In-Use Measurement of the Activity, Energy Use, and Emissions of a Plug-in Hybrid Electric Vehicle

Transcription

1 Frey, H.C., H.W. Choi, E. Pritchard, and J. Lawrence, In-Use Measurement of the Activity, Energy Use, and Emissions of a Plug-in Hybrid Electric Vehicle, Paper 2009-A-242-AWMA, Proceedings, 102nd Annual Conference and Exhibition, Air & Waste Management Association, Detroit, Michigan, June 16-19, In-Use Measurement of the Activity, Energy Use, and Emissions of a Plug-in Hybrid Electric Vehicle Paper 2009-A-242-AWMA H. Christopher Frey, Hyung-Wook Choi North Carolina State University, Department of Civil, Construction, and Environmental Engineering, Campus Box 7908, Raleigh, NC Ewan Pritchard, Advanced Transportation Center, MRC Suite 339, Raleigh, NC Josh Lawrence Advanced Energy, 909 Capability Drive, Suite 2100, Raleigh, NC ABSTRACT The purpose of this study is to demonstrate methodology for characterization of a plug-in hybrid electric vehicle (PHEV), taking into account gasoline and electricity consumption and emissions associated with each. Field measurements were made of a Toyota Prius with 1.5 liter gasoline engine, Hybrid Synergy Drive (HSD) system with an original battery, and retrofitted Hymotion plug-in system with a second battery. The PHEV initially operates in charge-depleting mode (CD) until the Hymotion battery charge reaches a set point, after which it operates in chargesustaining mode (CS) using only the original battery. Three systems were used for in-use monitoring of the PHEV: (a) electronic download from the hybrid control system interface for factors such as battery charge, voltage, and current, and on-board diagnostic (OBD) data such as engine RPM, manifold absolute pressure, intake air temperature, road speed, and others; (b) portable emission monitoring system (PEMS) measurement of exhaust gas concentrations; and (c) GPS monitoring of coordinates and of altitude using a barometric altimeter. These data were used to characterize the activity of the PHEV, the energy flow associated with the batteries and diesel engine, and the tailpipe emissions. Results are presented based on in-use data collection for real-world driving cycles, in order to demonstrate methodology for integrated analysis of a plug-in hybrid system. Fuel economies for CD and CS modes were approximately 60 and 40 mpg. The indirect electricity emission factors were estimated based on EPA egrid and National Emission Inventory data. An engine load-based model based on vehicle-specific power (VSP) was developed to explain variation in battery current, fuel use and emission rates based on the real-world data. INTRODUCTION Plug-in Hybrid Electric Vehicles (PHEVs) could mitigate two critical transportation problems: air pollution and energy security. Transportation accounts for 28% of U.S. energy use and contributes 34% of U.S. carbon dioxide (CO 2 ) emissions. 1 Furthermore, approximately twothirds of transportation fuels consumed in the U.S. are imported. Surface transport contributes 40% of national annual nitrogen oxides (NO x ) emissions, 56% of carbon monoxide (CO), and 1

2 28% of volatile organic compounds (VOC). 2 PHEVs could consume 50% less gasoline than conventional vehicles (CVs). 3,4 The purpose of this project is to develop and apply a methodology for real-world evaluation of PHEVs to estimate electricity use, fuel use, and emissions. The methodology features the use of portable emission measurement systems (PEMS) for the purpose of quantifying the activity, fuel use, and emissions of vehicles during actual duty cycles. To develop and demonstrate the methodology, field data collection was conducted for a selected PHEV. The selected PHEV is a 2005 Toyota Prius hybrid electric vehicle (HEV) that had previously been retrofitted with the A123 Hymotion plug-in conversion kit. The activity, electricity use, gasoline fuel use, and emissions for this vehicle were measured on a second-by-second basis during real-world operation on six selected routes in the Research Triangle Park, NC region. CONFIGURATION AND OPERATING MODES OF A PLUG-IN HYBRID ELECRIC VEHICLE HEVs are powered by an internal combustion engine (ICE) and one or more electric motors. The motors can use electricity discharged from a battery in order to provide power in addition to that from the ICE for the powertrain. Conversely, one or more of the motors can operate in reverse as a generator, to convert shaft output from the engine or from the powertrain during braking to generate electricity for recharging the battery. The advantages of an HEV can include: (a) reducing the peak power demand on the ICE for a given total vehicle power output, thereby enabling use of smaller engine than would otherwise be required; (b) reducing the transients on the power demand from the engine so that the engine can operate more efficiently; and (c) avoiding use of the engine entirely for situations that require low power demand, such as low speed driving. The Toyota Prius HEV has a 1.5 liter engine and two electric motorgenerators (MG). The gasoline engine produces 76 horsepower. The motors can produced a maximum output of 67 horsepower. Electric motors have some advantages over ICEs in terms of their torque. HEVs can achieve high fuel economy in citing driving, in particular, because the engine can be shut down during periods of low power demand and energy can be recovered during regenerative braking as part of stop-and-go driving. The A123 Hymotion retrofit for the Prius involves adding an additional battery that is recharged only from the electrical grid. The original battery of the Prius is referred to as the traction battery. Thus, the retrofitted Prius has two batteries: the original traction battery and the retrofitted Hymotion or plug-in battery. The traction battery is recharged during vehicle operation only, and is not recharged using any power from the electrical grid. The plug-in battery is recharged only from the electrical grid, and is not recharged during vehicle operation. The retrofitted Prius has two major battery usage modes: (1) charge depleting (CD); and (2) charge sustaining (CS). Charge depleting mode occurs from the time the vehicle is first operated after the plug-in battery is recharged from the grid. During this initial operating time, the plug-in battery is continuously discharged until it reaches a state-of-charge (SOC) set-point, after which time the battery is no longer used during vehicle operation until it can later be recharged from the grid. During CD mode, electrical power from the plug-in battery supplements the power available from the traction battery. Vehicle operations after the completion of CD mode are 2

3 referred to as charge sustaining. During the CS mode, the traction battery is discharged and recharged depending on vehicle power demand and operating conditions. However, the traction battery SOC is maintained between lower and upper set-points in order to prolong battery life. The peak electrical power available during CS mode is less than that available during CD mode. The major components of the retrofitted Prius are shown in Figure 1. The Prius has two motorgenerators (MGs) that are referred to as MG1 and MG2. MG1 is also used as an engine starter. The planetary gear (PG) enables transmission of power among the ICE, MGs, and drive shaft. The PG system has three sets of gears, including a sun (or central) gear, planetary gears that revolve around the sun gear, and a ring gear that circumnavigates the planetary gears. This complex configuration of gears enables operating modes with variations in the power split between the ICE and MGs. Examples of operating modes include: (a) low speed cruising during which only the traction battery is used to power the drive train; (b) moderate speed cruising during which the engine is used to power the drive train either supplemented by the traction battery or from which some power is diverted to recharge the traction battery; and (c) regenerative braking, during which MG2 is operated as a generator to extract power from the drive wheels, thereby slowing the vehicle and recharging the traction battery. Figure 1. Power Flow Diagram for Series-Parallel Plug-in Hybrid Electric Vehicle: MG = Motor-Generator. PG = Planetary Gear. While there is not a standardized definition of operating modes for a PHEV, they can be classified based on whether the vehicle is: (a) at zero versus nonzero speed; and (b) decelerating, cruising, or accelerating. Furthermore, operating modes can be stratified taking into account conditions under which the engine is on versus off and whether the traction battery is discharging or recharging. For simplicity, in this paper we focus on differentiating CD and CS modes. 3

4 METHODOLOGY The methodology for this research includes development of a field study design, instrumentation, data quality assurance, and data analysis. The data analysis includes characterization of the activity, energy use, and emissions of the selected PHEV and estimation of indirect energy use and emissions associated with use of electricity from the local power grid to recharge the plug-in battery. Field Study Design Field study design includes specification of all test conditions that are controllable include choice of test vehicle, driver, fuel, initial state-of-charge, routes, and time of day for each route. Because this is a field study, there are factors that are not controllable, including traffic and ambient conditions. The test vehicle is a 2005 Toyota Prius HEV retrofitted with the A123 Hymotion system for conversion to a PHEV. This vehicle, shown in Figure 3(a), had accumulated 103,000 miles prior to the date of the test. The driver for the test was a graduate research assistant. Figure 2. Real World Driving Cycles for a Plug-in Hybrid Prius tested in Raleigh-RTP Area, North Carolina Research Triangle Park (RTP) 1 RDU Airport 2 3 North Raleigh Six Forks Rd. A B C Capital Blvd. North Carolina State University (NCSU) The vehicle was operated on local retail gasoline. The fuel tank was filled and the the plug-in battery was fully-recharged overnight prior to the field measurements. Readily available 120 volt alternating-current is used to recharge the plug-in battery. After the measurements were 4

5 completed, the fuel tank was refilled and the plug-in battery was fully-recharged. The gallons of fuel required to refill the tank were recorded. The electricity required to recharge the battery was measured using a watt-hour meter. The selected routes for field measurements are shown in Figure 2. These routes were identified and used in previous field data collection with conventional light duty gasoline vehicles. The routes are based on two origin and destination (O/D) pairs: (1) North Carolina State University (NCSU) to north Raleigh (NR); and (2) NR to Research Triangle Park (RTP). For each O/D pair, there are three alternative routes. The routes from NCSU to NR are comprised primarily of signalized minor and major arterial roads. The routes from NR to RTP are comprised primarily either of major arterial roads or freeways. The NCSU to NR routes represent local commuting within Raleigh over a one-way distance of approximately 8 to 11 miles, while the NR to RTP routes represent commuting from a residential district to the RTP business district over a one-way distance of approximately 16 to 20 miles. The NCSU to NR routes are denoted as Routes A, B, and C. The average speed on all three of these routes is typically below 30 mph. However, because Route C includes some driving on a local interstate highway ring road, known locally as the Beltline, the average speed for this local route is higher than for the other two. The NR to RTP routes are denoted as Routes 1, 2, and 3. Routes 1 and 2 make maximal use of interstate highways in the region, whereas Route 3 maximizes the use of major arterials. Therefore, the average speeds of Routes 1 and 2 are typically significantly higher than for Route 3. Field data collection occurred during a weekday from 11:00 AM to 7:00 PM on July 15, Instruments There were three key instruments used for field data collection: (1) a data logger for the electronic control units (ECUs) of the retrofitted Prius; (2) a Portable Emission Measurement System (PEMS); and (3) a geographic position system (GPS) with barometric altimeter. There are two types of variables that were measured: externally observable variables (EOVs) and internally observable variables (IOVs). EOVs are variables that can be observed, in principle, by an observer outside of a vehicle. These include, for example, vehicle speed and acceleration, and road grade. IOVs are variables that can only be observed within the vehicle, such as engine RPM and electrical current to or from a battery. A combination of both EOVs and IOVs can be used to provide detailed quantitative characterization of the mass and energy balances for the vehicle, including pollutant emissions. EOVs could be used to develop modeling approaches that could be used with traffic simulation and emission factor models. The datalogger for the ECUs enables recording of both EOVs and IOVs from the on-board computer control systems of the vehicle. The Kviser Memorator was used as a data logger. Data were downloaded from the On-Board Diagnostic (OBD) interface of the vehicle, as shown in Figure 3(b). The data logger records several dozen parameters into a flash memory device. Data 5

6 are reported at various time intervals depending on the specific parameter, with some intervals typically approximately 0.7 seconds. Examples of data collected include vehicle road speed (miles per hour), voltage and current associated with electricity to or from the plug-in and traction batteries, SOC of each battery, and engine RPM. Figure 3. Installation of the Portable Emissions Measurement System (PEMS) and GPS in a Plug-in Hybrid Electric Vehicle: (a) side view of Toyota plug-in hybrid Prius; (b) ECU Memorator; (c) the portable unit on a passenger seat (frontview); (d) sampling lines; (e) sampling probe secured with a hose clamp; (f) GPS receiver on the roof; (g) accessing power from the extra batteries on the vehicle trunk; and (h) power outlet and watt-meter (a) (b) (c) (d) (e) (f) (g) (h) The Portable Emission Measurement System (PEMS) used here is the OEM-2100 Montana system manufactured by Clean Air Technologies International, Inc., as shown in Figure 3(c). 5 The Montana system is comprised of two parallel five-gas analyzers, a particulate matter (PM) measurement system, an engine sensor array, a global positioning system (GPS), and an onboard computer. The two parallel gas analyzers simultaneously measure the volume percentage of carbon monoxide (CO), carbon dioxide (CO 2 ), hydrocarbons (HC), nitrogen oxide (NO), and oxygen (O 2 ) in the vehicle exhaust. HC, CO and CO 2 are measured using non-dispersive infrared (NDIR). The accuracy for CO and CO 2 are excellent. The accuracy of the HC measurement depends on type of fuel used. 6,7 NO is measured using electrochemical cell. NO x is typically comprised of approximately 95 volume percent NO; therefore, NO emissions converted to an equivalent NO 2 mass basis (using the molecular weight of NO 2 ) are a good indicator of total NO x emissions. NO x emissions are typically reported as equivalent NO 2. The performance of the Montana system has been verified in comparison to that of an laboratory grade chassis dynamometer measurement system. 8 The PEMS is calibrated in the laboratory using a cylinder gas and in the field periodically recalibrates to ambient air to prevent instrument drift. The Montana System is designed to measure emissions during the actual use of a vehicle during its regular daily operation. The complete system comes in two weatherproof plastic cases, one of 6

7 which contains the monitoring system itself, and the other of which contains sample inlet and exhaust lines, tie-down straps, AC adapter, power and data cables, various electronic engine sensor connectors, and other parts. The monitoring system weighs approximately 35 lbs. Exhaust gas sample lines were run from the exhaust into the Montana system, as indicated in Figures 3(d) and 3(e). The system consumes 5-8 Amps at 13.8 V DC. For the PHEV tests, the Montana system was connected to a 12 VDC deep cycle marine batteries that were independent of the PHEV power system, as shown in Figure 3(g). Data from the PEMS for exhaust gas concentration and from the ECU data logger for fuel flow rate were used to estimate the exhaust flow rate of each pollutant. From the exhaust gas concentrations, the fuel-to-air ratio can be estimated based on the volume percentage of CO 2. Based on the fuel-to-air ratio and the fuel flow rate reported by the ECU, the inlet air flow and exhaust gas flow are estimated. Based on the exhaust gas flow rate and the exhaust gas concentrations, the mass per time emission rates are estimated. The emissions can also be easily estimated on a mass per distance (e.g., g/mile) or mass per unit of fuel consumed (e.g., g/gallon) basis. A Garmin GPS unit with a barometric altimeter was used to record GPS coordinates and elevation on a second-by-second basis. The coordinates were used to match vehicle location relative to each test route, and to estimate distances between elevation measurements. The differences in elevation over distance were used to estimate road grade. Road grade is a factor that influences vehicle-specific power (VSP) demand and is strongly correlated with the energy consumption of a vehicle. 9,10 The amount of power consumed to recharge the plug-in battery after the field measurements was measured using a watt-hour meter, as shown in Figure 3(h). Data Quality Assurance The initial processing of the raw data from the data logger, PEMS, and GPS with altimeter include the following steps: (1) convert data to a second-by-second basis, if necessary; (2) synchronize the data from multiple instruments into a single database; and (3) screen and evaluate the data for possible errors, correct errors where possible, and remove invalid data that cannot be corrected. Although the last step is focused on quality assurance, these three steps are referred to collectively as the quality assurance procedure for processing the data. For quality assurance purposes, the combined data set for a vehicle run were screened to check for errors or possible problems. If errors were identified, they were either corrected or the affected data were not used for data analysis. The procedure includes identifying and correcting problems associated with the ECU data, such as missing or invalid values of manifold absolute pressure (MAP), engine RPM, and intake air temperature (IAT) and associated with the gas analyzers of the PEMS, such as large discrepancies in simultaneous exhaust gas concentration measurements for a given pollutant when comparing the two gas analyzers, freezing of the analyzers (failure to update data), occurrence of negative values of emissions that are statistically significantly different from zero, 11 and air leakage into the sampling system based on assessment of the observed air-to-fuel ratio. The latter tends to occur in episodic transient situations when 7

8 the engine shuts down or turns on during vehicle operation, during which there may be residual infiltrated air in the tailpipe. A 19-step data screening and quality assurance process has been automated using Visual Basic macros in Excel. Raw data from the Montana system is processed via these macros to identify data quality problems. Data Analysis for Direct and Indirect Energy Use The consumption of gasoline by the ICE is reported by the ECU data logger in terms of microliters of fuel consumed during each reporting period, which is typically approximately an interval of 0.7 seconds. These data were integrated over time and then used to estimate the fuel consumption on a second-by-second basis. From the ECU data, the voltage and current either to or from each battery, and the battery state of charge, was reported. From these data, the total amount of current extracted from the plug-in battery during data collection was estimated. Since the plug-in battery can only discharge during operation, and can only recharge when plugged into the grid, the estimation of the net electrical discharge from the battery during charge depleting mode was straightforward and was estimated on a second-by-second basis. For the traction battery, the amount of electricity associated with either discharge or recharge, in Watt-hours, was estimated on a second-by-second basis. The net electricity obtained from the grid was used to estimate the gasoline-equivalent amount of fuel consumed by power plants to generate and deliver the electricity to the grid. The national average heat rate for power plants, which is the ratio of thermal energy input to electrical energy output, is 10,132 BTU/kWh. 1 The heating value of gasoline is approximately 125,000 BTU/gallon, and its density is 2,790 g/gallon. 11 Thus, each kwh of electricity obtained from the grid corresponds to the potential energy contained in 226 grams of gasoline. On this basis, a gasoline-equivalent fuel economy can be estimated even when the vehicle is operating in part on electricity stored in the plug-in battery from the electrical grid. The energy use was calculated on a second-by-second basis and also summarized on a round-trip basis for each route. Vehicle Specific Power VSP has been found by many investigators to be a useful basis for comparison of vehicle fuel use with key factors that govern power demand on a vehicle engine, including speed, acceleration, and road grade. VSP has typically been used to estimate the gasoline consumption of conventional light duty gasoline vehicles (LDGV). For a generic LDGV, VSP is estimated as: 13,14 r VSP = sin atan Where, v = vehicle speed (km/h) a = acceleration (km/h per sec) r = road grade (%) 3 v a v (1) 8

9 VSP = vehicle-specific power (kw/ton) The terms in the equation account for vehicle movement, acceleration due to gravity, rolling resistance, and aerodynamic drag. Here, VSP is used to account for the energy demand of a PHEV, rather than the fuel demand of an LDGV. VSP is an estimate of the power output required of the drivetrain in order to move the vehicle. In a PHEV, this power output can be met in many ways depending on the power split between the engine and electric motors/generators and whether the vehicle is consuming energy from the electrical grid stored in the plug-in battery. For a conventional LDGV, vehicle fuel use is typically constant for negative values of VSP, and increases linearly with positive VSP. For a PHEV, the relationship of gasoline fuel use versus VSP is not expected to be the same as for a conventional LDGV, since the battery-motor system will sometimes augment the power output of the engine during discharge and at other times may consume some of the engine output for recharging. Thus, a key goal of this work is to assess the relationship of gasoline fuel use and electricity use versus VSP. RESULTS Based on second-by-second data collected for six routes in the RTP area, results are given for quality assurance, engine utilization, energy use for both gasoline and electricity from the grid, tailpipe and indirect emission factors, and fuel and electricity consumption as a function of VSP. Data Quality Assurance A toal of 19,677 seconds of data were collected. Of these data, 7.8 percent were not used because of data quality issues, particularly with respect to large discrepancies in simultaneous exhaust gas concentrations when comparing the parallel gas analyzers and situations in which the exhaust gas was highly diluted, which leads to lower precision of estimated emissions factors. Both of these data issues are primarily attributable to the short-term episodes during which the engine is turning on or off. During these times, there are rapid transient changes in the exhaust concentration of CO 2, which may change from approximately 0 volume percent when the engine is off to almost 15 volume percent in a matter of a few seconds when the engine is turned on, and vice versa. Although both gas analyzers typically report the same CO 2 volume percent to within a few tenths of a percentage point, during these transients the discrepancy between the two gas analyzers can be more than a percentage point. At this time, we have chosen not to use such data, but are in the process of revisiting this issue to determine a potentially more suitable approach. Engine Utilization The ICE was found to idle at approximately 900 RPM. The engine turns on or shuts off periodically during vehicle operation depending on factors such as vehicle speed and acceleration and the traction battery SOC. Of course, tailpipe emissions occur only when the engine is on. Use of electricity that had been taken from the grid occurs anytime that the plug-in 9

10 battery is discharged during the CD mode. The frequency with which the engine was started during operation, and the duration of engine use with respect to distance driven, are summarized in Table 1. The results are listed based on the round-trip data for each route. The first route driven was Route A, during which CD mode occurred. The plug-in battery charge was depleted to its setpoint during travel on Route A; therefore, for all other routes, the PHEV was in CS mode. Route Table 1. Characteristics of Engine Use for Plug-in Hybrid Vehicle for each Route Total Travel Time (sec.) Average Vehicle Speed (mph) Total Distance (miles) Distance for Engine On (miles) Engine Starts per mile Charging Mode c A 2, a 9.5 (59%) b 7.64 CD B 3, (68%) 5.78 CS C 2, (73%) 5.91 CS 1 2, (89%) 2.61 CS 2 3, (82%) 2.19 CS 3 4, (75%) 3.26 CS a 5 miles of segment is not included due to loss of power for Montana. b Percent of travel distance for engine on c CD = charge depleting mode. CS = charge sustaining mode. The average speeds were lowest for Routes A and B, which are comprised mostly of minor and major arterials. The average speeds were highest for Routes 1 and 2, which are comprised mostly of interstate highway travel. Routes C and 3 have similar average speeds. Route C has a mix of interstate, major arterial, and minor arterial roads. Route 3 is primarily comprised of major arterial segments but with fewer signals per mile compared to Routes A and B. As expected, the engine was used less extensively during CD mode than for CS mode. Specifically, during CD mode, the engine was used 59 percent of the distance driven, versus 68 percent or more for the five routes during which the vehicle was in CS mode. However, during CD mode, the engine was turned on more frequently. This may be in part an artifact of the driving characteristics of Route A, which has the lowest average speed among all of the routes for which data were collected. The low average speed is associated with more stop-and-go operation compared to other routes. Nonetheless, the number of engine starts per mile for Route A was significantly higher than those for Routes B and C. Even though Routes B and C have significantly different average speeds, they had a similar number of engine starts per mile during CS mode. Therefore, the difference in the number of engine starts per mile for Route A versus either Route B or C may be in part due to CD mode, and in part due to characteristics of the duty cycle. The higher speed routes had a higher proportion of distance over which the engine was on, and a smaller number of engine starts, then the slower speed routes. For example, Routes 1 and 2, with average speeds of approximately 40 mph, had only 2.2 to 2.6 engine starts per mile, but the engine was on for 82 to 89 percent of the distance driven. This implies that the engine was used for longer durations compared to the lower speed routes. Routes 3 and C have similar average 10

11 speeds but different speed profiles, since the distance between traffic signals is much longer for Route 3 than for Route C. The similarity in average speed is associated with a similar proportion of distance driven during which the engine was on; however, the average number of engine starts per mile for Route 3 was almost half of that for Route C. These results indicate that the location and duration of engine use is highly dependent on short-term episodic events during driving. Energy Use for Gasoline and Grid Electricity The amount of grid electricity and gasoline used during vehicle operation on each of the six routes is given in Table 2. Grid electricity was used only during Route A, for which the vehicle was in charge depleting mode. A total of 1.4 kwh of electricity use was observed during driving on Route A, based on second-by-second estimates of the voltage and current from the plug-in battery. This corresponds to 0.32 kg of equivalent gasoline associated with fuel input to the power plants from which the grid electricity was extracted. Thus, when combined with the actual amount of gasoline consumed by the ICE, which was 0.75 kg, the total equivalent fuel use was 1.07 kg of gasoline. This is equivalent to 66 grams of gasoline equivalent per mile of operation, or 42.3 miles driven per equivalent gallon of gasoline consumed. Since only 70 percent of the total energy use was for direct consumption of gasoline, the fuel economy could also be reported as 60 miles per actual gallon of gasoline consumed. However, during charge sustaining mode, since there was no electricity used from the grind, the fuel economy was 37 to 41 miles per actual gallon of gasoline consumed. Hence, these results imply that the overall energy (not just fuel) economy of the vehicle is approximately the same regardless of the source of energy, which is to be expected. Route Table 2. Equivalent Fuel Use and Fuel Economy based on Fuel Use for Engine and Electricity Use for Hybrid System Total Fuel Use (kg) Net Electricity Use (kwh) Equivalent Fuel Use (kg) Total Equivalent Fuel Use (kg) Engine Fuel Use (g/mile) Electricity Use (kwh/mile) Equivalent Fuel Use for Hybrid (g/mile) Total Equivalent Fuel Use (g/mile) Total Equivalent Fuel Economy (MPG) A B C The rate of fuel consumption is likely to be highly sensitive to driver behavior. Here, only one driver operated the vehicle. At the time these data were collected, the driver was not experienced in driving an HEV or PHEV. Anecdotally, some drivers who are very familiar with the PHEV vehicles claim that they can achieve fuel economy as high as 100 miles per gallon or more during charge depleting mode, when only the direct consumption of gasoline is considered. 11

12 Direct and Indirect Emissions The emission factors for each route are shown in Table 3. The tailpipe emission factors are based on the PEMS measurements. The PEMS estimates emission rates every second for each pollutant in g/sec. These emissions were summed over the round-trip for each route and divided by the round-trip mileage in order to estimate average emissions per mile of vehicle travel. The tailpipe CO 2 emission rates are inversely proportion to fuel economy, since over 99 percent of the carbon in the fuel is emitted as CO 2. For the routes during which the vehicle operated in charge sustaining mode, the CO2 emission rate ranged from approximately 220 to 240 grams per mile. For the route during which the vehicle operated in charge depleting mode, the CO 2 emission rate from the tailpipe was only 147 g/mile, or slightly more than one-third smaller than during charge sustaining mode. Table 3. Measured Tailpipe and Estimated Indirect Emission Factors for each Route Route Source CO 2 (g/mile) NO x (mg/mile) CO (mg/mile) HC c (mg/mile) A a Grid b Tailpipe Total B Tailpipe C Tailpipe Tailpipe Tailpipe Tailpipe Note: a Route A is charge depleting mode and other routes are charge sustaining mode. b Emissions per electricity generation (g/kwh) are calculated based on EPA egrid for CO 2 and NO x and NEI data for CO, HC, and PM for 2005 data (EPA, 2007b and EPA, 2007c). egrid: CO 2 = 553 g/kwh and NO x = g/kwh. NEI: CO = 0.13 g/kwh, HC = g/kwh, and PM = g/kwh c The average exhaust concentrations for HC were below the gas analyzer detection limit of 10 ppm. For NO x, the emission factors ranged from 60 to 130 mg/mile during charge sustaining mode, versus a direct emission rate of only 38 mg/mile during charge depleting model. The two routes with the highest average speed had the highest average tailpipe NO x emission factors, and the lowest average tailpipe CO and HC emission rates. The CO emission rates were highest for Routes A, B, and C, which have many signalized intersections and corresponding stop-and-go driving patterns. The HC emission rates were highest for the two lowest average speed routes. For the grid electricity consumed during CD mode, the indirect emission factors for electric power generation are estimated on the basis of a gram of pollutant emitted per kwh of delivered electricity based on data from the EPA egrid database for CO 2 and NO x and from the National Emission Inventory for CO, HC, and PM. 15,16 These emission factors are estimated based on a 63 percent contribution of fossil fuel power generation to the total energy mix, which is typical in North Carolina. Except for NO x, the grid-based emission rates were low compared to those from the tailpipe. The emissions from the grid associated with electricity used to recharge the PHEV occur at a different time and location than the tailpipe emissions. The tailpipe emissions 12

13 occurred simultaneously with the onroad driving, and thus were distributed over the routes driven during the daytime. The indirect emissions occurred starting in the evening after the field measurements were collected, and thus took place overnight from those power plants whose marginal output was adjusted to compensate for the load on the grid. In the case of CO 2, CO, and HC, there is a net reduction in total emissions for the charge depleting case compared to the charge sustaining cases. For NO x, there may not have been a net reduction in emissions, but the change in timing of emissions might have some environmental benefits. For example, ozone formation in the troposphere may be more sensitive to NO x emitted during the day than that emitted overnight. Fuel Use and Electricity Consumption Versus Vehicle Specific Power An advantage of collecting real-world data for PHEVs using PEMS is the capability to quantify the effect of micro-scale events during a driving cycle, such as short periods of high speed or acceleration, on the localized emissions with respect to a transportation network. The development of a such a capability relies on the ability to predict energy consumption and emission rates as a function of the power demand on the vehicle. Here, the relationship between fuel and electricity use versus VSP is explored as the initial basis for development of a microscale modeling framework for PHEVs. In Figure 4(a), gasoline fuel use is plotted versus VSP, with mean values and confidence intervals shown for each range of 1.0 kw/ton of VSP. As expected, fuel use is not sensitive to VSP for negative values of VSP, which typically represent deceleration or can represent situations in which a vehicle is traveling at constant speed on a negative road grade. For positive VSP, fuel use increases. During charge depleting mode, the rate of gasoline consumption at a given value of VSP, such as 10 kw/ton, is less than that during charge sustaining mode. This is expected, since power from the plug-in battery is used during charge depleting mode to provide supplemental motor power in order to complement the power output of the engine, and partially offset the amount of engine power needed to meet the total demand for power. Figure 4(b) indicates the total battery electricity consumption for the plug-in and traction batteries versus VSP. Negative values indicate recharging of the battery, and positive values indicate discharging. During charge depleting mode, on average the battery is discharged under positive VSP, and recharged during episodes of negative VSP. This is expected, since the need for supplemental power from the electric motors is greatest during positive VSP, and increases with positive VSP. Conversely, either through regenerative braking or use of ICE output, recharging occurs during negative VSP. During charge sustaining mode, as shown in Figure 4(c), the battery is recharged during episodes of negative VSP. However, during positive VSP, the results here are not conclusive. On average, there is no net charging or recharging. However, there is significant variability in the current flow during positive VSP, with some cases of discharging and some cases of recharging. For example, discharging may occur during positive VSP and positive acceleration, whereas recharging might occur during positive VSP while the vehicle is cruising or decelerating. 13

14 CONCLUSIONS A methodology has been developed and demonstrated for field measurement of a PHEV that is based on recording or measuring both EOVs and IOVs. These data are used to characterize the consumption of gasoline fuel and the discharge or recharge of the vehicle batteries on a secondby-second basis. Furthermore, the tailpipe emission rates are estimated on a second-by-second basis. For the specific vehicle tested, which was a retrofitted Toyota Prius, a clear distinction was quantified between charge depleting and charge sustaining modes. The amount of gridbased electricity stored in the plug-in battery that was consumed could be estimated on a secondby-second basis during vehicle operation. Thus, it is possible to characterize the timing and location of vehicle tailpipe emissions on the road network, as well as the indirect emissions that take place from the grid during the time period in which the vehicle is recharged. Based on the quality assurance results, more work is needed to refine the QA procedures to be specifically applicable to a HEV and PHEV. In particular, the frequent engine starts and shutdowns create data quality problems that are not observed when conducting measurements of conventional LDGVs, such as large transients in the volume percent of CO 2 in the exhaust and dilution effects from infiltration of ambient air into the tailpipe that reduces the precision of estimated emission factors. The frequency and duration of engine use is sensitive to vehicle operations and to the state-ofcharge of the vehicle batteries. Driving cycles that require sustained high power demand have fewer engine starts but longer duration of engine use compared to cycles that might involve lower speeds but more deceleration and acceleration associated with stops and traffic delays. For the latter case, the rate of engine starts per mile is higher but the portion of distance over which the engine was used are smaller. This implies that PHEV emissions will not occur uniformly over the transportation network. The episodic nature of these emissions motivates the need for a micro-scale model that enables prediction of locations at which PHEV emissions will tend to occur, as well as locations or conditions under which such tailpipe emissions may be low or zero. The use of two sources of energy for a vehicle raise questions regarding how to report or characterize fuel economy. Here, we suggest that energy economy should be estimated taking into account the gasoline equivalent of the energy required to generate the electricity consumed by the vehicle. On this basis, the energy economy was found to be similar for CD and CS modes, even though the gasoline fuel economy during CD was approximately 50 percent higher than during CS mode. 14

15 Figure 4. Engine Fuel Use and Battery Electricity Consumption Rates vs. Vehicle Specific Power: Positive battery electricity use = discharge, Negative battery electricity use = recharge Vehicle Fuel Use (g/s) Charge Depleting Charge Sustaining Vehicle Specific Power (kw/ton) (a) Vehicle Fuel Use for Charge Depleting and Charge Sustaining Modes 4 Battery Electricity Consumption (Wh/s) y = x R 2 = y = x R 2 = Vehicle Specific Power (kw/ton) 2 (b) Battery Electricity Use for Charge Depleting Mode Battery Electricity Consumption (Wh/s) y = x y = 0.048x R 2 = R 2 = Vehicle Specific Power (kw/ton) (c) Battery Electricity Use for Charge Sustaining Mode 15

16 The direct tailpipe emissions varied depending on characteristics of the driving cycle, such as the average speed and the number of stops or delays. The indirect emissions were generally much lower than the tailpipe emissions for CO and HC. The indirect emissions of CO 2 were of similar magnitude but significantly lower than those from the tailpipe. The NO x emissions were comparable. However, the indirect emissions may typically occur overnight at predominantely rural sites where power plants are located, in contrast to the direct emissions that occur during vehicle operation on the transportation network. Given the differences in timing and location, the indirect emissions will have a different impact on local air quality. The ability to collect both EOV and IOV data provides insights regarding how vehicle energy use responds to VSP during CD and CD modes. The preliminary work shown here illustrates how such data can be used to describe the variation in gasoline use and to attribute the amount of electricity consumed from the grid to specific episodes that occur during real-world driving. Work is ongoing to further develop a micro-scale modeling approach for PHEVs based on the real-world data collected using the methodology described here. There are several limitations to this study that motivate additional work. One is that measurements were made only for one PHEV. There is inter-vehicle variability in energy use and emissions for the same year, make, and model of vehicle. Furthermore, there are differences in design among HEVs that are likely to carry over to PHEVs as the latter become commercially available as production vehicles in the near future. Thus, there is a need to expand the work here to include other HEVs and PHEVs where possible. A second limitation is the use of only one driver who was not experienced in operating a PHEV. Future work should involve comparison of multiple drivers. A third limitation is that only one day of measurements were conducted, limiting the observability of the CD mode. Since CD can occur only in the first half hour or so of operation of the PHEV that was tested, multiple days of operation are required in order to more fully characterize this mode. Furthermore, this mode should be measured on more than one route. Hence, future work will involve a more systematic round-robin of tests in which one PHEV will be operated by each of several drivers with multiple days of testing and day-to-day variation in the sequence of routes. ACKNOWLEDGMENTS This work has been supported in part by National Science Foundation grant no. CBET , for which Dr. Frey is the Principal Investigator. Advanced Energy, Inc. provided access to and use of their 2005 Toyota Prius with a retrofitted A123 Hymotion plug-in capability, and supplied the Kviser Memorator that was used for datalogging the vehicle ECU. The authors are solely responsible for the content of this paper. 16

17 REFERENCES 1. Energy Information Administration. Annual Energy Outlook 2008 With Projections to 2030; Report No. DOE/EIA-0383(2008); Energy Information Administration, U.S. Department of Energy: Washington, DC, Environmental Protection Agency Average annual emissions, all criteria pollutants, Current Emissions Trends Summaries from the National Emission Inventory; available on U.S. Environmental Protection Agency Web site, (accessed on August 21, 2006). 3. Markel, T. and Simpson, A. Cost-Benefit Analysis of Plug-In Hybrid Electric Vehicle Technology; Report No. NREL/JA ; National Renewable Energy Laboratory: Golden, CO, Gonder, J. and Simpson, A. Measuring and Reporting Fuel Economy of Plug-In Hybrid Electric Vehicles; Report No. NREL/JA ; National Renewable Energy Laboratory: Golden, CO, CATI. OEM-2100 Montana System Operation Manual; Clean Air Technologies International, Inc.: Buffalo, NY, November, Vojtisek-Lom, M. and Allsop, J.E. Development of Heavy-Duty Diesel Portable, On- Board Mass Exhaust Emissions Monitoring System with NO x, CO 2, and Qualitative PM Capabilities; ; Society of Automotive Engineers: Warrenton, PA, Andros Inc. (2003), Concentrations Measurement and Span Calibration Using n-hexane and Propane in the ANDROS 6602/6800 Automotive Exhaust Gas Analyzer, accessed on January 10, Battelle. Environmental Technology Verification Report: Clean Air Technologies International, Inc. REMOTE On-Board Emissions Monitor; U.S. Environmental Protection Agency: Columbus, OH, June, Jiménez-Palacios, J.L. Ph.D. Thesis, Massachusetts Institute of Technology, Cambridge, MA, Zhang, K.; Frey, H.C. Road Grade Estimation for On-Road Vehicle Emissions Modeling Using Light Detection and Ranging Data. J. Air & Waste Manage. Assoc. 2006, 56(6), Durbin, T.D.; Collins, J.R.; Norbeck, J.M.; Smith, M.R. Effects of Biodiesel, Biodiesel Blends, and a Synthetic Diesel on Emissions from Light Heavy-Duty Diesel Vehicles. Environ. Sci. Technol. 2000, 34(3),

18 12. Environmental Protection Agency. Inventory of U.S. Greenhouse Gas Emissions and Sinks: Fast Facts Conversion Factors to Energy Units (Heat Equivalents) Heat Contents and Carbon Content Coefficients of Various Fuel Types; EPA430-R ; U.S. Environmental Protection Agency: Washington, DC, Frey, H.C., Unal, A., Chen, J., Li, S., Xuan, C. Methodology for Developing Modal Emission Rates for EPA s Multi-scale Motor Vehicle & Equipment Emission System; EPA420-R-02-02; U.S. Environmental Protection Agency: Ann Arbor, MI, Koupal, J.; Landman, L.; Nam, E.; Warila, J.; Scarbro, C.; Glover, E.; Giannelli, R. MOVES2004 Energy and Emissions Inputs; EPA 420-P ; Office of Transportation and Air Quality, U.S. Environmental Protection Agency: Ann Arbor, MI, Environmental Protection Agency. egrid2006 version 2.1; available on U.S. Environmental Protection Agency Web site, (accessed on September 14, 2008). 16. Environmental Protection Agency. National Emissions Inventory (NEI) Air Pollutant Emissions Trends Data; available on U.S. Environmental Protection Agency Web site, (accessed on September 14, 2008). 18