Vehicles Research and Data

Vehicles: energy transition conclusions?

Vehicles transport people and freight around the world, explaining 70% of global oil demand, 30% of global energy use, 20% of global CO2e emissions. This overview summarizes all of our research into vehicles, and key conclusions for the energy transition.

Electrification is a revolution for small vehicles, a mega-trend of the 21st century. We also believe that large and long-range transportation will remain predominantly combustion-fueled due to unrivalled energy density and practicality. It is most cost-effectively decarbonized via promoting efficiency gains, lower-carbon fuels and CO2 removals.

(1) Electrification is a game changer for light-vehicles, with 3-5x greater fuel economy than combustion vehicles (database here), due to inexorable thermodynamic differences between electric motors and heat engines (overview note here).

(2) Electric vehicle technology continues to improve, in an early innings, with multi-decade running room, which makes it an interesting area for decision makers to explore. We have profiled axial flux motors, which promise 2-3x higher power densities, even versus Tesla’s world-leading PMSRMs while surpassing 96% efficiencies (note here). And SiC power electronics that unlock faster and more efficient switching in the power MOSFETs underlying EV traction inverters.

(3) New and world-changing vehicle types will also be unlocked by electric motors’ greater compactness, simplicity and controllability. e-mobility provides an example with the lowest energy costs per passenger on our chart above. Albeit futuristic, we have also written on opportunities in aerial vehicles, drones and droids, robotics, airships, military technologies, inspection technologies.

(4) EV Charging. Each 1,000 EVs will ultimately require 40 Level 2 (30-40kW) and 3.5 Level 3 (100+ kW) chargers (NREL estimates). But we wonder if EV charging infrastructures will ultimately get overbuilt (for reasons in our note here). Is there a moat in the patents of EV charging specialists, such as Chargepoint? Or Nio? As a result, we are particularly excited by the shovel-makers, the suppliers of materials and electronic components, that will feed into chargers, especially fast chargers. Granular costs of EV chargers are modelled here, and can be compared with the 17c/gallon net margins of conventional fuel retail stations here. An amazing statistic is that a conventional fuel pump dispenses 100x more fuel per minute than a 150kW fast-charger, and we wonder if fast-chargers will stoke demand for CHPs (note here).

(5) Large vehicles that cover large distances face different constraints. For example, once the battery in a heavy truck surpasses 8 tons, yielding c50% of the range of a diesel truck, then additional battery weight starts eating into cargo capacity (data here). And the range of a purely battery powered plane is currently around 90km (data here).

(5a) For trucks, an excellent data-file comparing diesel, LNG, CNG, LPG and H2 fuelling is here. There may be some niche deployments of electric trucks and hydrogen trucks. But we think the majority of long distance, inter-continental trucking will remain powered by liquid fuels, i.e., oil products. Although they may improve efficiency by hybridizing (energy saving data here), including using super-capacitors (note here).

(6) Economies of scale. Larger vessels, which carry more passengers and more freight are inherently more energy efficient. This is visible in the title chart. And we have modelled the economics of container ships, bulk carriers, LNG shipping, commercial aviation, mine trucks, electric railways, pipelines and other offshore vessels.

(7) Light-weighting also improves fuel economy, as energy consumption is a linear function of mass, and replacing 10% of a vehicle’s steel with carbon fiber can improve fuel economy by 16% (vehicle masses are built up here). Polymers research here. This can be compared with typical vehicle manufacturing costs.

(8) Automating vehicles can also make them 15-35% more efficient (note here), although we also wonder whether the improved convenience would also result in more demand for long-distance road travel…

(9) Changing demand patterns? Autos are c95% of <500-mile trips today, planes are c90% of >1,000-mile trips; while long distance travel is c50% leisure and visiting friends (data here). Travel demand correlates with income across all categories (data here). Travel speeds have also improved by over 100x since pre-industrial times (excellent data here, breaking down travel by purpose, vehicle and demographics). We estimate the distribution chain for the typical US consumer costs 1.5bbls of fuel, 600kg of CO2 and $1,000 per annum, across container ships, railways, trucks, delivery vans and cars (data here). But displacing travel demand delivers the deepest reductions of all, in the energy intensity of transportation. Thus we would also consider remote work (note here), digitization (category here), traffic optimization and recycling (here) together with vehicle efficiency technologies. They are all related. But as always, there are good debates to be had about future energy demand and fears over Jevons Paradox.

(10) Hydrogen vehicles. Despite looking for opportunities in hydrogen vehicles and e-fuels, we think there may be more exciting decarbonization opportunities elsewhere: due to higher costs, high energy penalties and challenging practicalities. This follows notes into hydrogen trucks; and data-files into Goldilocks-like fuel cells and hydrogen fuelling stations. A summary of all of our hydrogen research is linked here.

The data-file linked below summarizes all of our research to-date into vehicles, which follows below in chronological order. Note that we have generally shown energy use on a wagon-to-wheel basis and assume 2.0 passengers per passenger vehicle. You can stress test all of the different inputs ($/gal fuel, c/kWh electricity, $/kg hydrogen) in the data-file. Other excellent comparison files are here (EVs vs ICEs, ICEs vs H2 vehicles, diesel vs LNG vs H2 trucks). We have also published similar overviews for our research into batteries, electrification, battery metals and other important materials in the energy transition.

Electric vehicles: breaking the ICE?

Electric vehicles are a world-changing technology, 2-6x more efficient than ICEs, but how quickly will they ramp up to re-shape global oil demand? This 14-page note finds surprising ‘stickiness’. Even as EV sales explode to 200M units by 2050 (2x all-time peak ICE sales), the global ICE fleet may only fall by 40%. Will LT oil demand surprise to the upside or downside?

Global vehicle fleet: vehicle sales and electrification by region?

We have modeled the global light vehicle fleet, light vehicle sales by region, and the world’s shift from internal combustion engines (ICEs) towards electric vehicles (EVs) through 2050. Our base case model sees almost 200M EV sales by 2050, and a c40% decline to around 1bn combustion vehicles in the world’s fleet by 2050.

Our long-term forecasts for global oil demand see gasoline and diesel demand from light vehicles falling by 65% by 2050, as electric vehicles replace combustion vehicles. In thermodynamic terms, EVs are a superior technology. But how quickly do they ramp and replace internal combustion incumbents?

This data-file models the global vehicle fleet by region, starting with numbers from the OICA, adding our own estimates going back to 1990 and the running forecasts out to 2050.

The simple mathematical methodology is that the fleet in Year X equals the fleet in Year X-1 plus new vehicle sales minus retirements.

Today’s global vehicle fleet consists of 1.7bn light vehicles, of which 290M are in the US, 340M in Europe, 77M in Japan, 320M in China, 45M in India, 60M in Africa. This includes passenger vehicles and light commercial vehicles.

Global vehicle sales will rise by almost 3x, from c80M in 2022 to 230M in 2050. In per capita terms, this is an increase from 0.01 sales pp pa to 0.024 pp pa by 2050, mainly driven by GDP per capita rising from $13k pp pa to $22k pp pa, and especially in the emerging world.

Global vehicle ownership would thus increase c50%, from 0.2 pp to 0.3 pp. Vehicle ownership rates move sideways in the developed world; but rise in the emerging world. By 2050, car ownership rates are “only” 25% below today’s developed world levels in China, 50% below in other Asia, 60% below in LatAm, 75% below in India and 85% below in Africa.

An increasing portion of global vehicle sales are electric as part of our roadmap to net zero. Our base case is a scenario where effectively all vehicle sales in the developed world shift to electric, while the emerging world continues to buy a more even mixture (chart below).

Electric vehicle sales are forecast by region in the data-file. But the numbers can be varied in the data-file, to stress-test your own scenarios. This also drives the demand for key materials used in batteries, motors and traction inverters, such as lithium, fluorinated polymers, battery-grade nickel, graphite, copper, Rare Earth Metals and SiC.

Retirement rates of vehicles depend on their age and location. Further data on the typical retirement ages of vehicles is tabulated here. But in the model, we have had to assume a material acceleration in old vehicle retirement rates, to phase out ICEs faster.

Please download the data-file to stress test the evolution of global vehicle sales and the global vehicle fleet. Our base case sees a 40% reduction in the ICE fleet by 2050. But we think there are realistic alternative scenarios where the global ICE fleet may not decline materially from 2022 levels. Our own perspective from interrogating the model is that it also takes quite extreme assumptions to reduce the global vehicle fleet by over 75% by 2050.

Vehicle fleets: service life and retirement age by vehicle type?

The weighted-average combustion vehicle in the world has a current age of 12-years and an expected service life of 20-years. In other words, a new combustion vehicle entering the global fleet in 2023 will most likely be running through 2043. Useful data and notes are compiled here.

How long will new combustion vehicles remain in service and contribute to global oil demand? To answer this question, we have compiled data into different vehicle types.

One method to estimate the average service life of a vehicle is simply to take the average age of that type of vehicle in today’s fleet as a proxy. For example, the average passenger car in operation is around 12-years old, implying that half of cars might last less than 12-years and half last longer.

However this is a 25-50% under-estimate of ultimate vehicle life, because unless all global vehicles are suddenly scrapped tomorrow, then most of these vehicles will last a good while longer. The average age at retirement for vehicles in the US is closer to 16-years (charts below).

In many categories, the average has been pulled downwards by rapid expansion of the fleet in the emerging world. For example, some of the world’s youngest airplane fleets today belong to China’s Hainan Airlines (5.1-years), Saudia (5.1-years), Sichuan Airlines (5.7-years) and Brazil’s Azul Linhas (6.0-years). Likewise, the excellent “global bus survey” notes that the age of the average bus in service globally has effectively been halved by rapid growth of the bus fleet in China, Russia, Brazil, Indonesia, Mexico and Korea.

Another challenge is that vehicles getting retired in the developed world often simply get transported to frontier economies in the emerging world, where they have an entire second life. Africa imported 1.5M vehicles from the US, EU and Japan in 2018, and a significant share are over 15-years old at the time of arrival (!). The UN is advocating for tighter legislation here. For example, in 2022, Nigeria lowered the age limit on imported vehicles from 15-years to 12-years.

There are also broad distributions in vehicle ages, around the stated averages (chart below). And the rate of new vehicle purchases and old vehicle retirements also seems to fluctuate with economic conditions.

Hence we think the best overall estimates for vehicle lives is the targeted service life from the manufacturer, and some guidelines are tabulated in the data-file along with our notes. The average combustion vehicle entering service in 2023 is expected to last 20-years.

The shortest vehicle lifespans are expected for buses and heavy trucks, at c15-years, while the longest service lives are expected for new planes, ships and trains at 25-35 years.

The longest-lasting vehicles in our data-set include 0.5% of passenger cars in the US that are over 30-years old (chart above), and we also occasionally see a Soviet Zhiguli death-trap on the roads in Tallinn over here in the year of our lord 2023. Elsewhere, there are rail cars in service with half-a-century of history. In 2021, Mitsui noted it had scrapped its first LNG carrier, the Sunshu Maru, after a 37-year service life starting in 1984. And there are planes still taking to the skies that were constructed in 1978-81 before the oil shocks (e.g. from Iranian and Venezuelan carriers).

The full data-file contains estimates into the current fleet age (in years) and service life (years) for different types of vehicles, such as light vehicles, heavy trucks, airplanes, buses, bulk freighters, diesel trains, container ships, liquid tankers and other general ships.

Electric vehicles: motors and magnets?

This data-file assesses electric vehicle magnets, the use of permanent magnets and the use of Rare Earth materials such as neodymium (NdFeB). 80-90% of recent EVs have used Rare Earth permanent magnets, averaging 1.5 kg per vehicle, or 7.5g/kW of drive-train power, across the data-file. But the numbers vary vastly. From 0-4 kg per vehicle. 20 vehicles from different OEMs are tabulated in the data-file.

Electric vehicle motors produce torque as magnets in their rotors try to align with rotating magnetic fields that are created in their stator windings by traction inverters. For more details and underlying theory, please see our overview of magnets.

What magnets are used in electric vehicles? 80-90% of electric vehicles sold in the past half decade have used permanent magnets in their rotors, according to different sources.

This data-file tabulates the magnets used by different manufacturers, in kg/vehicle, kg/kW of power, and over time. The average EV in our screen used 1.5kg of Rare Earth Magnets per vehicle, which equates to 7.5 g/kW of power.

The scatter is broad. Some vehicles will use 4 kg of Rare Earth magnets, or as much as 15 g/kW. Almost all of the Rare Earth Magnets are NdFeB, although there are differing blends of other Rare Earths, including dysprosium and praseodymium.

Rare Earth magnets are preferred, because they are more efficient and more reliable. For example, one alternative is to use an electro-magnet in the rotor, but then this magnet needs a power supply (impacting efficiency by 1-3pp) and there is a risk of degradation on the rotating electrical connection (brushes) that feed this power supply in. Another alternative is to use induction, to magnetize the rotor core, but core losses and iron losses also leak power (impacting efficiency by 2-7pp) and generate heat via hysteresis.

Thrifting Rare Earths from EVs. One strategy we have seen from EV makers is to use less magnetic material. For example, this can be achieved by using a permanent magnet motor for the rear wheels, and induction machines on the front wheels. The induction motors are often idled for efficiency reasons, and only used when extra front-wheel power is required. Another approach to thrifting out Rare Earths optimizes the shapes or “slots” of the rotor, or winding configurations in the stator. The materials make-up of different EVs is charted below.

Supply chains. Another strategy we have seen is for EV makers to sign agreements that help guarantee security of supply. For example, in 2022 Hyundai-Kia signed an 1,000-1,5000 Tpa offtake agreement for NdPr oxides from Arafura Resources at the Nolans mine, 135km North of Alice Springs (reported here). Likewise, General Motors will source Rare Earths from MP Materials’ Mountain Pass mine project (reported here).

A final strategy is to use electro-magnets or induction motors in lieu of permanent magnets. Famously, this strategy was used by Tesla in the Model S and Model X, while Tesla said in early-2023 that it would also move away from Rare Earth magnets in future versions of the Model Y. Some European auto-makers are also adopting this strategy (details in the data-file).

There is some guesswork in the data-file. For some companies, clear data are available. For others, more painfully, googling the vehicle’s name and the keyword “magnets” was more likely to return hundreds of hits for promotional fridge magnets depicting the vehicle’s sleek design features, rather than reveal information about the use of Neodymium in its drivetrain.

Will Rare Earths be a bottleneck for electric vehicles? We think that the ramp-up of wind turbines and electric vehicles will require very large expansions in Rare Earth metals production, while in times of shortage, OEMs will sacrifice efficiency by thrifting out bottlenecked materials and relying on induction or electromagnetic machines. For more details, please see our broader research into Rare Earths.

Bulk shipping: cost breakdown?

Bulk carriers move 5GTpa of commodities around the world, explaining half of all seaborne global trade. This model is a bulk shipping cost breakdown. We estimate a cost of $2.5 per ton per 1,000-miles, and a CO2 intensity of 5kg per ton per 1,000-miles. Marine scrubbers increasingly earn their keep and uplift IRRs from 10% to 12% via fuel savings.

Bulk carriers and global trade. 13,000 bulk carriers, with 100MT of carrying capacity, transport over 5GTpa of bulk commodities ever year, in vessels with deadweight tonnage (dwt) of 4,000 – 400,000 tons. This is c50% of all global trade by mass. Of this dry bulk, c25% is iron ore, 20-25% is coal, c10% is grain, while the remaining 40-45% spans other metals and materials.

Economic modelling. This data-file models the economics of bulk carriers, including a breakdown of bulk shipping cost, across capex, opex costs, fuel, crew charges, port charges, maintenance, and insurance.

In our base case a large Capesize (or Newcastlemax) bulk tanker, with 200,000 dwt of capacity, must charge a total day rate of $67,000 per day to earn a 10% IRR (chart above) off of $60M pa capex costs (chart below).

Bulk shipping cost per ton? Costs per ton are estimated at $2.5 per ton per 1,000 miles, while CO2 intensity is estimated at 5kg per ton per 1,000 miles, as a large bulk carrier will consume 300-500bpd of oil products. Inputs and outputs can be flexed in the model.

Which oil products as used by bulk tankers? Oil products will comprise 30-50% of total shipping costs for a bulk carrier, depending on whether the vessel is consuming marine gasoil (0.1% sulphur, EU/North American limits), low sulfur fuel oil (0.5%, IMO limit) or heavier fuel oil (3.5% sulphur, but this requires a scrubber to be IMO-compliant).

Marine scrubbers are increasingly being installed. They might cost $2-6M (depending on the ship size), but pay for themselves in subsequent fuel savings. Numbers can be stress-tested in the model, but we estimate that a vessel with a scrubber will either earn 35% higher cash margins, 2% higher IRRs overall, or achieve 7% lower total shipping costs. Our flue gas desulfurization (scrubber) model is linked here.

Costs can also be compared to our models of container shipping, LNG shipping, and CO2 shipping.

Leading companies in bulk shipping? Some of the largest bulk shipping fleets are associated with global mining companies, such as Vale, which operates the largest vessels in the world, above 400,000 tons, ferrying iron ore from Brazil to China. Leading pure-plays include Golden Ocean (listed, US/Norway), Oldendorff (private, HQ’ed in Germany) and Star Bulk (listed, HQ’ed in Greece).

Offshore vessels: fuel consumption?

This database tabulates the typical fuel consumption of offshore vessels, in bpd and MWH/day. We think a typical offshore construction vessel will consume 300bpd, a typical rig consumes 200bpd, supply vessels consume 150bpd, cable-lay vessels consume 150bpd, dredging vessels consume 100bpd and medium-sized support vessels consume 50bpd. Examples are given in each category, with typical variations in the range of +/- 50%.

This data-file tabulates the typical fuel consumption for different types of offshore vesesel, across all of our research into the offshore and shipping industries.

Offshore construction vessels are especially used in the offshore wind industry, where installation costs for a large-scale wind project will average aroud $1,000/kW spread across 60-100 vessels during peak activity. The largest are offshore construction vessels which will tend to consume around 300bpd of fuel. This is also factored in our EROEI calculations for a wind turbine.

Cable lay vessels are also used in offshore wind and more broadly amidst the expansion of power grids and HVDC interconnectors. We think a typical cable lay vesel will consume 150bpd of fuel.

Offshore rigs also see a continued role in our energy balances, in order to provide 85Mbpd of long-term oil demand and 800 bcfd of long-term gas demand in our roadmap to net zero. A typical offshore oil rig consumes 200bpd of fuel. The numbers are lower for jack-ups and ultra-efficient drillships, but can be higher for larger and older semi-subs.

Elsewhere in our shipping research, we see the typical fuel consumption of a large container ship at 1400bpd, a bulk tanker at 420bpd and a LNG carrier at 270bpd.

The fuel consumption of dredging vessels and the fuel consumption of platform supply vessels (PSVs) are also covered in the data-file of offshore vessels’ fuel consumption.

Please download the data-file for additional datapoints into the fuel consumption of different ships, and individual data-points that led us to these numbers.

Hillcrest: ZVS inverter breakthrough?

Hillcrest Energy Technologies is developing an ultra-efficient inverter, which has 30-70% lower switching losses, up to 15% lower system cost, weight, size; low thermal management needs, high reliability, and confers up to 13% higher range than today’s inverters, especially for use in EV powertrains; but also in wind, solar, batteries and fast-chargers. It is based on SiC semiconductors. This Hillcrest technology review presents our conclusions from patents and technical papers.

Hillcrest was founded in 2006, is based in Vancouver, Canada with c15-20 employees. It is publicly listed, with market cap of $25M (Feb-23) and shares traded on the OTCQB Venture Market and Frankfurt Stock Exchange.

It is developing a Zero Voltage Switching (ZVS) inverter. What does this mean, and why does it matter?

Inverters convert DC power to AC power via pulse width modulation, which is covered in our primer into electricity. Specifically, electrical switches create bursts of current, of increasing frequency, then decreasing frequency, then increasing frequency, then decreasing frequency. When these power bursts are averaged out, they resemble an AC sine wave (chart below). This AC power signal can be used to feed power into the grid, or to drive the electric motors in an electric vehicle. The switches are MOSFETs or IGBTs.

The quality of the AC power signal depends on the switching frequency. Fewer pulses (each with longer length and longer gaps) creates a jagged AC sine wave. Whereas more pulses (with shorter length and shorter gaps) produces a smoother sine wave. A nice analogy is thinking about how video quality increases with a higher frame rate. So why don’t inverters dial up the switching frequency to the max?

Switches incur a small power loss every time they switch on and switch off. The reason is that when the switch is off, there is a potential difference (aka a voltage) across the junction (dark green, chart below). When the switch starts to turn on, current starts to flow from source to drain (light green, chart below). The current ramps up as the voltage ramps down. And thus, in the middle, power is dissipated, as power = voltage x current (yellow line, chart below). And so usually, the higher the switching frequency, the higher the switching losses.

Zero Voltage Switching, as the name implies, cuts the voltage from source->drain towards zero BEFORE the current from source->drain ramps up. Thus the power dissipated per switching event (VxI) is very close to zero (chart below). This is conventionally done using active snubber circuits or software on micro-controllers. In principle, ZVS enables faster switching frequencies without astronomical switching losses.

Hillcrest’s technology includes Zero Voltage Switching algorithms, which can be implemented in a micro-controller, and coupled with next-generation SiC semi-conductors, which are creating an exciting jolt forwards in the power electronics behind practically all of the core areas of the energy transition (TSE research note here).

Hillcrest’s white papers show that its algorithms achieve 30-70% lower switching losses than others using similar semi-conductors, especially at higher switching frequencies (chart below). They are also shown to operate over a wider operating range than existing solutions, and produce a particularly high quality output (low ripple, low harmonics, low EMI) .

Another benefit highlighted is that higher quality power signals should allow for downsizing of other components in the traction inverter; especially the DC link capacitor, which typically comprise 21% of the weight of the inverter, 14% of its cost, and 30% of the failures (chart below). This should be interesting for manufacturers of electric vehicles, and others in wind, solar, batteries, fast-charging, power grids.

The data-file linked below is our Hillcrest technology review. As usual, our goal is to review the company’s patents, and its White Papers, to assess (a) can we understand precisely how the company is achieving a technical breakthrough? (b) can we de-risk that breakthrough and (c) can we find a clear moat around the technology, conferring an edge for the company.

Auto manufacturing: the economics?

This data-file is a very simple model, aiming to break down the sales price of a typical mass-market automobile. Our numbers are informed by a survey of typical numbers for specific auto-plants in Europe and the US.

In typical times, a vehicle’s cost is estimated around $30k, of which c25% accrues to suppliers, c20% is sales taxes, c20% is dealer costs and logistics, c10% employees, c10% material inputs, c10% O&M, 1% electricity and c5% auto-maker margins. Numbers and calculations are in the data-file.

Amidst energy and industrial shortages, it is likely that the same vehicle could cost closer to $50k, representing c40% inflation, mostly due higher costs of materials and bottlenecks in supply chains.

EV fast charging: opening the electric floodgates?

power electronics for electric vehicle charging

This 14-page note explains the crucial power-electronics in an electric vehicle fast-charging station, running at 150-350kW, to charge up an entire EV in 10-30 minutes. Most important are power-MOSFETs, comprising c5-10% of charger costs. The market trebles by the late 2020s. We explore who benefits?

Cookies?