The jet-in-hot-coflow is a canonical combustion setup, which has been used in several studies to study Flameless/MILD combustion and auto-ignition of fuels. However, the NO_x and CO emission measurements from these combustion setups were not possible due to the entrainment of laboratory air and a lack of a well-defined physical system limit. These limitations have been overcome by a new enclosed jet-in-hot-coflow setup. The combustor was operated by injecting a mixture of CH₄-Air in the central jet, and the coflow comprised of hot products from CH₄-Air combustion in burners upstream. The coflow composition was further controlled by adding diluents such as N₂ and CO₂. Measurements were done using stereoscopic particle image velocimetry, suction probe gas analysis, thermocouples, and chemiluminescence imaging. Increasing central jet velocity and equivalence ratio led to lower NO_x and a reaction zone that enlarged and shifted downstream. The reduction in NO_x emission was attributed to the returning mechanism. Adding CO₂ and N₂ as diluents in the coflow resulted in a longer combustion zone and reduced temperatures in the combustion chamber, leading to decreased NO_x production and increased reburning. These experiments provide relevant flowfield and emissions data for modelers and help characterize combustion regimes such as Flameless/MILD.

@en

Flameless Combustion is an interesting low NO_x combustion technology for gas turbine engines. In order to design systems for stringent performance standards, it is important to understand emission formation in this regime. To this end, the characteristics of a combustor capable of operating in the Flameless regime are studied. Particle Image Velocimetry and thermocouple measurements were performed to obtain the velocity field and gas temperatures respectively, in the combustor under reacting conditions. Results from experiments were used to generate an "informed" chemical reactor network (CRN) model from which, temperature and species distributions were obtained. As such, this paper presents measured data and a methodology to combine it with CRN modelling to obtain gas composition and temperature. The temperature, NO_x, CO and O₂ mole fractions obtained at three different operating conditions shall be validated with gas composition measurements in the future.

@en

The Flameless Combustion (FC) regime has been pointed out as a promising combustion technique to lower the emissions of nitrogen oxides (NOx) while maintaining low CO and soot emissions, as well as high efficiencies. However, its accurate modeling remains a challenge. The prediction of pollutant species, especially NOx, is affected by the usually low total values that require higher precision from computational tools, as well as the incorporation of relevant formation pathways within the overall reaction mechanism that are usually neglected. The present work explores a multiple step modeling approach to tackle these issues. Initially, a CFD solution with simplified chemistry is generated [both the Eddy Dissipation Model (EDM) as well as the Flamelet Generated Manifolds (FGM) approach are employed]. Subsequently, its computational cells are clustered to form ideal reactors by user-defined criteria, and the resulting Chemical Reactor Network (CRN) is subsequently solved with a detailed chemical reaction mechanism. The capabilities of the clustering and CRN solving computational tool (AGNES—Automatic Generation of Networks for Emission Simulation) are explored with a test case related to FC. The test case is non-premixed burner based on jet mixing and fueled with CH4 tested for various equivalence ratios. Results show that the prediction of CO emissions was improved significantly with respect to the CFD solution and are in good agreement with the experimental data. As for the NOx emissions, the CRN results were capable of reproducing the non-monotonic behavior with equivalence ratio, which the CFD simulations could not capture. However, the agreement between experimental values and those predicted by CRN for NOx is not fully satisfactory. The clustering criteria employed to generate the CRNs from the CFD solutions were shown to affect the results to a great extent, pointing to future opportunities in improving the multi-step procedure and its application.@en