A Crystal Ball for Design & the Predictive Nature of Simulation




Ever since we started to move away from the drawing board and the vellum, ink, razor blade combo (remember those days? Just me? No? I’ll crack on then), we, as a collective of designers, engineers and manufacturers, have been talking about the ability to do more with our digital design tools than pure documentation.

The rise of 3D modelling tools made further promises in this regard. After all, if you have a 3D model, you should be able to do more with it than purely document the shape of an object. Right? Abso-bloody-lutely.

If you have a 3D model, you can then perform all manner of things as the design progresses. Whether that’s using it to drive the creation of views in drawings, the production of safe and efficient NC tool-paths for manufacturing, or as we covered a few weeks ago, the creation of photorealistic renderings.

What links all of these processes together is that they, typically, occur downstream from the creation of the 3D model. Sometimes in parallel as the design moves towards a final point and freeze, but they have to start afterward – once you’ve got cracked. Rarely do they inform the design and engineering process from the very earliest stages.

The benefit of 3D-associated processes is really to be found much earlier, long before you’ve got to the point of formalising the design – the experimental stage, the point where your product’s form and function is the most fluid. It’s here that simulation and analysis are fundamental to adding more value to your 3D-focussed development processes.

Whether you refer to it as simulation or analysis (I’ve never been too clear on the distinction between the two), whether you use thermal and fluid flow simulation tools or conduct stress/strain evaluations of your product, whether you use more simplified (but nevertheless, just as informative) 2D work.

If you want to try to predict the performance of your product according to your (or your customer’s) requirements, to formal legislated standards or to more informal needs, simulation holds the key.

Consider the ability, which many readers will be aware of, to take a 3D model at the early stages of design and run it through a series of tests according to real-world conditions and operational environments. To highlight areas that need more attention or further investigation. That’s something that we’re all on board with.

The design and engineering software technology market has always proposed the idea of upfront simulation and analysis. Consider the value and knowledge to be gained from not only having your product documented in three dimensions, but to also start to factor in knowledge of how it will perform under real world conditions – that’s incredibly powerful.

The reality is that simulation usage and adoption are still slow and I’m sure, annoyingly so to those tasked with trying to sell this process improvement to designers and engineers. Much of the issue is that many of these tools suffer from a couple of common issues.

Firstly, there’s a mismatch between the language used by the designers and engineers and the software. If you’re developing a pump, you’re interested in flow rate, in how to minimise the mass of your product, of how to find the perfectly specified power unit. You’re not really interested in Computational Fluid Dynamics per se. You want to investigate a part’s robustness under its expected operating condition – not its fatigue life.

The same is true of structural strength and performance. Does the part break? Will it deform? Is there an issue with vibration from the power unit? Not modal frequency analysis or Von Mises stress.

There’s also the issue of assumptions and the delineation of the different disciplines. If you’re going to simulate your product, you want to be able to do so with the most accurate representation of the product (both geometrically and in terms of loading and boundary conditions) and you want to be able to see the effect of all physics – rather than looking at stress/strain, thermal and fluid dynamics separately.

The good news is that this is starting to change. There’s a shift that sees much of the jargon of analysis removed. Tools are becoming more realistic in their language. There’s also the move to a multi-physics world (in part driven by the increase in the computational power needed to do this type of work) that breaks down the barriers and gives the design and engineering professional a deeper and more holistic understanding of how a product, even at the early stage of development, performs. While most tend to be skeptical about the power of a crystal ball, when it comes to the predictive nature of simulation in a design and engineering context, its potential is huge.