Gaining Traction with Markforged: Overmolding and Casting 3D Printed Wheels

Editor's Note: This guest post is written by Charles Guan, an MIT Mechanical Engineering graduate and former MIT machine shop & design instructor. He’s also the builder of the robot Overhaul on ABC’s BattleBots. He works as a engineering consultant in robotics and transportation, and is sponsored by Markforged to compete in BattleBots and other robotics competitions. Check out his website and Equals Zero Robotics for more!

Wheel Design

Wheels. With over 6,000 years of history behind them, you’d figure we would have gotten them perfect by now. As I found out the hard way during BattleBots Season 2, this is far from true in many ways. You can actually reinvent the wheel and get it wrong. I designed Overhaul 2.0 with a powerful and stable square 6-wheel-drive platform with the intention of using speed to my advantage in capturing opponents in its grappling lifter. I picked the wheels based on what I knew: medium hardness industrial caster wheels with thermoplastic elastomer (TPE) treads that had been used in robot combat since the beginning. Putting 16 horsepower motors through these wheels in a robot designed to reach 19mph in 1 second was uncharted territory.

This didn’t work out in my favor in the arena at all. The TPE wheels began coming apart, literally melting from the power being put through them. Their slick tread glided over the loose debris and particles in the box like the bot was driving on ice. My 16 horsepower of brushless drive motors was pretty much a total waste as a result. If you watch my match against Beta this past season, you’ll see that I mistook BattleBots for the D1 Grand Prix for a good part of the match.

After the BattleBots Season 2 tournament, I was left wondering about next steps. The wheels would need to change to put the power of the bot on the ground. This is where I began researching castable rubber compounds. I didn’t mind making some molds, pouring the material, then working on something else in the time it takes to cure. During Season 2 of BattleBots, I had watched some competitors – including Beta – pour their own urethane wheels on-site.

I knew nothing about tire and wheel design, nor molding and casting. Talk about a shot in the dark! Luckily, I had cornered some Smooth-On company representatives at the Detroit Maker Faire over the summer, and talked to them at length about their thoughts on which of their products was suitable for making traction wheels. Smooth-On is a company primarily geared to mold-making & casting for special effects, costumes, and props – not industrial uses. However, their distributors are nationwide so their materials are very easily available for experimentation – just 15 minutes’ drive away from Boston. The mission: learn the resin casting workflow by producing custom wheels for a 30lb-class scale model of Overhaul that I was designing at the time for a Fall 2016 competition.

Designing The Mold

The reps pointed me to two materials that had favorably high tear resistance and tensile strength, the Reoflex and Simpact series. These two traits are desirable in a wheel compound, where the rubber is constantly being sheared through its tread thickness and dragged along the ground. I got my hands on a sample of ReoFlex 50 (for 50A durometer, which is reasonably soft like a shoe sole) and set to designing the mold and hub.

‍After Overhaul’s performance in the box, I was pretty convinced some kind of debris-clearing feature was needed on the surface of the wheel. While we’re not trying to channel water or claw through mud like with car tires, there was still loose detritus and paint flakes from the arena itself to contend with. To start, I made a simple spiral tread pattern. In no way was this optimized – I just wanted to get to the molding stage quickly for now. I figured the helical grooves would tend to push debris off to the side away from the contact patch. The exact orientation and number of grooves is obviously still an unsettled science if you’ve ever gone tire shopping.

Next, I was faced with the task of designing the wheel hub. I took a look through my box of scooter and skateboard wheels for inspiration. They all had a common feature: through-holes or slots which the urethane flowed around and into. I was very keen on ensuring that the tread material could stay on the hub even with adhesion failure. I modeled a fairly basic cylindrical hub, but with many through-hole features in a tapered central rib to give maximum area and cross section for the urethane to stick to I chose Onyx over regular Nylon for the hub due to its higher rigidity and strength, factored with adhesion properties. The increased rigidity would allow me to make the wheel hub lightweight, while the resin would adhere better to the microscopically rough texture of Onyx over glossy nylon.

‍Since the Tiny Overhaul would also need some smaller front wheels, I used the parametric features of Autodesk Inventor to quickly generate a 2-inch model. This mold had a fairly constricted area to pour through, so I was interested in how the material would behave.

I printed a test mold using plain un-reinforced Nylon on the Mark Two. The mold was very sparse – 4 walls and 4 roof & floor layers, but only 25% fill, printed on the coarsest layer setting of 0.2mm. I modeled registration pin holes into the halves to ensure alignment.

Pouring the Mold

The mold halves are designed to be held together with a single regular hose clamp. Inside, the wheel hub sits on an adapter bushing to keep it centered in the mold, and this adapter bushing has a through-hole to the underside of the mold halves so I could run a retaining nut and bolt vertically to seal the bottom face of the wheel hub against the mold. Prior to assembling this, I coated the mold halves with spray-on mold release.

‍At the advice of friends who had more extensive molding & casting experience, I borrowed a vacuum pot to pull all the air bubbles out from the mixed resin. Smooth-on advertises their resins as “minimizing trapped air”, but watching the mixture boil over showed me that this was less than effective. Trapped air bubbles would reduce the integrity of the wheel tread since it would present many interruptions in the material. I degassed the cup of mixed material instead of putting the entire mold with liquid resin in the chamber, since all of the air inside the mold halves from being printed hollow would likely deform them or even leak and bubble out slowly.

It took about 2 minutes to pour the 3 inch wheel mold – I tilted the mixing cup just enough to maintain continuous liquid flow, letting the resin settle into the bottom and not pile up or crest over the mold. Letting the resin “find its waterline” is how to avoid trapping large air bubbles in the part.

I poured both a 3″ wheel and a 2″ wheel as a starting batch. After the first two wheels at least emerged solid, it was time to engage in some production. I printed additional molds such that I could pour 4 or more wheels at once – one of my mixing cups was perfect for two 3″ wheels and two 2″ wheels, so that became the standard.

The Curing Process

Most resin manufacturers recommend an initial room temperature slow cure, then heating the material afterwards to strengthen the polymer crosslinks more. For example, the datasheet for Reoflex 50 says to heat the material to 65 Celsius for 4 to 8 hours. I used a heated chamber at approximately 70 degrees Celsius: the new wheels that emerged from this heat-accelerated cure were definitely tougher and more resilient to the touch.

The result after a week of experimentation was half-dozen or so of each size wheel. I’d run through my supply of ReoFlex 50, so I determined this was a good place to stop and make sure the wheels actually worked in competition before spending more money on material. I got into a nice cycle the last few days before – I’d pour the molds as the last thing I did at night before leaving, demold them in the morning, and then bake them over the course of the day.

And that’s what they look like installed in the bot. If you’re interested in the development of “30-Haul” as we nicknamed it, you can check out its build thread on my website.

So what’s next? Now that I have the production process explored and a basic design in front of me, it’s time for the real science to begin. Stay tuned for Part II of this series, where I try making wheels of different compounds and testing their traction on a painted steel floor. For now, here’s what these wheels looked like after a few hard driving matches!

Learn more about Markforged printers and how they can suit your applications!