BIM evolucionário: aplicabilidade de algoritmos de inteligência artificial no desenvolvimento de projetos de edificações

Abstract

In the scope of the Architecture, Engineering and Construction (AEC) industry, parametric technologies such as Building Information Modeling (BIM) and Generative Design have supported new design systems. The increase in the use of BIM impacted the stages of detailed design and construction. The parameterization allowed for several advances in the generation, exploration and automation of these projects. However, although the generative and automated requirements assessment processes are highly useful, they are restricted to execution functions and do not offer evolutionary solutions. The integration of BIM technologies, generative design, automated design verification and evolutionary algorithms in this research demonstrates that it contributes to the increase of the optimized generation of solutions for building projects - especially in the Early Concept Design phase -, advancing the applicability of Artificial Intelligence in the field of civil construction. The approach Design Science Research was adopted as research methodology. This research design comprised the description of the problem and the formulation of a construct called Evolutionary BIM (E-BIM). The development and implementation of the construct occurred in three phases: Phase A - development, implementation and evaluation of a genetic algorithm inserted in the flow of BIM designs, as an evolutionary instrument; Phase B - generation, verification and optimization of an architectural object, integrating the technique of Shape Grammar and the computational tools Rhino-Grasshopper, Archicad, Solibri Model Checker and Python script; Phase C - verification of the assertion of the term Evolutionary BIM (E-BIM) proposed in the context of this application. The quantitative results demonstrated: the feasibility of automated generation of solutions in scale; the automated generative modeling of BIM IFC models; the optimization and increase in design performance indicators achieved with the use of evolutionary algorithms. Together, the results point that the implementation of an evolutionary algorithm integrated with the proposed design process, allows designers to explore, on a large scale, multiple solutions and improve the performance of these solutions, contributing to improving the productivity of AEC design.