A practical computational fluid dynamics (CFD) approach to modeling effusion orifices in gas turbine combustor liners is proposed specifically when liner metal geometry is not included and conjugate heat transfer is not invoked. The focus is on eliminating effusion orifices from the model while maintaining the imprint of the orifices on the cold and hot sides of the liner wall. The imprinted boundaries serve as embedded mass flow inlets and outlets on both sides of the wall and maintain the integrity of the wall geometry. An empirical model is then used to extract and inject mass from the cold and hot sides of the liner, respectively. The mass extraction and injection process is performed for each orifice based on local conditions such as pressure, temperature and discharge coefficient. The discharge coefficient is, in turn, dynamically computed for each orifice based on approach angle, approach Mach number, discharge Mach number and orifice length to diameter ratio. With this approach, the fidelity of the liner wall is preserved for better heat transfer predictions and easier near wall meshing. In addition, the discharge coefficient is not assumed but calculated allowing the redeployment of inherently inadequate effusion orifice mesh cells to other critical areas of the combustor. Presented are results of two combustor cases to demonstrate the practicality and accuracy of the proposed method as compared to standard effusion modeling and their comparison with rig data.

Modeling the interaction between gas turbine engine modules is complex. The compact nature of modern engines makes it difficult to identify an optimal interface location between components, especially in the hot section. The combustor and high-pressure turbine (HPT) are usually modeled separately with a one-way boundary condition transfer to the turbine inlet. This approach is not ideal for capturing all the intricate flow details that travel between the combustor and the turbine and for tracking hot streak migration that determines turbine durability. Modeling combustor-turbine interaction requires a practical methodology that can be leveraged during the engine design process while ensuring accurate, fast, and robust CFD solutions. The objective of this paper is to assess the effectiveness of joint simulation versus co-simulation in modeling combustor and turbine interaction. Co-simulations are performed by exchanging information between the combustor and the turbine stator at the interface, wherein the combustor is solved using Stress-Blended Eddy Simulation (SBES) while the stator is solved using RANS. The joint combustor-stator simulations are solved using SBES. The benefits of using SBES versus LES are explored. The effect of the combustor-stator interaction on the flow field and hot streak migration is analyzed. The results suggest that the SBES model is more accurate than LES for heat transfer predictions because of the wall treatment and the joint simulation is computationally efficient and less prone to interpolation errors since both hot section components are modeled in a single domain.

No abstract

No abstract

Controlling light-around and re-light presents design challenges for gas-turbine manufacturers. Researchers have studied the detailed phenomena in laboratory experiments to elucidate controlling factors and modes of behavior. Several groups have reported high-fidelity simulations of the fluid dynamics, turbulent mixing and light-around phenomena using large eddy simulations (LES) on highly refined computational meshes. While such simulations can reproduce experimental observations, they are computationally expensive and tend to be impractical for routine design analyses. In this work, we present a less computationally intensive CFD approach, which has been tested against laboratory experiments using both gaseous-fuel injections and liquid-fuel injections. Results show that a consistent practice of mesh and model settings can be used for all the test cases considered. The simulations generate light-around sequences and total-ignition times that agree well with experimental measurements. Observed trends are predicted when varying burner spacing as well as the fuel and injection method.