Skip to main navigation

2003 I/M Failure Rate Analysis Report

Introduction

The New York State enhanced I/M program applies to the nine counties of the New York Metropolitan Area (NYMA). The program consists of two components: a loaded-mode transient emissions test (NYTEST) and a single-speed idle emissions test (IDLE). The transient test is more widely used, as the idle test is applicable only to light duty vehicles (LDVs) and light duty trucks (LDTs) with model years prior to 1981, those equipped with four-wheel drive or non-disengageable traction control, and to heavy-duty vehicles (HDVs). Vehicles that are more than 25 model years old, less than 2 model years old, or diesel fuel-powered, are exempt from either emissions test.

The NYMA data which was analyzed for this report covers the period from January 1, 2003 through December 31, 2003. A total of 3,806,564 vehicles were emissions-tested during this time period.. Of the light-duty fleet, 3,291,283 vehicles (86.46%) were transient-tested, while 515,281 (13.54%) were idle-tested. Of the 3,806,564 tested vehicles, 2,599,991 (68.30%) were LDVs, 1,157,185 ((30.40%) were LDTs, and 49,388 vehicles (1.30%) were HDVs. These numbers excluded one 1980 model year and 13 2003 model year vehicles which were in the database.

In this report, failure rates are determined in two ways. The reported failure rate is based on counts of reported emissions test failures, with those failing the test denoted as "F" in our program database. The calculated failure rate is determined by cutpoint exceedances for each pollutant species (reported emissions value exceeding the cutpoint) as calculated by a computer program. This report presents the results of the calculated failure rates, unless otherwise noted.

A summary of the Enhanced I/M Program cutpoints used during 2003 is provided in Appendix A. New transient test cutpoints for HC and NO_x were implemented on April 1, 2003.

Transient Tests

A vehicle will fail the transient emissions test when at least one of the reported emissions values of HC, CO, or NO_x (measured in grams per mile [g/mi] for the entire driving cycle) exceeds the appropriate composite cutpoint value. However, such a vehicle will still pass the test if the reported emissions values for the failed pollutant species do not exceed their respective phase 2 cutpoints for the last 146 seconds of the test. It is important to remember that test-based failure rates are not the sums of the failure rates from each of the three pollutant species, since the test-based failure rates may contain multiple species failures.

Appendix B provides the calculated failure rates for transient-tested LDVs and LDTs, by model year, for each pollutant species, as well as the test-based failures; Figure 1 graphically displays that same information.

As can be seen from Figure 1, LDV failure rates differ from LDT failure rates for model years 1981-1983. This is largely a result of the higher failure rates for HC and CO for LDVs in those same model years, and may also be skewed higher than later model years because of the smaller sample sizes (# vehicles) represented by the earlier model years. For instance, the average number of transient-tested LDVs from 1981-1983 was 5,995, while the average number of transient-tested LDVs from 1984-86 was dramatically higher - 28,189.

The spike in HC failure rates for LDTs for model years 1983-84 can be explained by the cutpoint change for those model years, in which the HC "pass" threshold is reduced from 3.40 g/mile to 1.60 g/mile for model year 1984.

LDV failure rates for CO increase from model years 1982-83 because cutpoints for CO are reduced by half for model year 1983. LDT failure rates for CO increase sharply from 1983-84 because cutpoints decrease from 70 g/mile for model year 1983 to 40 g/mile for model year 1984. CO failure rates also show a modest increase for both vehicle types for model years 1995-1996, which reflects the lower cutpoints for 1996-and-newer vehicles.

NO_x failure rates for LDTs show a sharp spike from model year 1987 to model year 1988; this reflects a change in cutpoints for these model years from 4.50 g/m to either 2.50 or 3.50 g/m (depending on vehicle classification as LDT1 or LDT2).

Small differences were noted when we compared the calculated and the reported failure rates as shown in Tables 1 and 2. Since the same reported emissions readings, the same set of transient test cutpoints, and the same "two-way-to-pass" algorithm, as described above, were used by both the NYTEST software and our computer programs, the calculated and reported test-based failure rates should be exactly the same.

However, as shown in Table 1, small differences between the calculated and reported rates exist. The overall failure rate differences were smaller for LDVs (reported 7.92% vs. calculated 7.71%) than for LDTs (reported 7.73% vs. calculated 7.25%). As shown in Table 2, the reported transient test results were the same as the calculated results for 99.75% of the LDVs and 98.66% of the LDTs.

The differences (Table 1), as reflected by the higher fail rates for "reported" results, are likely a reflection of errors committed during the inspection process which are then reflected in the field data. Some vehicle information - such as model year, sales class, and test vehicle weight - are critical in determining the vehicle's emissions test pass/fail status. If these reported data fields differ from what would be indicated for those fields using the Vehicle Identification Numbers (VINs), they were subsequently corrected to the indicated VIN parameters during our quality assurance (Q/A) review of the data. For example, the reported field data might list a Dodge "Caravan" or "Grand Caravan" as being in sales class "V" (LDV), when, in reality, this vehicle is properly classified as an LDT. In addition, there are cases where the model name is incorrect; for example, an LDT may be listed as a "Dodge Intrepid", with a sales class of "V". However, according to the VIN, it is actually a "Dodge Caravan" which is classified as an LDT. As the cutpoints for LDVs are more stringent than those for LDTs, an LDT that is incorrectly identified as an LDV and tested as such, would likely fail the emissions test since the cutpoints are more stringent. Since our Q/A procedure corrects the sales class for these LDTs and calculates the pass/fail status based upon the LDT cutpoints, this may explain the larger difference seen between the calculated and reported failure rates for LDTs. The higher percentage of LDVs and LDTs with "reported = fail" and "calculated = pass" results indicates that a greater number of vehicles are being incorrectly assigned to the vehicle class which has a more stringent cutpoint, thus "failing" the vehicle.

Table 2 also reveals higher discrepancies between calculated and reported fail rates for LDTs as opposed to LDVs. This may arise from a requirement that LDTs which are 1996-and- newer model years have the "test vehicle weight" correctly assigned in order to determine the applicable cutpoints. Our Q/A work has shown that test vehicle weight classifications have been incorrectly made in numerous instances. The incorrect designation of the "test vehicle weight" would result in an incorrect cutpoint designation, and an improper "fail" or "pass", depending on how that assignment was made.

In short, the differences between calculated and reported fail rates can be attributed to the correction of vehicle information as a result of our data Q/A procedure, with those corrections being reflected in the calculated failure rates.

Table 3 provides the calculated failure rates for LDVs and LDTs by month, for each pollutant species, as well as the test-based results. Figure 2 graphically displays the same information.

Failure rates rise sharply from March to April for both LDVs and LDTs due to the cutpoint implementation on April 1, 2003. Failure rates, for both LDVs and LDTs, are also higher during the warm weather months (May to August), and then decline from September to December. For individual species, HC failure rates for LDVs spike sharply from March to April in response to the cutpoint change, then remain relatively static, (and even slightly lower) for the rest of the year. We observed a similar sharp spike in NO_x failure rates for LDTs from March to April. Both LDVs and LDTs show a slight increase in failure rates from April to August for NO_x.

One would expect higher failure rates for NO_x during the warmer months inasmuch as NO_x emissions are caused by high cylinder temperatures. Higher cylinder temperatures can, naturally, be expected during the warmer months of the year (May-August), leading to higher NO_x emissions and, consequently, higher failure rates.

We would also expect to see failure rates decline after August, 2003, as the effect of the cutpoint change implemented on September 3, 2002 completes its cycle through the vehicle fleet. Vehicles inspected prior to that cutpoint change returned to inspection stations to face these more stringent cutpoints for the first time if they were last inspected prior to September, 2002. We would expect to see failure rates drop after August, 2003, as all vehicles which failed the more stringent cutpoints implemented in September, 2002 would already have been subjected to inspections utilizing the September, 2002 cutpoints, and, if they failed the first time, would have been repaired or taken off the road. However, that explanation is complicated by the fact that all vehicles inspected after April 1st, 2003, were also subjected to the new cutpoints implemented on that date, thus mitigating the drop in failure rates that might otherwise be expected to occur in September, 2003.

We also would expect to see the spike in failure rates observed from March - April as a reflection of the cutpoint change implemented on April 1^st, 2003.

A summary of transient test failure rates by equipment vendor is presented in Table 4. SPX had the lowest overall failure rates among the three vendors, for both LDVs and LDTs, as well as for all individual species failure rates, except LDT failure rates for NO_x. The failure rates for ESP and SUN are very similar for HC and CO (LDVs), as well as LDT failure rates for HC. NO_x failure rates do not exhibit as close a correlation in failure rates between vendors for either LDTs or LDVs.

Finally, a summary of transient test failure rates by county is given in Table 5. The county referred to as "Others" consists of Orange and Putnam counties, as well as counties outside of the NYMA, and includes those tests for which no county was listed. This "others" category consists of 2,366 tests on LDVs and 953 tests on LDTs. The highest failure rates for LDVs are seen in Bronx, Kings, and Queens counties, possibly indicating a somewhat older fleet characteristic of less affluent owners. The highest failure rates for LDTs, however, are seen in Suffolk, Queens, Kings, Nassau and the "Others" counties, for which the above explanation for LDVs does not hold.

Idle Tests

Idle tests are required for:

• all gasoline-powered pre-1981 LDVs and LDTs not older than 25 years;
• those gasoline-powered 1981-and-newer LDVs and LDTs older than 2 years old equipped with four-wheel drive or non-disengagable traction control;
• and all gasoline-powered HDVs, defined by a gross vehicle weight rating of 8,501 pounds or more.

For an idle test, the HC emissions are measured in parts-per-million (ppm) and CO in percent (%). A vehicle fails the idle emissions test when at least one of the reported emissions values for HC or CO exceeds the appropriate cutpoint value. As with the transient test, the idle failure rates are not the sums of the failure rates for each pollutant species because the test-based idle failure rates can consist of multiple species failures.

Appendix C provides the calculated failure rates for idle-tested LDVs, LDTs, and HDVs, by model year, for each pollutant species, as well as the test-based failure rates. Figure 3 graphically displays these same results. All calculated failure rates were based on cutpoints which have been unchanged since 1981.

Table 6 compares the reported and calculated failure rates. As with the transient test, the reported failure rates were slightly higher than the calculated rates. Overall, the failure rates for LDVs and LDTs are substantially lower than those for HDVs.

As shown in Table 7, the differences between the reported and calculated failure rates for idle- tested LDVs (99.19% agreement) were found to be slightly larger for than for the transient test (99.75% agreement) as shown in Table 1. For LDTs, the reverse is true: Transient-tested vehicles show a greater difference between the reported and the calculated failure rates (98.66% agreement) than do idle-tested vehicles (99.28% agreement). As with the transient tests, the largest discrepancy is found with LDTs where the reported value is "fail" and the calculated value is "pass". These differences can be explained by our Q/A procedures which correct vehicle information entered during the inspection process.

Table 8 provides both the calculated failure rates for LDVs and LDTs by month, for each pollutant species, as well as on a test basis. Overall, the failure rates for all three vehicle types are fairly constant over all twelve months.

A summary of idle test failure rates by vendor is presented in Table 9. For LDVs and LDTs, the lowest failure rates were observed with SPX, and the highest rates were from SUN. However, the failure rates for HDVs were higher across the board for SPX than for the other two vendors.

Finally, the idle test failure rates by county are given in Table 10. Again, the county referred to as "Others" consists of Orange and Putnam Counties, as well as those in counties outside of the NYMA, and those with no county listed. The highest failure rates for LDVs were observed in Bronx, Queens and "Other" counties, with the lowest failure rates observed in New York (Manhattan) county. For LDTs, the highest failure rates were in Bronx, Kings and "Other" counties, while the lowest rate was again observed in New York county (Manhattan). The highest rates for HDVs were seen in Richmond, Suffolk, and Westchester counties, and the lowest rates observed in Queens and Kings counties. There were no HDV failures in the "Others" category, but the sample consisted of only 30 vehicles.

Conclusions

The 2003 data reveal an increase in failure rates for transient tests (7.59% vs. 6.67% in 2002) and a decrease in failure rates for the idle test (1.25% vs. 1.76% in 2002). For transient-tested vehicles, this is most likely the result of more stringent test cutpoints being implemented on April 1, 2003.

The lower reported idle test failure rate of 1.25% (which includes both light-duty and heavy-duty vehicles), compared to the corresponding transient failure rate of 7.59%, suggests that the idle test is less stringent and easier to pass than the transient test. The reasons for the reduction in failure rates for idle-tested vehicles in 2003 are not clear, but probably reflect an increasingly smaller number of pre-1981, but newer-than-1978 (less than 25 years old) vehicles as a component of the idle-tested fleet. In fact, there were 2063 vehicles which met that particular criteria for idle testing in 2002, and only 959 vehicles which met this criteria in 2003 (which this report covers), a 53% reduction from the previous year. As such, the idle-tested fleet of vehicles has become correspondingly "cleaner" than previous-year fleets, as better emissions-control technology is incorporated into vehicles. At the same time, this increasingly "cleaner" fleet is being tested against a set of cutpoints which has remained unchanged, and for which the "bar" has not been raised. As a result, failure rates continue to decline for idle-tested vehicles.

The differences between the calculated and the reported test counts of passes and failures for both transient and idle tests are small (less than 1% for all vehicle types) except for transient-tested LDTs, where the "agreement" between calculated and reported tests was only 98.66%, suggesting that some confusion may exist during the emissions inspection process concerning the proper identification of light-duty trucks vs. light-duty vehicles (passenger cars). These differences surface as the result of corrections of vehicle information during data Q/A procedures, and could be reduced by increasing the accuracy of data entry at the emissions inspection facilities.

We expect that as the Enhanced I/M Program continues, failure rates will continue to decline (assuming cutpoints remain unchanged) as cleaner pollution control technology is incorporated into more and more vehicles. Both cutpoints and tailpipe emissions from internal combustion engines have ultimate limits, but neither have yet been reached.

More about 2003 I/M Failure Rate Analysis Report :

• 2003 I/M Failure Rate Analysis Report - Appendices - Appendix pages of the 2003 failure rate analysis report
• 2003 I/M Failure Rate Analysis Report - Figures - Figures page of the 2003 failure rate analysis report

• 
• On This Page:
• • Introduction
  • Transient Tests
  • Idle Tests
  • Conclusions
• Contact for this Page:
• NYSDEC
  Division of Air Resources
  BAQA&R
  625 Broadway
  2nd Floor
  Albany, NY 12233-3259
  518-402-8402
  email us