Understanding Energy Conversion in Combustion
Master thesis
Permanent lenke
https://hdl.handle.net/11250/2787908Utgivelsesdato
2021Metadata
Vis full innførselSamlinger
Sammendrag
I dette prosjektet har tre forskjellige modeller for flammer blitt undersøkt: forblandet velblandetreaktor, forblandet fritt-spredene flamme, og ikke-forblandet motstrøms flamme. Måletmed prosjekter var å først bestemme en måte å beregne entropiproduskjoenen i flammer, samthvordan den fordeleres seg i rom, og deretter utføre en slik analyse på modellene nevnt over,med metan og syntese gass (syngas) som brensel. For hvert brensel ble det brukt detaljert,redusert, og globale mekanismer. GRI-mech 3.0 [41] ble brukt som den detaljerte mekanismenfor begge brenslene, mens DRM19 [24] og Davis et al. [15] var de reduserte mekanismenebrukt for metan og syngas, respektivt. For metan ble det bruke en global mekanisme med enreaksjon, laget av Westbrook og Dryer [47]. I midlertid ble det brukt to globale mekanismerfor syngas. For den vel-blandete reaktoren, og motstrøms flammen ble en global mekanismepresentert av Cuoci [11] brukt. En annen mekanisme presenter av Marzouk og Huckaby [32]ble brukt for den fritt-spredene flammen. I tillegg til variasjon av forenkling i den kjemiskemekanismen var trykket variert mellom 1 atm, 10 atm og 20 atm. Dette resulterte i totalt 54forskjellige tilfeller som skulle simuleres.Entropi produksjonen ble beregnet i en post-prosess analyse i en separat kode for den frittutvidendeflammen og motstrøms flammen, mens den var inkludert i koden for den vel-blandareaktoren. For å validere koden ble entropi produksjonen beregnet for metan ved bruk av dendetaljerte og redusert mekanismen, sammenlignet med Nishida et al. [35]. Dermed ble kodenefor, den fritt-spredene flammen, entropi produksjonen, samt bruken av GRI-mech 3.0 [41] ogDRM19 [24] for metan validert. Videre ble koden for motstrømsflammen, samt bruken av GRImech3.0 [41] og Davis et al. [15] for syngas, validert mot Som et al. [42]. Den vel-blandetereaktoren var antatt å være såpass enkel at en omgående validering ikke var nødvendig. Denvar derfor kun sammenlignet med manuelle beregninger.To større hindre, blant flere små, ble møtt i prosjekt. Den ene var at Cantera [16] ikke gir utflerkomponent diffusjons koeffisientene, nødvendig for entropi produksjon fra diffusjon, i hvertpunkt i løsningen. Derfor ble i stedet stoffenes molare diffusjon hentet ut i hvert punkt og brukti utregningene. Den andre var entropi produksjon fra intern varmeoverføring fra reaksjonenetil luft-brensel blandingen måtte tas med for reaktoren, i tillegg til produksjonen fra reaksjonenesom først var antatt å være den eneste kilden. Når hindrene var overkomne ble flammenesimulert, og entropi produksjonen og dens distribusjon i rommet ble beregnet.Fra resultatene ble det oppdaget at både for den vel-blandete reaktoren og den fritt-spredeneflammen, økte den integrerte entropi produksjonen med trykket for begge brenselene, ved brukav alle mekanismene. I motsetning sank produksjonen for begge brenselene, ved bruk av allemekanismer i motstrøms flammen. Grafene som viste den lokale produksjonen ble høyere ogsmalere for begge stoffene, og alle mekanismene i den fritt-spredene flammen. Profilen for denlokale produksjonen i motstrøms flammen var mindre avhengig av trykket.Videre ble det oppdaget at den redusert mekanismen for metan fungerte bra for å beregneden integrerte og den lokale produksjonen for den fritt-spredene flammen, men hadde noenavvik for motstrøms flammen. Den reduserte mekanismen for syngas fungerte bra i begge modellene.Den globale mekanismen for methan fungerte bra for å beregne den integrerte produksjoneni alle modellene. Den hadde noe avvik ved 1 atm for den fritt-spredene flammen, og ved10 atm og 20 atm for motstrøms flammen. Den globale mekanismen brukt for syngas i denfritt-spredene flammen fungerte ikke bra. Den globale mekanismen brukt i de andre modellenefungerte derimot overraskende bra. In this project, three different flame models have been investigated, namely premixed wellstirredreactor, premixed freely-propagating flame, and non-premixed counterflow flame. Theobjective of the project was to figure out a way to analyse the entropy production in flames,and how it is distributed in space, and thereafter perform such an analysis on the flame modelsmentioned. In the analyses, methane and syngas were used as fuels. For each fuel it wasused one detailed, one reduced and one global mechanism. GRI-mech 3.0 [41] was used asthe detailed mechanism for both fuels, while DRM19 [24] and [15] was used as the reducedmechanisms for methane and syngas, respectively. For methane, a global mechanism consistingof one equation created by Westbrook and Dryer [47] was used. Meanwhile, for syngas twoglobal mechanisms were used. For the well-stirred reactor, and the non-premixed counterflowflame, a global mechanism presented by Cuoci et al. [11] was used. Another global mechanismpresented by Marzouk and Huckaby [32] was used for the last model. In addition to varyingthe chemical mechanism, in degree of simplification, the pressure was varied between 1 atm,10 atm, and 20 atm. This resulted in a total of 54 individual cases to be simulated.The entropy production was estimated in a post-process analysis in a separate code forthe freely-propagating and counterflow flames, while it was included in the code for the wellstirredreactor. To validate the codes, the entropy production estimated for methane using boththe detailed, and reduced mechanisms were compared with results obtained by Nishida et al.[35]. Thus, the code for the freely-propagating flame, the use of GRI-mech [41] and DRM19[24] for methane, as well as the post-process analysis code were validated. The code writtenfor the counterflow flame, as well as the use of GRI-mech [41] and Davis et al. [15] for syngas,were validated against Som et al. [42]. The well-stirred reactor was considered to be of suchsimplicity that an extensive validation was unnecessary. The values obtained for entropy washowever crosschecked with manually. This was also done for the freely-propagating flame.Two larger obstacles, amongst more less time-demanding obstacles, were met during theproject. Firstly, the multi-component diffusion coefficient necessary for the calculation of entropyproduction due to mass diffusion was not given by Cantera [16]. Therefore, the speciesmole fluxes were retrieved rather than the coefficients. Secondly, it was discovered that the entropychange in the reactor was not caused by the chemical reactions alone, as initially thought.The entropy production due to internal heat transfer from the reactions to the fuel-air mixturealso had to be accounted for. With all obstacles sorted, the flames were simulated, and theentropy production, with its distribution in space was estimated. It was discovered that both in the well-stirred reactor model, and the freely-propagatingflame model the integrated entropy production increased with pressure for both fuels, with allmechanisms. In contrast, the production decreased for both fuels, with all mechanisms in thecounterflow model. The profiles of the local entropy production got thinner and taller for bothfuels in the freely-propagating flame. The counterflow model was less pressure dependent.Furthermore, the reduced mechanism for methane worked well for the integrated entropyproduction and the local production in the freely-propagating flame, but had some discrepanciesin the counterflow flame. The reduced mechanism for syngas worked well in both models.The global mechanism for methane worked well to calculate the integrated production in bothmodels, but had discrepancies for the local production at 1 atm for the freely-propagatingflame, and at 10 atm, and 20 atm for the counterflow model. The global mechanism usedfor syngas in the freely-propagating model did not work, and was not appropriate to use. Theglobal mechanism used in the reactor and counterflow flame models worked well.