This study has aimed at estimating effects of exhaust emissions during both stationary and transient conditions, and possible cold start effects. One side effect of this work has been a dynamic method for obtaining emission data in phase with engine load data. Unless this can be achieved, it is very difficult to determine exhaust emission response to engine loads. Input data such as torque and rpm levels together with exhaust emission data for carbon dioxide, nitrogen oxides, particles, hydrocarbons and carbon monoxide have been obtained from TRL. Non-linear regression analysis was applied in estimating the emission levels for the engines as functions of the torque and rpm levels and their levels of change over time. The results obtained show high correlations between measured emission levels and model values for carbon dioxide and nitrogen oxides in particular, and fairly high values for particles. Transient effects are estimated with and without using the changes in torque and rpm levels. Measured in this manner, they are fairly small, with a high value for particles which increase by 5.4% on the average. There are studies and practical experience reporting transient effects of 20-60% for particles, indicating a considerable underestimation in the approach used. An analysis of transient effects using cross tabulation and pairwise statistical hypothesis testing gave inconclusive results. The established significant effects turned out to be both positive and negative in the same cell, and were often so many that random effects cannot be ignored as the probable cause. Attempts to estimate the cold start effects using US transient data for both cold and warm starts showed a small, but insignificant increase in all emissions during cold start, except for nitrogen oxides. (A)
Samenvatting