Development of a CFD based methodology for predicting oil auto-ignition in gas turbine bearing chambers

The demand for increased efficiencies in modern aero-engines drives designs to higher pressure ratios, temperatures and shaft speeds. Consequently, higher cycle temperatures and parasitic heat loads are likely to become limiting factors in the design of bearing chambers. These chambers are lubricated and cooled with oil and pressurized using air taken from the main gas path. Historically, the temperature and pressure of the sealing air has been low enough to exclude the risk of bearing chamber oil fires caused by oil auto-ignition. However, temperatures are being driven towards the limits set by current design rules. With respect to oil fire risk assessment, current design rules are very conservative as they pessimistically assume the oil is continuously exposed to the maximum temperature expected in operation when determining the minimum residence time required for oil auto-ignition. Improved bearing chamber designs capable of tolerating higher temperatures could therefore be developed by applying a more rational level of conservatism, based on a more physics-based approach (such as considering the effect of an oil droplet being transported through a varying temperature field). In this paper a numerical methodology is developed to provide a pragmatic approach to addressing conservatism in bearing chamber oil fire risk assessment, with respect to oil auto-ignition. An unsteady Eulerian CFD prediction is used to compute the aerothermal flow field within a stylized bearing chamber. A Lagrangian discrete particle method is then used to track the oil droplet trajectories. The time-dependent droplet temperature histories are then used to compute the fractional accumulation of auto-ignition delay time using an empirically derived relationship. Finally, by defining an ‘auto-ignition (AI) energy’ accumulation factor, the methodology assesses whether an individual oil droplet has satisfied the criteria for auto-ignition. The present contribution examines the effects of various modelling parameters on the ‘AI energy’ accumulation factor. These include one/two-way coupling between the air flow and oil droplets, stochastic turbulence modelling of the droplet behavior, and the effect of droplet size and distribution. The work highlights that two-way coupling is required to ensure the thermal effect of the oil is modelled, despite the increased computational demand. Stochastic modelling of interactions between particles and the flow field is also required to capture the spread of particle trajectories and the resulting distribution of particle temperatures. A representative range of droplet sizes must also be simulated as the propensity for oil AI is a function of droplet diameter; the highest risk occurs for the smallest droplets whilst the largest droplets have greater cooling effect on the air flow. Given the extent of model simplification required to allow the work to be completed with a mid-spec desktop computer and the overall scope of the project, a validation of the findings has not been completed. Instead, an experimental validation is proposed as part of future investigation. The authors imagine that with enough investigation and validation, the understanding developed by the work could be applied as part of a computationally efficient industrial design toolset to inform the early stages of product design.