← Theses

3D-Printed Mini Hydropower Plant that Senses Its Own Faults

Open topic · MA · Advisors: Bruno Rodrigues

Build a hydropower plant small enough for your desk, break it on purpose, and teach it to hear exactly where the damage is.

Where it starts: our ESP32 microphone nodes (B to E) around a bench test rig.
Where it starts: our ESP32 microphone nodes (B to E) around a bench test rig.

Why this matters

Turbines make noise long before they break. Collapsing cavitation bubbles crackle between 6 and 25 kHz, and a leaking seal hisses. The vibration sensors that plants rely on today, standardised in ISO 20816-5, only measure the structure they are bolted to and stop at about 3.7 kHz, so they cannot hear these early warnings. Microphones can, and our group has shown that machine learning on turbine audio detects developing faults reliably [1, 2].

The challenge

Knowing that something is wrong is not the same as knowing where. Without a location, an operator has to open the machine to search, and unplanned downtime at a 295 MW plant costs upward of 28,000 EUR per hour. Sound gives the location away: it travels at about 343 m/s, so it reaches a nearby microphone a fraction of a millisecond before a distant one. With several microphones and accurate clocks, those tiny delays point to the source. This is called time difference of arrival (TDoA). Inside a turbine it gets hard: echoes off the housing, loud flow noise, sound that travels through the steel faster than through the air, and clock offsets that look exactly like real delays. How well it works in a running machine is still an open question.

What you will do

  • Build the plant. Assemble a scaled Francis turbine from an open 3D-printable design, with a spiral casing and an eight-blade runner [7]. Parts are already printed and the hardware is provided.
  • Give it ears. Mount and synchronise a microphone array, and measure how precise its timing really is.
  • Break it on purpose. Print damaged or unbalanced blades, so every fault has a known size and a position known to the millimetre.
  • Find the fault. Extend our existing localisation code in Python, or the language of your choice, using delay estimation (GCC-PHAT) [6] and geometric solvers.
  • Push it to its limits. Measure how accurately TDoA locates each fault, how many microphones and which layout it needs, and when it fails.

Why you will enjoy it

Where it leads: one of our microphones at the 295 MW Rodundwerk II plant of illwerke vkw
Where it leads: one of our microphones at the 295 MW Rodundwerk II plant of illwerke vkw

  • Hands-on from day one: 3D printing, microphones, real signals, real experiments.
  • A real-world impact: the rig scales down our nine-microphone array at Rodundwerk II, a 295 MW pumped-storage plant operated with illwerke vkw, so your results feed directly into an industrial deployment.
  • No background needed: you do not need to know hydropower or acoustics. We bring the hardware, the code base and weekly support.

Previous theses in this line

References

[1] Khamaisi, K., Keller, N., Krummenacher, S., Huber, V., Fässler, B., & Rodrigues, B. (2025). From Noise to Knowledge: Acoustic Anomaly Detection in Pumped-storage Hydropower Plants. arXiv:2509.22881.

[2] Zhang, J., Fässler, B., & Rodrigues, B. (2026). Multi-Modal Acoustic-Vibration Array Monitoring for Anomaly Detection and Source Localization in a Pumped-Storage Turbine. 23rd International Seminar on Hydropower Plants (ViennaHydro 2026), Vienna.

[3] Omarar, Z. (2026). Multimodal Sensing for Acoustic–Vibration Based Fault Localization in Hydropower. Master’s thesis, University of St.Gallen. https://sensing-group.com/files/theses/ma-zakaria-omarar.pdf

[4] Klindworth, L. F., & Berendt, C. W. (2026). BATradar: Acoustic Drone Localization Engine. Bachelor project, University of St.Gallen. https://sensing-group.com/files/theses/bp-loris-colin-batradar.pdf

[5] Huwyler, M. (2026). Distributed Acoustic-Vibration System for Industrial Predictive Maintenance. Master’s thesis, University of St.Gallen. https://sensing-group.com/files/theses/ma-maximilian-huwyler.pdf

[6] Knapp, C., & Carter, G. (1976). The generalized correlation method for estimation of time delay. IEEE Transactions on Acoustics, Speech, and Signal Processing, 24(4), 320-327.

[7] Martlul (2020). Francis turbine complete: Spiral hydro turbine. Printables model 234219, free STL. https://www.printables.com/model/234219-francis-turbine-complete-spiral-hydro-turbine-comp

Interested? Email the advisors: Bruno Rodrigues.