2. 2.

Combustion models Laminar and Turbulent premixed

2.1.1 Zeldovichs analysis of flame propagation The movement of the flame front in space depends on the heat transfer between the burning gases and the freshly mixed gases. model consists of simple combustion reactions. reactants and products are ideal gases. pressure diffusion and heat loss are not considered. It's a one dimentional model for laminar flames. It should be pointed out that the significance of the theory of laminar flame spreading is not on calculating the value of S L , but on revealing the effects of major physical and chemical parameters of various combustible mixed gases on the combustion process. 2.1.2 Turbulent flame propagation Model is used for turbulent state. It consist of 2 categories: 1. Flame surface folding model(turbulent flame is the result of folding occurred on the Laminar flame surface induced by turbulent flow): a. Small-scale turbulent flame (2300 Re 6000) When the turbulence intensity is low. The size of the fluctuating eddy inside the m fluid is smaller than the thickness of the laminar flame combustion zone. Large deformations of the flame surface would not be induced The flame surface and thickness of the flame region would increase. b. Large-scale weak turbulent flame When the fluid fluctuations increase and give rise to large-scale eddies, its turbulence scale is larger than the thickness of the combustion zone. the flame surface would become folded and deformed. c. Large-scale strong turbulent flame (Re > 6000) When the turbulence intensity is sufficiently large, the turbulent pulsation velocity exceeds the spreading rate of laminar flame. The fluid eddies would break through

the flame surface and tear it into dispersed small pieces. The surface of each piece has a thin layer of closed laminar combustion zone. 2. Control-volume model high turbulence intensity, the combustion reactions are not concentrated in the thin combustion zone, but penetrated into the deeper zone. The thickness of the flame zone is about several ten times of the laminar flame surface. This model is used to estimate the spreading rate of turbulent flame. 2.1.3 the concept of self-turbulence flow It is used In order to explain the difference between the experimental value and the calculated value of spreading rate of turbulent flame S T. the flame itself has fluctuations, giving turbulent flow that affects the combustion process and increases the value of S T. It has not yet been investigated thoroughly.

2.1.4 laminar premixed flame Fresh gases and products are separated by a thin reaction zone. A strong temperature gradiant is observed (T fresh gases/ T burnt gases = 5 to 7). Because of ability of premixed flame to propagate towards the fresh gases, fresh gases are preheated and then start to burn. It is applicable for one-step irreversible simple chemical reactions. Used for a fuel rich, without heat losses and compressibility effects. Therefor, mass and energy balance equations reduce to a single balance equation for the progress variable and by introducing the displacement speed of the iso-c surface. Molecular diffusion normal to iso-c surface and reaction rate may be modeled with the laminar flame speed. 2.1.5 finite rate chemistry model finite rate chemical reaction model involves tens of components which undergo several hundreds of intermediate basic reactions. the combustion process also includes heat transfer, radiation, convection and diffusion of the chemical components, and also anisotropy due to buoyancy. Since the chemical reaction rate is very fast, the time-scale is much smaller than that of the fluid transport. . The finite rate combustion model has a stiffness problem on the time scale. Since the chemical reaction rate is generally very fast, the time interval must be very short in using the explicit difference method. A much longer spatial increment can be used for the flow process. The explicit method cannot be adopted generally. The implicit approach has to be used. But this would make the diffusion of fuel difficult, and in turn greatly delay the combustion process.

2.2 Diffusion (non-premixed) combustion The rate of mixing that controls the combustion rate. 2.2.1 Laminar diffusion flames Chemical reactions take place in proper ratio of fuel and oxygen on the surface of the flame. The flows of fuel and oxygen are one-dimensional flows with uniform and same velocity. Mole and pressure are not changed throughout the whole process. The diffusion of fuel and oxygen in inert gases is regarded as the diffusion of two components, their diffusion coefficients are equal. The product density and diffusion coefficient of mixed gases is not related to temperature and is constant in radial direction. steady-state diffusion combustion process is considered. Reaction is faster than mixing process. 2.2.2 Fundamental Aspects of Laminar Diffusion Flames with Fast Chemistry (OneStep Reaction) Reaction time is much shorter than the mixing time. If the reaction is reversible, there will be both fuel and oxidizer at stoichiometric numbers close to the mixture fraction. Under the assumption of fast reaction, both the forward and backward reactions are faster than the mixing process, so that the mixed chemical components can attain dynamic equilibrium. Favre Probability Density Function (PDF) is often used to integrate the spatial distribution of mixture fraction f. This spatial distribution also includes the effect of turbulent fluctuations. 2.2.3 Simple chemical reacting system (SCRS) and mixture fraction Fuel and oxidizer will undergo a series of reactions to give combustion products. For example, the combustion of the simplest hydrocarbon methane CH 4 involves more than 40 basic dynamic reactions. The production and dissipation rate of the component in component transport equation and radiation loss or increase, pressure work and chemical energy in enthalpy equation should be considered and energy consumption due to viscosity can be neglected. Temperature can be calculated by calorific value. A simple chemically reacting system (SCRS) is adopted where the reaction rates are very fast compared with the mixing rates.Char oxidizes to CO at the particle surface; CO oxidizes to CO2 in the bulk gas. 2.3 premixed turbulent combustion 2.3.1 eddy break-up (EBU) model and time-averaged reaction rate of turbulent flow

EBU model is the earliest model used in turbulent premixed combustion, then for nonpremixed flames .It is based directly on a physical assumption on the turbulent reaction rate. As the reaction rate in laminar flow depends on the horizontal mixing, whereas the reaction rate in turbulent flow depends on the mixing of the turbulent eddies. The reaction rate is dependent to temperature and is calculated with Arrhenius equation for both laminar and turbulent state When the reaction consists of two components, fuel F and oxidizer O, and the reaction is assumed as a second-order reaction. The relationship between concentration and temperature pulsations, and the temperature pulsations itself have enhanced and increased the time-averaged reaction rate. Not only folding occurred, the surface was also broken into pieces of different sizes to expand the flame surface. Therefore, more combustible mixed gases were burnt per unit time. Although EBU model has already been widely applied, its chemical reaction dynamic process is still not very satisfactory. For example, some problems concerning the CO species cannot be represented by fast chemistry. 2.3.2 Eddy dissipation model the concept of EBU was extended to eddy dissipation model (EDM). The eddy dissipation model is best applied to turbulent flows when the chemical reaction rate is fast relative to the transport processes in the flow. There is no kinetic control of the reaction process. Thus, ignition and processes where chemical kinetics may limit reaction rate may be poorly predicted. In the EBU model proposed by Spalding, the reaction rate (as equation (3-4)) is only related to the turbulent dissipation rate of the premixed gas. Whereas in the EDM model, the reaction rate is related to the minimum value of turbulent dissipation rate of the premixed gas, oxidizer and combustion products with higher temperature.

2.3.3. Expansion of the EBU model It is assumed in the expanded EBU model that, in premixed reaction flow where both Arrhenius reaction dynamic mechanism and turbulent mixing mechanism have an effect. Because of lack of understanding of the reaction mechanism, when reaction time is larger than turbulent mixing time, average reaction rate is directly related to average Arrhenius rate, and when the turbulent mixing time is larger than reaction time, average reaction rate is directly related to turbulent rate. 2.3.4. simplified PDF model of fast reaction flow The reaction rate is much faster than turbulent mixing. The reaction flow is controlled by the comparatively slower diffusion process.

The basic assumption in diffusion combustion is that fuel and oxygen would not exist at the same time or in the same space. Considering equations in oxygen side which contains oxygen and combustion products with no fuel and equations in region near to the fuel which there is no oxygen,, and the assumption that fuel and oxygen composition on the flame surface are equal to zero model would be drived. 2.3.5 Flame surface model

The model is only applicable for two-feed system( fuel and oxidizer). Average temperature and composition of fuel and oxidizer can be calculated via instantaneous equations for each species and time-averaged equation for f. results show on the flame surface isolines. However, the results obtained are still too simplified and not very satisfactory. The reason is that an infinitely thin flame surface with zero concentration of fuel and oxygen does not exist in real turbulent diffusion flames. But on the flame surface with a finite thickness, the time-averaged values of fuel and oxygen concentration coexist in the same space. 2.3.6 Model of Partial Instantaneous Non-Mixing

The model describes the internal state of turbulent eddies. Rather than all above model equations and time-averaged equation for f, In addition, mean square equation of the mixture fraction pulsations has to be solved. There is a special problem about finding time averaged composition of species. results indicated that the equation sets combined with the simplified PDF agreed

better with the experimental data, when compared with the flame surface model. The cut-off Gauss distribution PDF model is slightly better than step functions pulsations PDF model, and the differences are very little. The diffusion flame mechanism of this model is similar to the dynamic equilibrium model. However, there are limitations in applying this kind of partial dynamic equilibrium model. The reaction might be too fast in practice to get a partial dynamic equilibrium.

2.3.7 Simplified PDF Assumptions of Finite Reaction Rate For more general situations, instantaneous laminar reaction is assumed within the turbulent eddies. Reaction rate is finitely fast and not under partial dynamic equilibrium state. Time averaged equations for species, mixture fraction, mean square mixture fraction and reaction rate are considered. Assuming both mixture fraction and mean square mixture fraction are the function of spatial location.

Solving steps are: 1. Solve for the continuity equation, momentum equation and the turbulent time-averaged flow to get the preliminary spatial distribution of 2. Calculate from the values of equations for the and g; and

and g, and the given values of

3. Put w s into the conservation equation (3-37) of species s to solve for Y s ; 4. Solve the conservation equation of enthalpy h; 5. Calculate T from h and Y s; 6. Solve again by including the effect of the estimated T on density variation, on the momentum equation and the equations;

7. Repeat the above steps until convergence.

2.3.8. Simplified PDF Model of Premixed Reaction Flow Controlled by Both Diffusion and Chemical Reaction) the reaction flow is controlled by both diffusion and chemical reaction.

(Reaction

Flow

The simplified PDF model was developed for studying premixed reaction flow. Assuming that the reaction is a one-step second-order reaction. 1. By assuming that inside each turbulent eddy is an instantaneous adiabatic laminar flamelet and PDF function of density, instantaneous temperature and compositions equal to a function of reaction level leads to the calculation of time-averaged reaction rate. 2. Relationship between the time-averaged reaction level and other time-averaged quantities From the energy equation and conservation equation of either instantaneous component Y F or Y o , the equation of the instantaneous reaction level can be obtained. 3. In solving the equations, apart from solving the continuity equation, momentum equation, and the k and equations, equations have to be solved jointly. To solve for these five equations, the method of trial and error has to be used. Statistical fluid dynamics and probability concept are applied to derive turbulent combustion. The calculations involved are too complicated, for engineering application. 2.3.9. Brief Introduction of Laminar Flamelet Model Laminar flamelet combustion model is different from SCRS in the way that it does not require a linear relationship between the mixture fraction, mass fraction and temperature. More complicated experimental correlation relationships between these variables are

considered. From the laminar flamelet relation, the species mass fraction can be derived. The mixture fraction of the transport equation can be solved. the concentration, temperature, soot concentration, viscosity, enthalpy and pressure of all species are all functions of the mixture fraction in the flamelet. A simple variation diagram of combustion products, oxygen and fuel concentration can be derived as functions of the mixture fraction. In the laminar flamelet model, it is assumed that chemical reactions only take place within the basic element in the laminar flame, that is, exists near the fire source If turbulent flame is believed to be composed of many tiny parts, then the concept of laminar flamelet can be applied to turbulent flames such as fires. Each tiny part is composed of undisturbed laminar diffusion flames. The chemical reaction time is shorter than other turbulent and transport processes. As an example of using this model, starting from the oxygen side, the mixture fraction is zero. While moving towards the center of the flamelet with increasing temperature, a certain quantity of combustion products including the components is taken from the fuel to determine the concentration of carbon and hydrogen atoms, or other related molecules such as O, CO 2 and free radicals H and HO 2 at that point. 2.4 Partially premixed combustion 2.4.1. Thickened Flame Models Thickened flame models for LES solve transport equations for chemical species, and use Arrhenius rate expressions to describe species reactions. resolving realistic flame structures is not affordable in LES, thickened flame models artificially broaden flame structures to ensure convergence of the species equations. The artificial flame broadening is accomplished by multiplying the diffusive terms in the scalar transport equations by a thickening factor, F, that may be as large as ten or twenty. This model has two perspectives: 1. One perspective, for example, would be that the approach is regime independent. This perspective would stem from the idea that no information about flame asymptotics is required in the implementation. It would suggest that the approach is very general, and it has indeed been applied in a variety of simulations where partially premixed combustion would be expected. 2. A second perspective, however, would be that this approach changes the nature of turbulence and chemistry interactions by reducing the time scale for transport and increasing the time scale for chemistry. This second perspective, then, emphasizes the idea that small-scale transport and chemistry processes are critically important, and that altering the associated time scales leads to significant errors in predictions.

2.4.2. Linear Eddy Models (LEM) Linear Eddy Models (LEMs) attempt to explicitly solve for a reduced representation of subfilter turbulence/chemistry interaction. Arrhenius rates are used to describe the subfilter chemical source terms, and once the 1-D solutions have been solved. Just as with thickened flame models, one of the particular advantages of LEM models is that they do not require a priori assumptions about the burning regime. Unlike thickened flame models, however, the 1-D subfilter meshes are capable of resolving and accounting for how transport and chemistry balance on the smallest scales. The LEM model therefore has the advantage of representing the multi-scale nature of combustion processes. In spite of its ability to capture the coupling between diffusion and reaction on the small scales, quantities such as the stirring frequency in the model are empirically determined. because chemistry is locally and explicitly solved for, LEM models for combustion can be relatively costly. While this empiricism has resulted in good agreement with experimental predictions in realistic combustor settings, its performance has not yet been fully characterized.

2.4.3. Conditional Moment Closure Models Conditional Moment Closure (CMC) approaches attempt to describe chemistry by solving for values of chemical species that have been conditioned on a particularly relevant scalar. An approach that is closely related to CMC is Conditional Source term Estimation (CSE). In this approach, independent variables are not added to the problems dimensionality or to the computational mesh. This approach requires solving transport equations for several reactive quantities, so that there are an adequate number of mean or filtered scalar values to accomplish the inversion at a given mesh point. If CSE were to be employed for partially premixed modeling, the nature of the inversion could grow considerably more complex due to the conditioning on progress variable as well as mixture fraction. Furthermore, separate premixed and non-premixed solutions might need to be used to deal with the ill conditioning along the separate directions. Another consideration of importance in a CSE method for partially premixed combustion LES is the issue of flame structure resolution. Reactive scalar equations need to be transported in these methods, but numerical errors are likely to contaminate these solutions if they are solved using traditional transport schemes.

2.4.4. Transported FDF Models Transported FDF models for LES (and PDF models for RANS) attempt to describe chemical reactions by explicitly solving for subfilter distribution functions. FDF approaches use a transport equation for the filtered density function (FDF) as a starting point. If a very accurate mixing model were in place, transported FDF approaches would be able to fully characterize partially premixed combustion, regardless of whether or not the typical mixture fraction and progress variable coordinates are aligned in any way. Indeed, the approach would then be fully closed and would not be subject to any a priori assumptions about the regime. Although the transported FDF approach is therefore very promising and powerful, the difficulties associated with describing mixing temper its current applicability. It may be possible to argue that the advantages of fully closed chemical source terms and detailed chemistry outweigh the inaccuracies of the mixing model in partially premixed flows. Conversely, errors in the description of mixing in a flow may overwhelm any advantages gained by the access to many chemistry realizations. 2.4.5. Flamelet Models These models attempt to describe the subfilter evolution of chemistry by mapping combinations of 1-D pre-computed asymptotic flame solutions into a 3-D flow field. Three advantages of the flamelet model have led to its relatively widespread consideration and study. These advantages include the methods minimal computational cost, the methods ability to consider arbitrarily detailed chemistry in the flamelet space, and the methods multi-scale nature that is based on the correct prediction of the balance that exists between chemistry and transport on the smallest scales.