Using genetic algorithm to automate the generation of an open-plan office layout. (September 2021)

• Record Type: Journal Article
• Authors: Chen, Chen; Chacón Vega, Ricardo Jose; Kong, Tiong Lee
• Is Part Of: International journal of architectural computing. Volume 19:Number 3(2021)
• Page Start: 449
• Page End: 465
• Publication Date: 2021-09
• Subjects: Automated process -- office design -- genetic algorithm -- open-plan office -- space layout planning; Architecture -- Data processing -- Periodicals; Architecture -- Informatique -- Périodiques; Virtual reality in architecture -- Periodicals; Computer-aided design -- Periodicals; Architecture -- Data processing; Periodicals; 720.2840285536
• DOI: 10.1177/1478077120943532
• Languages: English
• ISSNs: 1478-0771

Abstract

Today, the concept of open plan is more and more widely accepted that many companies have switched to open-plan offices. Their design is an issue in the scope of space layout planning. Although there are many professional architectural layout design software in the market, in the real life, office designers seldom use these tools because their license fees are usually expensive and using them to solve an open-plan office design is like using an overly powerful and expensive tool to fix a minor problem. Therefore, manual drafting through a trial and error process is most often used. This article attempts to propose a lightweight tool to automate open-plan office layout generation using a nested genetic algorithm optimization with two layers, where the inner layer algorithm is embedded in the outer one. The result is enhanced by a local search. The main objective is to maximize space utilization by maximizing the size of the open workspace. This approach is different from its precedents, in that the location search is conducted on a grid map rather than several pre-selected candidate locations. Consequently, the generated layout design presents a less rigid workstation arrangement, inviting a casual and unrestrictive work environment. The real potential of the approach is reflected in the productivity of test fits. Automating and simplifying the generation of layouts for test fits can tremendously decrease the amount of time and resources required to generate them. The experimental case study shows that the developed approach is powerful and effective, making it a totally automated process.