Topology Optimization Is Not a Style: Notes from a 45m Bridge Study
Our concept study for a pedestrian bridge cut steel by half, but the real lessons were about load definition, validation discipline, and fabrication limits, not organic forms.
Topology optimization has an image problem. Search for it and you will find renders of bone-like structures presented as if the algorithm itself were the design. Our 45m pedestrian bridge concept study taught us something less glamorous and more useful: the optimization is the easy part. Everything around it is the actual work.
The first lesson is that load definition is the design decision. We decomposed the design volume into 2.4 million voxels and ran over a thousand iterations, but the form was effectively decided the moment we fixed the load cases. Dead, live, wind, and seismic assumptions shape the force paths, and the force paths shape everything. If you get the loads wrong, the algorithm will faithfully optimize the wrong bridge.
The second lesson is that optimization without validation in the loop is decoration. Every iteration went through FEA checks; alternatives that exceeded L/500 displacement at mid-span were rejected regardless of how efficient they looked. An optimizer left unsupervised will happily converge on geometries that fail serviceability limits while winning the material-savings chart.
The third lesson is that fabrication constraints belong inside the algorithm, not after it. The 48 primary nodes of our study were only plausible because we planned for wire-arc additive manufacturing from the start and sized them against 1.8x design load in FEA. If your fabrication story is an afterthought, the optimized geometry is a sculpture proposal, not a bridge.
And the honest caveat: for short spans and standard load cases, a conventional truss with off-the-shelf sections is often cheaper, faster, and just as good. Topology optimization earned its 50% steel saving on this study because the span, the landmark ambition, and the material budget justified the complexity. The organic form is a consequence of forces, not a goal. When it becomes a goal, you are doing styling with extra steps.