Click for next page ( 58


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 57
Vehicle I Inspection and Maintenance Programs This chapter descnbes the basic components of the inspection and mainte- nance (I/M) program, the components that attempt to identify, diagnose, repair, and verify repairs for vehicles with high emissions. The following sections describe the venous network types, testing methods, and other elements of an I/M program. Also discussed are results from some previous evaluations of the effectiveness of existing programs. AM PROGRAM NETWORK TYPES The implementation structure of an I/M program, also known as "network type," can have a major impact on its operation. Three basic network types that are currently in operation are ~ Centralized. Decen~aTized. Hybnd. Remote sensing establishes yet another testing type. 1 Each program type States are now beginning to evaluate the feasibility of incorporating remote sensing as an integral part of their I/M programs. For example, the Denver Regional Air Quality Council has recommended beginning a "clean screen" program in January 2002, where on-road remote-sensing measurements would be used to exempt vehicles from scheduled testing (Regional Air Quality Council 2000~. Missouri has been operat- ing a remote-sensing clean screen program since early 2000. 57

OCR for page 57
58 Evaluating Vehicle Emissions I/M Programs has its strengths and weaknesses in terms of effectiveness, cost, and accep- tance by the public, the repair industry, and politicians. The following sections further describe the characteristics ofthe centralized, decentralized, and hybrid network types. Centralized Network A centralized network consists of a relatively small number (relative to a decentralized network) of stations that perform only emissions tests. Vehicles that fad! the inspection must be repaired elsewhere. This network typically is operated by a government entity or by a contractor with government adminis- tration. This system performs a high volume of emissions tests at Tow operating costs. The centralized network can achieve an economy of scale in terms of the investmentin equipment, inspector "raining, quality control, data collection, and reporting. Program management, consisting of administrative and opera- tional controls, is also more effective because there is better direct communi- cation with the testing stations. The smaller number of stations associated with a centralized testing network also simplifies program management. Disadvantages of a centralized network include the need to make inspec- tion stations convenient for the public while controlling costs for construction and operation. Property conveniently located for the motoring public is often diff~cultto find and/or very expansive. The centralized network wightbe more inconvenient to the public because of fewer testing stations and longer travel times to reach them. The centralized network also might tee more inconvenient to the public when a facility is experiencing high demand due to test expiration deadlines or lane closures due to equipment problems. An additional disadvantage ofthe centralized network is the 'ping-pony" effect. This happens when a vehicle fails the I/M test, obtains repairs at a separate location, and returns to the I/M centralized network but fails again. Some centralized networks have implemented measures such as the "repair effectiveness index," which rates the effectiveness of repair stations to mini- mize the ping-pony effect. Motorists can use this information to select a repair station to minimize the need to go back and forth between testing and repairs. Technicians in the repair industry might think they need to purchase emissions analyzers to verify that the emissions repairs they perform allow the vehicles to pass the test after repair. However, ina centralized network, this equipment

OCR for page 57
Vehicle I/M Programs 59 would be purchased without the opportunity to collect a fee for the emissions test verification. Decentralized Network A decentralized testing network consists of a larger number of Tow-volume stations that do both emissions testing and vehicle repairs. This type of net- work links testing to the repair process and is operated by private sector sta- tions. An advantage of the decentralized network is that it provides a revenue stream from testing fees, which in turn may enable the repair industry to ac- quire the training and skills needed to perform emissions-related vehicle re- pairs. In addition, the repair technician can use the emissions analyzer to verify the effectiveness ofthe repairs performed and eliminate the ping-pony effect that can occur for some vehicles in the centralized network. Program enforcement and quality control are more difficult in a decentral- ized network than in a centralized program because of the larger number of stations in the network.2 There can be more instances of fraud because of the difficulty of overseeing all test stations. Test-and-repair stations have addi- tional economic incentives notpresent in centralized test programs to fix vehi- cles to pass (to please the customer),3 or to fail (to get more repair business).4 2There is no recent comprehensive study, however, to indicate that there are more fraudulently passed vehicles occurring in decentralized programs. Testing fraud has been reported in both decentralized and centralized programs. Since there are many more stations performing inspections in the decentralized network, the number of stations cited for testing fraud will likely be higher compared to a centralized program. However, the number of inspections an individual station may be perfo~ing could be low whereas testing fraud at a high-volume centralized testing facility may impact a large number of tests. The committee could not find a rigorous comparison of these program types to state definitively that the number of vehicles impacted by testing fraud is greater in a decentralized program. 3Hubbard (1998) found that test-and-repair stations have an incentive to help vehicles pass inspections to increase the long-term demand for their inspections, even though they could increase short-term demand for emissions-related repairs by helping vehicles fail. 4Both test-and-repair and repair-only stations may provide more repairs than are actually needed (to make more money for the shop). Thus, some of the issues con- cerning repairs will happen in centralized testing as well.

OCR for page 57
60 Evaluating Vehicle Emissions I/M Programs However, advances in the design of emissions analyzers are thought to have made decentralized programs more effective by incorporating built-in quality control, making analyzers less prone to tampering, and linking station data to central data collection facilities. Hybrid Program A hybrid network is one that incorporates elements of both decentralized and centralized programs. One type of hybrid program incorporates both high- volume "test-only" stations and low-volume "repair-and-retest" stations. This approach achieves economy in enforcement, data and program management, and quality control for the initial test, which has the largest volume oftesting. It also provides an incentive to the repair industry by allowing them to perform the official retest and eliminates the problem of repaired vehicles having to return to a centralized facility for the retest. Another type of hybrid program sends a fraction of vehicles, such as those fitting the profile of a vehicle having high emissions, to a test-only station and allows others to choose to go to either a centralized station or a decentralized test-and-repair station. Such a program attempts to ensure that vehicles most likely to fail will undergo testing at facilities with the highest quality control. It also provides fairly convenient testing for most vehicle owners at the decen- tralized testing locations. VEHICLE-EMISSIONS TESTING Vehicle emissions tests vary in terms ofthe complexity of driving condi- tions represented. An important issue is the need for the test to obtain an accurate measurement of emissions while keeping equipment costs low and test duration short. Mass Emissions versus Concentration Measurements The two principal methods of measuring exhaust emissions are (~) directly measuring the mass of emitted pollutants, and (2) measuring the concentrations of pollutants in exhaust emissions. These methods are known as mass emis- sions tests and concentration tests, respectively.

OCR for page 57
Vehicle I/M Programs 61 Mass emissions tests quantify vehicle exhaust emissions by measuring the mass of various pollutants that are emitted. Generally, these emissions are expressed as the mass of pollutant emitted divided by the distance the vehicle is driven on a simulated driving cycle. In this type oftesting, a vehicle is driven on a dynamometer and the results are expressed in terms of grams of pollutant emitted per mile traveled. Concentration tests, on the other hand, measure the relative pollutant concentrations in a vehicle ' s exhaust. Because the measurement is a concen- tration measurement (generally expressed in terms of percentage or parts per million of total exhaust volume) little is known about the absolute amount of pollution generated. For a given exhaust concentration, vehicles with larger engines and higher fuel consumption will have higher mass emissions. To understand the magnitude of actual emissions, both pollutant concentrations and the volume of exhaust must tee known. The exhaust volume is measured in some, but not all, of the emissions tests used in I/M programs. By knowing the volume of emissions and airflow, it is possible to determine an average mass emissions rate. Converting concentration testresults to mass emissions introduces uncertainty in the estimates (Haskew et al. 1987~. Steady-State Versus Transient Testing Another way to differentiate vehicle emissions tests is by describing the conditions under which emissions are measured. Emissions can be measured under static or dynamic conditions, which are referred to as steady-state or transient tests, respectively. Steady-state tests measure vehicle emissions under one stable operating condition. Typically, a vehicle is tested at idle, when no dynamometer is used, or under steady speed with a simulated load when tested on a dynamometer. Dynamometer-based tests, such as the acceleration simulation mode (ASM) tests, are steady-state tests because they run the engine under a constant Toad instead of varying the Toad throughout the test, as is done in transient tests. Although steady-state tests do not simulate the range of driving conditions that are included in transient tests, they require smaller expenditures for testing equipment and may be performed in less time. Transient tests require a vehicle to operate under varying speeds and loads. They represent on-road driving conditions much better then steady-state tests, and they are transitory in nature. In emissions testing, typically the speed and acceleration ofthe vehicle are varied. By testing a vehicle under different

OCR for page 57
62 Evaluating Vehicle Emissions I/M Programs speeds and engine Toads, a broader range of emissions results is measured. To obtain a measurement thatis more representative of emissions when vehicles are driven on the road, test cycles have been developed that seek to replicate actual driving conditions. Exhaust Emissions Test Types A number of tests are commonly used to measure vehicle exhaust emis- sions. These range from unloaded idle tests to sophisticated transient-cycle mass emissions tests, such as the Federal Test Procedure (FTP). Some ofthe more common tests are described below. Mass Emissions Tests FTP City Driving Test This is a loaded-mode laboratory grade mass emissions test with transient (stop-and-go) driving conditions that is used by vehicle manufacturers to certify the emissions of prototype vehicles before they can sell the vehicle for the first time in the United States. It is usually considered the benchmark emissions test by which all other light-duty vehicle tests are measured. The FTP has extensive protocols, including specifying fuel parameters and environmental conditions and requires large expenditures of time, personnel, and capital. The test is split into various phases designed to measure the emissions from cold-start, urban driving, and hot-start operating conditions. To perform the City Driving Test (also known as the Urban Dyna- mometerDriving ScheduTe) end all other elements ofthe FTP (includingpre- paring the vehicle for testing), at least 2 days per vehicle are usually required. A problem with the FTP has been that the test does not measure emissions that occur during heavy acceleration or high-Ioad operating conditions that are sometimes observed in on-road driving (Kelly and Groblicki ~ 993; St. Denis et al. ~ 994; Cicero-Fernandez et al. ~ 997~. The Supplemental Federal Test Pro- cedure (SFTP) was proposed in ~ 996 (EPA ~ 996) to control emissions et high speed, at high load, and with the air conditioning operating. IM240 This test is a shortened version of the FTP, in which the vehicle is given minimal conditioning, and is assumed to be tested when fully warm. Thus, it can be conducted outside the laboratory in a well-equipped inspection station. It is a loaded-mode transient dynamometer test, which

OCR for page 57
Vehicle I/M Programs 63 measures the mass of emissions collected over a 240-second, 2-mile driving cycle that corresponds to the first 240 seconds ofthe City Driving Test ofthe FTP (see Figure 3-~. Many states that utilize the IM240 test have imple- mented a "fast/pass" or a "fast/fail" procedure orboth. This shortened version can reduce the testing time by several minutes. BARS ~This is a short, Toaded-mode dynamometer test utilizing similar equipment as the ]!M240. The driving cycle has been truncated to 3 ~ seconds, with the vehicle sharply accelerating and then decelerating through the test. A vehicle has three chances to pass the test. IM93/CT93 Connecticut 93 This test is a short version of the IM240 test cycle, utilizing the first "hill" or phase ofthe IM240. It consists of the first 93 seconds of the IM240. IMI47 This is also a shortened version ofthe IM240, specifically~e second phase (final 147 seconds). A major difference is the application of a retest aigonthm that determines whether a failing vehicle needs preconditioning before a final failure determination is made.5 A vehicle may be given up to three consecutive IMI47 drive cycles before it fails. VMASS The VMASS flowmeter system converts a concentration measurement to a mass measurement. The test methodology could use any transient I/M test cycle, such as the BARS I, CT93 , or IM I 47. In this system, BAR97 type equipment (see below) is coupled to a transient dynamometer Concentration Tests ~ idle test This steady-state unloaded test uses a tailpipe probe to measure directly the concentrations of CO, HC, and carbon dioxide (CO2) in exhaust emissions fromidling vehicles. A high-idle test,in which engine speed is manually increased to ~2,500 revolutions per minute (~pm), is sometimes performed in addition to the natural or "Iow-idle" test; in all cases, there is no Preconditioning refers to a vehicle that is fully warmed up so that it can give a valid result from an I/M emission test. Cutpoints, which determine passing or failing for such a vehicle, are based on testing a fully warmed-up vehicle in which the emissions control equipment, including the catalytic converter, are fully functional. If an owner drives a short distance to the test station or if the vehicle has to wait in the test station for a long time, the vehicle might not be fully warmed up, resulting in a false reading; a car that would have passed if fully warmed (i.e., fully preconditioned) would fail.

OCR for page 57
64 Evaluating Vehicle Emissions I/M Programs 0 ~ 30 ~ ~ 20 In 10 o FIGURE 3-1 IM240 driving cycle. r NO L: i, ~ .~.- . 0 50 100 150 200 2s0 300 Time (seconds) load applied to the engine. NOx concentrations are not usually measured as part of idle tests because NOx emissions are Tow if the engine is not under a load. The idle test is less expensive than Toaded-mode testing because no dynamometer is required. Although idle tests measure and report pollutant concentrations, Marr et al. (] 998) describes how mass emissions rates (i.e., grams of pollutant emitted per minute at idle) can be calculated from engine displacement volume, engine speed at idle, and measured tailpipe concentra- tions from the idle test. ASM This series of loaded-mode steady-state emissions tests mea- sures exhaust concentrations from motor vehicles operated on a dynamometer. The test series measures vehicle emissions under a loaded condition that simu- lates an acceleration event. The ASM steady-state test measures vehicle emissions at ~ 5 (ASM 5015) and 25 (ASM 2525) mph. The tests subject the vehicle to load conditions that are based on the maximum acceleration events in the FTP. The ASM 5015 subjects a vehicle to 50% ofthe maximum load conditions in the FTP test, and the ASM 2525 subjects a vehicle to 25% ofthe maximum Toad conditions in the FTP. ~ BAR97This refers to emissions testing equipment and software that meet the ~ 997 California Bureau of Automotive Repair's specifications for use in their I/M programs. The same test equipment may be used to perform either the ASM or the idle tests described above. This test equipment is nor- mally used for concentration measurements. When a BAR97 test analyzer is used in conjunction with a VMASS flow meter, it is then used to measure mass emissions. Earlier versions of the BAR analyzer specifications were issued in ~ 984, ~ 990, and ~ 994. Analyzers that met prey 997 specifications were usually used to perform idle tests only.

OCR for page 57
Vehicle I/M Programs 65 Remote sensing This is anonintrusive method of measuring emis- sions from individual vehicles as they drive by a sensor deployed at a roadside location. In normal operation, a beam of light is projected across a single lane oftraffic et tailpipe height, and light absorbed by pollutants is measured, usually at specific infrared wavelengths (ultraviolet light absorption is used in some systems to measure NO emissions). Remote-sensing measurements are Wpically coupled to a video image ofthe vehicle license plate, which can be used to obtain vehicle make, model year, and other relevant information from a central database. Comparison of Exhaust Emissions-Test Types .~ The California I/M pilot study (CARE 1996) provided the opportunity to compare emissions results from several candidate exhaust test types: the FTP, IM240, ASM5015, ASM2525, and low- and high-speed idle tests. In this program, 380 vehicles due for their biennia! inspection were given all six emis- sions tests. As a result, this sample was probably enriched in high emitters relative to the whole vehicle fleet. The data were analyzed to compare differ- ent emissions tests for measuring tailpipe CO and HC emissions. Figures 3-2 and 3-3 present the results, which are described further by [Lawson and Koracin (1996~. In Figure 3-2, the CO emissions from the 380 vehicles were rank-ordered from dirtiest to cleanest according to their FTP, IM240, ASM, and idle-test results. For the ASM and idle tests, the maximum value from either ofthe two ASM or idle tests was used. Figure 3-3 shows the results for HC plotted in the same manner. The correlation among the different test types is shown in Table 3- ~ . The statistic used is the Spearman rank-order correla- tion,6 which is a statistical method that measures the correlation between ranks oftwo sets of variables, rasher then their absolute values. The data from the California I/M pilot study illustrate Adhere is considerable correlation among different test types for measuring exhaust CO and HC. A similar comparison oftest types has not been done for NOX, which is Wpically not in an idle test. 6A correlation coefficient measures the degree to which two variables are related. For perfect positive correlation, the value ofthe correlation coeff~cientis +1; for perfect negative correlation, the value is -1. A correlation coefficient of O means there is no relationship between the variables. The Spearman rank correlation coefficient is a nonparametric (distribution-free) statistic measuring the strength of the associations between two variables when the variables are rank ordered.

OCR for page 57
66 Evaluating Vehicle Emissions I/M Programs 1.0 co . 0.~- co cq _ 0 0.6- V ~ 0.4- o A /~ _ - 00- ~_~ . O 38 76 114 152 190 228 266 304 342 380 Vehicles Ranked by Emissions Reading ['1'~ ..... TM240 ASM Idle Test FIGURE 3-2 Comparison of FTP, IM240, ASM, and idle-test results for CO. Replicate emissions tests were not performed for the 380 vehicles. Cor- relation of readings from two sets oftests with the same set of vehicles is far from perfect; Lawson (1995) reported an r-squared value of only 0.66 for IM240 test results between I/M lane data and laboratory data for the same vehicles. Another comparison of different test types canbe done with results ofthe Colorado's IM240 and idle-testprograms.7 Colorado's Automobile Inspection and Readjustment (AIR) program operates an IM240 program for 1982 and newer ears in the Denver metropolitan area, and a two-speed idle-test program in three other counties in Colorado. Vehicles older than ~ 98 ~ in the Denver metropolitan area are also tested with the two-speed idle test. A recent audit of this program (Air Improvement Resource ~ 999) using data for calendar year ~ 998 reported that the idle-test program given outside the Denver metro- politan area had a higher failure rate than the IM240 tests given in the Denver Additional comparison of test types could and should be done, including comparing an annual idle test program for HC and CO with a biennial IM240 test program. Emissions-reduction benefits of a biennial idle-test program could be compared with a biennial IM240 test program in which both had approximately the same failure rates.

OCR for page 57
Vehicle I/M Programs 67 1.0 0.8 a .g V 0.6- ~ 0.4- o , 80.2 0-0- ~ 0 38 76 114 152 190 228 266 304 342 380 Vehicles Ranked by Emissions Reading ~''m /"' '/ /.'/ ~ , b'1'~ IM240 ASM Idle Test FIGURE 3-3 Comparison of FTP, IM240, ASM, and idle-test results for HC. area ~ ~ 6.7% versus 5. ~ %), a smaller emissions-reduction benefit per repaired vehicle (23% versus 58%), and a similar overall emissions reduction (6.5% versus 6. ~ %) compared with vehicles tested with the IM240 test. Average repair costs for the idIe-tested vehicles were $95 per repair versus $2 ~ ~ for the IM240 tested vehicles. Both achieved about the same level of CO benefit in ~ 998. However, the idle test does not measure NOx, and while it fails more vehicles, average emissions reductions and costs per repair are less. No at- tempt was made to determine whether differences between the vehicle fleets in the Denver metropolitan area and the nonmetropolitan Denver counties caused any of these differences. Evaporative Emissions Tests Exhaust emissions are relatively easy to sample for routine inspection the exhaust exits from the tailpipe and a collection hose is attached to the tailpipe for testing. Evaporative emissions can occur from many places on the vehicle. The fuel tank, filler neck, and gas cap are typically at the rear of a vehicle. The engine's file] components are at the front, perhaps 12 feet away. If one tries to detect the presence of leaks from the system, the entire vehicle must

OCR for page 57
Vehicle I/M Programs 79 repair resulted in a range of estimates. Using the roadside pullover data, the IMRC calculated an emissions reduction of 17% for HC and 9/O for NOX. The IMRC also concluded that ~ 0/O of all vehicles that failed the initial Smog Check test never received a passing mark in later tests. This estimate is lower than that observed in Phoenix, where an estimated 29% of vehicles that failed the initial IM240 test never received a passing test or waiver in the following 3-15 months (Wenzel 1999a), and in Denver, where 27% of the e m i s s i o n s f a i 1 u r e s a r e u n r e s o l v e ~ ~ A i r ~ m p r o v e m e n t R e s o u r c e 1999 ~ . A n o t h e r 5- 10% ofthe vehicles observed on-road in California were eligible for Smog Check testing, but no records exist of these vehicles reporting for a test. Earlier independent evaluations of California' s ong~nal decentralized idle- testprogram showed no emissions-reduction benefit (Lawson 1993; Lawson et al. l 995,1996a). That result was based on data collected from California's random roadside surveys from 1989,1990, and 1991 and it is in contrast to the estimate that the program using the CARE I/M model (the CALIMFAC model) was producing emissions reductions of 18/O HC, 15% CO, and 7/O NOX (IMRC 1993) at that time, and to data from vehicles that were given emissions tests before and after repairs. Colorado AIR Program The 1999 audit of Colorado's AIR program used test data on the outcomes of failing vehicles as well as EPA's Serious CO Area Models to estimate emissions reductions (Air Improvement Resource 1999~. Directed at reducing CO emissions, the AIR program operates in metropolitan areas along the Front Range (Denver area, Colorado Springs, Fort Collins, and Greeley). As de- scnbed earlier, the program consists oftwo types oftests: a centralized bien- nial TM240 test used in the Denver area for 1982 and newer vehicles, and an annual basic idle test used in other areas and for vehicles older than 1982 in the enhanced area. The emissions reductions were estimated one time by analyz- ing the changes in emissions for fail-pass vehicles (vehicles that initially failed an I/M program and then passed a retest) tested during 1998 and part of 1999. iThe Serious CO Area model is a forerunner of the MOBILE6 model made avail- able by EPA to states completing CO planning activities. It utilizes the lower emissions deterioration rates that will be contained in MOBILE6 and reduces the credits for oxy- genated fuels.

OCR for page 57
80 Evaluating Vehicle Emissions I/M Programs The reductions were then used to estimate the benefits for the whole vehicle fleet. A second estimate of emissions reduction was made with EPA's Seri- ous CO Area Model. This mode! reflects the assumptions that will be con- tained in the new MOBILES, which probably will result in substantially lower credits for I/M programs than were estimated by MOBILES. Overall, the evaluation concluded that, depending on the method of analysis, the AIR pro- gramreduced 1999 CO emissions by 8-17%. The Tower estimate of emissions reductions was produced by using the in-program emissions data on fail-pass vehicles and by analyzing remote-sensing data from the area. The higher estimate was produced with the Serious CO Area Model. Earlier evaluations ofthe Colorado program estimated differentbeneiits. Stedman et al. ~ ~ 997) estimated a 4-7% CO emissions-reduction benefit based on remote-sensing measurements. This evaluation showed no HC and NOx emissions reductions and no CO emissions reductions for pre-1 982 vehicles. In another study, Stedman et al. (1997) estimated an 8-1 I/O benefit teased on the same measurements. ENVIRON International Corporation (1998) also funded a study that reported that the program obtained a 20-34% CO benefit using EPA's MOBILES model. Independent Evaluations of State I/M Programs In addition to the evaluations of I/M programs in Colorado and California described earlier in this report, numerous evaluations have been done by state agencies, EPA, and independent researchers. The following section describes a selection ofthese evaluations. Pierson (l 996) also summarizes earlier I/M program evaluations. Evaluation of Phoenix's Program Several studies in the past few years have examined the performance of the enhanced centralized IM240 program in Phoenix, Arizona. In addition to the thousands of records of program data gathered, extensive remote-sensing data were also collected. Several independent researchers such as Wenzel ( 1 999a) and Harnngton et al. ( l 99S, 2000) performed studies using this infor- mation. EPA also assessed the performance of the Phoenix program using only I/M test records (EPA ~ 997a). Results from the studies are shown in Table 3-4.

OCR for page 57
81 a, o v CQ To of N . _ Ct A N ._ o o o en X U. o V: ED I_ V g ._ ~ X ~ Z o .~ V .= a., C) 'S ~ ~ Z ~ .= 0 Al ^ V ~ ~ PI ~ o ~ :; c,, ~ - = E o E ~ ~ ~ ~ o ~ E U. V ~ ~ X a_ at Z Ce {t U. ~ Cal U. O ~ ~ ~ ~ ~ N ~ ~ C g O O ~ g \ ~o o o~ \o o .s o o 00 O. O _ o o \ \ \ ~ _ ON o o\ o\ _ O _ ~4 _ ~o o~ - \ o ~0 au O ~ O. O O _ C~` O Ct O O ~ ~ ~ _ ~o <: ~ ~ ~ ~ ~ E \ o - au ._ C) c'] _ "s: o ~ ._ ._ ~S o ~ ~ ~ o ax ~S ~ ~ ~ - ~ ~ Ct ~ N Ct O o ~ ~ ~ ~ ~_ _ o C) .^ ._ ox .^ C~ V __ +^ s0 . ~ .s _ o ox . ~ .E o V o a~ 1 - oo - U, 50 Ct ~_ ~0 . . U, Ct V ~ - tV O _ ~ O ~S C~ ~ U, Cd ,= 0 5 N ~ - ~ ~ ~0

OCR for page 57
82 Evaluating Vehicle Emissions I/M Programs The EPA analysis ( l 997a) compared predictions of the TECH5 compo- nent of the MOBILE mode! used in the Phoenix case with the emissions re- duction calculated from the IM240 test datable The results predicted by the mode} were only slightly greater for HC reductions, but they were substantially higher for NOx reductions (Table 3-4~. Chapter 5 provides a detailed discus- sion of the emissions-reduction benefits estimated for I/M with MOBILE. Also included in Table 3-4 are the original predictions for the IM240 pro- gram from EPA's regulatory impact analysis of enhanced I/M in 1992 (EPA ~ 992b). Even though these estimates are the results of comparisons between enhanced I/M and non-T/M programs, they are probably close to what Phoenix is actually being granted in SIP credits. This is due to the MOBILE mode} predictions of low emissions reductions (5% for HC) from basic I/M programs. As shown in Table 3-4, the MOBILE model forecasted greater HC and CO emissions reductions than were actually found using in-program or remote- sensing data. The large reductions predicted by the model were originally based on the assumption that all failing vehicles would be repaired. In Phoenix, however, the program data showed that roughly 25% of them had still not passed ~ year after failing the test (Ando et al. 2000~. Harrington et al. (2000) examined the costs and emissions reductions of the Phoenix program by using all IM240 test results over a period of ~ 7 months (January ~ 996 to May ~ 997~. This study used emissions data of initial and final tests for failed vehicles and assumed that emissions repairs lasted 2 years. Given these assumptions, the study estimated that HC emissions were reduced by ~ 3/O and NOX emissions by 7/O over the 2-year penod. This finding was very similar to the EPA (1997a) results from program data. Wenzel ~ ~ 999a) compared emissions reductions based on IM240 test data with a large sample of remote-sensing readings. The results ofthis study for the Phoenix program data were similar to those of the Harnngton et al. (2000) study, which was also based on test data from the program. Wenze} found, however, that the emissions reductions were lower when remote-sensing readings were used; HC reductions were ~ 1/O instead of 14%. Furthermore, the ~ 1/O remote-sensing readings included a relatively large share of pretest reductions, which would not be reflected in the program data. These pre- inspection repairs are observed in remote-sensing data, which show a reduc- tion in emissions In vehicles ~ -2 weeks before their inspection (Wenzel ~ 999a), lathe TECHS component of MOBILE was modified in this work to reflect emissions reductions that occur in a single I/M cycle. These model modifications are described in EPA (1997a, pp. 12-13~.

OCR for page 57
Vehicle I/M Programs 83 as well as survey data, which indicated that a significant fraction of motorists had their vehicles tuned up before inspection (IMRC 20001. The implication of these findings is that the remote-sensing results show substantially lower post-inspection emissions reductions than the in-program data. Evaluation of Minnesota's Program Scherrer and Kittelson ~ ~ 994) assessed the impact on air quality of an I/M programinitiatedinI99linM~nneapolis. Direct measurements ofambientCO data et three monitoring sites were used. Assessing the effectiveness of I/M using CO air-quality data is appearing because a high fraction of CO is emitted by light-duty vehicles subject to such testing, and CO is relatively unreactive in the atmosphere. Minnesota's centralized I/M program consisted of an idle- testprogram for HC and CO that failed about 9.4/O of vehicles dunng its first year(July 1991 to June 1992~. This study used amultifactorairegression that corrected for vehicle activity and meteorological factors to discern I/M bene- fits from time series observations of CO concentrations. The study collected hourly ambient CO monitoring data in the city and meteorological data at the regional airport. The average ambient reduction of CO attributed to I/M was ~ .3 ~ ~ .4/O, with individual sites showing a 5.~/O decrease, a ~ .5/O decrease, and a 3.4/O increase. Using air-quality data to evaluate I/M emissions benefits raises many issues. The committee recognizes that observing the effects of I/M programs on air quality is difficult because the level of emissions reductions have been relatively modest and there are numerous confounding variables. One of the issues encountered in this study related to the methods used to correct for the effects of changes in vehicle activity patterns and the vehicle fleet itself. The wide range of changes in ambient CO levels, estimated at the three monitoring sites, also suggests that using a limited number of monitoring sites for the purpose of program evaluation might tee unreliable. However, it is those moni- tors that define whether a locality is in nonattainment of the national ambient air-quality standards and trigger the need for an I/M program. Evaluation of Georgia's Program The Air Quality Laboratory at Georgia Institute of Technology used re- mote sensing to determine the influence of city characteristics on motor vehicle

OCR for page 57
84 Evaluating Vehicle Emissions I/M Programs emissions (Rodgers 2000~. The reference method of evaluating I/M's benefits (discussed in Chapter 6) compares vehicle emissions in an I/M city with those of a reference fleet in a city with similar socioeconomic and meteorological characteristics but with a different I/M program. This reference fleet could be from a non-~/M city in which case an evaluation would determine the reduction in emissions due to the I/M programor from a city with a bench- mark I/M program. In the latter case, the evaluation would compare how well the program performed relative to the benchmark. The Air Quality Laboratory studded whether the selection ofthe reference fleet could effect the evaluation. It looked at whether emissions in comparable cities were actually similar. Because of similar socioeconomic characteristics, fleet age distribution, and average model year emissions, the I/M cities of Nashville, Tennessee, and Atlanta, Georgia, had comparable fleets for application ofthe reference meth- od. Other city comparisons (Boston, Massachusetts compared with Burling- ton, Vermont; Macon, Georgia compared with Augusta, Georgia), however, suggested that characteristics outside of I/M program status can result in dissimilar model-year emissions. That result points to the difficulty in i inding a comparable fleet for use in the application of the reference method. The Air Quality Laboratory also used remote-sensing data to evaluate the emissions-reduction benefits for Atlanta's I/M program. Until 1999, the I/M program included a decentralized idle test in the four counties in the central Atianta metropolitan area. The Air Quality Laboratory study compared emis- sions in the I/M area with emissions in the surrounding nine counties, which were not subject to an I/M program at the time. From this study, a reduction in CO emissions of 7.5/O was estimated. Note that no attempt was made to correct for socioeconomic differences between the I/M and non-~/M areas. Atlanta' s I/M program expanded to include these nine surrounding counties in ~ 996. Using more recent data, the Air Quality Laboratory also estimated the benefit ofthe program by comparing emissions of vehicles that had been tested with those that had not been tested when the program expanded to include the surrounding nine counties. This method (the step method described in Chapter 6) yielded an ~ 1/O reduction in CO emissions. EPA's National Tampering Surveys For a number of years, from the latter ~ 970s to the early ~ 990s, the EPA conducted motor vehicle tampering surveys at various locations throughout the

OCR for page 57
Vehicle I/M Programs 85 country. EPA used data from these surveys to document the occurrence of tampering-related defectsi2 in the motor vehicle fleet and to compare different I/M program types for effectiveness. Random roadside surveys allow for inspection of vehicles as they are driven on the road, but many EPA surveys of centralized programs were done in the test lane, rather than on the road, which introduces bias into the results. Motorists were given no advance notice that they wilIbe stopped for aninspection. The date collectedin those surveys also give a representative sampling of actual vehicle miles traveled, because the more a vehicle is driven, the more likely it is to be stopped. Participation in the surveys was voluntary, so the survey results probably are biased toward vehicle data from compliant motorists who are generally willing to participate in such surveys. In each year's survey, EPA selected up to ~ 5 cities as sampling sites. To obtain statistically meaningful data, 300-500 vehicles were inspected et each location. The mix of inspected vehicles was assumed by EPA to be a self- weighting sample, and no attempt was made to approximate the national vehi- cle mix. The sampling location and the method of stopping individual vehicles varied for each location in accordance with the type of I/M program in place. Sampling also occurred in non-/M areas. The roadside inspection included the following: Basic vehicle identification data Check of all emissions-control system components Measurement ofno-Ioad, Tow-idle (~IOOOrpm) HC end CO emissions Collection offuel sample from unTeaded-only fueled vehicles forlead analysis Inspection of fuel inlet restrictor (designed to prohibit fueling with leaded gasoline) Test of tailpipe for lead deposits Lawson et al. (l 995) performed a series of analyses using EPA's national tampering survey data, obtained through roadside surveys, to compare the effectiveness of centralized and decentralized I/M programs with no I/M program. One study used pass-faiT rates for the 44,000 vehicles reported, according to tampering inspection and emissions-test results from the EPA "Tampering" is the malfunctioning of one or more emissions-control devices due to either deliberate disablement or mechanical failure.

OCR for page 57
86 Evaluating Vehicle Emissions I/M Programs survey data taken from ~ 985 to ~ 990. The data were adjusted for model year and mileage. The model-year categories are ~ 985- ~ 990, ~ 980- ~ 984, and pre- ~ 980, corresponding roughly to fairly homogeneous em~ssions-control technolo- g~es, although some differences for catalyst technologies and trucks and light- duty vehicles span these model-year groupings. Mileage was divided into five categories, the highest including 100,000 or more miles, as recorded on the vehicle's odometer. Finally, within each model-year group and mileage inter- val, vehicles were categorized according to the type of I/M program where they were sampled. As shown in Figure 3-4, each combination of mileage and year of manu- facture was compared for mean values of overall failure rates from different types of I/M programs: none, decentralized, and centralized. This plot displays the following features: failure rates tend to increase with odometer reading and vehicle age and to be highest for the oldest technology vehicles. Neither centralized nor decentralized programs showed a much Tower failure rate than vehicles from non-/M program areas. In a second study, data obtained in these surveys over an S-year period ( 1985- l 992) were adjusted for differences in vehicle age and odometer read- ings (Lawson et al. ~996a). The analyses also accounted for the type of I/M program in place in each ofthe areas where the surveys were made. Tamper- ing and emissions failure rates for different I/M configurations during the ~ 985- ~ 992 period are presented in Table 3-5, which shows that there were only small differences between different I/M program configurations. There was also only a 4/O difference in tampering or emissions failure rates between non- ]:/M and centralized areas, a small difference compared with the overall emis- sions and tampering failure rates. Because EPA discontinued its motor vehicle tampering surveys after ~ 992, more recent analyses with nationwide data have not been possible. SUMMARY Numerous variations of vehicle emissions I/M programs are in use today, each with its own attributes. One program type is the centralized program using transient emissions tests such as the IM240. Such programs enable the estimation of mass emissions of NOx, CO, and HC under simulated Unving conditions. Possible drawbacks from this type of program are higher testing equipment costs and greater inconvenience to the public because of fewer testing locations. Additionally, motorists requiring repairs must visit a separate

OCR for page 57
87 U. C`l ~ o o o . . ~ z o o o o o o o oo CC) ~ C~ U. ~ ~ ................................... - OD ~ ~ o ~ Z o , , , , o o o o o o o oo CD ~ C~ _ a~ u~ ......................................................................... ~ o ' ''' ''''''' '''''' ~ o~ ~ ~ ~ ~ ~ ~ o o o o o o o 0 ~ ~ ~ o~1~ ~o s~ . ~ Y ~ .. . . . ... Z 1 o o o o o o o 0 o x U. Ct - o OOr ~o _m ................................................................. ~ I ' , , Z o ~ ~ ~ ~ ~ ~ o o o o o o o oo ~ ~ CM C~ _ ,_ + C~ o o ~4 . . . , , , , o o o o ~ CO o o o ~ C~ - z cn ~ ~- o o . . . o o o o oo Co o o o z o s ~ ~ ~ l l l - l o =. o o o o o o o oD co ~ cM cD ~ ln o l ~o o oo cq s~ au r~ - o - o - o - o o' o o o O CO CD . . - o - o - o o - o - o - o o o ~ cM o=1 : ~o ; : z l o - o - o - o - o - o - o o - o - o - o - o - o - o o o o o o o oo co ~ c~

OCR for page 57
88 1 U. o U. ~ sm o ~ Z o o~ o o o o o o o oo ~ ~ ~ ~ ~ ~ Z o , t o o o o o o o oo CO ~ C~ U. ~ ~ N ~ Z O ~ ~ ~ ~ ~ o' O O o O O O O oc) co ~ ~1 U. ~ (S) ~ o ~ _ E ~ .. . .. ... . . `~ 1 _ o= ~ ~ ~ o~ O O O O o O O 00 CO ~t C~l ... . . . o~ o~ O O O O O O 0 0 CO ~ CM o U. _ ~0 Ct - .s Ct a' O =0 Ct ~ ._. ~ ~ U. ~ ~ .O ~ Ct ~ O =, O ~ ~9 Ct ~ Ct ~ ~ C.) O a' 0 E~ _ ~ ~ . ~ ~ O .= U, o g U, U, e~ o V C~0 ^ a~ ~ Ct O U, ~o ~ 11 U, ~ O ~ a~ 00 5- ~4 ao a' o o r~ o Ct o U) Ct U, - _ O O _ ~ ~ ~0 Ct ~S C~ C~ o o Ct

OCR for page 57
Tampering HC >100 HC > 400 Program Type Rate (%) ppm (%) ppm (%) CO > 1 (%) CO > 4 (%) Non-I/M 19.5 28.0 8.8 20.8 10.4 Vehicle I/M Programs 89 TABLE 3-5 Normalized Tampering and Emissions Failure Rates by program Type from EPA's 1985-92 Motor Vehicle Tampering Surveys Decentralized 16.6 25.7 7.9 18.5 8.7 Centralized 15.4 24.3 5.8 16.3 6.2 (on-road) Centralized 15.0 26.6 5.7 14.7 5.6 (I/M lane) repair facility before they return to be retested. Another type of I/M program uses a decentralized idle test in which exhaust concentrations of HC and CO are measured. Such programs provide greater convenience to motorists be- cause ofthe larger number oftesting facilities, andbecause testing end repairs can occur at the same place. These programs, however, do not simulate NOX emissions and are more difficult to oversee. Many variations on these two pro- gram types exist. For example, the current California Smog Check program is a hybrid network that uses an ASM test to estimate CO, HC, and NO concentrations. Previous evaluations of I/M's emissions benefits have been based on MOBILE as well as direct estimates of vehicle emissions. The committee believes those evaluations based on direct estimates of vehicle emissions are far superior to those taken from models. Estimates of I/M benefits from direct measurements of vehicle exhaust using test data, remote sensing, end roadside pullovers have shown reductions to be significantly smaller than model-pre- dicted reductions. This conclusion is based on a review of state-sponsored evaluations ofthe Colorado and California programs and independent evalua- tions of these programs and programs in Arizona, Minnesota, and Georgia. Although an exhaustive review of all previous evaluations is beyond the scope of this study, the committee believes those described represent some of the best examples of I/M evaluation. The committee recognizes that the number of evaluations will expand greatly in the future. As discussed in Chapter 1, this is the first phase of a two-part study. The second phase is expected to con- tinue to review evaluations and to rely on them as critical sources of informa- tion. x