10.3.2. Flamelet Generated Manifold (FGM) model

The Laminar Flamelet model (see The Diffusion Flamelet Models Theory) postulates that a turbulent flame is an ensemble of laminar flames that have an internal structure not significantly altered by the turbulence. These laminar flamelets are embedded in the turbulent flame brush using statistical averaging. The Flamelet Generated Manifold (FGM) [497] model assumes that the scalar evolution (that is the realized trajectories on the thermochemical manifold) in a turbulent flame can be approximated by the scalar evolution in a laminar flame. Both Laminar Flamelet and FGM parameterize all species and temperature by a few variables, such as mixture-fraction, scalar-dissipation and/or reaction-progress, and solve transport equations for these parameters in a 3D CFD simulation.

Note that the FGM model is fundamentally different from the Laminar Flamelet model. For instance, since Laminar Flamelets are parameterized by strain, the thermochemistry always tends to chemical equilibrium as the strain rate decays towards the outlet of the combustor. In contrast, the FGM model is parameterized by reaction progress and the flame can be fully quenched, for example, by adding dilution air. No assumption of thin and intact flamelets is made by FGM, and the model can theoretically be applied to the stirred reactor limit, as well as to ignition and extinction modeling.

Any type of laminar flame can be used to parameterize an FGM. ANSYS Fluent can either import an FGM calculated in a third-party flamelet code and written in Standard file format (see Standard Files for Flamelet Generated Manifold Modeling in the User's Guide), or calculate an FGM from 1D steady premixed flamelets or 1D diffusion flamelets.

In general, premixed FGMs should be used for turbulent partially-premixed flames that are predominantly premixed. Similarly, diffusion FGMs should be used for turbulent partially-premixed flames that are predominantly non-premixed.

10.3.2.1. Premixed FGMs

While the only possible configuration for a diffusion flamelet in 1D is opposed flow, 1D steady premixed flamelets can have several configurations. These include unstrained adiabatic freely-propagating, unstrained non-adiabatic burner-stabilized, as well as a strained opposed flow premixed flames.

1D premixed flamelets can be solved in physical space (for example, [398], [71]), then transformed to reaction-progress space in the ANSYS Fluent premixed flamelet file format, and imported into ANSYS Fluent. Alternatively 1D premixed flamelets can be generated in ANSYS Fluent, which solves the flamelets in reaction-progress space. The reaction progress variable is defined by Equation 9–2 where the sum is over all species in the chemical mechanism, denotes the species mass fraction, superscript denotes the unburnt reactant at the flame inlet, and superscript denotes chemical equilibrium at the flame outlet. The coefficients should be prescribed so that the reaction progress, , increases monotonically through the flame. By default, for all species other than , which are selected for hydrocarbon combustion. In the case where the element is missing from the chemical mechanism, such as combustion, ANSYS Fluent uses except , by default. The coefficients can be set in the text user interface when generating FGM using the text user interface.

The 1D adiabatic premixed flame equations can be transformed from physical-space to reaction-progress space [345], [288]. Neglecting differential-diffusion, these equations are

(10–3)

(10–4)

where is the species mass fraction, is the temperature, is the fluid density, is time, is the species mass reaction rate, is the total enthalpy, and is the species specific heat at constant pressure.

The scalar-dissipation rate, in Equation 10–3 and Equation 10–4 is defined as

(10–5)

where is the thermal conductivity. Note that varies with and is an input to the equation set. If is taken from a 1D physical-space, adiabatic, equi-diffusivity flamelet calculation, either freely-propagating (unstrained) or opposed-flow (strained), the -space species and temperature distributions would be identical to the physical-space solution. However, is generally not known and is modeled in ANSYS Fluent as

(10–6)

where is a user-specified maximum scalar dissipation within the premixed flamelet, and is the inverse complementary error function.

A 1D premixed flamelet is calculated at a single equivalence ratio, which can be directly related to a corresponding mixture fraction. For partially-premixed combustion, premixed laminar flamelets must be generated over a range of mixture fractions. Premixed flamelets at different mixture fractions have different maximum scalar dissipations, . In ANSYS Fluent, the scalar dissipation at any mixture fraction is modeled as

(10–7)

where indicates stoichiometric mixture fraction.

Hence, the only model input to the premixed flamelet generator in ANSYS Fluent is the scalar dissipation at stoichiometric mixture fraction, . The default value is , which reasonably matches solutions of unstrained (freely propagating) physical-space flamelets for rich, lean, and stoichiometric hydrocarbon and hydrogen flames at standard temperature and pressure.

The following are important points to consider about premixed flamelet generation in ANSYS Fluent in general and when specifying in particular:

  • It is common in the FGM approach to use unstrained, freely-propagating premixed flamelets. However, for highly turbulent flames where the instantaneous premixed flame front is stretched and distorted by the turbulence, a strained premixed flamelet may be a better representation of the manifold. The ANSYS Fluent FGM model allows a premixed flamelet generated manifold at a single, representative strain rate, .

  • Calculating flamelets in physical-space can be compute-intensive and difficult to converge over the entire mixture fraction range, especially with large kinetic mechanisms at the flammability limits. The ANSYS Fluent solution in reaction-progress space is substantially faster and more robust than a corresponding physical-space solution. However, if physical-space solutions (for example, from [398], [71]) are preferred, they can be generated with an external code, transformed to reaction-progress space, and imported in ANSYS Fluent's Standard Flamelet file format.

  • A more appropriate value of than the default (1000/s) for the specific fuel and operating conditions of your combustor can be determined from a physical-space premixed flamelet solution at stoichiometric mixture fraction. This premixed flamelet solution can be unstrained or strained. through the flamelet is then calculated from Equation 10–5 and is the maximum value of .

  • As is increased, the solution of temperature and species fractions as a function of tends to the thin-flamelet chemical equilibrium solution, which is linear between unburnt at and chemical equilibrium at . The minimum specified should be that in a freely-propagating flame. For smaller values of , the ANSYS Fluent premixed flamelet generator will have difficulties converging. When ANSYS Fluent fails to converge steady premixed flamelets at rich mixture fractions, equilibrium thin-flamelet solutions are used.

  • The ANSYS Fluent partially-premixed model can be used to model partially-premixed flames ranging from the non-premixed to the perfectly-premixed limit. In the non-premixed limit, reaction progress is unity everywhere in the domain and all premixed streams are fully burnt. In the premixed limit, the mixture fraction is constant in the domain. When creating a partially-premixed PDF table, the interface requires specification of the fuel and oxidizer compositions and temperatures. For partially-premixed flames, the fuel composition can be modeled as pure fuel, in which case any premixed inlet would be set to the corresponding mixture fraction, which is less than one. Alternatively, the fuel composition can be specified in the interface as a premixed inlet composition, containing both fuel and oxidizer components. In this latter case, the mixture fraction at premixed inlets would be set to one.

10.3.2.2. Diffusion FGMs

For turbulent partially-premixed flames that are predominantly non-premixed, diffusion FGMs are a better representation of the thermochemistry than premixed FGMs. An example of this is modeling emissions from a gas-turbine combustor where the primary combustion zone is quenched by rapid mixing with dilution air. If the outlet equivalence ratio is less than the flammability limit of a corresponding premixed flamelet, the premixed FGM will predict sub-equilibrium , even if the combustor is quenched (). A diffusion FGM, however, will better predict super-equilibrium for .

Diffusion FGMs are calculated in ANSYS Fluent using the diffusion laminar flamelet generator as detailed in Flamelet Generation. Steady diffusion flamelets are generated over a range of scalar dissipation rates by starting from a very small strain (0.01/s by default) and increasing this in increments (5/s by default) until the flamelet extinguishes. The diffusion FGM is calculated from the steady diffusion laminar flamelets by converting the flamelet species fields to reaction progress, (see Equation 10–2). As the strain rate increases, the flamelet chemistry departs further from chemical equilibrium and decreases from unity towards the extinction reaction, .

The FGM between , and the unburnt state , is determined from the thermochemical states of the final, extinguishing diffusion flamelet.


Release 18.2 - © ANSYS, Inc. All rights reserved.