Predicting Real Fuel Behavior

in IC Engine Simulations

March 14, 2010

REACTION DESIGN

www.reactiondesign.com

[phone]

Reaction Design

1

Seemingly small differences in how chemistry mechanisms are reduced can greatly affect 

the accuracy of results. 

The process by which fuel ignites and burns can be modeled effectively using a detailed chemical
mechanism of the fuel. Detailed mechanisms describe the thousands of short-lived species and chemical
reactions that dictate how a fuel ignites, how the flame propagates, and how emissions like NOx, CO, and
soot are formed. It is impractical to run detailed chemical mechanisms in available CFD packages because
of the complexity of the equations that need to be solved for each time step in a simulation. To attempt to
overcome this limitation, a common practice has been to severely reduce the number of chemical species
from the detailed mechanisms, and therefore the number of reactions that is actually solved for during a
CFD simulation run. Historically, only mechanisms with fewer than 50 species were thought to be
practical for use in CFD simulations. But a great deal of accuracy and predictive capability is lost with
severe mechanism reductions, which is driving the industry to pursue more accurate chemistry through
the use of High Performance Computing (HPC), chemistry lookup tables, and other approaches.

It is important to understand that elimination of just a few species from an accurate mechanism can
dramatically affect key simulation results. We’ll use the example of a kinetics simulation of a
fundamental experiment for measuring ignition delay, using CHEMKIN-PRO. We start with what many
in industry might consider a detailed mechanism with 102 species. This mechanism accurately predicts
ignition delay as can be seen in Figure 1. However, by removing just two species through mechanism
reduction, the mechanism no longer accurately predicts ignition delay. Similar behavior can be seen
when using reduced mechanisms targeted at flame speed calculations or pollutant formation predictions.

The impact of improperly reduced mechanisms on CFD calculations of emissions can be seen in Figure 2.
This figure shows emissions results for NOx, CO and unburned hydrocarbons using a popular, reduced
mechanism with 118 species. Because this mechanism is larger than mechanisms traditionally used in
CFD calculations, it is sometimes thought to be a detailed chemical mechanism and accurate enough to
predict emissions. However, the results indicate that not only does the mechanism fail to predict key
emissions values; it also fails to predict the trends accurately.

Often designers who are not kinetics experts do not recognize the sacrifices in accuracy that may result
from removing chemistry details from the mechanism to make it viable in CFD. They know that
capturing ignition behavior is critical to simulation effectiveness, so when results do not agree with data,
attention is often focused on compensating through calibration of turbulence models, spray models, or
turbulence-kinetics interaction models, while the real culprit may be inadequate detail in the chemical
mechanism. Further, since many designers don’t believe they have the option of incorporating more

Reaction Design

2

accurate chemical mechanisms, they typically do not contemplate the use of a larger, detailed mechanism
as an alternative.

Figure 1: Removing only two species can have a dramatic effect on chemistry accuracy for ignition

delay, as shown in these CHEMKIN ignition-delay simulations.

Figure 2: Dramatic errors in values and trends of emissions predictions with CFD results using what

is widely believed to be a "detailed" mechanism at 118 species.

i

(Triangles are experimental data.)

i

Puduppakkam, K.V., Liang, L., Shelburn, A., Naik, C., Meeks, E., and Bunting, B., “Predicting Emissions Using CFD

Simulations of an E30 Gasoline Surrogate in an HCCI Engine with Detailed Chemical Kinetics,” SAE Paper 2010-01-
0362, 2010.