Abstract
Environmental impacts of battery electric vehicles (BEV) and internal combustion engine vehicles have been broadly studied and compared. However, there is no evidence of studies comparing the potential effect of key factors such as vehicle mass reduction, life cycle inventories regionalization and electricity mix evolution in a Brazilian conditions scenario. The purpose of this study was to evaluate what would be the theoretical environmental impact of manufacturing BEVs and battery electric busses (BEB) in Brazilian south east and to compare those results to what is intended to be a global vehicle in 2015 and 2030. The methodology was based in adapting global life cycle inventories to local Brazilian south east conditions and then to make a comparison. Thus, a representative number of Ecoinvent V.3.02 datasets were adapted to better represent local conditions. The study established a comparison using 1 car and 1 km as functional unit. This research envisioned mass reduction setups for cars and busses in 2030, and additionally two material switching scenarios: plastic based and aluminum based. Bus results suggest that there is a tendency for Brazilian bus to show better results than its global counterpart except for freshwater eutrophication. A large environmental contribution from treatment of mining residues is common for all impact categories where local bus performed worse than global corresponding. This lead us to believe that in order to manufacture an environmentally competitive BEB, reduction of impacts on metal extraction, especially copper, and its residues must be prioritized. Electricity consumption is by far the main contributor for climate change for a functional unit of 1km. Hydroelectricity linked methane emissions from reservoirs and natural gas in the electricity mix are the main sources of GHG emissions. For BEV results in 2030, an unexpected result appears during ozone depletion examination; Brazilian aluminum-based BEV exhibits the largest impact, even though mass reduction was considered. Paradoxically, the large share of aluminum recycling in Brazil acts as a double edge sword since emissions arising from scrap treatment do have a significant impact. Maintenance stage proves itself as the largest contributor for Photochemical oxidant formation due to ethylene use. For human toxicity, metal depletion, freshwater eutrophication and ozone depletion contribution vehicle components count for nearly 80% of total impact per km. It must be considered that this model cannot capture the potential evolution of GLO inventories, especially Chinese manufacturing processes. Any future research must prioritize Brazilian inventories construction.
Introduction
Vehicle electrification is deemed as a way to mitigate greenhouse gas (GHG) emissions and to improve quality of air on large urban centers around the world. Air quality is a concern for many mega-cities, including São Paulo, the largest metropolitan area in the southern hemisphere. Andrade et al.(2017) conclude that the greatest air quality challenge currently faced by both, São Paulo State Environmental Protection Agency and the local communities is controlling secondary pollutants such as ozone and fine particles. Battery electric vehicles (BEVs) are expected to contribute for GHG mitigation emissions in urban areas since they present zero emissions during their use phase.
BEVs are still far from being a representative share of Brazilian automotive fleet. This is partly due to high ownership costs, driven by tax burdens which create disadvantages for BEVs with regards to conventional cars (AES Brasil, 2017) and lack of charging infrastructure. In an international context, BEVs registrations, including plug-in hybrid vehicles and fuel cell cars, hit a new record in 2016, with over 750 thousand sales worldwide (International Energy Agency, 2017). Arguments supporting BEVs adoption are that electric powertrains are more energy efficient for propelling vehicles than conventional internal combustion ones fueled by petrol, ethanol or diesel, besides electric propulsion barely emits noise (Sadek, 2012). In order to analyze the actual environmental benefits from BEV deployment, numerous studies have paid attention to their performance compared to internal combustion engine vehicles (ICEV) and hybrid electric vehicles throughout their entire life cycle. Research include fuel/electricity generation, use phase, vehicle production and in some cases End-of-Life (EOL) stage (Boureima et al., 2009; Faria et al., 2013; Hawkins, Singh, Majeau-Bettez, & Strømman, 2013; Helms, Pehnt, Lambrecht, & Liebich, 2010; Ma, Balthasar, Tait, Riera-Palou, & Harrison, 2012; Messagie, Boureima, Coosemans, Macharis, & Mierlo, 2014; Rajagopal et al., 2012). Results show lower GHG emissions for BEVs when a complete life cycle is considered. Conventional and hybrid busses have also been a matter of research (Buø, 2015; Olofsson & Romare, 2013) however, few evidence was found on battery electric busses (BEB) comparative LCA (Falco, 2017). Furthermore, there is evidence of BEV/Conventional cars LCA research within the Brazilian framework (Choma & Ugaya, 2013; Velandia Vargas, Seabra, Walter, Cavaliero, & Falco, 2016)
In a literature review containing conclusions from 79 papers, Nordelöf et al.(2014) stressed that vehicle LCA results vary greatly. It is also reported that only a few articles appropriately report the study time scope. Moreover, most of the studies focus on current BEV technology, which is rapidly evolving, meaning that there is a lack of future time perspective, e.g., evolution in materials, mass reduction and variations in electricity production. A conclusion, common to every study is that when the functional unit of comparison is defined as a travelled distance, for instance: 1 Km or 1 mi, electricity generation is the main cause of environmental impact for EVs. Consequently, they can reach their full potential in mitigating global warming only if the charging electricity is not fossil carbon intensive. Surprisingly, very few reports put emphasis in transmitting this conclusion as a core message.
Like all GHG mitigation actions, the implementation of EVs must be evaluated carefully to avoid environmental burden shifting or rebound effects. Skepticism is present, Frischknecht & Flury (2011) even point out that the role and contribution of electric cars to significantly mitigate the environmental impacts of transportation might be substantially overrated and that one core aspect to lower environmental impacts of individual mobility is a considerable evolution in terms of vehicle weight and performance.
Current BEVs manufacture has demonstrated to often increase GHG emissions when compared to ICEVs manufacture (Hao, Mu, Jiang, Liu, & Zhao, 2017; Kim et al., 2016). Battery evolution and mass reduction are promising opportunities to offset BEVs larger environmental burden during production stage.
Location has proven to be an important factor to be considered in both life cycle inventory (LCI) collection and even life cycle inventory analysis (LCIA). Studies have shown that location dependent impact assessment for categories like acidification and eutrophication provide more accurate results than site-generic assessments. LCI uncertainties related to geographical features are a matter of study. Mutel & Hellweg (2009) developed a method to couple existing regionalized characterization factors with large LCI databases which allowed them to obtain different total scores, identify different hotspots and even to vary distributions of the environmental impacts.
Lack of spatially differentiated LCI are specially challenging for large geographies like Brazil, for which most of its inventories are aggregated at country level in the best of cases. Regionalized inventories are essential in order to obtain more reliable results. In the context of the United States, Hao et al (2012) showed that the life cycle GHG and other air pollutants emissions induced by both gasoline and diesel vehicles differ to a varying extent among different regions when considering upstream life cycle stages: crude oil recovery, transportation, refining and distribution. Moreover, Brethauer et al (2015) adjusted both, LCI and characterization factors in order to obtain more spatially differentiated results. Their outcome: It was found that hybrid vehicles known as extended range electric vehicles presented an emission reduction in urban areas when compared to BEVs.
When comparing all LCA stages, the collection of LCI is generally the most effort intensive. This phase includes the quantification of inputs (primary and manufactured), by-products and environmental emissions to air, soil and water. The search for inventories, often results in no available data for a specific region, or in the best of cases LCA practitioners find data adapted to reflect global average values. This lack of geographical detail embodies a great concern for studies which are expected to be more accurate. Country overall data is usually used to represent regions geographically too distant or that present very different environmental conditions e.g. altitude, latitude, weather.
In the same way, LCA primary data is either confidential or it is scattered and difficult to find for researchers who usually do not have access to data for the entire life cycle of the car. Hence, data from diverse stakeholders is required for each stage, adding time and space uncertainties for the study. Furthermore, absence of transparency about the influencing factors of LCA in EVs creates great difficulty for boundary definition and makes the analysis prone to flaws. According to Egede et al. (2015) material composition of the vehicles, electricity mix and use patterns are considered to be the main influencing factors on BEVs environmental assessments.
The National Research Council of the National Academies (2011) points out three main methods to make steel structures lighter. One of them is to substitute lower-strength steel for higher-strength steel. Higher-strength steel can be made thinner thus reducing mass; however, its use can reduce the ability to meet design strength criteria. Furthermore, forming processes might imply an additional environmental burden even greater than that of the avoided mass. Another way to reduce vehicle mass is to substitute conventional steel for sandwich metal material. Sandwich material is light, stiff, and can be formed for many parts. As a downside, joining the parts may be difficult, expensive and it may need additional manufacture processes. Finally, the use of tubes which aim for an optimal use of steel (and aluminum) result in less mass without putting design criteria at risk. Although all of the previously stated methods may increase costs in the present day, this problem is expected to be overcome as mass production is achieved.
Lotus Engineering (2010) reported the technical feasibility for a 2017-2020 mass reduction development program. This model assumed a target total vehicle mass reduction of 40% to be achieved while considering a 50% upper limit constraint on total vehicle piece cost relative to the baseline car: Toyota Venza 2009. This development was intended for a 2020 Model. All technologies used to reduce mass at the vehicle had to be ready to use within the company in 2017 or earlier.
In a comprehensive review of technical literature Lutsey (2010) reports that by means of model redesign, automakers can achieve up to 20% of mass reduction in their vehicles at little or no additional cost and most surprisingly, without deeply shifting their manufacturing technologies. It also reports that a number of technical studies state that vehicle mass reductions from 20-35% in weight could be both feasible and affordable with technology shifts toward mass-reduction techniques. Finally, some automakers roadmaps indicate that mass-reduction technology with minimal additional manufacturing cost could achieve up to a 20% reduction in the mass of new vehicles in the 2015-2020 timeframe.
Reduction material approach is also merged with material switching alternatives. Das (2014) LCA research evaluated alternative lightweight vehicle designs in comparison to a baseline model. A high strength steel and aluminum design (“LWSV”) and an aluminum-intensive design (AIV) were considered. Results show AIV design achieved mass reduction of 25% (compared to baseline) consequently resulting in a decrease in total life cycle primary energy consumption by 20% and CO2 emissions by 17%. In contrast, LWSV have a mass reduction potential of only 15% which leads to higher overall life cycle energy consumption (9%) when compared to AIV design. Overall, the AIV design showed the lowest environmental impact per mile from both; climate change and primary energy consumption point of view.
In another study including mass reduction and materials switching aproach Ricardo AEA (2015b) aimed to understand the potential for automotive mass reduction in the EU market by means of a wide-ranging literature review. Among their conclusions it must be highlight that in spite of the fact that there are examples of vehicles produced almost entirely from high strength steels, aluminum or composite materials, future trends are likely to present a multi-material scenario. Therefore, material use predictions are bound to high levels of uncertainty.
Although there has been a growing interest on electric mobility options in Brazil there is no mass production of electric vehicles in Brazil currently. BEVs future market penetration along with tax regulations for imported goods (AES Brasil, 2017) could encourage automakers to manufacture the cars in the country. However, it remains unclear to what extent a Brazilian car is environmentally advantageous over an imported one. Environmental benefits of BEVs when compared to ICEVs depend greatly on electricity mix but also in the vehicle itself.
The purpose of this study was to evaluate what would be the environmental impact of manufacturing BEVs and BEBs in Brazil by adjusting LCIs to local conditions and then comparing to LCIs which employ data geographically intended to represent a global average. In order to do it, a representative number of Ecoinvent datasets were adapted to better represent Brazilian southeast conditions. Additionally, this research envisioned evolution scenarios for BEVs and BEBs, thus being able to compare Brazilian and global LCIs for a 2030 scenario. Finally, it was our intention to stablish a comparison of environmental impacts per travelled kilometer for each case.
Methodology
Goal and scope definition
This LCA study was carried out to compare the environmental impact of hypothetically manufactured BEVs and BEBs in Brazil versus their average global counterparts. The Brazilian BEV and BEB life cycle inventories were adapted to best represent a manufacturing process in Brazilian south-east conditions for what is considered to be a 2015 and a 2030 scenario. Then, those results were compared to what is intended to be a global BEV/BEB for 2015 and 2030 as well. For 2030 two lightweight scenarios for the BEV were considered. The functional unit was 1 electric vehicle and 1 electric bus. A cradle to grave product system was considered, thus EOL stage is included. The employed impact assessment method is Recipe Hierarchist midpoint (Goedkoop et al., 2009) while the software employed was SimaPro v8.3.0.(PRé-Consultants, 2014). An attributional approach was adopted based on Ecoinvent v3.02 (Swiss Centre for Life Cycle Inventories, 2015) datasets for the BEV, whereas for BEB material composition we considered information from Garcia Sanchez et al.(2013). Finally, as a way to further contextualize this research in Brazilian conditions we included a use phase for the vehicles, considering a functional unit of 1km.
Boundaries and evolution parameters
The system boundaries for a LCA study determine which processes and activities the overall analysis includes, in this case the boundaries were outlined firstly to evaluate the manufacture stage of both BEVs and BEBs and secondly to analyze the phase use of the vehicles. Ecoinvent LCIs for BEVs can be found following a Unit Process scheme. This scheme creates a hierarchy in which a given process is composed of several inputs which in turn are made of other several inputs. It was our intention to model as much of the vehicle production chain as possible for both BEVs and BEBs by adapting those Unit Process datasets that model the BEVs and BEBs manufacturing stage. Since Ecoinvent V3.02 do not represent the specific Brazilian conditions in most cases the vast majority of the processes must be adapted. We focused on raw materials production, especially steel and aluminum, Brazilian electricity mix, and transportation. A detailed description of all adjusted Ecoinvent processes is shown in the Appendix. General parameters for vehicles evolution are found in Table 1. Electricity mix specifics are presented in electricity generation section.
Table 1. Parameters for 2015 and 2030 scenarios for BEVs and BEBs
Battery Electric Bus Battery Electric Vehicle
2015 scenario
Vehicle w/o battery 11,010 Kg 1,243 Kg
Battery mass 3,289.43 Kg 296 Kg including heaters
Vehicle performance 1.66 Wh Km-1 (1.50 kWh Km-1*90% efficiency) 167 Wh Km-1 (150 Wh Km-1* 90% efficiency)
Energy density 11.40 E-2 kWh Kg-1 10.14 E-2 kWh Kg-1
Electricity mix Year 2014. EPE (2016)
Year 2014. EPE (2016)
Maintenance 17% of materials for assembly stage.
17% of energy required for assembly Ecoinvent dataset: Maintenance, passenger car, electric, without battery, Alloc Def, U. One maintenance for 150,000 km
Life expectancy 220,000 Km Battery
880,000 Km Bus w/o battery 100,000 mi Battery
120,000 mi Car w/o battery
2030 scenario
Bus w/o battery 9.469 Kg Aluminum scenario 1.156,48 Kg
Plastic scenario 1.035,74 Kg
Battery mass 2.857,1 Kg 228.6 Kg
Energy density 16.0 E-2 kWh Kg -1 35.0 E-2 kWh Kg-1
Vehicle performance 1.33 kWh Km-1 (1,20 kWh Km-1*90% efficiency) 133.3 Wh Km-1 (120 Wh Km-1*90% efficiency
Electricity mix Forecast for 2030. EPE(2016)
Forecast for 2030. EPE (2016)
Maintenance 17% of materials for assembly stage.
17% of energy required for assembly Ecoinvent dataset: Maintenance, passenger car, electric, without battery, Alloc Def, U. One maintenance for 150,000 km
Life expectancy 264,000 Km Battery
1’056,000 Km Car w/o battery 120,000 mi Battery
144,000 mi Car w/o battery
Battery Electric Vehicle
BEV Ecoinvent dataset is originally based in Habermacher (2011), whose starting point of analysis was the material content of a Volkswagen Golf A4, as it was modeled by Althaus & Gauch (2010). Habermacher (2011) created a baseline scenario and two lightweight scenarios for car glider, aiming to model future mass reductions based on synthetic and aluminum material substitutions. Glider refers to a vehicle without powertrain and battery. Both, 2015 and 2030 scenarios present the same basic Unit Process structure seen in Figure 1. Both datasets, “Car without battery Alloc Def, U” and the “Li-ion battery, rechargeable, prismatic Alloc Def, U” are adapted to mass data for 30kWh Nissan Leaf 2016 Accenta, Black edition, Tekna (NISSAN, 2016b) as presented in Table 1. Nissan Leaf was chosen for being the second bestselling car in the world, hence we considered it to be representative of BEV market (Cleantechnica, 2017).
Figure 1. Unit process visual scheme for a BEV
2015 scenario
Performance, Maintenance & Life expectancy
Current performance for the BEV was considered as 150 Wh Km-1 (NISSAN, 2016b) while 90% charging efficiency was assumed (Zackrisson, Avellán, & Orlenius, 2010), thus tank-to-wheel consumption is 0.167 kWh/km. For maintenance information we used Ecoinvent dataset Maintenance, passenger car, electric, without battery Alloc Def, U. Quantity used per Km is the same as in dataset Transport, passenger car, electric Alloc Def, U. Regarding life expectancy, we discovered that 30 kWh Nissan LEAF Accenta is backed by a limited warranty providing 100,000 miles Lithium-Ion Battery coverage (NISSAN, 2016a), we assumed this value as being the total life of the battery. Life expectancy for the rest of the car is assumed to be 120,000 mi due to the fact that Nissan Leaf maintenance booklet keeps records until this mileage.
End of life
In Ecoinvent inventories EOL stage is included in each dataset. For instance, the car without battery dataset includes outputs for glass, lubricants and rubber from the tires. Both powertrain and glider contemplate EOL outputs too. We kept these outputs in the datasets unaltered.
2030 Scenario
Vehicles evolution followed two main assumptions: There will likely be a mass reduction and a material switching in BEV components for 2030. Both main datasets were analyzed, li-ion battery and the car without the battery.
Li-ion battery
Researchers have studied the challenge of making batteries, cheaper, stronger, lighter and presenting a lower environmental burden. Wood et al (2015) presented a study aiming to prove the technical feasibility of generating a cost breakdown for battery electrodes by redefining design parameters and adopting different materials. Berckmans et al. (2017) analyzed a 2030 scenario for decrease in battery production cost whereas considering an improvement in material science and maturing of the market.
Moreover, abundant research has been carried out on the technical field. Last 25 years saw a sharp development in technical parameters like energy and power storage, battery safety and charging time (Blomgren, 2017). Li-ion batteries must evolve in terms of fulfilling technical requirements, specially driving range and hillside performance. In order to overcome these issues, improvements in battery capacity, energy density and power density must be achieved. Battery capacity is the main parameter influencing BEVs range (Mruzek, Gajdáč, Kučera, & Barta, 2016). Paradoxically, battery weight and energy consumption tend to increase significantly when driving range increases (Campanari, Manzolini, & Garcia de la Iglesia, 2009).
Nissan Leaf Accenta-Tekna 30kWh Li-ion battery does have a weight of 296 Kg, including heaters (NISSAN, 2016b). Heaters are included in the car to avoid it suffer from significant driving range loss in subzero temperature environments because of reduced energy and power capability (Wiesenfelder, 2011). Ji & Wang (2013) classify the Nissan Leaf heater as mutual pulse system. This heating scheme exhibits substantial benefits such as not requiring additional moving parts, not having any participation of fluids and minimal extra weight requirements. Having said this, we assume that battery heater weight is negligible, which means Nissan Leaf battery has an energy density of 10.135E-2 kWh kg-1. Reportedly, current BEVs have typical battery capacities of around 30 kWh and it is assumed they will reach a 60 kWh average by 2030, meaning a driving range between 350 and 450 Km (IRENA, 2017a). A range of at least 300 km is believed to be a requirement for general public acceptance. In contrast, Blomgren (2017) points out that some models like Chevrolet Volt BEV and Tesla model 3 already reached 60 kWh. In view of this we assumed the battery capacity in 2030 for the Nissan Leaf to be 80 kWh.
According to Gondelach & Faaji (2012) in the medium term, it is expected that only Li-ion batteries will manage to reach a specific energy density level of 400 Wh kg-1, leaving others such as Li-S and Li-air batteries behind. Notwithstanding, more recent opinions show that skepticism is common among battery researchers, who believe that improvements to Li-ion cells may squeeze 30% more energy by weight in the best case scenario (Van Noorden, 2014; Zu & Li, 2011). It means that Li-ion technology might never even reach 400 Wh kg-1. In order to keep a conservative approach, we assumed an energy density of 350 Wh kg-1 which results in a battery weight of 228.57 Kg.
Rest of the car
BEV mass reduction can be accomplished by substituting heavy materials, mostly ferrous metals, by lighter ones. The substituting materials could be either lighter metals, such as aluminum; or synthetic materials, like composite structures. Habermacher (2011) constructed two lightweight scenarios which reflect the material inventories of two prototype cars: Precept and ESX2 (Tonn, Schexnayder, Peretz, Das, & Waidley, 2003). Precept does contain a large share of aluminum substituting heavier materials, whereas composite materials, mainly resins, are used intensively in ESX2. Thus, the material content of the two prototypes matches what is expected to be a trend for the next 20 years in automotive glider manufacture. In fact, fiber-reinforced composite materials have been used for many years for body panels in low-volume vehicles whereas aluminum has predominantly been used for engine blocks, wheels and gearboxes but is expected to migrate to other components as well (Ricardo AEA, 2015b). Henceforth precept will be addressed as aluminum glider and ESX2 as plastic glider. Plastic glider LCI includes carbon fiber as an input, since this dataset is absent from Ecoinvent we modelled it as Schmidt & Watson (2013) did in a LCA study for ferryboats.
Mass reduction scenarios for the glider are based on Precept and ESX2 total material content. For electric powertrain we resolved to keep 2015 datasets unmodified compared to future scenarios, mainly due to electric powertrain LCIs being calculated per piece for each subset, therefore their material content is not scalable linearly. Ecoinvent passenger car, electric, without battery Alloc Def, U dataset originally comprises of glider and powertrain shares per Kg. 1 Kg of car in 2015 (without battery) contains 9% of powertrain and 91% in weight. For the lightweight 2030 scenarios Habermacher (2011) contemplated a reduction in glider mass while powertrain mass has no changes. Glider and powertrain masses for our aluminum and plastic prototypes were adjusted based on percentages shown in Table 2.
Table 2. Mass reduction parameters for BEV glider and powertrain.
Parameter Baseline Plastic Aluminum
Glider (Kg) 838 453 461
Powertrain (Kg) 78 78 78
Total mass (Glider + Powertrain) (Kg) 916 531 539
Total mass reduction compared to B/line (%) – 57.96% 58.84%
Glider share in Total mass (%) 8.52 14.68 14.47
Mass scaled to Nissan Leaf (Kg) 1,243 720.44 731.38
End of life, Performance and maintenance
Datasets are kept as in 2015 scenario because of lack of specific data on EOL evolution stage. Vehicle performance was assumed to be 20% higher (1.20 Wh Km-1) when compared to 2015 (1.50 kWh Km-1) conditions due to mass reduction. Maintenance LCI is kept unmodified for 2030 scenario with regards to 2015 scenario.
Life Expectancy
Manufacturing improvements along with materials evolution make vehicle longevity likely to increase as engineering techniques evolve, assuming planned obsolescence will not hamper lifespan improvements.
With regards to li-ion battery, a word should be given to the fact that in spite of intensive research on various electrode chemistries, ageing phenomena are not yet neither well understood nor quantified, and the combined impacts of temperature, depth of discharge and current intensity still remain difficult to quantify and manage (Barcellona, Brenna, Foiadelli, Longo, & Piegari, 2015). For the rest of the car material switching merged with mass reduction make lifespan forecasting full of uncertainties. Aluminum and plastic pieces present lower resistance to fatigue than their steel counterparts, however, due to lower overall car weight components might experience lower stresses having as result its lifespan actually extended. Due to lack of data for the specific conditions for Precept and ESX2 or the Li-ion battery defined by Habermacher (2011) we decided to assume life expectancy increase as having a 20 % increase when compared to 2015 scenario. This way, battery and rest of the car lifespans are 120,000mi and 144,000mi respectively.
Battery Electric Bus
2015 Scenario
So far, mass deployment of BEBs is scarce. For instance, in Germany, diffusion stage (when mass production is reached) is anticipated to arrive only by 2030. Even at this point, only a hundred pure BEBs are expected to be deployed. Notwithstanding, important actors like Chinese BYD are investing in changing public transportation paradigms including life expectancy and battery performance (Henderson, 2016) (Avid Technology Group Ltd, 2016). In 2016 the first double-decker BEB in the world was unveiled in London. It will manage to travel 180 miles on a single charge. These initiatives are expected to become more common in the next decades.
For the electric bus a different methodology was adopted, since there was no available data in Ecoinvent databases to describe a BEB we appealed to literature. BEB material inputs for one bus, are available at García Sanchez et al. (2013). Energy required for assembly stage is included as well. The LCI in this study managed to disaggregate the LiFePO4 battery from the rest of the bus. It also includes information about EOL stage and maintenance. The authors based their calculations on a BYD e-bus (China Buses, 2015), which is one of the few commercially available pure electric buses. Bus constituents are separated by type of material, e.g. steel, glass, aluminum, etc. End-of-life stage is established according to the type of disposal process e.g. dismantling, car shredding, etc. Detailed information on BEB LCI can be found in the Appendix.
Iron phosphate battery (LiFePO4)
Due to lack of data for LiFePO4 batteries, we employed the Li-ion battery dataset available in Ecoinvent;” Battery, li-ion, rechargeable, prismatic, Alloc Def U”. Same thing for infrastructure datasets, due to lack of information, data for conventional bus plant was included to model a BEB plant. Although not desired, this solution is reasonable, since it allocates the same infrastructure burden for both BEB; Brazilian and Global. Among Li-ion batteries LiFePO4 is recognized for presenting one of the best power performances (Miller, 2015), regarded as very important asset when facing hilly upsides. In contrast LiFePO4 energy performance is far from good (IRENA, 2017b). BYD answer to range, reliability and economy constraints for a BEB uses this battery that weights more than 3 tons, in a move that was called a “brute force approach” (Avid Technology Group Ltd, 2016). BYD managed to make this technically and financially feasible because they are mainly a battery manufacturer. According to Garcia Sanchez et al (2013) this battery presents a specific energy of 11.4E-2 kWh Kg-1 and a weight of 3,289.43 Kg resulting in a battery capacity of 375.0 kWh. Although this specific energy is 14% and 23% higher than reported by Julien et al.(2016) and Sullivan et al.(2012) respectively, we still considered it as feasible since it fits within the ranges proposed by Hassuani et al.(2005): 12.0E-2 Wh Kg-1 to 15.0E-2 Wh Kg-1 and Zu & Li (2011) 7.0E-2 Wh Kg-1 to 14.0E-2 Wh Kg-1.
Rest of the bus
Life cycle inventories in García Sanchez et al. (2013), reported a bus weight of 14,300 Kg, including battery. It is also reported that the average value of primary energy consumption or assembly stage goes from 17,400 to 22,100 kJ/kg. It is assumed this energy comes from electricity and thermal energy (split 50/50 share). We assumed a mean value of 19,750 KJ/Kg. It is worth noting that this value does not include the LiFePO4 battery assembly considered by the authors. LCI for the rest of the bus are presented in Table 3.
Table 3. Inputs required for manufacturing one glider of a BYD e-bus
Datasets Mass
Steel, rolled | market for | Alloc Def, U 5,959.24 kg
Polypropylene, granulate | market for | Alloc Def, U 1,362.22 kg
Cast iron | market | Alloc Def, U 1,021.59 kg
Aluminium, primary ingot | market for | Alloc Def, U 1,305.30 kg
Flat glass, coated | market | Alloc Def, U 170.27 kg
Flat glass, uncoated | market | Alloc Def, U 397.25 kg
Steel, chromium steel 18/8 hot rolled | market for | Alloc Def, U 454.03 kg
Particle board, for indoor use | Alloc Def, U 0.33 m3
Lubricating oil | market | Alloc Def, U 113.54 kg
Road vehicle factory {GLO}| market for | Alloc Def, U 8.73E-07 p
Electricity 2016, medium voltage | market for | Alloc Def, U 108,729 MJ
Heat, district or industrial, other than natural gas{RoW}| market |Alloc Def, U 108,729 MJ
End of life stage
Data collection on energy expended and residues generated due to the processes involved was obtained from the GaBi 4 database (García Sánchez et al., 2013). A word should be given to mass balance between LCI inputs and EOL outputs. As expected, lubricants and LiFePO4 battery mass inputs match liquid drain and LiFePO4 battery outputs. In contrast, a direct mass correlation for other inputs and outputs is not possible to establish. In order to keep coherency, we adopted a cut-off model for EOL as we did for other stages, which means only first life of a product appears in the boundaries of the LCI. Cannibalization material is considered to be part of a second life; therefore, it is not represented in the outputs. LiFePO4 EOL is included in the battery dataset represented by a Li-ion used battery dataset. This because of the lack of specific data for iron phosphate batteries EOL. There is evidence that municipal waste incineration in Brazil is negligible when compared to total solid residues production, mainly due to technical and financial unfeasibility (Machado, 2015), hence, municipal waste incineration share was considered to end up in landfills completely. Moreover, for lubricants, we assumed that is more accurate to model the final disposal as a treatment instead of just dumping it in a landfill. Brazilian regulation is becoming increasingly stringent on lubricant oil disposal (Canchumani, 2013). Table 4 presents the EOL outputs and the Ecoinvent processes used to represent them.
Table 4.End-Of-Life outputs and Ecoinvent datasets used for modelling.
Output Mass (Kg) Ecoinvent process used
Liquid drain 113.54 Waste, mineral| Alloc Def, U
Cannibalization 110.11 Not included
Dismantling (Subtotal) 4019.73
Tires 214.64 Used Tyre| treatment of| Alloc Def, U
LiFePO4 battery 3289.43 Used Li-ion battery| Alloc Def, U
Metal scrap 396.11 Used powertrain from electric passenger car, manual dismantling| Alloc Def U
Synthetic 119.41 (Residue). Plastic Waste
Car shredding (subtotal) 10056.62
Metal scrap (steel, aluminum, copper) 8289.14 Used glider, passenger car| treatment of, shredding| Alloc Def U
Landfill (waste for municipal disposal) 440.01 (Residue). Waste, Industrial
Municipal waste (incineration) 1327.47 (Residue). Waste, Industrial
Special attention should be paid to battery EOL contrasts between BEB and BEV. García Sanchez et al. (2013) point out that for each 1Kg of used battery it is required treatment for 1Kg of battery input. In contrast, Ecoinvent BEV li-ion battery EOL dataset considers used Li-ion battery output as being 0,745 Kg per 1 Kg of battery. This mass difference is probably due to Ecoinvent cut-off model, which does not include any second-life interactions within the main LCI. Thus, about 25 % of the BEV battery is likely assumed to have a second destination.
Performance, life expectancy & Maintenance
In order to define BEB performance in 2015 scenario we used the same values as in Garcia Sanchez et al. (2013): 1,5 kWh km-1 and 90% charging efficiency (Zackrisson et al., 2010).
Bus lifetime is projected to be a key requirement for BEBs penetration (Thielmann, Isenmann, Martin, Plötz, & Sauer, 2013). Nevertheless, prospects were not optimistic until very recently. An study for hybrid buses (Kellaway, 2007) reviews the principal causes of battery failure in LiFePO4 batteries and states that the most common problem with poorly specified batteries in hybrid buses, is that actual battery lifetime could be way much shorter than expected. In some cases, suppliers predicted a life time of 2–3 years, but only lasted 2–3 months. Garcia Sanchez et al (2013) assumed a 220,000 km lifespan for batteries (around 2,5 years) and argue this problem will be overcome in the future, while for the rest of the bus, the authors assumed a lifetime of around 10 years (880,000 km)based on Diesel busses. No great differences were anticipated between BEB and Diesel buses for elements other than battery and powertrain, thus, it is a reasonable assumption.
For vehicle maintenance stage, an energy consumption of 17% of the total energy required for materials production and for assembly stage of the buses was assumed. This proportional value was obtained from the scientific study (John L. Sullivan et al., 1998) and its selection was due to the lack of data about electric bus maintenance inventories.
2030 Scenario
LiFePO4 battery
In order to keep a conservative approach, we assumed the specific energy for 2030 scenario to be 14,0E-2 Wh Kg-1 this assumption is coherent with the results presented by Hassuani et al.(2005) and Zu & Li (2011). Battery capacity was assumed to improve till 400 kWh. An increase of around 7 % compared to current scenario. These conditions would produce a battery weight of 2857.1 Kg.
Rest of the bus and End-of-Life.
Mass reduction (total weight 13,550 Kg) was based on coach bus lightweighting estimation data reported by Ricardo AEA(2015a). Although this data is based in an ICEV coach bus and not an BEB we consider this comparison as acceptable. Firstly, because of expected mass reduction in powertrain system (engine, fuel, exhaust and transmission system) accounting for less than 1 % of total mass reduction, which means that almost all mass reduction would be linked to glider weight reduction instead of powertrain elements. Secondly because glider structure of conventional busses and BEBs is deemed to be similar. Weight comparison between Ricardo AEA (2015a) coach -11,900 Kg without powertrain- and Sanchez Garcia et al (2013) bus -11,010 without battery- ensure similarity and reveals our assumptions are coherent. Weight reduction for 2030 is forecasted to be 14%. No evidence of material substitution for 2030, thus 2015 LCI input shares remain unaltered, only total mass varies. No evidence was found for considering EOL stage is going to change for bus process, hence, 2015 datasets are kept unmodified.
Life expectancy, maintenance and performance
Analogously to other types of li-ion batteries LiFePO4 longevity prediction is still full of uncertainties. Although many battery ageing mechanisms have been described in the literature, these phenomena are complex and are prone interact with each other, resulting in different capacity loss and power decline (Prada et al., 2012). For 2030 scenario and due to lack of specific data for our BEB we assumed a longevity increase in 20% for both LiFePO4 battery and the rest of the bus. Thusly, battery lifespan is 264,000 Km whereas the rest of the car is considered to present a life of 1’056,000Km.
Maintenance stage approach for 2030 scenario follows the same principle as for 2015. We considered maintenance stage throughout the entire life cycle to represent an energy consumption of 17% of total energy required for assembly stage and 17% of total bus materials for pieces substitution. Since BEB mass was reduced for 2030 scenario then maintenance dataset is adjusted too.
BEB performance was assumed to be 1,20 kWh Km-1, which imply an improvement of 20% with regards to 2015 values. It also keeps the same percental difference as BEVs in 2015 and 2030.
...(download the rest of the essay above)