Best Practice Advice AC2-11

Front Page

Description

Test Data

CFD Simulations

Evaluation

Best Practice Advice

Delft-Jet-in-Hot-Coflow (DJHC) burner

Application Challenge AC2-11   © copyright ERCOFTAC 2024

Best Practice Advice

Key Fluid Physics

From the simulations performed so far, it is clear that the successful modelling of the DJHC flames is dependent on the correct representation of 1) the jet spreading rate, 2) ignition delay, and 3) entrainment. These aspects are not independent from each other. For example, the jet spreading rate affects fuel and oxidiser mixing, which affects ignition, or as ignition initiates reactions it affects the density and, thereby, the jet spreading.

Computational Domain and Grid Resolution

The computational domain and the grid resolution employed in the present simulations (Table 3) are adequate for the considered model approaches. The domain and mesh used here may help to guide future simulations. One should be aware that different approaches have different requirements for mesh-independence. It could be of interest to choose the start of the computational domain further upstream to better represent the flow at the first measurement position (see below at boundary conditions).

Physical Modelling

Turbulence model

In the present study the focus has been on comparison of three modelling approaches for turbulence-chemistry interaction and it has been shown that fairly standard RANS and LES approaches can provide satisfactory results. The selected model should predict spreading rate well and in this respect RSM brings improvement compared to standard k-?.

Turbulence-chemistry Interaction Approaches

It is clear that any model used to simulate the DJHC burner must be capable of representing mixtures of fuel, coflow and air, as well as predicting ignition. The three models that have been compared, EDC, CSE and DAFGM differ in the number of independent scalars in the turbulent flow and in the representation of the influence of turbulent fluctuations of these scalars. Furthermore EDC is more widely available. The success of an EDC model seems be due to parameter tuning and the proposed general formulation has not been widely tested. CSE and DAFGM provide better predictions and should be preferred. They are comparable to each other in the level of detail of chemistry carried to the turbulent flow calculation (i.e. number of independent scalar equations) and description of fluctuations (assumed PDF). The observed differences between the two might be due more to the implementation of scalar boundary conditions rather than in the scalar equations. It can be expected that other model variants with the same level of complexity (e.g. Flamelet Progress Variable (FPV)) provide similar accuracy. Higher accuracy can come from a more detailed representation of the scalar fluctuations. Kinetic scheme and chemistry reduction To date, an extensive comparison between different chemical reaction mechanisms has not been performed. An effect on the prediction of ignition is expected. However, if ignition and heat release characteristics are similar, the effect of other differences in the chemical reaction mechanism should be minor.

Boundary Conditions and Application Uncertainties

INLET BOUNDARY CONDITIONS

The coflow inlet conditions (at 3 mm) have been shown to greatly influence the results of the simulations. As one would expect, the more experimental data is employed to define the boundary condition, the better the (potential) quality of the results. The available data on radial profiles of mean temperature and temperature rms and mean oxygen mass fraction should be taken into account. The absence of measured data on the covariance of temperature and oxygen is a limitation of the experimental data set. It could possibly be overcome using information from separate simulations of the reaction flow inside the secondary burner. Special attention should be also given to the air inlet boundary. A problem arises because the measured profile does not extent into the air region and extra assumptions are needed. All simulations reported here have represented the air with a spatially uniform inflow velocity in axial direction. However, the imposed velocity can strongly influence the entrainment of air in the coflow, affecting the whole solution. It should be recalled that the outer wall of the coflow ends about 15 mm upstream of the location z=3mm. A more detailed representation of the flow development over this distance could be of interest to more accurately represent the air inlet condition.

OUTLET BOUNDARY CONDITIONS

Standard outflow boundary conditions can be applied if the computational domain has the size of the reported example simulations.

LATERAL SIDE BOUNDARY CONDITIONS

Standard symmetry and/or slip boundary conditions can be applied if the computational domain size has the size of the reported example simulations.

Recommendations for Future Work

Most of the simulations performed so far focused on one of the DJHC flames. It is important to test models for the other flames to validate whether they can predict trends with Reynolds number or coflow composition. Therefore, investigations should ultimately be made for all provided conditions. It has been concluded that the accurate representation of the boundary conditions plays an important role. Not only should all available experimental information be taken into account. But also separate simulations of the reaction flow inside the secondary burner could provide additional information, e.g. on the expected oxygen concentration fluctuations and the co-variance of temperature and oxygen. The use of other models especially capable of a better representation of scalar fluctuations should be further explored, e.g. RANS with transported PDF models and LES with the stochastic field method. Perpignan et al. (2018) give an overview of possible modelling approaches.

References

• Perpignan A.A.V., Gangoli Rao A., Roekaerts, D.J.E.M., Flameless Combustion and its Potential Towards Gas Turbines, Progress in Energy and Combustion Science 2018; 69: 28-62

---

Contributed by: André Perpignan, Dirk Roekaerts, E. Oldenhof, E.H. van Veen, M.J. Tummers, Hesheng Bao, Xu Huang — TU Delft

Contributed by: Jeffrey W. Labahn — Stanford University

Front Page

Description

Test Data

CFD Simulations

Evaluation

Best Practice Advice

© copyright ERCOFTAC 2018