COMBUSTION SIMULATION

Combustion models for CFD refers to combustion models for computational fluid
dynamics. Combustion is defined as a chemical reaction in which a hydrocarbon fuel
reacts with an oxidant to form products, accompanied with the release of energy in
the form of heat. Being the integral part of various engineering applications
like: internal combustion engines, aircraft engines, rocket engines, furnaces,
and power station combustors, combustion manifests itself as a wide domain during
the design, analysis and performance characteristics stages of the above-mentioned
applications. With the added complexity of chemical kinetics and achieving reacting
flow mixture environment, proper modeling physics has to be incorporated
during computational fluid dynamic (CFD) simulations of combustion. Hence the
following discussion presents a general outline of the various adequate models
incorporated with the Computational fluid dynamic code to model the process of
combustion.

Todays combustion system design objectives focus on efficiency, pollutant emissions, and fuel
flexibility. Combustion simulation can be a valuable aid to designers in meeting their goals but
only if the results of their modeling can give true insight into their engines behavior. Obtaining
accurate results from combustion simulation requires the capture of both the physical and
chemical characteristics that can change radically over a complete engine duty cycle. In an
internal combustion engine, for example, spray breakup and evaporation, turbulence, ignition
delay and flame propagation are all factors that must be modeled accurately to yield meaningful
results. Inaccuracies can sometimes be overcome by a substantial amount of tuning or
calibration of the simulation models, but the cost of calibration is usually significant and the
resulting calibrations are typically not transferable over the course of a designs evolution.
Thanks to massively parallel computers, engine geometries can now be represented with
amazing detail using computational meshes in CFD that approach 100 million cells. Advanced
turbulence modeling in CFD provides better simulation of turbulence-chemistry interactions,
albeit with severely reduced chemistry. But because the chemistry solver technology included
with most CFD packages is slow relative to the flow calculations, it is common for engineers to
use single-component, severely reduced fuel models (or mechanisms) in their combustion
simulations. These severely reduced fuel models lack the detail that is required to accurately
predict key engine performance factors such as ignition delay, flame propagation, NOx, CO and
PM (soot) emissions. Through the work of the Model Fuels Consortium (MFC), it was clearly

shown that the use of severely reduced fuel models can compromise the accuracy of important
parameters in a simulation.

In the figure below, the results of a simulation using a 34-species n-heptane fuel model that is
commonly used to model diesel combustion are shown. This fuel model was extensively tuned
to provide reasonable accuracy in standard diesel engine designs. However, it fails to accurately
predict trends for important combustion and emissions performance when compared to actual
engine measurements for high EGR cases. The more accurate 174 species mechanism is able
to accurately capture ignition and heat release for the high EGR cases.

The impacts of using oversimplified chemistry can be significant. Inaccurate results

require significant calibration or tuning such that the overall Time-to-Solution of a
combustion simulation is long and the results are not transferrable to new or
modified designs. Through the work of the MFC, the combustion simulation
community now has access to the most widely validated library of fuel model
components available. When coupled with Reaction Designs software suite, these
models have been shown to greatly increase the accuracy of results across a wide
range of operating conditions and fuel types without negatively impacting Time-toSolution.

Critical Considerations of Combustion

Phenomena
The major consideration during any general combustion process includes the mixing time scale
and the reacting time scale elapsed for the process. The flame type and the type of mixing of
flow streams of the constituents also have to be taken into account. Apart from that as far as the
kinetic complexity of the reaction is concerned, the reaction proceeds in multiple steps and what
appears as a simple one line reaction actually completes after a series of reactions. Also the
transport equations for mass fractions of all the species as well as the enthalpy generated
during the reaction have to be solved. Hence even the simplest combustion reaction involves
very tedious and rigorous calculation if all the intermediate steps of the combustion process, all
transport equations and all flow equations have to be satisfied simultaneously. All these factors
will have a significant effect on the computational speed and time of the simulation. But with
proper simplifying assumptions Computational fluid dynamic modeling of combustion reaction
can be done without substantial compromise on the accuracy and convergence of the solution.
The basic models used for the same are covered in the following paragraphs.

Simple Chemical Reacting System Model

This model takes into consideration only the final concentration of species and takes into
account only the global nature of combustion process where the reaction proceeds infinitely fast
as a single step process without much stress on the detailed kinetics involved. The reactants
are assumed to react in stoichiometric proportions. The model also deduces a linear relationship
between the mass fractions of fuel, oxidant and the non-dimensional variable mixture
fraction. The model also takes into account an additional assumption that the mass diffusion

coefficients of all species are equal. Owing to this additional assumption the model only solves
one extra partial differential equation for mixture fraction and after solving the transport equation
for the mixture fraction the corresponding mass fractions for fuel and oxidant are calculated.
This model can very well be applied to a combustion environment where laminar diffusion
effects are dominant and the combustion proceeds via non premixed fuel and oxidant streams
diffusing into each other giving rise to a laminar flame.

Eddy Break-up Model

Simulation of can combustor using eddy dissipation model.

This model is used when turbulent mixing of the constituents has to be taken into consideration.
The k/ turbulent time scale is used to calculate the reaction rate. A comparison between the
turbulent dissipation rates of the fuel, oxidant and products is done and the minimum amongst
all is taken as the rate of the reaction. The transport equations for the mass fractions of the

constituents are solved using this rate of reaction. Apart from this a mean enthalpy equation is
also solved and temperature, density and viscosity are calculated accordingly. The model can
also be implemented when finite rate kinetically controlled reaction is to be simulated. In such
situation while deciding the rate of the reaction the Arrhenius kinetic rate expression is also
taken into account and the rate of reaction is taken as minimum amongst the turbulent
dissipation rates of all the constituents and the Arrhenius kinetic rate expression. Since turbulent
mixing governs the characteristics of this model, there exists a limit to the quality of the
combustion simulation depending upon the type of the turbulent model implemented to
represent the flow. The model can also be modified to account for mixing of fine structures
during the turbulent reaction. This modification of the model results in the eddy dissipation
model which consider the mass fraction of fine structures in its calculations.

Laminar Flamelet Model

This model approximates the turbulent flame as a series of laminar flamelet regions
concentrated just around the stoichiometric surfaces of the reacting mixture. This model exploits
the use of experimental data for determining relations between the variables considered like
mass fraction, temperature etc. The nature and type of dependence of the variables is predicted
through experimental data obtained during laminar diffusion flame experiment and laminar
flamelet relationship is deduced based on the same. These relationships are then used to solve
the transport equations for species mass fraction and mixture composition. The model can very
well be implemented for situations where concentration of minor species in the combustion is to
be computed like quantifying the generation of pollutants. A simple enhancement to the model
results in the flamelet time scale model which takes finite rate kinetics effect into consideration.
The flamelet time scale model produces steady laminar flamelet solution when reaction
proceeds very fast and captures the finite rate effects when reaction chemistry is dominant.

Presumed Probability Distribution

Function Model
This model considers a statistical approach for calculating the variables like species mass
fractions, temperature and density while the mixture composition is calculated at the grids. Then
these all variables are calculated as functions of the mixture fraction around a presumed
probability distribution function. The model can produce satisfactory results for turbulent reactive
flows where convection effects due to mean and fluctuating components of velocity are
dominant. The model can be extended for adiabatic as well as non-adiabatic conditions.

Conditional Moment Closure

Conditional moment closure (CMC) is an advanced combustion model. The basic idea is to
model the chemical source based on conditional averages. The model was first introduced for
non-premixed flows and hence the conditioning is done in the mixture fraction.