Building a Disk Turbine
Turning the stator and backplate. The turbine shaft blank was used as an arbor to maintain concentricity.
|April 15, 2003
I decided to design the turbine housing in three main sections. A backplate/bearing support housing, a stator section, and a cover plate. These would allow me to make easy changes to the design. One stator might be designed to accommodate plug-in nozzles for driven turbine tests. Another might be relieved to accommodate scroll chambers for compressor mode tests. The faceplate could serve as either an exhaust or inlet, and feature either a nozzle or flared intake. Castings make changes relatively easy, and often a single pattern can serve to produce variations in the form of a part. The cost in materials for casting additional pieces is trivial.
Patterns for the housing were set into a single 12' x 12" greensand drag. The cope sand will be rammed up next. It's tricky to pour such a tight mold. The vertical pieces are sprue pins which are later removed to leave the pour openings.
Also because I was casting the housing, it was possible to make the design really solid. This is of course a stationary model, not an aircraft engine so weight is not a consideration. Burst strength is, however, and it is easily possible to work with 3/4" wall thickness for a turbine model of this size. The heavy housing provides a lot of additional burst protection for the rotor.
Writers with experience in model aircraft turbine design suggest that rotor failure is generally unspectacular at this scale. Apparently because of the close tolerances of the housing, a blade which is failing usually stretches after exceeding yield strength, contacts the housing and turbine grinds to a screeching halt. Nevertheless, plenty of housing thickness makes for a more secure feeling when testing. Model aircraft engines typically employ very thin steel or stainless steel tubing (used camping gas containers, oil filter housings, and stainless steel kitchen canisters have all been used) and apparently give satisfactory service with the extreme rates of rotation they manage to achieve.
(c) Copyright 2003, Stephen Redmond, all rights reserved
|April 16, 2003
I spent a lot of time thinking about the disks. Though I had already purchased both stainless steel and aluminum sheet material suitable for the washers and disks, I decided to see if I could cast the the blades to start with. For a small engine it seemed possible to eliminate the star washers and simply cast and machine a thin boss on each disk. Then I would mill out the ports through the thickened center area of the disks. This would make a stronger assembly, since the disks would be thicker near the shaft, and the rivets normally required could be eliminated.
I have seen a couple of Tesla style rotor assemblies with rivets located in the most highly stressed area of the disk spokes, just below the junction with the main disk flat, where the disk cross sectional area is at its minimum. These rivet holes would seem to concentrate stresses at the worst possible place on the disk, and I'm guessing that if the blades yield and stretch, this is the area where it happens.
In some designs, the star washers extend well past the ports and are riveted in the outer portions of the disk. These would be somewhat stronger than riveted designs with washer spokes that terminate near the ports. However the long star spokes block circular flow (not to mention laminar flow) which is the primary mechanism of the disk turbine. So the effective driven area of the disk is reduced by the length of the spokes. The spokes may nevertheless be providing drive as radial blades, but if the contribution is substantial, why not go to a radial design in the first place? Since I wanted to test a real disk turbine, long spokes didn't fit the requirement.
The main question I thought about was how to dimension the disks themselves. How thick should the disks be, and how far apart should they be spaced? A lot of disk machines I read about, (and reports of some of Tesla's original engines) specified a disk gap of about 1/32" or .032. But some recent theories suggest that this is much too wide for good efficiency. The latest recommendations are based on the assumption that the disk gap should be wide enough to accommodate laminar flow on each disk face, but not much wider. There are formulas to determine this, and two of the variables they use are the kinematic viscosity of the working fluid and the speed of rotation. If you plug values into these formulas, you find that the higher the kinematic viscosity. the wider the gap needed, and the faster the blades are turning, the narrower the gap.