A novel approach to kinetic mechanism reduction

L.M.T. Joelssona), C. Pichlera), and E.J.K. Nilssona)

a) Department of Physics, Lund University, Professorsgatan 1, Lund, Sweden

Chemical kinetic mechanisms is a tool to predict the concentrations of chemical species in a chemical domain. Such predictions could be useful for e.g. health hazard assessments in the urban environment. The complicated nature of organic chemistry and the large range of volatile organic carbons present in the troposphere makes an explicit mechanism of tropospheric chemistry inapplicable. The complexity of a certain chemical regime depend on the characteristics of the system (time scales, temperature, humidity, solar influx, and background concentration levels and emission rates of trace gases etc.) and the scope of the investigation allowing some extent of reduction of the mechanism. The primary challenge in choosing a reduced mechanism is the need to provide adequate accuracy in target compound predictions without the burden of superfluous computational costs. A new method to construct tailored chemical kinetic mechanisms for any atmospheric chemistry situation will be presented. Its performance was evaluated by applying the method to a number of idealized test cases previously described in the literature [1]. In addition, the methods sensitivity towards NMVOC to NOx ratio, time scales, and the choice of target compound set was investigated. The reaction type composition of the resulting mechanisms was examined.

The mechanism proved to successfully produce substantially reduced mechanisms (10 %–30 % of the relevant subset of the near-explicit chemical mechanism) with typically less than a 10 % loss in accuracy for the target compound set (O3, NO, NO3, NO3, OH, and PAN). The presence of isoprene in a system requires higher complexity in the reduced mechanisms.

Longer simulated time scales allows the chemistry to evolve further, especially hydrocarbon oxidation, which require the mechanisms to include the chemistry of more oxidation products. This proved particularly important for reactions on the form RO2 + NO → ROOH + NO2. Furthermore, mechanisms tuned to day time only does not contain as much oxidized nitrogen chemistry as a mechanism with simulated time spanning both night and day.

There is a simple monotone increase in number of reactions as the number of target compound increase, since each compound has a number of sink- and source reactions associated with it and a successful prediction of the compound need to include at least some of these. Also, compounds that are more intricately related to other compounds require more reactions. The dropping of NO3, OH, and PAN did cut the selected reduced mechanism for the test case by 30 %.

Finally, the NMVOC to NOx ratio impacted the process heavily. A case with low ratio did not require as detailed description of the HOx- and VOC chemistry, typically rendering in smaller mechanisms. 

The method has several potential paths for further development, where optimization of reaction rate coefficients, emission lumping, intermediate species lumping, and reaction lumping (shortcutting) deserves to be mentioned. However, the method already now delivers

fully functional accurate, reduced chemical kinetic mechanisms for all chemical situations evaluated thus far. Next step will be to incorporate the mechanism in coupled chemistry and flow dynamic models to enable realistic predictions of hazardous trace species concentrations in street canyon evironments.

References

[1] Emmerson and Evans, Atmospheric chemistry and physics, 9, 1831–1845 (2009).