Semester

Summer

Date of Graduation

2014

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Mechanical and Aerospace Engineering

Committtee Chair

Ismail Celik

Committee Co-Chair

V'yacheslav Akkerman

Committee Member

Christopher Atkinson

Committee Member

Dady Dadyburjor

Committee Member

Donald Ferguson

Committee Member

Gregory Thompson

Abstract

Combustion of fossil fuels is still the biggest source of power generation in the world. However, pollutants released to the atmosphere from combustion represent a risk for human health and the environment. Hence it is desirable to design a combustor that produces the maximum useful thermal power output while keeping low concentration levels of harmful emissions such as CO, P.M., NOx, and SOx. In the past, combustor design was aided by the compilation of large sets of experimental data and the development of empirical correlations which is an expensive process. Nowadays numerical simulations have become an important tool in the research and design of combustors. Numerical simulations allow the study of combustion systems under hazardous conditions and beyond their performance limits, and they are usually inexpensive and fast (compared to experiments). The main bottle-neck in combustion simulations is the accurate prediction of the concentration of the many species involved in combustion. Current computational fluid dynamic (CFD) simulations commonly use simplified versions of the chemical reaction mechanisms. But utilization of simplified chemical models comes with the associated inaccuracy while saving computational time.;In the present study the virtues of the chemical reactor network (CRN) approach are investigated and a new integration method is proposed to accelerate the calculation of species concentrations using reduced and detailed chemical mechanisms. Utilization of the CRN approach enabled the implementation of a detailed methane-air chemical mechanism that incorporates 53 chemical species and 325 reactions. The CRN approach was applied to two combustor configurations: a premixed methane-air swirl burner, and a non-premixed methane-air swirl burner. The CRN was built using results from the CFD simulations that were obtained using simplified chemical mechanisms with just one or two reactions. Numerical predictions of the premixed combustor behavior obtained using CRN simulations were compared with other CFD simulations that used mechanisms with more reactions and chemical species. The CRN results closely matched the CFD simulations with larger chemical mechanisms, the maximum relative difference of the predicted concentration for the major species (i.e. O 2, CO2, H2O, and N2) was 2.82% when compared to the CFD simulations. The calculation time of the CRN was greatly reduced, the maximum reduction of the CRN simulation took only one seventh of the computational time when compared with a CFD simulation. The CRN simulations of the non-premixed burner were also compared with experiments. Predicted spatial profiles of velocity, temperature, and mass fraction concentrations were compared with measurements. Results showed that the velocity and some mass fraction profiles matched the experimental measurements near the dump plane but it was found that downstream of the dump plane the temperature was overpredicted. Due to the temperature overprediction, the maximum difference was 250 [K], the nitrogen oxide (NO) concentration was overpredicted by 30 [ppm]. The relative difference of the predicted NO at the outlet of the combustor is 150% when compared with the experimental value.;Further, a novel integration method named log-time integration method (LTIM) was developed to calculate the solution of ideal reactors used in the CRN simulations. The integration method consists of the transformation of the time variable to the logarithmic space along with the use of variable time steps. The LTIM approach was applied to the solution of a perfectly stirred reactor (PSR) using a detailed chemical mechanism. PSR-LTIM results were compared with a commercial PSR code which is available in the CHEMKIN software package. The maximum relatively difference of the concentration of the species of interest was only 1%. Calculated species concentration using the PSR-LTIM matched the results from CHEMKIN with comparable computational time, the computational time of the PSR-LTIM was 5.3 [s] and for CHEMKIN was 3 [s]. The integration method was compared to higher order integration methods available in the literature producing satisfactory results with less CPU time, the LTIM approach took one fifth of the computational time of a higher order integration method. The LTIM was also applied to the solution of a premixed one dimensional methane-air flame, FLAME-LTIM, where a mechanism incorporating nine chemical species and five global reactions mechanism was used. Calculated temperature and mass fraction profiles matched closely the results obtained using the equivalent commercial code CHEMKIN PREMIX. The relative temperature difference at the outlet of the domain was 0.5% and the maximum difference in the chemical specie concentration at the outlet of the domain was 13.2%.;The outcome of the present research can be used to perform a rapid design analysis of gas turbines and similar combustors to achieve low levels of emissions.

Share

COinS