This experiment uses generative components as a parametric modeling software and Microsoft Excel to evaluate and optimize a steel truss. The objective consists in reducing the total steel weight in the truss.
The problem is simple. It is necessary to modify the position of some truss joints (nodes) in order to explore lightweight solutions (less material) while always satisfying the structural load.
The truss is composed of 2 chords, 5 vertical webs, and 6 diagonal webs. The top chord (green) is fixed to the wall, and the last joint receives a vertical load (f). The bottom chord (red) is composed of 6 joints, and these nodes are the ones to change position in order to explore lightweight design solutions. The first joint, represented by a red triangle, can only move along one axis parallel to the wall. On the other hand, the square red nodes have the flexibility to explore the space in different directions, but they are constrained to defined areas.
If no restrictions are applied to the red nodes, the solution will be a shape similar to a triangle, a well-known structural efficient form. The idea is to let the system generate efficient forms under complex constraints.
Planar steel truss diagram
The color areas represent the constraints where the node can be positioned. The black arrows represent the trajectory of the points to the optimized location. The tube radius reacts according to the structural load.
Planar trusses are typically hosted by a single plane, and their most common uses are for roofs and bridges. For this experiment, the truss is fixed at joint 1, and the load is applied at the furthest distance of the cantilever. The red color in the structure indicates that these segments are supporting tension, and the blue color represents slenderness.
Item 1 explains in diagrams how generative components send the information to Excel and how this information is managed. Item 2 is a brief description of Microsoft Solver, the Excel plugin used for optimization. Item 3 is a comparison of two trusses: on the left, before optimization, and on the right, after optimization.
he image above shows two screenshots before and after optimization.
The model before optimization weighed 16.795 tonnes, and the model after optimization weighed 11.607 tonnes. This represents a reduction of 5.188 tonnes.
This experiment was conducted over 3 days during the SmartGeometry conference and workshops, under the supervision of Steve Downing. San Francisco, March 2009.
