Analysis of Measured and Predicted Turbine Maps From Start-Up to Design Point

The turbine full performance map of Gas Turbine (GT) engine is seldom investigated, as the primary focus is on design conditions. The investigation of start-up and off-design conditions is not trivial as these are often transient points or areas where it may be difficult to ensure stable engine operation. Therefore, the acquisition of reliable data is more challenging with respect to a stable and stationary operating condition also because sensors may be operated outside their calibration range. Additionally, performance prediction methods, from correlation based to more sophisticated CFD based approaches, suffer in transient and highly off-design conditions where either the correlations may not hold valid or CFD gets unsteady and/or unstable, thereby requiring a considerable computational effort. Further difficulties stem from the shape of off-design speedlines. In fact, in the vicinity of the design corrected speed, the speedlines are similar in shape, while they deform and depart from each other for off-design conditions.

This paper addresses these problems by describing an experimental campaign conducted on a 1.5 stage turbine that investigated the full range of operating conditions, from start-up to design under variable expansion ratio and physical speed. The test is designed to maintain strict similitude to realistic conditions in terms or Reynolds number, corrected mass flow and corrected speed. The campaign includes specific activities to determine the start-up torque and dedicated measurements to assess the accuracy of both prediction methods and of measurement probes operated at the edge or outside their calibration range. The test rig is equipped with static and stagnation pressures and temperatures and flow angle probes with fixed rakes and traversing systems, flow rate in the main flow field and in the cavities as well. Around 50 operating conditions are investigated. A companion set of correlation and CFD analyses is completed in selected operating conditions, from steady to unsteady, with purge flows included to integrate and compare with measured performance, flow field, and torque. The prediction methods are discovered to provide a fairly realistic picture of turbine performance at both design and off-design conditions. The merits of unsteady v/s steady CFD results are discussed. Importantly, unsteady CFD predicts the start-up torque level, an important information especially for mechanical drive gas turbines.

The choice of variables to describe the speedlines is also addressed by using both measured and predicted data. A detailed discussion on velocity ratio versus corrected speed illustrates the advantages of the former parameter the adoption of which produces constant shape curves in a very wide range of operating conditions.