Atmospheric Emission Characteristics and Control Policies of Anthropogenic VOCs from Industrial Sources in Yangtze River Delta Region , China

A bottom-up inventory of atmospheric emissions of industrial MNVOCs in Yangtze River Delta (YRD) region was established for the period of 2010–2012. Results indicated that the total emissions of industrial VOCs increased from 3.34 Tg in 2010 to 3.99 Tg in 2012 at an annual average rate of 9.3%. Furniture manufacturing, machinery equipment manufacturing, architectural ornament, and petroleum refining were estimated to be the key sources, which contributed 58.6% of the total emissions in 2012. Based on the population database, the emission inventory was gridded into 1 km × 1 km grid cells with ArcGIS, which indicated that VOC emissions exhibited remarkable spatial and temporal characteristics. The most polluted areas were mainly centered in Shanghai, Nanjing, Ningbo, Hefei, and the Hangzhou and Taihu lake basin, which were the most developed and industrialized regions in the YRD. VOC emissions from specified anthropogenic sources under three different control scenarios for the different emission targets in 2030 were projected. The estimated average removal efficiency of the control technologies for each anthropogenic source should improve by 68.1% (scenario 2030B) and 79.0% (scenario 2030C). An entire process-control management scheme needs to be established based on the investigation of industrial processes, and the proportion of the application should be improved in the future.


INTRODUCTION
Over the past three decades, China experienced rapid economic development, as well as dramatic growth of energy consumption, extensive industries, and number of motor vehicles, accompanied by increased emission of ozone precursors and aerosols (Li et al., 2016); these factors cause frequent regional air pollution episodes, such as haze, photochemical smog, and high ozone concentration.The Yangtze River Delta (YRD) region is one of most developed and heavily-polluted regions in China.Located in the eastern part of the country, the YRD region contributes approximately 2.19% of the total territory area and more than 21.0% of the national gross domestic product (GDP) (National Bureau of Statistics of China, 2013).The data from long-term meteorological monitoring sites in the YRD region indicate high concentrations of fine particles and ozone (Gao et al., 2011;Xiao et al., 2011).People with long-term exposure to air pollution are more susceptible to acute or chronic respiratory, cardiovascular, and neurological diseases (Zhou et al., 2011;Zhu and Liu, 2014;Chen et al., 2016).
Volatile organic compounds (VOCs) are crucial precursors to secondary organic aerosol (SOA).An analysis of the influencing factors of VOCs indicates that a significant portion of SOA accounts for the oxidation of VOCs, which contribute 80%-90% to organic particulate matter (Xiao et al., 2011;Liu et al., 2012;Tsai et al., 2012;Huang et al., 2014;U.S. EPA, 2016).Studies have revealed that VOCs are also involved in the photochemical formation of ozone, the reactions of alkoxy radicals and the mechanisms of the gasphase reactions of O 3 with alkenes explained it (Atkinson, 2000).
VOCs are numerous, varied and ubiquitous, which are emitted from various sources, including motor vehicles, industries, and natural (biogenic) sources (mainly trees) (EPA, 2015).Some studies (Kesselmeier and Staudt, 1999;Baudic et al., 2016;Cappellin et al., 2017;Sobanski et al., 2017) have investigated the biogenic VOC emissions of different areas such as the Paris megacity, south-west Germany and so on, and analyzed the significant role they played in the photochemical reactions.For the anthropogenic VOCs, several VOC emission inventories were established in recent years, which provide information on emission sources and characteristics (Chen et al., 2014;Wang et al., 2014a, b).Basing on the emission inventory of China, a previous study determined that industrial and domestic solvent use contributed 28.6% to the total NMVOC (are identical to VOCs, but with methane excluded) emission in China (Wei et al., 2008), and Shandong, Jiangsu, Guangdong, and Zhejiang were the provinces with the highest VOC emissions (Qiu et al., 2014a;Wu et al., 2015).The inventory of VOCs in the YRD for 2007 also illustrated that industrial sources, including fuel combustion facilities and non-combustion processes, contribute roughly 69% of the total VOC emissions (Huang et al., 2011).Although significant progress was achieved in estimating VOC emissions and characterizing the spatiotemporal variations over the YRD region, some limitations persist.For example, integrated and detailed investigations of the VOC emission characteristics in atmosphere are hard to make, and the corresponding policies concerned with VOC control in the YRD region are difficult to implement.Till now, the control strategies to reduce emissions have been mentioned by only few researchers (Cheng et al., 2015).
In this paper, a bottom-up emission inventory from anthropogenic sources of high spatial resolution was established for 22 cities in the YRD region in 2010-2012.Population density-based Geographic Information System (GIS) was used to present the distribution of the VOC emissions in the YRD region.With the great practical significance regarding the aim of air quality pointed out in the government scheme, a scenario analysis of the expected reduction of VOC emissions in the target year 2030 was proposed, and the potential reduction of the key emission sources in the YRD region was calculated.

Methodology
In this study, the atmospheric emissions of industrial VOCs were calculated by applying a bottom-up methodology (Townsend-Small et al., 2015) with a refined activity database and an upgraded specific emission factor database., , , (1 ) where j represents the province, k represents a specific sector, E represents annual total emission of VOCs, A represents the activity data, EF represents the emission factor, and η represents the removal efficiency.In this study, the domain covered 22 cities, including Shanghai, Hangzhou, Ningbo, Jinhua, Shaoxing, Zhoushan, Jiaxing, Huzhou, Quzhou, Taizhou, Nanjing, Suzhou, Wuxi, Changzhou, Nantong, Yangzhou, Yancheng, Zhenjiang, Huaian, Taizhou, Hefei, and Maanshan, which all became members of the YRD Economic Coordination Association in 2010.
The source categorization listed in Table 1 is based on the concept of source-tracing (Chen et al., 2012;Wang et al., 2012;Zhao et al., 2012), which includes all types of sources from four major links: production of VOCs, storage and transport, industrial processes using VOCs as raw materials, and use of VOC-containing products.After reviewing the types of available activity data in the various cities, the industrial VOC source categorization considered in this study are aggregated into 78 particular industrial sources by relating city-specific activity data to VOC sources associated with these activities, which effectively demonstrate the present situation of VOC emission around the YRD region.

Compilation of Activity Data
Table 1 presents the types of urban activity data, which represent the activities associated with each source.Most of the data were obtained from the yearbooks of the cities in 2010-2012, which were developed by the local government.These data included product output and fuel consumption (National Bureau of Statistics of China, 2010-2012a, b,  c, d).In some cases such as the annual output of ethylene, benzene, synthesis, and other are adopted from official statistical records and available sources, such as data from the Urban Bureau of Statistics of China and Industry Association.In other cases, the data were obtained through estimation.For example, when the amount of the adhesives and the ratio for various purposes are available, the distributions of different sources were calculated.

Determination of Emission Factors
Emission factors describe the amount of emissions associated with one unit of a particular statistic, which is the foundation of an accurate emission inventory.In this study, we identified the emission factors of different sources.We consider the local measurement results first, followed by Chinese emission factors.However, when the local results and Chinese emission data were not available, the emission factors of other countries (e.g., USEPA (United States Environmental protection Agency) and EEA (European Environment Agency)) or regions (where the emission standards and the production process are the same in the domestic condition) were adopted.Emission factors and activity data of different sources are listed in Table S1.

Emissions and Source Contributions in 2012
The total industrial VOC emissions were 3.99 Tg in 2012.The top 10 sources are listed in Table 2. Use of VOC-containing products was the major contributor of the four links, which accounted for 68.8% of the total emissions in 2012.Industrial processes using VOCs as raw materials, the production of VOCs, and storage and transport contributed 15.7%, 10.3%, and 5.3%, respectively.The different rates of the four links in 22 cities are shown in Fig. 1.
Furniture manufacturing, architectural ornament, and machinery equipment manufacturing respectively contributed 36.0%,17.7%, and 16.9% to the use of VOCs-containing products, which were the largest emitters.About 33.7% of the VOCs in use of VOCs-containing products were emitted by Shanghai, the largest city in the YRD region, and with the most developed industry.
Chemical pesticide and synthetic fiber were the key sources of industrial processes that use VOCs as raw

Temporal-Spatial Distribution of VOC Emissions in YRD Region
Table 4 lists the VOC emissions during 2010-2012, The spatial distribution characteristics of YRD are shown in Fig. 4. Based on the population database, the emission inventory was gridded into 1 km × 1 km grids using ArcGIS.Fig. 4 illustrates that the VOC emissions exhibited remarkable spatial and temporal characteristics.The most polluted areas were mainly centered in Shanghai, Nanjing, Ningbo, Hefei, and the Hangzhou and Taihu lake basin, which are the most developed and industrialized regions in the YRD.These areas generated approximately 76.1% of the total industrial VOC emissions in three years although these areas cover 40.2% of the territory.Hangzhou (second), Nanjing (third), and Ningbo (fifth) account for 68.7% of the total gross economic production of the YRD region.
Fig. 4 indicates that the Hongkou (36,014 people per km 2 ) and Xuhui (20,292 people per km 2 ) districts had the highest emission densities with the largest population density in Shanghai.The highly polluted areas expanded from the coastal to the inland regions, particularly for the Hangzhou and Taihu lake basin.

Uncertainty Analysis
Compared with estimates of other air pollutants, those of VOC emissions from anthropogenic sources are uncertain because the activity data vary spatially and temporally.Monte Carlo simulation is employed to quantify the uncertainties of VOC emissions for different sectors and variable activity data (Zhao et al., 2011).For example, uncertainties of the activity data and emission factors in 2012 may be fitted with lognormal distributions (Wei et al., 2011).Thus, the uncertainties for the total emissions in 2012 was [-16.4%, 65.6%] at the 95% confidence interval in the YRD region.Huang et al. (2011).The results vary because of the different source classifications and areas covered in these studies.The differences of inventory uncertainty at 95% confidence interval were mainly due to the different industrial source classifications, emission factors, and activity data.Therefore, various manufacturing processes for each sector should be investigated thoroughly, and field tests for each process should be conducted to obtain accurate emission inventory estimates.

Scenario Projection and Proposals for Technology Applications
According to the collected information on activity data, GDP, urbanization, and population (Lei et al., 2005;Zhang et al., 2012;UNDP, 2013;Wu et al., 2015) for 2005-2012, the increasing rates of relevant activities were specified, and three scenarios (2030A, 2030B, and 2030C) were set to project the industrial VOC emissions in the YRD region.The activity data of the 2030A scenario were compiled based on the annual growth rate in 2005-2012, which was characterized by unchanged control technologies based on the level in 2010.For scenario 2030B, we assume that the VOC emissions of each anthropogenic source category in 2030 are equal to those in 2010.According to a previous report (Hao et al., 2014), scenario 2030C indicates the VOC emission of every anthropogenic source in 2030 decreased by 36% compared with that in 2010.The bars in Fig. 5 indicate the emission of VOCs for three scenarios, and the lines show the removal efficiency of the seven key anthropogenic source categories to meet the different scenario targets.
To reach the emission target, the average removal efficiency of VOC emissions from anthropogenic sources should be improved by approximately 68.1% (scenario 2030B) and 79.0% (scenario 2030B).In scenarios 2030B and 2030C, the removal efficiency of control technologies from machinery equipment manufacturing, furniture manufacturing, and synthetic leather should improve by 79.4%, 76.6%, 75.8%, and 86.8%, 85.04%, and 81.8%, respectively.To improve removal efficiency, advanced control technologies should be adopted for different anthropogenic sources.Traditional industries may combine different control technologies to reduce VOC emissions.
Condensing, adsorption, and absorption are traditional technologies to recycle VOCs.Membrane separation was recently developed.Thermal incineration, catalytic combustion, biodegradable, photocatalytic degradation, and plasma technologies control VOCs by decomposing VOCs.These technologies are widely applied in different anthropogenic sources.However, technologies such as adsorption and catalytic combustion must be hyphenated.
For example, the process of spraying is the main source of VOC emission in the furniture manufacturing industry.
According to an earlier investigation, the most widely applied control technologies are activated carbon adsorption and solution adsorption.The average removal efficiency of activated carbon adsorption is 50% (Luo et al., 2012).However, the field test shows that the removal efficiency of solution adsorption (solvent: water) is roughly 14.7% (Guo, 2013), which is lower than the laboratory data.To achieve the emission target in scenario 2030B, the average removal efficiency of VOC emissions from furniture manufacturing should be improved by 76.6%, which is higher than that of the control technologies applied.
Technologies must be hyphenated and developed.For example, the hyphenation of activated carbon adsorption and solution adsorption can improve the removal efficiency by 79% (Luo et al., 2012) and the hyphenation of activated  carbon adsorption and environment-friendly materials can increase removal efficiency by 85% (Yang, 2012).The control technologies and removal efficiency for sources are listed in Table S2.
In summary, combining technologies is the prospect of innovative VOC control technologies to achieve the target removal efficiency for petroleum refining, machinery equipment manufacturing, architectural ornament, printing, and synthetic leather industries.An entire process-controlled management system must be established, which includes selecting the optimum control technologies, maintaining the control system, reinforcing law implement, and regulation.

CONCLUSIONS
In this study, an inventory of industrial sector-based sources of VOC emissions was conducted for 2010-2012 in the YRD region using the most recent EFs and activity data at the city level based on the emission factor method.The estimated total industrial VOC emission in YRD increased from 33.4 Tg in 2010 to 39.9 Tg in 2012 at an annual average rate of 9.3%.The use of VOC-containing products is the major contributor of the four links, which account for 68.8% of the total emissions in 2012.Industrial processes using VOCs as raw materials, production of VOCs, and storage and transport contributed 15.7%, 10.3%, and 5.3%, respectively.Furniture manufacturing, machinery equipment manufacturing, architectural ornament, petroleum refining, printing, synthetic leather, storage and transport, chemical pesticide, and synthetic fiber were the key sources, which contribute 84.8% of the total industrial VOC emissions in 2012.Shanghai, Ningbo, Nanjing, Hangzhou, Suzhou, and Shaoxing generated the highest emissions in 2012, contributing 29.2%, 9.0%, 8.4%, 7.2%, 7.1%, and 5.7% of the total emissions.VOC emissions exhibited remarkable spatial and temporal characteristics in the YRD region.Shanghai, Nanjing, Ningbo, Hefei, and the Hangzhou bay area and Taihu lake basin are the most polluted areas, which generated roughly 76% of the total industrial VOC emissions.
The uncertainty for the total emissions in 2012 was -16.4% to 65.6% at the 95% confidence interval in the YRD region, which ranged from 6.5 Tg to 26.2 Tg for VOC estimates.Therefore, conducting further investigation on local industrial sources and additional field tests for emission factors on each sector in this region is crucial.
The projection of three scenarios indicates that the average removal efficiency of VOC emissions from anthropogenic   sources should be improved by 68.1% (scenario 2030B) and 79.0% (scenario 2030B) based on the data in 2012.The top seven anthropogenic sources were chosen and investigated according to the VOC emissions inventory.Activated carbon adsorption and solution adsorption are the main control technologies adopted by furniture manufacturing enterprises.
The removal efficiency of applied control technologies is lower than laboratory data because of the lack of effective management and maintenance.Combining technologies is key to improve removal efficiency, and an entire process control management system must be established.For the fugitive sources such as manufacturing industries of furniture, chinery equipment, and transportation equipment, which commonly use a large amount of VOCs-containing products in the process of painting, we have to improve the process of production, reduce the exposure duration of VOCs-containing products, and adopt suitable control technologies to collect and treat the exhaust gas.

Fig. 2 .
Fig. 2. Urban industrial VOCs emission of link in YRD region for 2012 (a) using VOC-containing products; (b) for industrial processes utilizing VOCs as raw materials.

Fig. 4 .
Fig. 4. Spatial distribution of VOC emissions in YRD region at a grid of 1 km × 1 km in 2010-2012.

Fig. 5 .
Fig. 5. Projection of industrial VOC emissions from specified anthropogenic sources (The bars indicate the emission of VOCs for three scenarios, and the lines show the removal efficiency of the seven key anthropogenic source categories to meet the different scenario targets).

Table 1 .
Source categorization and activity data of industrial VOCs emission inventory.

Table 2 .
Industrial VOC emissions of major sources.
Table5shows the uncertainties of the different sectors and the uncertainties of the use of VOC-containing products.The uncertainty is relatively high, particularly in synthetic leather, shoemaking, architectural ornament, and machinery equipment manufacturing, because of the inadequate source information and the limited field-test data of emission factors.

Table 3 .
VOC emission inventory of 22 cities in YRD region in 2012 (kt).

Table 4 .
VOC emission inventory of 22 cities in YRD region in 2010-2012.

Table 5 .
Uncertainty in emission inventories of each sector in YRD region, 2012.

Table 6 .
Comparison of results with those of other studies.