Improved Analytical Model for Bolted Joint Evaluation in Gas Turbines

Modern day gas turbines are very complex in construction consisting of a large number of smaller components connected and kept together by means of bolted joints. These joints operate at high temperatures, pressures and speeds. It is essential that the structural behavior of all the joints be assessed to address issues such as fatigue, creep and flange opening. During the preliminary design and concept optimization stage it is prudent to use simple analytical approaches rather than full fledged finite element simulation models. A number of analytical models are available in literature. Prominent among them are analytical model for “Flat Faces Flanges with Metal-to-Metal Contact beyond the Bolt Circle” by Schneider, which later become standard in ASME code and more recent one is by Galai and Bouzid that uses annular plate theory and thin cylinder theory in conjunction to simulate flange. However such analytical models do not account for all the operating parameters and it is essential to assess their adequacy. The main purpose of the proposed paper is to carry out critical assessment of the simpler analytical approaches and document the findings. The purpose is also to refine the approaches so that the shortcomings are addressed. To this end a 3D high fidelity finite element model including flanges, bolt and nut along with threads is developed simulating bolted joints in gas turbine cases and analyzed for different preload and external operating load conditions like pressure and axial blow off loads. The model is also analyzed for different geometric configuration by varying flange thickness, height and bolt spacing ratio. Different analytical approaches for fastener and member stiffness calculation are evaluated and identified the right combination for the complete model. Existing analytical model for flange opening and bolt stress is then updated by including the right stiffness models. The updated analytical model is compared with high fidelity FE model for different geometry configurations. Results show that the bolt stress variation between the updated analytical model and 3D FE model is around 10% as against 20% in the existing analytical model. Improved analytical method also provides good agreement with the 3D FE results for external loads up to about 60% of the bolt preload, which is well above the normal operating load.