Abstract

Combustor liners are subjected to high operating temperatures and high temperature gradients, which have an adverse effect on the durability of liners. Accurate prediction of liner wall temperature distribution can provide better insight into the design of effective cooling systems that have the potential to improve liner structure life.

When compared to RANS (Reynolds averaged Navier Stokes), LES (Large eddy simulation) framework provides better accuracy in resolving the large range of temporal and spatial scales of turbulent flow inside combustors. In simulations in which an unsteady LES fluid solver is interacting with an unsteady solid thermal solver, it would be impractical to advance and synchronize fluid and solid domains in physical time, due to a large difference between small fluid time scales set by turbulence and large solid time scales set by the thermal inertia of the solid. By advancing the fluid and solid solvers with different time step sizes, or by loosely coupling fluid and solid solvers such that they communicate at a defined frequency, convergence can be artificially accelerated. The convergence of the predicted temperature field solution is dependent on the implementation methodology of the acceleration techniques.

A combustor liner is subjected to hot turbulent gases on one side of its boundary and relatively colder air on the other side. This scenario is analyzed to understand the effect of accelerating convergence on the temperature field in a simplified 1D linear framework. A representative polychromatic temperature wave that a combustor liner is subjected to, is used in defining the boundary condition of a 1D implicit unsteady heat conduction solver.

This content is only available via PDF.
You do not currently have access to this content.