A multi-objective genetic algorithm approach to design optimal zoning systems for freight transportation planning

This paper contributes to the existing research on freight transportation, spatial and land use planning by investigating an improved spatial aggregation technique to delineate desirable freight traffic analysis zones. Zoning is a process of spatially aggregating several predefined basic spatial units (BSUs) into multiple zones. It plays a vital role in the transportation planning and decision-making process and is well-documented as the modifiable areal unit problem (MAUP). MAUP involves aggregating BSUs to obtain optimal zones satisfying specific criteria and constraints. This paper proposes an improved spatial aggregation methodology to develop a freight traffic analysis zone system by applying the multiobjective optimization concept using a genetic algorithm. The decision variables, namely, (i) Freight trip density; (ii) Number of establishments; (iii) Employment density; and (iv) Compactness, are chosen to represent the elements of freight, passenger traffic, and land use. The problem is formulated as a multiobjective network partitioning problem. The four objectives aim to create zones with better homogeneity and compactness. It is solved using a genetic algorithm with a weighted distance metric approach to prioritize the four objectives. Results show that zones resulting from the improved methodology are superior to the existing zones in terms of homogeneity of decision variables and compactness. The findings are expected to help the decision-making process of urban, transportation, and land-use planners in selecting appropriate freight traffic zone delineation strategies for a given region.