Publications / 2020 Proceedings of the 37th ISARC, Kitakyushu, Japan

Multiple Tower Crane Selection methodology utilizing Genetic Algorithm

Preet Lodaya, Abhishek Raj Singh and Venkata Santosh Kumar Delhi
Pages 558-565 (2020 Proceedings of the 37th ISARC, Kitakyushu, Japan, ISBN 978-952-94-3634-7, ISSN 2413-5844)

Equipment selection for a construction project is a complex decision making task that impacts the project cost significantly. The literature highlights the increasing focus of recent studies on tower crane (TC) optimization, largely due to the shift from horizontal to vertical construction. Many studies tend to select a single TC model to be used throughout the site and assume the number of cranes to be used as a priori knowledge or is calculated using heuristics and schedule demands. Selection of a combination of TC models is relatively unexplored. Existing literature focuses on finding a single most optimum solution to a multimodal problem. Need to look for multiple optimums arises from the uncertainties of an integral but disjoint simulation process and the local and specific nature of un-modelled constraints. Thus to address this gap the study presents a selection approach, aiming to leverage the long reach of TC, formulated as a function of the rental cost and lifting requirements. The selection of the equipment is of prime focus in this research with implicit consideration for the positioning of the TC. A Bi-level optimization problem is formulated involving task allocation to different TC models and minimization of required crane count for each model. Genetic Algorithm (GA) has been employed to work with non-differentiable multimodal function and obtain the preferable TC combinations from the available model variations. The results derived from the proposed model included optimal yet dissimilar TC combination options, task allocation to the utilized TCs and feasible regions for crane placement. The major limitations were parameter setting for the adopted algorithm and the inability of the distance metric to robustly capture phenotypic differences.

Keywords: Genetic Algorithm; Tower Crane selection; Bi-level optimization; Multimodal optimization