Design and Development of a High Performance LPT for Electric Turbocompounding Energy Recovery Unit in a Heavily Downsized Engine

Abstract

This thesis presents the design method and development of a high performance Low Pressure Turbine (LPT) for turbocompounding applications to be used in a 1.0L "cost-effective, ultra-efficient heavily downsized gasoline engine for a small and large segment passenger car". Under this assumption, the LPT was designed to recover the latent energy of discharged exhaust gases at low pressure ratios (1.05 - 1.3) and to drive a small electric generator with a maximum power output of 1.0 kW. The design speed was fixed at 50,000 rpm with a pressure ratio, PR of 1.1. Commercially available turbines are not suitable for this purpose due to the very low efficiencies experienced when operating in these pressure ratio ranges. A bespoke mean-line model was developed to evaluate the turbine performance and to generate a preliminary LPT design. Prior to the design work, the mean-line model was validated against an existing turbine. A good agreement between the predicted turbine performance and the test result was found and a minimum Relative Standard Deviation value of 1.62% was achieved. By fixing all the LPT requirements, the turbine loss model was combined with the geometrical model to calculate preliminary LPT geometry. The LPT features a mixed-flow turbine with a cone angle of 40˚ and 9 blades, with an inlet blade angle at radius mean square of +20˚. The exit-to-inlet area ratio value is approximately 0.372 which is outside of the conventional range indicating the novelty of the approach. A single passage Computational Fluid Dynamics (CFD) model was applied to optimize the preliminary LPT design by changing the inlet absolute angle. The investigation found the optimal inlet absolute angle was 77˚ and this was used to design the volute. Turbine off-design performance was then predicted from the mean-line model, single passage CFD and full turbine CFD model. The full turbine CFD and a refined single passage CFD were modelled to analyze the turbine flow field in the volute and rotor passage. A rapid prototype of the LPT was manufactured and tested in Imperial College turbocharger testing facility under steady-state and pulsating flow. The steady-state testing was conducted over speed parameter ranges from 1206 rpm/K0.5 to 1809 rpm/K0.5. The test results showed a typical flow capacity trend as a conventional radial turbine but the LPT had higher total-to-static efficiency, nt-s in the lower pressure ratio regions. A maximum total-to-static efficiency, nt-s of 0.758 at pressure ratio, PR≈1.103 was found, no available turbines exist in this range as parameters. A validation of the predicted off-design performance against the LPT test result found a minimum total-to-static efficiency Standard Deviation of ±0.019 points for mean-line model at 1206 rpm/K0.5 and the full turbine CFD model showed a minimum Mass Flow Parameter Standard Deviation of ±0.09 kg/s.K0.5/bar also at 1206 rpm/K0.5. The pulsating flow testing was carried out at LPT power of 1.0 kW for a pulsating frequency range of 20 Hz to 80 Hz over the turbine speed parameter range between 1206 rpm/K0.5 and 1809 rpm/K0.5. A hysteresis turbine performance encapsulated the steady state turbine map due to a ‘filling and emptying’ was shown for all frequencies. The pulse pressure amplitude and the chopper plate pulse frequency are found to have influenced the measured unsteadiness characterization. Strouhal number, St.* and [Lambda criterion symbol appears here] lambda criterion which commonly used in quasi-steady analysis were used to quantified the unsteadiness level. The LPT was implemented into a validated 1-D engine model to investigate the impact on the Brake Specific Fuel Consumption and Brake Mean Effective Pressure. The validated 1-D engine model was run at full load at three LPT locations: waste-gated, pre-catalyst and post catalyst. The study found the optimum location was at post catalyst and a maximum BSFC reduction of 2.6% can be achieved. The part-load 1-D engine model found that an installation of the turbocompounding unit increased the Pumping Mean Effective Pressure hence; it increased the Brake Specific Fuel Consumption up to 0.72%. However, as soon as the LPT power was reused into the engine, a maximum reduction of BSFC approximately 2.6 % can be achieved. Engine testing to analyze the effect of the LPT was carried out at Ricardo UK ltd, Shoreham Technical Centre. The LPT was installed at exhaust post catalyst of a heavily downsized 1.0L gasoline engine. A commercial compressor Garret GT28RS was attached to load the LPT. The results showed a maximum Brake Specific Fuel Consumption reduction of 2.6 % was achieved at an engine speed of 2500 rpm during the part load condition. Finally, the engine model results were compared against the engine testing. A comparison between the engine testing and 1-D engine model showed a good agreement at part load condition with a minimum Brake Specific Fuel Consumption Standard Deviation of 0.0238 at engine speed of 3000 rpm.