When Topology Optimisation Isn’t Enough

Topology optimisation tools are getting more powerful, offering greater control over outcomes, faster iteration, and increasingly sophisticated ways to generate geometry. The problem isn’t that geometry doesn’t matter, in fact, the ability to generate the right geometry is a critical differentiator. 

But production programmes don’t fail because geometry wasn’t clever enough. They fail because even advanced geometry wasn’t engineered for the realities that follow. Inspection. Distortion. Support strategy. Surface criticality. Joining. Sealing. Fatigue. Post-processing access. And ultimately, unit economics. 

These are not secondary considerations. They are the conditions that define whether a part succeeds or fails in production. 

And they are rarely captured fully within optimisation workflows. 

The Limits of Optimisation 

That’s why “optimised” parts still underdeliver. Topology optimisation workflows are highly effective at solving specific, well-defined objectives (reducing mass, improving stiffness-to-weight ratios, or improving thermal performance). Within those boundaries, they can produce elegant and highly efficient geometries. 

But they only address one part of a broader engineering challenge. 

In practice, their outputs are often applied to geometries that were never designed with additive manufacturing in mind in the first place. The starting point is typically a legacy CAD model (shaped by the constraints of machining, casting, or fabrication) and the optimisation process simply refines those assumptions rather than challenging them. 

The result is a familiar pattern in AM, optimised but basic manufacturing assumptions result in the component failing to meet expectations. 

The Legacy Design Trap 

This is where many AM programmes quietly lose their advantage. Engineers approach additive manufacturing using design logic inherited from traditional processes. Tool access, assembly constraints, and historical geometries shape the architecture of the part long before optimisation begins. 

By the time topology optimisation is applied, the design space has already been constrained. 

What follows is not true DfAM, but adaptation. Geometry may improve in simulation. Mass may be reduced. Performance metrics may shift in the right direction. 

But the design remains fundamentally misaligned with the realities of the chosen AM process, and the opportunities it offers. 

What the “Winning” Design Looks Like 

The “winning” design goes further. 

It integrates process behaviour, manufacturability, and functional intent from the outset. It recognises that geometry is not just a solution to a physics problem, but a response to a complete engineering system. Critically, it also introduces capabilities that are difficult (or impossible) to achieve with conventional manufacturing methods. 

This is where additive manufacturing becomes competitive. Not in refining existing parts, but in enabling entirely new performance. That might mean integrating multiple functions into a single geometry. Embedding flow paths within structural components. Designing for thermal behaviour and mechanical performance simultaneously. Or rethinking interfaces and assemblies entirely. 

In these cases, complexity is not aesthetic. It is purposeful. 

Engineering First, Optimisation Second 

The distinction is subtle, but fundamental. Optimisation explores a design space. Engineering intent defines it. Without that foundation, even the most advanced tools will converge on solutions that are locally optimal, but globally incomplete. 

The future of DfAM will not be defined by how efficiently we generate geometry, but by how intelligently we frame the problem in the first place. Because geometry is not the end point, it is the output of intent. 

Closing the Gap 

This is the gap between optimisation and engineering. And it is where most additive manufacturing programmes succeed or fail. 

Closing that gap requires a shift in mindset, from tool-driven workflows to intent-driven design. From adapting legacy geometry to rethinking it entirely. From treating manufacturability as a constraint to embedding it as a design driver. 

We’ve written this blog to explore where topology optimisation workflows fall short, and what engineering-first DfAM looks like in practice. Because the real promise of additive manufacturing is not in what we can generate. It’s in what we can engineer.