies. All cost terms sent to the optimizer, must be in the same currency to avoid nonsensically computations. My personal favorite is to use one of the real existing currencies like Dollar or Euro, instead of a virtual currency like “APO-Dollar”. It might have the draw-back, that you somehow have to adapt the conversions, but it is much simpler to work with since you easily can compare you planning cost with the cost which you see in the real world.
The next question is which cost of the real world should be passed to the optimizer. It is not a good decision to automatically model all costs. It is better to start with some simple costs. After you get all in running as expected, you can step by step enhance the complexity to model details in the cost structure. One more importance point, which is obvious but often negligent, is that the optimizer needs some cost which can influence its decision. If one item is constant in any planning, you do not need to modify it for the optimizer.
Now let us start with costs for trucks and trailers.
Carriers V.S. Own Fleet
If the resources belong to carriers, you can maintain the average achieved rates for the optimizer to keep it real and simple.
For your own fleet, some of the cost of a truck is independent of its usage, e.g. parts of fix cost for the maintenance and the inspections or the tax. Whatever a transport plan is, you have to pay these costs anyway. So you should compute something like a planning-relevant rate for your internal trucks, e.g. it should contain the cost for driver, distance dependent maintenance or other distance dependent cost. Doing so, you will usually have lower cost rates for your own fleets than for carriers. But this is exactly what we want. We want to prefer our own trucks, as long as there is capacity.
The first cost you maintain for the optimizer should be a rate for the distance.
Distance Cost: from planning cost profile V.S. from distance lane
If the rate is fixed, you can maintain the cost in the cost profile of the planning profile. This is a usual approach for your own fleet, especially if you act regionally. If you use carriers, you have usually different rates in different regions or in different directions. In this case you have to maintain the distance cost on the transportation lane. An average rate in the cost profile would be too rough.
If you maintain a transportation lane between two specified locations you can define the cost on the lane directly. If you have lanes from or to zones, which contain several locations, you can specify only a cost rate per distance. The Geo-Information-System will compute the distances and the optimizer will use the cost rate multiplied by the distance to compute the overall cost. Whenever you define such new lanes and start the optimizer again, you should have a look into the explanation tool to see if the distance computation is working well.
Calculation of Distance Cost
To compute the overall cost of a freight order the optimizer sums up the cost of all stages of the freight order. Dependent on information of the lane or the cost profile, the distance cost per stage is computed and summed up.
There is one exception: especially in the US instead the distances of the stages, the lane from the first to the last stop of the freight order decides the rate, even if the truck drives completely different stages because of additional stop-offs. You can define this behavior in the cost profile by enabling “Destination-Based Distance Costs” for such means of transports.
distance cost of freight order $1: 1459,564 = (581,7900+147,9920)*2
To finalize our cost settings to cover the distance we have to think about a last important issue: If you use carriers you usually pay only for the freight order itself, you don’t care about the following freight order or about the distance to drive home. To model this you create such resources as multi resources. They have the “beam”-feature. The optimizer doesn’t care about stages between freight orders.
For your own fleet you want to minimize such stages between freight orders. They are automatically created and considered by the optimizer, if you create “real” resources (not multi).
With such a setup you usually cover optimization of the distance. It should already produce good plans. Now we can use additional cost to model the reality a little bit better or to define some penalties to force the optimizer to compute specific plans. But keep in mind, that the cost should be oriented on the reality. So keep the additional cost small.
Fix Cost per Freight Order
Within the cost profile you can define a fixed cost for every freight order. This can help as additional penalty to reduce the number of freight orders and to use larger trucks or larger FOR’s instead. This can be part of our realistic cost modeling, too.
Fix Cost per Truck
We emphasized before that for your own fleet you do not need to maintain the real fix cost. But if in your scenario you have only own fleet and you want that optimizer use as few trucks as possible transport, you can set this cost to manipulate the optimizer results.
Intermediate Stop Cost and Maximum Number of Stop
Through this cost and/or max number of stops in planning costs profile you can regulate the size of a freight orders in term of number of stops. E.g. if optimizer creates Freight orders with too many stops than expected, you can enhance this value. Or e.g., if you want to have freight orders only with one customer, you can either set very high value to intermediate stop or use 1 for max number of stops.
This cost is part of realistic costs and can be defined in a distance-independent way or to multiply them by the distance. The distance independent way can be used, e.g. If your carriers charge you per quantity. The distance dependent way can be used as oil expense by optimizer; through this cost the heavier freight units have higher priorities and the optimizer will avoid to travel them around, but the routing may be not optimal.
If the optimizer creates longer tours than your expectation. You can use these two values to limit tours.
The last but not least: keep it simple. Scenarios can easily become sophisticated by over fine tuning of costs. It will be a barrier to figure out a sinful model for your scenarios. Important tips: you can design consecutive optimizer runs beginning with simple cost setting, or even use hard constrains instead costs if possible!