An assessment of technologies for reducing regional short-lived climate forcers emitted by ships with implications for Arctic shipping

Abstract

Evaluating potential cost–effectiveness of abatement technologies in parallel with emerging scientific evidence is important for better management decisions related to integrated environmental problems. This article evaluates six black carbon (BC) emissions reduction technologies for marine engines, including the net effect on a set of short-lived climate forcers from marine diesel combustion. Technologies evaluated include slide valves, water-in-fuel emulsion, diesel particulate filters, low-sulfur fuel, emulsified fuel and sea water scrubbing. Cost–effectiveness values for these technologies implemented alone or in combination are reported in terms of US$/metric ton (mt) for BC, particulate matter and CO₂ equivalents (CO₂eq), with the latter including CO₂ emitted directly due to parasitic fuel use and both warming and cooling short-lived climate forcers affected by control technology performance. The article finds that the most cost-effective strategy evaluated (i.e., the least US$/mt CO₂eq) occurs through a combination of technologies that achieve an approximate 60% BC reduction. Using the example of Arctic shipping, the cost to achieve this 60% BC reduction target may be approximately US$8–50 million per year, avoiding approximately 9–70 million metric tons CO₂eq per year. Uncertainty analysis using Monte Carlo simulation is used to demonstrate the robustness of these results.

Figure 1.  Two implicit triangle distributions from Table 2.

(A) BC emission rate (g/kWh) and (B) OC emission rate (g/kWh).

Figure 2.  Cost–effectiveness of black carbon control technologies, showing dominance of combination case.

BC: Black carbon; MDO: Marine distillate oil; SWS: Sea water scrubbing; WiFE: Water-in-fuel emulsion.

Figure 3.  Sensitivity of combination case technologies in terms of black carbon reduction cost–effectiveness.

BC: Black carbon; DPF: Diesel particulate filter; DFO: Diesel fuel oil; GWP: Global warming potential; PM: Particulate matter; SFOC: Specific fuel oil consumption; SWS: Sea water scrubbing; WiFE: Water-in-fuel emulsion; ULSD: Ultra low-sulfur diesel.

Figure 4.  Cost–effectiveness of particulate matter control technologies.

MDO: Marine distillate oil; PM: Particulate matter; SWS: Sea water scrubbing; WiFE: Water-in-fuel emulsion.

Figure 5.  Particulate matter reduction performance of control technologies, showing dominance of combination case.

mt: Metric ton; MDO: Marine distillate oil; PM: Particulate matter; DPF: Diesel particulate filter; WiFE: Water-in-fuel emulsion;

Figure 6.  Comparison of 20- and 100-year reductions in global warming potential across technologies.

DPF: Diesel particulate filter; GWP: Global warming potential; MDO: Marine distillate oil; SLCF: Short-lived climate forcer; WiFE: Water-in-fuel emulsion.

Figure 7.  Sensitivity of combination case technologies in terms of 20-year GWP cost–effectiveness.

BC: Black carbon; DPF: Diesel particulate filter; GWP: Global warming potential; MDO: Marine distillate oil; OC: Organic carbon; PM: Particulate matter; SFOC: Specific fuel oil consumption; SWS: Sea water scrubbing; WiFE: Water-in-fuel emulsion.

Figure 8.  Sensitivity of sea water scrubber technology in terms of 20-year global warming potential cost–effectiveness.

BC: Black carbon; GWP: Global warming potential; MDO: Marine distillate oil; O&M: Operation and maintenance; OC: Organic carbon; PM: Particulate matter; SWS: Sea water scrubbing.

Figure 9.  Optimized combination short-lived climate forcer targets in terms of black carbon and particulate matter US$/mt.

BC: Black carbon; mt: Metric ton; PM: Particulate matter.

Figure 10.  Optimized combination short-lived climate forcer targets in terms of global warming potential effectiveness.

GWP: Global warming potential; mt: Metric ton.

Integrated climate management depends in part upon jointly understanding primary GHGs together with non-GHGs that may both contribute to climate change, either regionally or globally. Pollutants that contribute significantly to climate forcing, but over much smaller time scales than CO₂, have been defined collectively as short-lived climate forcers (SLCFs). Combustion sources emit a subset of SLCFs that have much shorter residence lifetimes in the atmosphere than long-lived GHGs, such as CO₂, which has a lifetime on the order of centuries. These SLCFs may also contribute to regional climate impacts through interactions with clouds, ice or snow. Black carbon (BC) is chief among SLCF pollutants, at least for diesel engines.

Several factors are relevant to management considerations to control SLCFs. Scientists are focusing on these SLCFs in order to better understand impacts from long-term climate forcers, including CO₂[1]. Policy makers are increasingly recognizing that controlling SLCFs can be achieved by many technologies currently employed for reducing other pollutants, such as particulate matter (PM) or sulfur oxides (SO_x). Some policy advocates are considering whether controlling SLCFs may produce early benefits while longer term energy–climate decoupling strategies develop [2]. Among SLCFs, BC emissions have been quantified in first-order assessments being included in recent global inventories [3,4].

Control of SLCFs may not always mitigate climate change, and, like all SLCFs, the benefits of BC control are not uniform regionally (or temporally) [5–7]. Therefore, efficient BC management strategies may favor technologies that can be operated intermittently to reduce costs and/or to assess the benefits of technologies only when operated in a sensitive region. Either or both of these conditions can improve the cost–effectiveness of a control strategy. Mobile-source diesel engines, such as heavy-duty trucks, locomotives and marine vessels are of special interest because they are important sources of SLCFs, especially BC among other PM species.

The Arctic deserves special attention with regard to BC (and SLCFs generally) given northern hemisphere population densities, energy and economic activity, and international transportation. The polar regions are especially responsive to SLCFs and are sensitive to warming effects of light-absorbing compounds (LAC) such as BC [5,8–11]. In addition to substantial anthropogenic activity in the northern hemisphere, atmospheric transport mechanisms carry emissions into the Arctic region [12,13]. Moreover, ships of all types transit Arctic routes making some approximately 15,000 voyages annually, and international shipping routes transit northern ocean latitudes within and adjoining sensitive Arctic regions [13]. Recently, assessments have described current in-Arctic emissions from marine diesel engines and suggest that future growth in these emissions may be substantial [14].

Emissions from oceangoing shipping are a recognized contributor to regional and transboundary pollution [15–17]. Shipping (both international and domestic) has also been identified as a significant source of climate-forcing emissions, including CO₂, SO× and PM constituents such as BC and organic carbon (OC) [18–23]. However, to date, there are no regulatory measures in place to manage carbon emissions (either CO₂ or BC) from ship engines, although scientific and technology analyses and international policy efforts are focused on ship CO₂ reductions [24,25].

Carbon management decisions must recognize and account for significant uncertainties in considering whether control of SLCFs from shipping activity may be merited, feasible and cost effective. The potential contribution by ships to anthropogenic BC emissions quantified in first-order assessments are being included in recent global inventories [3,26]. Climate change impacts of SLCF emissions from shipping are now being studied. Scientists are recognizing and quantifying the warming importance of compression ignition (diesel) combustion emitting high particle numbers and small particle sizes – which is especially characteristic of high-temperature, high-pressure compression-ignition (diesel) combustion of residual fuels typical of oceangoing slow- and medium-speed marine engines [27–29]. Quantifying the contribution of shipping within sensitive regions such as the Arctic will help determine the value of managing regional operations versus implementing SLCF measures under all operating conditions.

This article assesses potential BC abatement technology of SLCFs within the broader SLCF management context. Cost–effectiveness values for BC control technologies implemented alone or in combination are reported in terms of US$/metric ton(mt) for BC, PM, and CO₂ equivalents (CO₂eq), with the latter including parasitic fuel loss and global warming potential (GWP) values of other important SLCFs affected by technology use. Given that incomplete information exists for the environmental sensitivities, technology performance and costs, uncertainty analysis using Monte Carlo simulation demonstrates the robustness of technology comparisons, and evaluates the importance of uncertain inputs.

We evaluate control costs and potential emissions reductions using current information about one sensitive region, the Arctic, where the largest transportation source is shipping. However, the technology assessment methodology and insights are not limited to Arctic strategies for several reasons, the first being that the general scientific evidence motivating abatement of SLCFs is still emerging and may include other regions where ships operate. This kind of integrated technology assessment may be needed to inform good carbon management decisions where achieving co-benefits (or avoiding conflicts) between pollution reduction and climate-scale goals are important.

Analytical decision framework

We construct a cost–effectiveness decision framework within which to compare and rank various control technologies, because this is often used in management decision making to support the selection of alternatives that maximize a given performance at least cost. Such an implementation context differs from benefit–cost analyses (BCA), which support decisions about whether to pursue a given strategy toward achieving climate change mitigation goals. Formally, a cost–effectiveness criterion is applied after “a desired and obtainable objective is selected, perhaps on non-economic grounds” [30]. We use this framework because it can provide the carbon management insights on technology performance despite the evolving scientific understanding of the potential impacts and mitigation value of reducing Arctic SLCFs. We also model this framework across many input samples using Monte Carlo simulation. This work applies decision support tools, including simulations and uncertainty assessment using Microsoft Excel^® and Oracle Crystal Ball^® products.

▪ Energy, costs emissions & cost–effectiveness calculations

We construct a typical ship activity profile that defines parameters common to all control technologies; for example, emissions rates, vessel characteristics, fuel prices and activity inputs such as operating days. The form of this function is general and used in many ship inventories [24,31,32]. This includes an explicit parameter describing operating time in a region that may require control, such as the Arctic, as a percentage of total annual operating time. This allows evaluation of cost–effectiveness for vessels dedicated to regional service versus vessels transiting through a sensitive region as part of a longer voyage. These parameters can define a baseline prior to control technology application from which control costs and emissions reductions can be estimated.

A general activity-based equation to estimate energy use (kWh) during operations is:

Equation 1

where Г is the kWh operating within a control zone (e.g., while in the Arctic region); Ф is the engine power in kW; Λ is the average engine load fraction (assumed to average 0.72 for this study); and t is the hours operating in the region of potential control (e.g., within the Arctic region)

This equation can be modified to depict the cost of energy for these operations by multiplying Equation 1 by the specific fuel oil consumption (SFOC) in g fuel/kWh, converting to metric tons (mt), and then multiplying by the price of fuel in US$/mt. Similarly, the emissions produced among any of several pollutants, including CO₂, SO₂, PM, OC and BC, is estimated by multiplying Equation 1 by the emissions rate in g/kWh.

The cost of control technology (T) is given as:

Equation 2

where T is the annual technology cost of control (US$); O is the operating and maintenance costs (nonfuel), in US$/kW; Ф is the engine power in kilowatts; K is the annualized capital costs at interest rate i using technology costs and lifetime details; F is the change in fuel consumption cost (if any) incurred during technology operation, identified as a ‘fuel penalty’ in later discussion and tables.

The estimated reduction in pollution (P_red), or avoided emissions, are estimated from:

Equation 3

where Г is from Equation 1; E is the emissions rate in g/kWh; and C is the percent reduction from the control technology.

Cost–effectiveness, then is the ratio of technology cost (T) to pollution controlled P_red, formally:

Equation 4

▪ Uncontrolled SLCF emissions factors

Baseline emissions factors (in the absence of control technologies) are estimated assuming marine distillate oil (MDO) use. International Maritime Organization standards entering into force will, among other things, mandate lower sulfur fuels that may induce some changes from heavy fuel oil to MDO [25]. These baseline emissions factors have evolved from recent studies that have examined pollutant emissions from ships, including direct measurement from the stack or observation of ship plumes. These studies present ranges, either through reporting standard deviations or confidence intervals, or by superposing the various studies, to assemble a multi-engine data set.

Estimated BC emissions factors based on observations and measurements are varied by an order of magnitude, from 0.08 to 0.97 g/kg fuel (equivalently from 0.015 to 0.18 g/kWh in energy units), depending on vessel type or on other factors still being studied [29,33–36]. Although these studies for the most part observed vessels using heavy fuel oil, these emissions factors may be applicable to vessels using MDO, since BC emissions are not well correlated with fuel sulfur content [29].

Estimated OC emissions factors also vary among reported measurements, although average at approximately 1.1–1.60 g/kg fuel (equivalently 0.20–0.30 g/kWh)[29,34–36]. Lack et al. observed a positive correlation between OC emissions and fuel sulfur content, with vessels using more than 0.5% sulfur (S) fuel emitting on average 0.30 g OC/kWh, and less than 0.5% sulfur fuel emitting approximately 0.17 g OC/kWh [29]. However, OC emissions vary by vessel and engine type, so we examine a range of emissions factors.

Particulate matter and SO₂emissions factors are obtained from the Second International Maritime Organization GHG Study, which summarized average values from observations and measurements assuming use of MDO (0.5% sulfur content) fuel. This study uses a PM emissions factor of 1.1 g/kg fuel (equivalently 0.2 g/kWh), and a SO₂emissions factor of 10 g/kg fuel (∼1.9 g/kWh) [24].

When use of low-sulfur marine fuel is assumed, at 0.05% sulfur (500 ppm sulfur), emissions factors are derived consistent with observed relationships between emissions and fuel sulfur content; BC emissions are unchanged compared with MDO, while OC, SO₂ and PM emissions are reduced accordingly. For 0.05% sulfur fuel, OC emissions are assumed at 0.5 g/kg fuel (∼0.09 g OC/kWh), SO₂ at 1g/kg (∼0.2 g SO₂/kWh), and PM at 1.6 g/kg fuel (∼0.3 g PM/kWh).

▪ Control cost estimation

Six control technologies evaluated independently or in combination include: slide valves, water-in-fuel emulsion (WiFE), diesel particulate filters (DPFs), low-sulfur distillate, emulsified fuels and sea water scrubbers. The performance and estimated costs of each technology are discussed later in this article. Methodologically, capital equipment and installation costs were converted to US$/kW and US$/kWh based on assumptions about equipment lifetimes, annual hours of operation and discount rate, and are shown in Table 1. Since these costs are vessel dependent, we conduct our analysis for a range of vessels, the characteristics of which are shown in Table 2. We also use these assumptions to generate operation and maintenance (O&M) costs. Costs for low-sulfur fuel are translated from US$/mt to US$/kWh using assumptions about the energy content of fuel and engine efficiency. All costs were developed from various data sources reported within the last decade, and are amortized using an interest rate of 7% (then varied with Monte Carlo simulation as discussed later). Costs from various years are presented in constant-dollar terms for 2008.

Parasitic fuel costs (i.e., fuel penalties) represent the reported costs of additional fuel needed to operate emissions control equipment. These are termed ‘parasitic’ by operators because they require energy for purposes that do not directly contribute to cargo transportation service except in terms of environmental performance. Fuel penalties are typically provided in terms of the percentage of additional fuel needed for operation. We combine those estimates with estimates of fuel costs and energy content to calculate a US$/kWh cost for fuel penalties. Once brought to a common reference unit (US$/kWh), all these values are summed to calculate total US$/kWh or each emissions reduction strategy.

▪ Modeling uncertain input parameters

Input parameters will vary with vessel size, service and market conditions (e.g., spot fuel prices); for example, some technologies may benefit from installation on vessels with very large engines, given economies of scale, while other technologies may be most cost effective in vessels with slower service speeds and smaller engines. Given that cost–effectiveness varies among these characteristics, we define lower and upper limits for activity profile parameters. In sampling simulations, the mid-value may be considered to be a ‘best-guess’, typical, or most likely value, and the bounding values may be less likely.

Table 2 shows the value ranges for parameters underlying each technology cost–effectiveness analysis. The input ranges are ‘first-order distributions’ within which technology performance may vary; these ranges derive from and are consistent with a number of previous sources [14,24,26,31]. Some input values can be considered more likely than others, given current and emerging information; in such a case a ‘triangle distribution’ is a convenient means of representing uncertainty that does not imply precision not yet available [30]. For example, where the mid-value falls between low and high values, a triangle distribution may resemble a normal or lognormal distribution of possible values; where the low and medium values are common and the high value differs, this would describe a wedge-shaped distribution favoring lower range values. Figure 1 illustrates two parameters: uncontrolled BC emission rate and uncontrolled OC emission rate. These allow for consideration of the sensitivity of control technology cost and performance to variable but common inputs. Technology-specific value ranges for control alternatives are discussed in the next section.

We do not address various constraints that may exist when considering the implementation of emissions controls on vessels. For example, there may be cases where certain emissions control devices may not be feasible or may cost more to install (for instance DPFs have been used primarily with four-stroke engines – their effectiveness with two-stroke engines is unknown) or where emissions control devices may not operate at full efficiency (e.g., due to vessel operating conditions). Such constraints may reduce the cost–effectiveness of the control device. The relative importance of certain cost items (e.g., capital costs) with respect to other cost items (e.g., parasitic fuel costs) will vary depending on vessel installed power (kW) and annual use (kWh). For this analysis, we presume that these issues are either less important than other uncertainties, and/or able to be addressed through technology design stages.

Technologies to reduce SLCF emissions in shipping

To evaluate the effectiveness of BC control, we consider technologies that have been shown to control PM emissions, and which evidence indicates will reduce BC emissions to some extent. A number of approaches have been shown to reduce PM emissions from diesel engines and ships, including switching to low-sulfur fuels, after treatment technologies and engine process modifications. While PM control performance has been estimated for many of these measures, the effectiveness of these technologies in controlling specific constituents (e.g., SLCFs such as BC or OC) is not known for all measures. The relative control effectiveness of individual PM components is important, since some PM reduction strategies (e.g., low-sulfur fuels and diesel oxidation catalysts) do not necessarily provide measurable BC reductions even when significant PM reductions occur.

As mentioned earlier, the focus here is to explore the overall effect that BC emissions reduction technologies would have on climate forcing, when considering: the BC emissions reductions themselves; fuel penalties associated with technology use, which increase CO₂ emissions; and the impact of reducing other SLCFs associated with BC control (e.g., OC and SO₂) that may have cooling effects based on their GWP. For each emission reduction measure, we present the following: potential PM and BC reductions and applicability for use in marine vessels, cost estimates derived from the best available data and other climate-forcing emissions reductions associated with the approach. We use available BC- and OC-specific performance data; we assume that the PM performance level applies to BC and/or OC if such data are not available, and if we have no reason to believe that BC and/or OC reductions would differ from PM. By assessing the contribution of uncertain assumptions to the uncertainty in the cost–effectiveness, we offer insights into the importance of testing and demonstration for verification compared with improved understanding of other uncertain inputs.

▪ Slide valves

Slide valves replace conventional fuel valves, facilitating more complete combustion at lower peak-flame temperatures and thus reducing NO_x and PM [37]. Slide valves are reported to reduce PM emissions by approximately 25% [38,39], although estimates of 50% PM control have been presented [40]. BC control performance estimates have not been reported; we assume that slide valves will reduce PM and BC similarly. Slide valves are already in use in a good portion of the shipping fleet, as new engines manufactured by MAN B&W (an engine technology company that represents a large share of installed power in the shipping market) come equipped with slide valves. Table 3 presents technology-specific input ranges for slide valves.

Cost estimates for the MAN B&W slide valves were obtained from Entec [37]. These costs include: incremental costs above conventional valves; installation costs; and retrofit design and installation expenditures for vessels older than 15 years [37].

Some researchers have suggested that there are no O&M costs associated with slide valves, and, in fact there may be benefits such as reduced fuel oil consumption for lubrication [37], and fuel economy improvements may result in low- and mid-speed engines. However, slide valves can be associated with a fuel penalty of up to 2% when low-NO_x behaviors are employed [38]. Other slide valve performance attributes are defined to include:

	▪ CO2: fuel economy improvements will decrease CO2 emissions, while fuel penalties will increase CO2 emissions (range: -1 to 1%);
	▪ OC: slide valve impacts on OC are unknown; we assume OC control is equal to PM control;
	▪ SO2/sulfates: slide valves have no known impacts on SO2 or sulfates.

▪ Water-in-fuel emulsification on demand

Water-in-fuel emulsification on demand involves introducing water into fuel prior to injection into the combustion cylinder. Unlike other water emulsification technologies or pre-emulsified fuels, WiFE can be deactivated when high power is needed [41, Hutchingame E, Pers. Comm.].

WiFE reportedly reduces PM emissions by two- to three-times the water content – so a 10% water emulsion would equate to 20–30% PM reductions, while 30% emulsion would result in 60–90% reductions [Hutchingame E, Pers. Comm.]. WiFE’s BC reduction performance has not been measured [Hutchingame E, Pers. Comm]. As noted above, a study examining PM reductions from water-emulsified fuel use in off-road equipment found that BC percentage reductions were actually higher than those of total PM [42]; therefore, it is reasonable to assume that BC reductions from WiFE will meet or exceed total PM reductions. Assuming that the PM:BC ratio observed in Noll et al. applies to WiFE, at 30% PM control [42], BC would be controlled by approximately 45–50%. Since the amount of water emulsion used and the duration used may vary with WiFE, we examine a range of PM and BC control performance assumptions, listed in Table 4.

On-demand WiFE capital cost estimates were provided by Sea to Sky Pollution Solutions [Hutchingame E, Pers. Comm.]. Capital costs of US$250,000 assume equipment and installation on a typical container ship and are expected to be similar for all main engines; auxiliary engine costs may be 40–60% lower (US$100,000–150,000), and cost reductions may be observed as the technology enters the maritime market. Fuel penalties of up to 1% are anticipated for newer vessels, while older vessels may not experience a fuel penalty, as WiFE can improve combustion efficiency. Other WiFE performance attributes are defined to include:

	▪ CO2: when water-in-fuel emulsification is associated with a change in fuel oil consumption, CO2 emissions will increase accordingly (i.e., best-estimate of 1%; range: -1–2%);
	▪ OC: assuming the PM:OC relationship observed by Noll et al.[42] applies to water-in-fuel emulsification, OC emissions may be increased up to 40%;
	▪ SO2/sulfates: water-in-fuel emulsification impacts on SO2 or sulfates are not documented, although they may be decreased slightly. We assume no change in fuel sulfur content and negligible change in SO2.

▪ Diesel particulate filters

Diesel particulate filter (DPF) systems are effective in controlling PM (achieving 70–95% total PM reductions), and are particularly effective at controlling BC emissions, achieving 95–99% BC reductions by mass [43,101].

There are two categories of DPFs: active, which require fuel burners or electric regeneration of filters; and passive, which use catalysts to regenerate without an external energy source [44,102]. Most DPFs demonstrated to date cannot be used if fuel sulfur content exceeds approximately 500 ppm, since DPFs become ineffective or may actually increase sulfate and PM emissions. Catalyzed DPFs become ineffective or counterproductive at far lower fuel sulfur content (15–50 ppm), and so are not applicable for use in marine applications [44,45,102, Roberts J, Pers. Comm.]. Marine DPFs have been designed to tolerate higher sulfur fuels (up to 1000 ppm – although less than 500 ppm is ideal); however, the use of conventional heavy fuel oil or MDO with higher sulfur content is not possible with these systems [Roberts J, Pers. Comm.]. We assume no size limitations for marine DPF systems, although applications with engines greater than 6000 kW were untested as of 2009 [Roberts J, Pers. Comm.]. Table 5 summarizes technology-specific inputs for DPF evaluation.

Costs of DPFs include capital costs for equipment and installation, O&M costs (filter regeneration and replacement) and fuel penalty costs [40,46,103]. Capital costs of DPFs are estimated at approximately US$22/kW, and O&M costs are estimated at US$19.6/kW. Fuel penalties have been estimated at 1–6%, with active systems having fuel penalties in the higher range [40,45,46,103]; we assume DPF fuel penalties of 1 and 4%.

Costs of switching to a low-sulfur fuel must also be considered for DPF use in marine applications, since DPFs are not compatible with MDO. We calculate the combined costs and emissions reductions of the fuel switch combined with the DPF. Costs of fuel switching were calculated using fuel price histories from the US Department of Energy [104]. We used prices for low-sulfur No. 2 diesel fuel (maximum 500 ppm sulfur). We consider the incremental cost of this fuel compared with MDO in 2008, using a range of average fuel prices from the following years: 2000 (low) to 2007 (high). Fuel cost assumptions are shown in Table 2; this equates to an incremental cost of approximately US$60–210/ton fuel. To illustrate potential costs of DPFs if fuel switching was not required (e.g. if DPFs compatible with MDO were developed, or if regulatory mandates required use of 500ppm fuel in shipping), we also examine a case with $0 incremental fuel cost.

Although DPFs are associated with a fuel penalty, the use of lower sulfur fuel may result in slight fuel economy gains (2%) since cleaner fuel is higher in energy content per ton; we also include the change in fuel consumption in our cost calculations. Table 5 summarizes technology-specific DPF inputs. Other DPF performance attributes are defined to include:

	▪ CO2: DPFs are associated with a fuel penalty. However, use of lower sulfur fuels results in a slight fuel economy gain, and there are also fewer emissions of CO2 per ton of fuel consumed during operation. The net effect is a slight increase (1%) in CO2 (with 4% DPF fuel penalty);
	▪ OC: DPFs remove approximately 50–90% of OC, with active DPFs in the lower end of the range;
	▪ SO2/sulfates: the sulfur content of fuel required for DPF operation is much lower than of MDO, so SO2 emissions and sulfate PM are reduced considerably. SO2 emissions of 0.05% sulfur fuel are 90% lower than MDO SO2 emissions.

▪ Emulsified fuel

Emulsified fuels (EMFs) are stable mixtures of fuel, water and additives for emulsification and stabilization; EMFs reportedly reduce PM emissions by up to 50–63% [40,47,48]. In a San Francisco Bay Water Transit Authority (SFBWTA) demonstration project, the use of the EMF PuriNOx™ in a ferry vessel reduced PM emissions by 42% [49]. PuriNOx has been verified by the US Environmental Protection Agency (EPA) for PM reductions in nonroad applications, although at a lower control performance [50].

PuriNOx can be used in new and old engines, does not require engine modifications, and can be used with or without aftertreatment technologies [49]. Although estimates of BC emissions reductions from the use of EMF in marine applications have not been reported, BC control (70–85%) was higher than total PM control (44–57%) from PuriNOx use in off-road equipment [42]. Therefore, it is reasonable to assume that EMF BC control will meet or exceed total PM reduction efficiency. Table 6 presents technology-specific input ranges for emulsified fuels. There is a price premium associated with EMF: for instance, SFBWTA paid a 14–18 US cents/gallon premium for PuriNOx fuel [49]. This equates to US$42–54/ton fuel.

Fuel penalties of up to 1.5% are also associated with EMF [40,47,105]. This is primarily due to the energy penalty per volume of fuel when lower-Btu emulsifiers displace hydrocarbons. The SFBWTA demonstration project reported a 15% fuel penalty when using PuriNOx in a ferry vessel [49], which presumably includes the water content of the fuel (15% more gallons of purchased product were consumed). For the purposes of this analysis we assume an approximately 1.5% increase in fuel consumption (i.e., MDO), but a 15% increase in gallons of EMF consumed; this accounts for increased financial cost of EMF consumption, but does not penalize EMF for any CO₂ emissions, which would be associated with fuel combustion only. As with DPFs, we use average MDO fuel prices (in 2008) from years 2000, 2004 and 2007. Other emulsified fuel performance attributes are defined to include:

	▪ CO2: EMF is associated with an approximately 1.5% fuel penalty, and thus an equal increase in CO2 emissions;
	▪ OC: according to a study of EMF use in off-road equipment, EMF actually increased OC emissions by 3–42% compared with baseline diesel fuel, although BC and total PM were reduced [42];
	▪ SO2/sulfates: EMF does not have any verified impacts on SO2 or sulfates.

▪ Sea water scrubbing

Sea water scrubbing (SWS) involves reducing ship emissions by exposing flue gases to seawater spray or other physical contact (e.g., bubbling) to dissolve traces of water-soluble components of the gas stream, including pollutants [51]. SWS reduces PM emissions by 25–80%, as verified in a recent demonstration project that showed 57% reductions in PM [52,53,106]. Although measurements of reductions in the BC fraction of PM have not been explicitly addressed, recent research indicates that SWS may reduce PM_2.5 (of which BC is a component) by 75% [24,54]. A trial of the EcoSilencer SWS system on an auxiliary engine using heavy fuel oil measured 98% reductions in PM₂, 74% reductions in PM_1.5, 59% reductions in PM₁ and 45% reductions in PM_0.05[52]. This research implies that SWS is at least as effective in controlling BC as in controlling PM. Table 7 presents technology-specific input ranges for sea water scrubbers.

Costs of SWS include capital costs for equipment and installation, and O&M costs (e.g., pump operation, maintenance and sludge disposal). Costs, obtained from Entec (UK) [52], were estimated from prototypes and expert elicitation. We convert Entec’s 2005 estimates into 2008 by first converting to 2005US$ using a conversion factor of 1.245, and then converting 2005 to 2008 by applying a conversion factor of 1.088. Significant portions of this cost go to pumping sea water through the scrubber [55–57]. This produces CO₂ emissions as discussed below. Other SWS performance attributes are defined to include:

	▪ CO2: SWS systems are associated with fuel penalties or changes in CO2 emissions, owing to the energy required to pump seawater through the exhaust treatment system. These fuel penalties are not well described by vendors or in public reports; however, they can be estimated to range between the optimistic estimates of approximately 1.5% increase in fuel and estimates that correlate to O&M estimates (as high as 10%). For this analysis we use a range of 1.5–5% to imply that improvements to the worst case estimate would prevent fuel penalty of SWS operation to exceed 5%.
	▪ OC: impacts on OC are unknown; we assume OC control is similar to but slightly less than PM control;
	▪ SO2/sulfates:SWS systems reduce SO2 emissions by 68–95% [51–53].

▪ Potential for combination synergies among technologies

Control technologies may be combined to provide synergistic control effectiveness. To explore whether the combination of less costly but less effective technologies could achieve performance similar to more costly technologies, we model a combination case in which slide valves and WiFE are combined. Other combinations are possible, and are discussed later in this article, but this case demonstrates that the two least-costly technologies together can achieve BC reductions that exceed all but DPF technologies and that the performance ranges overlap with DPF reduction capabilities.

Estimating SLCF emission control costs & cost–effectiveness

▪ Comparison of SLCF control technology effectiveness in terms of BC & PM

Using Equations 1–4, the cost–effectiveness of each technology is estimated in terms of BC and PM control (US$/mt). Figure 2 presents the cost–effectiveness of BC control for each technology, with the combination technology case (slide valves plus WiFE) represented in columns and each other technology represented by a separate line graph. These graphs report the statistical distributions of 10,000 Monte Carlo samples from the uncertainty ranges and sets of equations described above. Clearly, the cluster of costs per mt closest to US$0 (slide valves in red, WiFE in blue, combination case in grey) is preferred as more cost effective than technologies with distributions across higher costs per mt (e.g., DPF, SWS and EMF).

Figure 3 presents a rank correlation of inputs to results for the combination case, indicating the correspondence and significance of uncertainty or sensitivity of input parameters and cost–effectiveness results. Rank correlation coefficients are arranged in decreasing order of absolute value [58]. As shown, time in a potential control region is important to cost–effectiveness; this describes periods when the reductions are counted as beneficial and/or when equipment can be operated intermittently. For cost–effectiveness, increasing the time spent controlling emissions is negatively correlated with the cost–effectiveness ratio, given the fixed costs of technology implementation. In addition, uncertainty in the current understanding of the rate of BC emissions significantly influences the cost–effectiveness performance. Additional minor influences include fuel penalties, fuel prices and control efficacies, followed by specific fuel–oil consumption rates among others.

Cost-effective BC control is an indirect metric for technology performance; it does not, for example, include the associated PM reductions of the technologies. Figure 4 presents the estimated cost–effectiveness for PM reduced (US$/mt PM), inclusive of BC and other PM species. The PM cost–effectiveness metric allows for direct comparison with technologies considered to be cost–effective for controlling PM from other mobile sources (albeit to achieve air quality and health goals). The combination case of slide-valves and WiFE (grey columns) outperforms the individual technologies and, as expected, the inputs sensitivity follows the same rank correlation with similar values. Figure 5 illustrates the dominance of the combination case in terms of greater likelihood of reducing more emissions with a longer ‘tail’ distribution across higher PM reduction values.

▪ SLCF control effectiveness in terms of CO₂ impacts

We also consider impacts on CO₂eq emissions due to BC control strategies – accounting for not only the BC reductions themselves, but also for potential increases in CO₂ emissions (primarily due to fuel penalties) and for effects on other SLCFs. The overall impact of each control technology is evaluated using a 20- and 100-year GWP adjustment. This is similar to evaluations of BC reductions from Class 8 truck vehicles [59]. We assign BC a 20- and 100-year GWP of 2200 and 680, respectively [4]. For the Arctic-specific example discussed in this article, we use Arctic adjustments for GWP (20-year) ranging between 4500 and 7500 [7] (no adjustments are made to the GWP [100-year] values because we do not find any literature that has estimated regionally specific GWP [100-year] ratios for places such as the Arctic). Modest ranges are defined for the uncertainty in negative GWP (20-year) for OC and SO₂, although these uncertainties in the Arctic may be larger. The range of GWP values assumed in our analysis are presented in Table 8.

Using the full emissions effects of BC control strategies across pollutants, the cost–effectiveness of each strategy is presented in terms of US$/mtCO₂eq (Figure 6), where the SWS is presented in grey columns and the combination case of slide valves and WiFE is presented using a black line. The important observation here is that all technologies except SWS produce positive benefits in terms of GWP over most of their distributions. Sea water scrubber technologies do not achieve the greatest BC reductions, and SWS operations are energy intensive due to the water that must be pumped through the stack treatment system. This produces increased CO₂ emissions associated with the parasitic energy demand of SWS pumps, which offsets some or all of the potential GWP benefits of removing BC. From a carbon management perspective, this suggests that using SWS technology to reduce SLCFs may be less preferred than other technologies. Of course, this does not suggest that potential health and environmental benefits are insufficient to choose SWS as pollution control devices, merely that the climate co-benefits desired of technologies to reduce particles may favor other strategies.

Figure 7 ranks the most important inputs for the technologies combination case in terms of their influence on the CO₂eq cost–effectiveness, using the rank correlation of input ranges to output ranges. Figure 8 presents similar ranking for the SWS technology. The inputs most highly correlated with results are presented in rank order, producing a menu of prioritized input uncertainty that would be useful in pursuing improved information. Inputs related to operation rank highly for the combination case, for example, percentage of time operated in control region and fuel penalties for each technology; uncertainties in the emissions rates and GWP ratio conversion are similarly important. For the SWS technology, inputs related to SLCFs that may contribute negative climate forcing (cooling) are most important; for example sulfur and OC emissions rates. This is because slide valves and WiFE technologies can operate without assuming changes in SLCF emissions from fuel sulfur or OC. In other words, these combination technologies may selectively control BC, the primary warming SLCF from marine diesel engines.

Uncertain inputs influencing the GWP (20-year) results for the combined case also rank operating within a control region, energy use and emissions rates higher than the large uncertainty in the BC GWP (Figure 7). For example, container trips include portions of transoceanic voyages that transect the Arctic region; the entire Arctic voyages by most other ship types (icebreakers and bulk carriers, for example) are contained within region. Evaluating the cost–effectiveness of transecting vessels may be an important result, given the uncer- tainty declared around scientific understanding of SLCF effects on climate in ice and snow and transport of SLCFs from outside the region.

Table 9 summarizes the point estimates from the analysis for vessels operating in a control region 25% of the time, and Table 10 presents the same results if the vessel operated 100% of the time within a control region. (to continue with the Arctic shipping example, Table 10 may represent dedicated Arctic vessels, and Table 9 may represent vessels that serve routes that connect the Arctic region with other non-Arctic locations). These are ranked according to BC reduction, which corresponds monotonically with BC cost–effectiveness, PM cost–effectiveness, and total PM reduced. However, the effectiveness of these technologies in carbon management terms is not monotonic. Indeed, using the point-estimate values (‘best-estimates’), the SWS technology offers zero benefits and increases warming forcers on a GWP (100-year) basis, and achieves fewer GHG benefits on a GWP (20-year) basis at higher cost than all other technologies. More importantly, the combination case outperforms the individual technologies in terms of PM and GHG reductions.

SLCF management implications

Using a cost–effectiveness analysis framework to assess technology alternatives for reducing BC and other SLCFs from shipping, five important abatement management points can be discussed. First, scientific input is still required to confirm the premise that climate-scale impacts can be attributed to SLCFs from shipping, or more generally to shipping along routes that may be within atmospheric transport distance of sensitive regions such as the Arctic. For example, motivation for action to reduce SLCFs from Arctic shipping depends upon a firm scientific understanding of the role BC and other particles may have in ice- and snow-melting cycles and how these particles affect albedo from Arctic cloud formation. Moreover, this work is not a benefit–cost analysis of the merits of action to reduce SLCFs, but a technology assessment that provides cost–effectiveness values for technology strategies that may achieve SLCF reductions (e.g., BC) from ships. Given spatial variability, benefits of these reductions may be different for the same cost–effectiveness of control. For example, where densely populated regions are also in a sensitive region, controls may produce co-benefits of reducing health impacts and reducing GWP. Therefore, relatively cost-effective control strategies in urban regions outside of the Arctic may find more health benefits than climate benefits for the same abatement technology cost–effectiveness. As another example, the cost to abate would be a portion of total social costs that may be implied in a marginal abatement cost curve to model carbon price fundamentals. Thus, abatement technology insights reported here are necessary to support SLCF climate and health decision making, but may not be sufficient evidence for policy action without additional evaluation of benefits.

Second, technologies may be available for rapid action to control SLCFs and will be required to achieve cost-effective reduction targets. Technology performance values used in this analysis require further validation, and some of these technologies need to be developed for market-ready application. However, current information suggests that these technologies may be among the set of cost–effective strategies for reducing PM and SLFC emissions (Figure 9), although more research will be needed to fully quantify benefits. A comparison with diesel retrofit technologies shows similar cost–effectiveness of these technologies with cost–effectiveness of control technologies for heavy-duty nonroad diesel, ranging between US$18,700 and 87,600 per mt PM [60]. All ship technologies in Table 9 except SWS are within or less than this accepted cost–effectiveness range.

Third, the combination of technologies is not limited to slide valves and WiFE, but can include any combination of technologies (at least). We use the combination concept within an optimization routine using MS Excel Solver to find the combination of technologies that meet a target emissions reduction at the lowest cost. The resulting preferred ‘frontier’ can describe the most cost-effective fleet-wide technology combinations for a given reduction target in BC or PM.

Fourth, the transformation of these reductions in terms of SLCF benefits using GWP ratios indicates that there may be an ‘optimal’ control target in terms of maximizing reductions for a given cost. In other words, minimizing the cost per mt emission avoided (or maximizing cost–effectiveness) results from the combination of technologies that achieves emissions reductions to produce the lowest ratio of cost of control to abatement. This is illustrated in Figure 10, where the GWP minima occurs at approximately 60% BC control. In fact, performance consistently improves when moving from slide valves (ranging from 5–20% BC reduction) to our combination case (with performance ranging from 30–60% BC reduction) produces monotonically, but further BC controls incur direct CO₂ emissions (fuel penalties) and increasing costs to make them less cost effective.

Nonetheless, a reduction of approximately 70% can be achieved at approximately US$15–30 per mtCO₂eq (20 year), under conditions where the vessel spends 25–100% of the time in a sensitive region. From a carbon management perspective, the cost per CO₂eq (20 year) is very similar to recent spot prices for CO₂ trading (as of 13 October 2010).

Fifth, potential total abatement costs can be estimated for ships operating within a particular region, such as the Arctic, by multiplying optimal cost–effectiveness value of US$16,750/mt BC from Table 10 by annual emissions. For the Arctic example, annual emissions were recently estimated at 1200 mt/year BC under a 2020 ‘business-as-usual’ scenario [14]. The resulting cost would be approximately US$20 million per year to control Arctic shipping SLCFs and related emissions such that emissions of BC would be reduced by 63% and PM emissions would be reduced by 52%; with uncertainty ranges, the cost to control ranges from US$8 – 50 Million per year. If the number of Arctic vessels in 2020 is between 6000 and 10,000 (roughly similar to the 6000 vessels reported in the AMSA study for 2004, including fishing vessels), then this represents an annual GWP (20-year) avoidance of approximately 9–70 million mt CO₂eq per year at an average annual cost of approximately US$1200–8400 per vessel, depending on vessel size profile and operating time in a sensitive control region such as the Arctic. If only vessels similar to those studied here were included, only several hundred vessels would require controls and the per-ship annual cost of control would be in the range of approximately US$25–35,000 with an upper range of approximately US$55,000–115,000 for the larger ship sizes.

Conclusions

Addressing cumulative GHGs in terms of mitigation of climate change is a related but separate climate management issue to reducing SLCFs for near-term climate and/or environmental/health co-benefits. Climate policy may find value in evaluating both given limited resources and regionally sensitive areas such as the Arctic. The management implication of this work is that cost-effective technologies can reduce SLCFs with net GWP benefits. This result has direct implications for addressing SLCF emissions from Arctic shipping. This result is robust even given the significant uncertainties in technology performance, cost and environmental benefits. A second insight is that some pollutant control technologies, such as SWS, may offer few co-benefits in a climate-coupled emissions control strategy. More generally, the methodology to prescreen and rank the cost–effectiveness of abatement technologies for SLCFs in parallel with the emerging scientific evidence for action can support better management decisions related to integrated environmental problems.

Certain control technologies (specifically those shown to be the most cost effective) have not been sufficiently evaluated for BC control, as a specific performance within PM control. This is a limitation of the control technology performance data when considering the climate benefits of SLCF control for shipping. However, the ability to treat these with uncertainty bounds in rigorous Monte Carlo simulation is an advantage of the methodology used in this paper. We are able to provide important prospective insights using existing information, and the development of improved information can be readily incorporated into the decision framework for future refinements.

In addition, these technology abatement costs (and performances) could be compared with operational changes to see if the cost–effectiveness rankings include favoring behavior changes over technologies. For example, the cost–effectiveness of emissions abatement technologies may be compared in future work with slow steaming [61], although marginal abatement cost for behavior changes needs to include operational changes in revenue and opportunity costs.

Finally, this methodology can be repeated either prospectively for additional technologies or iteratively as new information becomes available. This offers some generalizable value to evaluate whether certain BC control technologies may support cost-effective strategies to address climate change in the short term, while simultaneously creating co-benefits of reduced PM and related health impacts.

Table 1.  Additional detail for cost estimates of short-lived climate forcers emission reduction technologies (2008).

Technology	Equipment and installation (US$)	O&M and/or fuel costs (US$)	Lifetime (years)
Slide valves	0.39–1.68/kW 270/valve 400 labor	0	2.5
Water-in-fuel emulsion	250,000 (100–150,000 for auxiliary engine)	0	20
Diesel particulate filter	21.8/kW	19.6/kW	20
Low-sulfur distillate		60–210/ton above MDO
Emulsified fuels		42–54/ton above HFO
Sea water scrubbing	160–228/kW	0.40–1.08/MWh	12.5–15

HFO: Heavy fuel oil; MDO: Marine distillate oil; O&M: Operation and maintenence.

Table 2.  Activity-based parameters common to all technologies, with value ranges.

Input parameter	Low	Best/typical	High
Baseline BC emission rate (g/kWh)	0.03	0.07	0.17
CO2 emission rate (g/kWh)	507	600	689
NOx emission rate (g/kWh)	8.96	14.66	18.49
PM emission rate (g/kWh)	0.28	0.30	1.08
SO2 emissions rate (g/kWh)	2.03	2.03	10.98
OC emission rate (g/kWh)	0.07	0.07	0.20
SFOC (g fuel/kWh)	160	188	215
MDO fuel cost (US$/mt)	226	354	595
Interest rate for capital financing (%)	4	7	10
Operating time in region(e.g., in Arctic)	0.10	0.25	1.00
Fuel cost ratio (ULSD/DFO)	1.10	1.30	1.50
Engine size (kW)	5500	10,000	49,504

BC: Black carbon; DFO: Diesel fuel oil; MDO: Marine distillate oil; OC: Organic carbon PM: Particulate matter; SFOC: Specific fuel oil consumption ULSD: Ultra low-sulfur diesel.

Table 3.  Specific input parameters and ranges for slide valve technologies.

Slide valve variables	Low	Best	High
Slide valve black carbon control (%)	10	25	50
Slide valve particulate matter control (%)	10	25	50
Slide valve organic carbon control (%)	10	25	50
Slide valve capital cost (US$/kW)	0.39	1.04	1.68
Slide valve fuel penalty (%)	-1	0	1

Table 4.  Specific input parameters and ranges for water-in-fuel emulsification technologies.

WiFE variables	Low	Best	High
WiFE black carbon control (%)	20	50	80
WiFE particulate matter control (%)	20	30	50
WiFE organic carbon control (%)	0	-20	-40
WiFE capital cost (total in US$)	100,000	150,000	250,000
WiFE fuel penalty (%)	0	0.5	1

WiFE: Water-in-fuel emulsification.

Table 5.  Specific input parameters and ranges for diesel particulate filter technologies.

DPF variables	Low	Best	High
DPF black carbon control (%)	70	85	90
DPF particulate matter control (%)	70	85	95
DPF organic carbon control (%)	50	60	70
DPF fuel penalty (%)	1	4	6
Organic carbon emissions for 500 ppm fuel (g/kWh)	0.02	0.06	0.11

DPF: Diesel particulate filter.

Table 6.  Specific input parameters and ranges for emulsified fuel technologies.

Emulsified fuel variables	Low	Best	High
Emulsified fuel black carbon control (%)	42	57	80
Emulsified fuel particulate matter control (%)	42	50	63
Emulsified fuel organic carbon control (%)	0	-20	-40
Emulsifued fuel fuel penalty (%)	0.5	1.5	2.5
Emulsified fuel incremental cost (US$/gal)	0.14	0.16	0.18

Table 7.  Specific input parameters and ranges for sea water scrubber technologies.

SWS variables	Low	Best	High
SWS black carbon control (%)	25	40	50
SWS particulate matter control (%)	50	75	85
SWS organic carbon control (%)	40	50	80
SWS SO2 control (%)	70	80	95
SWS capital cost (US$/KW)	160	194	228
SWS O&M cost (US$/MWh)	0.41	0.68	1.08
SWS fuel penalty (%)	1.5	3	5

O&M: Operation and maintenance; SWS: Sea water scrubbing.

Table 8.  Global warming potential values from literature.

GWP values	Low	Best	High
OC GWP (100-year)	-75	-75	-75
OC GWP (20-year)	-250	-200
SO2 GWP (100-year)	-40	-40	-40
SO2 GWP (20-year)	-200	-140	-80
tCO2eq GWP (20-year and 100-year)	1	1	1
BC GWP(100-year)	680	680	680
BC GWP(20-year)	2200	4500	7200

Relative cooling ratios to CO₂.

BC: Black carbon; GWP: Global warming potential; OC: Organic carbon.

Table 9.  ‘Best-estimates’ across seven SLCF control technologies and for one technology combination strategy, 25% in-region.

Technology	US$/mt BC	US$/mt PM	Total PM reduced (mt)	US$/mt CO2eq (20-year)	GHG reduced mtCO2 (20-year)	US$/mt CO2eq (100-year)	GHG reduced mtCO2 (100-year)
Slide valve	23,368	5452	0.89	5	894	39	126
WiFE (20%)	43,560	16,940	1.07	10	1869	70	260
Emulsified fuel	283,972	75,537	1.78	117	2059	1059	228
DPF (MDO vs MDO)	351,073	81,917	3.03	89	2796	1568	158
DPF (No. 2 diesel vs No. 2 diesel)	363,268	84,763	3.03	91	2812	1528	168
DPF (MDO vs No. 2 diesel)	363,268	84,763	3.03	176	2687	1970	240
SWS	735,928	91,582	2.67	335	730	0	(-153)
Combination case analysis	137,842	38,882	6.86	20	3977	149	536

Combination strategy includes slide valves and WiFE (20% water), the two least costly independent technologies combined.

BC: Black carbon; DPF: Diesel particulate filter; MDO: Marine diesel oil; mt: Metric ton; PM: Particulate matter; SLCF: Short-lived climate forcers; SWS: Sea water scrubber; WiFE: Water-in-fuel emulsification.

Table 10.  ‘Best-estimates’ across seven SCLF control technologies and for one technology combination strategy, 100% in-region.

Technology	US$/mt BC	US$/mt PM	Total PM Reduced (mt)	US$/mt CO2eq (20-year)	GHG reduced mtCO2(20-year)	US$/mt CO2eq (100-year)	GHG reduced mtCO2 (100-year)
Slide valve	5,842	1363	3.56	1	3756	10	503
WiFE (20%)	18,020	7008	4.28	4	7475	29	1038
Emulsified fuel	283,972	75,537	7.13	117	8238	1,059	912
DPF (MDO vs MDO)	121,324	28,309	12.12	31	11,184	542	633
DPF (No. 2 vs No. 2 diesel)	133,519	31,154	12.12	34	11,248	562	672
DPF (MDO vs No. 2 diesel)	133,519	31,154	12.12	115	10,749	1292	959
SWS	255,676	31,817	10.69	117	2919	0	(-612)
Combination case analysis	16,753	5144	22.57	3	11,585	21	1670

Combination strategy includes slide valves and WiFE (20% water), the two least costly independent technologies combined.

BC: Black carbon; DPF: Diesel particulate filter; MDO: Marine diesel oil; mt: Metric ton; PM: Particulate matter; SLCF: Short-lived climate forcer; SWS: Sea water scrubber; WiFE: Water-in-fuel emulsification.

Short-lived climate forcers

Pollutants that contribute significantly to climate forcing, but over much smaller time scales than CO₂. These include energy-absorbing particles, such as black carbon and energy-reflecting particles such as sulfate aerosols.

Particulate matter

Particulate matter is an inclusive term representing multiple species of nonvolatile and semi-volatile particles, in this context resulting from fossil fuel exhaust of engine combustion. Diesel particulate matter (PM) is typically in size ranges less than 2.5 µm diameter, with substantial particle numbers in the ultrafine size ranges (∼≤0.4 µm); we use the term PM without size differentiation as this characterizes ‘total PM’ from marine engines. Black carbon is one species of PM.

Cost–effectiveness

Formally, a ratio of cost to performance. In abatement technology assessments this is the ratio of direct costs of abatement to the pollution reduction, with monetary costs annualized to compare with avoided annual emissions in natural units (tons pollution avoided).

Compression-ignition (diesel) engine

An engine operating on the diesel cycle, that is, compression pressures and temperatures to produce ignition instead of a spark-ignition (Otto cycle). Such engines use any of a wide variety of fuels that can auto-ignite under the diesel cycle, ranging from solid–liquid emulsions of coal to residual heavy fuel oil to marine distillate oil to natural gas with small liquid–fuel autoignition charge.

Global warming potential

A relative scale that compares the gas in question to that of the same mass of CO₂[62]. “GWP is the time-integrated radiative forcing due to a pulse emission of a given gas, over some given time period (or horizon) relative to a pulse emission of CO₂” [63]. This is calculated across various time scales, typically 20 years and 100 years, labeled GWP(20 year) and GWP(100year), respectively.

Specific fuel oil consumption

Consumption of fuel in mass per power unit in 1 h, typically measured in grams per kilowatt-hour. This term is general, that is, not fuel- specific, applicable to either distillate or residual petroleum products, and sometimes termed specific fuel combustion. The term applies to both marine distillate oil and marine heavy fuel oil.

Benefit–cost analysis

The net difference between the benefits derived from a potential strategy and the costs of pursuing such strategy. A net positive benefit–cost analysis (BCA) implies that the action may be worth pursuing, and a larger net BCA for action A may be preferred to a lower net BCA for action B. BCA is therefore a decision support tool for choosing strategies that achieve fundamental objectives.

Executive summary

Short-lived climate forcers as a component of climate management

	▪ Many species of particulate matter (PM) in a climate context are considered to be short-lived climate forcers (SLCFs) that contribute to regional climate impacts through direct energy absorption and through interactions with clouds, ice or snow.
	▪ Individually, their impact on climate forcing may be positive (warming) or negative (cooling). SLCFs are emitted from sources in PM ‘bundles’ that can determine the net effect on climate (warming or cooling). Black carbon (BC) is chief among SLCF pollutants, at least for diesel engines.
	▪ BC is a warming SLCF that may be controlled through existing PM reduction technologies, providing climate change mitigation. For shipping, the net global warming impact of particulate emissions needs to be considered to understand the control effectiveness of address SLCFs.
	▪ The Arctic is a potentially sensitive region for BC and other SLCF impacts. Shipping activity within and proximal to the Arctic suggest that regional control of SLCFs from ships may be especially important.

An SLCF analytic decision framework for climate management

	▪ We construct a cost–effectiveness decision framework within which to compare and rank various control technologies, because this is often used in management decision making to support the selection of alternatives that maximizes a given performance at least cost.
	▪ The cost–effectiveness of abatement measures is constructed in terms of activity-based energy, costs and emissions, specific to ship diesel engine combustion. Uncontrolled emissions from ships are compared with seven abatement technologies, and combination(s), considering various control effects on different SLCF particles.

Estimating SLCF emission control costs & cost–effectiveness

	▪ These comparisons are evaluated using a comprehensive uncertainty analysis, and then the technologies are ranked in terms of cost–effectiveness of abatement, with a focus on BC removal and total PM removal.
	▪ Combination technologies – even the dual combination of the least-cost technologies – perform robustly in terms of cost-effective abatement for PM.
	▪ We also consider impacts on CO2equivalents (CO2eq) emissions due to BC control strategies – accounting for not only the BC reductions themselves, but also for potential increases in CO2 emissions (primarily due to fuel penalties) and for effects on other SLCFs.
	▪ All technologies evaluated, except sea water scrubbers, produce positive benefits in terms of global warming potential over most of their distributions.
	▪ Sea water scrubbers do not provide robust reductions in global warming potential, mainly because they selectively control SLCF particles that contribute to cooling effects on regional climate scales.

SLCF management implications

	▪ The SLCF decision management framework provides a context for discussing emerging scientific assessments of impacts, and potential abatement strategies. This is demonstrated for international shipping with a focus on sensitive regions such as the Arctic.
	▪ Uncertainties are ranked in terms of importance, providing scientists, engineers and technology-policy analysts with guidance on the key inputs that could be better understood to inform policy and management decisions.
	▪ Combination technologies continue to outperform individual technologies, even in terms of CO2eq cost–effectiveness of SLCF abatement.
	▪ A technology-combination frontier can be defined according to the most cost-effective ways to meet emission control targets. This frontier can be represented in terms of direct emission control of BC or PM.
	▪ When the abatement technology frontier is represented in terms of net climate benefits (i.e., mapped according to the most cost-effective strategies for reducing the net CO2eq emissions of SLCFs), an optimal BC control range appears for international shipping at approximately 60% control. This minimizes both the cost of control to meet a target and the tradeoffs with concurrent reductions of ‘cooling’ SLCFs.

Current insights into arctic shipping control of SLCFs

	▪ SLCF reduction from ships can be achieved at approximately US$15–30 per mtCO2eq (20 year), under conditions where the vessel spends 25–100% of the time in a sensitive region.
	▪ The per-ship annual cost of control would be in the range of approximately US$25,000–35,000 with an upper range of approximately US$55,000–115,000 for the larger ship sizes.
	▪ The cost to achieve this 60% BC reduction target in the Arctic region may be approximately US$8–50 million per year, avoiding approximately US$9–70 million metric tons CO2eq per year.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.