Www.lamar.edu



TO:????????????Texas Air Research CenterFROM:?????? Daniel Chen, Helen Lou, Xianchang Li, Christopher MartinLamar University daniel.chen@lamar.edu, (409)880-8786 SUBJECT:??????Annual Progress ReportPROJECT NUMBER: 413LUB0136APROJECT TITLE: CFD Study of Important Flare Operating ParametersPROJECT PERIOD: 09/01/2013-07/15/2015DATE:???11/07/2015????????????? BackgroundCurrent EPA regulations (40CFR60.18) require smokeless flaring, which motivates flare operators to over-steam or over-air to suppress smoke at the expense of combustion efficiency (CE) and destruction efficiency (DE). It is also well known that incipient smoke point (ISP) is a good indicator for good combustion, but the phenomenon is neither well understood nor scientifically defined [1-8]. Further, many factors affect soot emission and unburned/ produced VOC emissions [9-13]. In this project, computational fluid dynamics (CFD) methods based on CHEMKIN CFD-FLUENT [16-19] were used to study important flare operating parameters such as composition, vent gas net heating value (NHVvg), and exit velocity [2-7,10-13]. Lamar University has developed a series of 50-species mechanisms for C1-C4 hydrocarbon combustion that were validated with key performance indicators like laminar flame speeds, adiabatic flame temperature, ignition delay tests [20,21]. In this project, a new mechanism that contains soot precursors and C4 species, LU3.0.1, was developed and validated with experimental data. Lamar's CFD modeling has been validated with several laboratory flame (e.g., Berkeley flame, Sandia flame, and McKenna flat flame) data sets having detailed VOC composition profiles [17]. Currently, CFD methodology is used to model soot yield and combustion efficiency for 2010 TCEQ flare tests data. Further, response surface models were developed based on controlled flare tests data sponsored by EPA (1983, 1984) and TCEQ (2010) for which both soot and CE/DE data are available [2-7]. After simulating various flare scenarios, the validated data can be used to fill the experimental data gap. Response surface models are aimed at future applications for set point determination and flare optimization (e.g., to save steam and fuel gas).Objectives?The objectives of the completed project are1)Study important flare operating parameters: vent gas net heating value (NHVvg), exit velocity, and vent gas species with CFD simulations. 2)Develop combustion mechanisms with soot precursors and implement soot models in the CFD simulation. 3)Develop easy-to–use response surface models to estimate DE/CE and soot emissions with measurable and controllable operating variables. MethodologyComputational Fluid Dynamics (CFD)A computational fluid dynamics simulation is based on the application of fundamental physics along with turbulence and chemistry models. A CFD package such as FLUENT follows the finite volume approach to solve the governing transport equations for temperature, pressure, mole fraction and other fluxes [17-20]. Basically, the Navier-Stokes equations together with equations for mass, energy, and species transport need to be solved.Mechanism Decelopment using CHEMKINCHEMKIN, a reaction engineering software package, was used to develop the reaction mechanism files for use in the CFD software FLUENT. The complete combustion mechanisms are usually too complicated and have to be reduced to a maximum of 50 species to be used in certain Fluent models (such as EDC) and to save computation time [17-21]. Chemical kinetic mechanisms, LU 1.0, 1.1, 2.0 and 3.0.1 for the combustion of C1-C4 hydrocarbons have been generated and validated with data of key performance indicators like laminar flame speeds, adiabatic flame temperature, ignition delay tests using Chemkin. The results for the validation of the LU 1.0, 1.1, and LU 2.0 were published in the literature [17,20]. LU 3.0.1 contains C4 species and important soot precursors that can be used in conjunction with soot models in ANSYS Fluent. The manuscript for LU 3.0.1 is under preparation.Flare Operating Variables and Polynomial / Exponential Correlations The variables known to influence flare efficiencies (NHVCZ, exit velocity, tip diameter, and vent gas species) were used as input variables to correlate with performance indicators (DE/CE/soot) [2-7]. Generalized quadratic response surface models were established with Minitab, MATLAB statistics toolbox, and Microsoft Excel spreadsheets [22-27]. In addition, parameterized Sigmoid function models were also developed.AccomplishmentsMechanism Decelopment using CHEMKIN183451586360Fig. 1 Ignition delay time for propene00Fig. 1 Ignition delay time for propeneA new reduced reaction mechanism LU 3.0.1 was built upon earlier mechanisms (LU 1.0, LU 1.1, and LU 2.0) geared for C1-C3 hydrocarbons. LU 3.0.1 contains important soot precursor species and can handle C1-C4 hydrocarbons. The C4-species included are n-butane, 1-butene and 1,3-butadiene. Soot precursors species (acetylene, ethylene and benzene) employed in ANSYS Fluent soot models are also included. LU3.0.1, with 50 species and 310 reactions, has been compared to the full USC II mechanism and earlier combustion mechanisms for its accuracy, Table I. LU3.0.1 has been validated successfully against experimental performance indicators like laminar flame speed, ignition delay and adiabatic flame temperature using CHEMKIN, Fig. 1. LU3.0.1 has been used in conjunction with in the Moss-Brooks soot model built in ANSYS Fluent to predict black carbon emission in sooty flames (2010 flare study data provided by Aerodyne Research, Inc. [4-7]).Table I Comparison of prediction errors of reduced mechanisms for mole fraction of major species at residence time of 1 sec for C3H6 fuelSpeciesUSC IILU 3.0.1Abs. error %LU 1.0Abs. error %LU 2.0Abs. error %C2H21.46E-061.32E-069.721.19E-0618.531.21E-0617.23CH43.01E-073.02E-070.233.30E-079.683.91E-0730.07CO5.59E-035.70E-031.976.24E-0311.685.98E-037.06CO21.21E-011.21E-010.201.20E-010.641.21E-010.35H21.35E-031.37E-032.101.51E-0312.311.45E-037.41Average abs. error %2.8410.5712.42CFD Simulation of 2010 Flare Study-1111252834640Fig. 2 Measured vs. Predicted black carbon-PDF model00Fig. 2 Measured vs. Predicted black carbon-PDF model-111579114391Computational fluid dynamics (CFD) analysis of soot yield and combustion efficiency have been performed on controlled flare tests for which DE/CE/soot data are available. LU3.0.1 is used in conjunction with turbulence-chemistry models like non-premixed Probability Density Function(PDF) model and Eddy Dissipation Concept model(EDC) in Ansys FLUENT 13 to simulate air assisted flare tests in 2010 study. Both Probability Density Function (PDF) and EDC turbulence-chemistry interaction approaches have been adopted to simulate these flare tests. For 2010 study data, the PDF model provided good predictions for soot yield and CE, Figure 2 and Table II. However, the VOC yields are nearly non-existent when PDF model is used. EDC model results are shown in Figure 3 and Table III . EDC models give more reasonable VOC predictions. CE predictions are comparable to those of PDF. Soot yield predictions are In the right order of magnitude, also within a factor of 2. However, the EDC approach may lead to convergence difficulties, especially for low Btu, low flow rate test cases.Fig. 3 Measured vs. Predicted black carbon-EDC modelTable II Simulated (PDF) and experimental combustion efficiency of air-assisted flares in 2010 flare studyCase no.CE Exp. %CE Simu. %% errorA1.1 96.9 96.93 0.03 A2.1 95.9 95.09 0.84 A3.1 98.3 98.88 0.59 A4.1 97.1 99.23 2.19 A5.1 95.9 99.37 3.62 A6.1 99.4 92.9 6.54 Avg. 2.3 Table III Simulated (EDC) and experimental combustion efficiency of air-assisted flares in 2010 flare studyCase no.CE Exp. %CE Simu. %% errorA1.196.9096.900.99A2.196.8096.801.64A2.394.4094.403.50A2.491.3091.302.79A2.593.7093.704.10A4.394.8094.805.26Avg.3.05Response Surface ModelsThe experimental data for response surface models were collected from previous flare study reports: Flare Efficiency study by McDaniel (1983) [2], Evaluation of efficiency of Industrial flares by Pohl, Payne & Lee (1984) [3], 2010 TCEQ Flare study final report [4, 5]. The details of the data used for modeling is shown in Table IV. The purpose of this work is to develop a reasonably good model based on the experimental data which can be further used to predict both CE and DE under different scenarios. The models were developed in terms of measurable operating variables: vent gas heating values/ compositions, exit tip velocity, and assist rates. Combustion zone heating value is also an important input variable:??Where QVG: Volume flow rate of vent gas (scf/hr)Qf: Volume flow rate of supplementary fuel (scf/hr)Q a: Volume flow rate of assisted air (scf/hr)Q s: Volume flow rate of assisted steam (scf/hr)NHVVG: Net Heating Value of vent gas (BTU/scf)NHVf: Net Heating Value of the supplementary gas (BTU/scf)NHVCZ: Combustion Zone Heating Value (BTU/scf)xeff : Effective fraction (effective fraction of air-assist that causes the dilution), 2% is proposed for 2010 JZ Tulsa testsTable IV Data Range used for modelingReport year# data pointsFuelTip diameter(in)CE %DE %Exit velocity (ft/s)NHVCZ (BTU/scf)19832Propylene5.8690.45-92.2799.9914-252107-2140198458Propane/Nitrogen3, 6, 1286.93-99.8891.03-1000.2-428.2291-2099201089Propylene/TNG/Nitrogen3616.3-99.8921.7-99.90.4-295-405General Quadratic Models Since soot emission was not considered for the CE calculations in the 1983/1984/2010 flare studies, corrections were made based on soot data provided by ARI and designated as "corrected CE" or CCE [2-7]. In this work, quadratic response surface (RS) models between DRE/CE/soot emission and the design/operating parameters were developed based on the 2010 TCEQ flare study and 1983/1984 EPA test data using Minitab and MATLAB statistics toolbox [2-7,22-27]. The general quadratic response surface models were developed and given in Table V:Table V Response surface model equations for steam-assisted flaresModel Equations N R2 R2(Adj) R2(Pred) Log BC = -6.46 +?9.268E-03*V +?0.3991?*S +?0.1774?*MW +?16.23*CHR -?1.521E-03*NHVcz -?7.819E-03?*MW*MW -?3.8568E-03*?V*S -?1.054E-05*V*NHVcz +?0.0052*?S*MW -?1.2215*S*CHR +?2.133E-04*MW*NHVcz -?9.848E-03*CHR*NHVcz1230.960.950.94CE = 83.34 +?32.58?D -?4.541E-05*V -?1.061*S +?45.992*CHR +?1.011E-02*NHVcz +?3.621*D*D -?4.071E-04*S*S -?98.23*D*CHR +?2.212*S*CHR +?2.998E-04*S*NHVcz -2.805E-02*CHR*NHVcz1230.920.910.88DE =100.163 -?1.292?D -?0.0466?V +?0.1883?S -?0.0422?MW -?31.09?DB +?7.67E-04*NHVcz +?1.156?D*D -?4.821E-04*S*S +?0.1938*D*V -?4.492E-03*S*MW +?0.725?MW*DB1170.930.920.89where NHVcz: Combustion zone net heating value (BTU/scf),D: Flare tip Dia (inch); A: Air assist flow (lb/MMBTU), S: Steam assist flow (lb/MMBTU),CHR: vent gas Carbon to Hydrogen Atomic ratio, MW: vent gas molecular weight, DB: Double bond (0/1),V: Exit velocity (ft/s),BC: Soot yield (lb/MMBTU),VG: Vent gas heat flow (MMBTU/hr),CE: Combustion Efficiency (%),DE: Destruction Efficiency (%).Table VI General quadratic response surface model equations for air assisted-flaresModel Equations N R2 R2(Adj) R2(Pred) Log BC = 49.19 -?0.0270*V -?1.744E-04*A +?4.678?*MW +?58.27*?DB -?631*CHR +?1083.7?*CHR*CHR -?2.36E-05*V*A +?4.407E-06*A*MW -?9.437*MW*CHR -?123.8?*DB*CHR880.970.970.96CE = 91.9 +?0.0193?*V – 7.11E-04*A +?23.6?*CHR+?7.88?E-03*NHVcz -1.58*DB -?2.42E-04*?V*A -1.6E-04*A*DB-1.306E-02*CHR*NHVcz830.880.860.85DE = 98.03+0.0362*V -3.07E-3*A +?0.1265*MW- 2.8953*CHR +? 8.98E-05*V*A+5.062E-03*A*CHR740.870.850.84Parameterized Sigmoid ModelsThe two parameters used for the modeling are Combustion Zone Heating value (NHV cz) and exit velocity (v). Here, NHV cz accounts for the steam in the vent gas. Using the Flare test data with soot emission for the years 1983, 1984 and 2010, destruction and removal efficiency, and combustion efficiency were analyzed. CE= 100(1+3.561*exp(-0.01742*NHVcz+3.879*V-0.02*NHVcz*V+ 0.01*V^2))R2 = 0.9085The above model has been extended to DE to find the model constants in MATLAB[27]. DE= 100(1+2.836*exp(-0.01637*NHVcz+3.806*V-0.02*NHVcz*V+ 0.01*V^2))R2 = 0.9009The results are shown in Table VII. These model equations for CE and DE should be used within the range of the exit velocity(0.2-428.2 ft/s) and NHV CZ (95-2140 BTU/scf) used for developing the model itself. The contour plots of the model are shown for the cases CE and DE in Figures 4 and 5 respectively.Table VII Parameterized Sigmoid Response surface model equations for steam-assisted flaresModel Equations N R2 R2(Adj) CE= 100/(1+3.561*exp((-(0.01742*NHVcz-3.879*v+0.02*NHVcz*v-0.01*v^2))))1490.910.91DE= 100/(1+2.836*exp((-(0.01637*NHVcz-3.806*v+0.02*NHVcz*v-0.01*v^2))))1490.900.90 Fig. 4: Contour plot of CE vs. NHV cz and V for steam test casesFig. 5: Contour plot of DE vs. NHV cz and V for steam test casesList of Publications and PresentationsAjit Patki, Xianchang Li, Daniel Chen, Helen Lou, Peyton Richmond, Vijaya Damodara, Lan Liu, Kader Rasel, Arokiaraj Alphones, Jenny Zhou, "Numerical Simulation of Black Carbon (Soot) Emissions from Non-Premixed Flames," Journal of Geoscience and Environment Protection, v. 2, pp. 15-24, 2014.Kanwar Devesh Singh, Preeti Gangadharan, Daniel H. Chen, Helen H. Lou, Xianchang Li & Peyton Richmond (2014), “Computational fluid dynamics modeling of laboratory flames and an industrial flare”, Journal of the Air & Waste Management Association, 64:11, 1328-1340, DOI: 10.1080/10962247.2014.948229.Hitesh S. Vaid, Kanwar Devesh Singh, Helen H. Lou, Daniel Chen, Peyton Richmond, "A Run Time Combustion Zoning Technique towards the EDC Approach in Large-Scale CFD Simulations," International Journal of Numerical Methods for Heat and Fluid Flow, Vol. 24 No. 1, 2014, pp. 21-35. Kanwar Devesh Singh, Preeti Gangadharan, Daniel Chen, Helen H. Lou, Xianchang Li, P. Richmond, " Parametric Study of Ethylene Flare Operations and Validation of a Reduced Combustion Mechanism," Engineering Applications of Computational Fluid Mechanics, Vol. 8, No. 2, pp. 211–228 (2014).Daniel H. Chen, Xianchang Li, Helen H. Lou, Peyton Richmond, Matthew Johnson, "CFD and Response Surface Modeling of Flare Performance: DRE/CE vs. Soot,” AIChE Annual Meeting, November 16-21, 2014, Atlanta, GA. Daniel H. Chen, Kanwar Devesh Singh, Preeti Gangadharan, Xianchang Li, Helen H. Lou, Peyton Richmond, "CFD Study of Flare Operating Parameters,” AIChE Annual Meeting, November 3-8, 2013, San Francisco, CA. References[1] EPA, 40 CFR 60.18. General control device and work practice requirements. ()[2] McDaniel, M., Flare Efficiency study, EPA-600/2-83-052, July 1983. [3]Pohl, J., R. Payne, and J. Lee. 1984. Evaluation of the efficiency of industrial flares: Test Results. EPA-600/2-84-095. Prepared for U.S. EPA Office of Research and Development by Energy and Environmental Research Corporation (May).[4] Allen, & Torres. (2011). 2010 TCEQ Flare Study Final Report, The University of Texas at Austin, The Center for Energy and Environmental Resources, TCEQ PGA No. 582-8-86245-FY09-04 and Task Order No. UTA10-000924-LOAT-RP9, Aug. 1, 2011.[5] Allen, & Torres. (2011). “TCEQ 2010 Flare Study Final Report-- Appendices.” HYPERLINK "" [6] Aerodyne Research Mobile Laboratory Particulate Measurements”, TCEQ 2010 Flare Study Appendix --Particulates, HYPERLINK "" [7] Fortner, Brooks, Onasch, Canagaratna, Massoli, Jayne, Franklin, Knighton, Wormhoudt, Worsnop, Kolb, and Herndon (2012). Particulate Emissions Measured During the TCEQ Comprehensive Flare Emission Study. Industrial and Engineering Chemistry Research. 51. 12586 - 12592.[8] EPA. (2011). Industrial Flares. AP 42, Fifth Edition. Compilation of Air Pollutant Emission Factors, vol. 1: Stationary Point and Area Sources, CH 13: Miscellaneous sources.[9] Baukal, & Schwartz. (2001). The John Zink Combustion Handbook. New York: CRC Press.[10] Brzustowski. (1976). Flaring in the energy industry. Progress in Energy and Combustion Science, 129-141.[11] Castineira, & Edgar. (2008). CFD for Simulation of crosswind on efficiency of high Momentum Jet Turbulent Combustion Flames. Journal of Environmental Engineering, 561-571.[12] Levitsky. (2011). Black carbon and climate change considerations for international development agencies. Environmental Department Papers - Climate change series.[13] Rahimpour, M. R., & Jokar, S. M. (2012). Feasibility of flare gas reformation to practical energy in Farashband gas refinery: No gas flaring. Journal of Hazarodus Materials, 204-217.[14] Wang, Raj, & Chung. (2015). Soot modeling of counterflow diffusion flames of ethylene-based binary mixture fuels. Combustion and Flame, 586-596.[15] Nettles, R. (2014). TCEQ 2010 Flare Study & Supplemental Flare Operations Training.[16] R. J. Hall, M. D. Smooke, and M. B. Colket.?Physical and Chemical Aspects of Combustion.?Gordon and Breach.?1997.[17] K. Singh, P. Gangadharan, D. Chen, H. Lou, X. Li, P. Richmond, “CFD Modeling of Laboratory Flames and an Industrial Flare,” J. Air & Waste Management Association (in Press, 2014).[18] ANSYS FLUENT User's Guide Release 14.0 ; Theory Guide Release 14.0; ANSYS? Academic Research, Release 14.0, November 2011. ANSYS.Inc.[19] R.J. Kee, et al., CHEMKIN 4.1.1, Reaction Design, San Diego, CA, 2007.[20] H. Lou, D. Chen, C. Martin, X. Li, K. Li, H. Vaid, K. Singh, P. Gangadharan, Industrial & Engineering Chemistry Research, Industrial flares special issue, 51 (39), 12697-12705, 2012.[21] High-Temperature Combustion Reaction Model of H2/CO/C1-C4 Compounds- Hai Wang; Combustion Kinetics Laboratory, University of Southern California.[22]Minitab 17 User Guide; Matlab R2010a user manual.[23]Kathleen M. Carley, Natalia Y. Kamneva, and Jeff Reminga. Response Surface Methodology. CASOS Technical Report, Oct. 2004.[24] Boubakri, Hafiane, & Bouguecha. (2014). Application of response surface methodology for modeling and optimization of membrane distillation process. Journal of Industrial Engineering and Chemistry, 3163-3169.[25] Rostamiyan, Fereidoon, Mashhadzadeh, Ashtiyani, & Salmankhani. (2015). Using response surface methodology for modeling and optimizing tensile and impact strength properties of fiber oriented quarternary hybrid nano composite. Composites: Part B, 304-316.[26] Samimi, Modarresi, Dehghani, Rahimpour, and Bolhasani. (2015). Application of the response surface methodology for optimization of an industrial methyacetylene and propadiene hydrogenation reactor. Journal of the Taiwan Institute of Chemical Engineers.51-64.[27]MATLAB and Curvefitting Toolbox R2015b, The MathWorks, Inc., Natick, Massachusetts, United States. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download