Cannon, SM
1999
PDF Modeling of Lean Premixed Combustion Using In Situ Tabulated Chemistry
Cannon, S.M.; Brewster, B.S. and Smoot, L.D.
Combustion & Flame, 119:233-252 (1999).
The velocity-composition probability density function (pdf) model coupled with a k-?-based mean flow computational fluid dynamics (CFD) model was used to describe the turbulent fluid flow, heat transfer, chemistry, and their interactions in a bluff-body, lean, premixed, methane-air combustor. Measured data [1,2] including velocity, temperature, and chemical species concentrations were used to evaluate the model. The chemistry calculations were performed with an in situ look-up tabulation method [3]. A reduced, 5-step chemical mechanism [4] for describing fuel oxidation, CO, and NO chemistry was used in the model. NO formation from thermal, N2O-intermediate, and prompt pathways was included in the 5-step mechanism. An axisymmetric, unstructured grid was used for solving the Eulerian, mean flow equations and the vertices were used to store mean statistics for solving the Lagrangian, fluid particle equations. Predicted velocity and composition mean statistics were compared to measurements in the bluff-body combustor for a lean equivalence ratio of 0.59. The predictions of major species matched measured and calculated equilibrium values in the recirculation zone. Comparisons of mean CO throughout the combustor were always within an order of magnitude and showed marked improvements over past predictions. Maximum discrepancies between measured and predicted NO concentrations were between 5 and 7 ppm (~50%). The accessed composition space in this turbulent combustion simulation represented the values of species mole fraction and enthalpy for each fluid particle at each time step and was found to lie in a relatively small, uniquely shaped region that was dictated by the mixing, reaction, and heat transfer in the combustor. This accessed composition region was obtained in situ and required about 35 megabytes of storage once a steady state was reached. This memory requirement was more than three orders of magnitude less than would be needed in a standard, a priori table. The in situ tabulation approach allowed for technically correct and efficient chemical kinetic calculations by using the 5-step mechanism in this pdf-method-based, multidimensional combustor model.
1998
Stochastic Modeling of CO and NO in Premixed Methane Combustion
Cannon, S.M.; Brewster, B.S. and Smoot, L.D.
Combustion & Flame, 113:135-146 (1998).
The ability to use reduced CH4-air chemical mechanisms to predict CO and NO emissions in premixed turbulent combustion has been evaluated in a Partially Stirred Reactor (PaSR) model. CO emissions were described with reduced 4-, 5-, and 9-step mechanisms and a detailed 276-step mechanism. NO emissions from thermal, N2O-intermediate, and prompt pathways were included in the 5-, 9-, and 276-step mechanisms. Molecular mixing was described with a deterministic, Interaction-by Exchange-with-the-Mean (IEM) submodel. Random selection and replacement (without repetition) of fluid particles were used to simulate through-flow. The evolution of mean and root mean square (rms) temperature, CO, and NO in the PaSR was accurately described with the 9-step mechanism over a wide range in mixing frequency and equivalence ratio. Also, the 9-step mechanism provided accurate instantaneous reaction rates and concentrations for a broad region of the accessed composition space in the PaSR. The 5-step mechanism performed less reliably than the 9-step mechanism at phi = 1.0 but performed similarly to the 9-step mechanism at phi = 0.65. The 4-step mechanism underpredicted mean CO values and overpredicted instantaneous temperature reaction rates, most likely due to its inferior parent mechanism, partial equilibrium assumption for OH, and unallowed dissociation of neglected radical species. The detailed and reduced mechanism predictions of the accessed composition space in the PaSR covered only a small fraction of the allowable composition space, thus facilitating the use of an efficient in situ chemical look-up table for multidimensional, pdf-method calculations.
1997
Modeling of Lean Premixed Gaseous Turbulent Combustion
Cannon, S.M.
Modeling of Lean Premixed Gaseous Turbulent Combustion, Ph.D./BYU, December 1997. Advisor: Smoot
Stochastic Modeling of CO and NO in Premixed Methane Combustion
Cannon, S.M.; Brewster, B.S. and Smoot, L.D.
Combustion & Flame, (in press), 1997. Funded by ACERC.
The ability to use reduced CH4-air chemical mechanisms to predict CO and NO emission in premixed turbulent combustion has been evaluated in a Partially Stirred Reactor (PaSR) model. CO emissions were described with reduced 4-, 5-, and 9-step mechanisms and a detailed 276-step mechanism. NO emissions from thermal N2O-intermediate and prompt pathways were included in the 5-, 9- and 276-step mechanisms. Molecular mixing was described with a deterministic, Interaction-by-Exchange-with-the-Mean (IEM) submodel. Random selection and replacement (without repetition) of fluid particles was used to simulate through-flow. The evolution of mean and rms temperature, CO, and NO in the PaSR was accurately described with the 9-step mechanisms over a wide range in mixing frequency and equivalence ratio. Also, the 9-step mechanism provided accurate instantaneous reaction rates and concentrations for a broad region of the accessed composition space in the PaSR. The 5-step mechanism performed less reliably than the 9-step mechanism at phi = 1.0 but performed similarly to the 9-step mechanism at phi = 0.65. The 4-step mechanism underpredicted mean CO values and overpredicted instantaneous temperature reaction rates, most likely due to its inferior parent mechanism, partial equilibrium assumption for OH, and unallowed dissociation of neglected radical species. The detailed reduced mechanism predictions of the accessed composition space in the PaSR covered only a small fraction of the allowable composition space, thus facilitating the use of an efficient, in situ chemical look-up table in multi-dimensional, pdf-method calculations.
Comprehensive Model for Lean Premixed Combustion in Industrial Gas Turbines - Part I. Validation
Cannon, S.M.; Brewster, B.S.; Smoot, L.D.; Murray, R. and Hedman, P.O.
Presented at the Spring Meeting of the Western States Section/The Combustion Institute, Sandia National Laboratories, Livermore, California, April 14-15, 1997. Funded by US Department of Energy.
The velocity composition pdf model coupled with a mean flow CFD model was used to describe the turbulent fluid flow, heat transfer, chemistry, and their interactions in a swirling, lean premixed, methane-air combustor for which laser-based measurements of mean velocity and temperature were made. A flame was stabilized in this axi-symmetric, lab-scale, gas-turbine combustor (LSGTC. A reduced, 5-step chemical mechanism, for describing fuel oxidation and NO chemistry, was used in this LSGTC model. NO emissions from thermal, N2)-intermediate, and prompt pathways were described with this 5-step mechanism. The chemistry calculations were performed efficiently with and in-situ look-up table. An axi-symmetric, unstructured grid consisting of 2283 vertices and 4302 triangular elements was used for solving the Eulerian, mean flow equations and the vertices were used to store mean statistics for solving the Lagrangian, fluid particle (~310,000 fluid particles) equations. Predicted velocity and composition statistics were compared to measurements in the LSGTC for lean equivalence ratios of 0.8 and 0.65. The comparisons of predicted mean velocity and temperature were reasonable good throughout the combustor. The location and magnitude of peak axial velocity was well represented by the model at near inlet regions, through the negative mean axial velocity in the internal recirculation zone was over-predicted. The predicted maximum mean temperature and the penetration zone of the cold unburned fluid were in reasonable agreement with the experimental data. Correct trends in CO and NO with equivalence ration were predicted with the model. The in situ tabulation method was used to represent the chemical kinetics in this axi-symmetric combustor without requiring significant CPU time and memory. The model is currently being applied to simulate 3-dimensional, gas-turbine combustor geometries and is described in a companion paper.
1996
Prediction of CO and NOx in Lean Premixed Turbulent Combustion
Cannon, S.M.; Brewster, B.S. and Smoot L.D.
Proceedings of the Fall 1996 Meeting of the Western States Section/The Combustion Institute, The University of Southern California, Los Angeles, California, October 28-29, 1996. Funded by ACERC.
The ability to use reduced CH4-air chemical mechanisms to predict CO and NOx emissions in lean premixed turbulent combustion has been evaluated in a Partially Stirred Reactor (PaSR) model. CO emissions were described with mathematically reduced 4-, 5- and 9-step mechanism and a detailed 276-step mechanism. NOx emission form thermal, N2O-intermediate, and prompt pathways were described with the 5-, and 9-step reduced mechanisms provided accurate instantaneous reaction rate calculations for a broad region of the accessed composition space in the PaSR. The 4-step mechanism and the partial equilibrium assumption for OH. Practicality of using the 5- and 9-step mechanisms in industrial, 3-dimensional calculations may require the use of a novel, in situ look-up table
1994
Time-Resolved Temperature Measurements in an Elliptic Cross-Section, Turbulent, Diffusion Jet Flame
Cannon, S.M. and Queiroz, M.
1994 ASME Winter Annual Meeting, 296:37, 1994. Funded by ACERC.
An experimental study to determine the thermal structure of an elliptic cross-section, turbulent, diffusion jet flame using time-resolved temperature measurements was performed. The 2:1 aspect ratio fuel inlet allowed a fully developed turbulent flow (Reynolds number = 6000) of propane to exit into near ambient air. Measurements of mean and rms temperature, as well as power spectral densities (psd) and probability density functions (pdf) of temperature were obtained along the centerline and radially along the major and minor axis at axial stations ranging from 5<=z/Dh<=30. Similar to observed behavior in axi-symmetric diffusion flames, the non-axi-symmetric flame showed evidence of small-scale, momentum-driven vortices inside the flame zone and large-scale, buoyancy-driven vortices outside the flame zone. Differences in these turbulent shear layers along the major and minor axis were observed, as the minor axis had 25% higher temperature fluctuations along the fuel side mixing layer and the major axis had 20% higher temperature fluctuations along the air side mixing layer. A weaker fuel side shear layer along the major axis allowed more radial movement and a more radially stretched reaction zone in the near-burner region along this same major axis. This greater radial movement was sufficiently strong to cause a faster destruction of the inner vortex structures, such that less mixing would be observed along the fuel side of the major axis.
Cannon, SM
1999
PDF Modeling of Lean Premixed Combustion Using In Situ Tabulated Chemistry
Cannon, S.M.; Brewster, B.S. and Smoot, L.D.
Combustion & Flame, 119:233-252 (1999).
The velocity-composition probability density function (pdf) model coupled with a k-?-based mean flow computational fluid dynamics (CFD) model was used to describe the turbulent fluid flow, heat transfer, chemistry, and their interactions in a bluff-body, lean, premixed, methane-air combustor. Measured data [1,2] including velocity, temperature, and chemical species concentrations were used to evaluate the model. The chemistry calculations were performed with an in situ look-up tabulation method [3]. A reduced, 5-step chemical mechanism [4] for describing fuel oxidation, CO, and NO chemistry was used in the model. NO formation from thermal, N2O-intermediate, and prompt pathways was included in the 5-step mechanism. An axisymmetric, unstructured grid was used for solving the Eulerian, mean flow equations and the vertices were used to store mean statistics for solving the Lagrangian, fluid particle equations. Predicted velocity and composition mean statistics were compared to measurements in the bluff-body combustor for a lean equivalence ratio of 0.59. The predictions of major species matched measured and calculated equilibrium values in the recirculation zone. Comparisons of mean CO throughout the combustor were always within an order of magnitude and showed marked improvements over past predictions. Maximum discrepancies between measured and predicted NO concentrations were between 5 and 7 ppm (~50%). The accessed composition space in this turbulent combustion simulation represented the values of species mole fraction and enthalpy for each fluid particle at each time step and was found to lie in a relatively small, uniquely shaped region that was dictated by the mixing, reaction, and heat transfer in the combustor. This accessed composition region was obtained in situ and required about 35 megabytes of storage once a steady state was reached. This memory requirement was more than three orders of magnitude less than would be needed in a standard, a priori table. The in situ tabulation approach allowed for technically correct and efficient chemical kinetic calculations by using the 5-step mechanism in this pdf-method-based, multidimensional combustor model.
1998
Stochastic Modeling of CO and NO in Premixed Methane Combustion
Cannon, S.M.; Brewster, B.S. and Smoot, L.D.
Combustion & Flame, 113:135-146 (1998).
The ability to use reduced CH4-air chemical mechanisms to predict CO and NO emissions in premixed turbulent combustion has been evaluated in a Partially Stirred Reactor (PaSR) model. CO emissions were described with reduced 4-, 5-, and 9-step mechanisms and a detailed 276-step mechanism. NO emissions from thermal, N2O-intermediate, and prompt pathways were included in the 5-, 9-, and 276-step mechanisms. Molecular mixing was described with a deterministic, Interaction-by Exchange-with-the-Mean (IEM) submodel. Random selection and replacement (without repetition) of fluid particles were used to simulate through-flow. The evolution of mean and root mean square (rms) temperature, CO, and NO in the PaSR was accurately described with the 9-step mechanism over a wide range in mixing frequency and equivalence ratio. Also, the 9-step mechanism provided accurate instantaneous reaction rates and concentrations for a broad region of the accessed composition space in the PaSR. The 5-step mechanism performed less reliably than the 9-step mechanism at phi = 1.0 but performed similarly to the 9-step mechanism at phi = 0.65. The 4-step mechanism underpredicted mean CO values and overpredicted instantaneous temperature reaction rates, most likely due to its inferior parent mechanism, partial equilibrium assumption for OH, and unallowed dissociation of neglected radical species. The detailed and reduced mechanism predictions of the accessed composition space in the PaSR covered only a small fraction of the allowable composition space, thus facilitating the use of an efficient in situ chemical look-up table for multidimensional, pdf-method calculations.
1997
Modeling of Lean Premixed Gaseous Turbulent Combustion
Cannon, S.M.
Modeling of Lean Premixed Gaseous Turbulent Combustion, Ph.D./BYU, December 1997. Advisor: Smoot
Stochastic Modeling of CO and NO in Premixed Methane Combustion
Cannon, S.M.; Brewster, B.S. and Smoot, L.D.
Combustion & Flame, (in press), 1997. Funded by ACERC.
The ability to use reduced CH4-air chemical mechanisms to predict CO and NO emission in premixed turbulent combustion has been evaluated in a Partially Stirred Reactor (PaSR) model. CO emissions were described with reduced 4-, 5-, and 9-step mechanisms and a detailed 276-step mechanism. NO emissions from thermal N2O-intermediate and prompt pathways were included in the 5-, 9- and 276-step mechanisms. Molecular mixing was described with a deterministic, Interaction-by-Exchange-with-the-Mean (IEM) submodel. Random selection and replacement (without repetition) of fluid particles was used to simulate through-flow. The evolution of mean and rms temperature, CO, and NO in the PaSR was accurately described with the 9-step mechanisms over a wide range in mixing frequency and equivalence ratio. Also, the 9-step mechanism provided accurate instantaneous reaction rates and concentrations for a broad region of the accessed composition space in the PaSR. The 5-step mechanism performed less reliably than the 9-step mechanism at phi = 1.0 but performed similarly to the 9-step mechanism at phi = 0.65. The 4-step mechanism underpredicted mean CO values and overpredicted instantaneous temperature reaction rates, most likely due to its inferior parent mechanism, partial equilibrium assumption for OH, and unallowed dissociation of neglected radical species. The detailed reduced mechanism predictions of the accessed composition space in the PaSR covered only a small fraction of the allowable composition space, thus facilitating the use of an efficient, in situ chemical look-up table in multi-dimensional, pdf-method calculations.
Comprehensive Model for Lean Premixed Combustion in Industrial Gas Turbines - Part I. Validation
Cannon, S.M.; Brewster, B.S.; Smoot, L.D.; Murray, R. and Hedman, P.O.
Presented at the Spring Meeting of the Western States Section/The Combustion Institute, Sandia National Laboratories, Livermore, California, April 14-15, 1997. Funded by US Department of Energy.
The velocity composition pdf model coupled with a mean flow CFD model was used to describe the turbulent fluid flow, heat transfer, chemistry, and their interactions in a swirling, lean premixed, methane-air combustor for which laser-based measurements of mean velocity and temperature were made. A flame was stabilized in this axi-symmetric, lab-scale, gas-turbine combustor (LSGTC. A reduced, 5-step chemical mechanism, for describing fuel oxidation and NO chemistry, was used in this LSGTC model. NO emissions from thermal, N2)-intermediate, and prompt pathways were described with this 5-step mechanism. The chemistry calculations were performed efficiently with and in-situ look-up table. An axi-symmetric, unstructured grid consisting of 2283 vertices and 4302 triangular elements was used for solving the Eulerian, mean flow equations and the vertices were used to store mean statistics for solving the Lagrangian, fluid particle (~310,000 fluid particles) equations. Predicted velocity and composition statistics were compared to measurements in the LSGTC for lean equivalence ratios of 0.8 and 0.65. The comparisons of predicted mean velocity and temperature were reasonable good throughout the combustor. The location and magnitude of peak axial velocity was well represented by the model at near inlet regions, through the negative mean axial velocity in the internal recirculation zone was over-predicted. The predicted maximum mean temperature and the penetration zone of the cold unburned fluid were in reasonable agreement with the experimental data. Correct trends in CO and NO with equivalence ration were predicted with the model. The in situ tabulation method was used to represent the chemical kinetics in this axi-symmetric combustor without requiring significant CPU time and memory. The model is currently being applied to simulate 3-dimensional, gas-turbine combustor geometries and is described in a companion paper.
1996
Prediction of CO and NOx in Lean Premixed Turbulent Combustion
Cannon, S.M.; Brewster, B.S. and Smoot L.D.
Proceedings of the Fall 1996 Meeting of the Western States Section/The Combustion Institute, The University of Southern California, Los Angeles, California, October 28-29, 1996. Funded by ACERC.
The ability to use reduced CH4-air chemical mechanisms to predict CO and NOx emissions in lean premixed turbulent combustion has been evaluated in a Partially Stirred Reactor (PaSR) model. CO emissions were described with mathematically reduced 4-, 5- and 9-step mechanism and a detailed 276-step mechanism. NOx emission form thermal, N2O-intermediate, and prompt pathways were described with the 5-, and 9-step reduced mechanisms provided accurate instantaneous reaction rate calculations for a broad region of the accessed composition space in the PaSR. The 4-step mechanism and the partial equilibrium assumption for OH. Practicality of using the 5- and 9-step mechanisms in industrial, 3-dimensional calculations may require the use of a novel, in situ look-up table
1994
Time-Resolved Temperature Measurements in an Elliptic Cross-Section, Turbulent, Diffusion Jet Flame
Cannon, S.M. and Queiroz, M.
1994 ASME Winter Annual Meeting, 296:37, 1994. Funded by ACERC.
An experimental study to determine the thermal structure of an elliptic cross-section, turbulent, diffusion jet flame using time-resolved temperature measurements was performed. The 2:1 aspect ratio fuel inlet allowed a fully developed turbulent flow (Reynolds number = 6000) of propane to exit into near ambient air. Measurements of mean and rms temperature, as well as power spectral densities (psd) and probability density functions (pdf) of temperature were obtained along the centerline and radially along the major and minor axis at axial stations ranging from 5<=z/Dh<=30. Similar to observed behavior in axi-symmetric diffusion flames, the non-axi-symmetric flame showed evidence of small-scale, momentum-driven vortices inside the flame zone and large-scale, buoyancy-driven vortices outside the flame zone. Differences in these turbulent shear layers along the major and minor axis were observed, as the minor axis had 25% higher temperature fluctuations along the fuel side mixing layer and the major axis had 20% higher temperature fluctuations along the air side mixing layer. A weaker fuel side shear layer along the major axis allowed more radial movement and a more radially stretched reaction zone in the near-burner region along this same major axis. This greater radial movement was sufficiently strong to cause a faster destruction of the inner vortex structures, such that less mixing would be observed along the fuel side of the major axis.