One thing I have not yet figured out is how to confidently & efficiently make decisions about wall thickness of Al/PC/etc. tube for various applications. My approach now is to do research here and also look at some comparable applications on past robot CADs published by various teams. And then I’ll make an educated guess. I’d like to do better.
From reading other discussions, it’s apparent an experienced person will often be able to say, “1/16-inch 2x1 6061 will be fine” for this and “better use 1/8-inch” for that.
Is anyone aware of reference information that will help a less-experienced person make tube thickness choices? I could see something simple like a table of rules of thumb being useful. If not that, then it would be helpful to have a set of sources for strength & other material characteristics of the most common types of tube used in FRC and apply that information within engineering calculations. I have searched around, and I’m not having much luck (probably not looking in the right place).
If anyone is aware of helpful resources as described above, please share. I know some teams are moving to using the simulation features within Fusion 360 and other CAD tools. I’m looking into that in parallel. Thanks!
My general rule of thumb is as light as possible, unless I absolutely need the strength. On my team for aluminum tubing, that means 1/16" until I absolutely need 1/8". I know Ari’s new design calc has a beam bend calculator. I’ll see if can link it, although I haven’t experimented much with it myself
For some parts, you can use engineering formulas and FEA to figure out the dimensions of tube you need to stand up to the forces applied. This kind of analysis is good for things like climbers and arms where you can reasonably quantify static forces acting on the beam. You can use my the beam bend calculator on my design spreadsheet to see how much a tube will bend under a given load, but first you need to know the load that will be applied.
For most parts on the robot, the forces are mostly dynamic and hard to quantify. Coming up with accurate results for these parts is going to take a lot longer than we have for FRC purposes. That’s why you see most designers using intuition and past experience to determine these dimensions. A lot of good teams release CAD of their robot, and have points of contact for questions. My best suggestion would be to look over as many past robots as you can, then ask the teams what their experiences were with their choice (was it too thick, too thin, or just right). If you want to build experience quickly, that’s the way to do it.
In general, the answer to “how thick should this tube be” is “half as thick as you think it needs to be”.
We typically look at what the part is mating with. If we are going to be putting a lot of tapped holes into something, we go 1/8". If we are going to be putting bearings directly in the tubes, we typically go 1/8". Typically drivetrain rails are 1/8" because we weld our frames and we can’t afford to bend a tube since it’s essentially impossible to replace.
For polycarbonate, some of it has to do with application. We like to use polycarbonate for the natural spring, so we play with thickness to see how much spring we get out of it. Our intake was a 4 bar and we originally used 1/8" legs doubled up (one on each side of the 1/4" plate that had the bearings mounted in it), and although it never broke on us, we switched to 1/4" legs to make it a little bit stiffer, while still being able to take a beating.
Our team has learned with a lot of trial and error, although there are mathematical ways of figuring it out.
As a rule of thumb, most teams do .125" wall on the drive frame and .063" wall on all of the superstructure. Part of the reason for this is because the drivetrain takes extremely large impact shocks from the bumpers, and the other part is that these extrusions are often inconveniently located for replacement. Having done it once or twice before, I can tell you it absolutely sucks to try to replace drivetrain frame rails.
Meanwhile, most superstructures are within the frame perimeter and can typically handle these loads with grace. For any superstructure parts expecting high load, my team does FEA on them to find out if the wall needs thickening. For instance, we found that the elevator post for a level 3 suction climber needed .125" wall.
It’s worth noting that the common failure mode for extrusions, especially .0625" wall, is buckling. Buckling occurs when one or more of the walls fold up on themselves. Once a part has buckled, its strength is dramatically reduced since the load is forcing the walls to fold, rather than to stretch or compress as normal. While it takes some built-up intuition to find out what loads lead to buckling, my finding is that is most often happens when the robot encounters high impact shocks concentrated at small points.* A clever designer can put their robot together to reduce the frequency of these cases by shortening long moment arms, spreading out mounting features, and reducing the weight of parts high on the robot. Doing all 3 of those mean you can get away with far lighter structural components.
*As an aside, cases of high impact shocks concentrated at small points represent a lot of mechanical failures on FRC robots.
critical frame piece, go thicker (harder to get bent out of shape)
high on the robot, go thinner (helps keep your center of gravity low)
more prone to impact, go thicker (or go flexible!)
more force imparted through it, go thicker (supporting a climber, for example)
supporting a bearing directly, go thicker (bearing blocks can be a way around this)
Of course, I can think of all sorts of exceptions to each rule of thumb, and the rules don’t always agree with each other! In a real-world engineering situation, where a machine is expected to last through years of frequent use, there is a lot of analysis done using CAD models, and then more real-world testing. It’s not just a question of how thick to make the material, but also the exact composition of the material (There are many different types of aluminum, for example!) and exactly how it’s all attached together.
For the FRC world, experience is king. Some teams do go into analysis, but most just rely on experience to build something “good enough”. It might be over-engineered, with more material (weight) than needed, but it gets the job done and the extra effort to optimize it isn’t (in my opinion) worth the time and effort most of the time. I’d rather put that time and effort towards functional improvements, testing, and driver practice.
I’ll add on to the possibly disappointing sentiment that experience is the most useful for decision making in FRC, where loads can often be difficult to quantify (typically the largest loads in FRC are from impacts) and there is not time for extremely detailed analysis.
I will note, however, that when you can do detailed analysis, the experienced guess is often shockingly accurate. A roughly FRC-sized robot I worked on in college was constructed primarily from AL square/rectangular tube and had well-known maximum forces, so we did a lot of FEA. The material thicknesses chosen after FEA were very rarely different than what one experienced in FRC would have chosen on an experienced and educated guess. That is, designing experience is making educated guesses that I’m always surprised to find out are quite accurate.
Part of the reason for this is that there are only two common thicknesses of tube in FRC - 1/8" and 1/16". These can roughly be broken down into “very strong, but too heavy for most applications other than a drivetrain” and “safe to use for everything that doesn’t make alarm bells go off (climber, drivetrains (often), etc.)”
So, now that a lot of us have said experience is key, what do you (i.e., an inexperienced student, whether you are or not) do to gain experience?
Learn from the best - many great FRC teams release CAD files which you can learn from.
Learn from the experienced on your team - their educated guesses are likely more correct than you would think.
Do some math - If you work through the math (or use a calculator like Ari’s), you will get an idea of what is acceptable and what isn’t, and learn what factors influence that most.
Test mechanisms - When you have an unclear choice between one thickness and another, make sure to test the mechanism a lot, especially if you choose the smaller thickness.
Breaking stuff is the best way to learn. If you’ve got the facilities for it, try putting on your safety glasses and beating the ■■■■ out of some metal. Support some long-ish 1/8" and 1/16" bars at the ends and hang weights in the middle until they bend or break. Hit some things with hammers. If you put a 1/8" aluminum sheet in a vice and hit it with a hammer, how much does it bend? What about 1/4?" What does polycarb do? We’ve got a hydraulic press and we can force pretty much any part to fail, and see what kind of load it takes to do it.
It’s one of those rare but wonderful circumstances where “hit it with a hammer” is the educational, responsible thing to do. Take advantage of those circumstances.
I agree fully! I would be willing to bet this is the fastest way to get this experience everyone is talking about. I should definitely have included this in my list. The other things are what one can do when they don’t have access to scrap metal, a vice, and a hammer.
You can take a college statics class, maybe a dynamics class, then a mechanics of materials class, and then you’ll have a pretty good understanding of how it works.
That or continue as you are doing now, and also experiment, see what it takes to bend/break/damage specific materials in specific designs.
But putting numbers to it really requires doing some engineering. If you study the basics of calculating loads in members, then calculating stress in material due to those loads, you should be able to figure it out.
This past season, we tried out pieces of 2 x 1 for our cargo ball arm by leaning the pieces so one end rested on the floor and part of it rested on a table and then leaned on a point in between to see how much flex we got.
We also tested some 1/2" aluminum hex shaft by clamping one end in a vice and twisting the other end with a torque wrench with a 1/2" socket. It didn’t break but the elastic deformation was too much for our application.