Abstract
A new promising statistical approach based on principal component analysis is presented in this paper for the development of real use emission factors. It allows one to build in one model the relationship between two sets of variable: the set of emissions (CO, HC, NOX CO2 ) and the set of many explicative variables characterizing driving cycles, also if not so many tests are performed. Moreover, the approach considers that variables belonging to each set are correlated, this hypothesis is relevant in real use emissions where both emissions and driving cycles variables are highly related within each group. The model is built using the Partial Least Square approach. It is applied to different case studies considering three different gasoline car models equipped with three way catalyst. Common Driving Cycles defined in ARTEMIS project funded by European Union and emissions measurements data performed in the same project have been used,. Some different data obtained in preceding work with Naples driving cycles are used as well. The three cars have different technology: EURO II, EURO III. Trends resulted very different among the three cars but for the three data sets same significant quantities resulted to be the most important to explain emission variation. As a result, these quantities (idling time, percentage of time driven, percentage of time spent in first gear, percentage of time spent at velocity less than 20 km/h ) can be effectively used to build the statistical model, they have a precise physical meaning and a specific study can be done to put in relation quantities determined for each cycle, related aspects of driving cycle and instantaneous emission record to explain different emission behavior of different cars. Mean velocity, in spite the generalized use in all emission model proved to be not the most important variable in the projection. Moreover, considered quantities can be effectively used to classify driving cycles into the design of experiments, aiming to create an emission data base representative of wide driving conditions in a balanced way respect to factors affecting emissions, where the problem is that each observation (an emission measurement) is obtained from one driving cycle which is one combination of many different aspects