A Comparison of Mixed Integer Programming Models for the Construction Site Layout Problem

Site layout planning has been widely investigated in the literature as a means of minimising transportation costs, improving safety etc. The majority of the research conducted focuses on developing adequate formulations capable of producing optimum results. However, there is a lack of consensus on a particular format to implement for the problem, particularly in regards to the assumptions and constraints to be adopted. Distance between facilities, which is an important measure in site layout planning, has been modelled in several ways, all in the hope of producing reasonable approximations to the path followed by workers, machinery etc. between facilities. This paper aims to compare the performance of several models of the site layout problem. The objective function considered in this study is the transportation cost of material movement between facilities. A main mixed integer programming model is proposed to represent the site layout problem and a comparison is carried out against a relaxed variant, where travel distances are approximated. Further, a constraint generation algorithm is presented in a bid to improve the computational performance of the main model.