Fuel Consumption Potential of Different Plug-in Hybrid Vehicle Architectures in the European and American Contexts

Plugin Hybrid Electric Vehicles (PHEVs) have demonstrated the potential to provide significant fuel displacement across a wide range of driving cycles. Companies and research organizations are involved in numerous research activities related to PHEVs. One of the current unknown is the impact of driving conditions and standard test procedure on the true benefits of PHEVs from a worldwide perspective. To address this issue, Argonne National Laboratory (ANL) and IFP Energies nouvelles (IFPEN) have partnered under the IEA Annex XV task to evaluate the market specificities between Europe and U.S. Four different PHEV powertrain configurations with four All Electric Range will be analyzed under different standards (i.e., NEDC, UDDS, HWFET) and real world drive cycles (i.e. ARTEMIS…). The impact of different driving behavior for Europe and the US market will be analyzed through component sizing, fuel consumption benefits as well as Green House Gases (GHGs) considering the electricity production mix. The study will provide insight on how PHEVs can be designed to support worldwide market introduction of a limited number of vehicle options to maximize market penetration.


Introduction
National authorities all over the world have defined more stringent CO2 standards to decrease the overall fuel consumption of light duty vehicles. Figure 1 compares the normalized CO2 emissions from different countries up to 2020. In response to these constraints, car manufacturers and suppliers have developed numerous technologies to enhance vehicle drivetrain efficiency or shift a part of the energy consumption from fossil fuels to other primary energies (i.e., electricity, hydrogen…). Among the existing panel of possible solutions, hybrid electric vehicles (HEVs) and plug-in hybrid electric vehicles (PHEVs) constitute one of the most promising solution with dozens of HEV models already in production and some recently unveiled for PHEVs. Understanding the real potential of such drivetrains is a complex task as it depends on a high number of parameters, including: • Vehicle hybrid drivetrain architecture and functionalities (all electric range, plug in capabilities...); • Vehicle class (compact, sedan, SUV, 4WD...) and dynamic performances; • Vehicle usage (urban, extra urban, motorway, combined, type of standard procedures); • For PHEVs, the electricity mix considered for the battery charge from the mains; • Type of drivetrain components implemented • Type of vehicle energy management implemented.
In order to clarify the potential of HEVs and PHEVs both in Europe and in the US, ANL and IFPEN have collaborated to develop a specific methodology to precisely establish the fuel consumption and GHG emission potential of different HEV and PHEVs. For this purpose, the same vehicle body in white with similar drivetrain components technologies have been considered. The vehicles have been simulated through different American and European driving patterns. For the case of PHEVs, standard procedures such as US J1711 and EEC Regulation 101 have been considered.
This paper presents and discusses the results obtained for a large number of configurations simulated in both Laboratories.

Tools
Since the study was performed under a collaboration of two different laboratories, two simulation tools were used.
IFPEN used an in-house simulator developed under the LMS.IMAGINE.Lab AMESim® platform with components available in the IFP-Drive library [2]. This simulator is working under co-simulation with Simulink® for control algorithm. The models used for this study are steady-state efficiency depending on operating points, should it be for internal combustion engines, electric motors or transmissions and power electronics. The control is based on an online Equivalent Consumption Minimization Strategy (ECMS) principle. Although this approach leads to a higher computation time, it reduces the calibration process and fitted this study addressing several vehicles.
ANL used in-house developed software Autonomie, which is a MATLAB-based software environment and framework for automotive control-system design, simulation, and analysis [3]. The tool is designed for rapid and easy integration of models with varying levels of detail (low to high fidelity) and abstraction (from subsystems to systems and entire architectures), as well as processes (calibration, validation, etc. In previous studies [7], ANL already compared the instantaneous optimal control algorithm with the reference rules-based control to show the effect of different control strategies, and provided similar fuel economy results while properly managing the battery SOC. Thus we had to make sure the models gave consistent results. Figure 3 compare the behaviour of AMESim and Autonomie on the NEDC cycle, for the parallel HEV vehicle. AMESim generally tends to use higher gears, as it's based on fuel consumption minimization, whereas Autonomie integrates driveability constraints on the control calibration. Table 1 shows that the overall fuel consumption results are similar between both tools.

Component data Internal Combustion Engine (ICE)
IFPEN and ANL used in-house measured efficiency of an 1800cc spark ignition engine developed at IFPEN, equipped with VVT at intake and exhaust camshafts, direct-injection and turbocharger. The results from the test beds have been used to generate maximum Brake Mean Effective Pressure (BMEP) and Brake Specific Fuel Consumption (BSFC) of both the turbocharged and the naturally aspirated versions of the engine.
At this step of the study, the BSFC associated to an engine technology is considered as depending on the engine speed, BMEP, but not on its displacement. At the same time, the maximum BMEP does not depend on the engine's displacement. Finally, ANL provided an estimation of the engine's weight, depending on its maximum torque and maximum speed.

Electric Machine
An IFPEN in-house software (EMTool) was used to develop the efficiency maps of the different electric machines (EM). EMTool offers the capability to size and to characterize EM from basic requirements (maximum power and torque, maximum motor speed, input voltage). This tool is based on analytical models allowing to design an electric motor that meets the required specifications [8],[9], [10]. Electromagnetic parameters are then calculated from the geometry and are associated to quasi-static control strategy to evaluate electric motor performances and efficiency [10], [11]. A complete efficiency map can be then determined and integrated in the vehicle simulator. To validate the relevance of the results given by the EMTool, a comparison with an experimental efficiency map of the Toyota Prius II electric motor [12] is presented from Figure 4 to Figure 6 with the repartition of the error between simulation and experimental results on the whole operating conditions. Efficiency maps have a mean difference of 5% and a maximum of 17% in highly saturated regimes (saturation phenomena are not taken into account in EMTool for the moment). The EMTool is also able to calculate the mass and the volume of the different parts of the electric machine.

Battery
Batteries were the only energy storage systems used in this study, on the assumption that ultracapacitors alone could not provide sufficient available energy for the electric drive applications considered. We also considered that coupling ultra-capacitors with batteries would be costprohibitive and that Li-ion battery life would be significantly improved in the short term, making the combination ineffective.
The batteries used in the study as the reference have been provided by Argonne, Idaho National Laboratory, and major battery suppliers. A scaling algorithm developed by Argonne's battery experts is used for the high-energy cases [13]. The battery electrode materials are LiMn2O4 and Li4Ti5O12, which provide a cell area-specific impedance of about 40% of that of the commonly available lithium-ion batteries.

Drivetrain Architectures Considered
For this paper, several powertrain architectures have been compared, depicting the actual trend in conventional and hybrid vehicles: • Conventional 5 speed vehicle, with both automatic and manual gearbox • Pre-transmission parallel HEV and PHEV's.
Parallel PHEV's were also evaluated in a "mild-hybrid" version, with lower battery and electric motor power. This version does not respect the all-electric performance criteria but aims at limiting costs. The selection of the single-mode power-split hybrid and the parallel hybrid was based on the current sales volume of both Toyota and Ford hybrid vehicles.
The series engine configuration selected is the simplest one and has been used by many companies. For this option, the electric-range extended vehicles (E-REV) powertrain used in the GM Volt [14] offers significant advantages, especially during high-vehicle-speed operations.
Since the Volt uses a series-output split powertrain architecture, which provides benefits over the series architecture that typically has been considered for use in EREVs, it has been compared in this study.
Both simulation tools were used to simulate the conventional vehicles to ensure that the baseline vehicles provided similar fuel consumption. Autonomie was used to simulate the power split configurations (both input split and power split) as well as the battery electric vehicles. AMESim was used to simulate the pre-transmission and the series configurations.

Component Sizing
To properly evaluate the benefits of different powertrain configurations, one needs to ensure that their Vehicle Technical Specifications are comparable. All the vehicles have been sized to meet the same requirements: • Initial vehicle movement (IVM) to 100kph in 9 sec +/−0.1 sec with ICE + electric power, • Maximum grade of 5% at 110kph at gross vehicle weight (GVW) with ICE power only, • Maximum vehicle speed >150kph with ICE power only, and • All electric Range (AER) on UDDS (for US) or Artemis Urban (for Europe) The only requirement that is different from one architecture to the other is the acceleration capability in all-electric mode : • Energy recovery on urban cycles for HEV's and mild-hybrid parallels.
• Urban capability (based on UDDS in the US and Artemis Urban for Europe) for parallel and power-split PHEV's. • Highway capability (based on US06 in the US and Artemis Highway for Europe) for output-split. • Maximum performance available in allelectric mode for the series.
As detailed previously, the component's characteristics are determined by the constraints.
The main vehicle characteristics used in this study are summarized in Table 2.
Several automated sizing algorithms were developed to provide a fair comparison between technologies. These algorithms are specific to the powertrain (i.e., conventional, power-split, series-split, electric) and the application (i.e., HEV, PHEV).

Drive Cycles and Evaluation Procedures
This study aimed at evaluating results on US and EEC standard procedures, as well as real world driving cycles. The US standard test procedure for plug-in electric vehicle can be found in [15] while the EEC standard test procedure is described in [16]. For this study, we considered that all the European vehicles had a "Zero-emission" functionality: as long as the energy storage has enough energy, the user can decide to enter this mode and disable the ICE start. This functionality can be useful to limit emissions in city centres, and has also an impact on the evaluation of energy consumption on the EEC standard procedure.
Three daily missions were built in this study, using the Artemis cycles [17]. The objective is to represent daily trips outer to inner city and inner to outer city, for different distances:  Figure 8 shows the main component sizes for the PHEV50 when sized on the UDDS US drive cycle. One notices that the pre-transmission requires the smaller combined power of all electrified vehicles. The ability to follow a specific drive trace in electric only mode leads the output split configuration to have a large electric machine, similarly to the series. The series configuration, where the wheels are only powered from the electrical energy, shows the highest total power, about twice as high as for the pre-transmission parallel.  Figure 9 and Figure 10 show the fuel consumption ratio, in charge-sustaining mode, compared to the reference conventional vehicle for both European and US drive cycles. UL1 cycle has also been simulated [19]. It's a speed profile representing city center with traffic jam (mean speed is 3,8km/h).

Charge-Sustaining Fuel Consumption Results
As one notices, most of the electric drive powertrain considered lead to fuel consumption reduction. The exception is for the highway drive cycles (Artemis Highway and HWFET) where the pure series configuration shows a higher fuel consumption than for the reference conventional vehicle, due to high powertrain losses. The opposite tendency can be observed in the UL1 cycle, where the series architecture offers a better efficiency. As expected, the drive cycles with the lowest average vehicle speed (Artemis Urban and UL1) leads to the greatest fuel savings with a ratio lower than 0.4 (resp. 0.2). Most of the powertrain configurations however achieve similar fuel consumption ratio. Higher energy storage systems show higher fuel consumption in charge-sustaining mode, due to higher vehicle weight.  Figure 11 shows the fuel and electrical consumption for the different electric drive vehicles considered. Since the standard procedures do not provide a single energy value, both consumptions have been plotted on different axis. The powertrain configurations the closest to the origin show the highest overall efficiencies.

Results on US and EEC standard procedures
To properly analyse the results, special attention has been focused on obtaining similar charge depleting distance across powertrains with the same AER value (i.e. similar energy management strategies during charge depleting were used across powertrains). The input split HEV and PHEV configurations consistently demonstrate the highest powertrain efficiencies regardless of the standard driving cycles considered. The output split and the pretransmission configurations provide close results. The series configuration, however, demonstrate significantly higher losses than any other configuration.
In addition, for the same electrical consumption, the fuel consumptions achieved on the US drive cycle are consistently higher than the ones achieved for the European drive cycle. Detailed analysis of the results proved that this difference mainly comes from several factors including: • Fuel consumption is adjusted on the US cycles, while it is not on the EEC ones. It is the main reason why, for a same level of electricity consumption, fuel consumption is higher on the US norm. As Figure 12 shows, comparing EEC results to US results without using the adjustment factor results in a similar fuel/electricity consumption tradeoff, even if the driving patterns are different. • The utility factor usage in the US versus the EEC norm. The former aims at being an image of the average usage of a car overall the population in the country, while the second one represents a usage where each customer would buy a vehicle with an AER slightly lower (25km) than his daily trip. It explains why the electricity/fuel consumption trade-off goes to more electricity on the EEC norm with higher electric energy content.
This difference should be taken into account when comparing worldwide regulations (cf. Figure 1).

Results on Actual Use Daily Missions
As discussed earlier, several daily missions have been developed to represent the fuel displacement potential of each powertrain configurations under different driving conditions. In that case, no utility factor or weighting is applied. Figure 12, Figure 13 and Figure 14 respectively show the consumption for different mission profiles. One notice that the consumption values significantly vary based on the type of trip (city in mission 1 to extra urban in mission 2 and highway in mission3) compared to the standard drive cycle. For example, the input split HEV achieves 4.75 l/100km on the US standard and 3.34 l/100km on the NEDC while it varies from 3.6 to 5 l/100km.
In the majority of cases, for an equivalent amount of electrical energy, the PHEV configurations also lead to higher fuel consumption when simulating real world drive cycles.    In addition, the NEDC drive cycle provides lower fuel consumption values than the US standard, which should be carefully taken into account when comparing fuel consumption standards worldwide.

Consumption Results on Mission
When comparing powertrain configurations, the input split offers the highest efficiency and the series the lowest, which is consistent with the current technologies in the market. Parallel and output split configurations, which are being introduced in the market, also offer significant fuel displacement.
Future studies will include the impact of component optimum sizes and technology benefits focusing on the new worldwide drive cycle compared to current standard and different mission profiles.

Acknowledgments
This study was partially supported by the DOE Vehicle Technologies Office under the direction of David Anderson and Lee Slezak. The submitted report has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (Argonne). Argonne, a DOE Office of Science laboratory, is operated under Contract No. DE-AC02-06CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up, nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the Government.
This study was partially supported by the French ADEME Transport and Mobility Department under Contract No. 10 66 C0120.