Design, Numerical Simulation and Experimental Investigation of Radial Inflow Micro Gas Turbine

Abstract

The work arose initially from an interest in design of radial turbine for small scale gas turbine applications typically suitable for distributed power generation system which demands compact installations. The paper describes an investigation in to the design and performance of radial inflow turbines having a capacity of 25kW at 1,50,000 rpm. First a non-dimensional design philosophy is deduced to design a turbine rotor. The design approach is largely one dimensional along with empirical correlations for estimating losses used to obtain the main geometric parameters of turbine. From the proposed design approach, turbine total-to-static efficiency is calculated as 84.91% which is reasonably good. After that a modified vortex design procedure is developed to derive the non-dimensional volute geometry as a function of azimuth angle for actual flow condition. Once a specific turbine is designed, the flow is analyzed in detail using a three-dimensional Computational Fluid Dynamics (CFD) code in order to assess how accurately the performance is predicted by simple meanline analysis. Finally, a fully instrumented experimental setup is developed. The experimental investigations have been carried out to study the temperature and pressure distribution across turbine and total-to-static efficiency is calculated. The limitations of surging and choking in compressor as well as in the bearings to take up load at such high speed has allowed the tests to be conducted upto 70000 rpm only, with turbine inlet temperatures ranging from 900 K to 1000 K and a pressure ratio upto 1.79, which developed power in the range of 1.69 kW to 10.22 kW. The uncertainty bands are in order of ±13.76% to ±3.12%. It is observed that the CFD results are in good agreement with test results at off design condition. CFD models over predicted total to static efficiency by order of 7-8% at lower speed. These deviations are reduced as turbine runs close to design point.