Complexity is not a sign of quality!

In our business, we keep meeting decision makers who think geometrical complexity is a quality indicator on its own for mechanical components. This may be true from a marketing perspective - it's easier to get free publicity for a complex design (especially when it's 3D printed) - but definitely not from a structural point of view.

If you're following industrial 3D printing you can probably guess that the drawing above is an artistic impression of Airbus' metal printed wall. This was one of the metal printed structures to get the most attention last year. A cabin wall structure for an airplane printed in a number of smaller parts that have been bolted together.

I've from good sources that Airbus is fully aware, or at least their engineers are, that this design is far from optimal with respect to weight (not to mention cost). Thus, I think it should be considered nothing but a successful marketing stunt to get some brand awareness among the 99.99% of the population who are not stress engineers. For that purpose it worked brilliantly, with the design being presented in magazines such as The Wall Street Journal and Wired.

The wall is designed using "nature inspired"-algorithms (also known as genetic or evolutionary algorithms). This might be the way for the finance industry, where there are no laws of physics governing the behavior (you can use it to find the best "monkey" trader). But for structural design, you should use the information from the governing physics in an intelligent way. Whether there is a human or a computer designing the structure, it is not wise to use methods which can (to a larger or smaller degree) be compared to trial and error approaches. Approaches which require orders of magnitude more computational time (see e.g. Prof. Ole Sigmund's paper On the usefulness of non-gradient approaches in topology optimization), and in most cases will lead to inferior results.

For the Airbus wall, I think airlines buying A320s should demand discounts for the lack of proper structural optimization in the aircraft. Considering that Airbus claims that this design saves 30 kg compared to the original design, the original design must have been truly awful!

Unfortunately, avoiding software with "nature inspired"-algorithms does not guarantee structurally sane results. Even with the best topology optimization software, unwary users can easily end up with fancy, complex designs, which are far, far from optimal– but which the marketing department will love for the intriguing visual appeal.

The problem is, of course, that if you just use a sufficiently bad design as the baseline, even a poorly optimized design may appear good. This was probably the case with Arup's hyper efficient metalwork, which I have included a drawing of below. A quick look at the structure will set warning bells ringing for most structural optimization experts.

A good rule of thumb is that when you do structural optimization it should end up with a design that looks sensible. Else you're not solving the right problem or you've gotten a solution that is far from optimal. If you don’t have the technical experience to determine whether it’s sensible, don’t let yourself be fooled by complexity alone.

That said, a large part of our work is to aid clients make and print complex designs, but always with the focus that it should make sense.

Written by Erik Andreassen ·

Tagged with 3D print and Topology optimisation.