Robert Scharler

Validation of flow simulation and gas combustion sub-models for the CFD-based prediction of NOx formation in biomass grate furnaces

While reasonably accurate in simulating gas phase combustion in biomass grate furnaces, CFD tools based on simple turbulence–chemistry interaction models and global reaction mechanisms have been shown to lack in reliability regarding the prediction of NOx formation. Coupling detailed NOx reaction kinetics with advanced turbulence–chemistry interaction models is a promising alternative, yet computationally inefficient for engineering purposes. In the present work, a model is proposed to overcome these difficulties. The model is based on the Realizable k–ϵ model for turbulence, Eddy Dissipation Concept for turbulence–chemistry interaction and the HK97 reaction mechanism. The assessment of the sub-models in terms of accuracy and computational effort was carried out on three laboratory-scale turbulent jet flames in comparison with the experimental data. Without taking NOx formation into account, the accuracy of turbulence modelling and turbulence–chemistry interaction modelling was systematically examined on Sandia Flame D and Sandia CO/H2/N2 Flame B to support the choice of the associated models. As revealed by the Large Eddy Simulations of the former flame, the shortcomings of turbulence modelling by the Reynolds averaged Navier–Stokes (RANS) approach considerably influence the prediction of the mixing-dominated combustion process. This reduced the sensitivity of the RANS results to the variations of turbulence–chemistry interaction models and combustion kinetics. Issues related to the NOx formation with a focus on fuel bound nitrogen sources were investigated on a NH3-doped syngas flame. The experimentally observed trend in NOx yield from NH3 was correctly reproduced by HK97, whereas the replacement of its combustion subset by that of a detailed reaction scheme led to a more accurate agreement, but at increased computational costs. Moreover, based on results of simulations with HK97, the main features of the local course of the NOx formation processes were identified by a detailed analysis of the interactions between the nitrogen chemistry and the underlying flow field.

CFD simulation of ash deposit formation in fixed bed biomass furnaces and boilers

In order to describe and predict the formation of ash deposits in biomass fired combustion plants, a mathematical model is being developed and implemented into the CFD code Fluent® as a post processing tool. At the present state of development the model covers the release of coarse ash particles and ash-forming vapours from the fuel bed, the transport to furnace and boiler surfaces and the deposition of particles and vapours. Changes in the flue gas composition due to chemical reactions are considered by performing thermodynamic equilibrium calculations in order to determine the gas phase composition. The build-up of the deposit layer depending on wall temperature, deposit porosity and chemical composition is also taken into account. First results of modelling deposit formation for the furnace and fire tube of a grate fired combustion unit (boiler capacity 440 kWth) show plausible results but have yet to be validated.

A kinetic study on the potential of a hybrid reaction mechanism for prediction of NOx formation in biomass grate furnaces

This paper presents the verification of a hybrid reaction mechanism (28 species, 104 reactions) by means of a kinetic study with a view to its application for the CFD-based prediction of gas phase combustion and NOx formation in biomass grate furnaces. The mechanism is based on a skeletal kinetic scheme that includes the subsets for H2, CO, NH3 and HCN oxidation derived from the detailed Kilpinen 97 reaction mechanism. To account for the CH4 breakdown two related reactions from the 4-step global mechanism for hydrocarbons oxidation by Jones and Lindstedt were adopted. The hybrid mechanism was compared to the global mechanism and validated against the detailed Kilpinen 97 mechanism. For that purpose plug flow reactor simulations at conditions relevant to biomass combustion (atmospheric pressure, 1200–1600 K) for approximations of the flue gases in a grate furnace at fuel lean and fuel rich conditions were carried out. The hybrid reaction mechanism outperformed the global one at all conditions investigated. The most striking differences obtained in predictions by the hybrid and the detailed mechanism at the residence times prior to ignition were attributed to the simplified description of the CH4 oxidation in the case of the former. The overall agreement regarding both combustion and NOx chemistry between the hybrid and the detailed mechanism was better at fuel lean conditions than at fuel rich conditions. However, also at fuel rich conditions, the agreement was improving with increasing temperature. Moreover, it was shown that an improvement in the prediction of NOx formation by the N-subset of the hybrid reaction mechanism can be achieved by replacing its C–H–O subset with that of the detailed one.

A method for reduction of computational time of local equilibria for biomass flue gas compositions in CFD

In situ adaptive tabulation has been used to store equilibria states in a tree-like structure to be used in prediction of local condensation rates in CFD simulations of heat exchanger tubes. An interface to several tools to compute chemical equilibrium in flue gas compositions was developed. The results show a reduction of computation times at least by a factor of 8 (Table 4). Relative tolerances of 10−4 are sufficient to interpolate the table. Simulations indicate that mesh resolution is more important than a smaller tolerance.