Quantifying the Impact of High Emitters on Vehicle Emissions: An Analysis of Ecuador’s Inspection and Maintenance Program

Abstract

On-road vehicles are a primary source of urban air pollution. It is known that high-emitting vehicles represent a fraction of the fleet but contribute significantly to the total emissions. Usually, road transportation emission inventories do not capture the impact of these types of vehicles, underestimating emissions. This study introduces a simple method to refine vehicle emission inventories by incorporating data from Ecuador’s Inspection and Maintenance (I/M) program. We analyzed I/M data from Quito to develop a correction factor for the Vehicular Emissions INventory (VEIN) model, accounting for the higher emissions from vehicles that fail inspection. Our analysis shows that while less than 10% of gasoline and 20% of diesel vehicles failed inspection, their emissions were substantially higher; for instance, accounting for reproved vehicles produced 60% more Carbon Monoxide (CO), 18% more Non-Methanic Volatile Organic Compounds (NMVOC), 40% more Particulate Matter with aerodynamical diameter of 2.5 µm or less (PM_2.5), and 34% more or lower than 10 µm (PM₁₀). These findings demonstrate that incorporating I/M data is crucial for accurately quantifying vehicular pollution. The proposed methodology offers a way to create more accurate emission estimates, providing a tool for policymakers to manage air quality.

1. Introduction

On-road vehicles are a dominant source of air pollution, particularly in urban environments where traffic density is high [1,2,3]. The exhaust from cars, trucks, and buses releases a mixture of harmful pollutants that have been linked to significant public health issues, including respiratory and cardiovascular diseases [4,5]. Vehicle inspection and maintenance (I/M) programs are a critical strategy for mitigating the impact of road transport on air quality [6,7]. These programs, implemented in many cities worldwide, mandate periodic checks to ensure vehicles comply with established safety and environmental standards. A primary function of I/M programs is to identify vehicles with high emission levels and require necessary repairs, thereby reducing the overall tailpipe emissions of carbon monoxide (CO), hydrocarbons (HC), nitrogen oxides (NO_X), and particulate matter (PM). Recent studies have validated the effectiveness of I/M programs utilizing On-Board Diagnostics (OBD) and on-board sensing. Jiang et al. (2021) [8] demonstrated that repairing Heavy-Duty Vehicles (HDVs) based on OBD fault codes resulted in NOx reductions of 46–81% and opacity reductions of 43%. Also, Li et al. (2023) [9] utilized on-board sensors to characterize sediment hauling trucks, finding that low Selective Catalytic Reduction (SCR) temperatures significantly hampered NO_X conversion efficiency [9]. The effectiveness of I/M programs is rooted in their ability to address the deterioration of emission control systems that occurs over time with vehicle use. Indeed, the purpose is to enforce emission standards on the circulating fleet, contributing to cleaner air in urban areas.

A key challenge in managing vehicular pollution is the disproportionate impact of a small fraction of “high-emitting” vehicles, which can be responsible for approximately half of all harmful pollutants emitted by the entire fleet [10,11,12,13]. These high emitters are often older or poorly maintained vehicles whose pollution control systems are malfunctioning, although other reasons, including poor manufacture, contaminated fuel, and even tampering, may contribute [13,14]. Although emission inventory models like the Vehicular Emissions INventory (VEIN), which often use emission factor databases such as the COmputer Programme to calculate Emissions from Road Transport (COPERT), provide essential, large-scale estimates of vehicular pollution, they may not fully capture the real-world emissions from these high-polluting vehicles [15,16,17]. Therefore, incorporating data from on-the-ground I/M programs is crucial for refining these models and obtaining a more accurate characterization of a region’s emissions profile. Models typically either ignore higher emitters or they apply a common correlation, e.g., a multiplier, applied to the whole fleet, to up-shift emission factors globally, so treating high emitters as identically distributed everywhere. So, the ability to recalibrate these models at the city-level, where traffic management policy can often be most effectively targeted, could be of significant value when scoping, e.g., incoming emission zoning activities [18,19].

In Ecuador, the vehicle Periodical Technical Inspection and Maintenance (I/M) program is a mandatory regulatory framework designed to ensure that motor vehicles meet minimum safety and emission standards. The program is under the Resolution 925-ANT-DIR-2019 for Vehicle Technical Inspection Regulations in Ecuador [20]. The annual inspection is mandatory for all vehicles older than two years. The IM evaluates technical aspects, including pollutant emissions, concentrations of CO and HC, mileage, and safety characteristics, affecting gasoline and diesel vehicles. Vehicles that meet all inspection criteria receive a certificate of compliance. If deficiencies are detected, a conditional certificate is issued, requiring corrective repairs and reinspection, with up to four attempts permitted. Persistent non-compliance results in the banning of vehicles from circulation. Despite their comprehensive scope, I/M programs around the world exhibit limited effectiveness in detecting and addressing high-emitter vehicles, which are often responsible for a disproportionate share of urban air pollution [14].

Ecuador is a country with a growing vehicle fleet and faces significant air quality challenges, particularly in its densely populated urban centers like Quito and Guayaquil [21]. In 2021, Ibarra-Espinosa et al. [15] used the VEIN model to characterize vehicular emissions on a national scale, providing valuable baseline information. As detailed in the accompanying manuscript, these studies relied on COPERT emission factors, corrected for local conditions, to estimate emissions, assuming vehicles passing IM. Although Ecuador has a mandatory annual vehicle inspection program, the impact of failures and the prevalence of high-emitters have not been integrated into emission inventories. This new study will build upon the previous work by incorporating data from the I/M program. This will allow for the development of a correction factor to adjust the emission estimations from the VEIN model, accounting for the segment of the fleet comprising persistently high emitters and likely responsible for a larger share of pollution. It is hypothesized that this will result in higher and more realistic emission figures, providing a more accurate tool for policymakers to address air quality management and effectively target emission reduction strategies.

2. Materials and Methods

We obtained the I/M data from 2019 from the Environmental Secretary of Quito, Ecuador. The dataset was analyzed to understand temporal trends over the years. We also studied the distribution statistics. The idea of I/M programs is to advise on the repair or removal of high-emitting vehicles from circulation. However, there is a time before any car can be detected and then repaired, depending on the I/M specifications. So, even if the local I/M is highly effective, there is still likely to be a percentage of high-emitting cars in circulation at any given time. Here, we propose a methodology to correct emission factors accounting for the high-emitter cars in the emissions data. Since the I/M database includes exhaust measurements of all cars, approved and reproved during a given year, we calculated the emissions ratio of reproved/approved cars by year of use. Then, as we know the number of each type of vehicle, we know the percentage of high-emitting vehicles. Then, the final emission factor is weighted against high-emitting vehicles and their fleet participation, as shown in Equation (1):

E = EF·[PERC_A + (IM_R/IM_A) ·PERC_R]·VEH·LKM

(1)

where E is the emissions (g) and EF is the emission factors, representing vehicles with approved IM. For the purpose of this study, the emission factors come from Ntziachristos and Zamaras [17], available in VEIN v1.5 [16]. PERC_A and PERC_R are the percentage of the fleet that approves or reproves IM, respectively. IM_A and IM_R are the emissions for the approved and reproved vehicles from the IM program. VEH is the number of vehicles in circulation during the period of time, in this case, by year. LKM is the distance traveled by the vehicles by year in km. This equation is extended for all the possible combinations of vehicle type, fuel, technology, year of use, and deterioration, which are indeed considered when running VEIN. However, Equation (1) intends to express a generic formulation, but explicitly applied for each age of use. To further expand the reproducibility of this study, readers can find the project “ecuador_mdpi” at https://atmoschem.github.io/vein/reference/get_project.html (accessed on 23 December 2025). Since opacity is an optical measurement that reflects particulate matter [22], we applied our methodology for particulate matter.

VEIN is an open-source vehicular emissions model that was developed in Brazil and expanded with methodologies and emission factors from different parts of the world [23]. As a result, it currently has about 60 k downloads. VEIN [16] is an R package that imports spatial features from sf R packages [24,25], with bindings for GDAL (https://gdal.org/, accessed on 23 December 2025), GEOS (https://libgeos.org/ accessed on 23 December 2025), and PROJ (https://proj.org/, accessed on 23 December 2025) for spatial processing. Furthermore, VEIN also imports the data.table R package [26], which provides an optimized and fast approach to dealing with data. VEIN also incorporates Fortran subroutines with parallel processes to calculate the emissions efficiently.

To account for the effect of high-emitting cars on emissions, we developed two emissions inventories for Ecuador. The base scenario assumes emission factors with normal degradation over time and no high emitters. The second scenario includes the percentage of the fleet with high emitters. Finally, we perform a characterization of emissions and their comparison. We also included a comparison with road transport sector emissions from two global inventories, EDGAR v8.1 and CEDS v2024_07_08 [27,28].

3. Results and Discussion

3.1. Vehicle Inspection and Maintenance Program (I/M)

Diesel and gasoline cars show an expected pattern, with a larger proportion of approvals being awarded to newer vehicles, as shown in Figure 1. Data on older vehicles in circulation is very limited, around 0% for diesel before 1994 and 0% for gasoline before 1965. Hence, we limited the data on gasoline cars to that between 1965 and 2020, and diesel cars between 1994 and 2020. The average of approved gasoline vehicles was 91.22% while for the diesel was 81.99%.

In gasoline vehicles, the CO is measured as a percentage (%), while HC is measured in ppm. In diesel vehicles, opacity is measured as a percentage, which represents the amount of light blocked by the exhaust plume. Therefore, the opacity is a good indicator of particle matter; the darker the plume, the more soot. While there are studies that estimate particle matter from opacity measurements [16], in this study, we applied the ratio of reproved to approved as an indicator of particle matter emission factors for high-emitting diesel vehicles. The averaged values by age of CO, HC, and opacity are shown in Figure 2, showing higher values for reproved vehicles. Specifically, on average, the ratio reproved/approved for CO, HC, and opacity was 3.9, 6.2, and 3.6, respectively. Also, the CO and HC in gasoline vehicles pattern was similar, where the values of reproved vehicles are high in the first years, then drop with about 10 to 20 years of use, being greater the older the vehicle. The increase in emissions for cars from around the year 2000 and older is very likely due to the degradation of emission control systems, like the catalytic converter, over time.

3.2. Emissions Inventory

The emissions inventory included a combination of vehicles no older than 40 years, covering 12 months, 62 types of vehicles, 12 pollutants, 25 regions, and 2 scenarios, resulting in a database of 42 million estimations. The visualization of the emissions by scenario and pollutant is shown in Figure 3. The values in our study are labeled as IM, to account for the effect of high-emission vehicles only available in the IM dataset, and NOIM, which assumes all vehicles would pass the IM program. While less than 10% of gasoline and 20% of diesel vehicles did not pass the IM inspection, their impact on emissions is more significant. The average shows that emissions are 31% higher, including high emitters, but the percentage varies greatly for each pollutant. Specifically, the largest increment is 65% for CO, while the smallest change was 1% for NMVOC. Since I/M did not measure CO₂ or NO_X, no change was expected for these gases. Regarding CO₂, we also see that our estimation aligns well with values in the literature, which demonstrates the strength of our methodology, and the differences in our estimations are due to emission factors, finding small differences from CEDS of 11% and from EDGAR of 5% lower than our estimations. However, for all the other pollutants, our values are lower than those in the literature. Also, it is worth noting here that VEIN can directly estimate NO and NO₂ and more pollutants, applying the methodology of the European Emissions Guidelines [17]. Our estimations are separated by type; in this way, we see that exhaust emissions are the most important for all the pollutants, while evaporative emissions play a secondary role for NMVOC, the same as wear emissions for particulate matter PM₁₀ and PM_2.5.

The emissions by vehicle type can be seen in Figure 4, comprising 62 categories. Appendix A includes a definition for each vehicle category. First, we see that diesel vehicles change emissions only for particulate matter, Black Carbon (BC), and Organic Carbon (OC), as a result of using the opacity as a proxy for measurement. For CO and NMVOC, we see that emissions accounting for high-emitting vehicles affect only gasoline vehicles. Lastly, CO₂ and NO_X are not affected by the scenarios because there are no measurements of these gases. Nevertheless, NO_X is an important gas for atmospheric photochemistry [29] and should be added to the Ecuadorian IM program. Furthermore, NO_X is usually considered in IM programs, resulting in proper characterization and recommendations for control in Europe [30]. Furthermore, we see that electric vehicles do emit particulate matter from tires, brakes, and road wear.

Figure 5 shows the breakdown of total emissions by vehicle age for both the IM and NOIM scenarios. The highest emissions for most pollutants come from vehicles that are between 0 and 10 years old, which is likely because there are more of these newer cars in the fleet. Across almost all pollutants and age groups, the emissions are higher in the IM scenario, which accounts for high-emitting vehicles. This difference is most noticeable for pollutants such as CO, where older vehicles that fail inspection contribute significantly to total pollution levels.

4. Conclusions

This study presents a method for improving vehicle emission inventories by using data from Ecuador’s Inspection and Maintenance (I/M) program. Including data on vehicles that failed the inspections allowed us to more accurately account for emissions from high-emitting vehicles, which are often overlooked in standard models. Our analysis of the I/M data from Quito revealed that while a relatively small percentage of the fleet fails the annual inspection (less than 10% of gasoline vehicles and less than 20% of diesel vehicles), accounting for reproved vehicles produced 60% more CO, 18% more NMVOC, 40% more PM_2.5, and 34% more PM₁₀. Our CO₂ estimates were similar to those from global inventories, such as CEDS and EDGAR, suggesting that our basic fleet and activity data are correct. However, our estimates for other pollutants were lower, indicating that the emission factors used in global models may differ from those used locally in Ecuador. Our results also show that failing to include I/M data can lead to a significant underestimation of actual on-road emissions. We also propose that the method presented here offers a way to create more realistic emission figures and provides policymakers with a better tool for developing effective strategies for managing air quality and reducing transportation-related pollution and assessing the need for more targeted action to address high-emitting vehicles specifically. The contributions of high emitters predicted using this approach are broadly comparable to those reported in studies from other countries where dedicated campaigns have been undertaken to directly measure emissions [31,32,33]. The methods described in this paper are also readily transferable to other similar areas, where local traffic-related air-pollution management action is ongoing but local on-road evidence is limited, making this approach a viable option for policymakers looking to generate early estimates of the likely effectiveness of incoming environmental interventions.