Incompletely Stirred Reactor Network Modeling of a Model Gas Turbine Combustor

Abstract

The prediction of soot emissions is a requirement for the development of gas turbine combustors. Apart from global quantities related to soot mass, future regulations also call for the control of particle number. Therefore, theoretical models for soot from combustion devices must include various nucleation, growth, and oxidation mechanisms and aerosol physics in order to predict the soot particle number distribution. This paper introduces an approach based on Incompletely Stirred Reactor Network (ISRN) modeling that simplifies calculations and facilitates parametric analyses with very complex soot models. The method is based on the Conditional Moment Closure (CMC) combustion model and Incompletely Stirred Reactor (ISR) theory, which is here extended to a reactor network formulation. An ISR is a volume that is inhomogeneous in terms of mixture fraction but is characterized by homogeneous conditional averages, such as temperature conditioned on the mixture fraction having a particular value. A network of ISRs can then be deployed to separately capture soot production and oxidation regions exhibiting different degrees of micro-mixing rates and residence times, as typically observed in practical combustors. The ISRN approach is demonstrated in this paper for a model aero-engine combustor for which detailed CFD and experimental data exist, and it is found that reasonable accuracy is produced at a significantly reduced computational cost.

Soot Emission Simulations of a Single Sector Model Combustor Using Incompletely Stirred Reactor Network Modeling

“…Recent investigations in this combustor show that the mean soot location can be very well captured by the ISRN, consistent with both detailed CFD and experimental data [50]. This is probably a result of the different flow patterns and range of scalar dissipation rates in this combustor compared to the Cambridge an adequate balance between reaction-transport-diffusion is established.…”

Comprehensive soot particle size distribution modelling of a model Rich-Quench-Lean burner

“…Previous experimental studies in this burner have shown that these conditions drastically affect soot emission [25,26,27,28]. Similar trends have been observed via numerical investigations in similar conditions with a slightly leaner global equivalence ratio [48].…”

“…3. DNS studies of sooting flames [45,46], as well as LES studies [39,48], show that conditional fluctuations of soot quantities are large even when conditional fluctuations of soot precursors are small. This phenomenon takes place at both the large and small scales, and it is more pronounced when aromatics-based pathways are included in the soot model [47,29].…”

“…A similar equation is solved for the enthalpy with the conditional chemical source term being replaced by a conditional heat loss term ω h |η. Here, an opticallythin approximation is employed to account for the radiation losses emitted by soot particles, water vapour, CO 2 , CO and CH 4 , which has been proven a satisfactory approximation for the conditions studied herein [48]. Radiation from the gaseous-phase species is based on the RADCAL model with radiation properties taken from the online TNF workshop database (http://www.sandia.…”

“…A similar numerical setup to Refs. [31,48] is adopted here. The computational domain reproducing the experimental rig is shown in Fig.…”