Skip to main content

TUM Hyperloop Interview

The Technical University of Munich (TUM) Hyperloop team managed to set a new speed record with their pod IV. Toni Jukic, the team leader of the TUM Hyperloop team, giving insights into the development process of their winning pod. 

 

Congratulations on setting yet another speed record in the Hyperloop Challenge! How did you manage to set this amazing record?

We participated in the SpaceX Hyperloop Pod Competition before. This means, we have already gained quite some experience. Further, the competition rules give little restrictions on the design of the pod, which is a wonderful situation for an engineer, because it allows us to be very creative. We believe that we have now found an optimal concept. Our focus was on having a high power-to-weight ratio. We achieved this by using methods from lightweight construction and developing a powerful drivetrain. The pod weighs just 70kg and is propelled by eight electric motors with a total power of 320kW, which is more than that of a Formula E car. However, one thing is having high quality components, and another is being able to integrate and operate them properly. The experience gained from competing in previous years played an important role.

A big challenge when competing in such a competition is testing. You only have one pod and therefore, need to assure that you don’t damage it during tests in early stages. Also, it wasn't possible to fly to Los Angeles to conduct tests. How did you assure that the pod is working as intended with these restrictions?

That’s absolutely correct. The actual track is 10,000km away in Los Angeles, and the first and only time we can have a run with full power is during the finals. It’s like a rocket launch: Nothing can go wrong. There are no second chances. Therefore, it is even more important to thoroughly test the pod in advance. Starting with FEM, CFD, and system simulations including testing single components on test benches that we developed on our own. For example, a brake test bench for a Formula 1 cars goes up to 400km/h, which is too low for us. The same goes for the test bench of the propulsion units. We also had to invent a way to determine the discharge behavior of the batteries.

After thoroughly testing all subsystems, we started assembling the pod. In the final phase of testing, we tested the pod on our own 300m long track on campus, which is of course not comparable to the actual track in Los Angeles.

One more important question was how to advance the development of the electronics system and software, even though many mechanical components were still in an iteration loop and many sensors and actuators undefined? That’s where the Speedgoat machine came into play.

 

How did you use the Speedgoat Baseline real-time target machine?

One of the main goals was to test the electronic control units (ECU). The task being to read sensor data, process it, and send commands to the actuators in the pod. For that purpose, we created a detailed vehicle model with Simulink. We emulated all sensors and connected the I/O ports of the Speedgoat Baseline real-time target machine directly to our ECUs.

The most important sensors modeled were photoelectric sensors, hall switch sensors, and pressure sensors. Photoelectric sensors and digital outputs were used to measure the attitude of the pod, then we modeled hall switch sensors to generate pulse width modulation (PWM) signals with varying period to measure the velocity, and lastly pressure sensors and analog outputs were used to measure the brake chambers. Also, the commands from the ECUs were captured by the machine and fed to the vehicle model like the PWM signals for the motor controllers, analog inputs to control the pneumatic valves and digital inputs for opening and closing various switches.

Finally, there was the possibility to communicate directly with the ECU using the CAN port. It was possible to log every single input and output, and the collected data was then sent to SpaceX to show that the ECUs behaved as expected. It was crucial to test the functionality of the control units and to check whether the state transitions and abort criteria worked correctly.

 

Very Interesting! Would you say that using a real-time system from Speedgoat was one of the main factors that enabled you to set yet another speed record?

It’s obvious that the maximum achievable speed is limited by the built-in components. Not even the best design is worth anything if you can’t control the pod, especially if the software is not working correctly or if it shuts down due to inconsistent sensor data or noise. By testing with a Speedgoat real-time target machine we could simulate many scenarios without having to run the pod, which is awesome! In the end, we were able to identify and fix potential sources of error in a very early stage of development. This is how we gained confidence in our system and what allowed us to operate it without hesitation as well as learn its performance limits. Therefore, the real-time system from Speedgoat was a crucial factor in our success.

 

What did you personally like the most about working with your Speedgoat Real-Time Target machine and Simulink Real-Time™?

As an engineer, when prototyping and testing, you often just want to get started without losing time. This is especially true, when it comes to projects such as the Hyperloop, where time is very limited. On the one side, you have the machine itself which is very fast and easy to set up. There is no need to install endless packages and programs. Connecting the machine to the PC via ethernet couldn't be easier. It takes just a matter of minutes. On the other side there is the Speedgoat Simulink real-time driver block library. The great thing about that, is that it works just like any other Simulink library. You’re working in a familiar environment, blocks with input and output ports and their function are practically self-explanatory. So even inexperienced Simulink users will have no problems using it. If something is still unclear, you can check out the documentation, which is extremely good and detailed just like that from MATLAB-Simulink. But don’t be deceived by the simple operation. After all, you have an extremely powerful and reliable tool in front of you.

What’s the plan for next season – which improvements do you have in mind and how feasible are they?

We will definitely participate in the 2020 competition. Our current concept has proved very successful, so we will stick to it. We are very proud of our battery and drive system. A lot of development work has gone into it and we want to keep this without making any major changes. Perhaps we will make some optimizations in the pneumatic system to engage the brakes even faster. Further, at our speeds, we reach a point where the quality of the track has a big influence on the performance. Therefore, one focus will be to make the pod more mechanically resilient and less sensitive to unevenness. Our goal: half the speed of sound!

 

 

Sponsoring

Speedgoat encourages students to realize engineering and scientific breakthroughs by supporting students in science and technology competitions around the world. Together with MathWorks, we empower our student teams to overcome technological and scientific challenges.


Learn more about the student projects and our sponsoring activities