Imagine a 3D computer-aided design (CAD) on screen, covered in a fine mesh of small triangles. “FEA [finite element analysis] normally reveals all the safe areas of your design as blue and green, and the ‘at risk’ areas, where fatigue could cause failure, in yellow and red,” says Steve Cory, computer-aided engineering manager at Thatcham-based Xtrac.
The £40-million engineering business makes fail-safe parts, such as gearboxes, for top-end motorsport applications, including Formula 1, making extensive use of FEA as a fundamental tool.
Typically, a design engineer builds a component in CAD, tests it in simulation software, which is often, but not always, within the CAD suite, and imports it into a product lifecycle management (PLM) package when that part needs to be managed across users as data within a larger assembly across multiple users.
Simulation software functionality has improved greatly in recent years, driven by intense vendor competition and the demands of very computer-literate engineers. “Much simulation can be performed with a cursory understanding of physics; you don’t need a mechanical engineering degree,” says Mike McDermid, senior design engineer at Sheffield firm Performance Engineered Solutions (PES), which is used to designing anything from electric motorbikes to sports equipment.
Using simulated computer-aided design within a PLM system, in a fast-moving and flexible engineering function, is fundamentally important
This year Siemens PLM Software launched its CAD platform, NX 8. A favourite with the motorsport industry – Williams and Red Bull both use NX with TeamCenter, its PLM suite – NX 8 boasts 200 new capabilities over its previous version of NX, including improvements in the speed of geometry and analysis modelling, and topology optimisation and multi-physics analyses.
What does this mean for the designer and manufacturer? “The biggest improvements in simulation are in their usability and analysis processing time,” says Dan Fleetcroft, engineering director at PES. Core functions have remained fairly constant, but new multi-physics capability, such as thermo-mechanical solutions and non-linear analysis, is a step-change, he says.
In the real world, very few products are subjected to a singular physical force. “By simulating the interaction of these different forces we achieve greater fidelity between our analysis and reality, ultimately giving us greater confidence in the results,” says Mr Fleetcroft. “Multi-physics software can combine a range of mechanical, fluidic [aerodynamic-hydrodynamic], thermal and electromagnetic forces that may influence product performance.”
One improvement in FEA is linked to the processing power of the hardware. Software developers have enabled these memory-hungry programs to be more efficient and more “hardware-aware”, says Mr Cory.
“Our workstations use graphics cards and the recent version of Ansys [simulation tool] can recognise GPUs – graphical processing units – that live on the graphics card,” he says. “Because the card is concerned solely with presenting the best possible image on screen, it has to do a lot of number crunching to keep up. Partitioning the processing between core CPU [central processing unit] and card really speeds up the simulation.”
Simulation cannot be performed within PLM. But using simulated CAD within a PLM system, in a fast-moving and flexible engineering function such as Xtrac’s, is fundamentally important, says Mr Cory.
“It provides a stable platform for the ‘master model’ concept; that is a solid model of a component or assembly that drives all other data – drawings, CAE [computer-aided engineering] structural analysis and CNC [computer numerical control] manufacturing programs,” he says. “Re-use of existing data is simplified, streamlining the development process, and driving the rapid adoption of best-practice solid modelling techniques and company standards.”