More destructive air tank testing from 95

We managed to get our grubby little paws on some black air tanks and found a suitable location to do destructive testing of a metal tank.


The results are not too surprising, though we’re not sure how much of a difference temperature made. Clearly our tests are not full of scientific rigor, but I think they show the fundamental difference between metal and plastic tanks quite well.

Enjoy. And as always: let us know if you want to see something else tested!

-JamesCH95 & Andy A.

Science? Nahh you just like blowing things up don’t you :smiley: Good stuff though!

Nice work! Now I think we just need to see the non-clippard plastic tanks tested.

From this video, it’s clear that Clippard’s updated tanks (the black ones) can explode just like the old white ones. They removed the screw fitting, which removes one significant source of stress on the tanks that had been proven to cause issues, but they didn’t remove the real worry - the failure mode of the tanks.

Personally, I don’t really care about the rate of failure for any tank… it’s the specific failure mode that concerns me. The metal tank in the video pretty clearly has a better failure mode than the plastic one - it didn’t send shrapnel flying all over the place.

Given the abuse we put our robots through (I’ve seen things bent and broken on the field that you never would have expected to be damaged!), it’s really surprising we haven’t seen more of these plastic tanks explode. I know that my team will never again use plastic tanks on the robot - It simply isn’t safe enough to allow them to.

The damage to the metal tank almost looks like it was punctured by the plastic tank shrapnel, and didn’t fail on it’s own. The odds the the two tanks failing at the exact same time is pretty much zero.

So unfortunately this test doesn’t show how or when metal tanks will fail. I guess you have to do more destructive testing! :stuck_out_tongue:


Is it a crime to like science when it involves blowing stuff up? :smiley:

Well, when I shot the metal tank, the plastic tank was right behind it, so presumably spall and/or fragments from the bullet hitting the metal tank caused the plastic tank to fail at virtually the same time.

It would take some serious damage to make a metal tank fail. Here are a few pictures of the aftermath:

A .22LR ‘mini-mag’ round impact. No fractures. IMO it is highly unlikely that a plastic tank would ever take out a metal tank.

A 7.62mm bullet sitting at the impact site on the metal tank.

The bullet impacted almost tangential to the inside of the tank, in this picture you can see how bullet fragments spun around inside the tank’s ID.

We’d love to! Anyone want to donate to the cause?

Agreed on all points. While the black clippard tanks are an improvement over the white ones the potential for a very serious failure still exists. There was at least one team at NECHAMPS that elected to add lexan shielding around their exposed tanks after some tubing got ripped out of them during aggressive play. The line between tubing getting pulled out and the tank getting crunched seems pretty hairy to me. Under very slightly different circumstances that could have resulted in field crew getting hit with shrapnel.

If FIRST is to continue to allow these tanks they should, at a minimum, dictate rules about locating and shielding them from any possible robot/robot interaction. But if you can rationalize a rule set about that then, surely, you can rationalize just not allowing their use in the first place.

You guys need to get some slow motion footage of it bursting. That would be pretty awesome.

The moment that there is an affordable and non-terrible high-speed camera available, I’d love to. Alternatively, if someone would lend us one, that works to.

At the moment we’re SOL.

James, 125 has made heavy use of GoPro’s in a high speed mode for diagnosing shooter issues. It’s not OMG 1000FPS at 1080P but it’s better than nothing. If you promise not to shoot it I’m more than willing to loan you mine for any more tests you want to do.

I’d also like to see failures caused by realistic impact. While I am concerned about the failure mode I question the relevancy of testing. If we could focus on impacts with realistic energy and contact it might be more telling.

Thanks Andrew!

We shot everything on my GoPro, 720P @ 60FPS. Can yours do better? It might not take a whole lot more resolution to capture some of the really interesting behavior.

Do you have any thoughts on what a more realistic test would be?

FWIW: the pellet gun uses around 5% of the kinetic energy (0.002lb@1000ft/s) of a 140lb robot @ 16ft/s, so there is plenty more energy available from robot-robot interaction than what we used to initiate failure.

I don’t think so. I can check when I get home tonight.

I was thinking a more realistic test might involve blunt force. Maybe fail it with a spring loaded (or pneumatic actuated) hammer to simulate an intake slamming into it.

This article is relevant to this discussion… emphasis mine.

The rate of loading for a plastic material is a key component of how we perceive its performance. High strain rates—events that occur over a short period of time—tend to favor the elastic properties of materials. Elasticity is associated with load-bearing performance as embodied in properties such as strength and stiffness. However, low strain rates—events that occur over a longer time frame—favor the viscous or energy-damping aspects of material behavior. Viscous flow is associated with energy management, often referred to as impact resistance or toughness.


While these tests are interesting they are not scientific by any stretch. We (456) are currently working up a more scientifically rigorous test plan to be conducted at an actual blast containment test facility under controlled conditions. We will be meeting with Clippard at Championships to discuss these protocols and will initiate tests early this summer. We endeavor to test as many different variables as possible including all of the various plastic tanks from different vendors as well as the traditional metal Clippard tanks. We have a long list of tests we hope to conduct including impacts, mounting methods, over pressurization, etc. Rest assured these tests will NOT include setting up a pressurized tank in a field and shooting it with a pellet gun or high powered rifle. They will include scientific instrumentation including very high speed cameras, accelerometers, etc. The testing will be conducted this summer under the oversight of engineers who test the limits hardened structures for a living, a FRC Lead Robot Inspector (LRI), and a FIRST Technical Advisor (FTA) all from the viewpoint of how FRC teams are actually using these tanks in a competition environment. We are in a possibly unique position (facilities, expertise, manpower, and financial resources) to conduct these tests as a service to FIRST and the FIRST community. Before anyone jumps to any conclusions that ALL plastic tanks are simply unsafe and should never be used in FRC let’s see what some actual scientific and engineering data says over the coming few months. Please stay tuned.

1 Like


We could drop ~140lbs from some height onto a tank, with some representative strike surface to hit the tank (1x1 aluminum tube? 1/4in aluminum plate?). We could also strap a tank onto one robot and crash it into another robot, but that might get more than a little hairy.

There would be much to consider in making a ‘realistic’ test.

Very good read.

Figure 1 in the article is quite applicable to FRC (maybe not air tanks, strictly speaking) because it shows the difference in performance of a 5ft/s impact and a 15ft/s impact, two very real speeds likely to occur on the field.

Figure 3 shows PP (tank material) under various strain rates, but none of which are remotely close to strain rates that could develop from a robot-robot impact. However, it does show how strength stiffness change with strain rate. It would be interesting to compute the strain rate sensitivity and see what might happen at impact-level strain rates.

I noticed the 7.62 slug had an exposed lead tip. I am pretty sure that would be on the list of prohibited items for your robot. :]

Interesting study. thanks.

That’d actually be gray paint which denotes something about the bullet weight and intended use. For the curious, I believe it is a 184 grain FMJ bullet with a lead and steel core. Muzzle velocity is around 2400 to 2500 FPS. It was loaded in Hungary in 1977.

I agree- not FRC legal!

This is the part where I point to my sign which reads “ignorant software engineer”. I think what I had in my brain was less of a more realistic test more of a more realistic appearing test. It’s REALLY easy for me to rationalize away the risk right now because to me a bullet is far more destructive than our robots. Logically I know it’s not but it requires and understanding of physics. A more realistic looking test would remove even that interpretation.

TL;DR - disregard if your target audience is people who actually know stuffs.

You make a very good point. I think this is where I am going to take a step back and remind myself why Andy and I did this in the first place (and state it directly if it wasn’t obvious to everyone):

  1. Show how inadequate most of the proposed mitigation solutions were.
  2. Show, in a way tangible to a layperson, how much energy is contained in a storage tank.
  3. Show how far shrapnel is thrown after a tank failure.
  4. Raise awareness of the potential hazards associated with the use of plastic tanks.
  5. Hopefully spark a change in the FIRST community. This was the big objective - and I consider it accomplished. I know several teams changed out their white tanks, and I’m pretty sure that Clippard’s tank exchange program starting less than a week after our first video was not just coincidence.
  6. Have a little fun and provide a little entertainment.

[this is where I leave on a tangent]

We never set out to be scientifically rigorous or to test failure mechanisms. We stated as much in our videos and posts on several occasions. We know that failures can, and do, happen. We consider that aspect of this topic proven empirically by the experiences of numerous teams who have had tanks fail in service. If our videos motivate or inspire another team to do more rigorous testing, hey, that’s just icing on the cake as far as I am concerned.

So, I think we’re going to leave our efforts at the level we always intended: a demonstration, not a scientifically rigorous experiment. I’d love to test the Pneumaire and AndyMark tanks at some point, but thus far we have no takers to donate tanks.

A couple of top-of-the-line phones on the market have 120 fps speeds. It’s not amazing but it is double your 60 anyway. I know the iPhone 5s does it, as well as a couple of flagship Android phones. You can probably find someone who has one.