We work on many projects at SSA. Some aim really high, and some are more mundane. And a good many represent a companies first exploration of simulation as part of the design process. And this is often a culture shock to both sides.
A typical project may run something like this.. A customer asks if we can do some simulation for them, we say possibly and sign the inevitable NDA. Then dropbox works its cloud based magic and an impossibly complex CAD model arrives on our desktop. So then we make our first mistake. We start the analysis and simulation process without a review of how the design got to be the way it is. And often people say “there is a problem with the design, but we don’t want to tell you what it is.. Lets see if you find it.” Although this seems like a fun way to proceed we should, but don’t often, resist the temptation to play along with this. The world is an impossibly complex place, and even fairly simple designs have many, many, phenomena at work as they function – wanted or unwanted.
We do our best; idealise the geometry to a level which allows us to run some simulations, mesh it, work out sensible loads, and we interpret an appropriate material response from the scant data available. We do all these things, and more. And we find that whatever we do some failure or other is indicated. And at this point we get back to the customer and start discussing it with them.
“So what do the hand calculations you used to calculate what size everything should be tell you should happen?” we say.. And often there is something of a silence at this point. And the silence is very telling. At some point in the simulation process we’ve generally done a hand calculation to investigate why our analysis is highlighting such an obvious failure, so we already know that had some form of initial calculation been done the thousands of pounds spent on an FEA model would not have been quite so wasted.
And this situation would be much less crazy if the calculations needed to highlight the failure were the sort of thing you found in an arcane academic journal. But sadly they aren’t. 9 times out of 10 it’s F/A, My/I or PR/2t. And often the area of the model which shows the failure is something we’ve left out, because what we’ve done is have a best guess at what could be going on.
So this means we need to get more rigorous; not in our approach to the actual work, but in finding out where the designs in question are in terms of their design history, and design input. Like it or not we are slotting ourselves into somebody else’s design process, and the success of projects is judged on success, rather than who’s fault the failure is. Maybe the next development in our quoting process is to ask new potential customers to show how they arrived at the structural aspects of their design.. and never go for the “ we’d like to see what you come up with " approach again.. Because it really does seem that the more complex a model we are sent the nearer F/A the answer will be.