Bulk Plan Multi-stop Shipment Column Generation Algorithm

The column generation algorithm is selected by setting the multi-stop logic configuration parameter, MULTISTOP CONSOLIDATION ALGORITHM TYPE, to Column Generation.

The column generation algorithm is similar to complete enumeration in that it looks at several shipment consolidation combinations and selects the best set of shipments to cover all the order releases/order movements. However, the column generation algorithm, unlike complete enumeration, generates a much smaller subset of shipment combinations and can scale to larger problem sizes. The column generation algorithm employs either a concurrent, sequential, or look-ahead savings algorithm for generating the additional columns (or shipments). The underlying savings algorithm can be called multiple times to generate several different consolidation options. Each savings algorithm call takes the same amount of time as the original savings algorithm. The following combination of parameter settings is typical with column generation algorithm. Setting the iterations to 1, consumes the same amount of run time as the savings algorithm, with the advantage of getting similar or better solution quality as with the savings algorithm. Note that this is only true when cost savings is used instead of distance savings (i.e. when MULTISTOP COST SAVING CHECK TYPE is set to "4 - Calculate cost for all paired shipments" or MULTISTOP PERFORM THOROUGH SEQUENCE CHECKS is set to true).

• MULTISTOP CONSOLIDATION ALGORITHM TYPE = Column Generation
• MULTISTOP COLGEN SAVINGS ALGORITHM TYPE = Concurrent
• MULTISTOP COLGEN NUMBER OF ITERATIONS = 5

For this algorithm type, if the property glog.optimization.mip.setcovering.writeToFile is set to true, the set covering MIP problem will be written to the OTM logging directory as "SetCover.lp".

Note: Enable the log ID 'OptMipData' to export a CSV file of column generation information (like columns, dual, variable names and their OTM object references, assignment or shipment sequence) when it is used. This log can be used to analyse the solution obtained from the column generation.