Title | What can we learn about flare combustion efficiency from large eddy simulations? |
Creator | Smith, Philip J.; Thornock, Jeremy; Hradisky, Michal; Isaac, Benjamin J.; Jatale, Anchal |
Publication type | presentation |
Publisher | American Flame Research Committee (AFRC) |
Program | American Flame Research Committee (AFRC) |
Date | 2010-09-30 |
Type | Text |
Format | application/pdf |
Language | eng |
OCR Text | Show What can we learn about flare combustion efficiency from large eddy simulations? Philip Smith - Jeremy Thornock Michal Hradisky - Benjamin Isaac - Anchal Jatale The University of Utah TOTeM36 - Maui - 30 Sept 2010 universal flare combustion efficiency (!) correlation? universal flare combustion efficiency (!) correlation? flares are: • big • outdoors • burning • hard to measure ! • hard to calculate ! RANS collaborators: UT-Austin David Castineira & Tom Edgar therefore: universal flare combustion efficiency (!) correlation? flares are: • big • outdoors • burning therefore: Turbulence Model ∗ Reference model C.2.1 Standard k-ε model C.2.2 RNG k-ε model C.2.3 Realizable k-ε with modified C2 C.2.4 Realizable k-ε with wall functions C.2.5 k-ω model (CO2)out ηc (%) 137.94 149.01 134.80 162.37 171.31 55.30 95.3 103.0 93.1 112.3 118.5 37.6 !"#$%&'"()$*+$,"-&./012()$* Table C.2: Computed CO2 mass flows (10 kg/s) and resultant combustion efficiency for several turbulence models. -5 • hard to measure ! From Table C.2, we observe that the reference turbulence model (e.g., realizable • hard to calculate ! k-ε) and the RNG k-ε produce the best results, with a very good estimation of the experimental combustion efficiency (error < 4%). The standard k-ε model still gives an acceptable estimation of this combustion efficiency (error < 6.7%). Therefore all the different k-ε turbulence models (with default values) work very well for these flare simulations. local inhomgeneity • local (spatial & temporal) mixing, reaction, extinction !"#$%$&'"'$()*+,)-.,/(-,$"*$0*1$%/()-,$"*'00,1,'"1,')*,"*023.'*42(%')5* !"#$%&'()*&+)+,"-).&+-/)+,0&$,)*-"+,)0,"(#+,0,'$)$,($)1+&2+"0()) James G. Seebold 04/13/10 Flare flames are well known to produce plumes that are characterized by a distinctly inhomogeneous distribution of local combustion efficiencies. This inhomogeneity imposes the requirement of careful combustion efficiency integration over the plume both radially and axially to obtain an accurate assessment of emission control performance. For example, iQ WKH 86(3$¶V 1983-86 landmark investigation of the combustion efficiency of 1 industrial flares, the archival data produced by the agency-sanctioned extractive-sampling protocols demonstrated conclusively that to obtain the so-FDOOHG ³JOREDO´ FRPEXVtion efficiency, ZKLFK LV WR VD\ D FRPEXVWLRQ HIILFLHQF\ WKDW LV DFFXUDWHO\ UHIOHFWLYH RI D IODUH¶V overall emission control performance, requires detailed integration over the flare plume both radially and axially. In order to establish scaling principles, in that landmark 1983-86 USEPA investigation a homologous sequence of flare tips ( òƎ' 3ƎD, 6ƎD, 12ƎD) was tested that included IRXU ´' flare tips of which 3 were of commercial design. On the larger flare tips, the necessary integration of local combustion efficiencies required the use of an extractive-sampling rake that could be systematically traversed over the flare plume. On the smaller flare tips, hood-sampling WRGD\¶V safe ³SRLQW-and-VKRRW´ UHPRWH PHDVXUHPent techniques that are said to be able to measure real world flare combustion efficiencies. However, it is important to note that, today, NO successful blind-validation of ANY remote measurement technique has been carried out against reliable agency-sanctioned extractive-sampling results. local inhomgeneity In particular that is true of the PFTIR technique that has been used in recent field studies. Indeed, the PFTIR theory assumes, incorrectly, homogeneity of combustion efficiency in flare plumes. To the extent that any ³SRLQW-and-VKRRW´ WHFKQLTXH does not ... effectively integrate combustion efficiencies across the flare plume, or does not have an adequately oblique view of the flare plume, or is not effectively manipulated and pointed in such a way as to carry out a thorough integration of combustion efficiencies both radially and axially over the flare plume, ... LW LV LPSRVVLEOH IRU DQ\ ³SRLQW-and-VKRRW´ WHFKQLTXH, including PFTIR, to obtain a quantitatively accurate characterization of flare combustion efficiency. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! " !#$%&'%()*+!*,!(-#!#,,).)#+./!*,!)+0'1(2)%&!,&%2#13!'45678!16967:!#4;5<=4>7469?!@<=67A65=4!! !!!%B74AC3!*DD5A7!=D!%5<!EF9?56C!@?94454B!948!169489<8:3!#@%GHIIJKGLMGINO3!P9C!"NLMQ!#@%GHIIJKGLOG"IH3!! !!!17R67>S7<!"NLOQ!#@%GHIIJKGLOG"IH3!! !!!17R67>S7<!"NLO! ! opportunity: • use advanced diagnostics and advanced computer simulations together LES for utility flares in western Canada http://flaresimulations.org Wind Fuel Sulfur Case Velocity Velocity content (m/s) (m/s) (mol%) 1 2 5 6 7 8 3 4 9 10 11 12 13 15 15 15 15 0 0 0 0 7.5 7.5 7.5 7.5 7.5 15 1 8 8 8 8 15 1 15 15 1 1 8 15 15 30 0 30 0 15 15 30 0 30 0 15 Simulation Details 5.6M nodes V = 700 m3 fluid kinetic energy tsim = 4.6 sec 2nd Order time/space S4 DO Radiation Model 300 processors twall = 2.3 Days 17100 CPU Hours temperature LES: Turbulent Reaction Rate Spectra LES cutoff Eri(!) <ri'ri'|!> <ri'2> Da (tm/tr) flare fuel stripping Nyquist Limit !(l-1) • if low wave number process: LES offers direct resolution on mesh • if high wave number process: need multiscale model to accurately feed mesh scale heterogeneity no wind - high fuel velocity no wind - low fuel velocity heterogeneity Case 1 1.4 1.2 Carbon Efficiency 1 0.8 0.6 0.4 0.2 0 0 no wind - high fuel velocity LES carbon TimeïFiltered carbon 1 2 3 Time (sec) 4 5 no wind - low fuel velocity Flare Efficiency: fuel stripping • combustion efficiency: • stripping ratio: Flare Efficiency: fuel stripping • combustion efficiency: 1.6 1.4 Combustion Efficiency 1.2 1 0.8 0.6 0.4 • stripping ratio: 3456 0.2 0 0 0.5 1 1.5 2 2.5 Time, sec 3 3.5 4 4.5 5 Flare Efficiency: fuel stripping 789"7&(),$::$*;< • combustion efficiency: !&+=)$*9)$8*&>,8?&@7"?+&(),+)9A !&,"%$")$8*&+=)$*9)$8* !&?$=$*; !&,+"9)$8* !&%$72)$8* !&>2+7&+>>+9)( !&+>>+9)(&8>&%$>>+,+*)&%$72+*)( !&()+"?&@78B&,")+( !&"$,&@78B&,")+( !&>2+7&@78B&,")+( • stripping ratio: Flare Efficiency: fuel stripping • combustion efficiency: 1.6 1.4 Combustion Efficiency 1.2 1 0.8 0.6 0.4 • stripping ratio: 345CD 0.2 0 0 0.1 0.2 0.3 0.4 Time, sec 0.5 0.6 0.7 0.8 Validation Intended Use of Simulation Overarching Problem combustion efficiency & shape from industrial flares TCEQ(?)/other • hierarchical • advanced simulation • advanced diagnostics • subscale leveraged Pilot-Scale Validation Cases Plume Shape U of Alberta wind tunnel & John Zink test pad flare Component Scale Validation Cases Radiation Models participating media Molecularscale Models Combustion Efficiency CANMET wind tunnel flares Non-Reacting LES Reacting LES turbulent mixing (buoyancy-driven turbulent mixing) Sandia He Plume Sandia CH4 Kinetics & Thermophysical Fuel Properties chemical kinetics, transport properties, thermodynamic properties, and surrogate fuel formulation Plume Validation = Uncertainty Quantification ym ! x " < yexp ) u Validation = Uncertainty Quantification ym ! x " < yexp ) u Experimental Uncertainty (ye +/- ue) Verification Error - Numerics (yv +/- uv) Models / Model Parameters (xm +/- um) Scenario Parameters (xs +/- us) Quantification Methodology Validation = Uncertainty Quantification(Broad) ym ! x " < yexp ) u V/UQ Objective xs Modeling Activities Experimental Activities m tot fy fy Consistency Analysis fy fx|y Validation = Uncertainty Quantification ym ! x " < yexp ) u Consistency* = maximum value of γ subject to constraints ≤ −α −x j j xj ≤ βj −y (x ) + y ≤ l − γ m e e e ym (xe ) − ye ≤ ue − γ *see Feeley et al. J. Phys. Chem. A 2004, 108, 9573-9583 Validation = Uncertainty Quantification ym ! x " < yexp ) u • V/UQ algorithm: - from prior knowledge identify estimate of uncertainty bounds (hierarchical validation analysis) - build surrogate model spanning uncertainty space - estimate experimental uncertainty bounds - identify consistent bounds (validation) - predict new scenario (prediction) - integration of scales & physics - integration of simulation & experiments - integration of software & data • Sir Karl Popper: "scientific truth is always uncertain" Sim. & Exp. = V/UQ = ! • Manfred Drosg: 1. "All scientifically relevant data have an uncertainty." 2. "Data without uncertainty cannot be relevant scientifically" Manfred Drosg • flare combustion efficiency requires surface integral enclosing flare • local heterogeneity in combustion efficiency • ‘measure' combustion efficiency: - advanced diagnostics & advanced computer simulations |
ARK | ark:/87278/s6f241vc |
Relation has part | Smith, Philip J.; Thornock, Jeremy; Hradisky, Michal; Isaac, Benjamin J.; Jatale, Anchal (2010). What can we learn about flare combustion efficiency from large eddy simulations? Presentation given at TOTeM36, Maui, HI September 30, 2010. Institute for Clean and Secure Energy (ICSE). |
Format medium | application/pdf |
Rights management | (c)American Flame Research Committee (AFRC) |
Setname | uu_afrc |
ID | 1525779 |
Reference URL | https://collections.lib.utah.edu/ark:/87278/s6f241vc |