TU Darmstadt / ULB / TUbiblio

Large Eddy Simulation of Mixing and Combustion in Combustion Systems under Non-adiabatic Conditions

Pantangi, Pradeep :
Large Eddy Simulation of Mixing and Combustion in Combustion Systems under Non-adiabatic Conditions.
[Online-Edition: http://tuprints.ulb.tu-darmstadt.de/5827]
Technische Universität , Darmstadt
[Dissertation], (2016)

Offizielle URL: http://tuprints.ulb.tu-darmstadt.de/5827

Kurzbeschreibung (Abstract)

The application of combustion modeling in achieving the objective of predicting the combustion systems very precisely is increasing rapidly and it is a necessary tool. Combustion systems involve complex unsteady process such as flow, turbulence, mixing, chemistry, heat transfer and interaction of this phenomenon with each other. The present thesis work is focused on advanced modeling of combustion, mixing and flame wall interaction.

First part of the work aims at an evaluation of the ability of combustion-LES by FASTEST3D CFD code to correctly describe turbulent premixed combustion, especially a rod stabilized unconfined flame. For this purpose the flamelet generated manifold (FGM)-tabulated chemistry approach, in which a variable local equivalence ratio due to a possible entrainment of the environment air is included through a mixture fraction variable, and it is integrated into an appropriate complete model. Since the state of the local distribution of scalars is strongly dependent on the scalar flux known to generate or produce the scalar variance which is a measure of the mixedness, a newly developed Subgrid-scale (SGS) model for scalar flux is investigated. In general, SGS model assessment was mainly achieved through comparisons with Direct Numerical Simulation (DNS) results limited to low Reynolds numbers. Based on a comprehensive, highly resolved experimental database of SGS and mean scalar field quantities, an assessment of this new anisotropic SGS scalar flux model is carried out for a rod stabilized premixed methane V-flame configuration characterized by high Reynolds numbers for which DNS-data are not available. For that purpose, Large Eddy Simulations are performed using the dynamic Smagorinsky SGS model for the flow field. The new anisotropic SGS scalar flux model used to describe the turbulent scalar flux combines the conventional linear eddy diffusivity model with an additional contribution that couples in a thermodynamically consistent way the deviatoric SGS stress tensor and the gradient of the filtered scalar field. The combustion is modeled by a detailed tabulated chemistry based method following the FGM approach. To assess the prediction capability of the anisotropic SGS scalar flux model, LES achievements are compared against the highly resolved experimental data available and other simulation results performed under use of existing SGS scalar flux models. The behavior of SGS scalar fluxes is especially analyzed. It turns out that the new anisotropic model retrieves the overall expected features of the SGS scalar fluxes at both resolved and SGS levels and in both non-reacting and reacting premixed environments. It also allows achieving LES results for the flow and scalar field that are in better agreement with experimental data. A satisfactory agreement for the flow field quantities and species concentrations is achieved along with an assessment of the SGS scalar flux model used. In the second part, the FGM model based combustion modeling is extended to a complex fuel system for a reliable description of combustion in a gas turbine combustion chamber. In order to evaluate the capability of the model for predicting combustion processes induced by complex real fuels a high pressure single sector combustor (SSC) is investigated. This combustion chamber is fuelled with pre-vaporized kerosene fuel and features very complex unsteady swirling flow and partially premixed combustion properties. The validation of the designed tool along with the prediction analysis is carried out in terms of comparison between experimental data (achieved with a nozzle fired at 0.4 and 0.6 MPa) and numerical results. This reveals that the proposed LES model is able to capture satisfactorily the flow and combustion properties involving. In particular the flame is predicted to be not always attached to the nozzle. It fluctuates between a lifted and an attached regime. This agrees with experimental findings.

In the last part focus is given on development of the combustion modeling under non-adiabatic conditions. Initially the FGM table generation for non-adiabatic combustion is considered and evaluated with existing methods and thereby it was extended to generate non-adiabatic FGM table from non-unity Lewis number flamelets. A special focus is given to find the optimized reaction progress variable to construct the non-adiabatic FGM table. Firstly the suitable RPV was found, then validated for non-adiabatic combustion taking advantage of 1D flame by an independent (Cantera) CFD solver. The validated non-adiabatic FGM table is used for simulating a 3D laboratory scale non-adiabatic configuration.

The developed FGM table methodology was used to investigate non-adiabatic flame wall interaction (FWI) resulting from a premixed methane fuel jet impinging on a spherical disk. This is achieved by FASTEST3D explicit time stepping LES CFD solver. Here, nitrogen is used as co-flow in order to avoid the interaction with the surrounding air and combustion. The flow field is described by means of the Smagorinsky model with Germano procedure for SGS stresses. The SGS scalar flux in scalar transport equations is modeled by the linear eddy diffusivity model. Artificially thickened flame (ATF) combustion model is used to describe the combustion phenomenon more precisely by LES solver. Catalytic effects on the wall are not considered. Two aspects are especially addressed in this work. First focus is on the grid resolution required near the wall without including a special wall-adapted SGS modeling in reacting configurations. The second aspect is devoted to the integration of the near wall kinetic effects into the FGM framework. For this purpose the enthalpy has been considered as an additional variable in generating the FGM table. Such a tabulated non-adiabatic FGM combustion modeling is applied to FWI for a first time in the present work. The results for the flow field, mixing and combustion properties are presented and analyzed in terms of grid resolution, Reynolds number (in reacting and non-reacting case) and adiabaticity. The results obtained from applied artificially thickened flame combustion models based LES simulations were compared for flow, concentration, temperature and heat flux values from the experimental investigation. Comparisons with available experimental data show satisfactory agreement. An outline of the thermal and flow boundary layer analysis is subsequently provided and special focus is given on heat flux prediction dependency on the type FGM tables considered. This work is completed by conclusions and outlook which summarize the main findings.

Typ des Eintrags: Dissertation
Erschienen: 2017
Autor(en): Pantangi, Pradeep
Titel: Large Eddy Simulation of Mixing and Combustion in Combustion Systems under Non-adiabatic Conditions
Sprache: Englisch
Kurzbeschreibung (Abstract):

The application of combustion modeling in achieving the objective of predicting the combustion systems very precisely is increasing rapidly and it is a necessary tool. Combustion systems involve complex unsteady process such as flow, turbulence, mixing, chemistry, heat transfer and interaction of this phenomenon with each other. The present thesis work is focused on advanced modeling of combustion, mixing and flame wall interaction.

First part of the work aims at an evaluation of the ability of combustion-LES by FASTEST3D CFD code to correctly describe turbulent premixed combustion, especially a rod stabilized unconfined flame. For this purpose the flamelet generated manifold (FGM)-tabulated chemistry approach, in which a variable local equivalence ratio due to a possible entrainment of the environment air is included through a mixture fraction variable, and it is integrated into an appropriate complete model. Since the state of the local distribution of scalars is strongly dependent on the scalar flux known to generate or produce the scalar variance which is a measure of the mixedness, a newly developed Subgrid-scale (SGS) model for scalar flux is investigated. In general, SGS model assessment was mainly achieved through comparisons with Direct Numerical Simulation (DNS) results limited to low Reynolds numbers. Based on a comprehensive, highly resolved experimental database of SGS and mean scalar field quantities, an assessment of this new anisotropic SGS scalar flux model is carried out for a rod stabilized premixed methane V-flame configuration characterized by high Reynolds numbers for which DNS-data are not available. For that purpose, Large Eddy Simulations are performed using the dynamic Smagorinsky SGS model for the flow field. The new anisotropic SGS scalar flux model used to describe the turbulent scalar flux combines the conventional linear eddy diffusivity model with an additional contribution that couples in a thermodynamically consistent way the deviatoric SGS stress tensor and the gradient of the filtered scalar field. The combustion is modeled by a detailed tabulated chemistry based method following the FGM approach. To assess the prediction capability of the anisotropic SGS scalar flux model, LES achievements are compared against the highly resolved experimental data available and other simulation results performed under use of existing SGS scalar flux models. The behavior of SGS scalar fluxes is especially analyzed. It turns out that the new anisotropic model retrieves the overall expected features of the SGS scalar fluxes at both resolved and SGS levels and in both non-reacting and reacting premixed environments. It also allows achieving LES results for the flow and scalar field that are in better agreement with experimental data. A satisfactory agreement for the flow field quantities and species concentrations is achieved along with an assessment of the SGS scalar flux model used. In the second part, the FGM model based combustion modeling is extended to a complex fuel system for a reliable description of combustion in a gas turbine combustion chamber. In order to evaluate the capability of the model for predicting combustion processes induced by complex real fuels a high pressure single sector combustor (SSC) is investigated. This combustion chamber is fuelled with pre-vaporized kerosene fuel and features very complex unsteady swirling flow and partially premixed combustion properties. The validation of the designed tool along with the prediction analysis is carried out in terms of comparison between experimental data (achieved with a nozzle fired at 0.4 and 0.6 MPa) and numerical results. This reveals that the proposed LES model is able to capture satisfactorily the flow and combustion properties involving. In particular the flame is predicted to be not always attached to the nozzle. It fluctuates between a lifted and an attached regime. This agrees with experimental findings.

In the last part focus is given on development of the combustion modeling under non-adiabatic conditions. Initially the FGM table generation for non-adiabatic combustion is considered and evaluated with existing methods and thereby it was extended to generate non-adiabatic FGM table from non-unity Lewis number flamelets. A special focus is given to find the optimized reaction progress variable to construct the non-adiabatic FGM table. Firstly the suitable RPV was found, then validated for non-adiabatic combustion taking advantage of 1D flame by an independent (Cantera) CFD solver. The validated non-adiabatic FGM table is used for simulating a 3D laboratory scale non-adiabatic configuration.

The developed FGM table methodology was used to investigate non-adiabatic flame wall interaction (FWI) resulting from a premixed methane fuel jet impinging on a spherical disk. This is achieved by FASTEST3D explicit time stepping LES CFD solver. Here, nitrogen is used as co-flow in order to avoid the interaction with the surrounding air and combustion. The flow field is described by means of the Smagorinsky model with Germano procedure for SGS stresses. The SGS scalar flux in scalar transport equations is modeled by the linear eddy diffusivity model. Artificially thickened flame (ATF) combustion model is used to describe the combustion phenomenon more precisely by LES solver. Catalytic effects on the wall are not considered. Two aspects are especially addressed in this work. First focus is on the grid resolution required near the wall without including a special wall-adapted SGS modeling in reacting configurations. The second aspect is devoted to the integration of the near wall kinetic effects into the FGM framework. For this purpose the enthalpy has been considered as an additional variable in generating the FGM table. Such a tabulated non-adiabatic FGM combustion modeling is applied to FWI for a first time in the present work. The results for the flow field, mixing and combustion properties are presented and analyzed in terms of grid resolution, Reynolds number (in reacting and non-reacting case) and adiabaticity. The results obtained from applied artificially thickened flame combustion models based LES simulations were compared for flow, concentration, temperature and heat flux values from the experimental investigation. Comparisons with available experimental data show satisfactory agreement. An outline of the thermal and flow boundary layer analysis is subsequently provided and special focus is given on heat flux prediction dependency on the type FGM tables considered. This work is completed by conclusions and outlook which summarize the main findings.

Ort: Darmstadt
Freie Schlagworte: LES, Combustion, ATF, FGM, Non-Adiabatic Combustion, SGS Mixing, Combustor, Flame wall interaction, Kerosene, CHEM1D, Cantera, FASTEST3D
Fachbereich(e)/-gebiet(e): 16 Fachbereich Maschinenbau > Fachgebiet für Energie- und Kraftwerkstechnik (EKT)
Hinterlegungsdatum: 01 Jan 2017 20:55
Offizielle URL: http://tuprints.ulb.tu-darmstadt.de/5827
URN: urn:nbn:de:tuda-tuprints-58279
Gutachter / Prüfer: Sadiki, Prof. Dr. Amsini ; Dinkelacker, Prof. Dr. Friedrich ; Janicka, Prof. Dr. Johannes
Datum der Begutachtung bzw. der mündlichen Prüfung / Verteidigung / mdl. Prüfung: 20 Januar 2016
Schlagworte in weiteren Sprachen:
Einzelne SchlagworteSprache
LES, Combustion, ATF, FGM, Non-Adiabatic Combustion, SGS Mixing, Combustor, Flame wall interaction, Kerosene, CHEM1D, Cantera, FASTEST3DEnglisch
Alternatives oder übersetztes Abstract:
AbstractSprache
Die Verbrennungsmodellierung wird immer ein wichtiger Bestandteil der Auslegungs- und Optimierungsaufgaben von Verbrennungssystemen bleiben. Solche Systeme beinhalten komplexe instationäre Prozesse, wie Strömung, Turbulenz, Mischung, Chemie, Massen- und Wärmetransport in wechselseitiger Interaktion. Die vorliegende Arbeit beschäftigt sich mit der Entwicklung und Validierung von fortgeschrittenen Modellen zur Beschreibung der Verbrennungs- und Mischungsprozesse sowie der Flammen-Wand Interaktion mit Hilfe der Technik der Grobstruktursimulation (im Englischen Large Eddy Simulation, LES). Der erste Teil der Arbeit evaluiert die Fähigkeit der Verbrennungs-LES für Vormischverbrennung mit dem "FASTEST3D CFD" Code. Das Modell basiert auf einer tabellierten Chemie, welche im Rahmen des sogenannten „Flamelet Generated Manifold“ (FGM) tabellierten Chemie-Ansatzes generiert wird. In diesem Verfahren wird ein lokales aber variables Äquivalenzverhältnis integriert, um den Effekt der Umgebungsluft in den Mischungsgrad zu berücksichtigen. Da die Skalarflusskomponenten zur Produktion der skalaren Varianz eine entscheidende Rolle spielen, welche den Stand der lokalen Verteilung von Skalaren und somit die Güte der skalaren Vermischung in turbulenten Mischungsprozessen bestimmt, wird ein neues anisotropes Feinstrukturmodell (im Englischen Sub-Grid-Scalar (SGS) Model) für den Skalarflussvektor entwickelt. In dieser Arbeit wird eine umfassende, experimentelle Datenbank der SGS und gemittelten Skalarfelder aus einer turbulenten vorgemischten V-Flamme zur Verfügung gestellt, um das neu SGS Skalarflussmodell zu validieren. Die erzielten Resultate werden mit der Vorhersage von existierenden Modellen verglichen bzw. kritisch evaluiert. Das neue anisotrope SGS Skalarflussmodell kombiniert die Eigenschaft eines herkömmlichen linearen Wirbeldiffusivitätsmodells mit einem zusätzlichen Beitrag, welcher die Anisotropie in einer thermodynamisch konsistenten Weise durch die Kopplung des deviatorischen SGS Spannungstensors und des Gradienten des gefilterten Skalarfelds miteinbezieht. Es stellt sich heraus, dass das neue anisotrope Modell die erwarteten Merkmale der SGS skalaren Flüsse sowohl für den aufgelösten Anteil als auch für die SGS Größen unter nicht-reagierenden und reagierenden Bedingungen zufriedenstellend liefert. Im zweiten Teil wrd das auf dem FGM-Ansatz basierende Verbrennungsmodell erweitert, um die Verbrennung komplexer Kraftstoffe in Gasturbinen zuverlässig beschreiben zu können. Um die Leistungsfähigkeit des Modells zur Vorhersage von Verbrennungsprozessen mit komplexen realen Brennstoffen zu validieren, wird eine Hochdruck-Einzelsektor Brennkammer (SSC) untersucht. Diese Brennkammer ist mit vor-verdampftem Kerosin angetrieben und weist sehr komplexe instationäre Drallströmung und teil-vorgemischte Verbrennungseigenschaften auf. Die Validierung des Modells und die Analyse der Vorhersage werden durch Vergleiche zwischen experimentellen Daten und numerischen Ergebnissen erreicht. Diese zeigen, dass das vorgeschlagene LES-Modell in der Lage ist, die Strömungs- und Verbrennungseigenschaften zufriedenstellend zu erfassen. Insbesondere wird die Position der Flamme korrekt erfasst, die zwischen einen angehobenen Flammetyp und einem angehängten Flammeregime in Übereinstimmung mit experimentellen Ergebnissen schwankt. Im letzten Teil der Arbeit wird die Entwicklung eines Verbrennungsmodells für nicht-adiabatische Verbrennung beschrieben. Zunächst wird die FGM Tabellierung für nicht-adiabatische Verbrennungsbedingungen erweitert, wobei das bestehende Verfahren für die Generierung der FGM-Tabelle für Lewis-Zahl ungleich Eins zunächst verwendet wird. Dann wird eine Strategie entwickelt, um die optimalen Reaktionsfortschrittsvariablen für die Erzeugung der nicht-adiabatischen FGM Tabelle zu bestimmen. Diese Vorgänge werden mit der 1D Flamme aus einem klassischen CFD Löser (Cantera) validiert. Anschließend wird die entwickelte Methode für die Beschreibung der Flamme-Wand-Interaktionsprozesse (FWI) in einem vorgemischten Methankraftstoffstrahl eingesetzt. Hier wird Stickstoff als Co-flow verwendet, um die Wechselwirkung zwischen der Umgebungsluft und der Flamme zu vermeiden. Wie in anderen Teilen der Arbeit wird das Strömungsfeld mit dem Smagorinsky Modell und Germano-Verfahren für die Schließung des SGS-Spannungstensors verwendet. Insbesondere wird der SGS Skalarfluss in den skalaren Transportgleichungen durch das lineare Wirbeldiffusionsmodell hier modelliert, um die spezifischen Effekte des Verbrennungsmodells zu untersuchen. Das künstlich verdickte Flammen- (im Englischen Artificially Thickened Flame (ATF) Model) Verbrennungsmodell wird verwendet, um die Dicke der Flamme genauer mit LES auflösen zu können. Katalytische Effekte auf der Wand sind nicht berücksichtigt. Zwei Aspekte stehen vor allem im Zentrum der Untersuchung. Zunächst wird der Schwerpunkt auf die Gitterauflösung in Wandnähe einer reagierenden Konfiguration ohne spezielle Wandfunktion gelegt. Der zweite Aspekt beschäftigt sich mit der Integration der in der Nähe der Wand existierenden kinetischen Effekte in die FGM Tabellen. Zu diesem Zweck wird die Enthalpie als zusätzliche Variable bei der Generierung der FGM-Tabelle betrachtet. Eine so resultierende FGM-Tabellierungsstrategie wird für die Untersuchung der FWI im Rahmen dieser Arbeit erstmalig eingesetzt. Die Ergebnisse des Strömungsfelds, der Misch- und Verbrennungseigenschaften werden dargestellt und analysiert als Funktion der Gitterauflösung, Reynolds-Zahlen (im reagierenden und nicht-reagierenden Fall) sowie der Adiabatizität. Die erzielten Resultate für die FWI werden für die Strömung, Konzentration, Temperatur und den Wärmefluss mit den experimentellen Daten verglichen und beurteilt. Diese Vergleiche zeigen zufriedenstellende Übereinstimmungen. Eine Analyse der thermischen und Strömungs-Grenzschichten wird anschließend durchgeführt, wobei die Abhängigkeit der Wärmeflussvorhersage von der FGM Tabellen auch betrachtet wird. Zum Schluss werden die wichtigsten Resultate dieser Arbeit zusammengefasst und einen Ausblick gegeben.Deutsch
Export:

Optionen (nur für Redakteure)

Eintrag anzeigen Eintrag anzeigen