3.10.1 Defining the blade
The manufacture of the master blade pattern began with a turbine blade generating program which calculated the points on the surface of the blade from several initial condition parameters related to the size of the piping and headstock and the flow rate and pressure difference across the turbine. The points were then converted for use by the manufacturer.
The first potential manufacturing process that could be used to develop the blades is that of CNC machining. It was first proposed that the cutting tool path would have to run the length of the blade since the increment length along the blade is larger than that for the chord-wise increments. The ridges generated by the chord-wise increments should ideally run parallel to the fluid flow direction and avoid causing a turbulent boundary layer as would be promoted by the span-wise ridges. However, due to the required narrow radius of the ball-nose cutter, too many extra would be required to cut the blade.
This would increase both data entry and machining time. The surface that is finished by span-wise milling wiiJ therefore require a ftnishing process to minimise the ridging and resultant turbulent boundary layer formation. After exhaustive research into the feasibility of machining the turbine blades using a three ax_is TNC Maho milling machine, it was concluded that it was not possible. The machine was not sufficiently versatile to cut in three-dimensional space due to the limitation on the number of controlled axes.
Another manufacturing option was considered which involved the spark erosion process.
This typically requires a "profile" at the top and bottom of the billet from which the object is to be cut. These points are then joined by a straight line, the wire, which is then capable of tracing the independent patterns on the two separate surfaces. The problem with this process is the straight line joining the two profiles. Furthermore, the joining of the two points needed to be within a certain angle limit and a quick manipulation of the
program results revealed that this criteria was not met either. This was not true of the curvature of the blade caused by the twist of the profiles about the 30% stacking point.
The process would have caused gross error in the shape of the blade.
3.10.2 Profile fabricalion
The program output format of the co-ordinates was then changed to suit a number of planes in the z-axis, the spacing of which was determined by the thickness of the supawood which was in turn determined from the degree of resolution required to produce accurate results. The resolution proved to be a function of the radius of the profile, each [x y] point was dependant apon a z value, and the angle of the blade relative to the axis of rotation. This would be the equivalent of converting from cylindrical co- ordinates to Cartesian co-ordinates. High resolution was obtained by using a combination of linear interpolation and projection. It was endeavoured to use points that were as close together as possible to perform the interpolation since the points were on a curve defined by the twist of the blade and were therefore not linearly related.
The profiles were then placed next to each other with a constant spacing on a single sheet. The profiles would then be cut out of the single piece of supawood with a second cut, producing the blocks containing the profiles. The reason for this is that the design would allow the profiles to be stacked one on top of the other using the sizing of the profile blocks to locate the [x y] co-ordinates of the blade.
The problem with this method is that it provided the artisan with a negative of the master pattern, which did not prove to be conducive to the remaining stages of production. It was decided that it would be more appropriate to develop a positive master pattern from which the wax patterns could be developed. A major problem encountered in this decision was in the stacking of the profile patterns. Each pattern had to be located with respect to a single point on the profile and it became the task of the artisan to accurately aIjgn each profile with the angle associated with its z-displacement. Once this had been achieved, the profiJes were glued together.
3.10.3 Producing a master pattern
The stage to follow covered the finalisation of the master blade pattern. This began with filling the gaps of the profiles with body filler. Once it had dried, the body filler and some of the profile were sanded down to get a master pattern shaped as closel y as possible te the computer generated model. These patterns were then prepared for the production of the wax patterns, adding a dovetail, reservoir and sprues for the wax moulding process.
Once the blade shape had been completed it was necessary to determine what method was going to be used to attach the blades to the central hub. The initial consideration was to use dove-tailed slots that would slide into the hub for both the rotor and the stater. Two problems experienced with this design method were the limitation of space on the inner radius of the hub caused by the width of the dovetail and the second problem being the resultant lack of rigidity for the stator blades. The solution to these problems was a combination of reducing the size of the dovetail to allow a greater cross-sectional area between the dovetail slots on the hub and adding a "support ring" on the outer radius of the blades.
Further points that are necessary to take into consideration when defining the size of the dovetail slots:
1:1 The space limitation on the hub was partially solved by swopping the roles of the
"hub ring" and "outer ring"
o The space limitation at the hub needed to be compared to the tangential displacement of the blade camber to check interference
1:1 It was not required for a limitation to be placed on the blades to prevent displacement in an upstream direction. Although the force was not purely axiaJ, it did not cause the blades to move. However, a taper will be required for streamlining and will thus be used in a dual purpose as a preventative measure for the purpose of fluid flow.
The steps to follow were for the development of the wax patterns and required that two separate moulds were made for the rotor and stator blades master patterns. These moulds were fabricated from silicon rubber, by pouring fluid silicon into a coffer dam containing
firstly the rotor and then the stator blade and then allowing it to dry. The rubber was matched 10 the temperature tolerance of the wax that was to be used for moulding. The wax was then heated in a separate container and poured into the cavity. A sufficient number of waxes for both the rOlOr and stator were made plus two additional blades of each, in case of mistakes.
3.10.4 The blades
Once having made the wax patterns, the decision as to how they were going to be converted from waxes to moulds was investigated. The options that were available included sending the wax patterns out for production, developing the patterns into water or alcohol based shell-type investment, or to find another investment that required less in terms of lime and cost. Eventually, the decision made was based on cost and the remaining option was to purchase jewellery investment due to the reduced units of capacity in which it was sold relative to the amount required. The details of lh~
investment process are included in the appendix under manufacturing procedure for additional information.
The casting process was then attempted. It was initially intended that the blades be cast by an external company, but due to cost limitations, the blades were to be cast in house.
The first casting operation made use of the conduction furnaces, neither of which were large enough to heat both the casting and the crucible containing the bronze billets.
Hence two furnaces were used to heat the crucible and the casting, which was morc suitable considering that they required different heating cycles.
The heating cycle of the castings in order to remove the wax is provided in the appendix, but the period extended over some 10 hours in order to get the castings up to the pouring temperature. Considering that it took the furnace approximately the same time to heat to its maximum temperature, the process could not have been achieved much more quickly
Due to overshoot error on the controller, the furnace exceeded its temperature tolerance and burnt out. The furnace that was run in parallel was not capable of the required
maximum temperature and it was intended to be used for pre-heating the crucible. The entire operation being dependent on the main furnace was then abandoned when it failed.
The remaining option was to utilise the induction furnace. Before the operation was accepted, some concern was expressed as to the potential capacity of the furnace, since it only provided for small volumes. However the induction furnace promised to have a higher performance in terms of speed of heating. Another potential problem area of the induction furnace is that it does not have an automated controller. Any control process would have to be done manually and (he castings require that they be heated slowly to prevent temperature cracks. There is also the risk of the fluid solidifying on contact with the surface should the castings not be heated to the correct temperature. This solidification could adversely affect tbe surface tolerance of the blades and cause problems with the flow dynamics.
CHAPTER 4
PROGRAMMING FOR BLADE DESIGN