A new method for evaluating kinetic flame models, called net rate anal
ysis, was developed from existing mechanism screening techniques and a
modified form of reaction path analysis. In net rate analysis, the co
ntributions of individual reactions are gauged by comparing rates of r
eaction computed using experimental mole fractions, rather than mole f
ractions produced by a flame code simulation. The sum of these contrib
utions for a given species-the data-based predicted net rate-is compar
ed to the experimental net rate derived from measured mole fractions u
sing the laminar flame equations. Various features of a model's struct
ure and its predictions, such as interdependencies of submechanisms an
d net rates, can be revealed with this technique. Measured profiles of
mole fractions and temperature in flat, laminar, premixed H-2/O-2/C6H
6/Ar and C6H6/O-2/Ar flames were used to evaluate the new technique ag
ainst conventional reaction path analysis or mole fraction comparisons
. In some cases, the conventional methods appear to identify the exist
ence of a problem with the chemistry of a particular species when, in
fact, the chemistry is correct. Propagation of errors in the predicted
mole fraction of other species to the species under consideration are
the source of the discrepancy. Also observed in the use of the conven
tional analyses are cases where deficiencies in the chemistry involvin
g a particular species are offset by inaccurate predictions of other s
pecies, thus leading to the incorrect conclusion that the proposed che
mistry is sound. Net rate analysis is shown to lead to more realistic
conclusions than techniques based on the use of computed mole fraction
s because the net rate analysis eliminates the propagation of error fo
und in predicted mole fractions. (C) 1998 by The Combustion Institute.