2 – Computational Fluid Dynamics




Abstract




Humans have an insatiable desire to model physical phenomena and to continuously improve those models, be it modeling weather patterns for forecasting, molecular modeling in pharmaceutical research, or even biscuits baking in an oven.1 Of particular interest, especially given the nature of this book, is the modeling of combustion and combustion systems. Modeling of combustion comes in many forms, ranging from simple stoichiometry of global reactions, to detailed kinetic modeling of elementary reactions in a combustion mechanism, to one-dimensional reactor models, to full-blown transient three-dimensional models using Computational Fluid Dynamics (CFD).





2 Computational Fluid Dynamics



Edited by Mark Vaccari



Introduction




Mark Vaccari

Humans have an insatiable desire to model physical phenomena and to continuously improve those models, be it modeling weather patterns for forecasting, molecular modeling in pharmaceutical research, or even biscuits baking in an oven.1 Of particular interest, especially given the nature of this book, is the modeling of combustion and combustion systems. Modeling of combustion comes in many forms, ranging from simple stoichiometry of global reactions, to detailed kinetic modeling of elementary reactions in a combustion mechanism, to one-dimensional reactor models, to full-blown transient three-dimensional models using Computational Fluid Dynamics (CFD).


CFD is an area of fluid mechanics that deals with the computer simulation of fluid flow, heat transfer, combustion, etc. CFD is rooted in solving the fundamental governing equations for continuity, momentum, and energy. The volume of interest is discretized into a finite number of volume elements called cells. A single model may have many millions, or even billions, of cells. Appropriate models for turbulence, combustion, radiation, etc. are selected. The knowledge of which particular models are most applicable for a given simulation is a skill acquired through both schooling and experience running CFD simulations. Boundary conditions are then applied to the various boundaries in the model. Then the governing equations in each cell are solved. This is the most computationally intensive task, as all of the equations are coupled to each other in a complicated and nonlinear fashion. Even mid-sized steady-state models, with 20–30 million cells, can take several days on a large computing cluster to solve. If, instead of running the models on a computing cluster, they were run on a single desktop, these models could take more than a year to solve! For larger models and/or transient models (time varying), the computational requirements are even larger, often requiring tens of thousands or even hundreds of thousands of core-hours to solve.


CFD simulations are increasingly becoming a mainstay in the analysis of combustion systems, both in academic/research settings as well as industrial settings. Combustion CFD simulations allow for deeper insight into systems – solving for velocity, temperature, pressure, and species concentrations everywhere in the volume instead of just at a few probe points with experimental apparatus. Additionally, it allows for full-scale simulation of equipment, where full-scale testing is typically prohibitively expensive and challenging (if possible at all). Finally, because combustion CFD simulations predict variables in all locations, it allows for the prediction of quantities that cannot be measured in an experimental apparatus (e.g., vorticity).


The following images demonstrate a variety of applications of combustion CFD simulations from a diverse set of contributors. The images correspond loosely to the following categories (in order): gas turbines, burners in isolation, rocket engines, industrial applications of combustion, emissions reduction, and explosions.



Reference


1.D. Fahloul, G. Trystram, A. Duquenoy, I. Barbotteau, Modelling heat and mass transfer in band oven biscuit baking, LWT – Food Science and Technology, 27, 2 (1994), 119124, ISSN 0023-6438, .



Niveditha Krishnamoorthy , Chandraprakash Tourani , and Matthew Godo


Siemens Product Lifecycle Management Software Inc.

Artyom Pogodin
OOO Siemens Industry Software


Karin Fröjd
Siemens Industry Software AB

2.1 LES Simulation of a Gas Turbine Combustor


Figure 2.1

Transient LES simulation of turbulent combustion in a steadily burning annular gas turbine combustor, using Simcenter STAR-CCM+ from Siemens. The combustion chamber is 16.5 cm long with a height of 6 cm. The simulation is run for one sector and visually duplicated through a periodic transform. The sector is discretized using a 316,000-cell polyhedral mesh. An inlet mass flow of air at 500 K is provided through the plenum inlet, whereof a portion is reaching the combustion chamber through swirler and dilution holes, at velocities of around 100 m/s. The swirler walls and dilution holes are visualized in the illustration. JP10 is injected at 300 K at a mass flow rate of 1 g/s in the center of the swirler using a lagrangian injector. The LES WALE turbulence model is used and the mass averaged Reynolds number is 5 × 104. The Flamelet Generated Manifold combustion model is applied with FGM Kinetic Rate for flame propagation. The blue field visualizes areas with high fuel concentration, where the fuel–air mixture fraction is higher than 0.18. The flame front is visualized through volume rendering, highlighting areas where the combustion progress variable lies between 0.93 and 0.96. The redder the color, the higher the progress variable.




Eleonore Riber and Bénédicte Cuenot
CERFACS

2.2 Annular Burner Ignition


Figure 2.2

This image shows an instantaneous view of the propagating flame in the annular burner experimentally studied in Ref 1. A 0.74 equivalence ratio mixture of propane and air is injected at ambient temperature and pressure through 16 swirled injectors (swirl number 0.82). Results are analyzed in detail in Ref 2.


The authors would like to acknowledge PRACE for awarding access to computing resource of TGCC (France).



References


1.J.–F. Bourgouin, D. Durox, T. Schuller, J. Beaunier, S. Candel, Combust, Flame 160 (2013), 13981413.

2.M. Philip, M. Boileau, R. Vicquelin, E. Riber, T. Schmitt, B. Cuenot, D. Durox, S. Candel, Large eddy simulations of the ignition sequence of an annular multiple-injector combustor, Proceedings of the Combustion Institute 35, 3(2015), 31593166.



Damien Paulhiac , Bénédicte Cuenot , and Eleonore Riber
CERFACS

2.3 Numerical Simulation of a Spray


Figure 2.3

The atmospheric swirled spray flame corresponds to the experimental setup of Ref 1. The image is a visualization of the fuel droplets and the flame surface.


The liquid fuel is n-heptane, injected at a mass flow rate of 0.12 g/s and at ambient pressure and temperature. The global equivalence ratio of the burner is 0.17.


The resulting flame has a complex, partially premixed structure. In addition, a non-negligible part of droplets is able to cross the flame front and ignite in the burnt gas to form individually burning droplets.


The support of SAFRAN is acknowledged.



Reference


1.D. E. Cavaliere, J. Kariuki, E. Mastorakos, A comparison of the blow-off behaviour of swirl-stabilized premixed, non-premixed and spray flames, Flow, Turbulence and Combustion, 91 (2013), 347372.



David Barre , Lucas Esclapez , Eleonore Riber , and Bénédicte Cuenot
CERFACS

2.4 Flame Ignition and Propagation in Aeronautical Swirled Multi-burners


Figure 2.4

The experimental configuration was designed by CORIA in the context of the European project KIAI (Knowledge for Ignition, Acoustics and Instabilities – 7th Framework Program – 2009/2013). The setup allows to vary the separation distance between the burners. Methane and air mass flow rates are, respectively, 0.192 g/s and 5 g/s for each individual injector, leading to a global equivalence ratio of 0.66. The radial swirler is composed of 18 vanes inclined at 45°, leading to a swirl number of 0.76. The burner operates at ambient conditions. The images show a visualization of the flame surface from a side view (top image) and top view (bottom image). Results show the importance of burner spacing on the ignition sequence and the partially premixed flame structure.


The research leading to these results has received funding from the European Community’s 7th Framework Program (FP7/2007–2013) under Grant Agreement No. ACP8-GA-2009-234009. This work was granted access to the HPC resources of [CCRT/IDRIS] under the allocation 2013-x20132b5031 made by GENCI (Grand Equipement National de Calcul Intensif). This research is also part of a 2013 INCITE award of the Department of Energy, and used resources of the Argonne Leadership Computing Facility at Argonne National Laboratory, which is supported by the Office of Science of the U.S. Department of Energy under contract DE-AC02–06CH11357.




Anthony Ruiz , Raphaël Mari , and Bénédicte Cuenot
CERFACS


Laurent Selle
IMFT


Thierry Poinsot
IMFT/CERFACS

2.5 Direct Numerical Simulation of a Transcritical Flame


Figure 2.5

The image, colored by scalar dissipation rate, shows a cryogenic diffusion flame of H2/O2 attached to a lip of thickness 0.5 mm. Hydrogen is injected at 150 K while oxygen is injected at 100 K in a supercritical state. The density ratio between the two reactant streams is 80. The chamber pressure is 100 bar. Results are detailed in Ref 1.


Support for this research was provided by Snecma and CNES. The simulations used resources of the high-performance computing resources of CINES under the allocation 2011-c2011025082 made by GENCI.



Reference


1.A. M. Ruiz, G. Lacaze, J. C. Oefelein, R. Mari, B. Cuenot, L. Selle, T. Poinsot, Numerical benchmark for high-Reynolds-number supercritical flows with large density gradients, AIAA Journal 54, 5(2015), 14451460.



Annafederica Urbano and Laurent Selle
IMF Toulouse

Gabriel Staffelbach and Bénédicte Cuenot
CERFACS

Thomas Schmitt and Sebastien Ducruix
Centrale-Supelec

2.6 Large Eddy Simulation of a 42-Injector Liquid Rocket Engine


Figure 2.6

The image shown here is a temperature isosurface colored by velocity of the BKD operated at the P8 test facility at DLR Lampoldshausen (Ref 1). It is an H2/LOx burner, operating at 70 bar with cryogenic propellants injected in transcritical (LOx) or supercritical (H2) states. The oxidizer/fuel ratio is 4. In such configuration, the turbulent flame is in a purely non-premixed combustion regime. The work was published in Combustion and Flame (Ref 2).


This investigation was carried out in the framework of the French-German REST program initiated by CNES and DLR. All geometrical, operational, and measurement data related to the BKD were kindly provided by DLR Lampoldshausen. The authors are particularly grateful to Stefan Gröning and colleagues who performed the experiments and formulated the test case. Support provided by Safran (Snecma), the prime contractor of the Ariane rocket propulsion system, is gratefully acknowledged.


The authors acknowledge PRACE for awarding them access to resource FERMI based in Italy at Cineca. This work was granted access to the high-performance computing resources of IDRIS under the allocation x20152b7036 made by Grand Equipement National de Calcul Intensif. The support of Calmip for access to the computational resources of EOS under allocation P1528 is acknowledged. The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP/2007-2013)/ERC Grant Agreement ERC-AdG 319067-INTECOCIS.



References


1.S. Gröning, J. S. Hardi, D. Suslov, M. Oschwald, Injector-driven combustion instabilities in a hydrogen/oxygen rocket combustor, Journal of Propulsion and Power 32.3 (2016), .

2.Annafederica Urbano, Laurent Selle, Gabriel Staffelbach, and Bénédicte Cuenot, Exploration of combustion instability triggering using large eddy simulation of a multiple injector liquid rocket engine, Combustion and Flame 169 (2016), 129140.



Romain Bizzari and Antoine Dauptain
CERFACS

2.7 Numerical Simulation of a Rotative Detonation Engine


Figure 2.7

The geometry is a 105 mm diameter, 20 mm high, and 10 mm thick cylinder, as in Ref 1. A stoichiometric H2/O2 mixture is injected at the bottom at Ptotal = 10 bars and Ttotal = 300 K, and feeds six propagative detonation fronts. The exhaust is maintained at 0.4 bar. The isosurface of pressure colored by temperature shows the six detonative fronts. Shocks and mixing layer can be observed in the hot gases.



Reference


1.Yohann Eude, Dmitry Davidenko, Francois Falempin, and Iskender Gökalp, Use of the adaptive mesh refinement for 3D simulations of a CDWRE (continuous detonation wave rocket engine). AIAA paper 2236 (2011), 2011.




Steve Evans
Siemens Industry Software Computational Dynamics Limited


Piyush Thakre
Siemens Product Lifecycle Management Software Inc.


Karin Fröjd
Siemens Industry Software AB

2.8 RANS Simulation of an IFRF Coal Furnace


Figure 2.8

Steady-state RANS simulation of IFRF Coal Furnace (Ref 1) using Simcenter STAR-CCM+ from Siemens. The furnace is 6.5 m long, 2 m high, and 2 m wide. Pulverized coal of temperature 303 K is delivered at a mass flow rate of 0.104 kg/s and a velocity of 50 m/s with a transport air flow of 343 K at 75 m/s through a coal gun of inner diameter 11.5 cm and outer diameter 14 cm. An outer swirling co-flow of air at 300K is delivered through an annulus of inner diameter of 14 cm and outer diameter 23.3 cm at an axial velocity of 30–50 m/s and a swirl number of 1.16. The flow is turbulent with a mass averaged Reynolds number of 104. The k–ε turbulence model is used with lagrangian injection of coal particles, using a Rosin–Rammler particle-size distribution with a reference diameter of 0.2 mm. The particles react through devolatilization and char oxidation by O2, H2O, and CO2. The secondary gas phase reactions are modeled through the Eddy Break-Up combustion model. Radiation is accounted for through participating media radiation (DOM) with gray spectrum model. The red-yellow field indicates volume of significant CO mass fraction (> 1% by mass). Redder color indicates higher CO concentration. It can be noted that significant CO concentrations exist at the end of the furnace. The blue lines show tracks of large coal particles (particle diameter > 0.15 mm). Darker blue indicates larger particles. As can be seen, most particles are burnt out in the main flame and in the recirculation, but some particles escape as large particles, which is easy to see through this visualization. Those are the particles that travel more or less straight between the injector and the outlet and thus do not fully burn out, as well as particles with very high initial diameter, by outlet consisting of primarily ash (initial ash content is 8%).



Reference


1.André A. F. Peters and Roman Weber, Mathematical modeling of a 2.4 MW swirling pulverized coal flame, Combustion Science and Technology 122 (1997), 131182.




George Mallouppas
Siemens Industry Software Computational Dynamics Limited


Rajesh Rawat
Siemens Product Lifecycle Management Software Inc.


Karin Fröjd
Siemens Industry Software AB

2.9 RANS Simulation of a Glass Furnace


Figure 2.9

Steady-state RANS simulation of IFRF glass furnace (Ref 1) with a molten glass layer added at the base, using Simcenter STAR-CCM+ from Siemens. The furnace is 3.8 m long, 1.105 m high, and 0.88 m wide. Air is delivered at a velocity of 125 m/s and a temperature of 1,373 K through a 27.8 cm × 27.2 cm rectangular air inlet. Methane is delivered at a velocity of 10 m/s and a temperature of 283 K through a circular nozzle of 1.2 cm diameter, giving a global mixture with 10% excess air. The geometry is discretized using a 1,000,000-cell hexahedral mesh with local refinement by the fuel injection and the flame. Only half of the furnace is simulated using symmetry boundary conditions along the centerplane, and the picture is created using a symmetry transform. The flow is turbulent with a mass averaged Reynolds number of 8 × 105. The SST k–ω turbulence model is used with the complex chemistry combustion model, applying a skeletal 17 species mechanism with 73 reactions from Lu et al. (Ref 2). Radiation is accounted for through participating media radiation (DOM) with gray spectrum model. A molten glass layer of 15 cm thickness is included at the base of the combustion chamber using the VOF model. The red-yellow field visualizes the flame by showing volumes where temperature is higher than 2,000 K. The higher the temperature, the redder the color. The blue-red field indicates incident radiation at the molten glass surface, ranging from 2 MW/m2 in the dark blue areas to 3.4 MW/m2 in the red area just below the flame. It is easy to see where the incident radiation is lowest and consequently where there is highest risk for the glass to solidify.


The very same case but without the molten glass layer is presented in detail in Mallouppas et al. (Ref 3).



References


1.T. Nakamura, W. L. Vandecamp, J. P. Smart, Further studies on high temperature gas combustion in glass furnaces, Technical report, IFRF Doc No F 90/Y/7, 1990.

2.R. Sankaran, E. R. Hawkes, J. H. Chen, T. F. Lu, C. K. Law, Structure of a spatially developing turbulent lean methane–air Bunsen flame, Proceedings of the Combustion Institute 31 (2007), 12911298.

3.G. Mallouppas, Y. Zhang, R. Rawat, Modelling of combustion, NOx Emissions and radiation of a natural gas fired glass furnace, AFRC 2014 Industrial Combustion Symposium.




Carlo Locci
Siemens Industry Software GmbH


George Mallouppas
Siemens Industry Software Computational Dynamics Limited


Rajesh Rawat
Siemens Product Lifecycle Management Software Inc.


Karin Fröjd
Siemens Industry Software AB

2.10 LES Simulation of a Flameless Combustor


Figure 2.10

Transient LES simulation (Ref 1) of a flameless combustor from Verissimio et al. (Ref 2) in Simcenter STAR-CCM+ from Siemens. Methane is burnt in a cylindrical combustion chamber of diameter 10 cm and length 34 cm, with a convergent nozzle of length 15 cm by the end. Methane is injected at 0.2 g/s at 293 K through 16 fuel inlets of diameter 2 mm surrounding a central air inlet of diameter 10 mm. The air is preheated to 673 K and injected at a mass flow of 4.6 g/s. The flow is highly turbulent with a mass averaged Reynolds number of 6×104. LES with WALE sub-grid scale model is used to model turbulence. The combustion model is Flamelet Generated Manifold with kinetic rate for flame propagation. Thermal and prompt NOx are calculated. The blue field highlights fuel inlets by visualizing mixture fractions > 0.05. The orange field highlights the main combustion zone by showing areas where the combustion progress variable lies between 0.05 and 0.95. The redder the color, the higher the progress variable. The green areas highlight areas where dry NOx is > 50 ppm by mass. Here we can see that NOx is formed in the recirculation regions, where the residence time is high enough to allow for significant NOx production. We can also see that the residence time in the reactor overall is high enough to generate NOx > 50 ppm, indicated by the green volumes by the exhaust of the reactor.



References


1.C. Locci, G. Mallouppas, R. Rawat, Numerical analysis of NO and CO in a flameless burner, AFRC 2016 Industrial Combustion Symposium.

2.A. S. Verissimo, A. M. A. Rocha, M. Costa, Operational, combustion, and emission characteristics of a small-scale combustor, Energy & Fuel 25 (2011), 24692480.




Mark Vaccari
John Zink Hamworthy Combustion

2.11 Enclosed Ground Flare


Figure 2.11

Enclosed ground flares are used to supplement, and sometimes replace, elevated flares in the safe disposal of waste gases. Enclosing the flame has several advantages: no thermal radiation outside the enclosure, low noise, and no visible flame.


This image shows results from a computational fluid dynamics simulation of an enclosed ground flare burning ethene. The flame zone is colored by temperature using volume rendering. Air streamlines were added to help visualize how air entered the unit. A model of a person is added to visualize the scale of the unit.




Benjamin Farcy and Luc Vervisch
INSA Rouen Normandie


Pascale Domingo
CORIA CNRS

2.12 Control of Nitric Oxides Emissions


Figure 2.12

A spray of ammonia discharges into burnt gases downstream of non-premixed burners in a 15 MW furnace, to transform nitric oxides molecules into neutral products. Multi-physics large eddy simulation is performed with a spatial resolution of 170 µm. The figure is a close-up view of the spray injection; complex vortex rings are visible on the side of the spray jets, showing the dynamics of the vortices controlled by the shear layers surrounding the liquid spray. The spray injection is conical, with an angle of 30 degrees and a dispersion angle of 4 degrees. A total number of droplets of about 600 million evolve in the simulation. The droplets are colored by their velocity magnitude (0–50 m/s from blue to red), and the visualization of the vortices is colored by the fluctuations of temperature (0–300 K from blue to red).


This work is funded by ANRT (Agence Nationale de la Recherche et de la Technology) and SOLVAY under the CIFRE no. 73683; part of it was conducted during the 2014 Summer Program of the Center for Turbulence Research, Stanford. This work was granted access to the HPC resources of IDRIS under the allocation 2014-020152 made by GENCI (Grand Equipement National de Calcul Intensif).



Reference


1.B. Farcy, L. Vervisch, P. Domingo, Large eddy simulation of selective non-catalytic reduction (SNCR): a downsizing procedure for simulating nitric-oxide reduction units, Chemical Engineering and Science 139 (2016), 285303.



David Barre and Olivier Vermorel
CERFACS

2.13 Gas Explosion in a Medium-Scale Vented Chamber





Figure 2.13 Computer model of a gas explosion in a vented chamber (the red indicates isotherms and the blue indicates vorticity).


The medium-scale, 1.5 m-long chamber is initially filled with a stoichiometric methane–air mixture. After ignition, a laminar premixed flame develops first, which accelerates and transitions to a turbulent flame after crossing the obstacles.


The support of TOTAL is greatly acknowledged. Computational resources were provided through a grant awarded by the INCITE program of the US Department of Energy.




Randall J. McDermott , Glenn Forney , Matthew S. Hoehler , Matthew Bundy , Lisa Choe , and Chao Zhang
NIST


Christopher M. Smith
Berkshire Hathaway Specialty Insurance

2.14 Large-Scale Structure–Fire Interaction: National Fire Research Laboratory Commissioning Test, Experiment, and Modeling


Figure 2.14




The photograph on the left shows a large-scale experiment studying the interaction between fire and mechanically loaded building elements performed during the commissioning of the National Fire Research Laboratory (NFRL) at the National Institute of Standards and Technology (NIST) in Gaithersburg, Maryland, USA (Ref 1). The midspan section of a 6.2 m-long W16×26 structural steel beam is exposed to a 700 kW open flame from the 1 m2 natural gas burner located 1.1 m below the bottom flange of the beam. The purple-blue hue in the photo is caused by high-intensity, near-ultraviolet lighting used to illuminate the beam through the flames. The pattern of dots on the beam is imaged using two scientific cameras, and the images are processed using Digital Image Correlation (DIC) to resolve rigid body motion and deformation of the beam.


The image on the right shows a large-eddy simulation by the NIST Fire Dynamics Simulator (FDS) (Ref 2) of the NFRL commissioning test shown in the left image. The FDS results are visualized using Smokeview (Ref 3), a data visualization companion to FDS, also developed at NIST. The resolution of the simulation is 1 cm. The flame is depicted as a volume rendering of local heat release rate above a cutoff of 200 kW/m3. False color contours of adiabatic surface temperature are shown on the I-beam. This boundary condition is taken as input to a finite-element structural analysis code to predict the deformation of the I-beam under load in a realistic fire scenario (Ref 4). The FDS code is an open-source, explicit, low-Mach flow solver. Details of the solver may be found in Ref 2.


This work was funded by NIST STRS resources under the Fire Risk Reduction in Buildings Program.



References


1.L. Choe, S. Ramesh, M. Hoehler, M. Bundy, M. Seif, C. Zhang, J. Gross, National Fire Research Laboratory commissioning project: testing steel beams under localized fire exposure, NIST Technical Note TN 1977, National Institute of Standards and Technology, Gaithersburg, MD, 2017.

2.K. McGrattan, S. Hostikka, R. McDermott, J. Floyd, C. Weinschenk, K. Overholt, Fire Dynamics Simulator, Technical Reference Guide, Volume 1: Mathematical Model. National Institute of Standards and Technology, Gaithersburg, Maryland, USA, and VTT Technical Research Centre of Finland, Espoo, Finland, NIST Special Publication 1018-1, Sixth Edition, 2013.

3.G. P. Forney, Smokeview, a tool for visualizing fire dynamics simulation data, volume ii: technical reference guide. National Institute of Standards and Technology, NIST Special Publication 1017-2, Gaithersburg, Maryland, Sixth Edition, 2013.

4.C. Zhang, L. Choe, J. Gross, S. Ramesh, M. Bundy, Engineering approach for designing a thermal test of real-scale steel beam exposed to localized fire, Fire Technology 53, 4 (2017), 15351554.

Oct 6, 2020 | Posted by in Fluid Flow and Transfer Proccesses | Comments Off on 2 – Computational Fluid Dynamics
Premium Wordpress Themes by UFO Themes