Parallel Optimization for Competing Design Scheme Exploration: A competitive island-based evolutionary algorithm for early-stage design exploration

Evolutionary optimization has recognized potential for early-stage architectural design but is typically constrained to exploring variations of a single building typology and design scheme during an optimization process. This limitation hinders the comparative analysis of different design schemes, which is crucial in early design phases. This paper presents the Competing Steady-State Island Evolutionary Algorithm (C-SSIEA), which integrates a competitive evolutionary strategy with an island-based model. By leveraging the two strategies, C-SSIEA enables the parallel optimization of multiple distinct design schemes within a single optimization process, maintaining their diversity for comparative analysis rather than merely selecting the optimal one. A case study of a multi-functional building is presented to demonstrate its application. Results show how C-SSIEA can effectively reveal performance trade-offs, such as a high-rise tower excelling in contextual sky exposure but underperforming in solar heat gain balancing, which provides comprehensive insights to support designers in navigating competing design criteria and schemes from the outset of the design process.