Another entry today in the theme of “the combat robot engineering you don't see”, this time from my teammate Peter Lombardo and his full-body spinning drum robot Bee Roll!
Sometimes in robotics specifically, or any project with a lot of moving parts, you realize solid expectations and assumptions were actually wrong. Time and effort can be spent addressing these things and that time isn't wasted, even though it doesn't address the direct issue. At the end of the day, this hobby is about learning things and moving bottlenecks, and I'm happy to say the journey outlined below achieved both of those goals.
Back in June of this year, I went to NHRL to fight the newest revision of Bee Roll in the freestyle tournament, and to practice driving in a safe place. I had just completed a host of upgrades (printing a tube utilizing AR600 printed-in-place insert disks, a new material for the internal printed structural components, and a new bearing block) all with the goal of avoiding what happened against both Pinevictus (October 2024 freestyle) and Eruption (February 2025 Round of 32) -- where a hit to the drive shaft shattered both the shaft and the printed bearing holders. With everything assembled and set up I went to do an extended drive test... aaaaand the bearing housing prints melted. Now, 3D-printed plastics are not known for their thermal properties, and PLA+ is no exception. But I wouldn't have thought that a few minutes of driving in circles would pump enough heat into the pods to start turning my internal structure to goo.
Once this glaring weakness was identified, I swapped out what I could and got my fight in, totaling the frame from the inside out. The material had to be changed, but also I felt that the spinning nature of the robot would lend itself well to some sort of passive cooling solution. The oval-shaped frame should have built in high pressure areas (inner diameter) and low pressure areas (outer diameter) that, given the opportunity, could pass fresh air through, and at speed! But how could I design such a duct, and how could it perform?
Enter SOLIDWORKS Flow Simulation! Utilizing these tools, I was able to simulate different flow paths and entrances, optimizing for 3d printable shapes, possible debris ingestion, and motor clearance. See below for images of the simulation results. I simulated running at 3,000 and 7,000 RPM, noting that while the assumptions I made may not be perfect they would at least be consistent. As such, the results represent a comparison between each other, not necessarily a perfectly accurate simulation of reality. Some of the lessons learned were that wider ducts pass much more air, the interaction of the motor to the inside of the tube kicked up a lot of turbulence and was sensitive to the clearance between the motor and the pod, and that more fillets are more better for the airflow. To make things faster, I started simulating in simple 3d forms, then moved up to a full mockup, and tested the final design in a printed model. After running for 3 minutes at full speed, the ducting reduced motor stator temperatures by 5 °F (relative to spinning the "wrong" direction), and 10 °F (relative to no ducts at all). It would likely be even higher if not for the reduced drag of the ducts causing a higher motor current draw.
And at the end of the day, it turns out that the main driver of the heat buildup was ACTUALLY because I assembled the wheel hub too tightly to the bearing block. Not only were the motor magnets offset from the stator (reducing efficiency and output power) but what reduced power it had was fighting friction as the motor can rubbed against the supporting needle roller bearings. So, that got a little more clearance on assembly, and I got to learn a little bit about aerodynamics, materials science, and motors. But I'll still keep the ventilation for the happiness of a few extra degrees of cooling. Ain't systems integration grand?
