TY - CONF

T1 - A new conceptual design optimization tool for frame structures

AU - He, Linwei

AU - Gilbert, Matthew

AU - Shepherd, Paul

AU - Ye, Jun

AU - Koronaki, Antiopi

AU - Fairclough, Helen

AU - Davison, Buick

AU - Tyas, Andy

AU - Gondzio, Jacek

AU - Weldeyesus, Alemseged

PY - 2019

Y1 - 2019

N2 - Architects and structural engineers are increasingly embracing the use of optimization tools for use in the design of building structures. One tool is numerical layout optimization, which provides a powerful and highly efficient means of generating theoretically optimal lightweight structures. However, the layouts so obtained are often impractical due to their complex forms. In this paper, a means of generating families of more practical solutions for use as inspiration at the conceptual design stage is described. With layout optimization simpler solutions can be generated by running the optimizer with penalization parameters, e.g. to represent the cost of joints. Alternatively, a family of frame designs can be automatically generated by minimizing structural complexity indices (e.g. the total number of members, number of intersecting members at joints, etc) in a post-processing step. This is here shown to provide a pragmatic and computationally efficient means of identifying families of practical, near-optimal, truss designs. In the paper the aforementioned techniques are first briefly outlined; a Rhino-Grasshopper plugin tool incorporating the techniques is then used to generate families of 2D and 3D design concepts, which clearly indicate the promise of the methods involved.

AB - Architects and structural engineers are increasingly embracing the use of optimization tools for use in the design of building structures. One tool is numerical layout optimization, which provides a powerful and highly efficient means of generating theoretically optimal lightweight structures. However, the layouts so obtained are often impractical due to their complex forms. In this paper, a means of generating families of more practical solutions for use as inspiration at the conceptual design stage is described. With layout optimization simpler solutions can be generated by running the optimizer with penalization parameters, e.g. to represent the cost of joints. Alternatively, a family of frame designs can be automatically generated by minimizing structural complexity indices (e.g. the total number of members, number of intersecting members at joints, etc) in a post-processing step. This is here shown to provide a pragmatic and computationally efficient means of identifying families of practical, near-optimal, truss designs. In the paper the aforementioned techniques are first briefly outlined; a Rhino-Grasshopper plugin tool incorporating the techniques is then used to generate families of 2D and 3D design concepts, which clearly indicate the promise of the methods involved.

M3 - Paper

ER -