Policy Analysis pubs.acs.org/est
Air Quality and Climate Impacts Due to CNG Conversion of Motor Vehicles in Dhaka, Bangladesh Zia Wadud* Centre for Integrated Energy Research, University of Leeds, Leeds LS2 9JT, U.K. Civil Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh
Tanzila Khan Civil Engineering, Stamford University Bangladesh, Dhaka 1209, Bangladesh S Supporting Information *
ABSTRACT: Dhaka had recently experienced rapid conversion of its motor vehicle fleet to run on compressed natural gas (CNG). This paper quantifies ex-post the air quality and climate benefits of the CNG conversion policy, including monetary valuations, through an impact pathway approach. Around 2045 (1665) avoided premature deaths in greater Dhaka (City Corporation) can be attributed to air quality improvements from the CNG conversion policy in 2010, resulting in a saving of around USD 400 million. Majority of these health benefits resulted from the conversion of high-emitting diesel vehicles. CNG conversion was clearly detrimental from climate change perspective using the changes in CO2 and CH4 only (CH4 emissions increased); however, after considering other global pollutants (especially black carbon), the climate impact was ambiguous. Uncertainty assessment using input distributions and Monte Carlo simulation along with a sensitivity analysis show that large uncertainties remain for climate impacts. For our most likely estimate, there were some climate costs, valued at USD 17.7 million, which is an order of magnitude smaller than the air quality benefits. This indicates that such policies can and should be undertaken on the grounds of improving local air pollution alone and that precautions should be taken to reduce the potentially unintended increases in GHG emissions or other unintended effects.
1. INTRODUCTION Mega cities in most developing countries are often characterized by a poor air quality and motor vehicles are a major source of air pollution. Extensive research linked motor vehicle induced air pollution to adverse health impacts and premature mortality in the developed world,1−4 as well as in the developing countries (e.g., Delhi5). Vehicle activities are also a major source of carbon emissions, a potent greenhouse gas (GHG), adversely affecting the climate system. In developed countries, local air pollution problem from motor vehicles received attention decades ago, resulting in various strong policy measures (e.g., vehicle emissions standards, phasing out of leaded petrol, etc.), and a major concern now is the control of GHG emissions. The situation is reversed in most developing countries where local air quality is worsening, primarily because of increasing motor vehicle ownership resulting from a high economic growth and lax vehicle emissions control. While GHG emissions are also increasing and is of some concern, the priority to the policy makers in these countries or cities is reducing criteria pollutants from motor vehicles. Replacing the existing petroleum vehicles with compressed natural gas (CNG) vehicles or retrofitting them to run on CNG offers large air quality benefits6,7 and accordingly many of the megacities (Rio de Janeiro, Mexico City, Delhi, © 2013 American Chemical Society
Mumbai, Karachi) have successfully introduced a varied number of CNG vehicles in their vehicle fleet.8,9 Most of the fleet conversion was particularly for heavy duty diesel buses and trucks or two-stroke three-wheelers, which were the most polluting vehicles on the street in these cities. Dhaka, the capital of Bangladesh and a densely populated megacity of 13 million,10 has a poor air quality, with an annual average PM2.5 concentration of 80.4 μg/m3 in 2009; and motor vehicles were the major source of air pollution in Dhaka too (see Supporting Information (SI) for detail). The city was late to act on its poor air quality, but adopted a strategy of rapid conversion of its motor vehicle fleet to CNG. Dhaka had banned its fleet of two-stroke three-wheelers and replaced them by 9000 CNG three-wheelers and 12 000 CNG taxis in early 2002, resulting in a visible and measurable improvement in its air quality.11 However, unlike in Delhi, where the primary focus was on replacing para-transit (two-stroke petrol three-wheelers and taxis) and public transit modes (diesel buses) by CNG vehicles, Dhaka successfully aimed to replace or retrofit its Received: Revised: Accepted: Published: 13907
May 24, 2013 October 25, 2013 November 6, 2013 November 6, 2013 dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
Figure 1. Vehicle fuel distribution in Dhaka from authors’ spot survey. CNG (OEM) = original equipment manufactured CNG; CNG (diesel) = converted from diesel; CNG (petrol) = converted from petrol.
Figure 2. Impact pathway approach for modeling policy interventions in (a) air quality and (b) climate impacts.
entire vehicle fleet, including personal vehicles. The rapid introduction of 21 000 CNG taxis and three-wheelers in 2002 created an initial demand for CNG infrastructure and suitable policies (reduced CNG prices, zero taxes on retrofitting equipments, private sector participation in retrofitting and CNG supply) ensured that CNG demand and supply were not a constraint on each other and encouraged rapid conversion in a large scale. Local emissions reductions because of CNG conversion of vehicles have been analyzed in other megacities (Rio De Janeiro,8 Mexico City,9 New Delhi12), while the GHG benefits of such conversion for Delhi has also been studied.12,13 Reynolds and Kandlikar,12 however, showed that increased
CH4 emissions from CNG vehicles can reduce the GHG benefits in Delhi. The GHG benefits in Delhi primarily arose through conversion from diesel vehicles which are large emitters of particulates and black carbon, which is more potent than CO2 or CH4 as a GHG. Large scale conversion of petrol vehicles, which are not as large an emitter of particulates and black carbon as diesel vehicles, thus runs the risk of reduced GHG benefit or even negative GHG impacts. This paper aims to quantify such impacts for the widespread CNG conversion policy in Dhaka. In addition, air quality (AQ) benefits are modeled to understand the relative impacts of local air quality improvements and GHG emissions reduction. Unlike previous studies, focus is not on emissions reduction alone but on the 13908
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
The modeled grid-wise pre- and postconversion emissions are fed into an air quality model to determine the changes in ambient air quality (i.e., PM2.5 concentration) to which people are exposed. Rahman24 developed a gridded source-receptor matrix for Dhaka based on a year-long dispersion simulation of PM2.5 and its precursors using a Lagrangian puff model and local meteorology. The derived source-receptor matrix directly provides ambient concentration at a specific grid due to emissions at other grid locations (see SI). In the third step of the impact-pathway approach, the modeled improvements in ambient air quality in each grid is coupled with spatial population distribution and epidemiological concentration−response (CR) functions of the health impacts to determine the avoided health impacts of different types. Although there are various adverse health impacts such as asthma, chronic bronchitis, respiratory troubles, cough or sore throat (morbidity impacts), premature mortality impacts dominates (80−90%) other health impacts in monetary terms.25 We choose mortality as the only metric for health impacts due to its dominance in the monetary valuations, availability of Bangladesh specific data on mortality (relative to morbidity) and lack of information on morbidity valuation in Bangladesh. CR functions for long-term exposure to PM2.5 (our case) are an order of magnitude higher than short-term acute exposure.26−28 Although CR functions for increases in allcause mortality are generally used in modeling policy interventions in the developed countries,3,25 the causes of deaths vary significantly between developed and developing countries, and CR functions for all-cause mortality are not appropriate for Bangladesh.29 Therefore cause-specific CR functions with cause-specific mortality rates were used. In addition, recent evidence show that the CR functions can be nonlinear, with lower relative risk at higher ambient concentrations. The nonlinearity was modeled using Ostro’s30 log−linear approach combined with Dockery et al.31 CR functions, which results in a 3.9% and 4.1% increases in mortality risks of adults because of cardiovascular and respiratory diseases respectively for every 10 μg/m3 increases in the ambient PM2.5 concentration at the current concentration in Dhaka (see SI). Mortality risks for cardiovascular and respiratory diseases in Bangladesh are calculated to be 5.36 and 3.4 per thousand adults respectively (see SI).32,33 Population in Dhaka metropolitan area in 2009 was estimated to be around 13 million10 of which 4.67 million were above 30. Using the changes in concentration and population in each grid and the mortality and CR information, the number of avoided premature deaths in year 2010 due to CNG conversion of vehicles was determined. Following U.S. regulatory impact assessments,25 the avoided premature deaths are multiplied by the value of statistical life (VSL) to monetize these benefits. VSL in Bangladesh has not been derived in any dedicated study, therefore, we utilize VSL information from literature with necessary adjustments to reflect the differences in income and use a VSL of USD 200 000 for Bangladesh (see SI). Despite some practice of using loss of life expectancy,34 disability adjusted life year (DALY), or quality adjusted life year (QALY) to value premature mortality or morbidity, the valuation of such parameters have even larger uncertainty and therefore not followed here. 2.3. Climate/GHG Impacts. Quantification of the climate change benefits (costs) from CNG conversion follows a different path. Since the changes in GHG emissions is small
impacts of the reduction and its monetization. To our knowledge, this is the first study to model both the local and global environmental cobenefits of a large scale CNG conversion policy in terms of economic benefits (costs). We also conduct an extensive quantitative analysis using Monte Carlo simulation and sensitivity assessment to assess the uncertainty in the results. The research directly feeds into the current interest in simultaneously mitigating air pollution and climate change and integrated assessment of cobenefits14−18 and in introducing CNG vehicles in other developed and developing countries (e.g., Europe19 and Algeria20). There is also a renewed interest in natural gas as an energy resource, including as vehicle fuel, because of technological breakthrough, favorable economics, and regulations for unconventional (shale) gas.
2. METHODS 2.1. Current Conversion Status. The government statistics show that around 117 000 vehicles, around 40% of the vehicle fleet (excluding motor cycles) in Dhaka run on CNG, while nationwide the number is around 200 000 (in 2010). Although the absolute number may not appear large, vehicle ownership in Bangladesh is small, which puts Bangladesh among the top five countries in the world in CNG penetration (∼18% of vehicle fleet21). A spot survey of around 2000 vehicles at different locations in Dhaka reveals that almost 85% of all cars surveyed were running on CNG in June 2010 (Figure 1 and SI). The proportion of buses and minibuses running on CNG was above 75%. The taxis and three-wheeler autorickshaws run on CNG and are all original equipment manufactured. The category “others” contain human haulers (small pickup/truck converted to public transit for 10−15 people), covered vans, etc., and show large conversion too. One vehicle class which did not show significant uptake of CNG is the trucks. The large difference in CNG vehicle number with government statistics is possibly because of the slow updating of the CNG vehicle register or the inaccurate vehicle registration data (no vehicle attrition data, which deflates the share of CNG vehicles). We use the conversion statistics from the spot survey in this study (Figure 1). 2.2. Air Quality Benefits. Reduced emissions of criteria air pollutants and associated health benefits are the major local impacts of using CNG as vehicle fuel. CNG vehicles can have large reductions in CO, NOx and particulates compared to similar petrol or diesel vehicles.22 To model the health impact, the changes in emissions because of the conversion policy is linked to well-defined health impacts and associated monetary benefits through the impact-pathway approach, as described graphically in Figure 2a. The first step is to quantify the emissions (before and after the policy), which can be determined from a vehicle emissions inventory model. Since particulates, especially PM2.5, are the most hazardous for health,23 changes in PM2.5 and its precursors are modeled here. A spatially disaggregated, gridded, bottom-up emissions inventory for road transport in Dhaka has been developed in this study. In the absence of local emissions factors for vehicles, regional-literature-based factors are used. Pre- and postconversion emissions were distributed among grids based on road type and road length in each grid and relative vehicle volume in those roads using GIS (see SI for detail). Note that we are modeling the impact of CNG conversion of vehicles after 2002, therefore earlier benefits due to the introduction of CNG threewheelers and taxis are not included. 13909
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
Table 1. Impact of the CNG Conversion Policy vehicle types
column 1 a
number of CNG vehicles converted (converted + OEM) share of converted vehicles (%) ΔPM2.5 emissions (tonne/year) ΔSO2 emissions (tonne/year) ΔPM2.5 concentrations (μg/m3) average DCC grids avoided premature deaths, Greater Dhaka (DCC) AQ benefits (million USD/year), Greater Dhaka (DCC) ΔCO2 emissions (tonne/year) ΔCH4 emissions (tonne/year) ΔBC emissions (tonne/year) ΔOC emissions (tonne/year) ΔSO2 global emissions (tonne/year) ΔCO2-e emissions, CO2 + CH4 (tonne/year) ΔCO2-e emissions, all (tonne/year) GHG benefits (−ve implies cost) (million USD/year)
fuel conversion types
all vehicles
bus, minibus
car, SUV, microbus
all diesel CNG vehicles
column 2
column 3b
column 4b
203 040 (215 269)
9785 (13 808)
165 732 (172 143)
16 542
186 498
100 −2089 −628 −10.72 2045 (1665) 408.6 (332.7)
4.8 −1287 −141 −6.20 1104 (912) 220.6 (182.2)
81.6 −491 −361 −2.48 411 (326) 82 (65)
8 −1723 −263 −8.71 1532 (1252) 306 (250)
92 −366 −365 −2.01 344 (274) 68.7 (54.8)
−139 820 42 112 −1318 −528 −626 912 980 394 370 −17.7
42 085 7179 −788 −248 −141 221 560 −114 200 5.1
−148 672 27 255 −258 −177 −360 532 703 457 508 −20.6
29 430 10 125 −1155 −363 −262 282 555 −204 065 9.2
−169 245 31 987 −163 −165 −364 630 430 598 440 −26.9
column 5
all petrol CNG vehicles column 6
a
Excluding OEM taxis and three-wheelers. bThe values of columns 3 and 4 do not sum to that of column 2 because the other vehicle category is not included; the values of columns 5 and 6 sum to the value of column 2, except for avoided mortality and AQ monetized benefit estimates because of nonlinearity in the C-R functions.
for a policy as considered here, a full scale impact-pathway model coupled with climate impact models will not be able to pick up any differences. Also, modeling the changes in climate due to changes in emissions and modeling the corresponding damages require large and specialized resources (e.g., climate induced damages may include crop losses, coastal inundation, increased flooding, increased cyclones, increased diseases, etc., each of which requires separate, extensive damage pathways). Impact of different GHGs on radiative forcing balance and thus climate is also different. However, it is possible to normalize the changes in emissions (from the emissions inventory model) of different GHGs using global warming potentials35 and then use the market price of carbon or social costs of carbon emissions to monetize the avoided damages (Figure 2b). Since GHGs do not affect local population directly, spatial disaggregation is not important and only total emissions were calculated for different vehicle and fuel types. Among motor vehicle emissions, CO2 and CH4 are established GHGs, contributing directly to global warming.35 However, recent studies find aerosols, such as sulphates (SO2), black carbon (BC), and organic carbon (OC), also have important influence on the earth’s radiation balance.36 BC has a potentially large impact on warming, especially on short-term warming and is considered to be the second most important contributor to current global warming after CO2.37,38 On the other hand, SO2 (precursor to sulphates) and OC have cooling effects on the atmosphere through facilitating the formation of aerosols.12,36 Although NOx emissions can also have an impact on global warming through secondary effects, we assume, following Reynolds and Kandlikar,12 that the small NOx changes from fuel switching have negligible climate impacts. We therefore concentrate on five global emissions, CO2, CH4, SO2, BC, and OC, before and after the conversion of the vehicles (see SI). We use the 100 year global warming potentials of each of these pollutants to normalize them to a common CO2
equivalent metric, which allows the addition or subtraction of the different pollutants to generate net warming-weighted emissions. Although global warming potentials for CO2 (1) and CH4 (23) are well established in the literature, the factors for BC, OC, and SO2 are still discussed, and we use Reynolds and Kandlikar’s12 estimates for BC (455), OC (−35), and SO2 (−100). The global warming factors for OC and SO2 are negative, because an increase in these emissions results in cooling of the atmosphere. There is an ongoing discussion among the climate scientists about the appropriateness of GWPs, especially for the short-lived species, but as of now it is still the most widespread approach. Use of a different warming period will give different weight to the species, and their relative (and total) contribution can change. Net CO2-equivalent emissions are multiplied by unit value of carbon to monetize the GHG impacts. Instead of using the European Union’s Emission Trading Scheme prices for carbon, which depends on a “cap” for its prices, we opt to use social costs of carbon in the literature. Peer-reviewed literature on the social costs of carbon reveals that the mean social cost of carbon is USD 43 per tonne, with a standard deviation of USD 83 per tonne.39 We use a social cost of carbon of USD 45 in 2010 for our calculations.
3. RESULTS 3.1. Air Quality Benefits. Emissions inventory model reveals that there was a reduction of 2090 tonnes of PM2.5 and 630 tonnes of SO2 emissions because of the CNG conversion of the vehicles (Table 1). This resulted in an annual average reduction of ambient PM2.5 concentration of around 6.5 μg/m3 in the entire modeling domain and around 10.72 μg/m3 in Dhaka city area. The maximum annual reduction in a single grid was 35.63 μg/m3 at a traffic pollution hot spot. This improvement in PM2.5 concentrations has avoided around 1665 premature deaths in Dhaka City Corporation (DCC) and about 2045 premature deaths in greater Dhaka. At a VSL of 13910
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
of emissions, reduce the warming effects, but reductions of OC and SO2 act in an opposite direction. The total warming results above conceal important opposite effects of diesel and petrol vehicles. Although there is a small CO2 emissions benefit from petrol to CNG conversions for each conversion, increased CH4 emissions negate these benefits. Since BC emissions of petrol vehicles were not large, reduction in BC emissions is not large enough to reverse the warming impacts of CH4 (and reduced OC and SO2). Therefore, the net CO2-e impact remains a warming effect for petrol to CNG conversions (column 6 in Table 1). On the other hand, diesel to CNG conversion results in an increase in CO2 and CH4 emissions, and reductions in OC and SO2 emissions. However, large reductions in BC emissions make the net CO2-e impact beneficial (column 5). For the same reasons, climate impacts of the conversion of buses and minibuses are beneficial, while that of personal vehicles are not. The monetized costs (USD 17.7 Million) of damages because of global warming attributable to CNG conversion are relatively negligible compared to the health benefits. If the net increases in CO2-e emissions are balanced by buying carbon credits (i.e., if we use market price instead of social costs of carbon), the direct monetary loss is even less, around USD 5 Million. 3.3. Uncertainties. Benefits modeling results from policy interventions, such as ours as presented above, inherently have uncertainties, which depend on the performance of the underlying modeling techniques and input data. For both air pollution and GHG benefits, the successive underlying models are directly dependent on results from the previous ones, and therefore the uncertainties increase from left to right of Figure 2. Thus, the final estimates of the monetary benefits due to the policy intervention would generally be associated with more uncertainty than the estimates for changes in criteria air pollutant or GHG emissions. Our estimates of air pollution and climate benefits have a larger uncertainty than similar estimates in developed countries, primarily because of the larger uncertainty in input data for the emissions inventory model (emissions factors, grid wise activity etc.) and the rather simple air quality model. The CR functions also have uncertainties as they are determined using statistical techniques and in developed countries. VSL and social costs of carbon also have large uncertainties associated with them. Since the uncertainties in the inputs can have large impact on the results, we explicitly model these quantitatively. The primary purpose is to acknowledge and identify the gaps in knowledge and how that affects the results. The major sources of uncertainties are vehicle data, emissions factors, activity data, fuel-wise vehicle classification, air quality model, CR functions, GWP of nontraditional GHGs or precursors, VSL, and social costs of carbon. The uncertainties in vehicle data, vehicle activity data and fuel-wise distribution have been reduced following our spot survey which allowed the collection of some of the information directly. Effect of emissions factors on the outcomes is of particular interest since these were not specifically derived from vehicle test data in Dhaka. We hypothesize distributions for the emissions factors for various pollutants and vehicle types based on literature review on the high, low and most likely values (see SI). We then conduct a Monte Carlo simulation by randomly drawing 5000 emissions factors from the input distributions and running the impact pathway model to determine the output. As output, we use avoided premature mortality for local air quality and total
USD 200,000, the CNG conversion policy resulted in an air quality benefit of USD 409 million in the greater Dhaka area. In order to put the number in perspective, it is around 0.4% of the annual GDP of Bangladesh, and thus the air quality benefits are clearly large. Although the number of diesel bus and minibus conversions is smaller than personal vehicles, air quality benefits of these conversions are much larger than the personal vehicles (columns 3 and 4 in Table 1). Of the total 203 000 vehicles converted, around 9800 (5%) were buses or minibuses (13 800 if OEM buses and minibuses are included), but these buses and minibuses were responsible for around 62% of all PM2.5 emissions reduction. The large reduction is a result of larger travel of and higher particulate emissions from buses and minibuses, which used to run on diesel (or would have run on diesel). Including other diesel vehicles (four wheel drives and some vehicles in other category) we find total diesel to CNG conversion is around 16 500 (8% of all CNG conversions) but this represents a PM2.5 reduction of 82.5% (column 5). This large PM2.5 reduction benefits from a small number of diesel vehicles is not surprising, since diesel vehicles (especially older ones, as in Dhaka) emit more particulates than petrol ones, commercial vehicles are generally less well maintained and Dhaka’s diesel quality is particularly poor (5000 ppm sulfur). The conversion of around 166 000 personal vehicles (cars, SUVs, microbuses), which predominantly run on petrol reduces particulates by only 23.5% (column 4). A full-scale impactpathway run for only these conversions shows that around 411 avoided premature deaths can be attributed to this rather large number of conversions. Although the benefits per vehicle is much lower as compared to buses and minibuses, these conversions are still cost-effective from a social perspective (at a USD 1000 retrofitting cost per vehicle, total cost of conversion is USD 166 million, air quality benefits in 2010 alone is USD 82 million, which recur every year). Once we consider only the petrol to CNG conversions (column 6), the cost effectiveness diminishes further, since petrol vehicles emit much less PM2.5, compared to diesel vehicles. Impact pathway model runs for petrol to CNG and diesel to CNG conversions show that around 186 000 petrol to CNG conversions resulted in around 344 avoided premature deaths (column 6), while the 16 500 diesel to CNG conversions resulted in around 1532 avoided deaths (column 5). Clearly, returns on diesel to CNG conversions are much larger, although petrol to CNG conversions remain cost-effective (i.e., larger social benefits than individual conversion costs) as before. 3.2. Climate/GHG Impacts. The global warming impact of the CNG conversion is not as straightforward. Considering the most common GHG, CO2, the conversion policy is mildly beneficial (i.e., reduces warming) through a reduction in CO2 emissions of 140 000 tonnes (column 2 in Table 1). This results from small CO2 benefits due to a large number of conversions of the petrol vehicles (diesel vehicles have a CO2 penalty, see SI). However, since emissions of CH4 from all vehicles increase due to conversion, and CH4 has a larger GWP than CO2, the combined impact of CO2 and CH4 is harmful for climate (i.e., warming impact increases). Addition (subtraction) of the warming (cooling) impacts of aerosols and their precursors (SO2, BC, and OC) to the impacts of CO2 and CH4 counters the warming impacts of CH4 but still results in a mild net increase in warming emissions. Reductions in BC emissions, which have the largest impact on warming per unit 13911
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
Figure 3. Histogram for (a) avoided premature deaths and (b) changes in CO2-e emissions due to the uncertainties in the emission factors for 5000 random draws during Monte Carlo simulation.
CO2-e emissions for global climate impacts. Following Mahashabde,40 we kept the valuation of the impacts, VSL and social costs of carbon, outside the scope of this simulation, since they are policymakers’ choices, rather than model uncertainty (we include them in sensitivity analysis later). Figures 3a and b present the frequency distribution of outputs (premature mortality avoided and changes in CO2-e emissions) based on the input distribution of the emissions factors. The histogram of avoided premature mortality has a
mean of 1900 and standard deviation of 196. Ninety-five percent of the output in avoided premature mortality lies between 1517 and 2285, indicating that the benefits are highly likely even after considering the uncertainties in emissions factors. Avoided premature mortality in DCC area for our nominal (most likely) scenario was 1665, the difference between the mean and nominal is a result of the right skewed distribution. The histogram for changes in CO2-e emissions is nearly normally distributed, with a mean increase of 260 000 13912
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
Figure 4. Tornado charts for sensitivity of the monetary impacts due to uncertainties in inputs: (a) local air quality benefits and (b) GHG emissions costs.
tonnes/year and standard deviation of 485 000 tonnes/year. 90% of the output lies between −540 000 and 1 075 000 tonnes/year, indicating that the uncertainty is too large to allow reliable inference on the changes in CO2-e emissions. Given that 30% of the output sample shows a beneficial impact (i.e., negative CO2-e), we cannot rule out that possibility as well. In addition, a sensitivity analysis is carried out to understand the impact of uncertainty in other factors on the final result.
The sensitivity analysis is conducted by keeping all variables in the model constant at their nominal, most likely value and then changing one variable at a time to its plausible maximum and minimum values and recalculating the model output. The results of the sensitivity analysis are summarized in Figure 4a and b as Tornado Charts. VSL is the largest source of variation for the valuation of air quality benefits, followed by CR functions and PM2.5 emission 13913
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
CNG can be justified on the basis of local air pollution benefits alone. The local air pollution benefits would have also been larger had we included the improvements due to banning the twostroke three-wheelers which took place in early 2002 and resulted in a very large reduction of particulates (and likely beneficial net GHG impacts). Also, we considered only the annual benefit in year 2010. A typical net present value analysis of all the future years will increase the air quality benefits and GHG costs further. However, the relative impacts, our primary focus, would remain the same in our current modeling framework as long as the discount rate used for analysis remains the same for both impacts. We also did not consider the lifecycle emissions of the fuels, which would possibly reduce the GHG costs of CNG considering its indigenous availability and larger transport emissions for imported petroleum products. Clearly, diesel to CNG conversions offer a significantly larger air quality and GHG benefits as compared to petrol to CNG conversions. This indicates that the priority of petrol to CNG conversion of personal vehicles should have been lower than diesel to CNG conversion. However the lower price of diesel compared to petrol (since diesel is used in agriculture and is subsidized) and operating issues made it less lucrative for owners to convert diesel vehicles to CNG. There are still a large number of trucks in Dhaka city that have not been converted to run on CNG. These diesel trucks are highly polluting, and can offer opportunities for large air quality and climate cobenefits through CNG conversion. However, as the CNG infrastructure is still fragmented and the truck travel is generally longerdistance, conversion of trucks is difficult. Recent price rises of CNG made the diesel−CNG price differential even smaller and is therefore a wrong direction. Petrol to CNG conversions have mild negative impact on GHG emissions and positive health impacts and further strategies to encourage their conversion (most of which are personal vehicles) to CNG critically hinges on the remaining number of super emitters (it is likely that the super emitters, which are generally older vehicles and travel less, may not have been converted to CNG significantly). The personal vehicle fleet in Dhaka has added substantially new vehicles with better emissions performances over the past few years, despite lax emissions standards. This results from a ban on import of vehicles more than 5 years old. Considering the lower criteria emissions from these and future vehicles (especially particulates and BC), subsequent smaller air quality benefits, potential GHG costs of CNG conversion, potentially larger travel activities of CNG personal vehicles and large opportunity costs of CNG (natural gas fuels three-fourths of electricity generation in Bangladesh and current supply is constrained), the suitability of the CNG conversion policy, especially for the petrol vehicles, requires a critical revisit in a larger integrated framework including all these issues.
factors for preconversion petrol cars. The global warming impact is most sensitive to the uncertainties in CH4 and BC emissions factors for cars, GWP of BC, and social costs of carbon. Also, additional calculations show that if CH4 emissions can be reduced to zero through technical means (e.g., leakage prevention), then there can be a net reduction in CO2-e emissions. In addition to these quantitative uncertainties, another potential source of uncertainty can significantly affect the result. All of our estimates are based on the assumption that vehicles travel the same distances before and after conversion to CNG. However, the spot survey results indicate that the onroad CNG cars travel almost 30−50% more than the on-road petroleum cars, possibly because CNG is cheaper. In this case, local air quality benefits in greater Dhaka (DCC) area reduces to USD 380 million (310 million) for our nominal case, and GHG impact is further detrimental (USD 42.3 million). This cross-sectional snapshot can still be misleading because of the self-selection of vehicle conversion and vehicle travel decisions. It is highly likely that the CNG-converted vehicles were already running more before conversion and as such the additional retrofitting costs were justified for converting these vehicles only (since CNG is cheaper). Under this circumstance, our original estimates remain correct. However, the reality is likely to be in between and it will be useful to establish or disprove this self-selection hypothesis in future.
4. DISCUSSIONS Our analysis reveals that there are large air quality benefits accruing to the residents of Dhaka as a result of CNG conversion of the vehicles. Local air pollution benefit amounting 0.4% relative to the GDP for a single policy initiative appears large, which is due to motor vehicles’ large contribution to air pollution and high population density in Dhaka. These two factors combine to ensure any improvement in motor vehicles emissions performance directly and immediately affects the air quality, which then positively affects a large number of people, increasing total air quality benefits. The air quality benefits will likely be even larger since we did not consider any health impacts other than premature mortality or other benefits such as crop productivity and material damages (although these will still be small). However, CNG retrofitting increases ultrafine particulate emissions, which are harmful, but their impact cannot be quantified as yet. The monetized (adverse) GHG impacts of the CNG conversion policy are a few orders of magnitude smaller as compared to the local air quality benefits. The combined effects of changes in CO2 and CH4 emissions are clearly harmful for the climate. Therefore, carbon credit generation and associated financial benefits from such CNG conversion projects under the Clean Development Mechanism (CDM) is not possible unless CH4 emissions are brought under control. This is not impossible as there are opportunities to reduce the leakage CH4 emissions. Inclusion of aerosols and its precursors in calculating CO2-e emissions still results in a mild increase in warming emissions. However, considering the uncertainties in the model, it is possible that the climate impacts could still be beneficial (e.g., if CH4 emissions factor can be reduced significantly) and it is important to have a better understanding of CH4 and BC emissions from local vehicle in future. Nonetheless, GHG impacts (costs or benefits) are likely to be much smaller than the health benefits and the conversion of petroleum vehicles to
■
ASSOCIATED CONTENT
S Supporting Information *
Additional material as described in the text. This information is available free of charge via the Internet at http://pubs.acs.org/.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. 13914
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
Notes
(18) Zusman, E.; Srinivasan, A.; Dhakkal, S. Low Carbon Transport in Asia: Strategies for Optimizing Co-Benefits; Routlege: New York, 2012. (19) Engerer, H.; Horn, M. Natural Gas Vehicles: An Option for Europe. Energy Policy 2010, 38, 1017−1029. (20) Amrouche, F.; Benzaoui, A.; Harouadi, F.; Mahmah, B.; Belhamel, M. Compressed Natural Gas: The New Alternative Fuel for the Algerian Transport Sector. Proc. Eng. 2012, 33, 102−110. (21) IANGV (International Association for Natural Gas Vehicles). Current Natural Gas Vehicle Statistics. http://www.iangv.org/currentngv-stats/ (accessed October 2012). (22) Nijboer, M. The Contribution of Natural Gas Vehicles to Sustainable Transport; International Energy Agency: Paris, 2010. (23) Air Quality Criteria for Particulate Matter; U.S. EPA: Research Triangle Park, NC, 2004. (24) Rahman, S. M. Air Quality Assessment and Health Effects of Air Pollution in Dhaka City through Impact-Pathway Model. MSc (Engg) Dissertation, Department of Civil Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh, 2010. (25) Draft Regulatory Impact Analysis: Control of Emissions of Air Pollution from Locomotive Engines and Marine Compression-Ignition Engines Less than 30 Litres per Cylinder; Assessment and Standard Division, Office of Transportation and Air Quality, U.S. EPA: Washington, DC, 2007. (26) Krewski, D.; Burnett, R. T.; Goldberg, M. S.; Hoover, K.; Siemiatycki, J.; Jarret, M.; Abrahamowicz, M.; White, W. H. Reanalysis of the Harvard Six Cities Study and the American Cancer Society Study of Particulate Air Pollution and Mortality, Special Report; Health Effects Institute: Cambridge, MA, 1999. (27) Laden, F.; Schwartz, J.; Speizer, F. E.; Dockery, D. W. Reduction in Fine Particulate Air Pollution and Mortality: Extended Follow-Up of the Harvard Six Cities Study. Am. J. Respir. Crit. Care Med. 2006, 173, 667−672. (28) Pope, C. A., III; Dockery, D. W. Health Effects of Fine Particulate Air Pollution: Lines that Connect. J. Air Waste Manage. Assoc. 2006, 56, 709−742. (29) Cropper, M. L.; Simon, N. B. Valuing the Health Effects of Air Pollution, DEC notes, No 7; World Bank: Washington DC, April, 1996. (30) Ostro, B. Assessing the Environmental Burden of Disease at National and Local Levels, Environmental Burden of Disease Series, No. 5; World Health Organization (WHO): Geneva, 2004. (31) Dockery, D. W.; Pope, C. A., III; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G.; Speizer, F. A. An Association between Air Pollution and Mortality in Six U.S. Cities. N. Engl. J. Med. 1993, 329, 1753−1759. (32) World Health Organization. Disease and Injury Country Estimates: Burden of Disease. http://www.who.int/healthinfo/ global_burden_disease/estimates_country/en/index.html (accessed July 2009). (33) Bangladesh Bureau of Statistics. Bangladesh Data Book. http:// www.bbs.gov.bd (accessed July 2009). (34) Leksell, I.; Rabl, A. Air Pollution and Mortality: Quantification and Valuation of Years of Life Lost. Risk Anal. 2001, 21 (5), 843−857. (35) Global Warming Potentials. UNFCCC, United Nations Framework Convention on Climate Change, 2010. http://unfccc. int/ghg_data/items/3825.php (accessed April 2010). (36) Climate Change 2007: The Physical Science Basis, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change; Intergovernmental Panel on Climate Change: Geneva, 2007. (37) Ramanathan, V.; Carmichael, G. Global and Regional Climate Changes Due to Black Carbon. Nat. Geosci. 2008, 1 (4), 221−227. (38) Integrated Assessment of Black Carbon and Tropospheric Ozone: Summary for Decision Makers; United Nations Environment Programme (UNEP): Nairobi, Kenya, 2011. (39) Yohe, G. W.; Lasco, R. D.; Ahmad, Q. K.; Arnell, N. W.; Cohen, S. J.; Hope, C.; Janetos, A. C.; Perez, R. T. Perspectives on Climate Change and Sustainability, Climate Change 2007: Impacts, Adaptation and Vulnerability, Contribution of Working Group II to the Fourth
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was undertaken as a research project of the Centre for Advanced Studies and Research (CASR) of Bangladesh University of Engineering and Technology (BUET). Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of CASR or BUET.
■
REFERENCES
(1) Small, K. A.; Kazimi, C. On the Costs of Air Pollution from Motor Vehicles. J. Transp. Econ. Policy 1995, 7−32. (2) McCubbin, D. R.; Delucchi, M. A. The Health Costs of MotorVehicle-Related Air Pollution. J. Transp. Econ. Policy 1999, 33 (3), 253−86. (3) Kunzli, N.; Kaiser, R.; Medina, S.; Studnicka, M.; Chanel, O.; Filliger, P.; Herry, M.; Horak, F., Jr; Puybonnieux-Texier, V.; Quenel, P.; Schneider, J.; Seethaler, R.; Vergnaud, J.-C.; Sommer, H. PublicHealth Impact of Outdoor and Traffic-Related Air Pollution: A European Assessment. Lancet 2000, 356, 795−801. (4) Health Impacts of Transport Emissions in Australia: Economic Costs, Bureau of Transport and Regional Economics (BTRE) Working Paper 63; Department of Transport and Regional Services: Canberra, Australia, 2005. (5) Chattopadhyaya, V. Towards Clean Air: Delhi’s CNG Programme. Presented at 4th Asia Clean Energy Forum, Manila, June 17−19, 1999. (6) Maclean, H. L.; Lave, L. B. Environmental Implications of Alternative-Fueled Automobiles: Air Quality and Greenhouse Gas Tradeoffs. Environ. Sci. Technol. 2000, 34 (2), 225−231. (7) Cohen, J. T.; Hammitt, J. K.; Levy, J. I. Fuels for Urban Transit Buses: A Cost-Effectiveness Analysis. Environ. Sci. Technol. 2003, 37, 1477−1484. (8) Balassiano, R.; White, P. Experience of Compressed Natural Gas Bus Operations in Rio de Janeiro, Brazil. Transp. Res., Part D 1997, 2 (2), 147−155. (9) Schifter, I.; Diaz, L.; Lopez-Salinaz, E.; Avalos, S. Potential Impacts of Compressed Natural Gas in the Vehicular Fleet of Mexico City. Environ. Sci. Technol. 2000, 34 (11), 2100−2104. (10) Bangladesh Bureau of Statistics, Statistical Pocket Book of Bangladesh, 2008. http://www.bbs.gov.bd (accessed July 2009). (11) Begum, B. A.; Biswas, S. K.; Hopke, P. K. Impact of Banning of Two-Stroke Engines on Airborne particulate matter Concentrations in Dhaka, Bangladesh. J. Air Waste Manage. Assoc. 2006, 56, 85−89. (12) Reynolds, C. C. O.; Kandlikar, M. Climate Impacts of Air Quality Policy: Switching to a Natural Gas-Fueled Public Transportation System in New Delhi. Environ. Sci. Technol. 2008, 42 (16), 5860−5865. (13) Reynolds, C. C. O.; Grieshop, A. P.; Kandlikar, M. Climate and Health Relevant Emissions from In-Use Indian Three-Wheelers Fueled by Natural Gas and Gasoline. Environ. Sci. Technol. 2011, 45, 2406−2412. (14) Swart, R.; Amann, M.; Raes, F.; Tuinstra, W. A Good Climate for Clean Air: Linkages between Climate Change and Air Pollution. Clim. Change 2004, 66 (3), 263−269. (15) Mazzi, E. A.; Dowlatabadi, H. Air Quality Impacts of Climate Mitigation: UK Policy and Passenger Vehicles Choice. Environ. Sci. Technol. 2007, 41 (2), 387−392. (16) Bollen, J.; Hers, S.; van der Zwaan, B. An Integrated Assessment of Climate Change, Air Pollution, and Energy Security Policy. Energy Policy 2010, 38 (8), 4021−4030. (17) Hammingh, P.; Smekens, K. E. L.; Plom, A. J.; Koelemeijer, R. B. A. Co-impacts of Climate Policies on Air Polluting Emissions in the Netherlands, Final report of the Dutch Research Programme on Air and Climate; Netherlands Environmental Assessment Agency: The Hague, 2010. 13915
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
Environmental Science & Technology
Policy Analysis
Assessment Report of the Intergovernmental Panel on Climate Change; Parry, M. L., Canziani, O. F., Palutikof, J. P., van der Linden, P. J., Hanson, C. E., Eds.; Cambridge University Press: Cambridge, U.K., 2007; pp 811−841. (40) Mahashabde, A. Assessing Environmental Benefits and Economic Costs of Aviation Environmental Policy Measures. PhD Thesis, Massachusetts Institute of Technology, Cambridge, MA, USA, 2009.
13916
dx.doi.org/10.1021/es402338b | Environ. Sci. Technol. 2013, 47, 13907−13916
