Understanding gas turbine design at a system level presents a difficult challenge. Accurate predictions of gas turbine behaviour before whole engine tests are completed are invaluable in preventing costly design changes in the latter stages of the design life cycle. In this study a high fidelity whole engine model has been built — specifically, a thermally enabled structural model. This model can predict component displacements up to system level interactions across the whole engine. Knowledge from such a model can feed into multiple design areas contributing to performance, component design and structural understanding but can also be used to influence physical testing.

There are clear benefits in building such high fidelity models but also many challenges that need to be addressed, namely solver type, geometry interrogation, meshing, solver capability, computational power and finally, processing and validation of output data. Additionally, different applications have been used for thermal and structural modelling in order to utilise best capabilities in thermal and contact modelling but also enable scalability on high performance computing. However, utilising two different solvers involves meshes being tailored for each solver type but also introduces additional complexity of transferring information between the two models used.

The paper will discuss the challenges and analysis methodologies used to thermally solve the whole engine cycle, the mapping procedure to translate thermal data to a structural model, and the approach taken to solve the very large simulation model explicitly at a chosen condition to a pseudo-steady state. In order to validate the simulation results, a vast amount of time has been spent to compare the results to existing test data.

As model validation is a significant step in simulation to gain credibility of the results, a comparison of the predicted component displacements will be shown to X-ray data from a whole engine test. Results and limitations of this testing capability such as influence of engine vibration, shutter speed and noise in the data will be discussed and recommendations provided to improve accuracy of the results.

This content is only available via PDF.
You do not currently have access to this content.