PARTICLE-FILTER BASED UPSCALING FOR TURBULENT REACTING FLOW SIMULATIONS

Author(s):  
Shubham Srivastava ◽  
Tarek Echekki
2001 ◽  
Author(s):  
Edward Luke ◽  
Xiao-Ling Tong ◽  
Junxiao Wu ◽  
Lin Tang ◽  
Pasquale Cinnella

1998 ◽  
Vol 120 (1) ◽  
pp. 60-68 ◽  
Author(s):  
V. R. Katta ◽  
W. M. Roquemore

Spatially locked vortices in the cavities of a combustor aid in stabilizing the flames. On the other hand, these stationary vortices also restrict the entrainment of the main air into the cavity. For obtaining good performance characteristics in a trapped-vortex combustor, a sufficient amount of fuel and air must be injected directly into the cavity. This paper describes a numerical investigation performed to understand better the entrainment and residence-time characteristics of cavity flows for different cavity and spindle sizes. A third-order-accurate time-dependent Computational Fluid Dynamics with Chemistry (CFDC) code was used for simulating the dynamic flows associated with forebody-spindle-disk geometry. It was found from the nonreacting flow simulations that the drag coefficient decreases with cavity length and that an optimum size exists for achieving a minimum value. These observations support the earlier experimental findings of Little and Whipkey (1979). At the optimum disk location, the vortices inside the cavity and behind the disk are spatially locked. It was also found that for cavity sizes slightly larger than the optimum, even though the vortices are spatially locked, the drag coefficient increases significantly. Entrainment of the main flow was observed to be greater into the smaller-than-optimum cavities. The reacting-flow calculations indicate that the dynamic vortices developed inside the cavity with the injection of fuel and air do not shed, even though the cavity size was determined based on cold-flow conditions.


2021 ◽  
Vol 130 (11) ◽  
pp. 115702
Author(s):  
Malik Hassanaly ◽  
Hariswaran Sitaraman ◽  
Kevin L. Schulte ◽  
Aaron J. Ptak ◽  
John Simon ◽  
...  

Author(s):  
Viswanath R. Katta ◽  
W. M. Roquemore

Spatially locked vortices in the cavities of a combustor aid in stabilizing the flames. On the other hand, these stationary vortices also restrict the entrainment of the main air into the cavity. For obtaining good performance characteristics in a trapped-vortex combustor, a sufficient amount of fuel and air must be injected directly into the cavity. This paper describes a numerical investigation performed to better understand the entrainment and residence-time characteristics of cavity flows for different cavity and spindle sizes. A third-order-accurate time-dependent Computational Fluid Dynamics with Chemistry (CFDC) code was used for simulating the dynamic flows associated with forebody-spindle-disk geometry. It was found from the non-reacting flow simulations that the drag coefficient decreases with cavity length and that an optimum size exists for achieving a minimum value. These observations support the earlier experimental findings of Little and Whipkey (1979). At the optimum disk location, the vortices inside the cavity and behind the disk are spatially locked. It was also found that for cavity sizes slightly larger than the optimum, even though the vortices are spatially locked, the drag coefficient increases significantly. Entrainment of the main flow was observed to be greater into the smaller-than-optimum cavities. The reacting-flow calculations indicate that the dynamic vortices developed inside the cavity with the injection of fuel and air do not shed, even though the cavity size was determined based on cold-flow conditions.


Author(s):  
Sandeep Kedukodi ◽  
Srinath Ekkad

Established numerical approaches for performing detailed flow analysis happens to be an effective tool for industry based applied research. In the present study, computations are performed on multiple gas turbine combustor geometries for turbulent, non-reactive and reactive swirling flow conditions for an industrial swirler. The purpose of this study is to identify the location of peak convective heat transfer along the combustor liner under swirling inlet flow conditions and to investigate the influence of combustor geometry on the flow field. Instead of modeling the actual swirler along with the combustor, an inlet swirl flow profile is applied at the inlet boundary based on previous literature. Initially, the computed results are validated against available experimental data for an inlet Reynolds number flow of 50000 using a 2D axi-symmetric flow domain for non-reacting conditions. A constant heat flux on the liner is applied for the study. Two turbulence models (RNG k-ε and k-ω SST) are utilized for the analysis based on its capability to simulate swirling flows. It is found that both models predict the peak liner heat transfer location similar to experiments. However, k-ε RNG model predicts heat transfer magnitude much closer to the experimental values except displaying an additional peak whereas k-ω model predicts only one peak but tends to over-predict in magnitude. Since the overall characteristic liner heat transfer trend is captured well by the latter one, it is chosen for future computations. A 3D sector (30°) model results also show similar trends as 2D studies. Simulations are then extended to 3 different combustors (Case 1: full cylinder and Case 2 and 3: cylinders with downstream contractions having reduced exit areas) by adopting the same methodology for same inlet flow conditions. Non-reacting simulations predict that the peak heat transfer location is marginally reduced by the downstream contraction of the combustor. However the peak location shifts towards downstream due to the presence of accelerated flow. Reacting flow simulations are performed with Flamelet Generation Manifold (FGM) model for simulating premixed combustion for the same inlet flow conditions as above. It is observed that Case 3 predicts a threefold increase in the exit flow velocity in comparison to non-reacting flow simulations. The liner heat transfer predictions show that both geometries predict similar peak temperatures. However, only one fourth of the initial liner length experiences peak temperature for Case 1 whereas the latter continues to feel the peak till the end. This behavior of Case 3 can be attributed to rapid convection of high temperature products downstream due to the prevailing accelerated flow.


Sign in / Sign up

Export Citation Format

Share Document