Pump-to-Wheels Methane Emissions from the Heavy-Duty

Dec 22, 2016 - Pump-to-wheels (PTW) methane emissions from the heavy-duty (HD) transportation sector, which have climate change implications, are poor...
0 downloads 7 Views 1MB Size
Article pubs.acs.org/est

Pump-to-Wheels Methane Emissions from the Heavy-Duty Transportation Sector Nigel N. Clark,* David L. McKain, Derek R. Johnson, W. Scott Wayne, Hailin Li, Vyacheslav Akkerman, Cesar Sandoval, April N. Covington, Ronald A. Mongold, John T. Hailer, and Orlando J. Ugarte Department of Mechanical and Aerospace Engineering, PO Box 6106, West Virginia University, Morgantown, West Virginia 26506, United States S Supporting Information *

ABSTRACT: Pump-to-wheels (PTW) methane emissions from the heavy-duty (HD) transportation sector, which have climate change implications, are poorly documented. In this study, methane emissions from HD natural gas fueled vehicles and the compressed natural gas (CNG) and liquefied natural gas (LNG) fueling stations that serve them were characterized. A novel measurement system was developed to quantify methane leaks and losses. Engine related emissions were characterized from twenty-two natural gas fueled transit buses, refuse trucks, and over-the-road (OTR) tractors. Losses from six LNG and eight CNG stations were characterized during compression, fuel delivery, storage, and from leaks. Cryogenic boil-off pressure rise and pressure control venting from LNG storage tanks were characterized using theoretical and empirical modeling. Field and laboratory observations of LNG storage tanks were used for model development and evaluation. PTW emissions were combined with a specific scenario to view emissions as a percent of throughput. Vehicle tailpipe and crankcase emissions were the highest sources of methane. Data from this research are being applied by the authors to develop models to forecast methane emissions from the future HD transportation sector.



INTRODUCTION Natural gas, from traditional, shale, and renewable sources, is offered as an alternative fuel for heavy-duty (HD) transportation1 and has lower carbon content per unit of energy than diesel fuel. Production of natural gas in the U.S. has surpassed 27 trillion cubic feet per year.2 Natural gas vehicles and their associated fueling infrastructure currently release methane into the atmosphere and there is concern that methane from production and use has climate change implications.3,4 Methane is a substantially more potent greenhouse gas (GHG) than carbon dioxide (CO2), a main product of methane combustion in internal combustion (IC) engines. If overall methane emissions are sufficiently low or if the gas is certified as an ultralow carbon fuel (sometimes referred to as renewable natural gas, such as biomethane), then potential GHG advantages (from methane’s lower carbon content and lower CO2 emissions compared to diesel fuel) exist from increased adoption of natural gas as a HD transportation fuel.5,6 Studies supported by the Environmental Defense Fund and industry addressed methane emissions from the transmission, distribution, storage, and processing of natural gas.6−8 Argonne National Laboratory observed, “Limited publicly available data on total vehicle (or crankcase) methane emissions are available.”9 Methane emissions associated with fueling of vehicles and fuel stations are not well documented in the literature; however, vehicle operating emissions are included in © XXXX American Chemical Society

the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) model.10 Natural gas has been used to fuel IC engines, both stationary and mobile,11,12 including throttled lean and stoichiometric spark ignited (SI) engines13,14 and unthrottled High Pressure Direct Injection (HPDI) engines.15 Truck engines may be retrofitted with dual fuel kits that displace part of the diesel fuel with natural gas.16 During this study, the market was dominated by Cummins-Westport 9 L ISL G stoichiometric natural gas engines and Westport Innovations 15 L HPDI engines. Cummins-Westport 12 L ISX G stoichiometric natural gas engines were deployed during the study. Methane is present in the exhaust from all of these technologies. Natural gas engines without a closed crankcase system will emit methane: blowby fraction may be 0.5−1% of the intake flow,17 and blowby gas will be high in methane for homogeneous charge engines. Regulations from the U.S. Environmental Protection Agency (EPA) already limit GHG emissions, including crankcase methane emissions,18 from HD vehicle engines, and proposed rules for 202119 state directly that NG engine crankcases shall be closed. Cummins-Westport has released a closed crankcase Received: Revised: Accepted: Published: A

December December December December

11, 12, 22, 22,

2015 2016 2016 2016 DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 1. Well-to-wheels (WTW) diagram of the use of natural gas as a transportation fuel. Only CNG and LNG pathways were examined. The bounds of our Pump-to-Wheels (PTW) study are shown by the dashed boundary line, which includes the stations and the vehicles. Our study included emissions from the point that tankers or pipelines crossed station property lines through to the end use in the vehicle. The gray boxes indicate the sources of methane emissions that are possible.

engine for U.S. sale.20 HPDI engines currently emit methane via a dynamic fuel system vent to the atmosphere. Methane emissions from leaks can occur at compressed natural gas (CNG) and liquefied natural gas (LNG) fueling stations. Emissions occur during bulk fuel deliveries (offloads) to LNG stations. Since pressure rises in cryogenic LNG storage vessels due to various sources of heat gain, facilities may vent boil-off methane through a pressure relief valve (PRV) or regulator to control tank pressure safely. This is of particular note for underutilized stations where deliveries of fresh, cold LNG to the station tanks are infrequent. Codes, standards, and recommended practices for on-board vehicle fuel systems and natural gas dispensing and storage systems are found in National Fire Protection Association (NFPA) 52.21 CNG tanks typically store fuel at a pressure of 3600 pounds per square inch gage (psig) at 70 °F, whereas LNG vehicle tanks are cryogenic containers that store liquid vapor fuel at pressures up to 230 psig. Losses can occur due to manual or pressure relief venting from on-board fuel storage tanks. We gathered in-use methane emissions data from the HD natural gas transportation sector including sources from the stations (pumps) to the vehicle exhausts (wheels), see Figure 1. Our PTW study addressed methane emissions from the fueling infrastructure for HD vehicles and from the vehicles themselves, without considering sources that are upstream of the fueling station and that have been the focus of other studies.3,4,7 Significant gaps exist in the prediction of methane emissions from the HD transportation sector. Our goal was to provide a data set of emissions factors for the components of this segment for use in predicting future scenarios, rather than an existing inventory. Accordingly, we addressed only recent technology, and did not characterize methane emissions from older technologies such as lean-burn engines, although some of these technologies are still in use today. We did not seek to present an entire life cycle analysis (LCA) of the methane emissions but rather focused on new measurements of later

technologies, which can be combined with upstream data to complete a LCA using such models as GREET. Substantial Supporting Information has been provided to augment material in this paper including details on quantification methods, specific station configurations, data processing, and engine related emissions data.



EQUIPMENT AND METHODS We identified potential PTW methane leaks and emissions sources in the HD transportation sector through a review of current technologies and practices associated with LNG and CNG fueling stations, natural gas engine and exhaust aftertreatment systems, vehicle fuel delivery systems, and vehicle on-board storage. Methods were identified or developed for detecting and quantifying mass emission rates of different sources. Practices employed by vehicle and station operators were observed and reviewed. Vehicle and station technologies were recruited to represent those natural gas technologies most likely to be part of the near future HD transportation sector. Speed-time driving cycles for chassis testing and routes for onroad testing of vehicles were selected to produce a data set containing emissions and fuel consumption data representative of typical refuse, transit bus, and over-the-road (OTR) tractor operation. Data were gathered using applicable detection and sampling methodologies from in-service vehicles and stations. Laboratory experiments were developed to simulate real-world situations where in-use measurements were not feasible. Sources. Sources of methane emissions from this sector include leaks from mechanical fittings at stations and on vehicles, vents from storage tanks, compressors and fueling systems, releases during fueling hose disconnects, and vehicle tailpipe and crankcase vent emissions, as shown in Figure 1. Methane Quantification. Several different sampling systems were used to quantify methane emissions. Vehicle tailpipe methane emissions were quantified using either full-flow dilute exhaust or raw exhaust sampling systems where the methane concentration and molar flow of the exhaust were used B

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

known representativeness of measured values. Future work should examine additional samples to include with the data provided herein. Alternative methods such as bootstrapping could then be employed to assess uncertainty beyond the MU presented. Substantial detail on measurement methods and accuracy is provided in the Supporting Information Sections S1.1 and S1.2. Practices. Operational and maintenance practices were observed through a combination of fieldwork, interaction with fleet owners, vehicle operators, and station personnel, and a review of literature.24,25 These observations allowed for identification of practices that affect methane emissions from the HD transportation sector, both detrimental and beneficial. As examples, leaks and emissions from fittings or valves can be detected and eliminated through scheduled preventive maintenance, and vehicle operators may receive training such that manual venting of on-board LNG tanks is reduced. However, the practices associated with the small number of stations and vehicles in our study may not be nationally representative. Vehicle Selections and Recruiting. Eighteen vehicles were obtained by the researchers directly from public, private, and rental fleets, while two refuse trucks and two OTR tractors were provided by program sponsors, both of whom operated large HD fleets consisting of both diesel and natural gas powered vehicles. The variety of sources and ability to compare data between trucks assured that the test fleet was unbiased. The vehicles examined included five refuse trucks (9 L SI), four transit buses (9 L SI), three OTR tractors (9 L SI), three OTR tractors (12 L SI), four OTR tractors (15 L HPDI), and three OTR tractors with dual-fuel retrofit (DFR) kits. Station Selection and Recruiting. Methane emissions from fueling stations were gathered during research visits to six LNG stations, all fed by cryogenic tanker truck deliveries. Methane emissions were also characterized from eight CNG stations, seven fed directly from pipelines, and one fed from an LNG station (termed L-CNG). The LNG stations visited were all commissioned after 2011. The CNG stations represented cascade fast-fill (CFF), buffer fast-fill (BFF), and time-fill (TF) stations and represented a broader range of ages than for the LNG stations. Fuel-specific methane emissions from LNG station offloads were estimated from observations during six fueling events. Tank Venting Measurement and Analysis. A truck out of service for far longer than the tank design hold time would lose a high fraction of its fuel unless the fuel in the tank was recovered. LNG storage tank boil-off, PRV venting and manual venting of on-board LNG vehicle tanks by drivers prior to fueling were observed during the study. Vehicle tanks may also be vented prior to maintenance. Venting emissions from in-use LNG tanks could not be characterized directly in the field using the FFS because of the large volume of methane emitted and in the case of boil-off venting, the intermittent nature of events. However, we did record the tank pressure read from the installed pressure gauge before and after venting and the liquid fill level (%) of the tank. And we related this to tank mass loss using laboratory measurement data. Pressure, level, and mass data were collected before and after manual venting of two LNG vehicle tanks from different manufacturers (150- and 120gallon capacities). Existing pressure, level, and mass data were also available from industry for manual venting observations of 120-gallon capacity tanks.26 These data were used to develop an empirical relationship to predict methane mass from fill level and tank pressure before and after venting. An example data

to calculate instantaneous methane mass rates. The dilute exhaust sampling system employed a Horiba 7200D-exhaust emissions analyzer to measure dilute exhaust gas constituent concentrations and a subsonic venturi to measure dilute exhaust flow. Raw exhaust emissions were measured with a Horiba OBS-2200 exhaust emissions analyzer to measure exhaust gas constituent concentrations and the exhaust flow was measured using a pitot flow meter. Both emissions analyzers used heated flame ionization detectors to measure methane concentration and analyzer operation, calibration and verification met the requirements for HD engine emissions certification testing.18 Fuel consumption rates were determined using carbon balance methods where the carbon mass in the exhaust and, for dilute exhaust sampling, in the dilution air (carbon dioxide, carbon monoxide, and hydrocarbons) measured were chemically balanced using the carbon mass fraction of the fuel. The full flow sampling system (FFS − described below) was used to directly characterize crankcase methane emissions during chassis testing of SI engine powered vehicles. Crankcase methane emissions were not characterized separately from tailpipe emissions during on-road testing but, when practical, the vehicle was operated over the same driving route two times, once with the crankcase vent emissions routed to the tailpipe sampling system and once with tailpipe-only measurements. These replicate tests provided estimates of crankcase methane emissions using differencing from on-road testing. It should be noted that crankcase vent emissions are included with tailpipe emissions when performing EPA certification of HD engines as per Part 1065.130(i) of the CFR.18 For tailpipe and crankcase emissions and for fuel consumption, a conservative relative measurement uncertainty (MU) for an engine under power was ±5% and at idle was ±15% of the reported value. The higher relative MU for idle measurements is the result of higher uncertainties associated with measuring lower exhaust flow (raw exhaust measurements) and lower gas concentrations (dilute exhaust measurements) at idle. See Supporting Information Section S1.1 for additional details regarding vehicle emissions measurements. We used a hand-held methane detector to locate leaks from stations and on-board fuel storage and transfer systems. Methane emissions from leaks, compressor vents, engine crankcase vents during chassis dynamometer testing, and from LNG tank vents during laboratory experiments were quantified using a FFS that we developed, similar in concept to the constant volume sampler used for dilute exhaust emissions characterization. The FFS utilized a blower to draw a dilute mixture of the methane and ambient air using a sample hose while dilute sample flow and methane concentration were measured by a Los Gatos Research Ultraportable Greenhouse Gas Analyzer that employed cavity enhanced laser absorption. The MU for the FFS was ±4.4% of the reported value. Laboratory measurements, used to support modeling of vent mass for LNG vehicle fuel tank venting, were made by continuously measuring the mass of the suspended tank using calibrated strain gauges. We quantified methane emissions from HPDI vehicle fuel system vents using mass flow sensors. Methane emissions detection and quantification are detailed in Supporting Information Section S1.2 and in the literature.22,23 Measurement Uncertainty. MU is reported for the numerous sampling systems and methodologies used to quantify methane emissions and fuel consumption during this research. Due to small sample sizes, typical statistical uncertainty (SU) cannot be presented based on lack of C

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

⎛ ⎛g⎞ ⎞ ⎛ g ⎞ 0.5 ⎟· P + 0.0229g /s , R2 = 0.79, n = 21⎟ ⎜m̅ ⎜ ⎟ = 0.0026⎜ ⎝ s ·kPa 0.5 ⎠ ⎝ ⎝s⎠ ⎠

point was also included from an industry method to calculate boil-off emissions.27 Laboratory observations were also made using two LNG vehicle tanks to characterize the influence of ambient conditions, fill level, and initial tank pressure on pressure rise rates and to validate LNG tank pressure rise models based on heat transfer prediction. Thermodynamic and heat transfer models were developed to predict LNG station storage tank pressure and boil-off gas vent mass as a function of station design, operational parameters, and utilization. The goal in developing the pressure rise model was to identify the frequency at which threshold settings of PRVs would be exceeded while the vent mass model was developed to estimate the mass of methane emitted during such occurrences. See Supporting Information Section 1.3 for additional numerical model details. Performance of the pressure rise model was evaluated using data sets gathered from two LNG stations over extended periods of operation. While these models were partially validated against vent mass and pressure rise measurements of vehicle tanks in a laboratory setting, a more complete validation was not possible, as these experiments did not include removal/replacement of liquid fuel and return of vapors associated with station utilization. Vehicle Emissions. Speed-time driving schedules for chassis dynamometer measurements included the HD Urban Dynamometer Driving Schedule28 (UDDS) for all of the vehicles in the study, the Air Quality Management District (AQMD) collection and transit cycles for refuse trucks,29 the Orange County Transit Authority and Manhattan driving cycles30 for transit buses, and the high-speed cruise and transient modes of the Heavy HD Diesel Truck (HHDDT) schedule for OTR tractors.31 On-road measurements were performed over routes and with vehicles loaded such that typical vocational service was simulated. For example, emissions measurements from refuse trucks were made both during low-speed stop-and-go collection/compaction activity and during higher speed transit operation to/from transfer facilities. Crankcase methane emissions were not measured from some of the SI engines characterized during the early stages of the program but a predictive model for crankcase emissions from SI engines based on measured vehicle parameters was developed from those vehicles where crankcase vent emissions were characterized to augment the data set. We collected engine boost pressure data in parallel with crankcase vent emissions to develop an empirical model for use in estimating crankcase vent emissions for tests on stoichiometric 9 L natural gas engines where only tailpipe methane was measured. Boost pressure was selected as the predictive parameter as crankcase vent methane primarily arises from uncombusted methane leaking past the piston rings and into the crankcase with cylinder pressure as the driving force.32,33 Crankcase methane emissions rates and values of the square root of boost pressure for all available data points were averaged by vehicle and driving schedule for each individual test. A test-averaged model was developed considering transport delay differences between measurements of boost pressure, where response is near instantaneous, and crankcase methane, where the measurements were influenced by transport delay and diffusion. A simple linear regression of the available boost pressure (P) and crankcase methane emissions (m) data resulted in an empirical model to predict crankcase methane emissions from 9 L SI natural gas engines

This empirical model resulted in predicted methane mass emissions errors ranging from 16% under prediction to 22% over prediction. While our model would not be as accurate for an individual vehicle due to differences in engine wear and vocation, it serves well to predict for an entire fleet. Vehicle tailpipe and crankcase emissions data were segregated into short periods of activity termed microtrips.34 Each of these microtrips was then classified as idle (40 mph) based on average speed to facilitate fleet modeling based on composite activity and vehicle miles traveled (VMT).



MEASUREMENT RESULTS LNG Offload Losses. We characterized six bulk LNG deliveries at a single station. Average methane emissions from these offloads, estimated using physical dimensions of associated piping and fuel properties, were 0.128% (0.071%, 0.076%, 0.077%, 0.078%, 0.086%, and 0.381%) of the estimated delivered fuel. The primary source of these emissions was hose coupling disconnection where an estimated 11.1 kg of fuel was released during each offload. The average quantity of fuel delivered was 14 627 kg. Additional minor losses occurred but were lower than predefined study measurement threshold. LNG delivery loss estimation methods and uncertainties are provided in Supporting Information Section S1.3.7. LNG Station Boil-Off Emissions. Thermodynamic and heat transfer models were developed to examine relationships between throughput and boil-off. We collected utilization data and storage tank pressure and level data over three week periods from two LNG stations. The utilization data were segregated into 14 contiguous time segments ranging from 37 to 91.5 h in duration and used to predict storage tank pressure using a combined thermodynamic heat transfer station model. The model, on average, over predicted pressure rise rate by 19.7% with error from 0.7% under prediction to 37% over prediction. The model was modified to address the average error and then used as the method to predict station pressure rise and boil-off for a given scenario. We attributed model error to incomplete knowledge of station specifications, errors in estimating the mass/quality of vent gases returned to the station, and physical factors such as stratification in the tank. Losses due to transient cooling of lines were modeled only approximately. As a simplified verification, the results of the vent mass prediction model compared well to laboratory observations of LNG vehicle tanks, with the model over predicting vent mass by 2.9% on average, with an error range of −0.7 to 12.5% (see Supporting Information Section S1.3.8). Using the model it was determined that LNG boil-off emissions are a function of throughput and vehicle type. With an assumed station tank size of 25 000 gallons and fueling SI vehicles, the station must dispense 2000 LNG gallons per day to prevent boil-off emissions while for a station serving HPDI vehicles the required throughput is 3000 gallons per day. As reference, a severely underutilized station serving HPDI vehicles (throughput of only 1500 gallons per day) could experience a boil-off rate of 2% of throughput. However, comprehensive data for current station utilization are not known. LNG Station Continuous and Nozzle Emissions. Emissions rate measurements of detected continuous leaks at six LNG stations ranged from 0.01 to 53.1 g per hour (g/h)

D

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

venting was not collected, it was observed that a large majority of trucks did not vent their tanks at the station prior to fueling. Additional details on manual tank vent emission estimations and model development are included in Supporting Information Sections S1.3.12 and S1.3.13. The scenario below assumes that only 5% or 1 in 20 vehicle tanks would be vented as based on observations. CNG Fuel Station Compressor, Component, and Nozzle Emissions. Emissions were measured from eight CNG stations. Continuous unintentional leaks and intentional venting (valve actuators, regulator bleeds, continuous gas analyzer vents) from CNG station components, excepting those associated with compressors, averaged 35.7 g/h and ranged from 0.6 to 90.0 g/h (±4.4%). Leaks and vents associated with CNG compressors averaged 388.0 g/h during compressor operation and 51.8 g/h while the compressors were idle. Emissions from compressor crankcase and packing vents at two stations dominated the average CNG compressor emissions rate. The compressors at one station were not isolated from high-pressure CNG when idled resulting in compressor poweroff emissions that were orders of magnitude higher than measured at the other stations. Additional CNG station emissions details are found in Supporting Information Section S1.4. Fast-fill CNG vehicle fueling nozzle emissions were characterized for fueling of 36 vehicles. Sources of emissions associated with fast-fill fueling are vents from fueling system actuators, fuel hose depressurization vents, and nozzle disconnection dead volume. Hose depressurization vents were characterized separately from nozzle disconnects since only a single sampling system was available and sources were separated by a significant distance. The average methane mass emitted during fast-fill CNG fueling was 3.0 g/event from the nozzle vent, 0.5 g/event from nozzle disconnection (dead volume), and 0.3 g/event for fueling system actuators: total 3.7 g/event (MU < 0.15 g). Vehicle Fuel System Leaks. Leak audits were performed on all vehicles included in the emissions study. No continuous leaks were found from CNG vehicles while a single leak identified on an LNG vehicle that was below the quantification limit of 0.24 g/h. Engine Related Emissions. A total of 112 h of engine related emissions and fuel consumption data were collected from NG vehicles using a chassis dynamometer covering 3119 miles. Emissions and fuel consumption data were collected during 68 h of on-road operation covering 2014 miles. These data were segregated into microtrips and classified as idle, city, arterial, or highway operation. Engine related methane emissions measurements are detailed in Supporting Information Section S1.5. Figure 3 shows the average microtrip fuel-specific engine related methane emissions (tailpipe + crankcase vent) for all of the study vehicles powered by 9 L SI engines. Similar data were obtained for other engine types (12 L SI, 15 L HPDI, Retrofit). Engine related activity average fuel-specific methane emissions rates ranged from 0.5 to 3% of the fuel consumed for nonretrofit NG vehicles as shown in Figure 4 (Details on sample sizes and measurement uncertainty are presented in Table 1). Observed fuel-specific tailpipe methane emissions from three dual fuel retrofit vehicles (two manufacturers) over individual tests (chassis and on-road) ranged from 3% to 47%. Averages for substitution of diesel fuel by natural gas over microtrips ranged from 4% to 34% on an energy basis. Average fuel specific crankcase methane emissions from the retrofit vehicles were 1%.

with continuous leak rates at three stations below 1 g/h (average 12.8 g/h; MU < ± 0.6 g/h). LNG fueling nozzle emissions were quantified from 43 refills. Only recent technology dispensers were monitored as per study design. The largest methane mass emitted during a single refill exceeded the sampling system measurement range. This mass was estimated to be larger than the 330.4 g that was captured, and represented more than 43% of the total LNG nozzle emissions quantified in the program. Including this event, the average emitted methane was 17.7 g/event (MU < ± 0.8 g/ event). These emissions rate measurements and their distribution are detailed in Supporting Information Sections S1.3.9 to S1.3.11. Manual Venting of LNG Vehicle Tanks to Atmosphere. Vent mass is plotted against a pressure/fill variable for LNG vehicle tanks in Figure 2. These data include values measured

Figure 2. Mass vented as a function of pressure and fill level for 120/ 150 gallon capacity tanks.

under laboratory conditions by the authors (n = 31), values reported by the industry (n = 4), and a single value computed using an industry method for estimating vent mass.27 A logtransformed, least-squares regression of vent mass on pressure drop in the tank due to venting (dP), the square root of the initial pressure before venting (Pi), and the tank fill level (%Fill) from the available data is shown in Figure 2. The empirical model was used to estimate the vent mass from ten observations of manual venting of vehicle LNG tanks by drivers in the field. From tank level and pressure data collected for each observation, the average estimated vent mass was 5.0 kg per event. The average mass of fuel delivered to the vehicle tanks during a fueling event was 118.5 kg, therefore, the fuelspecific methane emissions were estimated as 4.2% for the vehicles that were manually vented. One of the LNG stations that served as a long-term observation station for this research employed a fueling technician and no manual vehicle tank venting was observed at this station. This fuel-specific rate is only for tanks that were observed being manually vented to the atmosphere and illustrates that this practice could be a significant contributor to overall methane emissions from LNG vehicles. While the number of tank venting events characterized serves to justify the reported average methane emissions per event, the statistical frequency of how often tank venting incidents occur throughout the PTW sector was not determined. Estimating incident/event related fuel-specific emissions across the PTW sector requires estimates or observations of incident/event frequency and the associated fuel throughput. While complete statistical data on manual E

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

chain, acquiring emissions factors for each major source associated with vehicles and fueling systems. Data were acquired from 22 vehicles, 8 CNG stations, and 6 LNG stations. By neglecting older technology, we did not address a current inventory. Moreover, a national inventory would require additional measurements and a better understanding of the current fleet and infrastructure and their operation. Generally, there is a need for more measurements at current and new fueling facilities to provide a larger statistical sample for developing emissions factors. However, our data can be used to augment or update models such as GREET or other LCA estimates and provide emissions factors for use in future predictions. GREET was recently updated to include new data for methane leaks and losses across the natural gas production and distribution portions that were beyond this scope of work,36 but could draw on these data for a more comprehensive prediction. To understand the relative contribution of emissions sources and areas for additional research we have used the values reported above with a specific scenario, see Supporting Information Section 1.6 for details on the Stasis Scenario. Figure 5 shows the relative contribution of each component, and shows that the tailpipe and crankcase emissions are dominant for the specific scenario−nearly 70% of all methane emissions. Implementing closed crankcase technologies offers the highest reduction potential and it is noted that new natural gas vehicles are incorporating this technology. It is also noted that implementing best practices can mitigate emissions from manual venting of LNG fuel tanks. Vehicle tank venting and station boil-off emissions can be reduced or eliminated by ensuring station and vehicle throughput remains high (>2000− 3000 gallons per day) and mitigation technologies are properly employed. Due to limited sample size and ranges of technologies, future research should include additional station leak measurements to improve confidence of the data, though it is clear that their contribution is relatively low. Future research might also quantify the effects of improving technology and implementing best practices. Our data represent new material, and we cannot find any prior data set for comparison. However, the vehicle emissions data can be compared to the corresponding factors in GREET. GREET 2016 employs methane emissions factors for Wells-toPump (WTP) and WTW for HPDI OTR trucks fueled by LNG, and transit buses, short and long haul OTR trucks, and refuse trucks fueled by CNG. The PTW value was calculated as the difference between WTW and WTP values. This value corresponded to the “operation” column in GREET. Note that the column for nonexhaust methane was not populated for any vehicle type or fuel; however, upon review of the 2015 update it was determined the GREET “operation” only column represented the sum of reported tailpipe plus crankcase methane emissions for 2010 and newer vehicles37,38 − the focus of this study. Our average data and the GREET factors are presented below in Table 2. For each vehicle type, its minimum, maximum, and average methane emissions are presented. The average is equally weighted for idle, city, arterial, and highway emissions and is normalized to fuel energy such that units are presented as grams of methane emitted for each MJ of fuel energy consumed. For OTR trucks, both 9 and 12 L data were averaged and compared with OTR long haul (LH) and short haul (SH). Note that the GREET reference report includes tailpipe and crankcase emissions for the total emissions for HPDI vehicles. We found that crankcase emissions from these

Figure 3. Fuel specific methane emissions (tailpipe + crankcase) during distinct microtrips for all vehicles powered by 9 L SI natural gas engines (n = 1482).

Figure 4. Activity-averaged tailpipe, crankcase vent, and HPDI fueling system vent methane emissions as a percentage of natural gas fuel consumed for vehicle types characterized in the study. These data are the average methane emissions of multiple on road and/or chassis cycles for each vehicle. Error bars represent the summation of measurement uncertainty weighted by emissions source.

These emissions values are high in comparison to current and projected future SI and HPDI engines, and the authors do not believe that trucks of this type will be prevalent in future populations. Variability of emissions data is due to vehicle age, catalyst temperature, engine speed, load, transient behavior, and emissions diffusion between neighboring microtrips.35 Average speed of a microtrip does not define average load: road grade, acceleration, fan power demand, vehicle weight, drivetrain design, wind speed, and frontal area affect load. The ratio of average power demand to engine rated output affects efficiency. These variations are separate from MU. Emissions differences between engine types are due in part to different duty cycles. Note that HPDI vehicles do not have significant crankcase methane emissions due to their injection strategy, but they have dynamic fuel system venting that contributes to methane emissions. This dynamic venting is related to transient operation of the vehicle as can be seen in the HPDI city operation. SI crankcase emissions are relatively high at idle compared to tailpipe emissions, but in other modes, they are more proportional to exhaust methane emissions.



DISCUSSION We conducted a study examining the current methane emissions from the PTW sector of the natural gas supply F

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology Table 1. Component Level Results for Methane Emissions under the PTW Studya methane emissions source eight CNG station audits across U.S.

six LNG station audits across North America

22 vehicles across U.S.d,e

samples

min

max

std. dev.

average

leaks/losses compressor on compressor off nozzle emissions nozzle vent emissions

8 8 8 22 14

0.60 0.00 0.80 0.02 0.51

90.00 1519.70 376.90 1.15 6.54

31.84 593.90 131.52 0.38 1.36

35.69 388.01 51.76 0.46 2.97

leaks nozzle emissions

6 43

0.00 0.12

53.10 165.20

20.73 25.91

12.80 17.7

station deliveryc

6

0.71

3.81

1.13

1.28

vehicle tank ventingb

31

29.16

250.19

49.75

143.96

transit bus 9-L (tailpipe) transit bus 9-L (CC) refuse truck 9-L (tailpipe) refuse truck 9-L (CC) OTR - HPDI (tailpipe) OTR - HPDI (dynamic vent) OTR - 9-L (railpipe) OTR - 9-L (CC) OTR - 12-L (railpipe) OTR - 12-L (CC)

12

2.08

10.48

2.73

7.83

12

6.38

14.09

2.60

9.45

20

0.32

9.79

2.58

2.81

20

5.37

20.93

4.76

9.68

16

3.53

21.03

4.47

8.12

4

0.00

22.10

9.51

9.27

12

1.38

13.35

3.84

5.55

12 12

3.61 0.18

22.26 5.07

5.78 1.53

9.90 2.45

12

3.72

39.05

9.54

10.19

measurement technique

units g/h

measurement uncertainty

full flow sampling system (FFS)

±4.4%

Estimates

±20%

Strain Gauge

±2%

MEXA 7200D and OBS-2200 FFS

±5% power, ±15% idle ±4.4%

MEXA 7200D and OBS-2200 FFS

±5% power, ±15% idle ±4.4%

MEXA 7200D and OBS-2200 FFS

±5% power, ±15% idle ±4.4

MEXA 7200D and OBS-2200 FFS OBS-2200

±5% power, ±15% idle ±4.4% ±5% power, ±15% idle ±4.4%

g/fueling event

g/h g/fueling event g/kg delivered g/PSI vented g/kg

FFS

a

These results can be applied as estimates to a scenario to predict overall mass emissions. The total numbers of stations and vehicles are presented along with the sample count for each component. The average values are presented for each sample population, which included all measurements (even zero emissions). bLaboratory measurements. cEstimates from observations. dThree dual fuel retrofit vehicles were tested, these vehicles were excluded from the table, fuel specific methane emissions ranged from 30 to 470 g/kg. eThe average fuel specific emissions values presented are an average from all vehicles of this time and equally weighted modes of operation.

Table 2. Comparison of Newly Collected Tailpipe Plus Crankcase Methane Emissions with GREET 2016 Dataa tailpipe + crankcasec HPDI transit bus OTR LH OTR SH refuse

WTPb

WTWb

PTWb

min

max

average

0.28 0.37 0.37 0.37 0.37

0.4 0.62 0.49 0.48 0.65

0.12 0.25 0.12 0.11 0.28

0.19 0.31 0.16

0.61 0.44 0.47

0.36 0.37 0.47

0.16

0.37

0.26

a

For comparisons, our results have been normalized to fuel energy and units are grams of methane emitted per MJ of fuel energy consumed. b GREET emission factors. cThis study.

Figure 5. Absolute and relative contribution of methane emissions by component for the base case scenario. Numbers represent the average methane loss per unit of fuel used (g/kg). The percentage values reflect the contribution of each source to the total PTW emissions.

crankcase emissions were higher than those currently employed by GREET. The GREET PTW values do not include emissions factors for sources beyond the tailpipe and crankcase. Inclusion

engines was negligible but did include the dynamic vent emissions within our totals for comparison with GREET data. In all cases but refuse vehicles, our average tailpipe plus

of station and fueling associated emissions, and vehicle fuel system emissions would increase these values. G

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology



ASSOCIATED CONTENT

TAP SC SAP NG MMT WTW WTP

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.5b06059. Additional information as noted in the text (PDF)





AUTHOR INFORMATION

Corresponding Author

*Phone: 304-293-6457; e-mail: [email protected].

technical advisory panel steering committee scientific advisory panel natural gas million metric tons Wells-to-Wheels Wells-to-Pump

REFERENCES

(1) Bennick, C. Natural gas charges into the future. OEM OffHighway, Jan−Feb 2013 http://www.oemoffhighway.com/article/ 10855398/natural-gas-charges-into-the-future. (2) United State Energy Information Administration, Independent Statistics and Analysis (Accessed March, 2016), U.S. Natural Gas Gross Withdrawals and Production, http://www.eia.gov/naturalgas/. (3) Brandt, A. R.; Heath, G. A.; Kort, E. A.; O’Sullivan, F.; Pétron, G.; Jordaan, S. M.; Tans, P.; Wilcox, J.; Gopstein, A. M.; Arent, D.; Wofsy, S.; Brown, N. J.; Bradley, R.; Stucky, G. D.; Eardley, D.; Harris, R. Methane Leaks from North American Natural Gas Systems. Science 2014, 343 (6172), 733−735. (4) Alvarez, R. A.; Pacala, S. W.; Winebrake, J. J.; Chameides, W. L. Hamburg S. P. Greater focus needed on methane leakage from natural gas infrastructure. Proc. Natl. Acad. Sci. U. S. A. 2012, 109 (17), 6435− 6440. (5) Cucchiella, F.; D’Adano, I.; Gastaldi, M. Profitability Analysis for Biomethane: A Strategic Role in the Italian Transport Sector. Int. J. Energy Econ. Policy 2015, 5 (2), 440−449. (6) Full Fuel Cycle Assessment: Well-to-Wheels Energy Inputs, Emissions, and Water Impacts. State Plan to Increase the use of Nonpetroleum Transportation Fuel; California Energy Commission (CEC-600-2007004-F), June 2007. (7) Allen, D. T.; Torres, V. M.; Thomas, J.; Sullivan, D. W.; Harrison, M.; Hendler, A.; Herndon, S. C.; Kolb, C. E.; Fraser, M. P.; Hill, A. D.; Lamb, B. K.; Miskimisns, J.; Sawyer, R. F.; Seinfeld, J. H. Measurements of methane emissions at natural gas production sites in the United States. Proc. Natl. Acad. Sci. U. S. A. 2013, 110 (44), 17768−17773. (8) Environmental Defense Fund. Gathering facts to find climate solutions: An unprecedented look at methane from the natural gas system. 2014 http://www.edf.org/sites/default/files/methane_ studies_fact_sheet.pdf (accessed July 14, 2014). (9) Cai, H., Burnham, A., Hang, W., Vyas, H. The GREET Model Expansion for Well to Wheels analysis of Heavy-Duty Vehicles; Argonne National Laboratories, Energy Systems Division, Report 15/9, Revision 1. (10) GREET Life-Cycle Model; Center for Transportation Research, Energy Systems Division, Argonne National Laboratory, October 3, 2014. (11) Whyatt, G. A. Issues Affecting Adoption of Natural Gas Fuel in Light- and Heavy-Duty Vehicles, PNNL-19745; Pacific Northwest National Laboratory, 2010. (12) Gladstein, C., Couch, P., Wake, M., Medlock, C. (2014) Pathways to Near-Zero-Emission Natural Gas Heavy-Duty Vehicles. Gladstein, Neandross & Associates: White Paper http://www. gladstein.org/pdfs/On-Road_Pathways.PDF (accessed July 14, 2014). (13) Kamel, M.; Lyford-Pike, E.; Frailey, M.; Bolin, M.; Clark, N. N.; Nine, R. D.; Wayne, W. S. An Emission and Performance Comparison of the Natural Gas Cummins Westport Inc. C-Gas Plus versus Diesel in Heavy-Duty Trucks. SAE Tech. Pap. Ser. 2002, 111 (1), 1409−1421. (14) Willner, K. Testing of Unregulated Emissions from Heavy Duty Natural Gas Vehicles; Swedish Gas Technology Centre Report, 2013, 289. (15) First Generation Westport HPDI Technology: Westport High Pressure Direct Injection Technology http://www.westport.com/is/ core-technologies/combustion/hpdi (accessed July 15, 2014). (16) Koehler, E.; Dahodwala, M. Dual fuel for on-highway HD applications, Automotive World Webinar, November 12, 2014. http:// www.automotiveworld.com/download/225635/.

ORCID

David L. McKain: 0000-0002-8874-4180 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Support was provided by the Environmental Defense Fund, Cummins, Cummins Westport, Royal Dutch Shell, the American Gas Association, Chart Industries, Clean Energy, the International Council on Clean Transportation, PepsiCo, Volvo Group, Waste Management, and Westport Innovations. Funding for EDF’s methane research series, including the West Virginia University study, is provided for by Fiona and Stan Druckenmiller, Heising-Simons Foundation, Bill and Susan Oberndorf, Betsy and Sam Reeves, Robertson Foundation, Alfred P. Sloan Foundation, TomKat Charitable Trust, and the Walton Family Foundation. Support was also provided by West Virginia University’s George Berry Chair endowment. We also acknowledge the support provided by the WVU Transportable Chassis Testing Laboratory personnel including Drs. Arvind Thiruvengadam and Marc Besch.



ABBREVIATIONS CC crankcase CNG compressed natural gas LNG liquefied natural gas HPDI high pressure direct injection SI spark-ignited HD heavy-duty GHG greenhouse gases PTW pump-to-wheels EPA U.S. Environmental Protection Agency NFPA National Fire Protection Association psig pounds per square inch gage FFS full flow sampler MU measurement uncertainty L-CNG liquefied and compressed natural gas CFF cascade fast fill BFF buffer fast fill TF time fill OTR over-the-road DFR dual-fuel retrofit gal/day gallons per day g/h grams per hour g grams kg kilograms dP change in pressure Pi initial pressure %Fill tank fill level mph miles per hour L liter TP tailpipe H

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology (17) Ebner, H. W.; Jaschek, A. O. The Importance of Blow-By Measurements, Measuring Equipment Required and Implementation. SAE Tech. Pap. Ser. 1998, 981081. (18) Code of Federal Regulations Title 40: Protection of the Environment, Part 1065: Engine Testing Procedures; Environmental Protection Agency, 2014. (19) Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty Engine and Vehicles-Phase 2, U.S. Federal Register, Vol. 80, No. 133, p 40208, 2015. (20) Cummins Westport. ISL G Near Zero Natural Gas Engine Certified to Near Zero - First MidRange engine in North America to reduce NOx emissions by 90% from EPA 2010 [Press release]; 2015. Retrieved from http://www.cumminswestport.com/press-releases/ 2015/isl-g-near-zero-natural-gas-engine-certified-to-near-zero. (21) NFPA 52: Vehicular Gaseous Fuel Systems Code; National Fire Protection Association, Quincy, MA, 2010. (22) Johnson, D.; Covington, A.; Clark, N. Methane Emissions from Leak and Loss Audits of Natural Gas Compressor Stations and Storage Facilities. Environ. Sci. Technol. 2015, 49, 8132. (23) Johnson, D.; Covington, A.; Clark, N. Design and Use of Full Flow Sampling System (FFS) for the Quantification of Methane Emissions. J. Visualized Exp. 2015, DOI: 10.3791/54179. (24) Powars, C. A. Best Practices to Avoid LNG Fueling Station Venting Losses; St. Croix Research, 2010. (25) Recommended Practices for LNG Powered Heavy-Duty Trucks, SAE Truck and Bus Alternative Fuels Committee, Society of Automotive Engineers, 2008. (26) Dixon, D., NorthStar, Inc. Personal Communication. May 2014. (27) Ursan, M. What is boil-off? Working paper presented at UNECE Task Force on Liquefied Natural Gas Vehicles, November 3, 2011. (28) Code of Federal Regulations Title 40: Protection of the Environment, Part 86: Engine Testing Procedures, Appendix I: Dynamometer Schedules; Environmental Protection Agency, 2014. (29) Walkowicz, K.; Proc, K.; Wayne, S.; Nine, R.; Campbell, K.; Wiedemeier, G. Chassis Dynamometer Emission Measurements from Refuse Trucks Using Dual-FuelTM Natural Gas Engines; SAE Tech. Paper 2003-01-3366, 2003. (30) SAE Surface Vehicle Recommended Vehicle Practice J2711, Recommended Practice for Measuring Fuel Economy and Emissions of Hybrid-Electric and Conventional Heavy-Duty Vehicles, September 2002. (31) Clark, N. N.; Gautam, M.; Wayne, W. S.; Thompson, G. J.; Lyons, D. W. California Heavy Heavy-Duty Diesel Truck Emissions Characterization for Project E-55/E-59 Phase 1.5, CRC Project E-55/ E-59 Report, Coordinating Research Council, August 2004. (32) Rose, A. H.; Stahman, R. C. The Role of Engine Blowby in Air Pollution. J. Air Pollut. Control Assoc. 1961, 11 (3), 114−144. (33) Heywood, J. B. Internal Combustion Engine Fundamentals; McGraw Hill, Inc., New York, 1988. (34) Gautam, M.; Clark, N.; Riddle, W.; Nine, R.; Wayne, W.; Maldonado, H.; Agrawal, A.; Carlock, M. Development and Initial Use of a Heavy-Duty Diesel Truck Test Schedule for Emissions Characterization, SAE Tech. Pap. Ser., 2002, SAE 2002−01− 1753.10.4271/2002-01-1753 (35) Madireddy, M. R.; Clark, N. N. Sequential Inversion Technique and Differential Coefficient Approach for Accurate Instantaneous Emission Measurement. Int. J. Engine Res. 2006, 7, 437−446. (36) Burnham, A. Updated Fugitive Greenhouse Gas Emissions for Natural Gas Pathways in GREET1_2016 Model, Argonne National Laboratory, 2016. https://greet.es.anl.gov/publication-updated-ghg2016. (37) Cai, H., Burnham, A., Wang, M., Hang, W., Vyas, A. The GREET Model Expansion for Wells-to-Wheels Analysis of Heavy-Duty Vehicles; Argonne National Laboratory, 2015. https://greet.es.anl.gov/ publication-heavy-duty. (38) Personal Communication. Email Correspondence with Michael Wang and Andrew Burnham, Argonne National Laboratory.

I

DOI: 10.1021/acs.est.5b06059 Environ. Sci. Technol. XXXX, XXX, XXX−XXX