Numerical Investigation on Bearing Chamber Wall Heat Transfer

Bearing chamber of a gas turbine engine is generally sealed by pressurized air, separating lubricant from the other zones of the engine. Heat transfer from the wall to air/oil mixture is a challenging engineering problem; predicting heat transfer rate from bearing chamber to oil is important to avoid oil coking and oil fires under high rotational speeds, pressure levels and turbine inlet temperatures. In this study, the inner wall temperature of bearing chamber which is located at the center of front engine structure was investigated numerically. The numerical study involved mainly two thermal modelling methods having two different empirical correlations was performed with finite element solver in order to calculate heat transfer on the wall. First method was based on rotational Reynolds number and Prantl number, in addition to these numbers second one, which is suggested in the literature, is based on oil related and sealing air related Reynolds number, mixture temperature and mixture mass flow. Second approach considers existence of a mixing of gaseous and liquid flow in the core flow unlike first modelling approach. The thermal model was solved by finite element solver and numerical model, assumptions were described with thermal boundary conditions. On the other hand, wall and air thermocouple readings were taken through engine test from the bearing chamber for real engine operating conditions having mainly idle, cruise and maximum power. DN number ranges from 712564 to 2742404, sealing air flow ranges from 46 to 78 g/s and oil flow ranges 22 to 40 g/s for these conditions. The calculated heat transfer coefficients were presented and discussed. The wall temperature predictions of the thermal models, and test measurements were compared. The comparison revealed that analysis results obtained with both correlations were in reasonable agreement with the test. In overall, the second approach predicted metal temperature slightly better at the front support and inner manifold wall, while first approach predicted much better at the rear support wall.