Optimization of facility locations for material transportation in a high-rise building construction site by binary-mixed-integer-linear-programming

Abstract

Inefficient facility layout inside a construction site is always one of the leading causes for productivity decrease in the construction field. To maintain efficient site management, a good facility layout plan for better working environment and safe operations is required. However, practical facility layout design is a very difficult task. Various design objectives and considerations are involved with complex interactions among site factors during specific construction periods. Existing methods lack extensions in terms of mathematical modeling, and the applications of probabilistic solution methods to deal with linear-type problems are very popular but inefficient. Facility location planning can be enhanced by introducing more binary type design variables and governing constraint sets. Thus, the feasible solution for better optimization results can be enlarged. The resultant facility location problem can be formulated as Binary-Mixed-Integer-Linear-Programming (BMILP) to supersede the Quadratic Assignment Problem (QAP), non-linear in nature, so as to assist practitioners to quantitatively evaluate the possible layout plans for efficient material delivery. Based on the basic concept to model material flows among facilities, the new BMILP formulations are developed to deal with the problems of tower cranes assignment, material storages in a high-rise building project and dynamic facilities allocations in a pre-casting yard during different stages. These basic models are extended with new constraints to satisfy more actual site conditions. The extensions in the first model include optimizing the locations of twin tower cranes with full-site coverage for material delivery with safety operation distance between each crane. The second model is extended to introduce a new dimension accounting for overtime working and operating cost. More safety considerations are introduced to the third model of the dynamic facility layout plan problem for different construction stages. These BMILP models are solved by a standard branch and bound (B&B) solution algorithm using LINGO™ to obtain the best solutions. New results are generated and compared with the solutions optimized by genetic algorithm (GA) to verify that the new proposed model approach and can ensure better numerical results as demonstrated.

Date of Award	2 Oct 2013
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Yiu Tommy LO (Supervisor) & Chi Ming TAM (Supervisor)