Estimating Full IM240 Emissions from Partial Test Results: Evidence from Arizona

The expense and inconvenience of enhanced vehicle emissions testing using the full 240-second dynamometer test has led states to search for ways to shorten the test process. In fact, all states that currently use the IM240 allow some type of fast-pass, usually as early in the test as second 31, and Arizona allows vehicles to fast-fail after second 93. While these shorter tests save states millions of dollars in inspection lanes and driver costs, there is a loss in information since test results are no longer comparable across vehicles. This paper presents a methodology for estimating full 240 second results from partial-test results for three pollutants: HC, CO and NOx. Using random sample of vehicles in Arizona which received full 240 second tests, we use regression analysis to estimate the relationship between emissions at second 240 and emissions at earlier seconds in the test. We examine the influence of other variables such as age, model-year group, and the pollution level itself on this relationship. We then use the estimated coefficients in several applications. First, we attempt to shed light on the frequent assertion that the results of the dynamometer test provide guidance for vehicle repair of failing vehicles. Using a probit analysis, we find that the probability that a failing vehicle will passing the test on the first retest is greater the longer the test has progressed. Second, we test the accuracy of our estimates for forecasting fleet emissions from partial test emissions results in Arizona. We find that forecast fleet average emissions are very close to the actual fleet averages.