Comparison of vehicle activity and emission inventory between Beijing and Shanghai.

ABSTRACT

Vehicle emission inventory is a critical element for air quality study. This study created systemic methods to establish a vehicle emission inventory in Chinese cities. The methods were used to obtain credible results of vehicle activity in Beijing and Shanghai. On the basis of the vehicle activity data, the International Vehicle Emission model is used to establish vehicle emission inventories. The emissions analysis indicates that 3 t of particulate matter (PM), 199 t of nitrogen oxides (N[O.sub.x]), 192 t of volatile organic compounds (VOCs), and 2403 t of carbon monoxide (CO) are emitted from on-road vehicles each day in Beijing, whereas 4 t of PM, 189 t of N[O.sub.x], 113 t of VOCs, and 1009 t of CO are emitted in Shanghai. Although common features were found in these two cities (many new passenger cars and a high taxi proportion in the fleet), the emission results are dissimilar because of the different local policy regarding vehicles. The method to quantify vehicle emission on an urban scale can be applied to other Chinese cities. Also, knowing how different policies can lead to diverse emissions is beneficial knowledge for other city governments.

INTRODUCTION

Since the 1990s, many large Chinese cities have experienced serious air pollution problems due to the rapid growth of automobile ownership. (1) Because of the absence of a database in China, the vehicle emission inventory is difficult to establish with high temporal or spatial resolution. The MOBILE emission model is currently used in China to develop vehicle emission inventory. (2) However, because the emission factors are based on the average speed, the unreliability of MOBILE-modeled results was indicated by a number of independent evaluation field studies. (3-5) The International Vehicle Emissions (IVE) model is instead based on driving cycle. It uses second-by-second driving data to obtain the bin distribution and related emission factors. (6) The basic theory of the model is credible, and several of the theories have been accepted by U.S. Environmental Protection Agency (EPA) to develop a next generation model. (7) Because the emission factors are based on detailed vehicle technology classification, the IVE model can be used by different countries if a detailed vehicle fleet is available. The temporal resolution of the emission inventory is by hour and the spatial resolution is by road types, which are better than with the MOBILE model. A systemic data collection method was created to obtain a basic input data for the IVE model.

Because of the increasingly serious air pollution and the important roles of both Beijing and Shanghai, this study was designed to support estimates of the air pollution impacts of on-road transportation in the two cities. It is also hoped that after the comparison between Beijing and Shanghai, the methodology, conclusions, and policy suggestions can be extrapolated to other Chinese cities.

METHODOLOGY

Because the basic data in China is absent, how to effectively collect representative data is a critical problem. The experimental methods to obtain activity data are discussed in this section. On the basis of the abundant input data, the IVE model is used to develop a mobile emission inventory.

Vehicle Activity Study

Test Area and Routes. To calculate the vehicle emission of an entire city, three representative sections were selected for experiment. The areas represented a generally lower income area, a generally upper income area, and a commercial area of the city. The imbalance between a northern and southern city is obvious in Beijing. A high-in-come area was selected in the northern part of Beijing that includes more expensive apartments, but not the most expensive. A low-income area was selected in the southern part of Beijing. The apartments here are older and relatively low priced. The commercial area is in the heart of the city with a high density of commercial and office buildings. The same method was used in Shanghai to select three areas. Three types of streets were selected in each area, which represent the typical highway, arterial, and residential roads. These streets meet the different levels of national roads standards. The vehicle technology survey was conducted in three selected areas, whereas the driving pattern tests and video study were conducted on nine selected routes; Table 1 shows the test routes selected in Beijing and Shanghai.

Vehicle Technology Distribution Data. The IVE model with basic emission factors included 1371 vehicle types. The vehicle technology distribution survey was designed to develop a representative distribution of vehicle type, engine size, age, and technologies of the operating fleet in Beijing and Shanghai; these parameters can greatly influence emissions. The technology distribution was then combined to calculate average emission factors for the vehicle fleet. Two approaches were used. Vehicles were videotaped on selected streets from 6:00 a.m. to 8:00 p.m. and the videotapes were reviewed to count the numbers of the various types of vehicles. Simultaneously with this data collection process, parking areas, taxi drivers, and bus and transportation companies were surveyed to collect specific technology information. Detailed methods and data are summarized in Table 2.

Driving Pattern Data. The main objective of obtaining driving pattern data was to collect second-by-second speed and altitude of the main types of vehicles operating in Beijing and Shanghai throughout the day. Vehicle driving patterns were measured using computerized global positioning satellite (CGPS) technology. This technology allows the measurement of vehicle location, speed, and altitude. (8) With the satellite number monitoring, the validity of second-by-second data can be judged. Several ordinary drivers with their vehicles were involved in this research. The passenger cars were driven along selected roads and followed cars randomly, whereas the trucks, taxis and buses operated on their normal daily routes and during their normal work time (Table 2).

Vehicle Start Patterns. Between 10% and 30% of vehicle emissions come from vehicle starts in the United States and likely elsewhere. (9) Thus, it is important to understand vehicle start patterns in an urban area to fully evaluate vehicle emissions. The vehicle engine start patterns were collected using devices known as vehicle operating characteristics enunciators (VOCEs). The VOCE units are equipped with microprocessors that sense and record vehicle system voltage on a second-by-second basis. A VOCE unit is plugged into the cigarette lighter or otherwise hooked into a vehicle's electrical system. The voltage fluctuations in the electrical system can be used to estimate when a vehicle engine is on and off. The VOCE data is then used to determine when vehicles start, how long they operate, and how long they sit idle between starts. Table 2 shows the valid data number of the start pattern study.

Emission Calculation

On the basis of the basic data collected above, the IVE model was used to develop the mobile emission inventory. The IVE model uses a calculation of the power demand on the engine per unit vehicle mass to correct for the driving pattern impact on vehicle emissions. This power factor is called vehicle specific power (VSP), which can be calculated from driving pattern data (eq 1).

VSP = v[1.1a + 9.81 (atan(sin(grade))) + 0.132] + 0.000302[v.sup.3] (1)

where grade is equivalent to ([h.sub.t = 0] - [h.sub.t = -1])/[v.sub.(t = -1 to 0 sec)], v is the velocity (m/sec), a is the acceleration (m/[sec.sup.2]), and h is the altitude in meters.

VSP is the most important variable related to the vehicle's base emission rate, and was selected as the basis for further analysis. (10) To further improve the emissions correction for vehicle driving, a factor denoted as vehicle stress was developed. Vehicle engine stress (ES) uses an estimate of vehicle revolutions per minute index (RPMIndex) combined with the average of the VSP in the 5 sec before the event. Ultimately, the driving data for each vehicle type studied is broken into one of 20 VSP classes and one of 3 ES classes. Thus, each point along the driving route can be allocated to one of 60 driving bins. The emission rates are different with these 60 bins. The IVE model allows the adjustment for base emission rate. About 20 items are optional, including fuel characteristics, inspection and maintenance class, road grade, local environmental variables, and so on. A large number of research results were summarized to get correction factors. (6)

With the base emission rate, correction factors, vehicle activity, and distribution, the emission inventory can be developed for different resolutions.

RESULTS

The results are analyzed from three aspects: road, vehicle, and emission inventory. The system of emission inventory is the administrative region of city. All the statistical data used here are fit for the administrative region of the city.

Comparison of Road Conditions

In the last 10 yr, both Beijing and Shanghai have constructed complex street networks to support vehicle traffic. Ring roads 3-5 were completed in Beijing from 1993 to 2003. The outer ring road and a very long overpass road were constructed in Shanghai between 1995 and 2002. These roads have greatly increased the highway proportions in the two cities. (11)

As Table 3 shows, Beijing has a higher percentage of highway and arterial roads than Shanghai. The percentage of each road type was used to weigh the vehicle type distribution and also the emission from each road type.

Comparison of Vehicle Fleet and Activity