Simulate Early, Simulate Often... In Rhino
We are using SnS now almost daily to help solve design problems with chair components. That alone saves us money; we can now discuss various options with our engineers in Germany rather than DHLing them samples and waiting weeks for a fail/nofail report.
This is an example from just yesterday. One of the BIFMA certification tests is a 134Kg load to the midback of a chair. It must sustain this for at least 1 minute.
Many older chairs were never designed for such stringent testing but must nevertheless be certified, even if to an older, 'softer' standard. In this case, the frame has been widened for 'heavier' persons and of interest is the back bracket (JBar) connecting the back frame to the mechanism.
The first step was to move the 134kg load to the top bar and adjust the load to suit the extra leverage (see first pic attached showing measurements of moments, i.e. 96Kg).
The second step was to model the existing bracket and compare it with an empirical test. The existing bracket yielded with only a 20Kg load!
The most significant change is a respecification of material thickness from 8mm to 10mm (ribbing may be another option but not all suppliers have the tooling). It was also widened as far as possible. We considered up rating the steel from 235MPa (plain) to 380MPa (structural). The problem there is that the rest of the frame starts to fail.
[In some components we have similarly uprated to structural steel, but only those elements that required it, which contains costs]
Halving the load, which more or less simulates the DIN standard, gives us something more reasonable to work with. The frames themselves have never failed in service, only the bracket. So if we can balance the strength of those, we're unlikely to have failures (see second image).
The next step is to have the part prototyped and test it physically. It will then go to R&D in Germany for them to test (pending discussing around DIN certification).
Something new we have started is to bring the SnS result in as a background image before PDF'ing the drawing (see third image). That may later be augmented by a list of hyperlinks to PDF'd SnS reports.
It is worth noting that since we started using SnS, not a single component analysed (in tens of thousands manufactured) has been returned with structural failure. Not one. It has however, meant a fair amount of time studying stresses (primarily Shiggleys textbook). Our overall reject rate has dropped from around 5% to 0.25%.
If ever I came across an application that earns it keep, this is it.
Ian, thanks very much for taking the time and sharing your experience! As many users expressed interest in furthering their understanding of structural analysis, what material in Shigley's you found particularly useful. (I assume that you mean this standard textbook.) I wonder if some of the material there may be illustrated with Scan&Solve models.
Also, what is at the top of your wish list?
One of the key learnings from Shigleys is understanding the different stress analysis models. For instance, what differentiates Maximum Shear Stress (MSS) theory from Von Mises? Why use Coloumb-Mohr? There are some things I still work out by hand, stresses through welds, for instance.
Wish list? A few come to mind:
(a) Ability to assign different material properties for testing, to simulate for instance, an over-moulded or reinforced component.
(b) Ability to select an area (or line) to measure microstrain, so that SnS can be compared directly with strain gauges.
(c) Ability to show the direction of strain (not just amplitude), something akin to using the isotactic lines of brittle lacquers like StressCoat. That would be useful for positioning strain gauges, so you know they're in near ideal tension/compression.
(d) In the interest of environmental certification, we are increasingly looking to reinforced polymers and composite materials. Those obviously have weaves, grains and so forth which can make strength highly direction oriented. That would obviously require some careful setup from the end user.
Easy enough to take some samples from Shigleys and run them through SnS. I have some more interesting examples than this with polymers but we first need to get them covered with IP protection and I need to machine out prototypes.
Thanks to your team too - every version of SnS includes useful enhancements.