Statistical uncertainty in automotive emissions testing

Abstract

Automotive exhaust emission testing according to legislated procedures and driving cycles is a routine process today in the United States, Europe and SE Asia. In practice, the selection of catalytic converters and the general design of exhaust aftertreatment systems are mainly aimed at the attainment of the legislated emissions limits with a reasonable safety margin. However, during the system optimization process, the catalyst and/or exhaust system manufacturer screens a large number of catalysts with a significant number of system design variations. In this process, certain levels of statistical uncertainty are present and must be seriously taken into account as stricter emissions standards are addressed and the safety margins become leaner. A methodology is developed and presented in this paper that is able to determine confidence levels in the attainment of specific emissions limits. This methodology combines results from experimental assessment of the distributions of engine-out emissions and catalyst activity by employing a previously developed and extensively validated model of a diesel oxidation catalyst. In this way, the emissions control system optimization is substantially aided and the required experimental effort is considerably reduced.