Thermal modelling and control of high pressure turbine subsystems

• Main Author: van Paridon, Andrew
• Other Authors: Ireland, Peter T. ; Bacic, Marko
• Published: University of Oxford 2016
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.728812

The control and monitoring of aircraft engine subsystems is one of the leading fields of research for improving overall performance. As engine sophistication increases, more sensor information becomes available to the digital control system, as well as options for active flow control technologies. It is important to develop control systems that can take full advantage of these developments. For instance, active control of the turbine casing can reduce excess rotor blade tip clearance; for every 0.0254mm of clearance eliminated, specific fuel consumption can be reduced by up to 0.1 and turbine entry temperature by up to 1K. However, the tip clearance changes between 0.1mm and 0.4mm throughout a flight due to differences in the expansion of the rotor disc and seal casing. Active tip clearance control can be used to mitigate this variation and thereby minimise fuel burn. Equally, turbine discs and blades experience large temperature ranges that affect their lives. These can be also be improved by active cooling flow modulation. Clearly, both tip clearance problems and cooling flow modulation require sophisticated real-time models and actuation systems. This thesis has two key objectives; the validation of an active impingement system, and the development of reduced order models for discs and casings. The thesis describes the design and development of a novel facility capable of reproducing typical conditions in a large civil engine's oversegment cavity. The facility recreates the air system at cruise-level temperatures (770K), pressures (1.35MPa) and mass flows (1.0kg/s), with a test casing that is instrumented to capture the complete three-dimensional thermal response. The facility has been designed specifically to validate a novel active clearance control concept - step climb alleviation (SCA). The SCA concept uses hot impingement jets from the oversegment cavity to heat the casing, providing a rapid thermal actuation system for tip clearance control during fast engine transients. Research at the facility allows better understanding of the circumferential heat transfer coefficients and the sealing effectiveness of the SCA impingement plates. The development of the facility considers different air system architectural solutions. A novel concentric double-vessel has been designed to accommodate the high temperatures and pressures using predominantly low cost steels. The electric heater and flow bypass system can deliver rapid changes to the test air flow rate without compromising the temperature. The rail system allows rapid change out of test geometries and accommodates the thermal growth of the facility. The commissioning results of the rig show that it can recreate engine realistic conditions. The facility has also been used to provide data for the development of reduced order models. In the second half of the thesis, a model of the test casing is developed using the newly defined LPV-SVD methodology. This method uses singular value decomposition (SVD) to identify modes of spatial coherence, and linear parameter varying (LPV) systems to model the non-linear dynamics. This is a low order thermal model capable of being run in real-time that requires only a small number of inputs already available in the engine measurement suite. When this is applied to the casing, an accuracy of +/-30K is achieved. The LPV-SVD model is capable of being applied directly to other engine components, using either experimental or simulated data for system identification. As such, it has also been applied to modelling the axisymmetric temperature of an IP disc using simulated data from SC03. In this context, the model is built and validated using flight relevant trajectories. The disc model uses only spool speed, pressure altitude, and compressor exit temperature as inputs, and produces temperature distributions to an accuracy of +/-32K, with all but a few points performing much better. The thesis also presents a Kalman filter augmentation to the LPV-SVD model, which helps track noisy trajectories outside the training data set. The filtered model has been validated against a previously unseen noisy flight profile, and achieves an accuracy of +/-30K. Finally, a second model of disc temperature has been developed using a physics based approach, a so called grey-box model. Heat transfer from air to disc is modelled using empirical rotor-stator aerodynamics matched to validated computational fluid dynamic analyses. Lumped capacitance models and simplified explicit conduction models simulate the heat diffusion through the axial and radial profile. This physics based model is also capable of being run in real-time, but proves to be less accurate than the LPV-SVD model in like-for-like comparisons. Overall, the LPV-SVD model recreates temperatures with far greater spatial resolution, making it a superior choice for disc life monitoring applications.