Smith, PJ
1993
Distributed Combustion Simulations
Sikorski, K.; Ma, K.-L.; Smith, P.J. and Adams, B.R.
Energy & Fuels, 7 (6):902-905, 1993. Funded by ACERC.
This paper reports research in progress. Two types of domain decomposition have been used in distributed computing with networked workstations for the numerical modeling of full-scale utility boilers. The numerical model is a three-dimensional combustion code that couples turbulent computational fluid dynamics with the chemical reaction process and the radiative heat transfer. Two approaches, here called microscale parallelism and macroscale parallelism, are proposed to study the intrinsic parallelism of typical combustion simulations. We describe the implementation of the microscale parallelism as well as its performance on networked workstations.
Parametric Sensitivity Study of a CFD-Based Coal Combustion Model
Smith, J.D.; Smith, P.J. and Hill, S.C.
AIChE Journal, 39, (10):1668-1679, 1993. Funded by Combustion Laboratory Consortium through Brigham Young University.
Parametric sensitivity of a two-dimensional pulverized-fuel (PF) combustion model is studied extensively for the effect of parametric uncertainty on model predictions. Results show that error in coal devolatilization/oxidation parameters has the dominant effect on predicted burnout, NOx formation, local gas temperature, and coal-gas mixture fraction. Uncertainty in the turbulent particle dispersion parameters appears to have a secondary effect, while error in the particle-gas radiation parameters has little impact on model predictions. Regions of the computational domain exhibiting sensitivity to specific parameters are identified. Specific parameter sensitivity implies the relative importance of various mechanisms in the overall process. Turbulent particle dispersion seems to be important early in the reactor with kinetic processes dominating at and following the predicted ignition point. Radiation appears to be of minor importance. These results indicate the need for a better method of predicting the overall volatiles yield and further understanding of the devolatilization/oxidation mechanism and its role in the overall PF combustion process. The study provides fundamental direction for future comprehensive model development and focuses on experimental work to better quantify critical input parameters.
Cloud Tracing in Convection-Diffustion Systems
Ma, K.-L. and Smith, P.J.
Proceedings of the Visualization 93 IEE/ACM SIGGRPHY Conference:253-259, San Jose, CA, October 1993. Funded by IBM and ACERC.
This paper describes a highly interactive method for computer visualization of simultaneous three-dimensional vector and scalar flow fields in convection-diffusion systems. This method allows a computational fluid dynamics user to visualize the basic physical process of dispersion and mixing rather than just the vector and scalar values computed by the simulation. It is based on transforming the vector field from a traditionally Eulerian reference frame into a Lagrangian reference frame. Fluid elements are traced through the vector field for the mean path as well as the statistical dispersion of the fluid elements about the mean position by using added scalar information about the root mean square value of the vector field and its Lagrangian time scale. In this way, clouds of fluid elements are traced not just mean paths. We have used this method to visualize the simulation of an industrial incinerator to help identify mechanisms for poor mixing.
1992
Virtual Smoke: An Interactive 3D Flow Visualization Technique
Ma, K.-L. and Smith, P.J.
Visualization Conference, Boston, MA, October 1992. Funded by International Business Machines and ACERC.
The paper introduces a new technique for computer visualization of simultaneous three-dimensional vector and scalar fields such as velocity and temperature in reaction fluid flow fields. The technique, which we call Virtual Smoke, simulates the use of colored smoke for experimental gaseous fluid flow visualization. However, it is noninvasive and can animate, in particular, the dynamic behaviors of steady state or instantaneous flow fields obtained from numerical simulations. Virtual Smoke is based on Volume Seeds and Volume Seedlings, which are direct volume visualization methods previously developed for highly interactive scalar volume data exploration. We use data from combustion simulations to demonstrate the effectiveness of Virtual Smoke.
1991
Detailed Model for Practical Pulverized Coal Furnaces and Gasifiers, Volume II, User Manual for 1990 Version Pulverized Coal Gasification and Combustion 3-Dimensional (90-PCGC-3), Final Report
Smith, P.J.; Smoot, L.D.; Hill, S.C. and Eaton, A.M.
Advanced Combustion Engineering Research Center, 1991. Funded by US Department of Energy, Consortium and ACERC.
The theoretical foundations, numerical approach and a guide for users of 90-PCGC-3 are presented. 90-PCGC-3 is a generalized, three-dimensional, steady state model that can be used to predict the behavior of a variety of reactive and non-reactive (isothermal) fluid flows. The model solves the Navier-Stokes equations in three-dimensional, Cartesian coordinates. Turbulence is accounted for in both the fluid mechanics equations and the combustion solution scheme. Major gas-phase reactions are modeled assuming local instantaneous equilibrium, and thus the reaction rates are limited by the turbulent rate of mixing. The gas phase is described by elliptic partial differential equations that are solved by an iterative line-by-line technique. Under-relaxation is used to achieve numerical stability.
1990
Furnace Design Using Comprehensive Combustion Models
Smith, P.J.; Sowa, W.A. and Hedman, P.O.
Combustion and Flame, 79 (2), 111-121, 1990. Funded by Brigham Young Unviersity.
A new design methodology is presented that allows for the use of comprehensive coal combustion codes in design applications and provides a priori information on the cost of the optimization. A statistical response surface methodology is used to determine appropriate sample points from the design space at which the computation for the comprehensive code is performed. Statistical regression analysis is used to provide interpolation functions for the optimization process. The optimum design point is then checked with a final comprehensive code calculation. The technique is demonstrated with simple examples for the design of two injectors for an entrained coal gasifier and a burner for a pulverized coal combustor. The three designs demonstrate the method as well as showing significantly different optima for different configurations. The importance of specifying operating conditions independently for different injectors or burners is demonstrated.
Algebraic, Multi-Zoned Radiation Model for a Two-Zoned Zero-Dimensional Cylindrical Furnace
Hobbs, M.L. and Smith, P.J.
Fuel, 69, 1990. Funded by Utah Power & Light and Tennessee Valley Authority.
A simple two-zone zero-dimensional combustion model that estimates the influence of impurities in the fuel on the radiative energy transport has previously been developed based on an overall energy balance coupled with a multi-zoned radiation model. This paper presents the equations of the model, illustrates the method of calculating the radiative exchange areas for the two-zone system, and presents predictions for pulverized-coal and fluidized-bed combustion. The model predicts thermal performance as a function of coal input and furnace operational parameters, steam mass flow rates, and superheated steam temperatures leading to the high-pressure turbine. Two wall ash deposit parameters, thermal conductivity and maximum deposit thickness, have been determined by a sensitivity analysis to be critical to furnace performance. These parameters have been obtained experimentally by others. The predictions from the two-zone model have been compared with predictions from an earlier single-zone model. The general trends from both models were the same, although the two-zone model predictions were closer to expected values.
An Evaluation of Three-Dimensional Computational Combustion and Fluid Dynamics for Industrial Furnace Geometries
Gillis, P.A. and Smith, P J.
Twenty-third Symposium (International) on Combustion, The Combustion Institute, France, 1990. Funded by Consortium and ACERC.
A three-dimensional gaseous combustion and computational fluid dynamics model is presented for simulating reacting flow in industrial furnaces and utility boilers. Data were obtained for non-reacting flow in both a tangential-fired and a wall-fired furnace. These two cases were simulated with a variety of grid resolutions to establish grid-independent solution requirements. A differencing scheme was used which assured that the numerical solution was exact for linear basis functions on an arbitrarily spaced mesh. This accuracy was demonstrated by comparing numerical results with analytic solutions of the fully coupled equation set. Several variations of the SIMPLE algorithm were incorporated into the flow model to study the importance of velocity/pressure coupling. These variations included SIMPLE, SIMPLER, SIMPLEC, SIMPLEST, and combinations of these algorithms. The robustness and speed of the SIMPLE-based methods were evaluated for a corner-fired furnace and a wall-fired furnace. The importance of temporal fluctuations due to fluid turbulence on the non-linear mixing in gaseous combustion was quantified for corner-fired furnace geometries and compared with similar studies on laboratory scale furnaces.
1989
Furnace Design Using Comprehensive Combustion Models
Smith, P.J.; Sowa, W.A. and Hedman, P.O.
Accepted for publication in Combustion and Flame, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and Brigham Young University.
A new design methodology is presented which allows for the use of comprehensive coal combustion codes in design applications and provides a priori information on the cost of the optimization. A statistical response surface methodology is used to determine appropriate sample points from the design space at which the computation for the comprehensive code are performed. Statistical regression analysis is used to provide interpolating functions for the optimization process. The optimum design point is then checked with a final comprehensive code calculation. The technique is demonstrated with simple examples for design of two injectors for an entrained coal gasifier and of a burner for a pulverized coal combustor. The three designs demonstrate the method as well as showing significantly different optima for different configurations. The importance of specifying operating conditions independently for different injectors or burners is demonstrated.
Algebraic, Multi-Zoned Radiation Model for a Two-Zoned Zero-Dimensional Cylindrical Furnace
Hobbs, M.L. and Smith, P.J.
Accepted for publication in Fuel, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and Utah Power & Light.
A simple two-zone zero-dimensional combustion model which estimates the influence of impurities in the fuel on the radiative energy transport has been developed based on an overall energy balance coupled with a multi-zoned radiation model. This paper presents the equations of the model, illustrates the method of calculating the radiative exchange areas for the two-zone system, and presents predictions for pulverized-coal and fluidized-bed combustion. The model predicts thermal performance as a function of coal input and furnace operational parameters, steam mass flow rates, and superheated steam temperatures leading to the high-pressure turbine. Two wall ash deposit parameters, thermal conductivity, and maximum deposit thickness, have been determined by a sensitivity analysis to be critical to furnace performance. Others have obtained these parameters experimentally. The predictions from the two-zone model have been compared to predictions from an earlier single-zone model. The general trends from both models were the same, although the two-zone model predictions were closer to expected values.
The foundation to describe coal-specific conversion behavior will be AFR's Functional Group (FG) and Devolatilization, Vaporization, and Cross linking (DVC) models, developed under previous and on-going METC sponsored programs. These models have demonstrated the capability to describe the time dependent evolution of individual gas species, and the amount and characteristics of tar and char. The combined FG-DVC model will be integrated with Brigham Young University's comprehensive two-dimensional reactor model, PCGC-2, which is currently the most widely used reactor simulation for combustion or gasification. The program includes: i) validation of the submodels by comparison with laboratory data obtained in this program, ii) extensive validation of the modified comprehensive code by comparison of predicted results with data from bench-scale and process scale investigations of gasification, mild gasification and combustion of coal or coal-derived products in heat engines, and iii) development of well documented user friendly software applicable to a "workstation" environment.
Success in this program will be a major step in improving the predictive capabilities for coal conversion processes including: demonstrated accuracy and reliability and a generalized "first principles" treatment of coals based on readily obtained composition data.
Large-Scale Predictability of Pulverized Coal Combustion Byproducts
Smith, P.J. and Gillis, P.A.
1st International Congress on Toxic Combustion Byproduct: Formation and Control, Los Angeles, California, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and Brigham Young University.
Mathematical model simulations are presented for pilot plant and laboratory test furnaces to quantify the predictability and accuracy of combustion by-product formation and destruction. The roles of numerical accuracy and of closely coupled physico-chemical processes are explored. Capabilities of 3-D furnace models to describe global, site-specific flow and fine scale mesh resolution is shown to be an important consideration in simulating flow patterns in utility boiler geometries, even for relatively simple configurations. Fabricated exact numerical solutions for the coupled particle differential equation set are shown to be useful in identifying algorithmic and coding errors in these large mathematical models. The coupling between local heat transfer and other physico-chemical processes occurring in coal combustion applications is emphasized. Differences of 50-70% are shown to be obtained if particle-gas convective/conductive heat exchange and gas radiation are ignored or if turbulent fluctuations are not accounted for. This variability is shown to have a large impact than the chemical kinetic reaction rate of CO to CO2. The importance of advancing chemistry submodels and their coupling to the turbulent fluid mechanics for more accurate predictability of combustion by-products is emphasized.
Three-Dimensional Computational Fluid Dynamics Modeling in Industrial Furnace Geometries
Gillis, P.A. and Smith, P.J.
Western States Section, The Combustion Institute, Livermore, California, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, and Utah Power & Light.
In recent years, advances in computer technology have revolutionized the methods used to analyze complex flow phenomena. Many practical flows, which were previously expensive and difficult to model, are now gradually yielding their mysteries through the power of today's computers. This process is aided by the continuing development of efficient numerical algorithms and further research into the modeling of physical phenomena that are not well understood, such as turbulence. Several three-dimensional comprehensive combustion models have appeared in the literature over the past three years. Nearly all these models employ similar numerical techniques, variations of the TEACH method. A review of these codes has revealed that they use coarse grid structures and typically lack validation through comparison with experimental data. The purpose of this paper is to describe and demonstrate a three-dimensional CFD (Computational Fluid Dynamics) code that has been designed for industrial geometries. This code includes new numerical techniques, has been validated with data comparisons in several industrial furnaces, and will be used to establish the grid resolution required to obtain grid independent solutions.
The CFD model was designed to be a fundamental element in a larger comprehensive pulverized-coal combustion code for utility boilers. The model handles Cartesian and cylindrical coordinate systems and allows for irregular grid spacing in order to efficiently model the large scale disparities encountered in industrial furnaces. The model includes the option of employing different variations of the SIMPLE algorithm to couple the Navier-Stokes equations. Coupling options include the SIMPLE, SIMPLER, SIMPLEC, SIMPLEST, and various combinations of these algorithms. First-order finite differencing is performed with a combined central and revised upwind scheme that has been adapted to handle grid irregularities. The resulting finite-difference coefficient matrices are solved with a vectorized relaxed Thomas algorithm. Several turbulence models have been incorporated into the code, include the k-e turbulence model. Significant portions of the model have been vectorized to improve code performance.
The emphasis of this paper is on evaluating model performance. Experimental data has been obtained from Consolidation Coal Company for a wall-fired pilot-scale furnace with four swirled burners. Combustion Engineering has also provided a large collection of velocity data from a tangentially fired furnace. Comparisons will be made between predicted and experiment velocities for both configurations. These comparisons will be performed in different locations within each furnace and model performance will be evaluated. Additional numerical experiments were performed to determine the grid resolution needed to achieve grid independent solutions. Recommendations will be made detailing the grid resolution needed to accurately model the major types of industrial furnaces. The different coupling algorithm options employed in the model will also be evaluated for both robustness and efficiency. Additional comparisons between the TEACH method and the model's numerical techniques will be made.
Detailed Model for Practical Pulverized Coal Furnaces and Gasifiers
Smith, P.J. and Smoot, L.D.
AR&TD Contractor's Meeting, Morgantown, West Virginia, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, and Utah Power & Light.
This project was jointly funded by a consortium of eight different industrial firms and governmental agencies of which PETC is a part. The objective of the project was to improve and extend a generalized two-dimensional, pulverized coal combustion and gasification code for application to large-scale practical configurations. The work initiated in this four-year project is being continued with NSF and private sponsorship under the Advanced Combustion Engineering Research Center. Over the past year the consortium project tasks were significantly scaled down from the preceding three years and focused on the evaluation of model and submodel development, integration of the pertinent algorithms into the 3-D model, and evaluation of the integrated code. This paper will report on work accomplished on this development over the last year and summarize the state of development of the 3-D modeling effort.
Three tasks were originally outlined for this four-year project. Task 1 - extend an available 2-D model to three-dimensional, large-scale furnace and gasifier configurations. Task 2 - evaluate the 3-D model a: through a statistical sensitivity analysis to determine parameters most influential on predictions, and b) through review, selection, documentation and comparison of a set of available data relating to practical, large-scale furnaces and gasifiers with model predictions. Task 3 - improve or develop submodel equations and/or data for the 3-D model.
All submodel work on task 3 was completed in the first three years of the contract and included the development of a new turbulent particle-dispersion model, a new two and three dimensional discrete ordinates model for radiative transfer, the generalization of the devolatilization submodel to include all known devolatilization models, advancements in the char reaction submodel, a new submodel for interface conditions in multi-phase reaction systems (i.e. coal or oil combustion), the incorporation of chemical kinetically limited gaseous reactions of CO - CO2, and the development of a crude but complete framework for incorporating mineral matter transformations and deposition in the combustion chamber. All of this work on this contract has been previously reported.
During this year work has been accomplished towards the incorporation of advanced numerical algorithms and submodels into the first-generation 3-D comprehensive coal combustion code. The code has been assembled for arbitrary geometries in Cartesian and cylindrical coordinate systems. It includes turbulent mixing with eddy diffusivity, mixing-length and k-e turbulence models. It incorporates dispersed particulate phases in an Eulerian description. The gas phase chemistry couples turbulent mixing-limited equilibrium reactions for all major species. Over the last year an ongoing evaluation has occurred to quantify three levels of error: numerical and algorithmic errors, submodel errors, and overall model or data comparison errors. Each of these areas of evaluation is presented in the paper.
A computer graphics package has been developed for the display of three-dimensional combustion data. The package is based on the FIGS graphics industry standard and thus will run on many different hardware platforms. The package allows menu driven access to all computed variables from the 3-D code and takes advantage of color capabilities to cue the user and display the results. A computer graphics pre-processor has been developed to facilitate the creation of the large set of input required for an industrial scale furnace. The complex three-dimensional mesh can be easily defined with menu driven color graphics input.
Calculations have been performed for pilot scale combustion furnaces including wall-fired and corner-fired units. Comparisons have been made with data for non-reacting fluid dynamics, gaseous combustion and coal combustion. Analysis of model performance has been completed for a variety of model options. Analyses of numerical accuracy and model robustness have shown that the formulation and implementation of this code to be more accurate, more computationally efficient and more robust than previous models from this and other laboratories.
An Analysis of Coal Particle Temperature Measurements with Two-Color Optical Pyrometers
LaFollette, R.M.; Hedman, P.O. and Smith, P.J.
Combust. Sci. and Tech., 66, 93-105, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates).
Two-color optical pyrometers have been used to measure the temperature of reacting pulverized-coal particles. An analysis of such measurements was performed to determine the effect of several possible conditions on the measured temperature. The conditions investigated were the use of a single photo-multiplier to alternately measure the radiant emission at the two selected wavelengths, the presence of soot, light extinction, the choice of wavelengths used to compute the two-color temperature, and non-uniform particle clouds. A computer model of a one-dimensional coal particle cloud was written for this analysis. Results of calculations showed that artificially high temperatures can result if a pyrometer with a single photo detector is used to measure temperatures in a rapidly fluctuating flame. Emission by soot in the coal particle cloud caused unrealistically high temperature measurements. Light absorption by soot lowered the two-color temperature, but not enough to compensate for the rise in observed temperature caused by soot emission. When the wavelengths used are in the visible spectrum, the hotter particles are weighted much more heavily than when the wavelengths are in the infrared region. The use of Wein's Law, a valid approximation to Planck's Law in the visible spectrum, causes substantial error for longer wavelengths. Finally, the two-color temperature of a non-isothermal cloud was weighted most heavily by the hotter particles, depending upon the wavelengths used for the measurement. From this analysis important questions have been raised as to the validity of such measurements under transient conditions and during devolatilization.
1988-1986
A Study of Two Chemical Reaction Models in Turbulent Combustion
Smith, P.J. and Fletcher, T.H.
Accepted for publication in Combustion Science and Technology, 1988. 24 pgs. Not externally funded.
Research efforts with comprehensive computer models that have tried to predict the performance of coal combustors have either neglected the effect of the turbulence on the mean chemical properties or have used one of two approximate methods. This paper focuses on the impact of the turbulence on the chemical reactions of the volatile products of coal combustion processes. It is shown that by ignoring the effect of the turbulence on mean combustion properties significant differences occur as compared to experimental data and predicted by both of the more rigorous models. The first method, the volatile reactances model, is an extension of an approach for premixed gaseous combustion presented by Magnussen and Hjertager. The second method, the statistical, coal-gas mixture fraction model, is an extension of gaseous diffusion flame approaches. These two methods are examined, analyzed and evaluated by comparing predictions from each method with experimental data from three laboratory furnaces. It is shown that while the first method takes only half as much computational time, the second method is required if species and temperatures in zones containing other than mixtures of pure fuel, pure oxidant and pure stoichiometric product are needed. The distribution of eddy mixtures as formulated in the second method is shown to be more consistent with existing limited experimental data.
Coal Combustion Modeling: Particulate and Gas Phase Heat Transfer Effects
Smith, P.J.; Baxter, L.L. and Jamaluddin, A.S.
AIChE Conference, 1988, New Orleans, LA. 15 pgs. All internal funding.
Heterogeneous heat transfer aspects strongly influence the performance of practical coal combustion systems since many of the subprocesses within the flame are highly temperature sensitive, and since the purpose of most furnaces is to extract energy from the flame. Coal combustion simulation or computer modeling permits investigation of the effect of various heat transfer mechanisms within flames on the many other simultaneous processes of turbulent fluid mechanics, coal conversion, gaseous reaction, etc. This paper examines two aspects of heat transfer on pulverized coal combustion processes: 1) the particle dominated radiation process and 2) the gas phase dominated turbulent convection processes. Comprehensive furnace modeling is used to study practical furnaces and the mechanisms are elucidated by comparing predicted results with experimental data from several coal and gas fired furnaces.
Predicting Particulate Deposition from Turbulent Streams
Jamaluddin, A.S. and Smith, P.J.
Engineering Foundation Conference on Mineral Matter and Ash Deposition from Coal, 1988, Santa Barbara, CA. 10 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
A theoretical model has been developed to predict the rate of deposition of particulates from flowing turbulent gas-solid suspensions. The predictions of the model compare favorably with available experimental data. Incorporation of this technique into a comprehensive combustion code demonstrates promise for modeling slagging propensity in coal-fired furnaces.
Discrete-Ordinates Solution of Radiative Transfer Equation in Non-Axisymmetric Cylindrical Enclosures
Jamaluddin, A.S. and Smith, P.J.
Proceedings of the 1988 National Heat Transfer Conference, 1, 227-232, 1988. 6 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
The discrete-ordinates approximation is used to solve the radiative transfer equation in non-axisymmetric cylindrical enclosures containing absorbing-emitting and scattering media, with and without the temperature profile known a priori. Since neither detailed experimental data nor predictions from a zone or Monte-Carlo model for three-dimensional cylindrical enclosures is available, cylindrical equivalents of three-dimensional rectangular enclosures, for which zone model predictions of radiative transfer are available, are used in model evaluation. Limited evaluation of the model shows that the discrete-ordinates method provides acceptable predictions of radiative transfer in non-axisymmetric cylindrical enclosures.
Turbulent Dispersion of Particles
Baxter, L.L. and Smith, P.J.
Western States Section, 1988, The Combustion Institute, Salt Lake City, UT. 21 pgs. Funded by ACERC consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
An approach to describing the transport of particles in turbulent flows is presented which is derivable from first principles. The approach is independent of the particular turbulence model used. It is valid in applications ranging from entrained flow dispersion to fluidized bed applications. It involves no adjustable parameters and requires a description of the turbulence in terms of the mean square velocities and Lagrangian autocorrelation functions.
The model has been rigorously evaluated by comparison to exact solutions, accurate alternative approaches, and experimental data. The model is able to reproduce the exact solutions both analytically and numerically, predict results which are both more accurate and precise than those of the leading alternative models, and predict experimental data within its inherent error under a range of different conditions.
The incorporation of the model into a comprehensive combustor simulator is also demonstrated. The model is found to be more efficient and robust, both in terms of the particle dispersion portion of the comprehensive code and in terms of the overall code performance. Predictions using this description of turbulent particle dispersion yield results that agree in detail with previous predictions using a calibrated particle dispersion model.
Measurement and Prediction of Entrained-Flow Gasification Processes
Brown, B.W.; Smoot, L.D.; Smith, P.J. and Hedman, P.O.
AIChE Journal, 34, 3, 435-446, 1988. 12 pgs. Funding source by Morgantown Energy Technology Center.
A detailed mathematical model is used to predict local and effluent properties within an axisymmetric, entrained-flow gasifier. Laboratory experiments were conducted to provide local properties for four coal types from a gasifier operating at near-atmospheric pressure. Effects of selected model parameters and test variables were examined and compared with measurements in most cases. The comparison of predictions and measurements provides the first evaluation of capabilities and limitations of a comprehensive model for entrained-flow gasifiers.
Three-Dimensional Fluid Dynamics Modeling in Furnace Geometries
Gillis, P.A. and Smith, P.J.
Western States Section, 1988, The Combustion Institute, Salt Lake City, UT. 12 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
A three-dimensional non-reacting flow model has been developed for predicting flow inside industrial furnace configurations. The code uses the SIMPLE algorithm to couple the Navier-Stokes equations and solves the resulting matrices with a vectorized Thomas algorithm. Model predictions have been compared with experimental data in cross-fired furnace geometry. The k-e turbulence model produced significantly superior data agreement than the simpler turbulence models. The effect of inlet condition variation and grid resolution were demonstrated. It was also shown that fine grid spacing is needed to resolve localized large-scale vortex structure.
Revised User's Manual: 87-PCGC-2
Smoot, L.D.; Smith, P.J.; Brewster, B.S. and Baxter, L.L.
Advanced Combustion Engineering Research Center, 1988. Funded by US Department of Energy, Electric Power Research Institute, Consortium, and ACERC.
A two-dimensional, steady state model for describing a variety of reactive and non-reactive flows, including pulverized coal combustion and gasification, is presented. Recent code revisions and additions are described. The model, referred to as 87-PCGC-2, is applicable to cylindrical axi-symmetric systems. Turbulence is accounted for in both the fluid mechanics equations and the combustion scheme. Radiation from gases, wall and particles is taken into account using either a flux method or discrete ordinates methods. The particle phase is modeled in a Lagrangian framework, such that mean paths of particle groups are followed. Several multi-step coal devolatilization schemes are included along with a heterogeneous reaction scheme that allows for both diffusion and chemical reaction. Major gas-phase reactions are modeled assuming local instantaneous equilibrium, and thus the reaction rates are limited by the turbulent rate of mixing. A NOx finite rate chemistry submodel is included which integrates chemical kinetics and the statistics of the turbulence. The gas phase is described by elliptic partial differential equations that are solved by an iterative line-by-line technique. Under-relaxation is used to achieve numerical stability. The generalized nature of the model allows for calculation of isothermal fluid mechanics/gaseous combustion, droplet combustion, particulate combustion and various mixtures of the above, including combustion of coal-water and coal-oil slurries. Both combustion and gasification environments are permissible. User information and theory are presented, along with same problems.
LDV Measurements in Simulated Entrained Gasifier Flows
Lindsay, J.D.; Hedman, P.O. and Smith, P.J.
Submitted to AlChE Journal, 1988. 15 pgs. Funded by Morgantown Energy Technology Center.
Entrained-flow gasification of pulverized-coal has the potential to become a competitive source of energy. One near-commercial application of entrained-flow coal gasification that has been receiving considerable attention is the use of an entrained-flow gasifier in an Integrated Gasification Combined Cycle (IGCC) (e.g., 1, 2). In order to better understand and to improve pulverized-coal gasification processes, a large body of gasification data from within a laboratory-scale entrained-flow gasifier has been collected at this laboratory (e.g., 3-7) and applied toward the development of a comprehensive computer model for pulverized-coal reactors (8-10). This paper summarizes a companion study of the flow processes in an isothermal flow facility that simulates the flow characteristics of the entrained-flow coal gasifier.
A laser Doppler-velocimeter (LDV) was used to make measurements of mean and turbulent velocities both at the inlet, and from within the flow chamber. Isothermal air flows were used to isolate the basic flow properties from such complications as density gradients and chemistry-turbulence interactions. This study emphasized the effects of inlet conditions on flow properties within the simulated reactor (e.g., axial velocity decay, location of recirculation zones, turbulence levels). A knowledge of the effect of inlet conditions on flow properties can lead to improved gasifier operating conditions, can assist in the interpretation of in situ chemical species data from the gasifier, and can guide modeling efforts.
Comparison of experimental measurements with predictions made by a specific computer model was a second objective of this study. The model, PCGC-2 (Pulverized-Coal Gasification and Combustion: 2-Dimensional), is a comprehensive code for pulverized-coal and coal-water slurry processes that has been developed at this laboratory (8-10). The code employs the k-e model for turbulent fluid mechanics. Much of the earlier experimental flow data was collected with intrusive probes, which in some cases seriously distorted the flow being measured. Furthermore, most of the earlier studies did no include measurements at the inlet. Documentation of the inlet boundary condition is needed if experimental data are to be properly applied to model development.
Predicting Radiative Transfer in Rectangular Enclosures Using the Discrete Ordinates Method
Jamaluddin, A.S. and Smith, P.J.
Western States Section, 1987, The Combustion Institute, Provo, UT. 19 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Discrete ordinates solutions of the radiative transfer equation in two and three-dimensional rectangular enclosures containing absorbing-emitting-scattering media have been obtained using S2, S4, S6 and S8 approximations. Evaluation against exact analytical and numerical solutions show that while all of these approximations provide acceptable predictions of the radiation fluxes in two-dimensional enclosures, use of the higher order (higher than S4) approximations is not justified due to substantial increase in computational time and negligible improvement in the accuracy of the predictions. However, for three-dimensional enclosures, the S2 approximation is grossly in error. S4, S6 and S8 approximations predict wall heat fluxes and the temperatures of the medium accurately in these enclosures, but, once again, S4 approximation is shown to be adequate. A study of the sensitivity of the predicted net heat absorption by the walls to the dimensions of the system, and radiative properties of the medium and the surrounding walls, based on Fourier analysis technique, indicates that the predictions are more sensitive to the radiative properties than to the dimensions of the enclosure.
Parametric Sensitivity Study of an Entrained-Flow Pulverized-Fuel Combustion Model
Smith, J.D.; Smith, P.J. and Hill, S.C.
Submitted for publication to Computer and Chem.Engineering, 1987. 35 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
An extensive parametric sensitivity study of a two-dimensional pulverized-fuel (pf) combustion model has been performed to investigate the effect of parametric uncertainty on model predictions. Results illustrate the dominant effect that error in coal devolatilization/oxidation parameters has on predicted burnout, Nox formation, local gas temperature, and coal-gas mixture fraction. Uncertainty in the turbulent particle dispersion parameters appear to have a secondary effect, while error in the particle-gas radiation parameters seems to have little impact on model predictions. Regions of the computational domain exhibiting sensitivity to specific parameters are identified. Specific parameter sensitivity implies the relative importance of various mechanisms in the overall process. Turbulent particle dispersion seems to be important early in the reactor with kinetic processes dominating at and following the predicted ignition point. Radiation appears to be of minor importance for the case under investigation. These results provide unique insight into the general pf combustion process. Specifically, they emphasize the need for a better methodology of predicting the overall volatiles yield. This implies additional understanding of the devolatilization/oxidation mechanism and its role in the overall pf combustion process. Although the results presented here are case specific, they provide unique insight into the overall pf combustion process and illustrate the effects that parameter uncertainty has on model predictions. This information provides fundamental direction for future comprehensive model development and focuses attention on pertinent experimental work to better quantify critical input parameters.
Comprehensive Modeling of Combustion Systems
Smoot, L.D. and Smith, P.J.
ASME/JSME Thermal Engineering Joint Conference, 1987, Honolulu, Hawaii. 13 pgs. Funded by ACERC (National Science Foundation and Associates and Affiliates).
This paper provides a brief review of comprehensive modeling of combustion systems. The focus is on continuous, subsonic flow systems. Supersonic flows and intermittent or unsteady flows with combustion were not considered. Where condensed phases are present, the focus is further placed on entrained systems where particulate and droplet collisions or interactions through motion can be neglected. Gaseous, liquid, solid and slurry fuels were considered. A brief review of related literature is included. Model foundations, elements (submodels), requirements, and potential uses are summarized. Recent technical work of the authors or colleagues in turbulent fluid mechanics of swirling flows, radiation, and gas-phase reactions is summarized. Example predictions for selected cases are shown from several investigators, mostly with comparisons to measured properties. Future directions, research needs and plans are also identified.
The Effects of Interfacial Conditions on Mass and Heat Transfer
Baxter, L.L. and Smith, P.J.
Western States Section, 1987, The Combustion Institute, Honolulu, HI. 3 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Classical assumptions about equilibrium conditions at phase interfaces during heat and/or mass transfer are shown to be approximations that can lead to errors in combustion environments. Specific illustrations involving droplet vaporization are given which show the nature and magnitude of the errors.
A method of relaxing these assumptions is discussed. Equations are developed from fundamental principles of physical chemistry to describe the actual conditions at interfaces. The corrections to the equilibrium assumptions depend on the rates of mass/heat transfer and physical constants that describe specific molecular behavior.
Computations illustrating the use of the nonequilibrium assumption reveal both the potentially large error associated with assuming equilibrium at the interfaces and the sensitivity of the predictions to the molecular parameters.
Statistical Distributions of Coal Particle Sizes in Pulverized-Coal Combustion and Gasification
McDonald, J.B.; Richards, D.O.; Smith, P.J. and Sowa, W.A.
Annual Meeting of the American Statistical Assoc., 1986. 18 pgs. Not externally funded.
Coal particle size distributions for a Utah bituminous coal and a Wyoming subbituminous coal were modeled using four different probability density functions (GB1, GB2, lognormal, and lognormal by method of moments) and two engineering approximations. The pdf models of the particle size distributions were discretized and compared to the engineering approximations by simulating a coal combustion case and a coal gasification case using the comprehensive combustion model, PCGC-2. The gasification case used the Utah coal and the combustion case used the Wyoming coal. Significant differences were noticed between predictions using engineering approximations of the particle size distribution and predictions using discretized pdf approximations of the particle size distribution. The different methods of describing the particle size distribution affected most notably the simulators prediction of NOx.
Prediction of Radiative Heat Transfer in Cylindrical Furnaces
Jamaluddin, A.S. and Smith, P.J.
Western States Section, 1986, The Combustion Institute, Banff, Canada. 20 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Two radiative heat transfer codes, one based on the six-flux radiation model, and the other based on the S4 discrete ordinates model, have been developed. Both the radiation models approximate the angular variation of the radiation intensity by solving the radiation transport equation in a predetermined number of directions so that the angular integral is removed, resulting in differential equations which can be solved simultaneously with the differential equations of fluid flow, chemical reactions and heat transfer. The limited evaluation of these two models indicates that the six-flux model under-predicts the radiant heat fluxes by up to a factor of two, while the discrete ordinates model affords sufficiently accurate predictions for engineering heat transfer calculations.
Effects of Swirling Flow on Nitrogen Oxide Concentration in Pulverized Coal Combustors
Smith, P.J.; Smoot, L.D. and Hill, S.C.
AIChE Journal, 32, (11), 1917-1919, 1986. 3 pgs. Funded by US Department of Energy.
Nitrogen-containing pollutants from pulverized coal conversion processes have been of concern for several years, and reviews of this subject have been published (Wendt, 1980; Chen et al., 1981). Swirl of the secondary stream is one technique of fuel-air contacting that causes significant changes in NO concentrations. In some cases, increased swirl increases NO emissions (Heap et al., 1973b; Brown et al., 1977; Pershing and Wendt, 1979). In other cases, increased swirl initially decreases NO emissions (Harding et al., 1982; Asay et al., 1983). Thus, the observed effects of swirl are varied, and indicate that additional parameters influence the NO emissions. In this note a model, described in detail elsewhere (Smith and Smoot, 1981; Smith et al., 1981, 1982; Hill et al., 1984), was used for interpreting effects of inlet stream swirl and velocity on NO emissions during combustion of pulverized coal.
Modeling of Swirl in Turbulent Flow Systems
Sloan, D.G.; Smith, P.J. and Smoot, L.D.
Progress Energy Combustion Science, 12, (3), 63?250, 1986. 188 pgs. Funded by US Department of Energy.
The standard k-e equations and other turbulence models are evaluated with respect to their applicability in swirling, recirculating flows. The turbulence models are formulated on the basis of two separate viewpoints. The first perspective assumes that an isotropic eddy viscosity and the modified Boussinesq hypothesis adequately describe the stress distributions, and that the source of predictive error is a consequence of the modeled terms in the k-e equations. Both stabilizing and destabilizing Richardson number corrections are incorporated to investigate this line of reasoning. A second viewpoint proposes that the eddy viscosity approach is inherently inadequate and that a redistribution of the stress magnitudes is necessary. Investigation of higher-order closure is pursued on the level of an algebraic stress closure. Various turbulence model predictions are compared with experimental data from a variety of isothermal, confined studies. Supportive swirl comparisons are also performed for a laminar flow case, as well as reacting flow cases. Parallel predictions or contributions from other sources are also consulted where appropriate. Predictive accuracy was found to be a partial function of inlet boundary conditions and numerical diffusion. Despite prediction sensitivity to inlet conditions and numerics, the data comparisons delineate the relative advantages and disadvantages of the various modifications. Possible research avenues in the area of computational modeling of strongly swirling, recirculating flows are reviewed and discussed.
A Study of Turbulent Particle Dispersion in Pulverized Coal Combustion and Gasification
Smith, P.J. and Baxter, L.L.
Western States Section, 1986, The Combustion Institute, Banff, Canada. 20 pgs. Funding source US Department of Energy, Brigham Young University and Electric Power Research Institute.
The location of the coal particles in pulverized coal combustion and gasification processes is a dominant factor in determining flame stability, combustion/gasification efficiency and pollutant concentrations. Experimental and predicted data show that particles do not mix at the same rate as the gas phase and that this difference in turbulent mixing is important to the coal reaction process. The environment around a given particle during heat up, devolatilization and heterogeneous reaction controls the overall combustion process. In flows of interest to pulverized coal combustion, the particle dispersion is dominated by the random fluctuations of the gas phase turbulence as opposed to the mean drag on the particle. A mixed Eulerian-Lagrangian mathematical model for predicting mean trajectories of expected values for ensembles of representative particles is presented, including the turbulent dispersion effects. The mean turbulent velocity is modeled as a Fickian diffusion process from the man number density as calculated from the Eulerian part of the formation. This model is discussed and compared to alternate procedures. Applications of the model are shown for particle dispersion in non-reacting flow and pulverized coal combustion. An evaluation by comparison with experimental data is presented. Conclusions regarding the significance of turbulent particle dispersion are discussed.
Prediction of High Intensity Pulverized Coal Combustion
Suzuki, T.; Smoot, L.D.; Fletcher, T.H. and Smith, P.J.
Combustion Science and Technology, 45, (3&4), 167-183, 1986. 17 pgs. Funded by Brigham Young University and Kobe Steel Company, Japan.
The overall characteristics of high-intensity pulverized coal combustion have been predicted by a one-dimensional model. The mixing of the primary stream of pulverized coal and transport air with secondary combustion air was estimated by a growth angle of the primary jet. The coal particle burnout was strongly affected by the extent of devolatilization, which varies amount coals. The extent of devolatilization as characterized by variation in a devolatilization coefficient was correlated with either proximate volatiles percentage or H/C mass ratio of the virgin coal. The resulting comparisons of predictions with measurements for eight coal types and tree different combustors show that observed trends are generally predicted. The data used for these comparisons were obtained from a wide range of high-intensity combustion experiments. The proximate volatile matter in the virgin test coals ranged form 16 to 40 percent while the coal feed rate was varied from 12 to 290 kg/hr. Combustion air temperature varied from 297 to 1483 K while residence time ranged from 3 to 140 ms. Comparative results suggest that the predictive method can be useful in interpreting high intensity combustion test results.
Optimization of Coal Devolatilization Model Parameters for Comprehensive Gasification and Combustion Models
Baxter, L.L.; Smith, P.J. and Smoot, L.D.
Western States Section, 1986, The Combustion Institute, Banff, Canada. 17 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Parameters for empirical devolatilization models are recommended for lignite and bituminous coals. These parameters were generated by combining optimization algorithms with rigorous particle and flow field modeling of experimental facilities.
PCGC-2, a comprehensive, axisymmetric, gasification and combustion code developed at the Brigham Young University Combustion Laboratory, was used to predict the velocity, temperature, and radiation fields of experimental apparatus used to collect devolatilization data. The effect of changing devolatilization models on these fields was neglected due to very low particle loadings. These fields were subsequently entered into a rigorous particle reaction model and the particle response to them was predicted.
A particle model predicted the trajectories, temperature, and reaction rates of coal particles in the flow fields described above. The particle model and its results are discussed. The particle model was coupled with an optimization algorithm to determine optimum parameters for the two-reaction devolatilization model.
Generalized reduced gradient (GRG) and sequential quadratic programming (SQP) methods were used in performing the optimization with OPTDES, an optimization program developed at Brigham Young University. The model parameters were optimized over six coal heat-up rates for each coal type. Statistical analyses of the residual sum of squares for the optimized two-reaction model revealed that there was no significant lack of fit of the data compared to the inherent variability of the data.
Furnace Design and Comprehensive Coal Combustion Models
Smith, P.J.; Sowa, W.A. and Hedman, P.O.
ASME Annual Meeting, 1986. 14 pgs. Funded by Morgantown Energy Technology Center and Brigham Young University.
Comprehensive coal combustion (including gasification) models have received little to no use in furnace design applications due, in part, to their large computational burden when used in conjunction with some design optimization strategy. This paper presents a new design methodology that allows for the use of comprehensive coal combustion codes in design applications and provides a priori information on the cost of the optimization. A statistical response surface methodology is used to determine appropriate sample points from the design space at which the computations for the comprehensive code are performed. Statistical regression analysis is used to provide interpolating functions for the optimization package. The final design point is then checked with a final comprehensive code calculation.
The technique is demonstrated with simple examples for design of two injectors for an entrained coal gasifier and of a burner for a pulverized coal combustor. The three designs show the utility of the method as well as showing significantly different optima for different configurations. The importance of specifying operating conditions independently for different injectors or burners is demonstrated. The utility of comprehensive coal combustion codes as another design tool is demonstrated.
Statistical Optimization of Feed Streams for an Entrained-Flow Coal Gasifier Using the PCGC-2 Simulator
Sowa, W.A.; Free, J.C.; Smith, P.J. and Hurst, T.N.
ASME Annual Meeting, 1986. 5 pages. Funded by US Department of Energy.
Concepts from statistical response surface methodology (RSM) and nonlinear optimization theory have been combined in a method for efficiently searching a "design space" when using large-scale analysis. The method is applied to a theoretical study of entrained-flow coal gasification, using PCGC-2, a two-dimensional, axisymmetric model developed at Brigham Young University, that predicts local properties in a reaction chamber. RSM was used to sample the design space and to construct a regression model, which was then linked to OPTDES, BYU, a program containing several nonlinear programming algorithms.
Two test plans were used to verify the performance of the RSM optimization algorithm in solving a gasification example problem. The method is shown to provide accurate results, while de-coupling the search and analysis phases of optimization. The importance of a proper test plan for conducting an un-biased exploration of design space is also demonstrated by comparing the results of the two test plans.
LDA Measurements in Simulated Gasifier Flows with Swirl
Lindsay, J.D.; Hedman, P.O. and Smith, P.J.
International Symposium on Laser-Doppler Velocimetry, 1986, Lisbon, Portugal. 6 pgs. Funded by Morgantown Energy Technology Center and US Department of Energy.
A laser-Doppler system was used in a cold-flow study of a simulated pulverized-coal gasifier. The study was designed to provide fundamental information about the behavior of flow in such a gasifier and to provide data for the validation of a computer model. Measurements in 21 flow cases with and without swirl were made. Results are compared with predictions from a comprehensive model that uses a k-e turbulence submodel. Several levels of replication were used in the testing in order to examine reproducibility and to permit statistical analysis of results.
Smith, PJ
1993
Distributed Combustion Simulations
Sikorski, K.; Ma, K.-L.; Smith, P.J. and Adams, B.R.
Energy & Fuels, 7 (6):902-905, 1993. Funded by ACERC.
This paper reports research in progress. Two types of domain decomposition have been used in distributed computing with networked workstations for the numerical modeling of full-scale utility boilers. The numerical model is a three-dimensional combustion code that couples turbulent computational fluid dynamics with the chemical reaction process and the radiative heat transfer. Two approaches, here called microscale parallelism and macroscale parallelism, are proposed to study the intrinsic parallelism of typical combustion simulations. We describe the implementation of the microscale parallelism as well as its performance on networked workstations.
Parametric Sensitivity Study of a CFD-Based Coal Combustion Model
Smith, J.D.; Smith, P.J. and Hill, S.C.
AIChE Journal, 39, (10):1668-1679, 1993. Funded by Combustion Laboratory Consortium through Brigham Young University.
Parametric sensitivity of a two-dimensional pulverized-fuel (PF) combustion model is studied extensively for the effect of parametric uncertainty on model predictions. Results show that error in coal devolatilization/oxidation parameters has the dominant effect on predicted burnout, NOx formation, local gas temperature, and coal-gas mixture fraction. Uncertainty in the turbulent particle dispersion parameters appears to have a secondary effect, while error in the particle-gas radiation parameters has little impact on model predictions. Regions of the computational domain exhibiting sensitivity to specific parameters are identified. Specific parameter sensitivity implies the relative importance of various mechanisms in the overall process. Turbulent particle dispersion seems to be important early in the reactor with kinetic processes dominating at and following the predicted ignition point. Radiation appears to be of minor importance. These results indicate the need for a better method of predicting the overall volatiles yield and further understanding of the devolatilization/oxidation mechanism and its role in the overall PF combustion process. The study provides fundamental direction for future comprehensive model development and focuses on experimental work to better quantify critical input parameters.
Cloud Tracing in Convection-Diffustion Systems
Ma, K.-L. and Smith, P.J.
Proceedings of the Visualization 93 IEE/ACM SIGGRPHY Conference:253-259, San Jose, CA, October 1993. Funded by IBM and ACERC.
This paper describes a highly interactive method for computer visualization of simultaneous three-dimensional vector and scalar flow fields in convection-diffusion systems. This method allows a computational fluid dynamics user to visualize the basic physical process of dispersion and mixing rather than just the vector and scalar values computed by the simulation. It is based on transforming the vector field from a traditionally Eulerian reference frame into a Lagrangian reference frame. Fluid elements are traced through the vector field for the mean path as well as the statistical dispersion of the fluid elements about the mean position by using added scalar information about the root mean square value of the vector field and its Lagrangian time scale. In this way, clouds of fluid elements are traced not just mean paths. We have used this method to visualize the simulation of an industrial incinerator to help identify mechanisms for poor mixing.
1992
Virtual Smoke: An Interactive 3D Flow Visualization Technique
Ma, K.-L. and Smith, P.J.
Visualization Conference, Boston, MA, October 1992. Funded by International Business Machines and ACERC.
The paper introduces a new technique for computer visualization of simultaneous three-dimensional vector and scalar fields such as velocity and temperature in reaction fluid flow fields. The technique, which we call Virtual Smoke, simulates the use of colored smoke for experimental gaseous fluid flow visualization. However, it is noninvasive and can animate, in particular, the dynamic behaviors of steady state or instantaneous flow fields obtained from numerical simulations. Virtual Smoke is based on Volume Seeds and Volume Seedlings, which are direct volume visualization methods previously developed for highly interactive scalar volume data exploration. We use data from combustion simulations to demonstrate the effectiveness of Virtual Smoke.
1991
Detailed Model for Practical Pulverized Coal Furnaces and Gasifiers, Volume II, User Manual for 1990 Version Pulverized Coal Gasification and Combustion 3-Dimensional (90-PCGC-3), Final Report
Smith, P.J.; Smoot, L.D.; Hill, S.C. and Eaton, A.M.
Advanced Combustion Engineering Research Center, 1991. Funded by US Department of Energy, Consortium and ACERC.
The theoretical foundations, numerical approach and a guide for users of 90-PCGC-3 are presented. 90-PCGC-3 is a generalized, three-dimensional, steady state model that can be used to predict the behavior of a variety of reactive and non-reactive (isothermal) fluid flows. The model solves the Navier-Stokes equations in three-dimensional, Cartesian coordinates. Turbulence is accounted for in both the fluid mechanics equations and the combustion solution scheme. Major gas-phase reactions are modeled assuming local instantaneous equilibrium, and thus the reaction rates are limited by the turbulent rate of mixing. The gas phase is described by elliptic partial differential equations that are solved by an iterative line-by-line technique. Under-relaxation is used to achieve numerical stability.
1990
Furnace Design Using Comprehensive Combustion Models
Smith, P.J.; Sowa, W.A. and Hedman, P.O.
Combustion and Flame, 79 (2), 111-121, 1990. Funded by Brigham Young Unviersity.
A new design methodology is presented that allows for the use of comprehensive coal combustion codes in design applications and provides a priori information on the cost of the optimization. A statistical response surface methodology is used to determine appropriate sample points from the design space at which the computation for the comprehensive code is performed. Statistical regression analysis is used to provide interpolation functions for the optimization process. The optimum design point is then checked with a final comprehensive code calculation. The technique is demonstrated with simple examples for the design of two injectors for an entrained coal gasifier and a burner for a pulverized coal combustor. The three designs demonstrate the method as well as showing significantly different optima for different configurations. The importance of specifying operating conditions independently for different injectors or burners is demonstrated.
Algebraic, Multi-Zoned Radiation Model for a Two-Zoned Zero-Dimensional Cylindrical Furnace
Hobbs, M.L. and Smith, P.J.
Fuel, 69, 1990. Funded by Utah Power & Light and Tennessee Valley Authority.
A simple two-zone zero-dimensional combustion model that estimates the influence of impurities in the fuel on the radiative energy transport has previously been developed based on an overall energy balance coupled with a multi-zoned radiation model. This paper presents the equations of the model, illustrates the method of calculating the radiative exchange areas for the two-zone system, and presents predictions for pulverized-coal and fluidized-bed combustion. The model predicts thermal performance as a function of coal input and furnace operational parameters, steam mass flow rates, and superheated steam temperatures leading to the high-pressure turbine. Two wall ash deposit parameters, thermal conductivity and maximum deposit thickness, have been determined by a sensitivity analysis to be critical to furnace performance. These parameters have been obtained experimentally by others. The predictions from the two-zone model have been compared with predictions from an earlier single-zone model. The general trends from both models were the same, although the two-zone model predictions were closer to expected values.
An Evaluation of Three-Dimensional Computational Combustion and Fluid Dynamics for Industrial Furnace Geometries
Gillis, P.A. and Smith, P J.
Twenty-third Symposium (International) on Combustion, The Combustion Institute, France, 1990. Funded by Consortium and ACERC.
A three-dimensional gaseous combustion and computational fluid dynamics model is presented for simulating reacting flow in industrial furnaces and utility boilers. Data were obtained for non-reacting flow in both a tangential-fired and a wall-fired furnace. These two cases were simulated with a variety of grid resolutions to establish grid-independent solution requirements. A differencing scheme was used which assured that the numerical solution was exact for linear basis functions on an arbitrarily spaced mesh. This accuracy was demonstrated by comparing numerical results with analytic solutions of the fully coupled equation set. Several variations of the SIMPLE algorithm were incorporated into the flow model to study the importance of velocity/pressure coupling. These variations included SIMPLE, SIMPLER, SIMPLEC, SIMPLEST, and combinations of these algorithms. The robustness and speed of the SIMPLE-based methods were evaluated for a corner-fired furnace and a wall-fired furnace. The importance of temporal fluctuations due to fluid turbulence on the non-linear mixing in gaseous combustion was quantified for corner-fired furnace geometries and compared with similar studies on laboratory scale furnaces.
1989
Furnace Design Using Comprehensive Combustion Models
Smith, P.J.; Sowa, W.A. and Hedman, P.O.
Accepted for publication in Combustion and Flame, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and Brigham Young University.
A new design methodology is presented which allows for the use of comprehensive coal combustion codes in design applications and provides a priori information on the cost of the optimization. A statistical response surface methodology is used to determine appropriate sample points from the design space at which the computation for the comprehensive code are performed. Statistical regression analysis is used to provide interpolating functions for the optimization process. The optimum design point is then checked with a final comprehensive code calculation. The technique is demonstrated with simple examples for design of two injectors for an entrained coal gasifier and of a burner for a pulverized coal combustor. The three designs demonstrate the method as well as showing significantly different optima for different configurations. The importance of specifying operating conditions independently for different injectors or burners is demonstrated.
Algebraic, Multi-Zoned Radiation Model for a Two-Zoned Zero-Dimensional Cylindrical Furnace
Hobbs, M.L. and Smith, P.J.
Accepted for publication in Fuel, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and Utah Power & Light.
A simple two-zone zero-dimensional combustion model which estimates the influence of impurities in the fuel on the radiative energy transport has been developed based on an overall energy balance coupled with a multi-zoned radiation model. This paper presents the equations of the model, illustrates the method of calculating the radiative exchange areas for the two-zone system, and presents predictions for pulverized-coal and fluidized-bed combustion. The model predicts thermal performance as a function of coal input and furnace operational parameters, steam mass flow rates, and superheated steam temperatures leading to the high-pressure turbine. Two wall ash deposit parameters, thermal conductivity, and maximum deposit thickness, have been determined by a sensitivity analysis to be critical to furnace performance. Others have obtained these parameters experimentally. The predictions from the two-zone model have been compared to predictions from an earlier single-zone model. The general trends from both models were the same, although the two-zone model predictions were closer to expected values.
The foundation to describe coal-specific conversion behavior will be AFR's Functional Group (FG) and Devolatilization, Vaporization, and Cross linking (DVC) models, developed under previous and on-going METC sponsored programs. These models have demonstrated the capability to describe the time dependent evolution of individual gas species, and the amount and characteristics of tar and char. The combined FG-DVC model will be integrated with Brigham Young University's comprehensive two-dimensional reactor model, PCGC-2, which is currently the most widely used reactor simulation for combustion or gasification. The program includes: i) validation of the submodels by comparison with laboratory data obtained in this program, ii) extensive validation of the modified comprehensive code by comparison of predicted results with data from bench-scale and process scale investigations of gasification, mild gasification and combustion of coal or coal-derived products in heat engines, and iii) development of well documented user friendly software applicable to a "workstation" environment.
Success in this program will be a major step in improving the predictive capabilities for coal conversion processes including: demonstrated accuracy and reliability and a generalized "first principles" treatment of coals based on readily obtained composition data.
Large-Scale Predictability of Pulverized Coal Combustion Byproducts
Smith, P.J. and Gillis, P.A.
1st International Congress on Toxic Combustion Byproduct: Formation and Control, Los Angeles, California, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and Brigham Young University.
Mathematical model simulations are presented for pilot plant and laboratory test furnaces to quantify the predictability and accuracy of combustion by-product formation and destruction. The roles of numerical accuracy and of closely coupled physico-chemical processes are explored. Capabilities of 3-D furnace models to describe global, site-specific flow and fine scale mesh resolution is shown to be an important consideration in simulating flow patterns in utility boiler geometries, even for relatively simple configurations. Fabricated exact numerical solutions for the coupled particle differential equation set are shown to be useful in identifying algorithmic and coding errors in these large mathematical models. The coupling between local heat transfer and other physico-chemical processes occurring in coal combustion applications is emphasized. Differences of 50-70% are shown to be obtained if particle-gas convective/conductive heat exchange and gas radiation are ignored or if turbulent fluctuations are not accounted for. This variability is shown to have a large impact than the chemical kinetic reaction rate of CO to CO2. The importance of advancing chemistry submodels and their coupling to the turbulent fluid mechanics for more accurate predictability of combustion by-products is emphasized.
Three-Dimensional Computational Fluid Dynamics Modeling in Industrial Furnace Geometries
Gillis, P.A. and Smith, P.J.
Western States Section, The Combustion Institute, Livermore, California, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, and Utah Power & Light.
In recent years, advances in computer technology have revolutionized the methods used to analyze complex flow phenomena. Many practical flows, which were previously expensive and difficult to model, are now gradually yielding their mysteries through the power of today's computers. This process is aided by the continuing development of efficient numerical algorithms and further research into the modeling of physical phenomena that are not well understood, such as turbulence. Several three-dimensional comprehensive combustion models have appeared in the literature over the past three years. Nearly all these models employ similar numerical techniques, variations of the TEACH method. A review of these codes has revealed that they use coarse grid structures and typically lack validation through comparison with experimental data. The purpose of this paper is to describe and demonstrate a three-dimensional CFD (Computational Fluid Dynamics) code that has been designed for industrial geometries. This code includes new numerical techniques, has been validated with data comparisons in several industrial furnaces, and will be used to establish the grid resolution required to obtain grid independent solutions.
The CFD model was designed to be a fundamental element in a larger comprehensive pulverized-coal combustion code for utility boilers. The model handles Cartesian and cylindrical coordinate systems and allows for irregular grid spacing in order to efficiently model the large scale disparities encountered in industrial furnaces. The model includes the option of employing different variations of the SIMPLE algorithm to couple the Navier-Stokes equations. Coupling options include the SIMPLE, SIMPLER, SIMPLEC, SIMPLEST, and various combinations of these algorithms. First-order finite differencing is performed with a combined central and revised upwind scheme that has been adapted to handle grid irregularities. The resulting finite-difference coefficient matrices are solved with a vectorized relaxed Thomas algorithm. Several turbulence models have been incorporated into the code, include the k-e turbulence model. Significant portions of the model have been vectorized to improve code performance.
The emphasis of this paper is on evaluating model performance. Experimental data has been obtained from Consolidation Coal Company for a wall-fired pilot-scale furnace with four swirled burners. Combustion Engineering has also provided a large collection of velocity data from a tangentially fired furnace. Comparisons will be made between predicted and experiment velocities for both configurations. These comparisons will be performed in different locations within each furnace and model performance will be evaluated. Additional numerical experiments were performed to determine the grid resolution needed to achieve grid independent solutions. Recommendations will be made detailing the grid resolution needed to accurately model the major types of industrial furnaces. The different coupling algorithm options employed in the model will also be evaluated for both robustness and efficiency. Additional comparisons between the TEACH method and the model's numerical techniques will be made.
Detailed Model for Practical Pulverized Coal Furnaces and Gasifiers
Smith, P.J. and Smoot, L.D.
AR&TD Contractor's Meeting, Morgantown, West Virginia, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates) and ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, and Utah Power & Light.
This project was jointly funded by a consortium of eight different industrial firms and governmental agencies of which PETC is a part. The objective of the project was to improve and extend a generalized two-dimensional, pulverized coal combustion and gasification code for application to large-scale practical configurations. The work initiated in this four-year project is being continued with NSF and private sponsorship under the Advanced Combustion Engineering Research Center. Over the past year the consortium project tasks were significantly scaled down from the preceding three years and focused on the evaluation of model and submodel development, integration of the pertinent algorithms into the 3-D model, and evaluation of the integrated code. This paper will report on work accomplished on this development over the last year and summarize the state of development of the 3-D modeling effort.
Three tasks were originally outlined for this four-year project. Task 1 - extend an available 2-D model to three-dimensional, large-scale furnace and gasifier configurations. Task 2 - evaluate the 3-D model a: through a statistical sensitivity analysis to determine parameters most influential on predictions, and b) through review, selection, documentation and comparison of a set of available data relating to practical, large-scale furnaces and gasifiers with model predictions. Task 3 - improve or develop submodel equations and/or data for the 3-D model.
All submodel work on task 3 was completed in the first three years of the contract and included the development of a new turbulent particle-dispersion model, a new two and three dimensional discrete ordinates model for radiative transfer, the generalization of the devolatilization submodel to include all known devolatilization models, advancements in the char reaction submodel, a new submodel for interface conditions in multi-phase reaction systems (i.e. coal or oil combustion), the incorporation of chemical kinetically limited gaseous reactions of CO - CO2, and the development of a crude but complete framework for incorporating mineral matter transformations and deposition in the combustion chamber. All of this work on this contract has been previously reported.
During this year work has been accomplished towards the incorporation of advanced numerical algorithms and submodels into the first-generation 3-D comprehensive coal combustion code. The code has been assembled for arbitrary geometries in Cartesian and cylindrical coordinate systems. It includes turbulent mixing with eddy diffusivity, mixing-length and k-e turbulence models. It incorporates dispersed particulate phases in an Eulerian description. The gas phase chemistry couples turbulent mixing-limited equilibrium reactions for all major species. Over the last year an ongoing evaluation has occurred to quantify three levels of error: numerical and algorithmic errors, submodel errors, and overall model or data comparison errors. Each of these areas of evaluation is presented in the paper.
A computer graphics package has been developed for the display of three-dimensional combustion data. The package is based on the FIGS graphics industry standard and thus will run on many different hardware platforms. The package allows menu driven access to all computed variables from the 3-D code and takes advantage of color capabilities to cue the user and display the results. A computer graphics pre-processor has been developed to facilitate the creation of the large set of input required for an industrial scale furnace. The complex three-dimensional mesh can be easily defined with menu driven color graphics input.
Calculations have been performed for pilot scale combustion furnaces including wall-fired and corner-fired units. Comparisons have been made with data for non-reacting fluid dynamics, gaseous combustion and coal combustion. Analysis of model performance has been completed for a variety of model options. Analyses of numerical accuracy and model robustness have shown that the formulation and implementation of this code to be more accurate, more computationally efficient and more robust than previous models from this and other laboratories.
An Analysis of Coal Particle Temperature Measurements with Two-Color Optical Pyrometers
LaFollette, R.M.; Hedman, P.O. and Smith, P.J.
Combust. Sci. and Tech., 66, 93-105, 1989. Funded by ACERC (National Science Foundation and Associates and Affiliates).
Two-color optical pyrometers have been used to measure the temperature of reacting pulverized-coal particles. An analysis of such measurements was performed to determine the effect of several possible conditions on the measured temperature. The conditions investigated were the use of a single photo-multiplier to alternately measure the radiant emission at the two selected wavelengths, the presence of soot, light extinction, the choice of wavelengths used to compute the two-color temperature, and non-uniform particle clouds. A computer model of a one-dimensional coal particle cloud was written for this analysis. Results of calculations showed that artificially high temperatures can result if a pyrometer with a single photo detector is used to measure temperatures in a rapidly fluctuating flame. Emission by soot in the coal particle cloud caused unrealistically high temperature measurements. Light absorption by soot lowered the two-color temperature, but not enough to compensate for the rise in observed temperature caused by soot emission. When the wavelengths used are in the visible spectrum, the hotter particles are weighted much more heavily than when the wavelengths are in the infrared region. The use of Wein's Law, a valid approximation to Planck's Law in the visible spectrum, causes substantial error for longer wavelengths. Finally, the two-color temperature of a non-isothermal cloud was weighted most heavily by the hotter particles, depending upon the wavelengths used for the measurement. From this analysis important questions have been raised as to the validity of such measurements under transient conditions and during devolatilization.
1988-1986
A Study of Two Chemical Reaction Models in Turbulent Combustion
Smith, P.J. and Fletcher, T.H.
Accepted for publication in Combustion Science and Technology, 1988. 24 pgs. Not externally funded.
Research efforts with comprehensive computer models that have tried to predict the performance of coal combustors have either neglected the effect of the turbulence on the mean chemical properties or have used one of two approximate methods. This paper focuses on the impact of the turbulence on the chemical reactions of the volatile products of coal combustion processes. It is shown that by ignoring the effect of the turbulence on mean combustion properties significant differences occur as compared to experimental data and predicted by both of the more rigorous models. The first method, the volatile reactances model, is an extension of an approach for premixed gaseous combustion presented by Magnussen and Hjertager. The second method, the statistical, coal-gas mixture fraction model, is an extension of gaseous diffusion flame approaches. These two methods are examined, analyzed and evaluated by comparing predictions from each method with experimental data from three laboratory furnaces. It is shown that while the first method takes only half as much computational time, the second method is required if species and temperatures in zones containing other than mixtures of pure fuel, pure oxidant and pure stoichiometric product are needed. The distribution of eddy mixtures as formulated in the second method is shown to be more consistent with existing limited experimental data.
Coal Combustion Modeling: Particulate and Gas Phase Heat Transfer Effects
Smith, P.J.; Baxter, L.L. and Jamaluddin, A.S.
AIChE Conference, 1988, New Orleans, LA. 15 pgs. All internal funding.
Heterogeneous heat transfer aspects strongly influence the performance of practical coal combustion systems since many of the subprocesses within the flame are highly temperature sensitive, and since the purpose of most furnaces is to extract energy from the flame. Coal combustion simulation or computer modeling permits investigation of the effect of various heat transfer mechanisms within flames on the many other simultaneous processes of turbulent fluid mechanics, coal conversion, gaseous reaction, etc. This paper examines two aspects of heat transfer on pulverized coal combustion processes: 1) the particle dominated radiation process and 2) the gas phase dominated turbulent convection processes. Comprehensive furnace modeling is used to study practical furnaces and the mechanisms are elucidated by comparing predicted results with experimental data from several coal and gas fired furnaces.
Predicting Particulate Deposition from Turbulent Streams
Jamaluddin, A.S. and Smith, P.J.
Engineering Foundation Conference on Mineral Matter and Ash Deposition from Coal, 1988, Santa Barbara, CA. 10 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
A theoretical model has been developed to predict the rate of deposition of particulates from flowing turbulent gas-solid suspensions. The predictions of the model compare favorably with available experimental data. Incorporation of this technique into a comprehensive combustion code demonstrates promise for modeling slagging propensity in coal-fired furnaces.
Discrete-Ordinates Solution of Radiative Transfer Equation in Non-Axisymmetric Cylindrical Enclosures
Jamaluddin, A.S. and Smith, P.J.
Proceedings of the 1988 National Heat Transfer Conference, 1, 227-232, 1988. 6 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
The discrete-ordinates approximation is used to solve the radiative transfer equation in non-axisymmetric cylindrical enclosures containing absorbing-emitting and scattering media, with and without the temperature profile known a priori. Since neither detailed experimental data nor predictions from a zone or Monte-Carlo model for three-dimensional cylindrical enclosures is available, cylindrical equivalents of three-dimensional rectangular enclosures, for which zone model predictions of radiative transfer are available, are used in model evaluation. Limited evaluation of the model shows that the discrete-ordinates method provides acceptable predictions of radiative transfer in non-axisymmetric cylindrical enclosures.
Turbulent Dispersion of Particles
Baxter, L.L. and Smith, P.J.
Western States Section, 1988, The Combustion Institute, Salt Lake City, UT. 21 pgs. Funded by ACERC consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
An approach to describing the transport of particles in turbulent flows is presented which is derivable from first principles. The approach is independent of the particular turbulence model used. It is valid in applications ranging from entrained flow dispersion to fluidized bed applications. It involves no adjustable parameters and requires a description of the turbulence in terms of the mean square velocities and Lagrangian autocorrelation functions.
The model has been rigorously evaluated by comparison to exact solutions, accurate alternative approaches, and experimental data. The model is able to reproduce the exact solutions both analytically and numerically, predict results which are both more accurate and precise than those of the leading alternative models, and predict experimental data within its inherent error under a range of different conditions.
The incorporation of the model into a comprehensive combustor simulator is also demonstrated. The model is found to be more efficient and robust, both in terms of the particle dispersion portion of the comprehensive code and in terms of the overall code performance. Predictions using this description of turbulent particle dispersion yield results that agree in detail with previous predictions using a calibrated particle dispersion model.
Measurement and Prediction of Entrained-Flow Gasification Processes
Brown, B.W.; Smoot, L.D.; Smith, P.J. and Hedman, P.O.
AIChE Journal, 34, 3, 435-446, 1988. 12 pgs. Funding source by Morgantown Energy Technology Center.
A detailed mathematical model is used to predict local and effluent properties within an axisymmetric, entrained-flow gasifier. Laboratory experiments were conducted to provide local properties for four coal types from a gasifier operating at near-atmospheric pressure. Effects of selected model parameters and test variables were examined and compared with measurements in most cases. The comparison of predictions and measurements provides the first evaluation of capabilities and limitations of a comprehensive model for entrained-flow gasifiers.
Three-Dimensional Fluid Dynamics Modeling in Furnace Geometries
Gillis, P.A. and Smith, P.J.
Western States Section, 1988, The Combustion Institute, Salt Lake City, UT. 12 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
A three-dimensional non-reacting flow model has been developed for predicting flow inside industrial furnace configurations. The code uses the SIMPLE algorithm to couple the Navier-Stokes equations and solves the resulting matrices with a vectorized Thomas algorithm. Model predictions have been compared with experimental data in cross-fired furnace geometry. The k-e turbulence model produced significantly superior data agreement than the simpler turbulence models. The effect of inlet condition variation and grid resolution were demonstrated. It was also shown that fine grid spacing is needed to resolve localized large-scale vortex structure.
Revised User's Manual: 87-PCGC-2
Smoot, L.D.; Smith, P.J.; Brewster, B.S. and Baxter, L.L.
Advanced Combustion Engineering Research Center, 1988. Funded by US Department of Energy, Electric Power Research Institute, Consortium, and ACERC.
A two-dimensional, steady state model for describing a variety of reactive and non-reactive flows, including pulverized coal combustion and gasification, is presented. Recent code revisions and additions are described. The model, referred to as 87-PCGC-2, is applicable to cylindrical axi-symmetric systems. Turbulence is accounted for in both the fluid mechanics equations and the combustion scheme. Radiation from gases, wall and particles is taken into account using either a flux method or discrete ordinates methods. The particle phase is modeled in a Lagrangian framework, such that mean paths of particle groups are followed. Several multi-step coal devolatilization schemes are included along with a heterogeneous reaction scheme that allows for both diffusion and chemical reaction. Major gas-phase reactions are modeled assuming local instantaneous equilibrium, and thus the reaction rates are limited by the turbulent rate of mixing. A NOx finite rate chemistry submodel is included which integrates chemical kinetics and the statistics of the turbulence. The gas phase is described by elliptic partial differential equations that are solved by an iterative line-by-line technique. Under-relaxation is used to achieve numerical stability. The generalized nature of the model allows for calculation of isothermal fluid mechanics/gaseous combustion, droplet combustion, particulate combustion and various mixtures of the above, including combustion of coal-water and coal-oil slurries. Both combustion and gasification environments are permissible. User information and theory are presented, along with same problems.
LDV Measurements in Simulated Entrained Gasifier Flows
Lindsay, J.D.; Hedman, P.O. and Smith, P.J.
Submitted to AlChE Journal, 1988. 15 pgs. Funded by Morgantown Energy Technology Center.
Entrained-flow gasification of pulverized-coal has the potential to become a competitive source of energy. One near-commercial application of entrained-flow coal gasification that has been receiving considerable attention is the use of an entrained-flow gasifier in an Integrated Gasification Combined Cycle (IGCC) (e.g., 1, 2). In order to better understand and to improve pulverized-coal gasification processes, a large body of gasification data from within a laboratory-scale entrained-flow gasifier has been collected at this laboratory (e.g., 3-7) and applied toward the development of a comprehensive computer model for pulverized-coal reactors (8-10). This paper summarizes a companion study of the flow processes in an isothermal flow facility that simulates the flow characteristics of the entrained-flow coal gasifier.
A laser Doppler-velocimeter (LDV) was used to make measurements of mean and turbulent velocities both at the inlet, and from within the flow chamber. Isothermal air flows were used to isolate the basic flow properties from such complications as density gradients and chemistry-turbulence interactions. This study emphasized the effects of inlet conditions on flow properties within the simulated reactor (e.g., axial velocity decay, location of recirculation zones, turbulence levels). A knowledge of the effect of inlet conditions on flow properties can lead to improved gasifier operating conditions, can assist in the interpretation of in situ chemical species data from the gasifier, and can guide modeling efforts.
Comparison of experimental measurements with predictions made by a specific computer model was a second objective of this study. The model, PCGC-2 (Pulverized-Coal Gasification and Combustion: 2-Dimensional), is a comprehensive code for pulverized-coal and coal-water slurry processes that has been developed at this laboratory (8-10). The code employs the k-e model for turbulent fluid mechanics. Much of the earlier experimental flow data was collected with intrusive probes, which in some cases seriously distorted the flow being measured. Furthermore, most of the earlier studies did no include measurements at the inlet. Documentation of the inlet boundary condition is needed if experimental data are to be properly applied to model development.
Predicting Radiative Transfer in Rectangular Enclosures Using the Discrete Ordinates Method
Jamaluddin, A.S. and Smith, P.J.
Western States Section, 1987, The Combustion Institute, Provo, UT. 19 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Discrete ordinates solutions of the radiative transfer equation in two and three-dimensional rectangular enclosures containing absorbing-emitting-scattering media have been obtained using S2, S4, S6 and S8 approximations. Evaluation against exact analytical and numerical solutions show that while all of these approximations provide acceptable predictions of the radiation fluxes in two-dimensional enclosures, use of the higher order (higher than S4) approximations is not justified due to substantial increase in computational time and negligible improvement in the accuracy of the predictions. However, for three-dimensional enclosures, the S2 approximation is grossly in error. S4, S6 and S8 approximations predict wall heat fluxes and the temperatures of the medium accurately in these enclosures, but, once again, S4 approximation is shown to be adequate. A study of the sensitivity of the predicted net heat absorption by the walls to the dimensions of the system, and radiative properties of the medium and the surrounding walls, based on Fourier analysis technique, indicates that the predictions are more sensitive to the radiative properties than to the dimensions of the enclosure.
Parametric Sensitivity Study of an Entrained-Flow Pulverized-Fuel Combustion Model
Smith, J.D.; Smith, P.J. and Hill, S.C.
Submitted for publication to Computer and Chem.Engineering, 1987. 35 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
An extensive parametric sensitivity study of a two-dimensional pulverized-fuel (pf) combustion model has been performed to investigate the effect of parametric uncertainty on model predictions. Results illustrate the dominant effect that error in coal devolatilization/oxidation parameters has on predicted burnout, Nox formation, local gas temperature, and coal-gas mixture fraction. Uncertainty in the turbulent particle dispersion parameters appear to have a secondary effect, while error in the particle-gas radiation parameters seems to have little impact on model predictions. Regions of the computational domain exhibiting sensitivity to specific parameters are identified. Specific parameter sensitivity implies the relative importance of various mechanisms in the overall process. Turbulent particle dispersion seems to be important early in the reactor with kinetic processes dominating at and following the predicted ignition point. Radiation appears to be of minor importance for the case under investigation. These results provide unique insight into the general pf combustion process. Specifically, they emphasize the need for a better methodology of predicting the overall volatiles yield. This implies additional understanding of the devolatilization/oxidation mechanism and its role in the overall pf combustion process. Although the results presented here are case specific, they provide unique insight into the overall pf combustion process and illustrate the effects that parameter uncertainty has on model predictions. This information provides fundamental direction for future comprehensive model development and focuses attention on pertinent experimental work to better quantify critical input parameters.
Comprehensive Modeling of Combustion Systems
Smoot, L.D. and Smith, P.J.
ASME/JSME Thermal Engineering Joint Conference, 1987, Honolulu, Hawaii. 13 pgs. Funded by ACERC (National Science Foundation and Associates and Affiliates).
This paper provides a brief review of comprehensive modeling of combustion systems. The focus is on continuous, subsonic flow systems. Supersonic flows and intermittent or unsteady flows with combustion were not considered. Where condensed phases are present, the focus is further placed on entrained systems where particulate and droplet collisions or interactions through motion can be neglected. Gaseous, liquid, solid and slurry fuels were considered. A brief review of related literature is included. Model foundations, elements (submodels), requirements, and potential uses are summarized. Recent technical work of the authors or colleagues in turbulent fluid mechanics of swirling flows, radiation, and gas-phase reactions is summarized. Example predictions for selected cases are shown from several investigators, mostly with comparisons to measured properties. Future directions, research needs and plans are also identified.
The Effects of Interfacial Conditions on Mass and Heat Transfer
Baxter, L.L. and Smith, P.J.
Western States Section, 1987, The Combustion Institute, Honolulu, HI. 3 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Classical assumptions about equilibrium conditions at phase interfaces during heat and/or mass transfer are shown to be approximations that can lead to errors in combustion environments. Specific illustrations involving droplet vaporization are given which show the nature and magnitude of the errors.
A method of relaxing these assumptions is discussed. Equations are developed from fundamental principles of physical chemistry to describe the actual conditions at interfaces. The corrections to the equilibrium assumptions depend on the rates of mass/heat transfer and physical constants that describe specific molecular behavior.
Computations illustrating the use of the nonequilibrium assumption reveal both the potentially large error associated with assuming equilibrium at the interfaces and the sensitivity of the predictions to the molecular parameters.
Statistical Distributions of Coal Particle Sizes in Pulverized-Coal Combustion and Gasification
McDonald, J.B.; Richards, D.O.; Smith, P.J. and Sowa, W.A.
Annual Meeting of the American Statistical Assoc., 1986. 18 pgs. Not externally funded.
Coal particle size distributions for a Utah bituminous coal and a Wyoming subbituminous coal were modeled using four different probability density functions (GB1, GB2, lognormal, and lognormal by method of moments) and two engineering approximations. The pdf models of the particle size distributions were discretized and compared to the engineering approximations by simulating a coal combustion case and a coal gasification case using the comprehensive combustion model, PCGC-2. The gasification case used the Utah coal and the combustion case used the Wyoming coal. Significant differences were noticed between predictions using engineering approximations of the particle size distribution and predictions using discretized pdf approximations of the particle size distribution. The different methods of describing the particle size distribution affected most notably the simulators prediction of NOx.
Prediction of Radiative Heat Transfer in Cylindrical Furnaces
Jamaluddin, A.S. and Smith, P.J.
Western States Section, 1986, The Combustion Institute, Banff, Canada. 20 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Two radiative heat transfer codes, one based on the six-flux radiation model, and the other based on the S4 discrete ordinates model, have been developed. Both the radiation models approximate the angular variation of the radiation intensity by solving the radiation transport equation in a predetermined number of directions so that the angular integral is removed, resulting in differential equations which can be solved simultaneously with the differential equations of fluid flow, chemical reactions and heat transfer. The limited evaluation of these two models indicates that the six-flux model under-predicts the radiant heat fluxes by up to a factor of two, while the discrete ordinates model affords sufficiently accurate predictions for engineering heat transfer calculations.
Effects of Swirling Flow on Nitrogen Oxide Concentration in Pulverized Coal Combustors
Smith, P.J.; Smoot, L.D. and Hill, S.C.
AIChE Journal, 32, (11), 1917-1919, 1986. 3 pgs. Funded by US Department of Energy.
Nitrogen-containing pollutants from pulverized coal conversion processes have been of concern for several years, and reviews of this subject have been published (Wendt, 1980; Chen et al., 1981). Swirl of the secondary stream is one technique of fuel-air contacting that causes significant changes in NO concentrations. In some cases, increased swirl increases NO emissions (Heap et al., 1973b; Brown et al., 1977; Pershing and Wendt, 1979). In other cases, increased swirl initially decreases NO emissions (Harding et al., 1982; Asay et al., 1983). Thus, the observed effects of swirl are varied, and indicate that additional parameters influence the NO emissions. In this note a model, described in detail elsewhere (Smith and Smoot, 1981; Smith et al., 1981, 1982; Hill et al., 1984), was used for interpreting effects of inlet stream swirl and velocity on NO emissions during combustion of pulverized coal.
Modeling of Swirl in Turbulent Flow Systems
Sloan, D.G.; Smith, P.J. and Smoot, L.D.
Progress Energy Combustion Science, 12, (3), 63?250, 1986. 188 pgs. Funded by US Department of Energy.
The standard k-e equations and other turbulence models are evaluated with respect to their applicability in swirling, recirculating flows. The turbulence models are formulated on the basis of two separate viewpoints. The first perspective assumes that an isotropic eddy viscosity and the modified Boussinesq hypothesis adequately describe the stress distributions, and that the source of predictive error is a consequence of the modeled terms in the k-e equations. Both stabilizing and destabilizing Richardson number corrections are incorporated to investigate this line of reasoning. A second viewpoint proposes that the eddy viscosity approach is inherently inadequate and that a redistribution of the stress magnitudes is necessary. Investigation of higher-order closure is pursued on the level of an algebraic stress closure. Various turbulence model predictions are compared with experimental data from a variety of isothermal, confined studies. Supportive swirl comparisons are also performed for a laminar flow case, as well as reacting flow cases. Parallel predictions or contributions from other sources are also consulted where appropriate. Predictive accuracy was found to be a partial function of inlet boundary conditions and numerical diffusion. Despite prediction sensitivity to inlet conditions and numerics, the data comparisons delineate the relative advantages and disadvantages of the various modifications. Possible research avenues in the area of computational modeling of strongly swirling, recirculating flows are reviewed and discussed.
A Study of Turbulent Particle Dispersion in Pulverized Coal Combustion and Gasification
Smith, P.J. and Baxter, L.L.
Western States Section, 1986, The Combustion Institute, Banff, Canada. 20 pgs. Funding source US Department of Energy, Brigham Young University and Electric Power Research Institute.
The location of the coal particles in pulverized coal combustion and gasification processes is a dominant factor in determining flame stability, combustion/gasification efficiency and pollutant concentrations. Experimental and predicted data show that particles do not mix at the same rate as the gas phase and that this difference in turbulent mixing is important to the coal reaction process. The environment around a given particle during heat up, devolatilization and heterogeneous reaction controls the overall combustion process. In flows of interest to pulverized coal combustion, the particle dispersion is dominated by the random fluctuations of the gas phase turbulence as opposed to the mean drag on the particle. A mixed Eulerian-Lagrangian mathematical model for predicting mean trajectories of expected values for ensembles of representative particles is presented, including the turbulent dispersion effects. The mean turbulent velocity is modeled as a Fickian diffusion process from the man number density as calculated from the Eulerian part of the formation. This model is discussed and compared to alternate procedures. Applications of the model are shown for particle dispersion in non-reacting flow and pulverized coal combustion. An evaluation by comparison with experimental data is presented. Conclusions regarding the significance of turbulent particle dispersion are discussed.
Prediction of High Intensity Pulverized Coal Combustion
Suzuki, T.; Smoot, L.D.; Fletcher, T.H. and Smith, P.J.
Combustion Science and Technology, 45, (3&4), 167-183, 1986. 17 pgs. Funded by Brigham Young University and Kobe Steel Company, Japan.
The overall characteristics of high-intensity pulverized coal combustion have been predicted by a one-dimensional model. The mixing of the primary stream of pulverized coal and transport air with secondary combustion air was estimated by a growth angle of the primary jet. The coal particle burnout was strongly affected by the extent of devolatilization, which varies amount coals. The extent of devolatilization as characterized by variation in a devolatilization coefficient was correlated with either proximate volatiles percentage or H/C mass ratio of the virgin coal. The resulting comparisons of predictions with measurements for eight coal types and tree different combustors show that observed trends are generally predicted. The data used for these comparisons were obtained from a wide range of high-intensity combustion experiments. The proximate volatile matter in the virgin test coals ranged form 16 to 40 percent while the coal feed rate was varied from 12 to 290 kg/hr. Combustion air temperature varied from 297 to 1483 K while residence time ranged from 3 to 140 ms. Comparative results suggest that the predictive method can be useful in interpreting high intensity combustion test results.
Optimization of Coal Devolatilization Model Parameters for Comprehensive Gasification and Combustion Models
Baxter, L.L.; Smith, P.J. and Smoot, L.D.
Western States Section, 1986, The Combustion Institute, Banff, Canada. 17 pgs. Funded by ACERC Consortium: Babcock & Wilcox, Combustion Engineering, Consol, Electric Power Research Institute, Empire State Electrical Energy Research Corp., Foster Wheeler, Pittsburgh Energy Technology Center, Tennessee Valley Authority, and Utah Power & Light.
Parameters for empirical devolatilization models are recommended for lignite and bituminous coals. These parameters were generated by combining optimization algorithms with rigorous particle and flow field modeling of experimental facilities.
PCGC-2, a comprehensive, axisymmetric, gasification and combustion code developed at the Brigham Young University Combustion Laboratory, was used to predict the velocity, temperature, and radiation fields of experimental apparatus used to collect devolatilization data. The effect of changing devolatilization models on these fields was neglected due to very low particle loadings. These fields were subsequently entered into a rigorous particle reaction model and the particle response to them was predicted.
A particle model predicted the trajectories, temperature, and reaction rates of coal particles in the flow fields described above. The particle model and its results are discussed. The particle model was coupled with an optimization algorithm to determine optimum parameters for the two-reaction devolatilization model.
Generalized reduced gradient (GRG) and sequential quadratic programming (SQP) methods were used in performing the optimization with OPTDES, an optimization program developed at Brigham Young University. The model parameters were optimized over six coal heat-up rates for each coal type. Statistical analyses of the residual sum of squares for the optimized two-reaction model revealed that there was no significant lack of fit of the data compared to the inherent variability of the data.
Furnace Design and Comprehensive Coal Combustion Models
Smith, P.J.; Sowa, W.A. and Hedman, P.O.
ASME Annual Meeting, 1986. 14 pgs. Funded by Morgantown Energy Technology Center and Brigham Young University.
Comprehensive coal combustion (including gasification) models have received little to no use in furnace design applications due, in part, to their large computational burden when used in conjunction with some design optimization strategy. This paper presents a new design methodology that allows for the use of comprehensive coal combustion codes in design applications and provides a priori information on the cost of the optimization. A statistical response surface methodology is used to determine appropriate sample points from the design space at which the computations for the comprehensive code are performed. Statistical regression analysis is used to provide interpolating functions for the optimization package. The final design point is then checked with a final comprehensive code calculation.
The technique is demonstrated with simple examples for design of two injectors for an entrained coal gasifier and of a burner for a pulverized coal combustor. The three designs show the utility of the method as well as showing significantly different optima for different configurations. The importance of specifying operating conditions independently for different injectors or burners is demonstrated. The utility of comprehensive coal combustion codes as another design tool is demonstrated.
Statistical Optimization of Feed Streams for an Entrained-Flow Coal Gasifier Using the PCGC-2 Simulator
Sowa, W.A.; Free, J.C.; Smith, P.J. and Hurst, T.N.
ASME Annual Meeting, 1986. 5 pages. Funded by US Department of Energy.
Concepts from statistical response surface methodology (RSM) and nonlinear optimization theory have been combined in a method for efficiently searching a "design space" when using large-scale analysis. The method is applied to a theoretical study of entrained-flow coal gasification, using PCGC-2, a two-dimensional, axisymmetric model developed at Brigham Young University, that predicts local properties in a reaction chamber. RSM was used to sample the design space and to construct a regression model, which was then linked to OPTDES, BYU, a program containing several nonlinear programming algorithms.
Two test plans were used to verify the performance of the RSM optimization algorithm in solving a gasification example problem. The method is shown to provide accurate results, while de-coupling the search and analysis phases of optimization. The importance of a proper test plan for conducting an un-biased exploration of design space is also demonstrated by comparing the results of the two test plans.
LDA Measurements in Simulated Gasifier Flows with Swirl
Lindsay, J.D.; Hedman, P.O. and Smith, P.J.
International Symposium on Laser-Doppler Velocimetry, 1986, Lisbon, Portugal. 6 pgs. Funded by Morgantown Energy Technology Center and US Department of Energy.
A laser-Doppler system was used in a cold-flow study of a simulated pulverized-coal gasifier. The study was designed to provide fundamental information about the behavior of flow in such a gasifier and to provide data for the validation of a computer model. Measurements in 21 flow cases with and without swirl were made. Results are compared with predictions from a comprehensive model that uses a k-e turbulence submodel. Several levels of replication were used in the testing in order to examine reproducibility and to permit statistical analysis of results.