Modeling¶

This document describes the carculator_truck model, assumptions and inventories as exhaustively as possible.

carculator_truck is an open-source Python library. Its code is publicly available via its Github repository. You can also download an examples notebook, that guides new users into performing life cycle analyses.

Overview of carculator_truck modules¶

The main module model.py builds the vehicles and delegates the calculation of motive and auxiliary energy, noise, abrasion and exhaust emissions to satellite modules. Once the vehicles are fully characterized, the set of calculated parameters are passed to inventory.py which derives life cycle inventories and calculates life cycle impact indicators. Eventually, these inventories can be passed to export.py to be exported to various LCA software.

Vehicle modelling¶

The modelling of vehicles along powertrain types, time and size classes is described in this section. It is also referred to as foreground modelling.

Powertrain types¶

carculator_truck can model the following powertrain types:

• Diesel-run internal combustion engine vehicle (ICEV-d)

• Gas-run internal combustion engine vehicle (ICEV-g)

• Diesel-run hybrid electric vehicle (HEV-d)

• Diesel-run plug-in hybrid electric vehicle (PHEV-d)

• Battery electric vehicle (BEV) with charging at depot

• Fuel cell electric vehicle (FCEV)

Size classes¶

Several size classes are available for each powertrain type. They refer to the maximum permissible gross weight of the vehicle (e.g., 32 tons). In addition, several application-specific designs are available for each powertrain-size class combination, namely: Urban delivery, Regional delivery and Long haul. They are associated with a given range autonomy: 150 km, 400 km and 800 km, respectively. This is particularly relevant for sizing the onboard energy storage unit. Some powertrain-size class-application combinations are not commercially available or technologically mature and are therefore not considered.

Battery electric vehicles are not considered for years prior to 2020. Additionally, carculator_truck may not find a solution for regional delivery and long haul use in 2020, as the volumetric density of batteries does not currently allow a range autonomy superior to 400 km without significantly sacrificing the cargo carrying capacity.

carculator_truck defines seven size classes, namely:

• 3.5t

• 7.5t

• 18t

• 26t

• 32t

• 40t

• 60t

Example of 3.5t truck, rigid, 2 axles, box body and 7.5t truck, rigid, 2 axles, box body

Example of 18t truck, rigid, 2 axles, box body and 26t truck, rigid, 3 axles, box body

Example of 32t truck, semi-trailer, 2+3 axles, curtain-sider and 40t truck, tipper-trailer, 2+4 axles

Example of 60t truck, semi-trailer + trailer, 2+4+2 axles, curtain-sider

Manufacture year and emission standard¶

For ICE vehicles, several emission standards are considered. For simplicity, it is assumed that the vehicle manufacture year corresponds to the registration year. Those are presented in Table 1.

Modelling considerations applicable to all vehicle types¶

Sizing of the base frame¶

The sizing of the base frame is based on p. 17-19 of [Nikolas and others, 2015]. Detailed weight composition is obtained for a 12t rigid truck and a 40t articulated truck. Curb mass and payload are obtained for all size classes, the rest being adjusted function of the gross mass. The masses of the vehicles and their subsystems are detailed in Table 2. These truck models have 2010 as baseline year. A 2% and 5% weight reduction factors are applied on on rigid and articulated trucks respectively, as indicated in the same report.

The following components are common to all powertrains:

• Frame

• Suspension

• Brakes

• Wheels and tires,

• Electrical system

• Transmission

• Other components

Other use and size-related parameters¶

HBEFA 4.1 is used as a source to estimate the calendar and kilometric lifetime values for European diesel trucks. Those are presented in Table 3.

Average loads for European trucks for long haul use are from the TRACCS road survey data for the EU-28 [G. and others, 2013]. We differentiate loads across driving cycles. To do so, we use correction factors based on the representative loads suggested in the Annex I of European Commission regulation 2019/1242. Such average loads are presented in Table 4.

The user can however easily change these values.

Other size-related parameters are listed in Table 5. Some of them have been obtained and/or calculated from manufacturers’ data, which is made available in the Annex A-C of this report.

The user can however easily change these values.

Time-dependent parameters¶

Several parameters that affect the performances of trucks (e.g., drag coefficient, etc.) are time-dependent, and based on various projections found in the literature. Table 6 lists some of them.

Modelling approach applicable to internal combustion engine vehicles¶

Traction energy¶

The traction energy for medium- and heavy-duty trucks is calculated based on the driving cycles for trucks provided by VECTO. Simulations were run in VECTO during the original study to obtain reference speed and performance profiles. A current carculator_truck run reads the bundled profiles; it does not invoke VECTO software.

The calculation of the total resistance to overcome at the wheel level is the sum of the following resistances:

• The vehicle inertia, calculated as acceleration * driving mass

• The rolling resistance, calculated as driving mass * rolling resistance coefficient * gravity

• The aerodynamic drag, calculated as frontal area * aerodynamic drag coefficient * air density * speed^2 / 2

• The gradient resistance, calculated as driving mass * gravity * sin(gradient)

Negative total wheel power indicates braking or deceleration. It is not an extra resistance added to the inertial force.

Figure 1 shows the contribution of each type of resistance as calculated by carculator_truck for the first hundred seconds of the “Urban delivery” driving cycle, for an 18t diesel truck.

Figure 2 shows the first two hundred seconds of the “Urban delivery” driving cycle. It distinguishes the target speed from the actual speed managed by the different vehicles. The power-to-mass ratio influences the extent to which a vehicle manages to comply with the target speed.

The optional rated power input fixes mechanical shaft output in kW; zero or an absent parameter retains power-to-curb-mass sizing. Explicit power constructor overrides take priority. The small FCEVs use documented 2020 motor-only proxies: 57 kW for 3.5 t from the Renault Master Z.E. and 129 kW for 7.5 t from the FUSO eCanter. These electric-commercial-vehicle ratings are transferred to FCEVs; they are not complete FCEV calibrations. The default reference ratings are retained after 2020 without assuming motor downsizing. Their mass, fuel-cell/buffer allocation, storage and consumption are recalculated in the native sizing loop. Both complete the Regional-delivery trace at default payload. The 57 kW design still cannot complete the default 800 km Long-haul configuration; different payloads, grades or driving cycles require a separate feasibility check. No wheel demand or fuel input is clipped to hide a power shortfall.

For regional delivery and long haul use, the “Regional delivery” and “Long haul” driving cycles of VECTO are used, respectively. They contain less stops and fewer fluctuations in terms of speed levels. The “Long haul” driving cycle has a comparatively higher average speed level and lasts much longer. Figure 3 shows the first two hundred seconds of the “Long haul” driving cycle.

Table 7 shows a few parameters about the three driving cycles used in the original study. Its padded durations and average speeds predate the cycle-duration correction and must not be used as current cycle statistics. For example, the corrected long-haul profile has 5,454 active seconds and covers 108.191 km, averaging about 71.4 km/h including stops. See the shared truck cycle diagnostics.

The energy consumption model is similar to that of passenger cars: different resistances at the wheels are calculated, after which friction-induced losses along the drivetrain are considered to obtain the energy required at the tank level.

VECTO’s simulations are used to calibrate the engine and transmission efficiency of diesel trucks. Similar to the modeling of buses, the relation between the efficiency of the drivetrain components (engine, gearbox) and the power load-to-peak-power ratio is used.

Indeed, once the power requirement at the wheel level for each second is known (and validated), inefficiencies from the transmission line and the engine need to be accounted for. Here again, second-by-second data from VECTO simulations are used. VECTO uses a complex gearshift model combined with an engine-specific torque map that are too complex to be implemented in carculator_truck. Instead, the relation between transmission and engine efficiency on one end, and the relative power load (i.e., power load over the rated power output of the engine) on the other end, is used. Such relations are shown in Figure 4, for a 40t diesel truck, where the efficiency of the drivetrain (left) and engine (right) in relation to the power load is plotted for each second of the “Urban delivery” driving cycle, with a loading factor of 100%. For example, Figure 4.a shows that the transmission efficiency (that is, from the wheels to the output shaft of the engine) is close to 85% at a power load of 20%. In fact, most of the time when the truck is driving, the transmission operates at above 80% efficiency. Similarly, Figure 4.b shows that the peak engine efficiency is reached at about 40% power load, after which it remains more or less constant. A curve is fitted on the data points (red line). Using such fit removes some of the complexity considered in VECTO, depicted here by the measurements that deviate for the red curve. Nevertheless, it allows obtaining a reasonable estimate of the efficiency of these drivetrain components.

Relation between transmission and engine efficiency on one end, and the relative power load on the other end

Such calibration exercise with VECTO for the diesel-powered 40t truck is shown below, against the “Urban delivery” driving cycle. The original exercise reported less than 1 percent difference over that cycle. This is historical simulator agreement, not an independent measured validation or a guarantee for the revised model and every truck configuration.

The original study did not use a VECTO model for compressed gas-powered trucks. The calibrated model for diesel-powered buses is used and a penalty factor of 10% is applied, based on findings from a working paper from the ICCT [Ragon and Rodríguez, 2021] showing that compressed gas-powered trucks have an engine efficiency between 8 to 13% lower than that of diesel-powered trucks.

Engine downsizing¶

Such approach allows also reflecting the effect of engine downsizing. As the relative power load observed during the driving cycle is higher as the rated maximum power output of the engine is reduced, it operates at higher efficiency levels. Figure 6 compares the engine efficiency between a conventional 40t diesel truck and a diesel hybrid truck of similar size, but where the power of the combustion engine is reduced by 25% in favor of an electric motor. This figure confirms that the combustion engine of hybrid-diesel truck (HEV-d) reaches higher efficiency levels. Of course, the difference in efficiency will be more pronounced on driving cycles with transient loads.

Compressed gas trucks¶

For compressed gas trucks, the energy storage is in a four-cylinder configuration, with each cylinder containing up to 57.6 kg of compressed gas – 320 liters at 200 bar.

The relation between the mass of compressed gas and the cylinder mass is depicted in Figure 7. This relation is based on manufacturers’ data – mainly from [Trucks, n.d., QTWW, 2021].

Inventories for a Type II 200 bar compressed gas tank, with a steel liner, are from [Daniele and others, 2021].

Exhaust emissions¶

Other pollutants¶

Emission factors for CO₂ and SO₂ are detailed in Table 8 - Table 9. Biofuel shares in the fuel blend are detailed in Table 10.

A number of fuel-related emissions other than CO₂ or SO₂ are also considered.

For trucks, two sources source of emissions are considered:

• Exhaust emissions: emissions from the combustion of fuel during operation. Their concentration relates to the fuel consumption and the emission standard of the vehicle.

• Non-exhaust emissions: abrasion emissions such as brake, tire and road wear, but also emissions of refrigerant and noise.

For exhaust emissions, factors based on the fuel consumption are derived by comparing emission data points for different traffic situations (i.e., grams emitted per vehicle-km) in free-flowing driving conditions, with the fuel consumption corresponding to each data point (i.e., MJ of fuel consumed per km), as illustrated in for a diesel-powered engine. The aim is to obtain emission factors expressed as grams of substance emitted per MJ of fuel consumed, to be able to model exhaust emissions of trucks of different sizes, masses, operating on different driving cycles and with different load factors.

Using these fuel-based emission factors, emissions for each second of the driving cycle for each substance are calculated.

To confirm that such approach does not yield kilometric emissions too different from the emission factors per vehicle-kilometer proposed by HBEFA 4.1, Figure 9 compares the emissions obtained by carculator_truck using VECTO’s “Urban delivery” driving cycle over 1 vehicle-km (red dots) for a 18t rigid truck with the distribution of the emission factors across different “urban” traffic situations (green box-and-whiskers) given by HBEFA 4.1, as well as its weighted average (yellow dots) for different emission standards for a rigid truck with a gross mass of 14-20 tons.

There is some variation across HBEFA’s urban traffic situations, but the emissions obtained remain, for most substances, within the 50% of the distributed HBEFA values across traffic situations. Special attention must be paid to EURO-III vehicles, for which emissions tend to be slightly over-estimated by carculator_truck. The comparison between the model’s emission results for the regional and long-haul driving cycles using trucks of different size classes and HBEFA’s emission factors for “rural” and “motorway” traffic situations shows a similar picture.

Modelling approach applicable to electric vehicles¶

Traction energy¶

Electric vehicles¶

VECTO does not have a model for battery or fuel cell electric buses that can be used. Therefore, similarly to the modeling of buses, static engine and drivetrain efficiency values are used. These values are based on [Schwertner and Weidmann, 2016] and are presented in Table 9-Table 10.

Energy storage¶

Non-plug-in hybrid traction buffers¶

From 2020, HEV-d has a lithium-ion traction buffer distinct from the lead-acid starter. Earlier model versions credited regenerative braking but supplied no traction battery. Capacity is now sized inside the mass, power and fuel iteration. It must support the existing electric-motor design rating and the driving cycle’s usable energy requirement. The resulting charging power and usable capacity limit regeneration; excess braking energy receives no fuel or brake-wear credit. The controller reuses stored energy at the next positive propulsion opportunity and reaches a periodic energy balance, without grid or engine charging. It is an engineering energy-balance approximation, not a calibrated hybrid supervisory controller.

The default is a high-power LFP profile available before 2020: A123 Systems AHR32113M1Ultra-B, 3.3 V, 4.5 Ah and 205 g per cell. The manufacturer’s 2014 Battery Pack Design, Validation, and Assembly Guide, document 493005-002 Rev 02, Tables 4 and 30–33, provides the cell specifications and current limits. Continuous charging/discharging limits at 25 degrees C are 33/200 A. Multiplying these by the 2 V discharge cutoff gives conservative cell power limits of 66/400 W. Short-pulse ratings are not applied continuously. A conditioned pack is assumed; low-temperature derating, thermal conditioning demand, ageing of resistance and series/parallel electrical layout are not separately modelled.

The engineering assumptions are a 60% usable SOC window and a 70% cell mass share, giving about 19.7 kg per nominal kWh. hybrid_battery={...} overrides the cell power, energy, mass and pack-envelope assumptions (see HybridBatteryProfile). An explicit energy_storage['capacity'] retains the requested capacity if it can supply the motor rating; additional braking curtailment is permitted. Explicit alternative chemistries retain their physical properties, with the high-power electrical envelope disclosed as a transfer assumption in hybrid_battery_profile_transfers.

Cycle-life values retain the existing chemistry defaults (2020 LFP: 7000 nominal-capacity equivalent full cycles). They are not manufacturer guarantees for these high-rate conditions. Replacement demand is lifetime distance times actual stored-energy discharge per km, divided by nominal capacity and cycle life, minus the initial pack, bounded to 0–3. There is no separate calendar-age model. Supply, disposal and replacement costs use the same count. The starter is separate in vehicle mass and inventory, with the inherited five-year life and at least one initial battery. Other truck types retain their existing starter-inventory assumptions.

Existing chemistry-specific prices are extended to HEVs as explicit cost proxies (2020 LFP: 90 EUR/kWh), not quotations for high-power packs. Users can override them through battery_costs, including an explicit zero. Future native model years retain this cell envelope; it is not a forecast of high-power cell performance. Premise’s separately generated scenarios must also be distinguished from native future-year runs. The Premise adapter uses the user-selected MIX capacity market for nominal kWh and replacements and documents its chemistry and mass mismatch with this physical pack.

Battery electric trucks¶

Battery electric vehicles can use different battery chemistry (Li-ion NMC, Li-ion LFP, Li-ion NCA and Li-LTO) depending on the manufacturer’s preference or the location of the battery supplier. Unless specified otherwise, all battery types are produced in China, as several sources, among which BloombergNEF [Henze, 2020], seem to indicate that more than 75% of the world’s cell capacity is manufactured there. Accordingly, the electricity mix used for battery cells manufacture and drying, as well as the provision of heat are assumed to be representative of the country (i.e., the corresponding providers are selected from the LCI background database). The battery-related parameters considered in this study are shown in Table 11. For LFP batteries, “blade battery” or “cell-to-pack” battery configurations are considered, as introduced by CATL [Xinhua, 2019] and BYD [Mark, 2020], two major LFP battery suppliers in Asia. This greatly increases the cell-to-pack ratio and the gravimetric energy density at the pack level.

Overall, the gravimetric energy density values at the cell and system levels presented in Table 11 are considered conservative: some manufacturers perform significantly better than the average, and these values tend to change rapidly over time, as it is being the focus of much R&D.

The sizing of energy storage for BEV trucks is sensitive to the required range autonomy, which is specific to each driving cycle (or defined by the user).

The expected battery lifetime (and the need for replacement) is based on the battery expected cycle life, based on theoretical values given by [Dietmar and others, 2018] as well as some experimental ones from [Yuliya and others, 2020]. Although the specifications of the different battery chemistry are presented in Table 11, they are also repeated in Table 12.

The historical NMC-622 example denotes a nickel:manganese:cobalt atomic ratio of 6:2:2, not an elemental mass ratio. Current truck defaults use NMC-622 at 2020, NMC-811 at 2025 and 2030, and NMC-955 from 2035. Between the listed milestones, the truck hook uses the closest earlier chemistry, rather than interpolating chemistry labels. Explicit selections override these defaults.

For trucks, battery replacement demand is derived from lifetime energy use:

\(L_{vehicle}\) is lifetime distance in km, \(F_{ttw}\) stored-energy use in kJ/km, \(C_{battery}\) capacity in kWh and \(N_{cycles}\) battery cycle life. This factor applies to charger-equipped vehicles and is evaluated after battery capacity and driving energy have converged. Each zero supplied input requests this calculation; nonzero supplied inputs are retained separately for each vehicle, year and sample. A calculated result does not become an override when the model is run again. Fuel-cell replacements likewise preserve supplied values per cell and otherwise use the final operating-hours calculation. The battery factor can be fractional and can be zero. There is no mandatory one-pack replacement in the truck calculation. Bus defaults use a different, explicitly documented assumption.

For plug-in hybrids, the electric driving share multiplies lifetime energy before subtracting the initial pack and applying the replacement cap:

Here \(u_e\) is electric driving share and \(F_{ttw,e}\) is the electric mode’s net stored-energy use. Both mode costs and the combined PHEV retain this one physical pack’s count, without a second driving-share weighting. Nonzero explicit counts on the PHEV or its intermediates apply to that physical pack; conflicting counts are rejected. Cycle life uses nominal-capacity equivalent full cycles, without an additional depth-of-discharge divisor.

This calculation retains a net-energy approximation: it does not count the extra discharge associated with regenerative cycling, including combustion-mode battery use. It also has no separate calendar-age limit. These limitations must be considered when interpreting a zero calculated replacement count.

Replacement production is allocated to the vehicle, with no second-life credit. The following table preserves the original study’s configurations; inspect the completed model’s battery lifetime replacements for a current case.

Plugin hybrid trucks¶

The number of commercial models of plugin hybrid trucks is limited. In this study, plugin hybrid trucks are mostly modeled after Scania’s PHEV tractor [Scania, 2020]. It comes with three 30 kWh battery packs, giving it a range autonomy in battery-depleting mode of 60 km, according to the manufacturer. These specifications in terms of battery capacity are used to model plugin hybrid trucks of different size classes (i.e., roughly based on their respective gross mass).

Knowing the vehicle battery storage capacity and its tank-to-wheel efficiency when powered on battery, it is possible to calculate its resulting range autonomy in battery-depleting mode. Furthermore, it is assumed that, in the context of urban delivery, the truck is used in battery-depleting mode in priority, resorting the combustion mode to complete the driving cycle (i.e., 150 km). This approach is used to calculate the electric utility factor for these vehicles. Energy storage capacities and electric utility factors for plugin hybrid trucks are described in Table 14.

Fuel cell electric trucks¶

All fuel cell electric vehicles use a proton exchange membrane (PEM)-based fuel cell system.

Table 15 lists the specifications of the fuel cell stack and system used in carculator_truck. The durability of the fuel cell stack, expressed in hours, is used to determine the number of replacements needed – the expected kilometric lifetime of the vehicle as well as the average speed specified by the driving cycle gives the number of hours of operation. The environmental burden associated with the manufacture of spare fuel cell systems is entirely allocated to vehicle use as no reuse channels seem to be implemented for fuel cell stacks at the moment.

The energy storage unit of fuel cell electric trucks is sized based on the required amount of hydrogen onboard (defined by the required range autonomy). The relation between hydrogen mass and tank mass is derived from manufacturers’ specifications, as shown in Figure 10.

We start from the basis that fuel cell electric trucks are equipped with 650 liters cylinders, which contain 14.4 kg hydrogen at 700 bar, for a (empty) mass of 178 kg. Hence, the requirement in term of tank mass for a long haul fuel cell electric truck that needs 74 kg of hydrogen is 0.1916² + 14.586*14.4 + 10.8 * (74/14.4) = 1’068 kg, excluding the hydrogen mass.

The hydrogen tank is of type IV, a carbon fiber-resin (CF) composite-wrapped single tank system, with an aluminium liner capable of storing 5.6 kg usable hydrogen, weighting 119 kg per unit (of which 20 kg is carbon fiber), which has been scaled up to 178 kg for a storage capacity of 14.4 kg to reflect current models on the market [Quantum, 2019]. The inventories are originally from [Thanh and others, 2010]. The inventories for the supply of carbon fiber is from [Alicia and others, 2021]. Note that alternative hydrogen tank designs exist, using substantially more carbon fiber (up to 70% by mass): this can potentially impact end-results as carbon fiber is very energy-intensive to produce.

Charging stations¶

The depot charger reference inventory and current allocation inputs are summarized in Table 16. BEVs and the electric operation of PHEVs receive a share of charger production according to grid electricity use, depot share, charger lifetime and annual fleet throughput. The latter is limited by both fleet demand and charger capacity. See Depot charger inventory allocation for the equation, hybrid handling, boundaries and completed model/inventory verification.

These are allocation assumptions for shared depot service. Charger production is spread over its own lifetime; the truck lifetime determines how much service is allocated to that vehicle. The 0.94 factor limits throughput and is not applied as a second electricity loss. Public-charger production is not separately included. The former 24-year, 2,100-kWh/day description does not represent the current input defaults.

Finding solutions¶

Truck sizing converges driving mass, available payload, energy-battery mass, fuel mass and TtW energy together. Each active vehicle, year and sample must change by no more than a relative tolerance of 1e-6 (0.0001%), with a small absolute tolerance for values near zero. max_iterations limits the number of attempts; nonfinite values or nonconvergence raise an error identifying the affected configuration. A fixed mass or a converged fleet total alone is not sufficient.

Cargo is limited to the available payload during sizing. Each iteration starts from the requested load, so capacity released by a lighter powertrain remains available. Once sizing converges, the model refreshes the energy trace at the final driving mass before calculating consumption, costs and direct emissions. Stored energy and the requested range agree within the sizing tolerance when storage is sized for range; an explicit capacity override remains an override.

The model checks actual driving mass against gross mass. The former description of a mandatory 10% residual cargo capacity is not an enforced check in the current code. The size class denotes gross vehicle mass, not the mass of cargo. Convergence establishes numerical stability; it does not prove that the assumed battery or duty cycle is commercially feasible.

Validation¶

Diesel trucks¶

Figure 11 compares the fuel economy of trucks of different size classes modeled by carculator_truck with those found in HBEFA and ecoinvent v.3.

Battery electric trucks¶

Table 17 compares some of the modeled parameters for battery electric trucks with the specifications of some commercial models disclosed by manufacturers. These manufacturers’ specifications can also be found in Annexes.

Fuel cell electric trucks¶

Inventory modelling¶

Once the vehicles are modeled, the calculated parameters of each of them is passed to the inventory.py calculation module to derive inventories. When the inventories for the vehicle and the transport are calculated, they can be normalized by the kilometric lifetime (i.e., vehicle-kilometer) or by the kilometric multiplied by the passenger occupancy (i.e., passenger-kilometer).

Road demand¶

The demand for construction and maintenance of roads and road-related infrastructure is calculated on the following basis:

• Road construction: 5.37e-7 meter-year per kg of vehicle mass per km.

• Road maintenance: 1.29e-3 meter-year per km, regardless of vehicle mass.

The driving mass of the vehicle consists of the mass of the vehicle in running condition (including fuel) in addition to the mass of passengers and cargo. Driver/passenger and cargo masses come from the selected vehicle inputs. For freight comparisons, tkm divides impacts by cargo tonnes, not by gross vehicle mass or passenger count.

The demand rates used to calculate the amounts required for road construction and maintenance (based on vehicle mass per km and per km, respectively) are taken from [Spielmann and Scholz, 2005].

Because roads are maintained by removing surface layers older than those that are actually discarded, road infrastructure disposal is modeled in ecoinvent as a renewal rate over the year in the road construction dataset.

Fuel properties¶

For all vehicles with an internal combustion engine, carbon dioxide (CO₂) and sulfur dioxide (SO₂) emissions are calculated based on the fuel consumption of the vehicle and the carbon and sulfur concentration of the fuel observed in Switzerland and Europe. Sulfur concentration values are sourced from HBEFA 4.1 [Benedikt and others, 2019]. Lower heating values and CO₂ emission factors for fuels are sourced from p.86 and p.103 of [for the Environment, 2021]. The fuel properties shown in Table 19 are used for fuels purchased in Switzerland but should be applicable for other areas/countries.

Exhaust emissions¶

Emissions of regulated and non-regulated substances during driving are approximated using emission factors from HBEFA 4.1 [Benedikt and others, 2019]. Emission factors are typically given in gram per km. Emission factors representing free flowing driving conditions and urban and rural traffic situations are used. Additionally, cold start emissions as well as running, evaporation and diurnal losses are accounted for, also sourced from HBEFA 4.1 [Benedikt and others, 2019].

For vehicles with an internal combustion engine, the sulfur concentration values in the fuel can slightly differ across regions - although this remains rather limited within Europe. The values provided by HBEFA 4.1 are used for Switzerland, France, Germany, Austria and Sweden. For other countries, values from [Miller and Jin, 2019] are used.

The amount of sulfur dioxide released by the vehicle over one km [kg/km] is calculated as:

where \(r_{S}\) is the sulfur content per kg of fuel [kg S/kg fuel], \(F_{fuel}\) is the fuel consumption of the vehicle [kg/km], and \(64/32\) is the ratio between the molar mass of SO2 and the molar mass of sulfur (S).

Country-specific fuel blends are sourced from the IEA’s Extended World Energy Balances database [(IEA), 2021]. By default, the biofuel used is assumed to be produced from biomass residues (i.e., second-generation fuel): fermentation of crop residues for bioethanol, esterification of used vegetable oil for biodiesel and anaerobic digestion of sewage sludge for bio-methane.

For exhaust emissions, factors based on the fuel consumption are derived by comparing emission data points for different traffic situations (i.e., grams emitted per vehicle-km) for in a free flowing driving situation, with the fuel consumption corresponding to each data point (i.e., MJ of fuel consumed per km), as illustrated in Figure 12 for a diesel-powered engine. The aim is to obtain emission factors expressed in grams of substance emitted per MJ of fuel consumed, to be able to model emissions of passenger cars of different sizes and fuel efficiency and for different driving cycles.

Hence, the emission of substance i at second s of the driving cycle is calculated as follows:

where:

• \(E(i,s)\) is the emission of substance i at second s of the driving cycle,

• \(F_ttw(s)\) is the fuel consumption of the vehicle at second s,

• and \(X(i, e)\) is the emission factor of substance i in the given driving conditions.

NMHC speciation¶

After NMHC emissions are quantified, EEA/EMEP’s 2019 Air Pollutant Emission Inventory Guidebook provides factors to further specify some of them into the substances listed in Table 22.

Non-exhaust emissions¶

A number of emission sources besides exhaust emissions are considered. They are described in the following sub-sections.

Engine wear emissions¶

Metals and other substances are emitted during the combustion of fuel because of engine wear. These emissions are scaled based on the fuel consumption, using the emission factors listed in Table 23, sourced from [Agency, 2019].

Abrasion emissions¶

We distinguish four types of abrasion emissions, besides engine wear emissions:

• brake wear emissions: from the wearing out of brake drums, discs and pads

• tires wear emissions: from the wearing out of rubber tires on the asphalt

• road wear emissions: from the wearing out of the road pavement

and re-suspended road dust: dust on the road surface that is re-suspended as a result of passing traffic, “due either to shear forces at the tire/road surface interface, or air turbulence in the wake of a moving vehicle” [Beddows and Harrison, 2021].

[Beddows and Harrison, 2021] provides an approach for estimating the mass and extent of these abrasion emissions. They propose to disaggregate the abrasion emission factors presented in the EMEP’s 2019 Emission inventory guidebook [Agency, 2019] for two-wheelers, passenger cars, buses and heavy good vehicles, to re-quantify them as a function of vehicle mass, but also traffic situations (urban, rural and motorway). Additionally, they present an approach to calculate re-suspended road dust according to the method presented in [EPA, 2011] - such factors are not present in the EMEP’s 2019 Emission inventory guidebook - using representative values for dust load on European roads.

The equation to calculate brake, tire, road and re-suspended road dust emissions is the following:

With:

• \(EF\) being the emission factor, in mg per vehicle-kilometer

• \(W\) being the vehicle mass, in tons

• \(b\) and \(c\) being regression coefficients, whose values are presented in Table 24.

The respective amounts of brake and tire wear emissions in urban, rural and motorway driving conditions are weighted, to represent the driving cycle used. The weight coefficients sum to 1 and the coefficients considered are presented in Table 25. They have been calculated by analyzing the speed profiles used in the original study. The table is a historical summary, not a current list of all cycle-dependent coefficients. Current two-wheeler energy calculations also use driving cycles; their earlier use of owner-reported consumption should not be inferred from this table.

Finally, for electric and (plugin) hybrid vehicles (with the exception of two-wheelers), the amount of brake wear emissions is reduced. This reduction is calculated as the ratio between the sum of energy recuperated by the regenerative braking system and the sum of negative resistance along the driving cycle. The logic is that the amount of negative resistance that could not be met by the regenerative braking system needs to be met with mechanical brakes.

The sum of PM 2.5 and PM 10 emissions is used as the input for the ecoinvent v.3.x LCI datasets indicated in Table 27.

Finally, we assume that the composition of the re-suspended road dust is evenly distributed between brake, road and tire wear particles.

Figure 13 below shows the calculated abrasion emissions for trucks in mg per vehicle-kilometer, following the approach presented above. These amounts will differ across driving cycles. For example, the amount of brake wear emissions is higher for the urban delivery cycle than for the regional delivery cycle, because the urban delivery cycle has a higher share of braking events.

Refrigerant emissions¶

The use of refrigerant for onboard air conditioning systems is considered for trucks until 2021. The supply of refrigerant gas R134a is accounted for. Similarly, the leakage of the refrigerant is also considered. For this, the calculations from [P. and others, 2016] are used. Such emission is included in the transportation dataset of the corresponding vehicle. The overall supply of refrigerant amounts to the initial charge plus the amount leaked throughout the lifetime of the vehicle, both listed in Table 28. This is an important aspect, as the refrigerant gas R134a has a Global Warming potential of 2’400 kg CO₂-eq./kg released in the atmosphere.

Noise emissions¶

Noise emissions along the driving cycle of the vehicle are quantified using the method developed within the CNOSSOS project [Kephalopoulos and others, 2012], which are expressed in joules, for each of the 8 octaves. Rolling and propulsion noise emissions are quantified separately.

The sound power level of rolling noise is calculated using:

With:

• \(v_m\) being the instant speed given by the driving cycle, in km/h

• \(v_{ref}\) being the reference speed of 70 km/h

• and \(A_{P,i,m}\) and \(B_{P,i,m}\) are unit-less and given in Table 29.

The propulsion noise level is calculated using:

With:

• \(v_m\) being the instant speed given by the driving cycle, in km/h

• \(v_{ref}\) being the reference speed of 70 km/h

• and \(A_{P,i,m}\) and \(B_{P,i,m}\) are unit-less and given in Table 29.

A correction factor for battery electric and fuel cell electric vehicles is applied, and is sourced from [Pallas and others, 2016]. Also, electric vehicles are added a warning signal of 56 dB at speed levels below 20 km/h. Finally, hybrid vehicles are assumed to use an electric engine up to a speed level of 30 km/h, beyond which the combustion engine is used.

The total noise level (in A-weighted decibels) is calculated using the following equation:

The total sound power level is converted into Watts (or joules per second), using the following equation:

The total sound power, for each second of the driving cycle, is then distributed between the urban, suburban and rural inventory emission compartments.

Typically, propulsion noise emissions dominate in urban environments, thereby justifying the use of electric vehicles in that regard. Rolling noise become dominant above 50 km/h. The sound power [W] over time is expressed in joules [or W.s] over the course of the driving cycle.

The study from [Stefano and others, 2019] provides compartment-specific noise emission characterization factors against midpoint and endpoint indicators - expressed in Person-Pascal-second and Disability-Adjusted Life Year, respectively.

Electricity mix calculation¶

Electricity generation shares are averaged over each vehicle and sample’s operating lifetime, assuming equal annual distance. The default uses national Ember observations followed by GECO 2025 Reference projections. Other GECO pathways and a TYNDP 2026 option can be selected explicitly. These generation choices are separate from the background-impact scenario. In particular, scenario="static" does not freeze national generation shares.

The operating lifetime is lifetime distance divided by annual distance; the calculation retains whole years and uses at least one year. Refreshed generation shares are held at their final values after 2070. Background technology impact coefficients are selected for the manufacturing year, with their own time horizon; this is not a year-by-year dynamic LCIA calculation.

See the shared electricity guide for geographic coverage and scenario assumptions, and its lifetime calculation for averaging and custom mixes.

Historical illustration: the Poland table below uses TYNDP 2020 and the original study’s lifetime assumptions. It illustrates averaging but does not contain the current default generation mix or current climate coefficients.

Inventories for fuel pathways¶

A number of inventories for fuel production and supply are used by carculator_truck. They represent an update in comparison to the inventories used in the passenger vehicles model initially published by [Brian and others, 2020]. The fuel pathways presented in Table 32 are from the literature and not present as generic ecoinvent datasets.

Inventories for energy storage components¶

The source for the inventories used to model energy storage components are listed in Table 33.

Life cycle impact assessment¶

The inventory uses a labelled array A with four axes: sample, product, activity and year. A sample is a set of uncertain inputs, not an iteration of the sizing procedure. The product and activity axes have the same length and include the background suppliers and foreground processes needed for the model.

For each year and sample, the model solves the linear system

\(f\) specifies demand for a vehicle or transport service; \(s\) contains the required activity amounts; \(h\) contains impact scores. The result is normalized to vehicle-km, passenger-km or tonne-km as requested.

B has axes year, impact category and activity. It contains precomputed supply-chain impact coefficients for background activities and characterization factors for elementary flows represented in the inventory. It is not a raw biosphere-exchange matrix. Its ordering must match A and the input index.

The bundled background was rebuilt from ecoinvent 3.12 cutoff with premise. Supported background scenarios are SSP2-NPi, SSP2-PkBudg1000, SSP2-PkBudg650 and static. The old 1150/500 names are rejected. ReCiPe supplies midpoint and endpoint collections; EF 3.1 supplies midpoint indicators. In the recipe midpoint collection, climate change specifically uses IPCC 2021 GWP100 excluding biogenic CO2. See Understanding inputs and results for units and the separate category that includes biogenic CO2.

See Inventory export with Brightpath for Brightway Excel, SimaPro CSV and foreground-only openLCA JSON-LD exports. The destination application needs matching background providers and elementary flows. A successfully written file does not prove that those links have been resolved in the destination database.