**1. Introduction**

The European Union (EU) has agreed on a range of policies aiming to reduce greenhouse gas emissions in various sectors of society. Since transport largely contributes to these emissions by a share of 27% of the EU's total emissions in 2016, these emissions have to be reduced. For the year 2030, this policy implies that in the EU fleet-wide CO2 emissions of passenger cars should be reduced by 37.5% as compared to 1990 levels. For new vans and trucks, the emissions should be reduced by 31% [1]. Therefore, new strict targets require the reduction of average CO2 emissions of new vehicles that will enter the market. Consequently, the year 2020 is widely expected to bring dramatic changes to the automotive market. Due to the aforementioned targets, manufacturers are forced to invest intensively in innovative technologies of sustainable mobility. Therefore, many automotive players focus on battery electric vehicles (BEVs). In recent years, a large number of environmental impact studies were published, analyzing the potential environmental benefits of electric vehicles (EVs). The overall conclusion is that BEVs are preferable over petrol and diesel vehicles, however only if charged by renewable energy [2]. A possible solution is to charge these cars with low-emission renewable energy technologies such as photovoltaic systems. This could be achieved by charging stations which are powered by PV systems or by photovoltaic solar modules which are built in a car's body parts, also called vehicle integrated photovoltaics (VIPV). Some vehicle manufacturers already aim at integrating PV cells in body parts of their passenger cars. One of the most recent solar powered electric vehicles is the Lightyear One of the Dutch company Lightyear. The vehicle has an integrated silicon PV array of more than 5 m<sup>2</sup> with a nominal installed powered of 1250 Wp. Similarly, Munich-based producer Sono Motors is planning on launching their solar electric vehicle, named Sion. The Sion's PV array has a nominal power of 1200 Wp. Solar charging in summer can add 34 km to the drive range of 255 km. With Audi´s e-tron Quattro with a nominal PV power of 400 Wp and the Toyota Prius P with a PV array of 860 Wp, two of the car industry's major players recently entered the market as well. Especially for light utility electric vehicles (LUV), VIPV could be an attractive feature due to their predictability of utilization, in particular their moments of use and daily travel distances, and their significantly larger and flat roof surface which, if covered by solar cells, can potentially yield sufficient amounts of solar power. LUVs are usually vehicles with a gross vehicle weight of no more than 3.5 metric tonnes and are optimized to be tough-built, have low operating costs, and to be used in intra-city operations. Though prior studies have often indicated that VIPV will result in lower CO2 emissions, actual life cycle assessments (LCAs) of VIPV are barely available, and most claims until now have not been quantified or validated for the specific situation of VIPV of LUVs [3–5]. Thus, the goal of this work is to analyze how PV-powered vehicles can contribute to sustainable mobility. Therefore, an LCA focused on determining the CO2 emissions of a German VIPV LUV called StreetScooter will be conducted. The results of this research could be useful for car manufacturers, to calculate emissions per vehicle, for political institutions to estimate environmental impacts for the transport sector, and for business parties in the solar market to identify further application possibilities and yield useful data to identify critical areas for the improvement of VIPV for LUVs.

This LCA study was executed in the framework of a project called STREET, which was funded by the German Ministry for Economic Affairs and Energy and a German logistics company of Deutsche Post DHL Group named StreetScooter, which is currently working on the integration of PV on electric light utility vehicles (see Figure 1). Forschungszentrum Jülich as an organization for applied research supports the project by equipping the vehicle with PV modules and by analyzing the energy yield of PV modules on this vehicle by the analysis of data measured by radiation sensors on the vehicle under real shading and reflection conditions.

**Figure 1.** StreetScooter Work L Reprinted from: CC-BY-SA-4.0 (via Wikimedia Commons), Superbass, 2017.

This paper is structured as following: in Section 2 the LCA method will be explained and all input parameters of the LCA will be described. Major assumptions regarding the operation phase and technology choice for the on-board vehicle application are discussed. The results, sensitivity analyses, and limitations of the study are reported in Section 3. Finally, Section 4 summarizes the results, presents the conclusions, and offers recommendations for future studies.

#### **2. Method and Data**

This section presents the general methodology used to execute the LCA, defines the e fficiency of the VIPV investigated, and quantifies the resulting CO2 emissions. Additionally, key parameters that limit the environmental performance of the electricity produced by the PV system integrated into the vehicle are shown. Assumptions about these critical parameters for the reference case are clarified.

#### *2.1. Life Cycle Assessment Method*

LCA is a useful tool to quantify environmental performance, considering a holistic perspective. LCA is generally understood as a compilation and evaluation of the inputs, outputs, and potential environmental impacts of a product system throughout its life cycle [6]. LCA studies always consist of four main phases, which are covered through ISO standards (DIN 14044; ISO 14040:2006). The first step of the LCA is used to define the goal and scope of the study. The second step is a life cycle inventory (LCI) model through which data is collected and organized. The third step is the life cycle impact assessment (LCIA), used to understand the relevance of all the inputs and outputs in an environmental framework. The fourth step is the interpretation, which is a systematic technique to identify, check, and evaluate information resulting from the LCIA (see Figure 2).

**Figure 2.** Life cycle assessment (LCA) framework (DIN 14044; ISO 14040:2006).

The environmental impact assessment for this study is completed at the mid-point level. Midpoints are considered to be connections in the cause–e ffect chain of di fferent impact categories, also known as the problem-oriented approach or classical impact assessment method. Greenhouse gas emissions (kgCO2-eq) were used as an indicator of climate change contribution. The 100-year global warming potentials based on the latest IPCC 2013 were assumed, according to their radiative, forcing capacity relative to the reference substance CO2. Global warming potential (GWP) during the life cycle stages of a PV system was estimated as an equivalent of CO2 containing all the significant emissions CO2 (GWP = 1), CH4 (GWP = 25), N2O (GWP = 298) and chlorofluorocarbons (GWP = 4750–14,400). The calculations were performed using LCA software GaBi with Ecoinvent v2.2+ as back-ground database. GaBi is a process-oriented software, examining the material and energy flows of each step of the production chain. The applied methodology of this study can be divided into four sections which will be briefly described below.

#### **(A) VIPV Use Case Parameters**

In the first section, the system boundaries and the input parameters for the operation in urban delivery were clarified and collected for the Use Case StreetScooter. Promising VIPV configuration was defined as the result.

#### **(B) LCI of VIPV—Manufacturing**

The second section was dedicated to compiling an inventory of energy and material inputs and outputs over the life cycle of VIPV. Life-Cycle Inventory was completed based on the data from literature. Based on the Inventory, a GaBi model was developed.

#### **(C) LCI of VIPV—Operation and VIPV Energy flow model**

The third part included the simulation of VIPV contribution to charging. To simulate the reduction of grid power demand, an energy flow model for the identified reference case was developed.

#### **(D) Evaluation of environmental impacts**

In the last part, potential environmental impacts (GWP) related to identified inputs and releases were evaluated. LCA results were compared to grid charging by means of their environmental impacts. The characterization factors were based on IPCC (2013) and should be incorporated for the impact category of global warming potential, which is tracked in kg CO2 eq.

#### *2.2. Functional Unit, Goal, and Scope*

The use case is the light utility battery electric vehicle Work L of StreetScooter. The functional unit for this study is 1 kWh of electricity supplied by the PV system to the battery of the StreetScooter. In comparison to the functional unit of 1 km driven, the emissions of 1 kWh can be calculated more accurately. Furthermore, the chosen functional unit of 1 kWh allows for a direct comparison of effects of charging by PV modules to those due to charging by the grid. Thus, the emissions of VIPV and grid charged BEV can be evaluated more precisely referring to the same functional unit. The operation of the electrical vehicle is set in Cologne, Germany and starts in 2017. Within the scope of this project, the environmental impacts of VIPV are to be studied according to the standard of life cycle assessment ISO 14040:2006. The PV system configuration is based on the first generation of the VIPV panels for the STREET Project with heterojunction silicon PV modules manufactured in China. The analyzed VIPV configuration includes three panels and three control units including the cables mounted on the vehicle roof. The overall capacity of the VIPV system is 930 Wp. The system of the VIPV electricity includes raw material extraction, wafers, crystalline silicon-based heterojunction solar cells and module manufacturing, mounting structures manufacturing, inverters manufacturing, system installation, and the operation.

#### *2.3. Input Parameters for the Life Cycle Inventory*

The production process of a typical commercial crystalline silicon solar cell is modelled based on the existing datasets describing the supply chain [7] (see Figure 3). Input parameters of the manufacturing of the PV control unit (PVCU) as well as the vehicle integration process were added based on internal communication in the project STREET. The electricity consumption on all process levels is modelled following specific electricity mixes corresponding to China (CN) or Germany (DE), respectively, based on the Ecoinvent datasets.

**Figure 3.** Vehicle integrated photovoltaics (VIPV) system value chain: process flow diagram. CN means China, DE Germany.

All input parameters of the manufacturing and vehicle integration process are described in Table 1. The exact location of manufacturing plants is undocumented and unknown. However, it can be assumed that the location of these plants is somewhere within China. Modelling of the transportation was based on the standard distances as suggested in the Guideline for PC LCA [7]. Metal parts were commonly reported with 200 km train and 100 km truck transportation in China. Additionally, transoceanic transport from China to Belgium was estimated to be 19,994 km based on searates.com data. In Europe, lorry transport from Antwerp (Belgium) to Cologne (Germany), a total of 500 km, was used.


**Table 1.** Input parameters of the manufacturing process and vehicle integration Ppocess.

(a) Guideline (Frischknecht et al. 2015) [7]; (b) STREET Internal Expert Judgement; (c) LCI on SHJ Cells (Louwen et al. 2016; Olson et al. 2013) [8,9].

For the solar cells in the VIPV, heterojunction technology (SHJ) was chosen due to the best trade-off between efficiency and costs. Thus, in the LCA, the cell heterojunction process was described by the following process steps taken from [8,10] and shown in Table 2. The metallization of the front side requires a double print of the standard amount of silver paste and sputtered aluminum closed back side. The LCI data on material and energy consumption were added for heterojunction cell processing, referring to [8–10].


**Table 2.** Process steps of heterojunction cell.

#### *2.4. Input Parameters for the Energy Flow Model*

The main factor for the estimation of PV electricity generation is the effective solar irradiance, which depends on the route and location, season, time, and module configuration and orientation. For the reference case of the LCA, the location for the operation was set in Cologne, Germany. The hourly global horizontal solar irradiance was defined by averaging hourly incident global horizontal radiation data extracted from the PVGIS database. The on-board generation of electricity was simulated based on degradation, system losses, and shadowing factor (see Table 3). A 19.7% module efficiency was assumed [9]. In line with IEA-PVPS methodology guidelines [7], degradation of 0.7% per year was applied. Operation time of the reference case was set to 8 years, based on data of LUVs in delivery services [11].

**Table 3.** Input parameters for the operation of the VIPV.


According to the literature guidelines, efficiency for the VIPV system was estimated. Due to dynamic shading, an average 70% performance compared to residential PV was assumed [3]. Furthermore, generated energy cannot be used directly for traction of the vehicle and must be stored in the battery, where DC-Charging/discharging loss of 2% appears. Additional loss of 5% was considered due to the DC/DC converter. The loss of the MPP tracking additionally limits its efficiency in the model to 95% [3]. A performance loss of 9% due to temperature increase and low irradiance was assumed [5]. The overall average efficiency losses of the VIPV system is to be found in Table 4.

**Table 4.** VIPV system efficiency.


#### *2.5. Input Parameters of the Grid Charge*

The grid mix in the location of the charge was analyzed regarding its carbon intensity. The emissions of the grid can vary massively depending on the different power plants. Fossil power plants dominate the power generation in Germany. Acknowledged studies usually consider annual average carbon footprints of the grid power plants caused by the life cycle (construction, fuel production, operation, etc.) [12]. Hourly average emissions of the German electricity mix vary depending on the day and night times. The German electricity mix was modelled using SMARD electricity generation data from 2017 and utilized for the projection of the future scenario [13].

The reference scenario follows the pathway of technological development as far as possible, according to the goals set by politics. The target of the electricity sector in Germany for 2030 is 180–186 Mio t. Until 2028, the annual electricity mix GWP is expected to decrease by 2% per year [1]. Table 5 gives an overview of the emissions of different electricity sources, found in [14].


**Table 5.** Emission factors of electricity sources.

#### *2.6. Reliability of the Data*

The LCI in this study was based on extracting the data from reliable literature. The commercial LCA software GaBi Version 8.7.1.30 was used to model and calculate the LCI and impact assessment results. Essential materials, electricity mixes were calculated based on data represented by the Ecoinvent database unless otherwise noted. The International Energy Agency (IEA) developed guidelines to make the LCAs of PV systems more consistent and to enhance quality and reliability. Data on production is mainly based on these guidelines and LCIs of photovoltaics [7], additionally considering the heterojunction process of [9,10]. Some values for the Vehicle Integration Process and PVCU were adjusted after internal communication in STREET. The reason for adjustment was mainly a lack of access to the supply chain model data. The data used for this LCA varies in quality and reliability. To limit the resulting uncertainty, the differences of the data sources were analyzed and scored referring to the Quality Pedigree Matrix Flow Indicators determined by DIN 14044. Due to the above-mentioned conditions, the scores for each step were evaluated in Table 6. The highest score shows the lowest uncertainty and data scored with 5 shows the highest uncertainty.


#### **Table 6.** LCA data quality.
