Simulation and Optimization of Thermal Fluid-Structure Interaction in Blade-Disc Configurations of Aircraft Engines

In the turbine stage of aircraft engines, most components are purged by cooling air to resist the hot core flow of around 1800 °C that exit the combustors. The supplied cooling air of around 1100 °C is bled from the core flow in the compressor stages. Hence, this part does not contribute directly to the engine thrust. Likewise, up to 20% of the core flow is extracted to serve such functions in secondary air systems in order to enable a safe operation. Therefore, it is of great interest to minimize such secondary air flows because of their disadvantageous impact on the turbine efficiency. Correspondingly, it is necessary to prove their cooling effect by most accurately determining the correct temperature distribution in the turbine discs.

In the European Union sponsored project MAGPI the heat transfer in rotor stator cavities has been investigated. In particular, the dependency of the cooling effectiveness on the interaction between the cooling flow and the hot main annulus gas has been the main focus of attention. This PhD project investigates the impact of structural deformations under hot running condition on the cooling effectiveness by the use of Thermal Fluid-Structure Interaction methods. Such a coupled numerical approach solves the fluid-solid heat transfer in conjunction with the structural deformations that appear due to centrifugal and thermal expansion of rotating and stationary parts. Especially the hot running clearances at the interstage labyrinth seal, blade tip and rotor stator rim seal can be addressed to capture the correct mass flows at these locations.

The validation test case is a two-stage axial turbine rig, originated at the University of Sussex. Figure 1 shows the flow structure in the representative rotor stator cavity. In contrast to methods that disregard deformations under hot running condition, the TFSI approach decreases the discrepancies in solid temperatures in the rotor stator cavities. A 3D sector model as well as a reduced surrogate 2D model are validated and compared regarding the influence of non-axisymmetric features.

Within the commercial software package ANSYS it is possible to simulate Thermal Fluid-Structure Interaction (TFSI) by two different approaches. First, the implicit TFSI approach employs the MFX interface, which couples the fluid solver CFX and the structural solver Mechanical in a bi-directional, implicit and partitioned way. Hereby, the exchange of thermo-mechanical variables is located at fluid-solid interfaces among several coupling iterations. In the second approach, an explicit TFSI approach, the finite element solver Mechanical calculates the deformed state subsequent to a fully converged CFX run, which in turn can be executed on the corresponding deformed mesh. In contrast to the first approach, CFX is set up as a Conjugate Heat Transfer (CHT) model solving not only for the heat convection in the fluid region but also for the convective heat transfer at the walls and the heat conduction in the solid region. If the uni-directional exchange of temperature distribution and deformation of the solid domain is operated multiple times, this approach works as a manually coupled bi-directional approach.

In the optimization scenario, the explicit TFSI coupling between a CHT and FE solver is utilized, separating the solution of the heat transfer from the corresponding deformation. This manual coupling has proven to be more efficient than the MFX coupling that exchanges the thermomechanical variables implicitly at the fluid-solid interface and solves the heat transfer in parallel to the stress-strain state.

A set of four design variables is chosen to optimize the cooling efficiency in a rotor stator cavity of a turbine rig. Different optimization algorithms are compared regarding their accuracy and effiency. The gradient-based SQP method is affirmed to be the superior optimization technique in comparison to the gradient-free Nelder-Mead simplex algorithm, which reaches the optimization goal in twice as many evaluations and significantly longer computational time. Figure 3 compares the default with the optimized setting, revealing an optimized rotor geometry with an angled and axially shifted cooling entry, radially shifted drive arm and slightly adjusted interstage seal clearance. Hereby, the rotor temperature is reduced and the coolant is distributed more efficiently to the desired area. Additionally, the optimized setting is proven to be more robust against worst case scenarios with hot gas ingestion in dynamic behavior as a drop of coolant mass flow rate or an increasing interstage seal clearance as a result of a reduced rotor speed.

Multi-Physics – Special Coupled Systems: Fluid- and Thermodynamics, Structural Mechanics – Optimization

Hannes Lück

Dipl.-Ing.

Address: | Dolivostraße 15 |

D-64293 Darmstadt | |

Germany | |

Phone: | +49 6151 16 - 24401 or 24402 |

Fax: | +49 6151 16 - 24404 |

Office: | S4|10-312 |

Email: | |