Numerical simulation of heat transfer around film-cooled
gas turbine blades

G. Medic and P.A. Durbin


Three-dimensional numerical simulation of film-cooling phenomena is becoming a part of the complex procedure of gas turbine blade design and is slowly replacing more simplistic two-dimensional methods. However, the correct prediction of film-cooling and associated heat transfer is directly related to the prediction of cross-flow jet mixing, which represents the major difficulty.

The majority of turbulence models fails for this configuration. Because of the presence of multiscale flow phenomena, more fundamental approaches such as LES or DNS are computationally impractical.

Another alternative is the improvement of existing (simpler) eddy-viscosity models in order to take into account the effects present for this configuration and achieve more accurate predictions of film-cooling. These results can then be used for optimal blade design.

Here at FPCD, Stanford University, we have conducted the preliminary analysis of an experimentally documented film-cooling test case using different turbulence models, with the final goal of proposing an improvement of existing models for this configuration.

The future step is to link the improved flow prediction code with the optimizitation procedure in order to achieve an optimal shape or/and positioning of film holes.


Figure 1. Film cooling, pressure side cooling, jets penetrates into the mainstream.




Test case:
  • VKI experiment (Camci & Arts, 1985),
  • film-cooled gas turbine rotor blade,
  • double row of staggered cooling holes,
  • heat transfer along the suction side of the blade.

  • transonic flow, inflow conditions:
    Ma = 0.25, Re_c = 8.5e5, TU = 5 %, T = 409.5 K, inflow angle = 30 deg,
  • wall temperature: T_w = 298 K,
  • variable coolant temperatures and blowing rates,

  • heat transfer coefficient h comparisons,

Numerical computations:

References:
Downloads:
  • STARCD user subroutine sorkep.f implementing Kato-Launder k-epsilon model with two-layer formulation
  • STARCD user subroutines sorkep.f and vistur.f implementing Kato-Launder k-epsilon model with two-layer formulation, as well as limitors for the time scale T (effecting turbulent viscosity). When using limitors for the time scale (through user defined turbulent viscosity together with k-epsilon model), SWITCH 12 in STARCD has to be set on. These files are provided by Jeong Min Seo (FPCD, Stanford University).


    FPC Home Page  Gorazd Medic's Home Page