"SCOPE OF THE EPA DATA BASE The EPA data analysis procedures will be discussed in this section. In a departure from classical fuel economy measurement techniques@ the EPA determined fuel economy using the mass carbon monoxide and carbon dioxide concentrations obtained from the 1972 Federal Test Procedure. The fuel economy expressed in miles per gallon was computed via a mass carbon balance equation@ and averaged for tests made in a given model year at a given inertial weight. For example@ 21 1967 model-year vehicles of 3500-pounds inertial weight were evaluated; the average mpg was 13@8@ with a spread of 11.5 to 18.6. In interpreting its data@ the EPA assumed that vehicle weight was the only parameter having an influence on fuel economy. The EPA developed a data analysis model@ using the product of vehicle inertial weight and fuel economy (mpg)@ which was held to be a constant@ ""C@"" for a given model year. The variability in C (inertial weight X fuel economy) is used by EPA as a measure of the fuel economy penalty incurred in a given year relative to a reference point. For example@ the C value for the uncontrolled (1957-1967) vehicle fleet is 52129@ while that for 1973 is 48667; the % decline is 6.6%. Using the same methodology@ the average loss from the 1963-1973 fleet compared to the uncontrolled base is calculated at 7@75%. The EPA pointed out in a footnote that no attempt has been made to analyze these data statistically and that their observations were based solely on observed means of the samples taken. The critique of this procedure which follows treats the Task Force reservations in a detailed fashion. It is worth noting in this section that there are also deficiencies in the EPA data base@ These deficiencies are discussed in the Attachments@ and it will suffice to add only general observations at this point. ? The distribution of vehicles tested by EPA is not representative of the actual U.S. vehicle population. As pointed out in Attachment 3@ Table I@ the distribution of vehicles tested by weight@ at least for 1972 models@ is at variance with 1972 sales data for cars of U.S. manufacture. The EPA fleet contains a higher percentage of the lighter vehicles@ which@ as will be shown@ have not been greatly affected by the emission control effort. The EPA data are biased toward the 1973 certification fleet@ which is more suspect than other groups@ since these vehicles are at very low mileage and little deterioration has taken place. ? The average inertial weights@ engine displacements@ and ratios of automatic to manual transmissions vary widely from year to year; the use of the C factor on this type of data base will tend to minimize differences which could be ascribed to factors other than inertial weight. ? The EPA data were limited to model year@ inertial weight test (not car weight)@ engine displacement@ transmission type@ odometer miles@ and emissions (hydrocarbon@ CO@CO2@ NO^) other desirable data were not available@"
API PUBL 4195-1973 history
1973API PUBL 4195-1973 A TECHNICAL ANALYSIS OF FUEL ECONOMY AND EMISSION CONTROL