WORKING PAPER 2019-03

Transitioning away from heavy fuel oil in Arctic shipping

Author: Bryan Comer, PhD Date: February 18, 2019 Keywords: Arctic shipping, heavy fuel oil, hydrogen, zero emission vessels

1. Introduction socioeconomic, and environmental shipping goods between Europe and costs of spilling even a small amount Asia compared to the Suez Canal In this paper, we present five case of fuel outweigh these savings. In the route, but the time and cost savings studies of Arctic shipping routes short-term, all ships in this study could come with detrimental environmental serviced by ships that can use heavy cease using HFO and immediately use impacts (Yumashev, van Hussen, Gille, fuel oil (HFO). The five cases include (1) distillate fuels, and some ships could & Whiteman, 2017). a tanker carrying liquified natural gas use LNG. In the longer term, zero- (LNG) from Norway to South Korea, emission vessels (ZEVs) using renew- Heavy fuel oil is the most commonly (2) a cargo ship carrying wind turbine able fuels can provide an alternative used fuel in Arctic shipping, repre- equipment from Shanghai to the to HFO and 0.5% S-compliant fuels. senting 57% of fuel used and 76% of Netherlands, (3) a small container ship Hydrogen appears to be the most fuel carried on board ships in 2015 servicing western Greenland, (4) a bulk promising solution for zero-carbon (Comer et al., 2017). When spilled, carrier transporting nickel ore from long-range Arctic shipping. In most of HFO persists in the marine environ- Canada to the Netherlands, and (5) a the cases we analyzed, a ship would ment. Most of it does not evaporate, 20-night northern Europe and Arctic need to refuel once or twice along the unlike lighter petroleum-based fuels, cruise originating from Amsterdam. voyage to operate on liquid H . The and it can emulsify in the water. This For each case, we compare the costs of 2 Arctic may be the natural showcase creates a mixture that is much larger using HFO, 0.5% sulfur (S)-compliant for ZEVs that use renewable energy than the original volume spilled, and fuel, distillate fuel, LNG, electricity, and because they obviate the spill risk, one that is nearly impossible to com- hydrogen (H ). We also compare the 2 eliminate black carbon emissions, and pletely clean up. In his speech to the relative cleanup, socioeconomic, and avoid GHG emissions. 7th Symposium on the Impacts of environmental costs of spilling these an Ice-Diminishing Arctic on Naval fuels. Lastly, we explore the operational and Maritime Operations in 2017, consequences of using electricity or H 2 2. Background United States Coast Guard Admiral by estimating the number of times each ship would need to be refueled Dwindling Arctic sea ice is opening new Paul Zukunft, who coordinated the or recharged in order to complete the Arctic shipping routes, some of which response to the Deepwater Horizon voyage or voyages in each case. are navigable year-round. Recent years oil spill, stated that the Coast Guard have seen increased interest in both could not recover all of the oil if a spill We find that while in some cases oper- intra-Arctic and trans-Arctic shipping. happened in the Arctic. In fact, only ating on HFO or 0.5% S-compliant fuels For example, Russia’s Northern Sea 15 percent of the crude oil from the results in fuel cost savings, the cleanup, Route offers an enticing shortcut for Deepwater Horizon spill was removed

Acknowledgements: The European Climate Foundation generously supported this study. Thank you to Charlotte Inglis, Sian Prior, Joseph Pratt, Nikita Pavlenko, Dale Hall, and Amy Smorodin for their detailed review and advice. Thank you to exactEarth for providing Automatic Identification System data and to IHS Fairplay for providing ship characteristics data.

© INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION, 2019 WWW.THEICCT.ORG TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

despite favorable weather and sea carbon dioxide (CO2) emissions from because of their higher energy density conditions (Zukunft, 2017). Crude oil ships by at least 50% from 2008 and ease of handling. However, due is less dense than HFO and should levels by 2050 and to phase out GHG to the high toxicity of ammonia, we therefore be easier to recover. emissions from international shipping focus on liquid H2 but recognize that as soon as possible in this century other H2 carriers could be used in Heavy fuel oil is a “residual fuel” and has (Resolution MEPC.304 (72), 2018). the future. Despite the barriers there an average S content of approximately Therefore, ships will eventually need are several ZEVs in operation today, 2.6% by mass. Beginning in 2020, all to be zero-emission if the IMO is to mainly smaller ships such as ferries ships will need to either use fuels with achieve its ambitions. but also a small number of cargo ships, a maximum S content of 0.5% or use fishing vessels, and cruise ships (Hall, cheaper, high-sulfur HFO but employ All ships are capable of using distil- Pavlenko, & Lutsey, 2018). There is an exhaust gas cleaning system called late fuels; ships commonly switch from also increasing interest in developing a “scrubber.” While scrubbers are a HFO to distillates as they enter and zero-emission fuels and propulsion popular compliance option, given leave Emission Control Areas (ECAs) technologies for longer distances. the time and financial costs of retro- near the coasts of North America, the fitting and bans on washwater dis- Caribbean, and western Europe. Some charges from open-loop scrubbers in ships have dual-fuel engines that can 3. Methodology some ports, most ships will use 0.5% use HFO, 0.5% S-compliant fuels, dis- In this paper, we consider five Arctic S-compliant fuels. We expect many tillates, or LNG. Unfortunately, all of shipping routes with ships that can different types of 0.5% S-compliant these fuels emit CO and black carbon 2 use HFO or an alternate fuel. For each fuels to be available starting in 2020, and, in the case of LNG, methane. case, we compare the costs of using including desulfurized residual fuels These pollutants warm the climate and HFO, 0.5% S-compliant fuel, distillate and blends comprised of lighter dis- will eventually need to be eliminated fuel, LNG, electricity in the form of bat- tillates and heavier residual fuels. under the IMO’s GHG strategy. teries, or liquid H with fuel cells in the Unfortunately, we do not yet know if 2 a spill of 0.5% S-compliant fuel will act Ships could also use energy that year 2023. Additionally, we compare more like a distillate fuel and evaporate results in zero direct emissions and, the relative cleanup, socioeconomic, and break down over time, or more if produced using low-carbon renew- and environmental costs of spilling like a residual fuel and sink, emulsify, able electricity, could result in near- these fuels. Lastly, we estimate how and persist. Research to answer this zero lifecycle emissions. These fuels many times each ship would need to question has been proposed by other include electricity from batteries and be refueled or recharged in order to complete the voyage when operating research organizations. H2. The main challenge for batteries is their energy density: powering a on each fuel. Ships are currently free to use any ship requires a considerable amount fuel in the Arctic provided it meets of energy that can be difficult to pack Beginning in 2023, unless a ban on the minimum global fuel quality stan- inside the hull of a ship. The highest using HFO in Arctic waters is imple- dards or by using a scrubber. The energy density batteries on the mented, ships will need to use a International Maritime Organization market today are about 250 watt- scrubber to operate on high-sulfur (IMO) is working to develop a ban hours per kilogram (Hall, Pavlenko, HFO. The scrubber costs are not con- on using or carrying HFO for use as & Lutsey, 2018), about one-fiftieth sidered in this analysis, nor are the costs associated with retrofitting a a fuel in Arctic waters. It is possible the energy density of HFO. For H2, that the ban will apply to both HFO the main barriers to widespread use ship to operate on a different fuel. and fuels that may behave like HFO include its high price and limited While we recognize that retrofitting when burned or spilled, which could availability for maritime operations. a ship with a new engine, fuel system, or both, can result in significant costs, include 0.5% S-compliant fuels. IMO In addition, while H2 provides about Member States are aiming to agree to three times more energy per kilogram there are few estimates of what it ban HFO by 2021, with enforcement than HFO, it is so light, even when costs to retrofit a ship to operate on in 2023, although the timing is depen- liquefied, that one needs about eight LNG, batteries, or fuel cells. It is espe- dent on the outcome of upcoming times as much space for the same cially difficult to estimate retrofit costs IMO meetings. Previously, the IMO amount of energy. Liquid hydrogen for ships because of the heterogeneity agreed to an initial greenhouse gas carriers, such as ammonia, are being of the shipping sector: ships come

(GHG) strategy that aims to cut total discussed as potential sources of H2 in many different shapes and sizes,

2 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2019-03 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

Table 1. Arctic shipping cases and associated ship and cargo characteristics.

Fuel options Fuel Engine Engine without Cargo capacity power volume Case Ship Name Origin Destination Flag Ship Class retrofit Cargo Capacity (m3) (kW) (m3)b Significance First LNG carrier to Snøhvit LNG 170,000 transport HFO, 0.5% Christophe platform, Boryeong, Liquefied cubic Yamal project 1 Cyprus S, distillate, LNG 6,734 64,350 1,155 de Margerie Hammerfest, S. Korea Gas Tanker meters LNG to Asia; LNG Norway (m3) gas transited the Northern Sea Route in 2017. First Chinese 19,460 cargo ship to Wind turbine Rotterdam, General HFO, 0.5% deadweight sail the NSR; 2 Yong Sheng Dalian, China Hong Kong tower and 1,186 7,860 110 Netherlands Cargo S, distillate tonnes transited NSR blades (dwt) in 2013, 2015, and 2016. 424 Irena Aalborg, Denmark HFO, 0.5% Containerized twenty-foot Busiest Arctic 3 Greenland Container 804 5,880 74 Arctica Denmark (DIS)a S, distillate cargo equivalent container ship. units (TEU) Raglan Mine, Most powerful Nunavik, Rotterdam, Marshall Bulk HFO, 0.5% Ore from 4 Nunavik 31,750 dwt 2,020 21,770 367 icebreaking Quebec, Netherlands Islands Carrier S, distillate Raglan Mine bulk carrier. Canada Shetland MS Amsterdam, Islands, HFO, 0.5% 843 Popular Arctic 5 Netherlands Cruise Passengers 2,061 21,120 355 Prinsendam Netherlands Iceland, S, distillate passengers cruise ship. Greenland Notes: [a] DIS = Danish international ship registry, which is separate from Denmark’s national ship registry; [b] Estimated by the following equation based on Minnehan and Pratt (2017): Engine volume [m3] = (engine power [kW] – 1906)/54.066 [kW/m3]. each with their own particular chal- to South Korea. The Christophe de Case 3 considers the 2017 voyages of lenges. Retrofitting may also not be Margerie has dual-fuel main engines the Denmark DIS1-flagged ship the Irena worth the investment on an older ship. that can operate on HFO, 0.5% Arctica container ship as it sailed from Additionally, these five routes could be S-compliant fuel, distillate, or LNG. It is Denmark and then up and down the serviced by a replacement vessel that not known which fuel the Christophe western Greenland coast transporting operates on cleaner fuels, requiring no de Margerie used on this particular containerized cargo. The Irena Arctica retrofit costs. Therefore, we focus this voyage and, because the ship was is the busiest container ship operating analysis on the fuel costs, spill costs, carrying LNG, it may have used LNG in the Arctic, sailing more than 50,000 and refueling needs. as its fuel source. However, because nautical miles (nm) in 2017, equivalent this ship is capable of using HFO in to about two-and-a-half trips around its dual fuel engines, we examine the the world along the equator. The Irena 3.1. CASES tradeoffs of using HFO compared to Arctica is capable of using HFO, 0.5% Many different kinds of ships sail the alternative fuels. S-compliant fuel, or distillate. Arctic. Here, we consider five real- Case 4 considers the 2017 voyage of word cases, summarized in Table 1 and Case 2 considers the 2015 voyage of the Marshall Islands-flagged Nunavik mapped in Figure 1. the Hong Kong-flagged Yong Sheng general cargo ship on a westbound as it sailed from Glencore’s Raglan Case 1 considers the 2017 east- journey from China to the Netherlands Mine in Nunavik, Quebec, to Rotterdam bound voyage of the Cyprus-flagged via the NSR carrying wind turbine Christophe de Margerie, the first of materials. In 2013, this ship became 1 DIS is the Danish International Register of Shipping, which includes both Danish 15 ice-class tankers that will transport the first Chinese cargo ship to sail the and foreign merchant ships, excluding LNG from Russia’s Yamal Peninsula NSR. The Yong Sheng is capable of fishing vessels, above 20 gross tonnes. More information can be found here: to Asian markets. In 2017, it made using HFO, 0.5% S-compliant fuel, or https://www.dma.dk/SynRegistrering/ its maiden NSR transit from Norway distillate. SkibsregistreringAfgifter/DIS/Sider/default.aspx

WORKING PAPER 2019-03 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 3 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

carrying a load of nickel ore. With a 22-megawatt engine, Nunavik is the Nunavik Bulk Carrier: Ore from Raglan Mine to Rotterdam, Mar 2017 world’s most powerful icebreaking Christophe de Margerie LNG Carrier: LNG from Norway to S. Korea, Jul/Aug 2017 bulk carrier. Yong Sheng General Cargo Ship: Wind turbine equipment from Shanghai to Sweden, Jul/Aug 2015 MS Prinsendam Cruise Ship: 20-night cruise from Amsterdam, Aug/Sept 2017 Case 5 considers the 2017 voyage Irena Arctica Container ship: Western Greenland container services, year-round 2017 of the Netherlands-flagged MS IMO Polar Code Arctic Prinsendam cruise ship as it completed Sea Ice Extent (Aug 2017) a 20-night round-trip cruise from Sea Ice Extent (Mar 2017) Amsterdam with stops in the Shetland Islands, Iceland, and Greenland.

3.2. FUEL CONSUMPTION, ENERGY USE, AND FUEL COSTS This analysis uses Automatic Identification System (AIS) data from exactEarth to track and identify each ship’s speed and position. The AIS data includes a ship identification number that can be linked to the IHS Fairplay database to determine ship characteristics. including main engine power, main fuel type, and maximum speed. This information is used to estimate each ship’s HFO consumption along each journey using the method described by Olmer et al. (2017).

We then estimate the mass of non-HFO fuels needed to supply the equivalent amount of energy based on each fuel’s energy density (Table 2) and the effi- ciency of converting the fuel to propul- sion (Table 3). The equation for deter- mining the amount of fuel use needed for each case is as follows:

EDHFO ηICE FC = FC × × i,j i,HFO η © International Council on Clean Transportation, 2019 EDj p,j Figure 1. Five Arctic shipping case study routes. FCi,j = fuel consumption of ship i when operating on fuel j, in kg ηp,j = the efficiency of the propulsion The 2023 fuel costs for each case are

FCi,HFO = fuel consumption of ship i equipment associated with using calculated using fuel price estimates when operating on HFO, in kg fuel j (Table 3) found in Table 2. Heavy fuel oil, dis- ED = energy density of HFO in tillate, 0.5% S-compliant fuels, and HFO In the results, we present energy kWh/kg (Table 2) LNG prices are consistent with Roy needed to complete the voyage(s) and Comer (2017) and Faber et al. ED = energy density of fuel j in j under each case when using differ- (2016). Electricity and H prices are kWh/kg (Table 2) 2 ent fuels. To do this, we multiply the and consistent with Hall, Pavlenko, fuel consumption calculated using the ηICE = the thermal efficiency of an and Lutsey (2018). internal combustion marine engine, equation above by the energy density which we assume is 50% of the fuel (Table 2).

4 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2019-03 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

Table 2. Fuel characteristics and fuel price assumptions. Table 3. Efficiency assumptions.

Energy Density 2023 Price 2023 Price Efficiency of propulsion 3 Fuel Density (kg/m ) (kWh/kg) ($/t) ($/kWh) Fuel equipment (ηp) HFO 991a 11.1e 538g 0.048 HFO 50% (ICE) Distillate 890a 11.7e 712g 0.061 Distillate 50% (ICE) 0.5% S-compliant 910a 11.6e 688g 0.060 0.5% 50% (ICE) S-compliant LNG 456b 13.9e 462g 0.033 LNG 50% (dual fuel engine) Electricity (lower) 839c (battery) 0.11c (battery) N/A 0.035f 90% (95% DC/AC convertera Electricity (higher) 839c (battery) 0.11c (battery) N/A 0.088f Electricity x 95% electric motor) Hydrogen from fossil 40d 33.3f 4,547f 0.137 54% (60% fuel cell x 95% fuels Hydrogen DC/AC convertera x 95% Hydrogen from 40d 33.3f 7,828f 0.235 electric motor) renewables Source: Lloyds Register and UMAS (2018) and Notes: [a] ISO 8217 2017 except 0.5% S-compliant which the author estimated as the weighted average of [a] author’s assumption HFO (20%) and distillate (80%) densities; [b] U.S. Department of Energy (U.S. DOE, 2004). We assume this is the fuel system density for LNG, including insulated fuel tanks, although it may be higher or lower than this in practice. The American Bureau of Shipping (2012) estimates that 1.6 times the storage volume of HFO is needed to provide the same amount of energy with LNG; our assumption equates to 1.74 times the storage space for the equivalent amount of energy; [c] Based on SMAR-11N battery system outlined in Table 2.4 of 3 Minnehan and Pratt (2017); [d] Minnehan and Pratt (2017): While liquid H2 has a density of 71 kg/m at boiling 3 point, we must consider the H2 fuel system as a whole, which has a density of approximately 40 kg/m . [e] Author’s assumptions based on typical marine fuel energy densities; energy density of 0.5% S-compliant fuel is the weighted average of HFO (20%) and distillate (80%) energy densities; [f] Hall, Pavlenko, and Lutsey (2018); this is the density of liquid H2; [g] consistent with Roy and Comer (2017) and Faber et al. (2016).

3.3. COSTS OF FUEL SPILLS and colleagues maintain these spill for this analysis are found in Table 4. cost estimates are conservative when Liquefied natural gas spills result in Fuel spills result in cleanup, socio- applied to Arctic oil spills, as they are no shoreline cleanup costs because economic, and environmental costs. based on worldwide data. We have LNG largely evaporates when spilled. For this analysis, we assume that the estimated 0.5% S-compliant fuel Although there may be some environ- HFO and distillate 2023 spill costs cleanup costs based on the assump- mental impacts, the costs are assumed are the same as the 2018 costs esti- tion of a 20% HFO and 80% distillates to be negligible in this analysis. mated by DeCola et al. (2018). DeCola blend. Fuel spill cost assumptions However, spilling LNG, which is mostly composed of methane, a potent GHG, Table 4. Fuel spill cost assumptions. does have socioeconomic costs. Based Shoreline on the U.S. EPA (2016), we estimate b Tonnes cleanup Socioeconomic Environmental Total costs the social cost of methane in 2023 at Fuel Type spilleda (2023 $/t) (2023 $/t) (2023 $/t) (2023 $/t) approximately $1,750 per tonne.2 We 380 to 64,000 72,000 14,000 150,000 assume that electricity, which cannot HFO 3,850 be “spilled” per se, and H , which is >3,850 31,000 63,000 13,000 107,000 2 non-toxic, quickly disperses, and is not 380 to 19,000 40,000 7,000 66,000 a GHG, result in zero spill costs. 0.5% S 3,850 compliant >3,850 8,000 33,000 5,000 46,000 Fuel spill costs are estimated by mul- 380 to 8,000 32,000 5,000 45,000 tiplying the quantity of fuel spilled by 3,850 Distillate the total costs reported in Table 4. >3,850 3,000 25,000 4,000 32,000 Any LNG 0 1,750 0 1,750 2 The U.S. EPA estimates the social costs of amount methane at approximately $1,300 per tonne Electricity Any in constant 2007 U.S. dollars in the year 2023, 0 0 0 0 which is roughly $1,600 per tonne in 2018 U.S. or H amount 2 dollars; assuming inflation of 2% per year, we Notes: [a] Spills <380 tonnes have costs, as presented in DeCola et al. (2018), but the potential spills in estimate the social costs of methane in 2023 this analysis all exceed 380 tonnes; [b] costs are slightly different than DeCola et al. due to rounding. U.S. dollars at approximately $1,750 per tonne.

WORKING PAPER 2019-03 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 5 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

3.4. NUMBER OF REFUELS OR PME – 73.331 voyage(s) for each case when using V = RECHARGES NEEDED TO FC liquid H 55.944 2 COMPLETE EACH CASE E = energy input needed for ship i V = volume (m3) the fuel cell system i,LH2 Depending on the specific fuel used, FC to operate on liquid H in kWh would need to occupy to provide 2 a ship may need to refuel to complete an equivalent output power as the D = density of liquid H in kg/m3 its entire voyage. Based on the volu- LH2 2 existing main engine (Table 2) metric density (kg/m3) and energy ED = energy density of liquid H in density (kWh/kg) of each fuel, we PME = main engine power at MCR LH2 2 can estimate how much energy could kWh/kg (Table 2) Second, we estimate the remaining be provided within the space avail- V = the volume available to store space available on board for liquid LH2 able for the existing fuel and engine liquid H on board H fuel storage.4 We first assume 2 systems on board each ship. For each 2 case, we compare the energy needed that the combustion engine room is For batteries, we assume that the to the energy that can be provided five-times larger than the volume of battery system can take up two- to estimate how many times the ship the engine (J. Pratt, personal com- and-a-half times the volume of the would need to refuel or recharge to munication, December 18, 2018). We engine and all the volume currently complete the voyage(s). then assume that the fuel cell engine taken up by the fuel tanks, consistent room is twice the size of the fuel cell with Minnehan and Pratt (2017). We For HFO, 0.5% S-compliant, and dis- system plus the volume taken up by also assume that the battery capacity tillate fuels, we estimate the number the existing fuel tanks to allow enough needs to be greater than the energy of refuels needed to complete the space to operate and maintain the fuel the ship will use to avoid damaging the voyage(s) in each case as follows: cell systems. Therefore, the volume of battery. We assume that the ship can

liquid H2 that can be stored on board is use 75% of the energy that the batter- Ei,j R = calculated as follows: ies can store. Therefore, the number i,j D × ED × V j j f of recharges needed to complete the V = V × 5 – V × 2 + V LH2 e FC f voyage(s) when operating on batteries Ri,j = number of times ship i needs to be refueled to complete the V = the volume available to store is as follows: LH2 voyage(s) for each case when using liquid H on board 2 Ei,elec fuel j R = V = volume taken up by the existing i,elec e (Dbatt × EDbatt × [Ve × 2.5 + Vf] ) × 0.75 3 Ei,j = energy input needed for ship i to engine in m (Table 1) operate on fuel j in kWh Ri,elec = number of times ship i needs VFC = volume of the fuel cell D = density of fuel j in kg/m3 to be refueled to complete the j V = volume taken up by the existing (Table 2) f voyage(s) for each case when using 3 fuel tanks in m , equivalent to the electricity in batteries EDj = energy density of fuel j in ship’s fuel capacity (Table 1) kWh/kg (Table 2) Ei,elec = energy input needed for ship i Finally, we estimate the number to operate on electricity in kWh Vf = volume taken up by the existing 3 of refuels needed to complete the 3 fuel tanks in m , equivalent to the Dbatt = density of batteries in kg/m voyage(s) when using liquid H as ship’s fuel capacity (Table 1) 2 (Table 2) follows: ED = energy density of batteries in E batt For liquid H , we first estimate how i,LH2 2 R = kWh/kg (Table 2) much space the fuel cell system would i,LH2 D × ED × V LH2 LH2 LH2 need to occupy to provide an equiva- Ve = volume taken up by the existing 3 lent amount of power as the existing R = number of times ship i needs engine in m (Table 1) i,LH2 main engine at maximum continuous to be refueled to complete the Vf = volume taken up by the existing rating (MCR). Based on Minnehan and fuel tanks in m3, equivalent to the 3 Pratt (2017), we estimate the space ship’s fuel capacity (Table 1) needed for the fuel cell as follows: 4 Note that these assumptions are different than those published in Minnehan and Pratt (2017) but are informed by personal communication 3 Minnehan and Pratt (2017) equation 2.8 can with J. Pratt. The methodology in this working be modified to estimate the volume of the paper attempts to avoid underestimating both fuel cell needed to provide a given amount of the combustion engine room volume and the installed power. volume available to the fuel cell.

6 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2019-03 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

4. Results Table 5. Activity, fuel consumption, and fuel cost for each case. A summary of each case and the esti- Maximum mated HFO fuel cost to complete each petroleum HFO fuel Distance fuel HFO cost voyage is presented in Table 5. Note Voyage sailed Voyage onboard used (2023 that, with the exception of Case 1, these Case Ship Name Description (nm) days (t) (t) US $) ships spend a portion of the voyage Christophe LNG from Norway 1 6,300 19 6,734 2,985 1,606,000 within ECAs where using HFO is pro- de Margerie to S. Korea hibited unless the ship can demon- General cargo 7,700 strate equivalent compliance with ECA 2 Yong Sheng from Shanghai 28 1,186 578 311,000 fuel quality regulations. One means to Sweden of compliance is using an exhaust gas Year-round container service Year- cleaning system, also called a scrubber. 3 Irena Arctica 54,000 804 3,212 1,733,000 in western round For this analysis, we assume these Greenland ships could use HFO along the entire Ore from Canada 4 Nunavik 3,100 19 2,020 597 321,000 voyage in 2023, which implies the use to Rotterdam of scrubbers. In practice, these specific 20-night Shetland MS ships may comply by using distillate 5 Islands, Iceland, 4,500 21 2,061 469 252,000 Prinsendam fuels when operating in ECAs, which Greenland cruise would reduce the amount of HFO used along the journey but increase the overall fuel costs along the voyage. Electricity (low) 0.40 All five ships can operate on cleaner fuels in addition to HFO and 0.5% LNG 0.69 S-compliant fuels. In the short-term, operators could carry and use distillate Electricity 1.00 fuels and, in the case of the Christophe (high) de Margerie, LNG fuel. In the medium- term, the other four ships could be HFO 1.00 retrofit to operate on LNG, which largely solves the spill problem but Fuel 0.5% 1.23 ignores climate impacts. Additionally, S-compliant these ships could be retrofit to use Distillate 1.26 electricity, H2, or some other low- or zero-carbon fuel. Hydrogen 2.59 (fossil) 4.1. FUEL COSTS Hydrogen 4.46 Figure 2 shows the voyage-level costs (renewable) of operating any of the five ships on different fuels in 2023, relative to the 01 2345 cost of operating on HFO, taking into Fuel costs relative to heavy fuel oil account the energy densities of the Figure 2. Voyage-level fuel costs relative to heavy fuel oil, accounting for energy density fuels (Table 2) and the propulsion of each fuel and propulsion equipment efficiency. equipment efficiencies found in Table 3. Heavy fuel oil is not always the least- using batteries could be as expensive Using hydrogen could be two-and- cost fuel option for ships that can as HFO or up to 60% less expensive. a-half to four-and-a-half times more use other fuels. Operating on LNG is Using 0.5% S-compliant fuels or distil- expensive than using HFO for fossil- expected to cost 31% less than HFO. lates is expected to be roughly 23% fuel derived H2 and renewable-derived

Depending on the electricity price, to 26% more expensive than HFO. H2, respectively.

WORKING PAPER 2019-03 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 7 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

4.2. SPILL COSTS COMPARED TO consider the costs associated with H2 because using LNG or electricity FUEL COSTS a spill. The total cleanup, socioeco- would be less expensive. For Cases While using distillate and hydrogen nomic, and environmental costs of an 1, 2, 4, and 5, because cleaning up would result in higher voyage-level HFO spill of between 380 and 3,850 residual fuel spills is more expen- fuel costs compared to HFO or 0.5% tonnes are estimated to be $150,000 sive than cleaning up distillate spills, S-compliant fuels, a spill of distillates per tonne (Table 4). The total costs of spilling more than 0.1% of the HFO or hydrogen would result in low or no a distillate spill of the same size are or 0.5% S-compliant fuel that could cleanup, socioeconomic, and environ- approximately $45,000 per tonne, a be carried during that particular mental costs. Therefore, when con- difference of $105,000 per tonne. If voyage would result in spill costs that sidering which fuel to use in Arctic less than one tonne of HFO is spilled exceed the fuel cost savings of using shipping, one should consider not along the route, the extra costs would these fuels. Similarly, because there only the fuel costs, but also the costs outweigh the fuel cost savings from are no costs associated with spilling of a potential spill. HFO, especially given that smaller H2, spilling more than ~1% of the HFO spills tend to be costlier per tonne or 0.5% S-compliant fuel that could Consider a concrete example: trans- than larger spills.5 be carried would result in spill costs porting approximately 30,000 tonnes exceeding fuel cost savings for the of ore from Canada to Rotterdam Table 6 indicates how much HFO or voyages in each case. (Case 4) requires nearly 600 tonnes 0.5% S-compliant fuel would need of HFO at a cost of approximately to be spilled before the cost savings Case 3 models one full year of activity $321,000 (Table 5). Using distillate associated with operating on the less rather than a one-off journey. In this fuel, the most likely replacement fuel expensive fuel were overcome by the case, the Irena Arctica would need to for the Nunavik bulk carrier, would costs of spilling the fuel. Heavy fuel avoid spilling more than 0.53% of the cost approximately $405,000 to oil and 0.5% S-compliant fuels are HFO or 0.32% of the 0.5% S-compliant complete the same voyage. Now, compared only to distillate fuel and fuel the ship could carry in its fuel

Table 6. Tonnes of HFO or 0.5% S-compliant fuel that, if spilled, results in spill costs outweighing fuel cost savings of using these fuels.

0.5% 0.5% 0.5% S-compliant S-compliant HFO instead HFO instead S-compliant instead of instead of HFO instead of hydrogen of hydrogen instead of hydrogen hydrogen Case Ship Name Voyage of distillate (fossil) (renewable) distillate (fossil) (renewable) LNG from 4.0 t (0.06% of Christophe 17 t 37 t 2.4 t 33 t 79 t 1 Norway to S. fuel capacity de Margerie (0.26% of FC) (0.55% of FC) (0.03% of FC) (0.50% of FC) (1.2% of FC) Korea [FC]) General cargo 0.8 t 3.3 t 7.2 t 0.5 t 6 t 15 t 2 Yong Sheng from Shanghai (0.07% of FC) (0.28% of FC) (0.61% of FC) (0.04% of FC) (0.55% of FC) (1.3% of FC) to Sweden Year-round Irena container service 4.3 t 19 t 40 t 2.5 t 36 t 85 t 3 Arctica in western (0.53% of FC) (2.3% of FC) (5.0% of FC) (0.32% of FC) (4.5% of FC) (11% of FC) Greenland Ore from 0.8 t 3.4 t 7.5 t 0.5 t 7 t 16 t 4 Nunavik Canada to (0.04% of FC) (0.17% of FC) (0.37% of FC) (0.02% of FC) (0.33% of FC) (0.78% of FC) Rotterdam 20-night MS Shetland 0.6 t 2.7 t 5.9 t 0.4 t 5.3 t 12 t 5 Prinsendam Islands, Iceland, (0.03% of FC) (0.13% of FC) (0.28% of FC) (0.02% of FC) (0.25% of FC) (0.60% of FC) Greenland cruise

5 DeCola et al. (2018) estimate that the total costs of an HFO spill of less than 2 tonnes is $246,000 per tonne compared to $77,000 per tonne for a distillate spill.

8 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2019-03 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

tanks in order to ensure that the fuel 10,000 HFO instead HFO instead of HFO instead of cost savings would outweigh the of distillate hydrogen (fossil) hydrogen (renewable) costs of spilling them. If operating on 2,161 HFO or 0.5% S-compliant fuel instead 1,664 of H , the most the Irena Arctica could 1,009 2 1,000 775 spill would be 85 tonnes (~11% of fuel 504 capacity) over the course of the year 388 235 232 to guarantee that the fuel cost savings 182 178 of using the residual fuels would 108 123 100 84 exceed the spill costs.

In each case, fuel costs savings will 29 continue to accrue for every voyage 13 where the quantity of HFO or 0.5% 10 S-compliant fuel spilled is less than

those indicated in Table 6. However, by a one large HFO spill (log scale) Voyages of fuel cost savings erased one large spill could erase those accu- mulated savings. If, for example, we 1 define a “large spill” as one where 65% Case 1: Case 2: Case 3: Case 4: Case 5: LNG from Norway General cargo Year-round Ore from Canada 20-night Shetland of the maximum fuel capacity of the to S. Korea from Shanghai to container service to Rotterdam Island, Iceland, ship is spilled, we can estimate how (6,300 nm) Sweden in western (3,100 nm) Greenland cruise (7,700 nm) Greenland (4,500 nm) many voyages of fuel cost savings (54,000 nm) would be erased from a large spill. Figure 3. Number of voyages that must be completed before using heavy fuel oil is As shown in Figure 3, if the Irena Arctica guaranteed to result in net cost savings assuming that, in the event of a “large spill”, 65% of the maximum fuel capacity is lost. (Case 3) managed to sail 122 voyages without a major spill, the savings asso- 0.5% S-compliant 0.5% S-compliant instead 0.5% S-compliant instead ciated with operating on HFO instead instead of distillate of hydrogen (fossil) of hydrogen (renewable) of distillate would be erased if a spill 10,000 occurred on the 123rd voyage. Because 3,623 each voyage of the Irena Arctica repre- 2,790 sents one year of activity, this is equiv- 1,692 alent to avoiding a large spill for more 1,151 1,000 than a century. In Case 1, the Christophe de Margerie would need to avoid a 257 large spill for 775 voyages and, in Case 206 198 5, the MS Prinsendam would need to 120 108 make over 2,000 20-night voyages 100 90 84 without a large spill. If the Nunavik 51 38 (Case 4) operated on HFO instead of

H2, it would have to complete more 15 than 178 voyages to ensure that the 10 fuel cost savings of operating on HFO 6

instead of renewable H2 were realized, Voyages of fuel cost savings erased by a or more than 388 voyages if HFO was one large 0.5% S-compliant spill (log scale) used instead of fossil H2. If not, a large 1 HFO spill would cost more than oper- Case 1: Case 2: Case 3: Case 4: Case 5: ating on H . LNG from Norway General cargo from Year-round Ore from Canada 20-night Shetland 2 to S. Korea Shanghai to container service in to Rotterdam Island, Iceland, (6,300 nm) Sweden western Greenland (3,100 nm) Greenland cruise Figure 4 shows that a similar pattern (7,700 nm) (54,000 nm) (4,500 nm) is observed if a ship operates on 0.5% Figure 4. Number of voyages that must be completed before using 0.5% S-compliant fuel S-compliant fuel instead of distillate is guaranteed to result in net cost savings assuming that, in the event of a “large spill”, or H2. 65% of the maximum fuel capacity is lost.

WORKING PAPER 2019-03 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 9 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

4.3. REFUELING OR RECHARGING Table 7. Number of refuels or recharges required to operate on each fuel in each case. REQUIREMENTS Energy that can Refuels or For all five cases, the routes could be be provided within Energy recharges needed serviced by ships that operate on any the existing space needed for over the course of of the fuels analyzed. In some cases, for fuel and engine the case the voyage(s) Case Fuel systems (GWh) (GWh) in each case the ship will need to refuel to complete HFO 74.1 33.2 0 the voyage(s) that make up each case. 0.5% Table 7 indicates the number of times 77.1 33.2 0 Case 1: S-compliant the ship will need to refuel or recharge LNG from Distillate 77.9 33.2 0 (for batteries) to complete the voyage Norway to within each case. This assumes that S. Korea LNG 42.6 33.2 0 the ship does not sacrifice cargo space Electricity 0.67 18.3 27 to operate on the alternative fuel. Hydrogen 13.6 30.5 2 HFO 13.1 6.4 0 In Cases 1, 2, 4, and 5, no ships 0.5% 13.6 6.4 0 require refueling when operating on Case 2: S-compliant petroleum-based fuels (HFO, 0.5% General Distillate 13.7 6.4 0 S-compliant, or distillates) or LNG. cargo from Shanghai to LNG 7.5 6.4 0 Sweden Electricity 0.10 3.5 34 To complete voyages using H2, Cases 4 and 5 would need to refuel once and Hydrogen 1.9 5.9 2 Cases 1 and 2 would need to refuel HFO 8.9 35.8 3 twice. Again, this assumes that no 0.5% Case 3: 9.2 35.8 3 cargo or passenger space is sacrificed Year-round S-compliant to carry more fuel. Case 5 provides container Distillate 9.3 35.8 3 a reasonable opportunity to use H service in 2 LNG 5.1 35.8 6 given that the MS Prinsendam and western Greenland Electricity 0.07 19.7 288 similar cruise ships make frequent port calls. Case 1 and Case 2 could also Hydrogen 1.3 32.9 25 be good candidates if the Christophe HFO 22.2 6.6 0 0.5% de Margerie and the Yong Sheng or 23.1 6.6 0 Case 4: S-compliant similar ships were able to bunker H2 in Europe, along the Northern Sea Ore from Distillate 23.4 6.6 0 Canada to Route, and in Asia. Case 4 could also Rotterdam LNG 12.8 6.6 0 use H2 if the Nunavik or a similar ship Electricity 0.20 3.7 17 sacrificed ore capacity to carry more Hydrogen 4.1 6.1 1 fuel. Using batteries to provide all HFO 22.7 5.2 0 the energy needed to complete the Case 5: 0.5% 23.6 5.2 0 voyages in all cases is out of reach 20-night S-compliant with current technology, but batteries Shetland Distillate 23.8 5.2 0 could provide supplemental propul- Islands, Iceland, LNG 13.1 5.2 0 sion energy and auxiliary energy. Greenland Electricity 0.20 2.9 13 cruise In Case 3, we find that with HFO, 0.5% Hydrogen 4.1 4.8 1 S-compliant, or distillate fuels, the Irena Arctica would need to refuel at would mean 288 recharges, essen- detailed analysis would be required least 3 times to complete a season tially requiring daily recharging. The to confirm this. Batteries could also of year-round container service to Irena Arctica completes many short be used to supplement main engine western Greenland. Using LNG would journeys along the Greenland coast power and smooth out variations in require 6 refuels. If the ship used each year and some may be short main engine load when the ship is

H2, it would need 27 refuels. Using enough to use battery power for most operating at variable speeds, such as fully battery electric propulsion or all of the trip, although additional near port.

10 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2019-03 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

5. Conclusions would result in fewer air pollutants (e.g., are limited applications for their use. sulfur oxides) and climate pollutants, There simply is not enough space on Given IMO’s efforts to ban HFO use including black carbon. most ships to make long fully electric and carriage as fuel in Arctic waters journeys possible at this time. Batteries and to phase out GHGs from inter- Liquefied natural gas, a major export could be used for short routes in the national shipping, we presented five commodity in the Arctic, is another Arctic, perhaps on voyages that take case studies of Arctic shipping routes viable alternative that is predicted less than one day to complete, and serviced by ships that can use HFO or to be less expensive to use than any as part of a hybrid system where the 0.5% S-compliant fuels. We explored other fossil fuel in the Arctic in 2023. battery provides power when the ship the costs and benefits of using fossil The Christophe de Margerie is the is operating at variable engine loads, fuels that have less severe conse- only ship in our five cases that can such as when the ship is maneuvering quences when spilled (distillates and use LNG immediately due to its dual in and near port. LNG) and alternative fuels that elimi- fuel engine. The other ships would nate spill risk and emit zero emissions need to be retrofitted, which would Hydrogen appears to be the most from the ship (electricity [batteries] mean completely overhauling the promising solution for zero-carbon, and H ) for the year 2023. We also 2 fuel and engine systems, or an LNG- long-range Arctic shipping. In most of examined what it would take, logisti- powered ship would need to be sub- the cases we analyzed, a ship would cally, to use alternatives to HFO and stituted along this route. While there need to refuel once or twice along the 0.5% S-compliant fuels on these routes. are socioeconomic costs associated voyage to operate on liquid H2. The MS Prinsendam cruise ship could be a In each of the five scenarios, all the with LNG spills, there are zero spill- prime candidate for using H because ships can stop using HFO and avoid related cleanup costs because it evap- 2 it requires only one refueling for a the use of 0.5% S-compliant fuels by orates. It also releases few harmful 20-night journey and makes frequent 2020. All ships can immediately use air pollutants and nearly zero black port calls. If the Christophe de Margerie distillate fuels and the Christophe de carbon. The downside of using LNG tanker and the Yong Sheng general Margerie tanker can use LNG. In the is that it is a fossil fuel, so combust- ing it releases CO that contributes cargo ship or similar ships were able long term, these ships can be modified 2 to bunker H in Europe, along the to use fuels that emit zero emissions to climate change; more importantly, 2 Northern Sea Route, and in Asia, these from the ship, including electricity LNG is mainly methane, a potent ships and routes could also be good and H . These alternatives emit fewer climate warming pollutant. Methane 2 candidates for using H . In the case and potentially zero lifecycle air and leakage throughout the fuel’s lifecycle 2 of the Irena Arctica, it would have climate pollutants compared to HFO and methane slip from the marine to refuel only 25 times per year to or 0.5% S-compliant fuels and they engine can result in LNG being more provide year-round H -powered con- reduce or eliminate the costs associ- damaging to the climate than petro- 2 ated with fuel spills. leum-based fuels. Therefore, while tainer service to western Greenland LNG avoids many of the negative con- if it were retrofitted with a fuel cell. Using distillate fuels is the simplest way sequences of a bunker fuel spill, we do While there is growing interest in for all ships to immediately stop using not consider LNG a long-term solution using H2 for maritime shipping appli- HFO in the Arctic. Using distillates to decarbonize the shipping sector. cations, current barriers include a requires only minor mechanical modi- limited supply, especially from sus- fications, and ships routinely switch The use of batteries avoids the costs of tainable sources, limited fueling infra- from HFO to distillates as they enter fuel spills, but the source of electricity structure, and higher costs relative will determine the overall health and and depart ECAs. Distillate fuels are to other fuels. For now, H2 is mainly expected to be slightly more expen- climate impacts compared to other limited to ferry operations and other sive than 0.5% S-compliant fuels and fuels. It is possible to produce electric- domestic shipping applications, but 26% more expensive than HFO in ity from low- or zero-carbon sources that could change depending on the

2023. However, distillate spills are esti- today, including in and near the Arctic. relative costs of using H2 compared to mated to be 30% less costly than 0.5% Additionally, electricity is often less other fuels. There is interest in using

S-compliant spills and 70% less costly expensive on a per-unit-energy basis ammonia as an H2 carrier but there than HFO spills when the cleanup, socio- than fossil fuels and H2. However, are serious health and safety risks economic, and environmental costs are because of the low energy density of that must be overcome given that it considered. Additionally, distillate fuels batteries relative to other fuels, there is highly toxic at low concentrations.

WORKING PAPER 2019-03 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION 11 TRANSITIONING AWAY FROM HEAVY FUEL OIL IN ARCTIC SHIPPING

As Arctic shipping increases, pressure DeCola, E., Robertson, T., Fisher, M., and Resolution MEPC.304(72), Initial IMO will mount to take actions to protect Blair, L. (2018). Phasing out the use and Strategy on Reduction of GHG the environment and its peoples carriage for use of heavy fuel oil in the Emissions from Ships. April 13, from the consequences of using and Canadian Arctic: Impacts to northern 2018. Retrieved from http://www. carrying fuels that are harmful when communities. Report to WWF-Canada. imo.org/en/OurWork/Documents/ burned and spilled. If the use and Det Norske Veritas (2013). HFO in the Resolution%20MEPC.304(72)%20 carriage of HFO is banned in Arctic Arctic—Phase 2 (No. 2013–1542– on%20Initial%20IMO%20Strategy%20 on%20reduction%20of%20GHG%20 waters, there are alternatives that may 16G8ZQC–5/1). Retrieved from emissions%20from%20ships.pdf. be less expensive to use (e.g., LNG or https://oaarchive.arctic-council. org/ electricity) or less costly when spilled handle/11374/1315. Roy, B. and Comer, B. (2017). Alternatives to heavy fuel oil use in the Arctic: (e.g., distillates, LNG, electricity, or H2). Etkin, D. (2004). Modeling oil spill In the short-term all ships are capable response and damage costs. Economic and environmental trad- of operating on distillate fuels, which Environmental Research Consulting. eoffs. International Council on Clean are less damaging when spilled, and Prepared for the U.S. Environmental Transportation. Retrieved from https:// some ships can operate on LNG, which Protection Agency. Retrieved from www.theicct.org/publications/alter- natives-heavy-fuel-oil-use-arctic-eco- avoids spill cleanup costs. Both of https://archive.epa.gov/emergen- nomic-and-environmental-tradeoffs. these fuels are fossil fuels and do not cies/docs/oil/fss/fss04/web/pdf/ etkin2_04.pdf. offer a long-term climate solution for U.S. DOE. (2004). Liquefied natural gas: shipping. In the longer term, ships can Faber, J., et al. (2016). MEPC 70/INF.6— Understanding the basic facts. U.S. use renewable energy stored in bat- Assessment of fuel oil availability: Final Department of Energy. Retrieved teries or used in fuel cells. The main report. The International Maritime from https://www.energy.gov/fe/ barriers to using these technologies Organization. Retrieved from www. downloads/liquefied-natural-gas- understanding-basic-facts. are the relatively low energy density of cedelft.eu/publicatie/assessment_of_ fuel_oil_availability/1858. batteries and the relatively high price U.S. EPA. (2016). Addendum to techni- Hall, D., Pavlenko, N., and Lutsey, N. cal support document on social cost of H2, barriers that should lower over time. Eventually, ships will need to be (2018). Beyond road vehicles: of carbon for regulatory impact zero-emission if the IMO is to achieve Survey of zero-emission technol- analysis under Executive Order 12866: its goal of eliminating GHG emis- ogy options across the transport Application of the methodology to sions from the international shipping sector. International Council on estimate the social cost of methane Clean Transportation. Retrieved from sector. The Arctic may be the natural and the social cost of nitrous oxide. https://www.theicct.org/publications/ U.S. Environmental Protection Agency. showcase for ZEVs that use renewable zero-emission-beyond-road-vehicles. Retrieved from https://www.epa.gov/ energy because they obviate the spill sites/production/files/2016-12/docu- risk, eliminate black carbon emissions, Lloyds Register and UMAS. (2018). Zero- ments/addendum_to_sc-ghg_tsd_ and avoid GHG emissions. emission vessels 2030: How do we get there? Retrieved from https:// august_2016.pdf. www.lr.org/en/insights/articles/ Yumashev, D., van Hussen, K., Gille, J., 6. References zev-report-article/. and Whiteman, G. (2017). Towards a balanced view of Arctic shipping: American Bureau of Shipping. (2012). LNG Minnehan, J. J. and Pratt, J. W. (2017). Estimating economic impacts of emis- fuel systems: Certification & approval. Practical application limits of fuel sions from increased traffic on the Retrieved from http://www.glmri.org/ cells and batteries for zero emission downloads/focusAreas/presentations/ vessels. Sandia National Laboratories. Northern Sea Route. Climatic Change, gumpel2-2012.pdf. Retrieved from https://energy.sandia. 134:143-155. gov/wp-content/uploads/2017/12/ Comer, B., Olmer, N., Mao, X., Roy, B., Zukunft, P. (2017). Keynote Address. SAND2017-12665.pdf. and Rutherford, D. (2017). Prevalence Presented at the 7th Symposium on of heavy fuel oil and black carbon Olmer, N., Comer, B., Roy, B., Mao, X., and the Impacts of an Ice-Diminishing in Arctic shipping, 2015 to 2025. Rutherford, D. (2017). Greenhouse gas Arctic on Naval and Maritime International Council on Clean emissions from global shipping, 2013- Operations. U.S. National Ice Center Transportation. Retrieved from https:// 2015. International Council on Clean and U.S. Arctic Research Commission, www.theicct.org/publications/preva- Transportation. Retrieved from https:// Washington, DC. Retrieved from lence-heavy-fuel-oil-and-black-carbon- www.theicct.org/publications/GHG- https://www.star.nesdis.noaa.gov/ arctic-shipping-2015-2025. emissions-global-shipping-2013-2015. Ice2017/program.php.

12 INTERNATIONAL COUNCIL ON CLEAN TRANSPORTATION WORKING PAPER 2019-03