Comparison of a 4000-reaction chemical mechanism with the carbon bond IV and an adjusted carbon bond IV-EX mechanism using SMVGEAR II

Abstract

The well-known carbon bond IV (CBIV) chemical mechanism (33 species, 81 reactions) is compared with an adjusted carbon bond mechanism (ACBM) (109 species, 233 reactions) and a more explicit master chemical mechanism (MCM) (1427 species, 3911 reactions) in tests of their predictions of O 3, NO x (=NO+NO 2), HCHO, HNO 3, H 2O 2, and peroxyacetlynitrate (PAN). The ACBM was developed from a fourth mechanism, the expanded carbon bond mechanism (CBM-EX), by explicitly including the decomposition of C 2H 6, C 3H 8, and C 3H 6. All three mechanisms tested were updated with the inorganic chemistry from the ACBM and implemented into the sparse-matrix, ordinary differential equation solver, SMVGEAR II. Sparse-matrix treatment in SMVGEAR II reduced the number of calculations during matrix decomposition for the MCM by a factor of 15,000 (99.995%), or from an estimated 154 h to 37 s of cpu time per simulation day in one grid cell on an SGI origin 2000, in comparison with a full-matrix solution. Computer time for each mechanism was linearly proportional to the number of species in the mechanism. It is shown that the three mechanisms agreed closely when aromatic concentrations were initially low in comparison with alkane, alkene, and aldehyde initial concentrations. When aromatic concentrations were initially high (higher than that observed in urban air), the yields of O 3, HCHO, and PAN differed significantly among the three mechanisms although the daily maximum concentrations of these species agreed better. The aromatic representation in MCM appears to lead to systematic overprediction of ozone, according to a comparison with smog chamber data. For initial conditions taken from measurements at nine sites in Los Angeles, the daily maximum concentrations of O 3, HCHO, PAN, and H 2O 2 predicted by the three mechanisms differed by 30–50%, 10–40%, 15–40%, and 60–80%, respectively. The relative differences between the daytime series of O 3, HCHO, H 2O 2, and PAN predicted by the three mechanisms were 7–68%, 7–46%, 35–150%, and 10–64%, respectively. The use of the aromatic scheme of ACBM in MCM significantly reduced the disagreement with respect to ozone. The measurement of H 2O 2 in smog chamber experiments would be useful in validating chemical mechanisms.