Was just wondering if anyone is aware of any papers or docs on good pocketing practices that are based on either testing or FEA. “Oh that looks good” or “add more webs” aren’t exactly based in any hard fact. In my own experience tweaking with FEA, simply changing the termination point of a web to a different location can allow for much lower stress distributions (and thus longer part life or lower weight).
I do have FEA packages at my personal disposal (SW and ANSYS) to verify my own designs, but I was looking for something that just gives good strategies.
I was not able to find anything with a precursory search through CD (and I kinda assume you searched this before you started the thread, but I was curious), but I think if you look up truss design, you might find many more resources in that area. While truss bridges are NOT the same as gearbox or drivetrain plates, or whatever application for FRC, a lot of the theory behind truss patterns can be applied to pocketing patterns.
Another thing to check might be some of the softwares available for topology optimization. I haven’t worked much with topology optimization, and I doubt many in FRC ever use it while designing their robots, but I imagine you could use topology optimization to design better pocketing patterns. And if you can do it in software, I’m sure you can find some guidelines or best design practices to write down.
I just watched their tutorial video for this, and it looks amazing! I have used SolidThinking Inspire(a similar tool) before, but not enough to become proficient. Shape Generator looks much easier and the integration into Inventor is a definite plus. I will have to try it out when I get the chance.
Honestly, build season is so quick and I’m so busy that I’d rather just over-build the mechanism a little bit and go conservative with the pocketing than do rigorous analysis to optimize the part for weight. The few extra ounces this saves is rarely worth the design time. If you have a general understanding of trusses, how your mechanism is loaded, etc. and you are conservative with your pocketing you should be okay.
It’s very easy to misuse FEA tools to give you a false sense of confidence, so I would be careful with them.
FRC pocketing is something I heavily rely on anecdotal experience for (and occasionally remaking parts lighter or stronger in rev 2).
Even if you application of FEA is perfect, it’d be very difficult to quantify the forces involved if the part is exposed to impacts or the dynamic reactions of attached mechanisms. This pretty much just leaves small internal gearboxes that don’t take much load other than their internal torques (which don’t need much plate strength at all) that are easy to analyze.
I’ve used FEA in FRC some for A-B comparison of methods combined with anecdotal experience, but I’ve never determined strength required in a given plate/gusset then FEA’d to drive the pocketing to match.
That’s the page for SolidWorks’ information about doing simulations on entire assemblies. Theoretically, the way to get the most accurate FEA data is to create your entire assembly (fasteners, sheet metal features, chains, belts, weldments, the whole nine yards) and then run your simulation on the whole thing. However, in my experience, that’s more trouble than it’s worth, especially in a 6-week timeframe. What I do these days is just take it one critical part at a time, like the main drive rails or the gearbox plates and just place constraints and loads as best I can. For that, I run lots of different simulations, including drop test, static, and cyclic simulations. For me, running 10 (maybe not actually 10, but it’s a nice number) simulations on each critical component is easier and more accurate than running one giant simulation on the whole assembly, especially while I’m still designing. Granted, I’m not the most familiar with FEA yet, so take my advice with a grain of salt.
Oh, and remember that accurate FEA data and inaccurate FEA data often look very similar, and it can be difficult to get all of your constants and constraints and loads just right.
As someone who teaches SOLIDWORKS FEA and does consulting work with it for a living I can say this…
Students will only have access to static FEA analysis in their student SOLIDWORKS version so you need to check the following:
-You are using a linear static material (no rubber or non-linear polymers)
Ex. HDPE is ok but PE is not because PE plastic does not have a yield strength. Any metal alloy you use commonly in FRC are covered here but you need accurate material models from the manufacturer, MATWEB.com, or somewhere reliable.
-The loads and deflections are linear. Double the load, double the deflection.
-Loads are applied slowly and gradually. No dynamic loads.
The software can do all of these things above with the correct version of SOLIDWORKS Simulation! But not with a student version!
It is also best practice to simplify. You could try to build the entire assembly in a simulation and wait 3 days for it to mesh and run to get good results or you can take it part by part and simulate that parts environment instead. The highest number of components I have ever tried to simulate simultaneously was about 600 parts. It took about 4 days to run on a $80,000 machine that most people could not fathom owning. I don’t recommend that route if you are just starting out with FEA.
With that being said, I have actually started to produce content at work relating FRC, SOLIDWORKS, 3D Printing, and FEA analysis. One of the videos on FEA should be posted this week. When I have the link I will share it here.
As promised, this was a webinar I did featuring our robot, 3D printing, and some basic FEA analysis that we used. Sorry it took so long to get up there.
Just to be clear, FRC was not the target audience here. SOLIDWORKS users were. There are some additional robotics topics on the channel as well if you guys have not seen those yet from my other postings.
If there is some additional interest in FEA within the FRC community I might tape some examples with my students doing the work and post them for you guys.
I know we use it but is this something you guys would use?
We call it swiss cheesing. Often it goes on the scale and gets the hole saw treatment until it’s light enough. Nothing scientific. :ahh:
As the students design, cam, and shop skills have improved we have done more milled triangles into our plates. This saved half the weight of our side plates. Can’t say they were engineered but rather guess-gineered.
The isogrid looks attractive due to considerable weight savings but it does really drive up the material cost and machine time. If your milling away 90% of your material you better have deep pockets. A lot of isogrids and orthogrids appear to be on lightweight pressure vessels, like air and space craft. Is this because they work well in tension? I wonder how well this transfers to arms and such used in FRC. It would be neat to run some through FEA and see what it does.