As many new areas are going into District format, I would like to remind folks about designing for durability.
In the past, if you went to one regional, you could expect 8 qualifying matches and hopefully at least 4 Elims matches (making it to semis). Basically, you could plan on 12-20 matches.
Now, playing districts, you might see 12 qualifying, and have an even higher likelihood of making it into elims. Plus you will be doing at least 2 events which means a high likelihood of 30-40 matches. Plus another 12-20 if you make it to a district championship. plus another 20-ish if you play a third district. Plus another 12-20 if you make it to championship…
Last year FRC33 had 92 official matches at teh end of the World Championship. Throw in a few practice matches at each event, and they likely exceded 100 matches.
Then with offseasons, teams may see another 20-30 matches. Suddenly you have a need for a robot to last 150 matches instead of 15-ish…
Total run time changed from under 1 hour to around 7 hours.
What (if anything) are teams doing now that they are seeing a lot more matches?
Our team generally tends to make things very durable (arguably too durable). Our frames are made to stand up to the most abuse imaginable (even a pyramid fall). We make our drive-trains low maintenance and easily swap-able (we usually have an entire fully assembled extra set of wheels/motors/gearbox/bracket assemblies on hand every regional). Our systems are independent and modular to allow for easy repair/replacement. We have generally been going to 2 regionals and an off season competition (worlds eludes us every year) and our robots end the season more than capable of spending 2-3x longer under stress.
During 2013 and going into 2014 we are keeping maintenance in mind.
This past year our shooter was very easy to maintain with an easy access panel for changing shooter wheels. Our shooter motors were mounted to a separate plate that we could undo four bolts to drop out the motors and swap it with a spare assembly if a motor died. Our climber was also attached using 3 bolts so a replacement was an easy swap.
Our drivebase wasn’t the easiest to keep up over the course of two regionals, Champs, IRI, and 2 off-seasons so we are working on a simpler base that is reliable but makes it easy to replace parts.
This is one of the main reasons I argue for a practice bot. Not only does it provide spare parts when something inevitably breaks, it provides a much better sense of how much wear and tear your parts will see (just like a durability car). Practicing at least 4 or 5 hours a week (as most teams who build 2 robots do) means you’ve hit or exceeded your competition bots life cycle in 1-2 weeks. If the design failed after 5 hours of practice you know what you have to watch for on the real robot.
This hits on a very important point IKE hinted at but didn’t outright say. Yes you’re competing in more matches, AND the turn-around between your matches is shorter. I know there have been times in Districts where we only had a four match break. So easy access for quick repairs is something we’ve started to highly prioritize.
After building FRC robots for years it still amazes me how cars, that only cost 4 times as much can stand up to 100’s of thousand of miles of driving over years and still hold up reasonably well when our robots are always barely hobbling along by the last off season.
A few things to consider first is that most of the parts FRC teams use are rated in industry to be used much longer then they will ever see in FRC. For example ball bearings have life ratings in the millions of rotations (As long as they are used within load and speed ranges) and pneumatics are rated in miles of equivalent travel. So we don’t need to change much because we are already operating well within the expected life of those parts.
Bumpers also made a huge difference in longevity.
If I had to pick one thing, i’ll echo what has been brought up already and that is access access access. But we should all be designing that way whether you plan on playing 20 matches or 200.
Last year (our rookie year), during our regional, we never had to fix anything until elims.
Of course, during elims is when our slider attachment method broke. LOL.
At Worlds, we only had 1 instance where we had to repair some damage due to a crash with the pyramid. That doesn’t mean we didn’t have any problems at Worlds, but ours were almost entirely field/software issues.
For the most part our bot is still running at nearly 100%.
So going back to what the O.P. asked, if you design for durability you should have fewer issues.
High volume production cars (50,000+/yr) work as well as they do for as long as they do due to testing, refinement, and good processes. Over the years, as problems arise, they build new tests to run cars through. For instance, several decades ago, there were a lot of suspension failures in a particular area of Mexico… So the proving ground engineers brought Mexico to Chelsea Michigan. Here is a link about them “Fixing the broken road” http://blog.chryslergroupllc.com/blog.do;jsessionid=21659A6728297351755767EC9663E034?p=entry&id=1303
Cars will go through various “life tests” that attempt to simulate the life of a vehicle in a short timeframe. Sometimes these are referred to as “HALT” or Highly Accelerated Life Tests. In the short time span of developing a new car, they will run a couple iterations of early product through these tests in order to verify that they would be good for customers. For a car, there are thermal test, vibration tests, electrical interference tests, several different body/chassis durability tests, several component-subsystem-vehicle level powertrain tests…
They also had us (when I worked there), design for 150,000 miles. What that used to mean was for a normal distribution of failure modes, but the average part life was 150,000 customer miles. This typically meant few failures under 3 years/36,000 miles, but due to normal distributions, you would frequently have some issues creeping in around 90,000 miles. This is in part why a used car looses so much value around the 100K miles.
The final round of testing on “production” cars typically runs into the first 4-6 motnhs of production. That is why you will often hear about a new car getting service bulletins to add or modify a brake line clamp and a wire harness rub shield. It takes that long to get those final cars made in the plants and tested, and reviewed. Guards and clamps will be added to reduce the risk of failure due to rubbing through (hose, wire insultation or …)
I find it interesting the “issues” taht sometimes arise with customers. For instance, My father-in-law had a trunk seal wear out on his car with only about 40K miles on it… I did a quick estimate on the number of opens/closes I would think a trunk would see in the course of 150K mile vehicle life. 2/open/close cycle on 50% of trips with an average trip around 10 miles thus about 15,000 open close cycles/150K of mileage. I then asked him about his usage. He opens an closes his trunk to put his lunch box in the trunk. this means 4 open/close per day. He only uses that car to commute to work, thus he was at about 4X my estimated usage… or another way to put it, he had 160K worth of open and closes in 40K of driving… (sometimes you get lucky with your math). After that discussion he felt better about his seal wearing out so quickly…
Never underestimate the price breaks of super mass production. Also, engineers spend A LOT of time just shaving off any dollar they can. For example, in 2012, over 400000 Ford F-150s were sold. So for every dollar off their BoM cost saves $400,000, well worth spending the time on. However, in FRC, price has the habit of playing second fiddle to other things like size and weight. If we REALLY tried to, I bet we could really drive down the price of an FRC robot. It’s just not a priority.
You had to go pick on my product, didn’t you . Companies like Ford pay a fraction of the price that a person would for most components. When they work out a contract with a supplier, the supplier is guaranteed a certain number of parts at a certain price. As a result, they can supply at a much lower cost because they know the amount of product they will sell.
Some commodities like fasteners can be an order of magnitude less expensive for a major manufacturer.
$500 robots sound nice, don’t they?
They’d also be made at a rate of a thousand or two a day…
Not that the other explanations aren’t very true as well, but personally, I also try to crash my car less than I do my robot. (I also drop it off fewer pyramids and drive it off fewer bridges, even accounting for the vastly different velocities and distances.)
You guys spend more than $500 on your robots? If we add the price of all three years robots our team wouldn’t hit the $3500 season limit… This year, I actually listed KOP items (including the cRIO) on our BOM to try to make it and still didn’t get there.
Maybe we need to step up our fundraising a notch or two to find out what we are really capable of…
At the same time, we’ve never had a major breakdown that couldn’t be fixed with a little time or a few rubber bands. We had two repairs this year, a pneumatic solenoid block and two wheels. The pneumatics weren’t properly tested and we had a defective solenoid, and the wheels were the result of a 6 foot dismount from the pyramid. Even then, we replaced the wheels because the had a small crack and we had an hour to kill, not because they were destroyed.
I think in average design for these robots, teams just need to think about what they are asking the machine to do. If it is strong enough to do the task, it should be able to handle any abuse a 16 year old with a joystick can throw at it.
Yes, they should last but remember, this program is about high school kids with hammers. Proper installation (no inner race pressure) and support pressure is a big part of this. ALSO TEAM STOP APPLYING SO MUCH SIDE LOAD TO YOUR CYLINDERS, and if you do pack 4 spares.
racecars don’t last long, if your cim motor can outlast a Civic then your prolly not pushing it to is max output
Estimating the full value of a FRC robot is a fun exercise! To get close to the real amount you should also count the volunteered labor. Each dedicated engineering mentor is probably donating $5-10K of his or her time annually.