The reactor network model is used to simulate species and temperature fields in a combustor using a detailed chemical kinetic mechanism. The reactor network is constructed from a converged Ansys Fluent simulation modeled with a fast chemistry combustion model, such as the Non-Premixed, Partially-Premixed, or Eddy-Dissipation model. A full chemical mechanism in CHEMKIN format can be imported into Fluent and solved on the reactor network. The combustor volume is automatically subdivided into a small number of connected perfectly stirred reactors. The mass fluxes through the network are determined from the CFD solution, and the species and temperatures in the reactor network are solved together. Hence, the reactor network model is used to simulate finite-rate chemistry effects with detailed kinetic mechanisms, in particular pollutant emissions such as NOx, CO, and unburnt hydro-carbons. Since the specified number of stirred reactors is much smaller than the number of CFD cells, the reactor network model allows much faster simulations of the species and temperature fields than solving for detailed chemistry in every cell, as in the Laminar, EDC, and PDF Transport models.
Typically, the reactor network model is executed on a converged steady-state RANS solution or a time-averaged unsteady solution. The model can also be run on an unsteady flow representing a “snap-shot” in time. Since there is no backward coupling of the reactor-network solution to the flow, the model is useful for predictions where the detailed chemistry has little impact on the flow, such as pollutant formation. The model is inappropriate for highly unsteady flows such as flame ignition or global extinction and also for flows that are strongly influenced by the chemistry, such as soot with significant soot-radiation interaction.
For more information about using reactor network, see Reactor Network Model in the Fluent User's Guide.
The first step in a reactor network simulation involves agglomerating
the CFD cells into the specified number of reactors, 
.
Since each reactor is a perfectly mixed representation of a region of the combustor, ideally cells that are closest in composition space should be grouped together. For optimal performance, the CFD cells grouped together in each reactor should have temperatures and species mass fractions that are as similar as possible.
By default, for Non-Premixed and Partially-Premixed cases, Fluent groups cells that have similar temperatures and mixture fractions, and for Species Transport cases, Fluent groups cells that have similar temperatures and mass fractions of N2 and H2O. These defaults should provide good cell clustering in most cases. However, Fluent allows user-controlled clustering through custom-field functions when these defaults are not sufficient.
After similar cells have been clustered, Fluent splits non-contiguous
groups, then agglomerates clusters with the smallest number of cells
to their closest neighbors until the specified number of reactors, 
, is obtained.
The second step in a reactor network simulation involves the solution of the reactor network, which proceeds as follows.
The mass flux matrix, 
, is calculated from the cell fluxes in the CFD solution, where each
matrix component 
is the mass flux from reactor 
to reactor 
. The 
th species
mass fraction in reactor 
, 
, is governed by the algebraic equation:
 | (7–146) |
where 
is the volume of reactor 
, 
is the 
th species reaction rate
in the 
th reactor, and 
is a mass source term. 
is the net mass flux into reactor 
and is calculated as:
 | (7–147) |
The mass source term, 
, accounts for contributions from the CFD inlet species
mass fluxes through the CFD boundaries and from volume sources, such
as the Discrete Phase Model (DPM).
The system of equations for 
, Equation 7–146, has the dimension of 
, where 
is the user-specified number of reactors,
and 
is the number of species in the chemical
mechanism. Ansys Fluent solves this system with a segregated algorithm
by default, although the option to use a fully coupled algorithm is
available.
An energy equation is not solved in the reactor network. Instead, by default, the temperature in each reactor is calculated from the equation of state. The reactor pressure is fixed and determined as the mass-averaged pressure of the CFD cells in the reactor. Note that the density of each reactor is fixed since both the reactor volume and the reactor mass are constant. This approach ensures that heat loss (or gain) in the CFD simulation is appropriately accounted for in the reactor network. Fluent also provides an option to not solve for temperature, in which case the temperature is fixed as the mass-average temperature of the CFD cells in the reactor.