In this research we consider an approach for improving the efficiency and tightness of column generation (CG) methods for solving vehicle routing problems. This work builds upon recent work on Local Area (LA) routes. LA routes rely on pre-computing (prior to any call to pricing during CG) the lowest cost elementary sub-route (called an LA arc) for each tuple consisting of the following: (1) a customer to begin the LA arc, (2) a customer to end the LA arc, which is far from the first customer, (3) a small set of intermediate customers nearby the first customer. LA routes are constructed by concatenating LA arcs where the final customer in a given LA arc is the first customer in the subsequent LA arc. A Decremental State Space Relaxation (DSSR) method is used to construct the lowest reduced cost elementary route during the pricing step of CG. We demonstrate that LA route based solvers can be used to efficiently tighten the standard set cover vehicle routing relaxation using a variant of subset row inequalities (SRI). However, SRI are difficult to use in practice as they alter the structure of the pricing problem in a manner that makes pricing difficult. SRI in their simplest form state that the number of routes servicing two or three members of a given set of three customers cannot exceed one. We introduce LA-SRI, which in their simplest form state that the number of LA arcs (in routes in the solution) including two or more members of a set of three customers (excluding the final customer of the arc) cannot exceed one. We exploit the structure of LA arcs inside a Graph Generation based formulation to accelerate convergence of CG. We apply our LA-SRI to CVRP and demonstrate that we tighten the LP relaxation, often making it equal to the optimal integer solution, and solve the LP efficiently without altering the structure of the pricing problem.

In this research we consider an approach for improving the efficiency and tightness of column generation (CG) methods for solving vehicle routing problems. This work builds upon recent work on Local Area (LA) routes. LA routes rely on pre-computing (prior to any call to pricing during CG) the lowest cost elementary sub-route (called an LA arc) for each tuple consisting of the following: (1) a customer to begin the LA arc, (2) a customer to end the LA arc, which is far from the first customer, (3) a small set of intermediate customers nearby the first customer. LA routes are constructed by concatenating LA arcs where the final customer in a given LA arc is the first customer in the subsequent LA arc. A Decremental State Space Relaxation (DSSR) method is used to construct the lowest reduced cost elementary route during the pricing step of CG. We demonstrate that LA route based solvers can be used to efficiently tighten the standard set cover vehicle routing relaxation using a variant of subset row inequalities (SRI). However, SRI are difficult to use in practice as they alter the structure of the pricing problem in a manner that makes pricing difficult. SRI in their simplest form state that the number of routes servicing two or three members of a given set of three customers cannot exceed one. We introduce LA-SRI, which in their simplest form state that the number of LA arcs (in routes in the solution) including two or more members of a set of three customers (excluding the final customer of the arc) cannot exceed one. We exploit the structure of LA arcs inside a Graph Generation based formulation to accelerate convergence of CG. We apply our LA-SRI to CVRP and demonstrate that we tighten the LP relaxation, often making it equal to the optimal integer solution, and solve the LP efficiently without altering the structure of the pricing problem.