Atmospheric chemistry mechanisms are the most computationally intensiv
e components of photochemical air quality simulation models (PAQSMs).
The development of a photochemical mechanism, that accurately describe
s atmospheric chemistry while being computationally efficient for use
in PAQSMs, is a difficult undertaking that has traditionally been purs
ued through semiempirical (''diagnostic'') lumping approaches. The lim
itations of these diagnostic approaches are often associated with inac
curacies due to the fact that the lumped mechanisms have typically bee
n optimized to fit the concentration profile of a specific species. Fo
rmal mathematical methods for model reduction have the potential (demo
nstrated through past applications in other areas)to provide very effe
ctive solutions to the need for computational efficiency combined with
accuracy. Such methods, that can be used to ''condense'' a chemical m
echanism, include ''kinetic lumping'' and ''domain separation''. An ap
plication of the kinetic lumping method, using the direct constrained
approximate lumping (DCAL) approach, to the atmospheric photochemistry
of alkanes is presented in this work. It is shown that the lumped mec
hanism generated through the application of the DCAL method has the po
tential to overcome the limitations of existing semiempirical approach
es, especially in relation to the consistent and accurate calculation
of the time-concentration profiles of multiple species.