Tesla Turbine
Compressed Air Energy Storage is one option available for scaling energy storage up to the levels necessary for smoothing out inconsistent renewable energy sources like wind and solar. Typically, a conventional turbine is used to convert the stored compressed air back into electricity when needed.
Image source:
Buffa, F. et al., 2013. Exergy and Exergoeconomic Model of a Ground-Based CAES Plant for Peak-Load Energy Production. Energies, 6(2), pp.1050–1067.
An alternative, and potentially cheaper, approach to these multi-staged turbines is the Tesla Turbine; a single-staged turbine first conceived by Nikola Tesla in the early 20th century. These simple machines consist of multiple discs spinning at high speeds inside an outer case and employ viscous forces to transfer the kinetic energy to the blades. This approach is less destructive for the blades than the typical impulse turbine, and could mean less maintenance over the lifetime of the product.
Image source:
Krishnan, V.G., 2015. Design and Fabrication of cm-scale Tesla Turbines.
This is the initial drawing of what I planned to build. I chose steel for the discs, aluminum for the side covers, and acrylic for the circumference. The keyed shaft kept the discs from turning independently and the bearing holders could be disassembled for easier removal.
The discs were spaced using spacers (machined from steel shim stock) around the shaft. Machining the shim stock meant building a few jigs since the metal needed to remain flat by the end of the process.
The nozzle was, at first, machined from brass since different air hose connections would need to be switched out depending on the source of air.
Due to their availability, strength, and consistency of manufacture, I moved to compact discs for the inner discs. Polycarbonate proved a brittle material to machine.
After an initial test, I quickly found that the polycarbonate discs were not stiff enough to maintain a consistent spacing when air was blown into the gaps between them, so I placed small machine screws and washers around the circumference to keep the thin polycarbonate from vibrating during operation (not shown in picture).
One crux of the Tesla Turbine design is the dynamic seals around the side exhaust holes. They must be as airtight as possible without introducing much friction. My first attempt involved the loop side of hook-and-loop tape.
In order to measure power out, I opted for a tachometer and Prony Brake. This meant I could get total power out by multiplying angular speed and torque.
My first tachometer was an optical one utilizing an IR LED. The signal was processed using a Beaglebone Black single board computer.
Getting consistent measurements with the DIY optical tachometer ended up being very difficult, so I moved to a Hall Effect sensor with a magnet on the shaft, and an interrupt service routine to count turns.
For the torque measurement, I started with a Prony Brake. The first iteration used an aluminum lever with an aluminum braking surface.
I soon realized how quickly the two surfaces with identical properties led to galling, so I moved to an oak lever on a brass braking surface. Here's the braking pulley; the oak lever is shown below.
The friction brake used to exert a torque on the lever was not consistent enough for this application, so I tried an eddy current brake.
I cut an aluminum disc and mounted it to the spinning shaft, then mounted the load cell next to it with some neodymium magnets glued to it. This allowed me to adjust the distance from the magnets to the aluminum in order to vary the amount of force exerted by the eddy currents.
This method provided very smooth power curves since the torque didn't spike from catching imperfections in a braking surface.
Once I had a reliable power measurement and was able to run a few tests (and see how low of an efficiency I was getting), I decided the brass nozzle didn't direct the air as precisely as I wanted, so I drew up a nozzle in Solidworks and 3D printed it in PLA on a LulzBot.
The final assembly of the turbine testing rig included a mass-flow meter and pressure gauge to calculate the energy in, and a tachometer and eddy current brake (pressing on a load cell for torque measurement) to calculate the power out. I mounted everything on a piece of plywood for portability and included a screen for monitoring the status of the datalogger program.
The following four charts are RPM, Torque, Power out, and Efficiency Data of three trials at different flow rates: 113, 169, and 226 standard liters per minute (SLPM). The flow was stopped before the datalogger was turned off to see the spin down.
Efficiency was similar for the 169 SLPM and 226 SLPM trials, so 8.5% is likely the peak efficiency of this configuration.
Areas to improve:
The exhaust hole was originally intended to have a labyrinth seal around it to limit the amount of air that escapes without going through the discs, but the move to CDs and the limited time made this impossible to implement. The CDs I used were not rigid enough on their own, so I had to reinforce them around the perimeter with screws that limit airflow between the discs. The nozzle was designed with almost no fluid dynamics considerations, so I'm confident that an analysis of an efficient nozzle design would yield a much better product.
The areas where I can see the largest improvement being made with minimal effort are the discs, the nozzle, and the exhaust seal. I hope to get the chance to build another iteration using the lessons learned and knowledge gathered during this project.
2016