Numerical Research On The Clocking Effect Of The 1+1 Counter-Rotating Turbine And Predicted Model

It is well-known that turbine internal flow field has strongly unsteady features which cover kinds of unsteady flow phenomenon, such as wake interacting with main flow, secondary flow interacting with main flow and leaking flow interacting with main flow. Clocking position is a key factor of influencing turbine unsteady flow phenomenon and additional aerodynamic benefits can be optained by changing clocking position of turbine. Many researches have been conducted to get insight into clocking effects on axial flow turbine. However, basic flow principles of some flow phenomena are still confused to us and method of applying clocking effect still need to be further developed. For that, numerical research on cloking effect of 1+1 counter-rotating turbine 1.5 stages will be conducted and a predicted model to find optimal clocking position of axial flow turbine will be constructed. Primary research contents are as follows.1. Unsteady simulation on 5 clocking positions of 1+1 counter-rotating turbine 1.5 stages is performed. The results state that time-averaged efficiency of 1+1 counter-rotating turbine 1.5 stages is increased by 0.16% through changing clocking position between upstream vane and downstream vane. The aerodynamic benefits of 1+1 counter-rotating turbine are evidently lower than traditional turbine, which are attributed to three dimensional effect of clocking effect and unique characteristic of 1+1 counter-rotating turbine, named little turning angle.2. Wakes of high pressure stage rotor are studied. In one time period, as downstream stator turning direction is in alignment with rotor’s turning direction, main flow direction in the downstream stator passage is parallel with linear wakes of rotor and rotor wakes present linear distribution from thick to thin, which are beneficial to the promotion of time-averaged efficiency. In the meantime, a new wake ball will be generated due to staggered mixing between upstream rotor wakes and downstream stator wakes when the linear wakes continued to develop downstream. Then time-averaged efficiency of 1+1 counter rotating turbine can be further increased by changing axial gap between stator and rotor of low pressure stage to make the new wake ball impinges the leading edge of low-pressure stage rotor.3. When we focus on the specific flow field of turbine, difference of wake and secondary flow’s flow principle is found, which is the main reason of two-peak phenomena that efficiency varied with clocking position has two-peak points. When both wakes and secondary flow hit the downstream vane leading edge, time-averaged efficiency achieves the highest point. When neither wakes nor secondary flow hits the leading edge, time-averaged efficiency appears the lowest point. Another peaking point will be get when secondary flow hits the leading edge and wakes migrates in the passage.4. The predicted model is constructed by combining results of steady simulation and formulas derivation on the basis of wake evolution principle and previous research results in the field of clocking effect. In the meantime, two unsteady numerical research examples, including 1+1 counter-rotating turbine and Aachen turbine, are carried out to compare with the results of the predicted model. The final numerical results state that the optimal clocking position of 1+1 counter rotating turbine shown by unsteady numerical simulation is that downstream vane relative to corresponding upstream vane circumferentially moves 1.6 pitches while the predicted model estimates circumferential moving distances is 1.606 pitches, relative tolerance is 0.4%. As for Aachen turbine, the unsteady numerical simulation presents that wakes of upstream vane appear in the position of corresponding downstream vane circumferential 3.0 pitches and circumferential moving distance of wakes predicted by model is 3.098 pitches, relative tolerance is 3%. Additionally, the prediction of optimal clocking position by traditional unsteady simulation using computer clusters(12 cores×2.0 GHz) in this paper needs at least 4 days to complete while the time consumed by the model can be nearly neglected compared with above predicted time.