A BIM-interoperable optimisation plug-in for structural design of whole-roof timber truss systems

Abstract

Optimal design of timber roof truss topology, geometry, and cross-sections is limited in practice due to the absence of integrated, code-aware design tools. This paper presents a Grasshopper 3D-based optimisation plug-in. The plug-in combines site-specific action generation with Eurocode 1, finite element analysis in Karamba3D, Eurocode 5 member checks in Python and genetic algorithm (GA)-based topology and geometry search within a single parametric workflow. Considering three truss types Fink, Pratt, and Howe cross-section sizing is performed against a discrete library of solid timber sections. Optimised models export to Industry Foundation Class format, supporting interoperability with building information modelling based coordination tools. The plug-in is applied across six Norwegian locations, as well as four inventory scenarios consisting of one new and three reclaimed component stocks. This results in location-specific, minimum mass roof configurations that satisfy Eurocode ultimate limit states. Cross-section optimisation with Eurocode 5 compliance checks executes in under 20 s on a standard office computer. The overall computational cost is dominated by the GA, with total runtimes varying according to truss type and stock composition, ranging from a minimum of 1 h 15 min to a maximum of 5 h 54 min. The applicability of the plug-in is constrained by predefined geometric limits, including a roof pitch between 20 (Formula presented) (Formula presented) and 35 (Formula presented) (Formula presented), bay spacing from 1.55 m to 2.8 m, and web member spacing between 0.7 m and 1.25 m. Cost and embodied carbon (EC) assessments show that minimising mass alone is an insufficient objective when connection contributions are included. Plug-in extensions to broader geometry, more locations, direct cost and EC optimisation, and reliability-based design for reclaimed stock are identified as priorities for future work.