Automation of Simulation and Engineering Processes
Generations of engineers have spent their lives trying to get things designed faster, cheaper. Not only that but the world expects every reducing development timescales, and ever increasing numbers of projects. I remember when I was working for a certain superconducting magnet company in Oxfordshire I had the wonderful title of Value Engineer.
As a Value Engineer the whole purpose of my existence was to reduce costs by redesigning products or processes, whilst retaining the quality of the end product. Although I had access to a then up to date FEA tool, (IDEAS and some NAG routines for those with an interest in history) I had to manually look at the sensitivity of my design to the variation in the position of the coils within the magnet. This was a tedious process as each possible design took weeks to generate, and as you’d expect most didn’t give any form improvement.
Perhaps it was just my inability to engineer anything but I always felt that there must be a better way of doing this. Automating this process would have been a huge gain in time and therefore money, however this might have been just a bit poineering given that it was the late 1980’s. Looking at modern tools available now I could probably have completed a design in 2 to 3 weeks rather than the 3 months it took me. At the time the idea of using some of the stochastic analysis was not considered although if I had been aware of Design of Experiments, Sis Sigma and Taguchi methods it would have really helped.
There was, and still is, an amount of paranoia from employees within companies regarding automation of systems and processes in the workplace, and I know some of my colleagues would have said “but doesn’t that make you redundant “. However the reality would have been very different. If I saved myself 10 weeks of work then I could have looked at many more possible solutions to the problem, and consequently, I suspect, created a much better design than the one which was finalised and manufactured. And for the avoidance of doubt, which worked.
Coming back to the present day there are now a number of software products around which can look at the variability of parameters within your design, and we are fortunate to have one in our portfolio, called Isight. With its open system architecture you can integrate your design tools (typical design tools include packages such as Excel, Mathcad, Abaqus, CAD, visual basic programs, et al) and put together an engineering process in the form of a workflow. A number of analyses can then be run to look at your engineering process. For example:
- Design Optimisation – considering either geometry or functional variables of the design space the analysis can be performed to reduce the weight, performance or cost of the design.
- Design of Experiments (DOE) – Explore the design space using various DOE methods
- Six Sigma – Measure the quality of the design
- Taguchi – Strive to achieve performance targets whilst minimising performance variation
- Data Matching – The ability to effectively reverse engineer the defined process using a graphical set of results data, often created by test data, and determine the best fit for the variables in the process.
Finally, given the complexity and volume of the data these things generate, there are a number of tools to review the results, and visualise and understand whats happening in your design space. There is a lot more information about this on both, on our website.
