Environmental Impact Assessment and Classiﬁcation of 48 V Plug-in Hybrids with Real-Driving Use Case Simulations

: Plug-in hybrid electric vehicles (PHEVs) are commonly operated with high-voltage (HV) components due to their higher power availability compared to 48 V-systems. On the contrary, HV-powertrain components are more expensive and require additional safety measures. Additionally, the HV system can only be repaired and maintained with special equipment and protective gear, which is not available in all workshops. PHEVs based on a 48 V-system level can offer a reasonable compromise between the greenhouse gas (GHG) emission-saving potential and cost-effectiveness in small- and medium-sized electriﬁed vehicles. In our study, the lifecycle emissions of the proposed 48 V PHEV system were compared to a conventional vehicle, 48 V HEV, and HV PHEV for individual driving use cases. To ensure a holistic evaluation, the analysis was based on measured real-driving cycles including Global Position System (GPS) map-matched slope proﬁles for a parallel hybrid. Optimal PHEV battery capacities were derived for the individual driving use cases. The analysis was based on lifecycle emissions for 2020 and 2030 in Europe. The impact analysis revealed that 48 V PHEVs can signiﬁcantly reduce GHG emissions compared to vehicles with no charging opportunity for all use cases. Furthermore, the ﬁndings were veriﬁed for two vehicle segments and two energy mix scenarios. The 48 V PHEVs can therefore complement existing powertrain portfolios and contribute to reaching future GHG emission targets.


Introduction
One quarter of the greenhouse gas (GHG) emissions in Europe originates from the transport sector [1]. One of today's main targets for the automotive industry is therefore a reduction of the environmental impact of vehicles. The electrification of the powertrain is a widely accepted method to reduce the amount of GHG emissions and can be scaled on different levels [2]. It was shown that 48 V hybrid electric vehicles (HEVs) offer a good compromise between additional investment costs and fuel-saving potentials [3]. However, to substitute more fossil fuel with electric energy, high voltage (HV) systems are necessary, which can recuperate more energy due to their lower current [4,5]. Plug-in hybrid electric vehicles (PHEVs) and battery electric vehicles (BEVs) utilize an external charge plug to use electrical grid energy and increase the electric driving share. Fuel cell vehicles are operated similarly to HEVs or PHEVs and allow the substitution of fossil fuel with hydrogen, which can reduce the CO 2 emissions further if the hydrogen is based on renewable energy sources. different optimal degrees of electrification. In cases of states with a carbon-intensive electricity mix, HEVs even showed lower life cycle GHG emissions compared to PHEVs and BEVs. However, on a national level, the largest GHG reduction potential was achieved with a PHEV. Despite the focus on the United States and high battery production emissions in the past, the analysis demonstrates the importance of renewable energies for the mobility sector. The authors in [26] underlined the LCA-saving potential of PHEVs, especially in combination with renewable fuels, which can be an additional measure to reduce greenhouse gases.
Extending a 48 V hybrid to a PHEV with a larger battery and an external charge plug can be a cost-efficient fuel-saving technology, which shall be further investigated in this article on an LCA basis for the first time. To this aim, our study also includes the emissions during the production phase of the vehicle and the well-to-tank emissions. Unlike previous studies, we conduct an LCA of a 48 V PHEV and compare it to results for a conventional vehicle, 48 V HEV, and HV PHEV to classify the proposed system against broadly existing technologies. Measured year-round speed and slope profiles from different driving use cases ensure the applicability towards real-world driving. Furthermore, the outcome of two different EMSs is shown to evaluate the influence of possible strategy improvements. The results are evaluated for a 2020 and 2030 energy mix scenario and include a comparison to the emission targets of the European Union (EU). Finally, based on the simulation results, policy implications for PHEVs are discussed.
The paper is organized as follows: The overall system model is presented in Section 2. This includes the derivation of the selected driving use cases, the powertrain model, the investigated EMSs of the vehicles, and the underlying assumptions for the LCA. The results of the LCA are shown in Section 3, followed by a discussion in Section 4. Finally, a conclusion is given in Section 5.

Research Method
The working principle of the conducted LCA is shown in Figure 1. First, vehicle speed and Global Position System (GPS) measurements were collected for 104 different vehicles. Four typical driving use cases were selected out of this database afterwards. The year-round trips served as an input for the powertrain simulation layer. The results were calculated for two different vehicle segments, two EMSs, and systems with different degrees of hybridization. Finally, the operation emissions were added to the GHG emissions from the production and recycling of the vehicle and battery. The input and powertrain simulation layers are explained in the following subsections. electrification. In cases of states with a carbon-intensive electricity mix, HEVs even showed lower life cycle GHG emissions compared to PHEVs and BEVs. However, on a national level, the largest GHG reduction potential was achieved with a PHEV. Despite the focus on the United States and high battery production emissions in the past, the analysis demonstrates the importance of renewable energies for the mobility sector. The authors in [26] underlined the LCA-saving potential of PHEVs, especially in combination with renewable fuels, which can be an additional measure to reduce greenhouse gases. Extending a 48 V hybrid to a PHEV with a larger battery and an external charge plug can be a cost-efficient fuel-saving technology, which shall be further investigated in this article on an LCA basis for the first time. To this aim, our study also includes the emissions during the production phase of the vehicle and the well-to-tank emissions. Unlike previous studies, we conduct an LCA of a 48 V PHEV and compare it to results for a conventional vehicle, 48 V HEV, and HV PHEV to classify the proposed system against broadly existing technologies. Measured year-round speed and slope profiles from different driving use cases ensure the applicability towards real-world driving. Furthermore, the outcome of two different EMSs is shown to evaluate the influence of possible strategy improvements. The results are evaluated for a 2020 and 2030 energy mix scenario and include a comparison to the emission targets of the European Union (EU). Finally, based on the simulation results, policy implications for PHEVs are discussed.
The paper is organized as follows: The overall system model is presented in Section 2. This includes the derivation of the selected driving use cases, the powertrain model, the investigated EMSs of the vehicles, and the underlying assumptions for the LCA. The results of the LCA are shown in Section 3, followed by a discussion in Section 4. Finally, a conclusion is given in Section 5.

Research Method
The working principle of the conducted LCA is shown in Figure 1. First, vehicle speed and Global Position System (GPS) measurements were collected for 104 different vehicles. Four typical driving use cases were selected out of this database afterwards. The year-round trips served as an input for the powertrain simulation layer. The results were calculated for two different vehicle segments, two EMSs, and systems with different degrees of hybridization. Finally, the operation emissions were added to the GHG emissions from the production and recycling of the vehicle and battery. The input and powertrain simulation layers are explained in the following subsections.

Real-driving measurements
Year-round speed and GPS map-matched slope profiles

Life-cycle emissions
Summation of production, recycling and operation emissions

Life-cycle assumptions
• Vehicle/Battery production and recycling emissions • Well-to-wheel emissions for energy carriers

Fleet Analysis and Use Case Definition
The battery sizing problem for PHEVs is largely dependent on the use case [21]. It is important to consider the different trip distance shares during the usage of the vehicle for a realistic evaluation. Therefore, this analysis does not focus on individual cycles, and instead uses continuous trip data information.
The database used in this study consists of 104 vehicles operated in Germany for which we extracted the data until 30.03.2020 to eliminate influences of the pandemic. Four vehicle trip datasets were extracted from this subset, based on the following criteria: • Short-distance driver, high share of trips less than 10 km; • Average driver, annual mileage is close to German passenger car average of 13.602 km (2019) [27]; • Long-distance driver, high share of trips longer than 100 km; • Commuter, prominent peak at a certain trip distance.
The key parameters of the selected drivers are shown in Table 1. The recorded timespan of more than 285 days ensures a robust extrapolation towards the annual distance. Individual routes on one driving day are merged for the simulation and defined as a trip in the following. Furthermore, the driving style can influence the energy consumption of a vehicle, which was shown in [28]. To ensure the selection of realistic data, RDE testing criteria were used to evaluate the driving style [29]. During RDE testing, two boundary conditions need to be fulfilled, which are shown in for the four selected vehicles in comparison to the complete database. The 95 percentile of the product of vehicle speed and positive acceleration in (a) is used to identify high trip dynamics and is an upper limit, calculated as where v is the vehicle speed and a pos is the positive acceleration. The relative positive acceleration (RPA) in (b) is used as a lower boundary to filter out cycles which are too smooth, and is described as where d i is the distance of a daily trip. The comparison of the selected drivers to the complete fleet data in Figure 2 shows that the drivers have an averaged driving style and stay within the limits for RDE testing for most daily trips, which ensures that no extremely aggressive or passive drivers are selected. The energy consumption of PHEVs strongly depends on the share of electric-driven mileage compared to the total distances. The utility factor (UF) is a key parameter to determine this share and is used in many regulations worldwide [30]. The UF is dependent on the charge depleting (CD) range of a PHEV, and is calculated as where R CD is the CD range and n is the total number of trips. Ideally, each driving day is assumed to start with a charged battery. The UF curves of the four drivers are shown in Figure 3 in comparison to the Worldwide Harmonized Light Vehicles Test Procedure (WLTP) curve for Europe [29]. Except for the short-distance driver, the resulting curves are underneath the European WLTP UF curve. This is in accordance with other studies, where the real-life UF is lower, especially for company cars. The annual mileage and UF curves for the average driver and commuter fit to typical privately owned cars, whereas the long-distance driver curve matches a company car profile. The energy consumption of PHEVs strongly depends on the share of electric-driven mileage compared to the total distances. The utility factor (UF) is a key parameter to determine this share and is used in many regulations worldwide [30]. The UF is dependent on the charge depleting (CD) range of a PHEV, and is calculated as where is the CD range and is the total number of trips. Ideally, each driving day is assumed to start with a charged battery. The UF curves of the four drivers are shown in Figure 3 in comparison to the Worldwide Harmonized Light Vehicles Test Procedure (WLTP) curve for Europe [29]. Except for the short-distance driver, the resulting curves are underneath the European WLTP UF curve. This is in accordance with other studies, where the real-life UF is lower, especially for company cars. The annual mileage and UF curves for the average driver and commuter fit to typical privately owned cars, whereas the long-distance driver curve matches a company car profile.   The energy consumption of PHEVs strongly depends on the share of electric-driven mileage compared to the total distances. The utility factor (UF) is a key parameter to determine this share and is used in many regulations worldwide [30]. The UF is dependent on the charge depleting (CD) range of a PHEV, and is calculated as where is the CD range and is the total number of trips. Ideally, each driving day is assumed to start with a charged battery. The UF curves of the four drivers are shown in Figure 3 in comparison to the Worldwide Harmonized Light Vehicles Test Procedure (WLTP) curve for Europe [29]. Except for the short-distance driver, the resulting curves are underneath the European WLTP UF curve. This is in accordance with other studies, where the real-life UF is lower, especially for company cars. The annual mileage and UF curves for the average driver and commuter fit to typical privately owned cars, whereas the long-distance driver curve matches a company car profile.

Powertrain Simulation
The investigated systems in this study are shown in Figure 4. The conventional vehicle has no electrical driveline components, so the driving energy is obtained from the liquid fuel tank only. The HEV and PHEV systems are parallel hybrids, with an electric machine (EM) located between the ICE and transmission. The ICE can propel the vehicle together with the EM in a parallel hybrid mode, but can also be decoupled and turned off for pure electric driving. Both systems can harvest kinetic energy during regenerative braking and load shifting of the ICE. In contrast to the HEV, the PHEV systems have an external charge plug to charge the battery with additional electrical grid energy. cle has no electrical driveline components, so the driving energy is obtained from the liquid fuel tank only. The HEV and PHEV systems are parallel hybrids, with an electric machine (EM) located between the ICE and transmission. The ICE can propel the vehicle together with the EM in a parallel hybrid mode, but can also be decoupled and turned off for pure electric driving. Both systems can harvest kinetic energy during regenerative braking and load shifting of the ICE. In contrast to the HEV, the PHEV systems have an external charge plug to charge the battery with additional electrical grid energy. The main parameters of the two examined vehicles are shown in Table 2. In this study, the results for a subcompact class vehicle are compared to a compact class vehicle. The sales price of a subcompact class vehicle is typically lower, which makes the introduction of HV PHEVs in that vehicle segment difficult; 48 V PHEVs may reduce the cost barrier. Furthermore, the lower vehicle weight leads to lower required power demands. The occurrences of power demands exceeding the limits of 48 V systems are consequently lower, which allows larger driving shares in pure electric mode. Nevertheless, we also show the results for a compact class vehicle, because it addresses a larger market share [31].
The powertrain equations are solved in a kinematic backward simulation in MATLAB. The mechanical power demand on wheel level is calculated as where ℎ is the vehicle mass, is the equivalent rotational inertia, is the gravitational acceleration, is the tire-rolling resistance coefficient, the climbing angle, the air resistance coefficient, the frontal area, and the air density. The vehicle mass ℎ is separated into three parts to address the component differences between the three hybrid systems and scale with increasing battery capacity: where 0 is the vehicle base weight, 0, is the electric powertrain weight without the battery, is the battery energy, and is the battery gravimetric energy density.  The main parameters of the two examined vehicles are shown in Table 2. In this study, the results for a subcompact class vehicle are compared to a compact class vehicle. The sales price of a subcompact class vehicle is typically lower, which makes the introduction of HV PHEVs in that vehicle segment difficult; 48 V PHEVs may reduce the cost barrier. Furthermore, the lower vehicle weight leads to lower required power demands. The occurrences of power demands exceeding the limits of 48 V systems are consequently lower, which allows larger driving shares in pure electric mode. Nevertheless, we also show the results for a compact class vehicle, because it addresses a larger market share [31]. The powertrain equations are solved in a kinematic backward simulation in MATLAB. The mechanical power demand on wheel level is calculated as where m veh is the vehicle mass, Θ is the equivalent rotational inertia, g is the gravitational acceleration, f R is the tire-rolling resistance coefficient, α the climbing angle, c w the air resistance coefficient, A the frontal area, and ρ Air the air density. The vehicle mass m veh is separated into three parts to address the component differences between the three hybrid systems and scale with increasing battery capacity: where m 0 is the vehicle base weight, m 0,Edr is the electric powertrain weight without the battery, E Bat is the battery energy, and ρ Bat is the battery gravimetric energy density. The values for electric powertrain parameters are summarized in Table 3. Since 48 V PHEVs have a higher share of electric driving compared to HEVs, it is assumed that the weight increases due to enhanced cooling requirements and larger wire diameters. The weight increase towards the HV PHEV is based on the higher EM power. The energy densities for the 48 V HEV and HV PHEV are derived from the average of current pack models on the market. The power-to-energy ratio of 48 V PHEVs is in the range of typical BEV applications, which allows the usage of cells with a higher energy density. If the fill factor of a PHEV battery pack remains, a change from PHEV cells with 180 Wh/kg to a BEV cell with 250 Wh/kg leads to an energy pack density of 130 Wh/kg. The utilization of high energy cells is particularly interesting because the increasing amount of BEVs leads to scaling effects and therefore cost reductions in the battery cell production. The recuperation power P EM,Gen is higher than the propulsion power P EM,Mot for the electrified vehicles to consider the increasing power availability of the EM due to the voltage increase during recuperation. The conventional vehicle compensates the electrical system load with an alternator, which does not provide motoric power. The system load increases with a higher degree of hybridization due to the additional control units as well as higher thermal management requirements. As no heating and air condition model is utilized, the system load is chosen high enough to represent the required annual average power [32]. The propulsion power at the transmission input can be split up between the EM and the internal combustion engine (ICE) for the three hybrid types. During propulsion the power demands are defined as where P EM is the EM power, P ICE is the ICE power, and ψ split is the power split between the EM and ICE. The engine used for this study was a 90 kW turbocharged gasoline engine with direct injection. The efficiency map with the maximum torque and power curve is shown in Figure 5. Temperature effects were not considered in this study. The electric energy consumption for propulsion is calculated as where P Bat is the battery terminal power, P Pnt is the 12 V electrical system load, and η DC/DC is the converter efficiency between 12 V and 48 V or HV. Charging losses for the PHEV systems are considered with the charging efficiency η Chg and are included in the electric energy consumption values. where is the EM power, is the ICE power, and is the power split between the EM and ICE. The engine used for this study was a 90 kW turbocharged gasoline engine with direct injection. The efficiency map with the maximum torque and power curve is shown in Figure 5. Temperature effects were not considered in this study. The electric energy consumption for propulsion is calculated as where is the battery terminal power, is the 12 V electrical system load, and / is the converter efficiency between 12 V and 48 V or HV. Charging losses for the PHEV systems are considered with the charging efficiency ℎ and are included in the electric energy consumption values.

Battery Pack Modeling
In this study, the battery cells were modeled with measured internal resistance curves. The open-circuit voltage (OCV) and direct current (DC) resistance curves, extracted after a pulse length of five seconds, are shown in Figure 6. Note that the resistance values are normalized by the cell capacity. The high-power 18 Ah 48 V HEV cell showed the lowest internal resistance, whereas the 108 Ah 48 V PHEV cell is a high-energy cell for BEV applications with the highest resistance. The 42 Ah HV PHEV cell has balanced characteristics, so the power-to-energy ratio and internal resistance is located between the other curves. Only discrete battery pack capacities can be configured when the investigated nickel manganese cobalt oxide cells are connected in series or parallel. To comply with the voltage limits and stay in reasonable voltage ranges for the inverter, the number of serial cells was fixed to 12 cells for the 48 V systems and 96 for the HV system. The number of parallel cells varied during the optimization, and was chosen so that the discrete pack capacity matched the required one as closely as possible. For the remaining difference, the cell was scaled to various, continuous capacities where is the scaled battery cell capacity, is the scaling factor, and , is the measured battery capacity. The cell resistance is expressed as Figure 5. ICE efficiency map, maximum torque (black), and maximum power (blue).

Battery Pack Modeling
In this study, the battery cells were modeled with measured internal resistance curves. The open-circuit voltage (OCV) and direct current (DC) resistance curves, extracted after a pulse length of five seconds, are shown in Figure 6. Note that the resistance values are normalized by the cell capacity. The high-power 18 Ah 48 V HEV cell showed the lowest internal resistance, whereas the 108 Ah 48 V PHEV cell is a high-energy cell for BEV applications with the highest resistance. The 42 Ah HV PHEV cell has balanced characteristics, so the power-to-energy ratio and internal resistance is located between the other curves. Only discrete battery pack capacities can be configured when the investigated nickel manganese cobalt oxide cells are connected in series or parallel. To comply with the voltage limits and stay in reasonable voltage ranges for the inverter, the number of serial cells was fixed to 12 cells for the 48 V systems and 96 for the HV system. The number of parallel cells varied during the optimization, and was chosen so that the discrete pack capacity matched the required one as closely as possible. For the remaining difference, the cell was scaled to various, continuous capacities where C Cell is the scaled battery cell capacity, δ is the scaling factor, and C Cell,measured is the measured battery capacity. The cell resistance is expressed as where R Cell is the scaled battery cell resistance, and R i,5s is the normalized cell resistance. The pack resistance is then calculated by R Pack = R Cell n ser n par (12) where n ser is the number of serial cells, and n par is the number of parallel cells. where is the scaled battery cell resistance, and ,5 is the normalized cell resistance. The pack resistance is then calculated by = where is the number of serial cells, and is the number of parallel cells.

Energy Management Strategies
The optimal sizing problem is closely related to the EMSs. In literature, Dynamic Programming (DP) [33,34] and ECMS [35,36] are commonly used strategies for powertrain simulations. The former is based on Bellman's principle of optimality [37] and often used as a benchmark solution, as it indicates the optimal fuel consumption if the cycle is known beforehand. A MATLAB function is presented in [38]. The cost function can be written as = ∑ ( ( ), ( )) −1 =0 (13) where is the total fuel consumption, is the fuel consumption matrix, and is the number of steps in the driving cycle. The costs vary with the state vector ( ), the state of charge (SOC) in this case, and the control input ( ). The control variable determines the engine state and power split between the drivetrain components. ( ) and ( ) are discretized into discrete grids and evaluated during the backward calculation for each time step. The cost-to-go matrix for one timestep can be written as * ( ( )) = ( ) [ (( ( ), ( )) + +1 * ( ( + 1)) when the initial and final states are defined, all optimal control paths can be found by the algorithm with the lowest possible fuel consumption for this cycle. Implemented constraints for the driveline components ensure the operation within the physical limits. However, we recognized the influence of a nonoptimal but real-life implementable strategy and compared the DP results to an A-ECMS. The equivalent fuel consumption was calculated as

Energy Management Strategies
The optimal sizing problem is closely related to the EMSs. In literature, Dynamic Programming (DP) [33,34] and ECMS [35,36] are commonly used strategies for powertrain simulations. The former is based on Bellman's principle of optimality [37] and often used as a benchmark solution, as it indicates the optimal fuel consumption if the cycle is known beforehand. A MATLAB function is presented in [38]. The cost function can be written as where J is the total fuel consumption, L is the fuel consumption matrix, and N is the number of steps in the driving cycle. The costs vary with the state vector x(t), the state of charge (SOC) in this case, and the control input u(t). The control variable determines the engine state and power split between the drivetrain components. x(t) and u(t) are discretized into discrete grids and evaluated during the backward calculation for each time step. The cost-to-go matrix for one timestep can be written as when the initial and final states are defined, all optimal control paths can be found by the algorithm with the lowest possible fuel consumption for this cycle. Implemented constraints for the driveline components ensure the operation within the physical limits. However, we recognized the influence of a nonoptimal but real-life implementable strategy and compared the DP results to an A-ECMS. The equivalent fuel consumption was calculated as . m f ,eq ψ split , t = . m f uel ψ split + s(t) P Bat ψ split Q LHV (15) where . m f ,eq ψ split , t is the equivalent fuel consumption rate, . m f uel is the ICE fuel consumption rate, s(t) is the s-factor, and Q LHV is the lower heating value of the fuel. The m f ,eq ψ split , t in each time step. The s-factor is controlled by a P-controller where s 0 is the initial s-factor, K p is the proportional gain, SOC(t) is the current SOC and SOC Target is the target SOC. HEVs drive in charge-sustaining mode (CS), meaning the SOC at the end of the cycle is the same as the starting SOC. Therefore, SOC Target is the starting SOC at SOC(t 0 ). The two PHEV systems are operated in a CD at the beginning of the cycle until the battery reaches SOC Min . Afterwards, the operating mode switches to a CS mode, where the SOC fluctuates around the lower limit in a narrow range. Accordingly, the target SOC is SOC Min .
To prevent frequent ICE starts in a short amount of time, the minimum ICE runtime was implemented as five seconds in the A-ECMSs. For the first start, the minimum runtime was increased to three minutes to heat up the ICE and exhaust gas treatment.

Life Cycle Variables
To evaluate the LCA emissions, it would be inaccurate to consider the fossil fuel consumption only. PHEVs in electric mode and BEVs have no tailpipe emissions, but the generation of electric energy produces GHG emissions that influence the LCA results significantly [39]. Additionally, the larger battery contributes to the total vehicle lifetime emissions due to the energy-intense production.
This study considers the CO 2 -equivalent emissions of the production and use-phase of a vehicle. Environmental impacts of the recycling phase for the vehicle and battery are included in the production values coming from other LCA studies. The emissions for the production phase are normalized by km with an assumed vehicle lifetime of 225,000 km. The assumptions are summarized in Table 4. Table 4. LCA emissions for the production phase and energy sources.

2030
Conv./48 V HEV production (t CO 2,eq /t vehicle) 5.20 4.42 PHEV production (t CO 2,eq /t vehicle) 5.70 4.85 Battery production (kg CO 2,eq /kWh) 90 35 Gasoline (g CO 2,eq /MJ) 90.4 90.4 Electric energy (g CO 2,eq /kWh) 438 254 A weight-based approach from [40] was used to simulate the vehicle production emissions. In our study, the LCA emissions for the production and recycling of the glider and the powertrain from [41] were divided by the average mass of lower medium cars in Europe. For a conventional vehicle, 5.2 t CO 2,eq /t vehicle were emitted in 2020, which is assumed to be the same for the 48 V HEV. A PHEV produces slightly more GHG gases, with 5.7 t CO 2,eq /t vehicle excluding the battery. The value is expected to decrease by 15% towards 2030 [40]. Emissions for the battery production have changed in recent studies due to more available production data, production plant location shifts, and different chemical compositions. With further improvements, the value is expected to decrease from 90 kg CO 2,eq /kWh in 2020 to 35 kg CO 2,eq /kWh in 2030 [41], which is a feasible range compared to the extensive literature review in [42].
The emissions of the use-phase include the well-to-tank processes as well. The gasoline emissions are based on a publication of the EU [43] with 90.4 g CO 2,eq /MJ, and are expected to remain constant throughout the investigated time horizon. It is important to quantify the year for the GHG emissions of the electric energy, as the electricity mix is moving towards lower emissions due to a higher share of renewable energy sources. In the LCA of the European Commission report [41], the EU 28 CO 2 -equivalent emissions for electricity are assumed to be 438 g CO 2,eq /kWh for 2020 and 254 g CO 2,eq /kWh for 2030. The assumptions in this study rely on the baseline scenario, which includes all planned or already implemented EU and national policies.

Results of Year-Round LCA
In this chapter, the LCA simulation results are discussed. Figure 7 displays the results for a subcompact vehicle and Figure 8 for the compact class vehicle. In the figures, the electric and fuel emission bars represent the A-ECMS simulations. The possible improvements due to an optimal EMS are shown with the white error bars, which indicate the LCA result for the DP results. The optimal battery capacities with the lowest GHG emissions are marked with diamonds; red for the A-ECMS and white for DP. The results for the year 2020 are shown in the top part for each driver; the lower part demonstrates the LCA results for the year 2030. The relative saving potential compared to a conventional vehicle of the corresponding year is given on the right side of each bar. The next subchapters discuss the influence of individual changes on the optimal battery capacity and LCA results.

EMS Influence
All electrified vehicles were simulated with an A-ECMS and DP, whereas the conventional vehicle only has one result, as the operation strategy has no opportunity to split up the power between different components. The emissions due to the production are not dependent on the EMS, so only the electricity and fuel consumption bars are different. For the 48 V HEV, the GHG reduction potential is solely based on reducing the fuel consumption, as the system does not consume electric energy as a PHEV does. For the PHEVs, the DP-saving potentials can originate from a lower electricity or fuel consumption. However, the emissions due to electricity consumption are only 0.4% smaller on average for a DP subcompact class PHEV compared to the ECMS variant, so the main factor for overall lower GHG is the reduction of the consumed fuel as well.
The reduction potential of the DP relies on three effects. In contrast to DP, the A-ECMS includes parameters that prevent switching the ICE on and off frequently and ensure a minimum runtime. The runtime for heating up the engine has a significant influence on short trips. Secondly, the DP propels the vehicle with a higher efficiency because the cycle is known beforehand and adjusts the SOC accordingly. Finally, the battery is discharged slowly if the cycle length exceeds the electric range for PHEVs. If the electric energy could be used more efficiently at the end of a long trip, the A-ECMS might have already depleted the battery charge.
The benefit of an optimal EMS is the greatest for the 48 V HEV, where it is 5.6% on average for the subcompact vehicle and 5.3% for the compact class vehicle. It shrinks with larger battery capacities and HV systems. If a cycle is shorter than the electric range of the vehicle, it can be completed in pure electric mode, and the two EMS deliver the same result. As this occurs more often for higher battery capacities and an HV system, the difference shrinks between the two EMSs.
The previously mentioned points also affect the optimal battery size, indicated by the red and white diamonds. The optimal battery capacity for a DP-controlled vehicle is slightly smaller for all systems and drivers compared to the ECMS minimum. The optimal EMS propels the vehicle with a higher efficiency, especially at lower capacities, so a larger energy storage is not useful. In some cases, the optimal capacity is outside the range of typical capacities below 30 kWh. In those cases, the results for 30 kWh are extracted, because a further capacity increase would reflect a heavy and expensive BEV with dual powertrains. The values for the optimal battery capacity and corresponding LCA emissions can be observed in Table 5.

Influence of Investigated Year
The GHG emissions for the conventional and 48 V HEV systems were reduced overall due to the lower vehicle production emissions, but remained constant in the usage phase throughout the years. As the emissions from the vehicle, battery, and electric energy production decreases in the 2030 scenario, the overall life cycle GHG impact for PHEVs drops significantly in future years. The emissions during the production of the battery have an especially minor influence on the total LCA emissions for PHEVs beyond 2030.
The change of LCA emissions throughout the years depends on the electric driving share because the fuel emissions remain constant. The lowest share of electric driving can be noted for the compact class long-distance driver and a 48 V 5 kWh battery. The relative reduction potential of 16.2% increases to 21.4% in 2030. In contrast to this, the relative potential of 38.4% for the short-distance driver with the largest HV battery, where almost every daily distance can be completed in pure electric mode, increases to 58.8%. Similar values can be seen for the subcompact class.

Use Case Influence
Four typical driving use cases were defined in Chapter 3 which show a different GHG reduction potential. The baseline emissions for the conventional vehicle are almost the same, except for the short-distance driver. Taking the compact class vehicle as an example, the GHG emissions for the average, long-distance, and commuter driver are between 191 gCO 2,eq /km and 194 gCO 2,eq /km in 2020. The short-distance driver results in 219 gCO 2,eq /km. As the driving style is slightly more aggressive and no start-stop is possible during the heating time, the fuel consumption is higher during short trips. The opposite behavior can be seen for the 48 V HEV, where the relative reduction potential is the largest for the short-distance driver in both vehicle classes. In short trips, the share of mechanical energy for acceleration is higher compared to the rolling, climbing, and air resistance parts. Similarly, the share of recuperation energy is higher compared to the total trip deceleration energy due to the frequent speed changes for the short-distance driver. Consequently, the advantage of a 48 V HEV is very large during short distance trips and decreases with higher trip lengths, which usually have a higher mean speed and are less dynamic.
The results of the LCA for the PHEVs correspond to the statistical driving behavior in the UF curves. The largest relative fuel-saving potential can be achieved by the shortdistance driver and the lowest by the long-distance driver. The higher share of electric driving even overcompensates the higher baseline value of the short-distance conventional vehicle so that the absolute GHG emissions are lower for most of the systems.
The commuter shows no large leap at a certain battery capacity, even though many trips are in a specific distance range. Typical of a work commute, 67% of the trips are within a range of 25 and 45 km. Combined with shorter trips, 76% of the trips have a distance below 45 km and could be completed in mostly pure electric mode with a battery capacity of less than 10 kWh, even for the compact class vehicle. Nevertheless, in 2030, the optimal battery capacity exceeds 30 kWh, although this capacity would electrify 90% of the trips purely. The remaining 10% of the trips are in a range of 210 and 270 km, where battery capacities larger than 30 kWh still reduce the GHG emissions. This is partly due to the assumption that the vehicle is not charged intermediately during the day. A recharge of the battery during long trips would be possible and lead to lower optimal capacities. Nonetheless, this example demonstrates that it is important to consider the trip length, which cannot be driven in charge-depleting mode, and not only the number of trips.

Vehicle Class Influence
The overall GHG emissions increase for all systems and drivers towards the compact class vehicle because the additional weight has a larger influence on the required wheel energy compared to the slightly lower drag resistance. Furthermore, vehicle production emissions increase due to the additional weight.
The relative potential for the 48 V HEV only varies within 0.3 percentage points between the two classes. This value is larger for the PHEV variants, where the difference can be up to 4 percentage points less for the compact class vehicle. The emissions due to electric energy consumption increase only slightly towards the compact class vehicle because the power demand on wheel level increases. The amount of fuel emissions increases proportionally more, because in addition to the increasing power demand, the electric range decreases, so the vehicle is driven in CS mode more often. For the 48 V PHEV, a higher difference is noticeable because more operating points exceed the power limits of the EM and the number of ICE starts increases.

Road towards Climate Neutrality and Potential of 48 V PHEVs
The transport sector needs to achieve lower GHG emissions in the coming years to keep global warming below 1.5 • C. Current regulations within the EU limit the tailpipe emissions and do not consider the whole life cycle of the vehicle. Therefore, a direct comparison between regulatory fleet target limits and the LCA simulation results for individual vehicles is not possible. However, the overarching climate goals are summarized in the European Green Deal, which aims at climate neutrality until 2050 and a GHG reduction of 55% by 2030 compared to 1990. As emissions have already dropped by 24% until 2019 [44], a further reduction of 41% from a 2019 baseline is necessary.
In the current phase of transformation towards electrification, the concept of HEV respective to PHEV has to be regarded as a supplement to BEVs. Hybrids can typically cover those use cases which can't be managed by battery electric vehicles because of, e.g., range limitations, infrastructure, or power requirements. Considering the relatively low costs of a 48 V HEV system, these systems offer an easily implemented solution with a comparably large saving potential and no need for charging infrastructure. The small batteries do not increase the emissions in the production phase significantly, and only a small quantity of additional resources would be necessary. Within a fleet mix, the 48 V PHEV in combination with BEV and other hybrids can significantly contribute to a reduction of GHG emissions and reach future emission targets. Even when incorporating future climate protection efforts, PHEV concepts with their 2030 LCA potential can reduce the overall emissions of the transport sector.
PHEVs rely on frequent charging to electrify as many kilometers as possible. Due to the high daily distances of the long-distance driver, the CD share is smaller, indicated by the lowest share of electricity emissions. Consequently, the improvements in electricity production have a smaller impact compared to the other drivers. Still, a GHG reduction of at least one-third in 2030 is possible when the vehicle is equipped with a 20 kWh battery. For any upcoming fleet emission scenario, PHEVs are vital to achieve the objectives in 2030 on an LCA data basis. Even small battery capacities help to reduce the overall LCA emissions significantly, but larger capacities compared to today's market average can increase the relative GHG savings. In contrast to the results for 2020, the battery should be as large as possible for the lowest LCA emissions in 2030. Only the short-distance driver has an optimal battery capacity below 30 kWh, but with a negligibly small difference in overall GHG emissions.
As expected, the 48 V PHEV cannot reach the same GHG-saving potentials as an HV PHEV. It still has a large benefit compared to an HEV system, and the overall LCA emissions are similar to the HV vehicle emissions for all use cases and vehicle classes. While HV PHEVs can cover trip distances below the electric distance in pure electric mode, the 48 V PHEV needs to turn on the ICE if the power demand exceeds the limits of the less powerful EM. On the contrary, during very long daily driving distances, the fuel consumption of the 48 V PHEV is lower than the HV PHEV. A large share of the possible recuperation energy can be covered by the 48 V EM and the lower weight results in lower energy demand on wheel level. As these long trips have a higher influence on the year-round fuel consumption, the 48 V PHEV can compensate parts of the higher fuel consumption in short trips. This can be validated in the figures by comparing the relative reduction potential between the 48 V and HV PHEV at different capacities. With 5 kWh, many trips exceed the electric range and the 48 V PHEV results are therefore close to the HV values. With increasing battery capacities, the difference between the two system voltage levels increases.
In addition to CO 2 targets, the 48 V PHEV technology can also be expected to be beneficial with respect to real-driving emissions. The additional battery capacity compared to HEVs offers an increased electric driving share at the start of the vehicle when the exhaust gas aftertreatment system is not conditioned. At the same time, electric catalyst heaters for 48 V systems can be required to meet future emission regulations. Catalyst heaters can ensure the required temperature before the ICE is started. As the ICE is started more frequently compared to HV PHEVs, the average catalyst temperature can be expected to remain at a higher level. This can reduce the additional heating and therefore electric energy demand if the EMS is required to switch on the ICE.
The lower additional powertrain costs for 48 V PHEVs could help to increase the PHEV share in the cost-driven subcompact class market. Substituting more conventional and HEV vehicles with PHEVs reduces the life cycle GHG emissions while the vehicle range remains constant due to the CS phase. Especially in inner-city traffic, where the distances and power requirements are typically low, the ability to drive electrically results in high emission reduction potentials. The lack of HV components also reduces the maintenance and safety issues, which requires specially trained personnel and tools in workshops. On the contrary, electric driving is only possible with limited power, which leads to limited driving dynamics towards heavier vehicle classes. Additionally, the GHG emission-saving potential relies on regular charging of the vehicle.
Although we focused on the European market in this study, other markets around the world are attractive for a 48 V PHEV system as well. Especially in Asian countries, megacities struggle with a dense traffic volume and thus emission issues. At the same time, the traffic jams lead to slow vehicle speeds and therefore low power requirements. In combination with low average vehicle weight, such as in India, large trip shares can be electrified with a 48 V PHEV at a cost-attractive level.

External Influencing Factors
The results of this study show that next to the dimensioning, several external factors influence the resulting GHG emissions of PHEVs as well. All implications can be applied to 48 V and HV PHEVs in the same manner as the influences have been shown to be similar. First, the analysis of the real driving behavior showed that only the short-distance driver UF curve is located above the EU WLTP curve. Since the average annual driving distances in the European fleet are in the range of the average driver, the CD shares of most customers are below the estimated shares by the current European UF curve. Though we only compared four drivers to the regulation curve, other studies with larger driving trip databases have shown similar results [45,46]. This issue is crucial for trust in the ability of PHEVs to contribute to the climate goals because the fuel consumption values of the customers diverged from the certified values. However, this study demonstrates that significant fuel consumption and GHG savings are possible with the plug-in technology if the systems are used under the mentioned assumptions. In the future, all new vehicles in the EU are monitored with the introduction of the on-board fuel consumption monitoring systems under real-life conditions. It is planned to use the gained data in future fuel consumption regulations.
Second, the LCA demonstrates the importance of the emissions due to electric energy consumption. Even with the current electricity mix, life cycle GHG emissions can be reduced with PHEVs, but a further inclusion of renewable energy sources into the grid is necessary to increase the GHG reduction potential. The assessment of life-cycle emissions for passenger cars in future years is very important, and its inclusion in the regulation process would help to reach the climate goals on a comprehensive level. The consideration of electricity production GHG is especially important to evaluate the potential of hybridization because the impact on LCA emissions is larger compared to the differences in the production phase. The additional greenhouse gases of the battery production influence the LCA emissions in the future marginally, so the electricity GHG share is even more dominant. Together with increasing market shares of electricity-consuming BEVs and PHEVs, a low-emission electricity mix is vital for the transportation sector.
Finally, the LCA emissions of PHEVs in future years indicate an optimum at much larger battery capacities than the current market average. Considering the effects of the GHGs of the electricity mix show a decreasing trend during the vehicle lifetime, PHEVs with higher battery capacities can help to reduce the emissions today as well. Shrinking battery prices can even allow larger battery capacities for the same additional costs as today, while improvements in the battery cell itself can reduce the weight and volume of the battery packs at the same time [47]. In addition to larger battery capacities, frequent charging increases the electric driving share and thus reduces the LCA emissions of PHEVs. Convenient access to charging stations, fast charging along highways, and competitive energy prices are therefore essential to electrify more driving shares [48]. Besides, it can be expected that additional savings in fuel consumption are achieved by the next generation of ICEs due to stricter emission regulations. The application of dedicated ICEs, which are optimized for the operation in Hybrid-Powertrains, offers further relevant potential [49].
The presented approach covers many driving trips for different vehicle types and EMSs to give a holistic view on the GHG-saving potential. Nevertheless, the LCA results are depended on the input assumptions, which may vary, especially for the 2030 scenario. The GHG-saving potentials need to be investigated for further regions. Also, temperature effects and emission modelling besides CO 2 , which influence the environmental impact of PHEVs, have been neglected in our powertrain simulation model [50,51]. Alongside technical aspects, such as the thermal and degradation behavior of a 48 V PHEV battery, the economic analysis of this system bears further research potential [52].

Conclusions
In this study, the environmental impact of different electrified systems was investigated and their potential to contribute to climate goals was analyzed. To ensure a realistic and comprehensive comparison, we utilized real-drive data from different drivers and focused on life cycle GHG emissions.
The LCA results reveal that PHEVs can significantly reduce GHG emissions for passenger cars compared to the conventional and 48 V HEV systems. Since the relative reduction potential of 48 V HEVs is limited to 23% in our case, vehicles with higher degrees of electrification are necessary to meet future regulatory targets and climate goals. Even 48 V PHEV vehicles with a small 5 kWh battery capacity reduce the greenhouse gas emissions by 16 to 34% for the 2020 assumptions. This value is expected to increase up to 46% in the 2030 scenario. Optimal sizing of the energy storage system can increase the emission saving potential even further. Therefore, the PHEV is one promising vehicle technology on the road toward zero-emission mobility.
The 48 V PHEV shows relative saving potentials close to the HV PHEV values. Considering the lower costs, weight, installation space, safety, service, and manufacturing requirements, this technology could open the PHEV market in low-cost vehicle segments. As the results for the subcompact class and compact class vehicles are similar, the GHG savings can be expected to work for many vehicle types. Next to the technological aspects, boundary conditions need to be fulfilled, nonetheless. Regular charging and an increasing share of renewable energies in the grid are key factors for the reduction of GHG emissions. Further research will investigate the battery requirements of 48 V PHEVs with a focus on the aging behavior and effects of BEV cell utilization in this application. Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest. Total number of trips for one driver n ser /n par number of serial/parallel cells K p Proportional gain P EM,Mot /P EM,Gen

Nomenclature and Acronyms
Maximum motoric/generator power (kW) P Pnt /P EM /P ICE /P Bat 12 V electrical system/EM/ICE/Battery terminal power (W) Q LHV Lower heating value (J/kg) R CD Charge-depleting range (m) R Cell scaled battery cell resistance (Ω) R i,5s normalized cell resistance (ΩAh) r dyn Tire radius (m) SOC Max /SOC Min /SOC/SOC Target Battery upper limit/lower limit/current/target SOC