Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Emerging Emissions Testing Technologies New technologies continue to be developed to provide faster, more convenient, end more accurate emissions testing. Since inspection end maintenance (~/M) programs have been in operation for approximately 30 years, the accumulated data can be used for motor vehicle profiling. In addition, on-board diagnostics (OBD) systems and remote-sensing technologies may eliminate traditional I/M testing for some vehicles. These emerging emissions technologies are de- scnbed below. MOTOR VEHICLE PROFILING Motor vehicle profiling can be used to help determine whether a vehicle is likely to be a high emitter. By examining past emissions performance of a vehicle (through existing T/M records), as well as performance for similar makes and models, the likelihood of whether a vehicle will pass or fail an I/M inspection may be determined. There are two types of profile: ~ Low-emitter profile (LEP:The LEP attempts to assign a probability that a vehicle or group of vehicles will pass an I/M test. It relies on I/M re- cords indicating which vehicle makes, models, and model years have low incidences of T/M failure. 90
Emerging Emissions Testing Technologies 91 · High-emitter profile (HEP)The high-emitter profile similarly at- tempts to predict the probability that a vehicle or group of vehicles will fait an EM test. Examples ofthis type of profiling are provided by Wenzel and Ross (] 996), who used California remote-sensing data, and Wenze! (! 999a), who used Arizona IM240 data to identify vehicle makes and models with high malfunction rates. However, data from random roadside surveys in California have shown significant differences between the HEP and on-road data. These profiles might be used to make I/M programs more cost-effective by targeting more frequent inspection to those vehicles most likely to need repair and exempting from testing vehicles with a small likelihood of high emis- sions. For example, depending on the state, many programs exempt vehicles 2-5 model years old (MECA ~ 999~. The developers of one LEP estimated that such a "clean screening" model could screen out at least 50% of the vehicles from testing with little impact on emissions benefits (Klausmeier and Kishan 1998~. The California Bureau of Automotive Repair uses a HEP in their en- hanced I/M program to help determine whether to send vehicles to test-only facilities, which are thought to perform more accurate tests. Eastern Research Group developed and maintains the computer program for performing this analysis (Eastern Research Group ~ 997~. This HEP uses the prior inspection history of a vehicle; historical failure rates of vehicles ofthe same year, make, and model; and remote-sensing measurements ofthe vehicle. The California I/M Review Committee (IMRC 2000) reported the failure rates and initial emissions test results for HEP versus non-HEP vehicles. As shown in Table 4- I, the HEP made only a small improvement in identifying vehicles likely to fail, and the improvement was only for post-1986 model-year vehicles. The benefits of the HEP model are much smaller than those estimated by Klaus- meier and Kishan ~ ~ 998) from a small sample of data from California. How- ever, updating the profile for recent test data might improve its performance. If a HEP is used in I/M program design, it is important to consider the regressive impacts that will occur to low-income motorists who may be more frequently targeted for testing. To the extent such a policy is regressive, it could be combined with the most cost-effective supporting policies, such as repair assistance or scrappage programs, in an attempt to mitigate its negative effects.
92 Evaluating Vehicle Emissions I/M Programs TABLE 4-l Results of the HEP Used in the California Smog Check Program HEP Directed Random Directed Not Directed Failure Rate 26% 22% 23% Initial emissions HC (ppm) 71 63 68 CO (%) 0.45 0.41 0.46 NOX (ppm) 520 485 494 Note: Vehicles could be directed using the HEP, randomly directed, or not directed to a test-only facility, but directed to a test and repair facility. Source: IMRC 2000. ON-BOARD DIAGNOSTICS The use of OBD technologies in I/M programs represents a major shift from current practice. As described in Chapter 2, the OBDIT system on ~ 996 end newer vehicles uses sensors to monitor end modify the performance ofthe engine and emissions control components. Itis possible to identify problems with OBD sensors or emissions control components by interfacing a diagnostic analyzer ("scan tool") with the vehicle ' s electronic processor to download any diagnostic trouble codes. These codes identity emissions-control systems and components that are malfunctioning. If the OBD I/M program is operating properly, OBDIT inspections will fail vehicles if either the vehicle's emissions- con~o} components are, or have been, malfunctioning or if the sensors monitor- ing emissions-control components are malfunctioning. This program is in con- trast to a traditional I/M emissions-testing program where a vehicle is in- spectedto determine if itis emitting, et the time of its appearance et the testing station, more pollutants than a standard set by individual states. The U.S. Environmental Protection Agency (EPA 200 ~ ~ has recently finalized an OBD rule that requires states to begin implementing OBD testing in I/M programs for ~ 996 and newer OBD-equipped vehicles. The OBDII system does not actually measure emissions. Because this system does not measure emissions but rather alerts drivers that there is a problem that might result in excess emissions, evaluating the benefits from such a system is not straightforward. Direct emissions reductions from the repairs resulting from an OBDIT alert cannot be estimated without subjecting the vehicle to at least two tests that measure emissions, one before and one after the repair. In addition, the objective of OBD is to prevent vehicles from
Emerging Emissions Testing Technologies 93 becoming high emitters. It is difficult to quantify the emissions benefits of preventing vehicles from becoming high emitters. An important aspect of the OBD system is that it is required to maintain a memory of past events, and store a history of previous problems, even though the problem no longer exists. Thus, a vehicle in an OBD I/M program might fait due to an intermittent problem. Another aspect of the OBD I/M system is that it might fait a vehicle due to a failed sensor. Though the infor- mation collected by the sensor might be used to minimize emissions, the on- board computer on OBDTI-equipped vehicles can use an alternate strategy to control vehicle emissions until the failed sensor is repaired or replaced. The emissions using the alternate "fail-safe" strategy can be almost as low as if the control system was using the primary sensor. Thus, some repairs, particularly repairs to sensors, will result in little, if any, immediate emissions reductions. This can make OBD-directed repairs look ineffective in the short term. As descnbed in several studies below, the design of the OBDII system means that the malfunction indicator light (MIL or "check engine" light) will illuminate and diagnostic trouble codes issued for a potentially significant num- ber of vehicles despite vehicle emissions being below the state's I/M emissions cutpoints and even the vehicle's certification standards. Furthermore, OBDIT's strict malfunction criterion (the MTE illuminates if a problem is de- tected that could cause emissions to exceed ~ .5 times the vehicle's emissions certification standards) were not set by individual states to meet their air-qual- ity needs, but were set nationally according to the needs of areas with the most stringent requirements for vehicle emissions control. If people respond to the MIL and the OBD system operates properly, there is little need for aperiodic inspection. Vehicles wiliremainatIowemis- sions levels throughout their lifetimes. However, there is a question as to whether drivers will seek repairs as the vehicle ages, and runs out of warranty. As described in several studies below, there is also the concern over the large number of vehicles with high emissions and no MID illuminated. There are a number of issues related to the OBD I/M programs that will first be highlighted. They are · Readiness codes before and after repairs. · Failure criteria. · Emissions or pollutants of concern. · Fast pass using OBDIl. · Human response to the MTL.
94 Evaluating Vehicle Emissions I/Al Programs We discuss each in more detail below. In addition, we also discuss recent regulations and results of studies related to the use of OBD in I/M. Readiness Code Through the use of various sensors, the OBDIT system tests the compo- nents of the emissions-control and fuel-management systems to ensure that they are operating correctly. There is a specific criterion for each emissions- control system component that must be met before an OBDIT system sensor is considered ready, meaning the component in question has been monitored. These criteria are defined and implemented by each manufacturer. Forexam- ple, certain components are monitored as soon as the engine is fumed on, after as many as 40 engine restarts, or only after the vehicle is driven under a cer- tain Toad and speed. In the case of monitoring the evaporative emissions- contro} system, readiness codes are not set until the vehicle is exposed to cer- tain ambient temperatures. A significant part ofthe problem is that the detection limits are set to very sensitive levels. If, for instance, the system was just to detect if the gas cap was in place, the test could be completed almost on every trip. Because the intent ofthe evaporative control system monitoring is to reliably detect a very small leak (e.g., a I-millimeter diameter hole), the conditions necessary to make such a determination might not occur very often and perhaps never in the normal use of a certain vehicle. If the OBD detection limits were set at higher levels, the tests could be completed more often. EPA's OBD rule (EPA 2001) requires that all but two ofthese readiness codes be set to indicate that a particular emissions control component has been monitored by the OBDIT system. If more then two readiness codes are found unset, then the vehicle would be tested by an emissions tailpipe test or rejected from OBD testing. EPA (2000c, 200 ~ ~ estimates that the frequency of finding more than two readiness codes unset will be small. However, McCTintock (2000a) found in Colorado a much higher level of readiness codes being unset than the EP A ' s e s tim ate s . The readiness codes are turned off when the battery is disconnected, possibly occurring as the result of repairs. If that occurs, they must be reset before the vehicle can be subjected to another OBD check. Resetting might be an issue for motorists attempting to retest their vehicles immediately after repairs. Additionally, if it is easier to pass an emissions tailpipe test than an
Emerging Emissions Testing Technologies 95 OBDIT check and the tailpipe test is an option for vehicles with unset readiness codes, motorists that fear failing an OBDIT check may attempt to avoid this test by simply disconnecting the battery. The current OBDIT system has no "stay-alive" memory that can persist through a battery disconnection. Future OBD systems might have such memory components and carry records of vehicle identification number, odometer reading, and a record ofthe MILs and their source and date. This information might provide a solution to the issue of unset readiness codes. Failure Criteria Most I/M programs fail vehicles for excess emissions that are much higher than vehicles certification standards. The OBDIT system, by design, illumi- nates the MIL (and fails a vehicle) if a problem is detected that might cause emissions to exceed 1.5 times the certification standard. In addition, MIL illumination occurs if the system determines that a monitor or sensor is not responding properly, even without increased emissions. Emissions or Pollutants of Concern Many state I/M programs are designed to address a particular air-quaTity problem. Colorado's program is designed to control carbon monoxide (CO). The nonattainment areas in Texas are concerned about controlling nitrogen oxides (NOx). Several states are concerned about hydrocarbons (HC) orNOx, while others are concerned about both HC and NOx, but one to a greater degree than the other. I/M cutpoints and standards are designed to control the emissions or pollutants of concern for each state. However, an OBDII pro- gramwill fad] a vehicle if any ofthe emissions (HC, CO, end NOx) exceed the failure criteria. Fast Pass Using OBDI! States may consider using the OBDIT system as a fast-pass test. This means that vehicles that demonstrate a clean OBDII system will be passed and not subjected to an emissions test. If the MIL is illuminated or there are unset readiness codes, the vehicle will not fail, but the vehicle owner will be
96 Evaluating Vehicle Emissions I/M Programs given information for suggested repairs, and the vehicle will be required to undergo an emissions test. Under the current regulation (EPA 2001) the fast- pass test may be used only in the start-up of an OBDTI test and for one cycle of vehicle inspection requirement. The fast-pass test could also be used for areas that are not required to implement an I/M program but are considering an OBDIl test as a preventive control measure. Human Response to OBDIl The intent ofthe MTE is to inform motorists that they need to check their emissions control systems because there is an indication of some malfunction. However, many motorists are unlikely to spend the time and money to bring vehicles in for repair voluntarily, especially if the vehicle's emissions systems are outside the warranty period and the vehicle's operation is not affected. This might become especially problematic as the vehicle ages. Warranty periods -for most components are relatively short compared with current vehi- cle lifetimes. The federal emissions control warranty is 96 months/80,000 miles for major emissions control components (such as the catalyst), and 24 months/24,000 miles for other components (such as sensors, PCV valve, EGR valve). Auto manufacturers have extended these warranties to 3 years/36,000 miles and 10 years/100,000 miles. If motorists are required to have the OBD system checked as part of an I/M test, as with any I/M testing system, there is the possibility of cheating to avoid test and repair. We discussed above the possibility that motorists will simply disconnectbatteries, causing unset reediness codes, which would allow them to take a more lenient IM240 tailpipe test rather than an OBD test. Other avoidance methods might also arise, although it is too early to determine how serious or widespread this problem could be. Perhaps the most serious problem with motorist response to the MIL is the confusion that is likely to occur about what an illuminated check-engine light represents and what its relationship is to emissions and the cost-effectiveness of emissions reductions. Current studies, discussed later in this section, indi- cate little consistency between a MIL illumination and the probability of an IM240 failure. In addition, repairs of many vehicles withilluminatedMILs do not produce substantial emission reductions, atleastin the short-run. In part, it could be that many of the problems caught by OBDII are early-stage prob- lems. Repairing these vehicles might result in much lower emissions later in the vehicles' lifetimes. Failures could also be due to evaporative emissions
Emerging Emissions Testing Technologies 97 problems. Even after repair ofthese problems, improvements are difficult to measure. The inconsistencies between results of IM240 and OBD tests might produce motorists' confusion and serious skepticism about OBD I/M pro- grams. Recent OBD I/M Regulations EPA recently finalized a rule concerning the use of OBD in I/M (EPA 2001~. It mandates the introduction of OBD tests for cars equipped with OBDIT (1996 model year and later) end provides states the flexibility to com- pletely replace traditional I/M tests with OBD checks. ~ The rule also extends the deadline by ~ year (untiT January I, 2002) for states to implement OBD checks, and loosens some criteria for performing OBD inspections without all readiness codes being set. It also allows the states to phase-in the OBD test. Technical Analyses Regarding OBD I/M Tests Table 4-2 contains a summary description of several recent studies regard- ing OBDIl's use in I/M programs. Three EPA studies were summarized in a draft technical support document (EPA 2000c) that accompanied the OBD rule. The objectives of the EPA studies were to assess the effectiveness of OBD I/M testing for exhaust and evaporative emissions and to investigate implementation issues through the information collected from the Wisconsin enhanced T/M test lanes. The first study evaluated the effectiveness of OBD I/M for tailpipe testing. EPA recruited 20 ~ OBDIl-equipped vehicles with either a malfEmction indica- tor light (MIL) illuminated ~ ~ 94 vehicles) or no MIL illuminated but suspected of having high emissions (eight vehicles).2 Once recruited, each vehicle was iThe Clean Air Act Amendments of 1990 and later regulations mandated that OBD checks be incorporated into I/M programs by January 1, 2001. However, until this proposed rule, states would have had to implement OBD checks in addition to tradi- tional I/M programs. EPA concluded that there is no reason to subject vehicles to both IM240 and OBD checks (EPA 2000c). 2The discrepancy between the stated number of vehicles recruited (201) and the total number of vehicles in these two categories (202) exists because one vehicle was recruited twice for separate problems.
98 Evaluating Vehicle Emissions I/M Programs TABLE 4-2 Studies of OBD Issues Related to UM Programs Vehicle Study Recruitment EPA study of Solicitation, mainly 193 vehicles MIL illuminated from rental car vehicles agencies EPA study of I/M lanes 8 vehicles high emissions, no MIL vehicles EPA Same, with in- 30 vehicles evaporative duced evaporative emissions study emission failures Vehicles Tested Comments Tailpipe comparison of IM240 with OBDII for vehicles with MIL on (EPA 2000c) Vehicles suspected of high emissions with MIL off (EPA 2000c) Evaporative emissions effectiveness of OBDII (EPA 2000c) CE-CERT Rental car fleet 75 vehicles Tailpipe comparison of (University of and newspaper IM240 with OBDII MIL on California, ads (Durbin et al.2001) Riverside) Wisconsin I/M lanes 116,667 1996- Tailpipe comparison of 1998 vehicles IM240 with OBDII MIL on (EPA 2000c) Colorado I/M lanes (fails 3,162 vehicles Tailpipe comparison of CDPHE study back to back in 2000; 1,466 IM240 with MIL on, IM240 fails and/or vehicles in 1999 including retest data MIL on) (Barrett 2001 ~ Colorado I/M lanes (fails 21 vehicles (out Tailpipe comparison of CDPHE study back to back of a study IM240 with OBDII MIL IM240 fails and/or design of 100) on, including repairs MIL on) up to April 2001 (Barrett 2001 ~ given a Federal Test Procedure (FTP) emissions test as well as a laboratory IM240 emissions testy Table 4-3 shows that ofthe ~ 94 vehicles with the MIL 3The difference between a lane IM240 test, those conducted as part of an I/M program, and a laboratory IM240 test is that the latter can control better for factors such as the quality and calibration of the test equipment, ambient conditions, and tire pressure.
Emerging Emissions Testing Technologies 99 TABLE 4-3 Results of EPA Study of Tailpipe Emissions for Vehicles with MIL Illuminated and Vehicles Suspected to Have High Emissions FTP> 1x FTP> 1.5x No. of MIL Self- Certification Certification Vehicles Cleared Standards Standards Vehicles with MIL illuminated 194 Vehicles with no MIL 8 illuminated but suspected to have high emissions Source: EPA 2000c. 11 58 31 ~ 4 illuminated, ~ ~ had the MIL self-clea~A before any emissions tests were com- pleted, and 58 had emissions greater then the certification standards. Ofthe 58 vehicles with emissions greater then the certification standards, only 3 ~ had emissions greater than 1.5 times the certification standards. Table 4-3 indicates that 70°/O of the vehicles ( l 3 6) had the MIL illuminated but emissions below certification standards. Ofthese ~ 36 vehicles, 97 vehicles were identified to have a broken part (EPA 2000c). The remaining vehicles might have had an intermittent problem, but it no longer existed when the vehicle was tested. The problem might have been the result of a loose gas cap, a fuel-injector problem, or an intermittent electrical problem, or the vehicle could have been driven under an extreme driving condition. The OBDII sys- tem still stores much ofthis information in its memory and allows technicians to review the pest history ofthe originalproblem. The large number offailing vehicles with relatively Tow emissions was also noted by Durbin et al. (200 ~ ~ in a study of 75 vehicles with MILs illuminated. That study found that 63°/O of vehicles with MILs illuminated had emissions below their certification standards, and 79°/O had emissions below 1.5 times their certification standards. The eight vehicles in Table 4-3 with no MIL illuminated were recruited because of either high IM240 test lane results or other characteristics such as high mileage or driveability problems that suggest high emissions. Of those eight, four had emissions greater then ~ .5 times their certification standards for 4The MIL self-cleanng occurs when an intermittent problem is not detected on later system scans by the OBD system and the MIL tutus off.
~ 00 Evaluating Vehicle Emissions I/M Programs CO anchor NOX on the FTP test but no MIL illuminated. The OBDIT catalyst monitoring system verifies the efficiency for HC. Thus, if the vehicle had high CO or NOx, and low HC, it would not trigger the MIL. This study also identified 21 vehicles with FTP emissions greater then two times their certification standards, and compared whether the OBD or labora- tory IM240 test was better able to identify them. Of the 21,19 were correctly identified by the OBD system, whereas only ~ 3 were correctly identified by the laboratory IM240 test. A second EPA study induced failures in the evaporative system on 30 vehicles to determine whether the OBD system could detect a range of such failures. These included missing, loose, or leaking gas caps and disconnected purge lines used to overload the carbon canister. Of these 30 vehicles, MTE illumination occurred in over 80°/0 (25 vehicles). It should be noted that tradi- tional I/M tests for evaporative emissions (the purge and pressure tests de- scribed in Chapter 3) are difficult to perform. Thus, the results of this study are encouraging in terms of the ability of OBD to identify problems in the evaporative system. The third EPA study used data gathered in Wisconsin to assess failure rates and other issues associated with implementing an OBDI/M program. This study examined the relative failure rates for OBDI/M versus lane IM240 testing. For 1996 model-yearvehicles, the OBD failure rate fromtheWiscon- sin lane data was 2.4%, and the IM240 failure rate was 2.~%. However, the percentage of vehicles that failed both was only 0.2%, which indicates that only a small fraction (about ~ 0%) of vehicles failing one also failed the other. Figure 4-l shows data for the number of vehicles failing IM240 and OBD tests for model-year ~ 996- ~ 998 vehicles. The large discrepancies between TM240 and OBD test failures are obviously of major concern. The Wisconsin study also estimated the average time to perform an OBD I/M inspection (31 seconds) and identified atypical data link connector loca- tions. The issue here is that the connector is placed out of sight and is difficult to locate as one moves from mode} to model. The study also looked at OBD readiness code data. Current regulations for OBD I/M testing require I/M programs to reject a vehicle with two or more readiness codes unset. Readi- ness codes were found unset in 5.~°/0 of 1996 model-year vehicles, 2.3% of ~ 997 model-year vehicles, and I.4% of ~ 998 model-year vehicles. The reason for these lower "not ready" rates on newer vehicles could be due to manufac- turers; building better OBD systems or to lower mileage accumulation on these newer vehicles. Allowing states to proceed with OBD I/M testing with two
Emerging Emissions Testing Technologies ~ 01 6,667 mode! year ~ 996-1 998 vehicles screened - 1,479 OBD failures \ - - - ~ 73 vehicles failed Moth OBD and IM240 1,344 IM240 failures - FIGURE 4-1 Number of OBD and IM240 failures from Wisconsin lane data for 1996 and newer vehicles. or more readiness codes unset lowered the percentage of vehicles that would be rejected from 3.2% to I.4°/O. At the time ofthis report, the Colorado Depa~l~entofPublic Health and Environment(CDPHE) is investigating various aspects of OBDIl. Under the Colorado' s enhanced I/M program, new vehicles are exempt from testing for the first 4 years or until a change of ownership. If change of ownership oc- curs, the vehicles are inspected et that time end then are subject to the biennial inspection program from that time forward. A vehicle in the Colorado program can fait either if emissions are higher than the cutpoints or if the MIL is illumi- nated. Two recent presentations examine I/M program data from testing of ~ 996 and newer vehicles in Colorado's I/M program (McClintock 2000b; Barrett 20014. Table 4-4 summarizes the results. The data show that, for 1996 and newer vehicles, about nine times more vehicles fad! the MIL illumination test than the IM240 emissions test. Only a small fraction (about 2%) of the total vehicles tested failed both tests. Barrett (200 ~ ~ further reported that repair data for vehicles that fad] either the MTE illumination or the IM240 tests showed similar repair cost per vehicle.
102 Evaluating Vehicle Emissions I/M Programs TABLE 4-4 Number of ~ 996 and Newer Model-Year Vehicles Failing the IM240 and MIL Illumination Tests in Colorado's Enhanced IM240 Program Number of 1996 Model Year and Newer Vehicles Calendar Year Failed IM240 Test Failed IM240 + MIL Failed MIL 1999 182 36 1,320 2000 393 66 2,835 Sources: McClintock 2000b; Barrett 2001. Both reported average repair costs at about $220 per vehicle (about 13°/O of vehicles that failed these tests reported repair costs). As shown in Table 4-2, the number of OBDIl-induced repairs likely exceeded the IM240-induced repairs by almost an order of magnitude for these relatively new vehicles. However, the emissions reductions associated with OBDIT repairs are much less than those resulting from IM240 repairs. For example, Barrett (2001) reported that for the retested IM240 failures, CO emissions were reduced from 47. ~ g/mi to 5.7 g/mi, whereas for the retested MTE illuminated vehicles, CO emissions were reduced from 4.7 g/mi to 3.3 g/mi. This result is not sur- prising since the OBDTT system will cause MTE illumination for a number of problems that do not cause high emissions in the short term but could lead to higher emissions or the nondetection of an emissions problem in the future Summary of Technical Issues Regarding OBD I/M Tests The combination of much higher failure rate, Tower emissions reductions, and comparable cost of repair for OBDIl-failed vehicles is likely to lead to higher repair costs and lower cost efficiency associated with an OBD I/M program. It should be noted, however, that the failure rate and repair cost information currently available come from first-generation OBDll systems and young vehicles with low overall failure rates. Additionally, the cost-effective- ness of an OBDIl-based inspection system is difficult to compare with that of a traditional IM240 program end might require comparing emissions of OBDIl- equipped vehicles in an I/M program with OBDIl-equipped vehicles operating in an area that does not have an inspection program. OBDIT-based inspection programs can be expected to have much greater amounts of pre-inspection repair, as one can expect very few motorists with a MIL illuminated to go to
Emerging Emissions Testing Technologies 103 an inspection testing station. If an OBD I/M program is operating as it is supposed to, very few OBDII-equipped vehicles eligible for testing would be operating with high emissions in the area. These studies raise many important issues that should be reviewed by an independent group. In particular, both the Colorado and Wisconsin studies with recruitment of large numbers of vehicles from I/M lanes found many vehicles that failed the IM240 test but did not have the MIL on. This finding is a seri- ous problem that needs to be thoroughly analyzed, because the IM240 failures are from higher-emitting vehicles. The problem could arise in manufacturer' s design of OBDII systems, in the reproducibility of the IM240 test, or some other factor. In any case, this problem needs to be understood and corrected before I/M programs operate using OBDIT alone. REMOTE SENSING Remote sensing is a technique used to measure emissions from individual vehicles as they drive by a roadside sensor. Light of suitable wavelengths is projected across the roadway at tailpipe height and is partially absorbed by pollutants present in vehicle exhaust. Passing vehicles block the light beam as they drive by. Ratios of individualpollutants to CO2 presentin vehicle exhaust are determined by analysis of a series of sensor scans of the exhaust plume made after a vehicle has driven by. Background corrections are made by using readings taken just before the sensor beam is blocked by each passing vehicle. These ratios are used to calculate end report exhaust concentrations similar to those measured by a probe inserted into the tailpipe. Remote sensing of vehicle emissions was pioneered by Stedman and coworkers at the University of Denver (Bishop et al. 1989~. Researchers at General Motors (GM) developed a similar instrument (Stephens and Cadle ~ 991~. More recent approaches to remote sensing of vehicle emissions (Nel- son et al. ~ 998; Baum et al. 2000) have made it possible to measure a wider range of exhaust constituents, including ammonia, NO2, NO, and some individ- ual organic compounds. Other advances that have facilitated the collection and interpretation of remote-sensing measurements include pattern recognition software to read vehicle license plates automatically (lack et. al ~ 995) and sensors to measure speeds and accelerations of passing vehicles. A typical remote sensor measures between 3,000 and 10,OOOvehiclesperdayandpro- vides the only test type that can be operated unmanned.
04 Evaluating Vehicle Emissions I/M Programs Although remote-sensing readings are commonly reported as the concen- trations of exhaust constituents (e.g., % CO, or the amount ofCO as a fraction oftotal exhaust gas volume), the underlying measurement is actually the mole ratio of the pollutant of interest (CO, HC, or NO) to carbon dioxide (COO. Tailpipe concentrations are calculated from the ratios measured by the remote sensor (Stedman et al. ~ 99 ~ ). Ratios are determined by measunug the exhaust plume repeatedly within an interval of 0.5 second after the vehicle drives by and plotting the amount of pollutant detected versus the amount of CO2. An example ofthe correlation of NO and CO2 signals measured by remote sens- ing in the plume of a passing vehicle is shown in Figure 4-2. The remote sen- sor measures the amount of pollutant emitted relative to the amount of CO2 because dilution of the exhaust plume varies with time, wind speed, vehicle speed, and other factors. From remote sensor measurements of exhaust emissions ratios (CO/CO2, HC/CO2, and NO/CO2) and knowledge of fuel properties, it is possible to derive mass emissions rates per unit of fuel burned by carbon balance (Stedman et al. ~ 99 I; Singer and Harley ~ 996~. Therefore, although remote sensing is described above as a concentration test, it may also be used to provide mass emissions results but only per unit of fuel burned. Fuel economy (not measured by remote sensors) must be estimated to obtain mass emissions rates per distance traveled. Remote-Sensor Accuracy Carbon Monoxide CO is the pollutant for which remote-sensing capabilities are best devel- oped and demonstrated. Typically, this is the most abundant pollutant in vehi- cle exhaust, facilitating its measurement. One of the first assessments of remote-sensor accuracy in measuring CO emissions involved double-blind comparisons of remote-sensor readings with a specially equipped vehicle that had on-board instruments to measure exhaust emissions (Lawson et al. ~ 990~. An observer in the vehicle manually selected different air-to-fuel ratios and recorded the on-board CO measurement as the vehicle passed by the remote sensor. The remote sensor was highly correlated with simultaneous on-board CO measurements (r2 = 0.94), with a regression slope of ~ .03 and an intercept of 0.08% CO, over a series of 34 vehicle passes with speeds ranging from approximately ~ 5 to 50 mph.
100- o ._ ct a) c' 0 ~ ~ `~ 40- u) 0 ct CD a, ,,, 60- 20- O- 0.025- 0.020- O 0.015- z - a ° 0.01 0- 0.005- O Emerging Emissions Testing Technologies ~ 05 (a) ~ ~ . . · · ~ NO · · ·A · .A 2 a-` . ~ · .~. ~ C02 . ted ·~s teed 0 0.1 0.2 0.3 Time (seconds) 0.4 0.5 (b) If- I · ~~ - · ,/e .~ . - '' 1 1 1 1 1 0 0.5 1.5 ~ % CO2 2.5 3 FIGURE 4-2 Correlation of NO and CO2 signals measured by remote sensing. Source: Popp 1999. Reprinted by permission of the author. The comparisons between remote-sensor readings and on-board emissions measurements from an instrumented vehicle were repeated in ~ 99 ~ (Ashbaugh et al. ~ 992, Stedman et al. ~ 994) in a study involving University of Denver (DU) remote sensors, a GM remote sensor, and an instrumented vehicle. Correlations between remote-sensor readings and the instrumented vehicle for
~ 06 Evaluating Vehicle Emissions I/M Programs CO were high: r2 = o.gg for the DU remote sensor, and r2 = 0.96 for the GM remote sensor. Corresponding slopes of linear regression lines for the DU and GM remote sensors relative to the instrumented vehicle were 0.98 it 0.02 and 0.96 ~ 0.02, respectively. These results confirm the accuracy of remote sens- ing in measuring CO emissions. Hydrocarbons Measuring exhaust HC emissions by remote sensing is more challenging than measuring CO emissions for several reasons. HC emissions are typically much lower than CO emissions, and their infrared extinction coefficients are much less, making them more difficult to detect. Instead of dealing with a single well-defined molecule, HC encompasses hundreds of different organic compounds with different infrared spectra. Therefore, it is not surprising that in the intercomparison study(Ashbaughet al. 1992) of remote sensing with en on-board infrared analyzers on an instrumented vehicle, remote-sensing mea- surements of HC emissions were found to be less accurate than those of CO. Correlations between remote-sensor HC measurements and instrumented vehicle readings were somewhat lower (r2 = 0.85 for DU, r2 = 0.87 for GM) than those for CO, and linear regression slopes were further away from ~ .0 (slope was 1.08 ~ 0.06 for DU, and 0.85 ~ 0.05 for GM). The first DU re- mote sensor read systematically higher than the GM remote sensor by approxi- mately 3 5°/O on average. The infrared filter used for HC measurements in the DU remote sensors was changed later (Guenther et al. 1995) and now matches more closely the filter used in the GM sensor. The earlier DU remote-sensor HC channel might have suffered from interference due to absorption by water vapor in vehicle exhaust. Further evaluation of remote- sensing capabilities for HC is needed. Infrared analyzers that are used to measure HC emissions are optimized to measure absorption by the carbon-hydrogen bonds present in alkanes; typi- cally, propane or hexane is used to calibrate the analyzers. Other compounds present in vehicle exhaust, such as alkenes and aromatics, have additional peaks in their infrared spectra at frequencies different from those of alkalies; however, no small set of remote-sensor channels is able to measure all the hydrocarbons present in vehicle exhaust. Therefore, remote sensors (as well as infrared analyzers typically used in idle and acceleration-simulation-mode (ASM) tests) detect only a fraction of total HC in vehicle exhaust (Stephens et al. 1996~. To obtain a more accurate estimate of mass emissions rates,
Emerging Emissions Testing Technologies 107 Singer recommended that infrared analyzer results for HC should be multiplied by a factor of about 2 (Singer et al. ~ 9984; the scaling factor may vary depend- ing on the chemical composition ofthe HC emissions. Emissions tests, such as the FTP and IM240 described above, use a flame ionization detector (FID) to measure HC emissions. This detector is known to respond similarly to alkalies, alkenes, and aromatics and thus is better suited for measuring the HC mass present. Nitrogen Oxides NOX is defined as the sum of nitric oxide (NO) and nitrogen dioxide (N02), although the direct emissions from internal combustion engines are dominated by NO (Kirchstetter et al. ~ 996; Jimenez et al. 2000~. Remote-sensing capa- bilities for NO have been developed more recently than those for CO and HC. Measurements of NO emissions are challenging because of overlapping ab- sorption by other exhaust constituents: water vapor in the infrared and aro- matic HC in the ultraviolet (UV). Various approaches have been used, includ- ing tunable infrared diode lasers (Nelson et al. ~ 998; limenez et al. ~ 999) and UV spectroscopy (Zhang et al. ~ 996a; Popp et al. ~ 999b). By modifying the UV absorption technique used in earlier DU remote sensors, Popp et al. were able to achieve a lower detection limit (i.e., increased sensitivity) and eliminate interference due to UV absorption by aromatic HC. A limited comparison of NOX concentrations by remote sensing and on- board measurements in heavy-duty diesel truck exhaust has been reported (Jimenez et al. 2000~. Remote-sensing measurements of the NOX/CO2 ratio agreed win similar measurements made on-board an instrumented diesel Suck, although the authors note that remote-sensor accuracy was assessed over only a limited portion ofthe likely range of NOX/CO2 emissions ratios. An instru- ment comparison and further evaluation of remote-sensing methods for mea- suring NOX emissions is needed. Remote-sensing site selection is especially problematic for measuring NOX emissions accurately because oftheir depend- ence on operating conditions, but it might not be critical for identifying high NOX emitters. Particulate Matter Development ofcapabilities for remote sensing of particulate matter (PM) emissions is an area of active research. The Coordinating Research Council
~ 08 Evaluating Vehicle Emissions I/M Programs is currently sponsoring a study to evaluate the effectiveness of remote sensing to measure PM emissions from heavy-duty diesel vehicles (CRC Project E- 56~. Qualitative measurements of PM emissions using remote sensing will likely be available soon. Quantitative measurement of mass emissions rates for PM will be more difficult to achieve because the scattering and absorption of light by airborne particles are complex functions of particle size and chemi- cal composition. Site Selection and Effects of Engine Load An important consideration when measuring vehicle emissions by remote sensing is careful selection of roadside monitoring sites. Sites where vehicles might tee sampled during cold-start operation should tee avoided, because vehi- cle emissions are higher than normal until the engine and emissions-control systems have warmed up. Sites where driving conditions involve frequent heavy acceleration and high-Ioad driving (e.g., accelerating on steep uphill grades) might yield unrepresentative emissions results associated with opera- tion in commanded enrichment modes. (Enrichment modes have lower air/fuel ratios and increased CO and HC emissions.) To correct for effects of engine Toad on exhaust emissions, the roadway grade at remote-sensing sites should be noted, and the speed and acceleration of each vehicle passing by the remote sensor should tee measured. This is now common practice, but information on vehicle speed and acceleration is rarely available in older remote-sensing studies. Ashbaugh et al. (1992) and Zhang et al. (1993) noted that tailpipe HC concentrations measured by remote sensing were elevated at a site where vehicles were decelerating. The significance of this result is not that such driving modes are an important source of HC (fue! consumption is Tow under these conditions, hence mass emissions rates are also low), but rather that such sites should be avoided in remote-sensing studies. Jimenez (l 998) and McCTintock (l 998) developed more formal approaches to estimating engine load based on readily observed vehicle-operating parame- ters, such as speed and acceleration in addition to roadway grade. Vehicle- specific power (VSP) is estimated as the sum of loads due to aerodynamic drag (wind resistance), vehicle acceleration, rolling resistance (tire-roadway friction), and hill climbing, divided by the mass ofthe vehicle and commonly reported in kilowatts per metric ton. On a fleet-average basis, CO emissions
Emerging Emissions Testing Technologies 109 appear to be less sensitive to engine load than other pollutants; they remain fairly constant over a VSP range of -5 to 20 kilowatts per metric ton (kW/t) (Bishop et al. ~ 999~. HC emissions decrease with increasing engine load, over a VSP range of-15 to 15 kW/t (Bishop et al. 1999~. NOX emissions often increase over the same VSP range, but further study of this relationship is needed. Coverage it remote sensing is used to screen the vehicle fleet to help identify high- emitting vehicles, a significant issue is the need to measure by remote sensing the emissions from most of the vehicles operating in a given area. This need might require multiple remote sensors, which must be moved to different road- side sampling locations every few days. Current constraints, such as the need to measure across a single lane oftraffic, make it difficult to achieve complete coverage of the fleet. A remote-sensing study conducted by the Bureau of Automotive Repair (Amlin 1995) using 10 remote-sensing vans in a 3-month period was able to obtain CO emissions measurements matched to readable license plates for 380,000 vehicles registered in Sacramento County, California. Emissions from 58°/O ofthese vehicles were measured more than once in this study. About 2 million remote-sensing measurements obtained at multiple roadside sampling sites were needed to achieve a 47°/O level of coverage for the ~ ~ 0,000 vehicles registered in Sacramento County. Improved coverage of the vehicle fleet could have been obtained by increasing the number of sites and days where remote sensors were operated; when the study ended, 30°/O of vehicles driving by the remote sensors were being observed for the first time. Note that remote sensors provided measurements for unregistered and out-of-county vehicles operating within Sacramento County; some ofthese vehicles would not be covered by traditional I/M programs. Need for Quality Assurance and Quality Control As with any emissions testing program, a critical element is data quality. Success in using remote sensingin field studies has been mixed, with problems often apparent when multiple remote-sensing instruments and measurement teams are involved. Walsh and Gertler (1997) reviewed remote-sensing data
~ ~ O Evaluating Vehicle Emissions I/M Programs collected in Texas during ~ 996 in Houston, DalIas/Ft. Worth, and E! Paso. For the most recent 10-15 vehicle model years, HC emissions measured by one of the two remote sensors used in Houston were systematically lower by factors of 2-3 than those measured with the other. They concluded that the most probable cause was "related to calibration gases used by the individual instru- ments during the field study." There was also evidence of systematic differ- ences in CO measurements from Houston relative to those measured in Dallas/Ft. Worth with different remote-sensing instruments. Walsh and Gertler (1997) noted that although the study sponsor was aware ofthe utility of a side-by-side comparison ofthe various remote-sensing units being used, it was not done because of budget and time constraints. The experience described above with remote sensing in Texas is not unique: similar problems arose when multiple contractors and remote-sensing instruments were used in field studies in Phoenix and Sacramento. In such cases, there appears to have been too much emphasis on the number of vehi- cles and sites sampled and insufficient time and effort devoted to quality con- trol. Use of Remote Sensing to Identify High Emitters The question of whether remote sensing should be used to identify high emitters in I/M programs has been controversial. Proponents of remote sens- ing argue that current I/M programs waste time and money because about 20 vehicles have to be tested to identify one high emitter that is a candidate for repair; some vehicles might be adjusted by their owners or technicians to pass scheduled emissions tests, but they do not remain clean or are difficult to re- pair; and motorists might register their vehicles in ways that avoid I/M require- ments. Opponents argue that remote sensing is not a reliable way to measure vehicle emissions, motorists might avoid known remote-sensor locations and/or take steps to frustrate remote-sensing device measurements, and use of re- mote sensing might lead to an unacceptably high rate of false failures, reducing public acceptance of I/M programs. Several valuable studies have been conducted in which both remote sens- ing and standard I/M program tests have been used to measure emissions from the same vehicles. The most useful comparisons are made with I/M program tests administered at the roadside on vehicles pulled over immediately after their emissions were measured by remote sensing. Some analysts emphasize
Emerging Emissions Testing Technologies 111 perceived inadequacies in the correlation between remote sensing and I/M test results for individual vehicles as a reason why remote sensing should not be used in I/M programs. More appropriate categorical analyses focus on the ability of remote sensing to identify high-emitting vehicles that will fait UM program tests, as discussed below. When comparing remote-sensing readings with roadside emissions inspec- tion results, it is important to remember that the vehicle itself can be a signifi- cant source of variability, especiallyin the case of intermittent malfunctions in the emissions-control systems. Therefore, it is unrealistic to expect any emis- sions test to be ~ 00°/0 repeatable in making pass-fai} determinations for individ- ual vehicles, even for the same test administered repeatedly. For example, Knepperet al. (l 993) measured emissions from TO "normal" and7 "high-emit- ting" vehicles, all ~ 986 or later mode} years. Emissions were measured in the laboratory using the FTP loaded-mode test described in Chapter 3. Knepper found that relative to normal emitters, the high-emitting vehicles showedgreat- er emissions of CO and HC and greater variability of emission rates within each test vehicle. Variability in emission rates for the high-emitting vehicles was traced to changes in air/fuel ratio from test to test. Hawthorne (Los Angeles) 1989 Study Lawson et al. ( 1990) describe a study in which vehicles were pulled over, and roadside inspections made immediately after vehicle emissions had been measured by remote sensing. Of 50 vehicles that were identified by remote sensing as having high emissions (more than 2% CO in their exhaust), 28 failed the CO portion ofthe roadside inspection and 15 failed for other reasons, for a total of 43 failing vehicles out of 50. The rate of false failures (error of commission) was 14%. Of the ~ 5 vehicles in the remote-sensing/pullover study that failed the roadside inspection for reasons other then high CO emissions, ~ werepre-1975 models that also would have failed for CO if a tailpipe concentration of less than 2% had been required to pass. (The actual CO concentration required to pass the roadside test ranged from 2.5% to 7°/O for pre-1975 vehicles, in contrast to levels of approximately ~ % CO required for ~ 980 and newer cars.) Of ~ 0 additional vehicles that were measured by remote sensing to have Tow CO emissions, all passed the CO portion ofthe roadside inspection (2 of these ~ O vehicles failed for reasons other than high CO emissions). The error
~12 Evaluating Vehicle Emissions I/M Programs of omission for this small sample of vehicles was 20%: remote sensing ofCO emissions alone did not pick up other problems that leaf to failure ofthe road- side inspection. Rosemead Boulevard (Los Angeles) 1991 Study A larger-scale combined remote-sensing/roadside inspection study was performed during summer 1991 in El Monte, California, as described by Stedman et al. ~ ~ 994~. Vehicles were identified as high emitters based on two remote sensors both reading greater than 4% CO. There was a preference for purling over post- 980 mode] vehicles for roadside inspections because of the less stringent emissions requirements for older vehicles. Of 307 vehicles that had both remote-sensor and roadside inspection data available, ~ 5°/0 failed the exhaust emissions (idle test) portion of the roadside inspection, and the overall failure rate (including vehicles with tampered or noncomplying emissions-control systems identified during an underhood inspection) was 92%. Michigan 1992 Study Stephens et al. ~ ~ 995) assessed variability in vehicle emissions by examin- ing correlations between multiple remote-censor readings of CO and HC emis- sions and between remote-sensor and roadside IM240 emissions test results for 170 vehicles. In general, the correlations between remote-sensor and IM240 results improved as the number of remote-sensor measurements of a vehicle's emissions increased from ~ to 4. This finding indicates that the rate of false failures is likely to decrease when more than one high remote-sensing reading is required to identify a vehicle as a high emitter. It is unclear, how- ever, whether simply raising the cutpoint would be a more effective method for identifying high emitters and reducing the number of false failures. Orange County (Los Angeles) 1995 Study Lawson et al. ( 1 996b) conducted a study in Orange County, California, in 1995, where high-emitting vehicles were identified by remote sensors and pulled over for repairs. In that study, measurements from two remote sensors
Emerging Emissions Testing Technologies 113 separated by 100-150 feet were used to identify high emitters, whose criteria were average readings of 4°/O and 0. ~ °/O for CO and HC, respectively. During a 10-day period, remote-sensing readings were obtained for 19,000 vehicles at the two locations. Nearly ~ 0°/O of the vehicles transiting the remote-sensing devices at the two sites exceeded the high-emitter cutpoint criteria, with a high emitter passing the remote-sensing devices every 2. ~ minutes. More than 600 vehicles were pulled over for possible participation in the repair program, and 140 were selected for repairs and testing. Once a car was chosen for participation in the program, it was given an IM240 test on EPA's transportable dynamometer; in nearly all cases, the TM240 test was given on the same day the vehicle was selected for program participation. Once the vehicle was given the IM240 test, it was transported to a repair garage where it was given the two-speed BAR90 idle tests and visual Smog Checkinspection. In this unique data set, remote sensing, IM240, and BAR90 data were available for the same vehicles, with emissions readings taken on nearly the same day. The average pre-repair IM240 emissions rates ofthe vehicles in this study were 70, 6.2, and 2.0 g/mi for CO, HC, and NOX, respectively. Eighty-six percept ofthe vehicles in the program failed the IM240 testusingEPA's 1997 standards for the IM240; 66%, 78°/O, and 27% failed for CO, HC, and NOX (even though the vehicles were not stopped for NOX emissions), respectively. Ninety-six percent of the vehicles that participated in the program failed the BAR90 inspection. This percentage includes those that failed a functional test, shown as an underhood failure in Figure 4-3. Of the five vehicles that passed the BAR90 test, four failed the IM240. Seventy-three percent ofthe vehicles in the program were classified as being "tampered" with or "arguably tam- pered." Additional data on pass/faiT rates are shown in Figure 4-3. This study showed that remote-sensing identifications of high CO- and/or HC-emitting vehicles were confirmed in 86% to 96% of IM240 and BAR90 emissions tests administered on the same or the next day. Arizona High-Emitter Program The state of Arizona implemented a remote-sensing program to identify high-emitting vehicles in the Phoenix area, starting in ~ 995. The program was terminated ~ years later by state legislators, because of problems including costs in the final year of over $300 per high-emitting vehicle identified and
114 100% 80% 60% ·~ 40% to Ohio Evaluating Vehicle Emissions I/M Programs IM240 Test Result IM240 Test Result (RSD C0>4%) (RSD HC>O. two) ~/~ BAR90 Idle Test BAR90 Idle Test (RSD C0>4%) (RSO HC>O.1%) ~ Fail CO and HC · Fail CO only E3 Fail HC only E! Fail UH only ~ Pass FIGURE 4-3 Results from Orange County study showing percentage of vehicles identified by remote-sensing device (RSD) that failed IM240 and BAR90 emissions tests. UH denotes under-hood inspection. false failures attributed to cold-start operating conditions and mismeasured HC emissions. Appropriate remote-sensing sites were scarce, sites were lost as freeway ramps were widened, and the program was continually looking for new sites is. Walls, Arizona Department of Environmental Quality, personal commun. February 16, 2001~. During a period from mid-May ~ 998 through early June ~ 999, over 2 mil- lion valid remote-sensing testrecords were collected, but only 2,987 vehicles were identified as high emitters (Wrona ~ 999~. Owners of vehicles identified as high emitters were sent letters ordering them to submit their vehicles for IM240 testing within 30 days. Abouthalf(55°/O) of vehicle owners responded within this time period; en additional 15-20% of vehicle owners compliedIater, after their vehicle registration was suspended. Of vehicles that reported for testing, 42% passed the initial IM240 test, although a survey indicated that one- third of these vehicles underwent repairs prior to the test (Wrona ~ 999~.
Emerging Emissions Testing Technologies ~15 Use of Remote Sensing to Screen for Clean Vehicles Remote sensing is being used to identify clean vehicles so that they may avoid visiting an emissions test station for scheduled testing. For example, in the St. Louis area, a clean-screen program has been in operation since April 2000. If two or more successive low-emissions readings have been measured by remote sensing, the vehicle owner can opt to be excused from scheduled emissions testing. Daily locations end hours of operation of remote-sensing vans are advertised vie a web site. This clean-screenapproachmaybeprefer- able to issuing blanket exemptions to all vehicles of specified model years. ALTERNATIVE APPROACHES FOR CONTROLLING LIFETIME EMISSIONS An I/M program attempts to ensure that a vehicle's emissions-control system is operating properly throughout the vehicle ' s lifetime. There are other approaches to controlling lifetime emissions. Shifting the burden of responsibil- ity from vehicle owners to the manufacturers is one approach. Although requiring manufacturers to maintain vehicles throughout their lifetime might tee unlikely, increasing warranties on emissions-controT systems to 200,000 miles could accomplish at feast pert ofthis objective. Adopting national policies that force older vehicles from the fleet is another option. Vehicle scrappage pro- grams have been used on only a limited scale in the United States. Other countries have taxation and inspection policies that help to maintain a young vehicle fleet, but these policies tend to be in effect in places with a large do- mestic automobile production sector (European Commission 1997; JAMA 2000~. Developing outreach and financial-support programs targeting owners of high-emitting vehicles could be used to reduce the negative incentives for those needing emissions repairs. These and the methods described earlier in this chapter could be used to control lifetime emissions from vehicles and reduce or eliminate the need for traditional I/M testing. SUMMARY Traditionally, I/M programs have used tailpipe emissions tests, often ac- companied by visual underhood inspections, to assess vehicles registered in the
116 Evaluating Vehicle Emissions I/M Programs program area. This approach is inefficient and costly because of the skewed distribution of emissions across the vehicle fleet; 10-20 mustbe tested to iden- tify one high-emitting vehicle that is a candidate for repairs. A variety oftechnologies that matured during the ~ 990s will affect emis- sions testing regimes in the future. These approaches include developing profiles of vehicles likely to have high or Tow emissions, use of OBD systems to detect and help guide repairs of emissions-related malfunctions, and the use of remote sensing to identify vehicles most likely to fail traditional tai1nine emissions tests. ~ ~ -r -r - The most significant form of profiling to date has been the excusing of new vehicles (typically up to 4 years of age) from regular I/M program testing. Smaller effects, if any, on emissions benefits and program costs have resulted from profiles that rely on inspection history and results oftesting for vehicles of the same make, model, and mode! year. Although OBD systems are present on an increasing number of vehicles, unresolved questions remain concerning their usefulness as a replacement for traditional emissions tests. These systems detect malfunctions "likely" to lead to increases in emissions above certification levels, but no actual emissions measurements are made. Studies of emissions levels on vehicles with MTEs illuminated have shown that most of these vehicles do not have emissions much above their certification standard. A separate EPA study done with data from Wisconsin's I/M program showed very little overlap between vehicles that failed the IM240 and the OBD tests. The CDPHE also found a similar result. instituting an OBD I/M program that fails a large number of marginal emitters could undermine a commitment to find high emitters and ensure that they are the repaired. Instituting an OBD I/M program that failed to detect high emitters could do the same. Furthermore, a critical human factor for OBD systems is the motorist's response to the MIL. The results of these initial studies emphasize the inadequacy ofthe current data set for assessing the effectiveness of OBD for I/M testing. The results also emphasize that much additional information is required before OBDIT's reliability and effec- tiveness canbe quantified in MOBILE. The modeling of OBD I/M options in MOBILES is discussed in the following chapter. Roadside remote sensing has been shown to measure CO emissions reli- ably, with less certain results now available for HC and NOX. Development of remote measurement capabilities for PM remains an important research priority. A variety of issues require careful attention in remote-sensing study design: site selection, effects of engine Toad, attention to quality assurance and
Emerging Emissions Testing Technologies 117 quality control, and achieving adequate coverage of the in-use vehicle fleet. Studies in which vehicles suspected to have high emissions based on remote- sensing measurements are pulled over for further roadside testing have con- firmed that remote sensing can identify vehicles likely to fail emissions tests with a success rate of 80-96%.
