The reaction progress variable is defined as a normalized fraction
of product species (see Equation 9–2),
namely . Note that the denominator,
, is only a function of the local mixture fraction.
When the Flamelet Generated Manifold model is
enabled, ANSYS Fluent solves a transport equation for the un-normalized
progress variable,
, and not the normalized progress variable,
. This has two advantages. Firstly,
since there are usually no products in the oxidizer stream,
is zero and
is undefined here (in other words, burnt air is
the same as unburnt air). This can lead to difficulties in specifying
oxidizer boundary conditions where oxidizer is mixed into unburnt
reactants before the flame, as well as into burnt products behind
the flame. Solving for
avoids these issues and the solution is independent of the specified
boundary value of
for pure oxidizer inlets. The second advantage of
solving for
is that flame quenching can be modeled naturally. Consider a burnt
stream in chemical equilibrium (
) that is rapidly quenched with
an air jet. Since the
equation in ANSYS Fluent does not have a source term
dependent on changes in mixture fraction,
remains at unity and hence the
diluted mixture remains at chemical equilibrium. Solving for
can capture quenching since
changes with mixture fraction, and the normalized
reaction progress
is correctly less than unity after mixing. To model
these two effects with the normalized
equation, additional terms involving
derivatives and cross-derivatives of mixture fraction are required [55]. These terms do not appear in the transport equation
for
:
(10–8) |
where is laminar thermal conductivity of the mixture,
is the mixture specific heat.
Typically, in the FGM model, the mean source term is modeled as:
(10–9) |
where is the Finite-Rate flamelet source term from the flamelet library, and
is the joint PDF of reaction-progress (
) and mixture fraction (
).
The source term determines
the turbulent flame position. Errors in both the approximation of
the
variance, as well as the assumed shape Beta PDF, can cause inaccurate
flame positions. ANSYS Fluent has the option of two other turbulence-chemistry
interaction models for the source term
. One option is to use a turbulent flame speed,
(10–10) |
which is essentially the same source term when using the chemical equilibrium and diffusion flamelet partially-premixed models. An advantage of using a turbulent flame speed is that model constants can be calibrated to predict the correct flame position. In contrast, there are no direct parameters to control the FGM finite-rate source (Equation 10–9), and hence the flame position.
The third option is to use the minimum of the Finite-Rate and Turbulent-Flame-Speed expressions:
(10–11) |
The idea behind Equation 10–11 is to use the turbulent flame speed model to predict the flame location and the finite-rate model to predict post-flame quenching, for example, by dilution with cold air.
The joint PDF, in Equation 10–9, is specified as the product of
two beta PDFs. The beta PDFs require second moments (that is, variances).
The variance of the un-normalized reaction progress variable is modeled
either with a transport equation
(10–12) |
where ,
or with an algebraic expression
(10–13) |
where is
the turbulence length scale and
is a constant with a default value of 0.1.
In the FGM model, the species concentration is obtained from reduced variables, such as the mixture fraction, progress variable, and their variances. Since the definition of the progress variable is predominantly based on the global species, the progress variable tracks the evolution of the major species and heat release reactions. In a reacting mixture, however, not all species evolve at the same rate due to the differences in their time scales. For example, the global progress variable may approach the unity, which marks the completion of a reaction. Meanwhile, a slow forming species, such as nitric oxide NO, may still be evolving. Therefore, mapping that is performed solely on the basis of the global progress variable may lead to inaccurate results for the slow-forming species.
The progress variable is a reduced-order modeling approach that is based on a representative time scale of a reacting mixture. In reality, different species within a reacting mixture may have large disparities in their reaction time scales. Reduced-order models cannot capture these disparities. Solving for multiple separate transport equations for slow-forming species is one of the alternative approaches that will account for differences in the time scales of species evolution. Within the framework of the FGM model, ANSYS Fluent provides an option to solve generic transport scalar equations, in addition to the global progress variable equation. These generic scalar transport equations can be solved for minor species that are low in concentration and, therefore, have no feedback on the main flow field, and that have time scales much slower than those of the major species in the mixture.
The transport equation of a generic scalar solved with the FGM model in ANSYS Fluent has the following form:
(10–14) |
In this equation, the scalar that is to be transported can be a mass fraction of any species in the mixture. The last term is the reaction source term. It represents the net production rate of the scalar
and is computed as the difference between its forward and reverse reaction rates:
(10–15) |
The forward reaction rates of the scalar are not dependent on the value of
. Therefore, they can be pre-tabulated and stored in the PDF table. On the other hand, the reverse reaction rates of the scalar are dependent on the concentration of
. Therefore, the pre-tabulated values for the reverse reaction rates need to be corrected based on the solved transport scalar equation [197].
In ANSYS Fluent, the reverse reaction rates are first calculated using the prevailing values of the scalar , and then they are stored in the PDF table. At the run time, the reverse reaction rates are scaled using the value of the solved transport scalar
in the following way:
(10–16) |
where is an averaged value of the transported scalar stored in the PDF table. This methodology allows the consumption of the scalar
to respond to the slow evolution of the scalar solved.
The scalar transport model assumes that all the species, except for those solved as scalars, will occur over relatively short time scales. The scalar solved in Equation 10–14 is treated as a passive scalar, with no impact on the flow field, temperature, or mixture properties. Therefore, the scalar transport option should be used only for species that are low in concentration and that also have a slow formation rate. A typical example of such a species (for which a mass fraction can be solved as a scalar) is nitric oxide (NO). In the case of multiple scalar equations, the scalar equation for each species is solved independently, and the concentration of the remaining species is taken from the PDF tables.
Note: The scalar equations are treated as passive scalars that do not provide feedback to the flow equations or mixture properties. Therefore, for steady-state simulations, these equations can be solved in a post-processing mode, as for example, when modeling pollutant formation in ANSYS Fluent.