Reciprocal frame (RF) architecture design based on statistical analysis of geometrical form parameters from built cases and parametric simulations

Based on empirical case data and parametric simulations, this study employs statistical induction to explore a design methodology of reciprocal frame (RF) at the initial design stage. Specifically, 3 primary parameters that influence the RF global morphology are identified as the maximum span (L1), the minimum radius of curvature (C), and the span-to-arch ratio (λ); 2 principal parameters affecting the dimensions of RF components are collated as the maximum length (L2) and the maximum cross-sectional area (S). Following an analysis of the tectonic relationship and data distribution in built cases, RF are classified into 2 primary types — node-type & unit-type — and further divided into 5 subtypes. The conclusive findings illustrate the superiority of the unit-type over the node-type in terms of greater L1 and spanning efficiency (L2-S), but the unit-type's adaptability to λ and C within curved surface is considerably lower than that of the node-type. Furthermore, the study unearths prediction domains for the 2 primary RF types across L1-L2, L1-S, C-L2, and λ-L2, derived from parametric modelling and structural finite element analysis. These domains define the structurally permissible extremes of component volumes, serving as effective tools for the initial design inspiration, assisting in the overall design reference of structure-morphology synchronized design. •two-tiered RF typology encompassing configurational and statistical diversion.•Various morphological features and configurational performances in 5 RF types.•Prediction intervals based on frame morphology and component configuration.•Prediction intervals applied as preliminary design reference of RF form and scale.•Integrated design via interaction of built experience and parametric simulation.