Study on the potential of BML-approach and G-equation concept-based models for predicting swirling partially premixed combustion systems: URANS computations

Abstract

In this work the potential of two combustion modeling approaches (BML and G-equation based models) for partially premixed flames in combustion systems of various complexities is investigated using URANS computations. The first configuration consists of a nonconfined swirled premixed methane/air flame (swirl number 0.75) exhibiting partially premixed effects due to coflowing. The system is studied either in the isothermal case or in the reacting mode and for different thermal powers. The second configuration represents a model GT combustion chamber and features the main properties of real GT combustors: a confined swirled flow with multiple recirculation zones and reattachment points, resulting in a partially premixed methane/air aerodynamically stabilized flame and an additional diffusion flame formed by the fuel and oxidizer not consumed in the premixed flame. This makes it possible to subject the modeling to variation of different parameters, such as confinement, Re-number or flame power, or adiabatic or nonadiabatic conditions. For this purpose an extended Bray–Moss–Libby model and a G-equation-based approach, both coupled to the mixture fraction transport equation to account for partially premixed effects, are used following the so-called conditional progress variable approach (CPVA). The radiation effects are also taken into account. To account for the turbulence–chemistry interaction, a (multivariate) presumed PDF approach is applied. The results are compared with LDV, Raman, and PLIF measurements. Beyond a pure validation, the URANS is used to capture the presence of the precessing vortex core and to analyze the performance of different modeling strategies of partially premixed combustion in capturing the expansion ratio, species formation conditioned on the flame front, and flame front stabilization. It appears that the combustion models used are able to achieve plausible results in the complex combustion systems under study, while the BML-based model affords less computational time.