Interleaved optimization

Continuous moves can be modeled with interleaved shipments. In this type of model, a route can have deliveries and pickups interspersed.

The solver generates routes where each route is a schedule for a vehicle. The vehicle starts from its home base, visits a sequence of pickups and deliveries, and then goes back to the home base if a round trip is required.

The solver seeks to route the shipments with the available asset resources while attempting to minimize the total cost.

The interleaved shipments problem adheres to the various constraints that you define. Asset maximum capacity constraints are required, but there are a number of other optional constraints that apply, such as business hours at locations and the various constraints defined in the Transportation Assets table.

Additionally, there is a Last-In First-Out When Loading And Unloading constraint in the Transportation Optimization options. When this option is selected, the shipment that is loaded last will be the first to be unloaded. The unloading order for other shipments on the route follow this pattern. This constraint is used to model a vehicle with a single loading door. This setting represents a vehicle with a single loading door in which the first items loaded are placed furthest from this door.

Interleaved optimization and direct shipping cost

If you populate the Direct Shipping Cost in the Shipments table, and the Direct Shipping Cost turns out to be lower than the routed cost, those shipments are unrouted. Therefore, you will not see routes where the Selected Mode is "Direct Ship".

Limitations for interleaved optimization

The following are current limitations when working with interleaved optimization.

• Compartment 2 values currently do not work with interleaved shipments.
• Asset Search Distance is not used with interleaved optimization. The solver enumerates the routes using all the assets that it thinks are reasonable.

Interleaved optimization high level algorithm summary

The following provides a high level review of the interleaved optimization algorithm:

• Route enumeration
  • Cluster the shipments to create smaller problem sets to solve. The clustering uses the pickup and dropoff locations to group similar shipments into the same cluster.
  • For each asset type and pool site, generate a set of candidate routes by inserting shipments onto initial routes and being guided by key performance indices.
  • Use column generation technique to iteratively include new candidate route in the route pool and discard any bad routes.
• Initial route selection, asset assignment and initial solution generation
  • Solve a mixed integer program (MIP) to selects the routes and asset types to be used to build the initial solution.
  • If the Optimize Fleet Size setting is selected: Heuristically string the routes together to create a continuous path for each asset ID.
• Solution improvement phase
  • This phase utilizes improvement phases similar to those built for the standard optimization in addition to some modifications and unique improvements specific for the management of the assets and the in structure of the interleaved problem.
• Post processing and reporting