Full text of DOI:10.1111/j.1559-3584.2002.tb00113.x

T E C H N I C A L PA P E R
Commercial Marine Emissions and
Life-Cycle Analysis of Retrofit Controls in a
Changing Science and Policy Environment
■ James J. Corbett, P.E. and Paul S. Fischbeck
ABSTRACT
This paper considers the technical feasibility of achieving NO reductions through engine controls on existing ocean-
x
going ships. International transport ships account for most of the NO emissions from shipping globally and regional-
x
ly (Corbett and Fischbeck 1997; Corbett et al. 1999). Recently, regulatory efforts to control NO from ship propul-
x
sion have been adopted at the international level (IMO 1998)) along with related efforts at national levels (EPA
1998a; Lemieszewski 1999; National Swedish Maritime Administration 1996; Swedish Port and Stevedores
Association 1999). The long working life of modern commercial ships, with world fleet averages greater than 20 years
(UNCTAD 1995), implies that new-engine
policies have limited value in meeting air-quali-
ty objectives in the near term. This has moti-
vated efforts to control shipboard emissions
with retrofit technologies. Nine technologies,
able include in-engine measures, mixtures of
water in air or fuel, and selective catalytic
reduction technology. Only selective catalytic
reduction would be considered a secondary
control method as defined above. Another
way to consider these technologies is by
whether control technologies require engine
modifications (in-engine controls) or whether
control strategies are implemented in the fuel
or air system (pre-engine technologies) or in
the exhaust system (post-engine technolo-
gies). These second definitions are used in the
discussion below.
demonstrated to control NO from existing
x
engines, are described in terms of their ability
to reduce emissions at the lowest life-cycle
costs. The costs for these technologies are with-
in the cost range of similar controls proposed
for new marine engines and within the cost
ranges for various land-based NO control
x
efforts, suggesting that these are not only tech-
nically feasible but also are economically rea-
sonable technologies for pollution control.
Nine specific NOx control-technologies are
considered, because of their application to
existing marine engines or similar systems.
In-engine NOx control strategies include: 1)
aftercooler upgrades, 2) engine derating, 3)
injection timing delays, 4) fuel system modifi-
cations to increase supply pressure, 5) fuel
injector upgrades, and 6) exhaust gas recircu-
lation. Pre- and post-engine technologies that
will be considered for NOx control include:
1) water added to the combustion air, 2)
water in fuel emulsions, and 3) selective cat-
alytic reduction.
Controlling NO in
x
Marine Diesel Engines
NOx emissions can be reduced through pri-
mary and/or secondary control mechanisms.
Primary methods affect the engine process
directly and can reduce emissions by 10 to 50
percent; secondary methods reduce emissions
without changing the engine performance
from its full optimized setting and typically
use equipment that is not integrally part of
the engine itself (MAN B&W 1996). The
types of methods that are commercially avail-
The literature discusses other engineering
control strategies, some of which are
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 9 3

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
proven to reduce NOx in new marine
This alternative is reported to result in an
increase in fuel usage of approximately 2%
(NAVSEA 1994). Moreover, over-optimized
“supercooling” might actually increase NOx
emissions, because ignition delay will be
larger at reduced temperatures. This could
result in a larger premixed fuel charge, with
corresponding increases in pressure and tem-
perature at the beginning of combustion
(Alexandersson et al. 1993).
engines or are being actively researched
(Alexandersson et al. 1993; DOE 1997;
EPA 1998b; NAVSEA 1992; NAVSEA 1994;
NIAG 1996; Venkatesh 1996; Wartsila NSD
1998). These include direct in-cylinder water
injection, in-cylinder ceramic coatings, non-
thermal plasma systems, variable geometry
turbocharging, etc. Since most of these tech-
nologies require substantial engine redesign
that makes them infeasible for retrofitting
on existing propulsion systems, they are not
considered as options for achieving signifi-
cant NOx reductions in the existing Fleet
within the next couple of decades.
Engine derating. NOx emissions from diesel
engines can sometimes be reduced by select-
ing a larger size engine (in terms of power
capacity) than needed and using lower-rated
fuel injectors (Sierra Research 1991). Space
and weight requirements for this technology
are negligible. This concept can be applied to
existing engines, with an estimated NOx
reduction between 5% and 23% (NAVSEA
1994). A nominal NOx reduction for this
evaluation is taken to be 14%, the midpoint
of this range. However, operating a marine
diesel engine at less than design-rated load
tends to increase the brake-specific fuel con-
sumption (bsfc); a nominal 4% increase in
fuel consumption is reported (NAVSEA 1994;
Sierra Research 1991). Smoke and particulate
matter (PM) also increase because of less effi-
cient operation (NAVSEA 1994).
In-Engine NOx Controls
Aftercooler upgrades. Nearly all medium-
speed and slow-speed marine engines have
turbochargers with aftercooling to optimize
fuel economy (Alexandersson et al. 1993).
An aftercooler is used between the tur-
bocharger and the engine to cool the charge
air (Genovesi and Browning 1998). This
cooling increases the intake air density and
lowers the charge-air temperature. Both of
these factors act to reduce the peak combus-
tion temperature and NOx emissions. NOx
reductions over naturally aspirated engines
have been achieved through better cooling
(Sierra Research 1991; Venkatesh 1996).
However, additional NOx reductions may be
achieved through additional upgrades to the
aftercooling system. Based on an average of
the range of estimates reported in the litera-
ture (NAVSEA 1992; NAVSEA 1994; Sierra
Research 1991; Venkatesh 1996), upgrading
the aftercooler systems on ship engines
might achieve a nominal NOx reduction of
10%, with a lower bound of 0% and an
upper bound of 22%. Space and weight
requirements for this technology are
assumed to include an additional cooler and
corresponding piping, with an estimated vol-
ume of five cubic meters and an estimated
weight of 4,500 kg. However, this may not
be applicable to those engines already
equipped with advanced cooling systems.
For these plants, the degree of optimization
is already high and improvements are not
expected to reduce current emission levels
by more than a few percent (Port of Los
Angeles et al. 1994).
Injection timing delays. Reducing the pres-
sure at auto-ignition by retarding the timing
of fuel injection will lower the peak flame
temperature and reduce NOx; however, it
also results in higher fuel consumption
(Hellmann 1997; MAN B&W 1996). This is
one of the simplest control strategies to
implement, particularly on marine propul-
sion engines with electronic controls that
allow the operator to “dial in” the injection
timing without engine shut down (Broman
1998). Space and weight requirements for
this technology are negligible. NOx reduc-
tions ranging between 10% to 30% are
reported, with an average reported reduc-
tion of 19% (Broman 1998; Hellmann
1997; MAN B&W 1996; NAVSEA 1994;
Sierra Research 1991). Fuel penalties are
estimated to result in about a 4% increase
(MAN B&W 1996; NAVSEA 1994), and
increases in PM, hydrocarbons, and carbon
monoxide have been reported (Heywood
1988; MAN B&W 1996).
9 4 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Fuel system modifications to increase supply
pressure. As discussed above, diesel engines
operate with excess air because the fuel and
air are not homogeneously mixed when
autoignition occurs. One way of obtaining a
better fuel-air mixture is to improve fuel
atomization. This can be done either by
upgrading the fuel injectors (discussed next)
or by increasing the injection pressure
(Alexandersson et al. 1993). For many large
marine engines, this means upgrading the
common rail fuel system to accommodate a
25% to 50% increase in maximum injection
pressures (Alexandersson et al. 1993).
Avoiding excessive stresses and deformations
on the fuel camshaft for existing engines
may define the upper limit for existing
engine retrofit and limit increases in fuel
injection pressure. Space and weight require-
ments for this technology are negligible,
although strengthening the camshaft may
modify the engine dimensions. NOx reduc-
tion ranges are similar to those from fuel
injector upgrades discussed next, achieving
reductions between 6% and 23% (MAN
B&W 1996; Port of Los Angeles et al. 1994)
with an average reported reduction of about
14% (slightly greater than for engine derat-
ing above). However, one report suggested
that NOx may increase under this alternative
(NAVSEA 1994), and smoke may also
increase (MAN B&W 1996). The average
fuel penalty for this alternative is approxi-
mately 2% (MAN B&W 1996).
cussed above (MAN B&W 1996). A nomi-
nal value is taken for this analysis to be
about 16% (slightly greater than fuel pres-
sure increases above).
Exhaust gas recirculation. Exhaust gas recir-
culation (EGR) entails recirculating some of
the exhaust gases into the cylinder with the
intake air. This increases the specific thermal
capacity and decreases the temperature in
the combustion space, and displaces oxygen
that might otherwise participate in the for-
mation of NOx (Alexandersson et al. 1993).
Space and weight requirements for this tech-
nology are negligible. While this technology
has been shown to reduce NOx by 19% to
50% without affecting fuel consumption,
engine manufactures do not recommend
EGR for engines that use residual fuels
(Alexandersson et al. 1993; MAN B&W
1996; Venkatesh 1996). This is because
recirculating the levels of particulate matter
and soot found in residual fuel exhaust
increases engine wear. Moreover, EGR fur-
ther increases emitted PM and soot by as
much as 50% (Alexandersson et al. 1993).
For this analysis, the nominal NOx reduc-
tion is 34% (the average of the reported
reductions).
Pre/Post-Engine NOx Controls
Water added to the combustion air.
Conceptually similar to increased aftercool-
ing discussed above, another way to cool the
charge air is to inject water into the intake
air. Water injection is a proven method to
reduce NOx in gas turbines (Sierra Research
1991; Urbach et al. 1997). Water injection
into the air system (also referred to as fumi-
gation) is easier to implement than water/fuel
emulsions discussed next, but can cause cor-
rosion of engine parts, and water quality is a
greater concern (Sierra Research 1991;
Venkatesh 1996). This technology is expect-
ed to require increased water distillation
capacity and storage to supply up to 4,000
gallons per day. Water quality requirements
are expected to require either a two-pass
reverse osmosis (RO) or a waste heat distilla-
tion system. While the RO system is larger
than the waste heat system, average values
for the space and weight requirements of 38
m3 and 19,000 kg, respectively, are used here
Fuel injector upgrades. Similar to increased
fuel pressure, upgrading the fuel injectors
has the objective of improving atomization.
The number and distribution of injector
holes is limited partly by the strength of the
injector tip and partly by manufacturing
limits on hole size; another limitation is the
ability for the spray pattern to avoid oxygen
deficient areas (Alexandersson et al. 1993).
Space and weight requirements for this tech-
nology are negligible. According to one
report, a NOx reduction of 20% is possible
at a fuel penalty of 3.5% (Alexandersson et
al. 1993). In addition, one report suggested
that smoke may increase (MAN B&W
1996). However, the range of data for differ-
ent types of nozzles suggests that NOx
reductions could be expected to be in the
same range as fuel pressure increases dis-
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 9 5

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
to get an order-of-magnitude requirement
(Choules 1999). This includes volume and
mass for a separate day-tank for distilled
water storage. A benefit of this technology is
that the percent of water injected into the
intake air can be varied to achieve various
levels of NOx reduction. However, smoke
and PM emissions may increase, and a nomi-
nal fuel penalty of 3% can be expected (Port
of Los Angeles et al. 1994). Reports indicate
that NOx reductions can range from 5% to
60% (Hellen 1998; Port of Los Angeles et al.
1994; Sierra Research 1991; Venkatesh
1996; Woodyard 1998); an average value of
28% is used here.
(Alexandersson et al. 1993; MAN B&W
1996; Port of Los Angeles et al. 1994;
Venkatesh 1996). An average value of 42%
NOx reduction was chosen for this assess-
ment. In addition to these reductions, this
technology is reported to reduce PM and
carbon monoxide through better fuel atom-
ization as the water vaporizes explosively
creating secondary atomization
(Alexandersson et al. 1993).
Selective catalytic reduction (SCR). SCR pro-
vides the greatest reductions in NOx emis-
sions of any of the technologies discussed
above, and marine application of this tech-
nology has been the focus of considerable
interest (Alexandersson et al. 1993; Cooper
and Peterson 1995; Gibson and Groene
1991; MAN and B&W 1997; MAN B&W
1996; Venkatesh 1996; Wartsila NSD 1994;
Woodyard 1998). Catalytic reactions with
ammonia or urea reduce the oxidized nitro-
gen to nitrogen gas according to the follow-
ing reactions (Alexandersson et al. 1993):
Water in fuel emulsions. Water emulsifica-
tion in the fuel is a proven technique to
reduce NOx, and it lends itself to application
with residual fuels that often already contain
some emulsified water and other blended
fuels (MAN B&W 1996). A standard engine
design permits the addition of about 20%
water (as a percent of fuel mass) at full load,
although more that 50% water has been
tested (MAN B&W 1996). Additional distil-
lation capacity may not be needed at 20%
or less water injection. However, if it is
needed, the quality of water is not as impor-
tant to this technology as it is for water
added to the intake air, discussed above. If
additional distillation is required, average
values for the space and weight require-
ments of 25 m3 and 13,000 kg, respectively,
are used here to get an order-of-magnitude
requirement (Choules 1999). These values
are somewhat smaller than above, because
emulsification is not expected to require a
separate tank for distilled water storage.
Emulsification may require some special
equipment and emulsifying agents to ensure
fuel stability and reliable engine starts. Fuel
consumption using this technology has been
shown to vary from a fuel savings of 6% to
fuel penalties of similar size (Alexandersson
et al. 1993), although most reports show an
increase in fuel consumption (MAN B&W
1996; NAVSEA 1994; Venkatesh 1996).
This analysis used 2%, an average of report-
ed fuel penalties. NOx reductions ranged
widely for this technology (from 20% to
73%), mostly as a result of the ability to
select the water content in the emulsification
and therefore the NOx control
For ammonia catalyst:
4 NO + 4 NH3 + O2
4 N + 6 H O
(1)
(2)
→
2
2
6 NO2 + 8 NH3
7 N + 12 H O
2 2
→
And for urea catalyst:
6NO + 2(NH ) CO
5N + 4H O + 2CO (3)
2 2 2
→
2
2
6NO + 4(NH ) CO
7N + 8H O + 4CO (4)
2 2 2
→
2
2
2
Although both types of catalyst systems have
been installed on marine vessels, urea may be
favored because it is non-toxic and biologi-
cally harmless and can be transported with-
out problems (Woodyard 1998). Installation
of SCR systems imposes additional space and
weight requirements, with an average of 33
m3 and 3,100 kg reported taken from various
reports in the literature cited above.
However, Wartsila NSD has been develop-
ing a “compact SCR” system that equals
the size of and replaces current silencer sys-
tems installed aboard ships (Mullen 1998).
Reported NOx reductions can be as high as
98%, although the reduction at lower
engine loads can be as low as 57% (MAN
B&W 1996). An average of 81% from
9 6 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Table 1
Summary of NOx Control Technology Performance Attributes
CONTROL TECHNOLOGY
NOMINAL NO
x
NOMINAL REDUCTION IN PM
NOMINAL INCREASED
REDUCTION (%)
AND OTHER POLLUTANTS (%)*
FUEL USE (%)
Aftercooler upgrade
Engine derating
Fuel pressure increase
Injector upgrade
10
14
14
16
19
28
34
42
81
-1
-10
-21
-21
-11
1
-51
15
0
2
4
2
2
4
3
0
2
0
Injection Timing Retard
Water in combustion air
Exhaust gas recirculation
Water/fuel emulsion
Selective catalytic reduction
* Negative values in this column represent an increase in emissions of other pollutants.
reported values is used here as the nominal
NOx reduction for SCR. While SCR does
not increase fuel consumption and can be
installed on engine systems using high-sul-
fur residual fuel (MAN B&W 1996), the
technology involves the consumption of
ammonia or urea. Catalyst consumption
rates equal about 2% of the fuel consump-
tion (about 16 kg/h of operation)
but less than linear as reported by another
(MAN B&W 1996). Therefore, the intent
here is that data included in this study,
while quantitative in form, serve to inform
the reader qualitatively that a given NOx
control technology may have greater
trade-offs compared to other control tech-
nologies, in that they increase either other
pollutants or fuel consumption — or both.
(Alexandersson et al. 1993).
Cost and Pollution
Reduction Assessment
Summary of NOx
Control Technologies
This section describes the approach to esti-
mating life-cycle costs per ship of including
existing engines in NOx control policies.
Because these technologies are currently avail-
able for marine application, costs were direct-
ly taken or derived from publicly available
sources. The results, presented as net present
values, define a set of preferred technologies
according to their NOx reduction potential.
Table 1 presents a summary of the technol-
ogy discussion above, focused on the attrib-
utes most important in considering the NOx
-control potential. The determination of
nominal values for NOx reduction was
described above. Nominal values quantify-
ing the effects of each technology on other
pollutants were developed similarly from
the literature, but with much less informa-
tion available. These quantitative values
are taken from the available literature;
however, it should be acknowledged that
the emissions trade-off between NOx
reduction and increased emissions for
other pollutants, especially PM, is not
always reported.
Cost Estimating Methodology
Two basic types of cost data are considered:
fixed and annual. Fixed costs of NOx con-
trol include retrofit design, hardware and
equipment, installation. Annual costs result
from fuel penalties associated with retrofit
NOx controls, other materials such as
ammonia or urea in SCR systems, and other
operating expenses (mostly labor) resulting
directly from the retrofit technology.
For example, increases in PM and smoke
are consistently reported for injection tim-
ing retard; these increases are reported to
be nearly linear with reductions in NOx
according to one source (Heywood 1988),
Data for these cost estimates is taken as
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 9 7

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Table 2
Retrofit Design Costs for Existing Engine NOx Control Technologies
CONTROL TECHNOLOGY
RETROFIT DESIGN (R&D-LIKE)
RETROFIT DESIGN (R&D-LIKE)
COSTS ($) PER ENGINE MANUFACTURER
COSTS ($) PER SHIP
$700
Aftercooler upgrade*
$336,000
$0
Engine derating**
$0
Fuel pressure increase*
Injector upgrade*
$1,000,000
$1,500,000
$0
$2,100
$3,300
$0
Injection Timing Retard**
Water in combustion air†
Exhaust gas recirculation*
Water/fuel emulsion†
$834,000
$500,000
$834,000
$834,000
$5,838,000
$1,800
$1,100
$1,800
$1,800
$12,700
Selective catalytic reduction†
Total Retrofit Design Costs
* Used U.S. EPA R&D costs for these technologies.
** Engine derating and injection timing retard are not considered to incur retrofit design costs.
† Used average of U.S. EPA R&D costs for all technologies they reported.
reported where possible and derived below
otherwise. This analysis uses new engine
technology costs developed by U.S. EPA for
the largest marine engines addressed by the
proposed rule (EPA 1998a), except where
other technology cost data was considered
to be more detailed. U.S. EPA costs were
used for the following technologies: 1) after-
cooler upgrades, 2) fuel system modifica-
tions to increase supply pressure, 3) fuel
injector upgrades, and 4) exhaust gas recir-
culation. The primary exception to this is
that where U.S. EPA calculated only the
incremental costs on top of a new engine
without NOx controls, this analysis used the
full price of equipment. Another modifica-
tion is that labor rates are assumed to be
$500 per person-day, typical of labor rates
(plus benefits) in public shipyards
(Rivenbark 1996). Costs developed here are
idealized to some degree, as they apply to
the “typical” engine propelling large ships
operating internationally. However, these
costs were discussed and were generally vali-
dated through discussions with representa-
tives of Wartsila NSD Corporation (Broman
and Koivisto 1999).
including both manufacturer costs to design
a general solution to the problem, and spe-
cific implementation costs for various types
of ships in the fleet. This analysis estimated
the retrofit costs on a per ship basis for each
technology, in order to determine how
design costs would be distributed across the
fleet. For this calculation it was assumed
that all transport ships in the existing world
fleet are engaged in international maritime
transportation (LMIS 1996). The number of
ships to be retrofit was distributed across
the primary engine manufactures for large
marine propulsion engines. According to
Lloyd’s Registry, the top nine engine manu-
facturers provide the main engines to more
than 66% of the world’s ships (LMIS 1996).
(The next two manufacturers, Deutz and
S.K.L., account for about 4% each; the last
two manufacturers, Akaska and Niigata,
account for about 3% each.) The top five
manufactures, MAN-B&W, Wartsila-NSD,
Hanshin, Mitsubishi, and MaK, account for
more than 50% of the main propulsion
engine systems on transport ships. These
manufactures account for an even larger per-
cent of recently built ship engines—engines
with significant remaining working life and
therefore most likely to be retrofit with NOx
controls. A ten-year retrofit implementation
schedule was used to estimate the number of
retrofits that would occur in the first year.
Similar to the U.S. EPA method, this
A retrofit-design cost-estimating method
that is similar to the approach used by the
U.S. EPA to estimate new engine R&D costs
(EPA 1998b). was applied. Retrofit design
costs were considered to be similar to new
engine research and development costs,
9 8 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
approach places retrofit design costs up
front and makes the estimate larger (more
conservative). These costs are shown in
Table 2.
annual costs were taken from the various
sources that provided technology costs.
Fixed, Annual and NPV Costs
Hardware costs for engine derating assumed
that new injectors were needed that would
limit the maximum engine power to the
lower rating. Injection timing retard has no
hardware costs, only labor costs to retime
the engine. Additional water distillation
capacity was assumed in the costs for
humidification of combustion air, with dis-
tillation costs provided through an industry
quote (Choules 1999) and other costs taken
from a report for U.S. EPA (Venkatesh
1996). The same increased distillation
capacity was assumed for water/fuel emulsi-
fication, with equipment costs for emulsifi-
cation provided by a Swedish Transport
Research Board report (Alexandersson et al.
1993). Costs for SCR technology onboard
ship were also taken from the Swedish
Transportation Research Board report
(Alexandersson et al. 1993).
Table 3 presents the fixed and annual costs
estimated for each NOx reduction technolo-
gy considered. Several observations can be
made at this point. First, hardware and
installation costs vary considerably, but gen-
erally increase with increased NOx reduction
(compare Table 1 and Table 2). The excep-
tions are the costs estimated for injection
timing retard and exhaust gas recirculation.
Second, retrofit design costs are generally
similar across technologies. Third, annual
maintenance and operating costs appear
generally small compared to increased annu-
al fuel costs, with the exception of SCR.
This is expected since SCR does not increase
fuel consumption but does involve signifi-
cant consumption of either urea or ammo-
nia, as discussed above.
The net present value (NPV) of these esti-
mates was calculated using a 15% interest
rate. Adopting a very conservative approach,
we assume that all existing ships, when
retrofitted, would have a remaining life span
of 23 years, equal to the average life of a
new ship. This is considered to be conserva-
tive since a longer period increases the NPV
and the current average age of the oceango-
ing fleet is greater than 18 years (LMIS
1996). (Ages reported from 1996 registry
data. The average age of container ships in
1996 was about 11 years (LMIS 1996).) In
other words, this would represent the NPV
To estimate annual fuel costs, estimated
average daily fuel usage was converted to an
annual basis and multiplied by the percent
increase in fuel usage presented above for
each technology. This annual increase in fuel
usage was then multiplied by a representa-
tive price for fuel based on a review of
prices published during the last several years
in a weekly industry publication (MGN
1999). A price of $66 per ton fuel was used
for residual fuels, and a price of $130 per
ton was used for marine distillate oil. Other
Table 3
Summary of Fixed and Annual Costs for NOx Control Technologies
FIXED COSTS
ANNUAL COSTS
CONTROL TECHNOLOGY
Retrofit Design
(R&D-like) Costs
Hardware and
Increased Maintenance
Increased Fuel Costs
Installation Costs
and Operating Costs
Aftercooler upgrade
Engine derating
$700
$0
$11,000
$34,000
$34,000
$38,000
$250
$0
$1,500
$0
$27,000
$54,000
$29,000
$24,000
$54,000
$34,000
$2,600,000
$31,000
$0
Fuel pressure increase
Injector upgrade
$2,100
$3,300
$0
$0
Injection Timing Retard
Water in combustion air
Exhaust gas recirculation
Water/fuel emulsion
Selective catalytic reduction
$3,000
$2,600
$0
$1,800
$1,100
$1,800
$1,800
$130,000
$2,500
$117,000
$283,000
$1,000
$30,000
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 9 9

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
FIGURE 1:
costs of a ship that was one year old when
retrofit. With a 15% discount rate this
assumption has less effect on the NPV costs
than a lower discount rate would. Table 4
presents the NPV costs for each technology.
the cost for that equipment when provided
in a new engine package (EPA 1998b).
NPV (i=15%) of Control
Costs (Fixed + Annual)
of NOX Control for
Existing Engines
Figure 1 graphically presents the NOx con-
trol technologies according to their relative
costs and pollution control and illustrates
the sensitivity of these concerns on the NPV
costs. Horizontal bars describe the range of
reported NOx reductions and vertical bars
describe the range of NPV costs if retrofit
design costs were ignored and if fixed costs
were three times greater (consistent with
after-market cost factors reported by U.S.
EPA). In addition, the lower cost bound for
water in combustion air and water/fuel
emulsions has removed costs associated with
adding distillation capacity, assuming that
existing capacity may be sufficient.
(emission control
assumed during 100%
of operations)
Discussion and Summary
The results in Table 4 include the nominal
NOx reductions drawn from the literature.
However, these NOx reduction values may
vary across the ranges discussed above,
either as a result of uncertainty in technolo-
gy effectiveness or as a result of intentional
variations in engineering design. For exam-
ple, the percent of water in water/fuel emul-
sions may be adjusted to achieve different
levels of NOx control. Additionally, while
annual costs should be relatively robust,
since they depend on the increased annual
fuel usage more than any other factor, the
NPV cost estimates in Table 4 may be sensi-
tive to errors in estimated fixed costs.
It should be emphasized that all costs shown
in Figure 1 assume that control technologies
are operated at all times. This study does
not explore the potential impact of scenarios
where operation of NOx control equipment
is more limited.
Specifically, U.S. EPA reports that after-mar-
ket engine equipment may cost three times
1 0 0 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Table 4
Summary of Expected NOx Reductions and Fixed, Annual and NPV Costs
for NOx Control Technologies Applicable to Existing Engines*
FIXED COSTS
ANNUAL COSTS
CONTROL TECHNOLOGY
NOMINAL NOx
REDUCTION (%)
HARDWARE, INSTALLATION,
MAINTENANCE, OPERATING,
NPV COSTS
(15% interest annually over 23 year
AND RETROFIT DESIGN
AND FUEL
s)
Aftercooler upgrade
Engine derating
10
14
14
16
19
28
34
42
81
$12,000
$34,000
$36,000
$41,000
$250
$27,000
$55,000
$29,000
$24,000
$57,000
$184,000
$386,000
$220,000
$192,000
$363,000
$365,000
$16,900,000
$325,000
$475,000
Fuel pressure increase
Injector upgrade
Injection Timing Retard
Water in combustion air
Exhaust gas recirculation
Water/fuel emulsion
Selective catalytic reduction
$134,000
$3,500
$36,000
$2,640,000
$32,000
$30,000
$119,000
$285,000
• Some differences from previous numbers may appear due to rounding of calculations.
engine be retrofit individually. Depending on
the method, this could also apply to the
water in combustion air technique.
Relative Feasibility of
NOx Control Technologies
As seen in Figure 1, the point estimates
developed in this paper suggest a clear pro-
gression of preferred technologies that pro-
vide increasing NOx reductions at the least
cost. The least-cost control curve would
include 1) aftercooler upgrade, 2) fuel injec-
tor upgrade, 3) water/fuel emulsion, and 4)
selective catalytic reduction. The technolo-
gies that simply appear too expensive
include 1) engine derating, and 2) exhaust
gas recirculation. EGR appears to be beyond
serious consideration for existing engines
unless it can become feasible during engine
operation with residual fuel.
However, water/fuel emulsions and selective
catalytic reduction can be retrofit in the
common fuel system before both engines or
in a common exhaust system, respectively.
This implies that costs for higher NOx con-
trol using pre- or post-engine technologies
could be more economical for multiple-
engine configurations.
Perhaps more important are the horizontal
ranges for each technology. Particularly for
the technologies with lower nominal NOx
reductions, the overlap is significant. This
may imply that where costs are similar, an
operator may select the technology with
greater nominal NOx control, even if it
exceeds the minimum required control. This
could ensure greater confidence in the
amount of NOx reduction and some insur-
ance against future regulatory changes. The
most interesting technologies in this regard
may be humidification of intake air or
water/fuel emulsions. With such a wide
range of NOx reduction, the ability to con-
trol the technology performance during
operation may be desirable for an operator.
The order of these results does not appear to
change if the fixed costs shift together, either
higher or lower, within the ranges shown.
Even if fixed costs increase by a factor of
three, representing typical differences
between OEM price and retail mark-up for
parts (EPA 1998b), the relative cost ranking
of these technologies does not substantially
change. An anomaly of the method present-
ed here is that these per-ship costs assume
one main engine per ship. For ships with
multiple main engines, the relative cost dif-
ferences could change—but not for all tech-
nologies. For example, the five technologies
with lower NOx reductions (aftercooler
upgrade, derating, fuel system upgrade,
injector upgrade, and injection timing
Technology Cost Impact
on the Ship Operator
These technology costs of control will not be
negligible, especially to the ship operator.
Table 5 identifies the general operating costs
retard) would typically require that each
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 1 0 1

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
FIGURE 2:
for an international bulk carrier, noting that
fuel is the largest factor affecting total operat-
ing cost (Wartsila NSD, 1997). As a first-
order analysis of the annual cost impact of
NOx control technologies on vessel opera-
tions, the annualized cost for SCR operated
during 100% of main engine operations (the
technology with the maximum annual cost of
all preferred technologies) was used to get an
potential upper bound estimate. Using an
annualized cost of about $86,000 and an esti-
mated annual fuel cost of $1.34 million, NOx
controls equal approximately 6% of annual
fuel costs. If fuel costs equal nearly 60% of
the operating cost of a ship, then NOx control
costs equal less than 4% of total operating
costs. This could be more or less significant to
a shipping company depending upon the rate
of policy implementation requiring NOx tech-
nology retrofit, because the ship operator will
need to factor this cost into shipping rates.
Operators at the margin could be most affect-
ed by aggressive policies in this regard.
The impact on the price of goods is expect-
ed to be lower. Freight rates equal between
4% and 13% of the import price of low-
value bulk commodities, with a worldwide
average of about 9% of import value
(UNCTAD 1995). The commodity prices
of higher-value goods such as manufac-
tured items or containerized cargoes are
less affected by transaction costs such as
freight rates. Although freight rates include
other costs (such as bunker surcharges and
currency adjustment factors, port delay
and additional port surcharges) that are
not explicitly defined in Table 5, it is
assumed here that these are contained in
the other operational costs (UNCTAD
1995). With shipping contributing about
9% to the cost of goods and the assump-
tion that in the long run 100% of operator
costs will be passed on to the consumer,
the 4% change in operating costs could
eventually increase by about 0.3% the
costs of goods.
Cost-effectiveness of
NO Control
x
Technologies ($ per
ton NO )
x
1 0 2 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Moreover, if fuel prices were to increase so
would the annual costs associated with fuel
penalties for NOx control technology.
However, the direct effect of increasing fuel
Table 5
Operational Costs for a Bulk Carrier
[Wartsila NSD, 1997]
prices on total operating costs would
Total Operational Costs
Fuel Costs
100%
59%
24%
7%
increase both the baseline fuel cost and the
annual technology cost by the same percent-
age (or less for capital intensive technologies
like SCR). However, while the fuel penalties
of most of the NOx control technologies
considered in this work could be motivation
enough, any significant increase in fuel
Crew Expenses
Repair and Maintenance Costs
Other Operational Costs
9%
period does not change significantly any of
the insights resulting from the longer annu-
alization period.
prices would promote research into minimiz-
ing these fuel-related costs of NOx control.
The second number that is needed is the
estimated annual NOx emissions reduced
from the inventory through control. This
can be estimated on a per-ship basis, or by
using the estimated NOx reduced in the
actual region of control. A comparison of
the two methods using our previous invento-
ry work (Corbett and Fischbeck 1997;
Corbett et al. 1999) showed that the differ-
ences between these approaches are relative-
ly small, although cost-effectiveness on a
per-ship basis provides the highest costs. We
present the per-ship cost-effectiveness results
in Figure 2. It is worth noting that the most
cost-effective technology is SCR. Although it
is the most costly retrofit technology consid-
ered here, the significant NOx reduction
potential (denominator) offsets its higher
technology cost (numerator). The minimum,
median, and maximum cost-effectiveness
values accepted by the California Air
Cost-effectiveness Comparison
With Other NOx Control Efforts
A cost-effectiveness comparison of retrofit
NOx controls can be made with NOx control
strategies for new ships and with NOx con-
trol strategies for other combustion sources
(both mobile and stationary). Cost-effective-
ness is defined by regulators as the annual-
ized cost of control divided by the annual
mass of emissions reduced (Koeberlein et al.
1997). Policymakers use cost-effectiveness
primarily to compare control technologies
and regulations in terms of relative costs to
benefits. For purposes of this analysis, cost-
effectiveness represents the annual technolo-
gy cost divided by the annual amount of
NOx emissions avoided.
Two numbers are needed to estimate cost-
effectiveness. First, the annual cost of each
technology is computed by annualizing its
net present value. This analysis uses NPV
costs that were annualized over 23 years
(the same period used to estimate the NPV
costs in Figure 1). A sensitivity analysis
annualizing the 23-year NPV over a ten-year
period increased the annualized costs for
capital intensive technologies (e.g., 10-year
annualized costs increase by 26% for SCR,
by 25% for water/fuel emulsion and water
in combustion air, and by 14% for after-
cooler upgrade). (Absolute annual cost dif-
ferences were less than $700 in all cases. For
simplicity, this analysis ignores the transient
costs that may occur during implementation
of NOx control and the effects of fleet
Resources Board for stationary engines
(Koeberlein et al. 1997) are presented in
Figure 2 as well, and discussed below.
Cost-effectiveness estimates for existing
marine diesel engines compare well with
other NOx control efforts for both mobile
and stationary sources. As shown in Table 6,
NOx regulations for stationary internal
combustion engines (which are most similar
to marine diesel engines) have ranged in
cost-effectiveness between $140 per ton
NOx removed and $25,000 per ton NOx
removed (Koeberlein et al. 1997). CARB
has also reported cost-effectiveness for
mobile sources, including those with diesel
engines (Koeberlein et al. 1997). The U.S.
EPA estimated the cost-effectiveness of Tier
growth or emissions increase over time.)
However, this difference using a 10-year
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 1 0 3

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Table 6
Summary of Cost-Effectiveness for NOx Control from Existing Ships (This Work) and Other Recent NOx Control Programs
NOx RULE
POLLUTANTS CONSIDERED
COST-EFFECTIVENESS ($/ton NOx removed)
This study, all nine control technologies
(low = SCR, high = EGR)
NO
x
$200 - $16,000
U.S. EPA proposed Tier 2 for new
marine engines (including IMO)
U.S. EPA proposed Tier 2 and Tier 3
for new marine engines (excluding IMO)
Clean fuel fleet program (heavy-duty)
2.5 g/hp-hr NMHC* + NOx standards
for highway heavy-duty engines
Locomotive engine standards
Non-road Tier 2 standards
HC + NO
x
x
$76 - $738
$30 - $280
HC + NO
NO
x
$1,300 - $1,500
$100 - $600
NMHC* + NO
x
x
NO
NMHC* + NO
Both THC + NO
x
$160 - $250
$480 - $540
Various CARB mobile source
emissions control measures
Various CARB stationary IC
x
and NO
x
only
$180 - $38,000
NO
x
$140 - $25,000
engine measures
2 and Tier 3 standards for new marine
diesel engines, and provided four other
recent U.S. EPA mobile source regulations
that required NOx reductions (EPA 1998b).
These cost-effectiveness estimates are sum-
marized in Table 6.
controls proposed for new marine engines
and controls for other mobile or stationary
sources on land. These results, based on
technologies currently installed in shipboard
applications, suggest that retrofit NOx con-
trol technologies can be feasible, both techni-
cally and economically.
Conclusion
Acknowledgements
This paper discussed the mechanisms of NOx
formation and presented nine NOx control
technologies that are applicable to existing
marine engines. These technologies can be
ranked in terms of their emission-control
potential and their cost to define a technolo-
gy “feasibility frontier” that includes four
control technologies (in order of increasing
control and cost): aftercooler upgrade, injec-
tor upgrade, water/fuel emulsion, and selec-
tive catalytic reduction. Estimated annual
costs of these technologies are shown to be
less than 4% of annual operation costs.
Under appropriate policy implementation,
freight rates could include these NOx control
costs with only a marginal impact on bulk
commodity prices and a negligible effect on
the prices of higher-value commodities.
Moreover, the NOx control costs for pre-
ferred technologies according to this study
are clearly within cost-effectiveness ranges
for other NOx control strategies, such as
This work was supported by the Department
of Engineering and Public Policy and the
Center for Integrated Study of the Human
Dimensions of Global Change at Carnegie
Mellon University, by the National Science
Foundation (SBR-9521914), and the
University of Delaware Graduate College of
Marine Studies. ■
REFERENCES
Alexandersson, A., E. Flodstrom, R. Oberg, and P.
Stalberg, “Exhaust Gas Emissions from Sea
Transportation,” Mariterm AB, Swedish
Transportation Research Board, Gothenburg,
Sweden, 1993.
Broman, C., “The Diesel Engine and the
Environment,” in 61st Annual International Joint
Conference of the Canadian Shipowners
1 0 4 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Association and Lake Carriers Association,
Wartsila NSD Corporation, Motebello, Quebec,
Canada, 1998.
Low-Speed Diesel Engines,” in General Meeting of
the Institution of Diesel and Gas Turbine
Engineers, pp. 18, The Institution of Diesel and
Gas Turbine Engineers, London, UK, 1997.
Broman, C., and O. Koivisto, Meeting to Discuss
NOx Technologies and Costs for Existing Engines,
Wartsila NSD Corporation, 1999.
Heywood, J.B., Internal Combustion Engine
Fundamentals, 930 pp., McGraw-Hill, Inc., New
York, NY, 1988.
Choules, P., Information about shipboard desalin-
ization units, MECO, 1999.
IMO, Annex VI of MARPOL 73/78 and NOx
Technical Code, 150 pp., International Maritime
Organization, London, UK, 1998.
Cooper, D., and K. Peterson, “Emission
Measurements From a Urea-Based SCR/OXI
Catalytic NOx/HC Exhaust Gas Treatment System
on Board a Diesel Powered Passenger Ferry —
Operation After 12,000 Hours Service,” Swedish
Environmental Research Institute (IVL),
Gothenburg, Sweden, 1995.
Koeberlein, D., B. Riddell, E. White, J. De Vita, M.
Bekken, J. Guthrie, K. Nguyen, and T. Ferreira,
“Sources and Control of Oxides of Nitrogen
Emissions,” California Environmental Protection
Agency Air Resources Board (CARB), Sacramento,
CA, 1997.
Corbett, J.J., and P.S. Fischbeck, “Emissions From
Ships,” Science, 278 (5339), 823-824, 1997.
Lemieszewski, S., “Swedish Incentive Scheme of
Differentiated Fairway and Harbour Dues for
Ships Related to their Environmental
Performance Profile: Experiences from the First
Year of Operation,” in The Motorship 21st Annual
Marine Propulsion Conference, Swedish Maritime
Administration, Athens, Greece, 1999.
Corbett, J.J., P.S. Fischbeck, and S.N. Pandis,
“Global Nitrogen and Sulfur Emissions
Inventories for Oceangoing Ships,” Journal of
Geophysical Research, 104 (D3), 3457-3470, 1999.
DOE, Abstracts of Papers Presented at the 1997
Diesel Engine Emissions Reduction Workshop, in
1997 Diesel Engine Emissions Reduction
Workshop, edited by J. Fairbanks, DOE, San
Diego, 1997.
LMIS, Dataset of Ships 100 GRT or Greater, Lloyd’s
Maritime Information Services, Stamford, CT,
1996.
MAN, and B&W, “Emission Control: Two-Stroke
Low-Speed Diesel Engines,” MAN B&W, 1997.
EPA, 40 CFR Part 94: “Control of Emissions of Air
Pollution From New CI Marine Engines at or
Above 37 kw;” Proposed Rule, Environmental
Protection Agency, Ann Arbor, MI, 1998a.
MAN B&W, “Emission Control: Two Stroke Low-
Speed Diesel Engines,” MAN B&W Diesel A/S,
Copenhagen, Denmark, 1996.
EPA, “Draft Regulatory Impact Analysis: Control of
Emissions From Compression-Ignition Marine
Engines,” U.S. EPA Office of Air and Radiation;
Office of Mobile Sources; Engine Programs and
Compliance Division, Ann Arbor, 1998b.
MGN, M.G.N., MGN Newsletter and HK Shipping
News International, Maritime Information
Systems, Inc.(MIS), 1999.
Mullen, E.R., “Modern Trends in Medium Speed
Diesels,” in Trans Marine Propulsion Systems, Inc.,
3rd Annual Diesel Seminar, Wartsila NSD
Corporation, 1998.
Genovesi, K., and L. Browning, “Incremental Cost
Estimates for Marine Diesel Technology
Improvements,” EPA OMS, Ann Arbor, MI, 1998.
Gibson, J., and O. Groene, “Selective Catalytic
Reduction on Marine Diesel Engines,” Automotive
Engineering, 99 (10), 18-22, 1991.
National Swedish Maritime Administration,
Summary of a Proposal to the Swedish
Government Concerning Environmentally
Differentiated Shipping Dues, Sjöfartsverket,
National Swedish Maritime Administration,
Norrkoping, Sweden, 1996.
Hellen, G., “Technologies for Diesel Exhaust
Emission Reduction,” Wartsila NSD Corporation,
Crystal City, VA, 1998.
NAVSEA, “Internal Combustion (Gas Turbine and
Diesel) Engine Exhaust Emission Study,” Naval Sea
Hellmann, J.O., “Emission Control: Two Stroke
NAVAL ENGINEERS JOURNAL
WINTER 2002 ■ 1 0 5

Commercial Marine Emissions and Life-Cycle Analysis of Retrofit Controls in a Changing Science and Policy Environment
Systems Command, and Westinghouse Machinery
Technology Division, Pittsburgh, PA, 1992.
Purposes,” Swedish Port and Stevedores
Association, 1999.
NAVSEA, “U.S. Navy Marine Diesel Engine and
Gas Turbine Exhaust Emissions,” Naval Sea
Systems Command 03X31, Washington, DC, 1994.
UNCTAD, Review of Maritime Transport 1994,
United Nations, New York and Geneva, 1995.
Urbach, H.B., D.T. Knauss, J. Emory, B.L. Wallace,
J.H. Wasser, M.R. Sexton, and J. Frese, “The
Reduction of NOx Emissions From Marine Power
Plants,” in Air and Waste Management
Association’s 90th Annual Meeting and Exhibition,
Air and Waste Management Association, Toronto,
Ontario, Canada, 1997.
NIAG, “Prefeasibility Study on a NATO
Environmentally Sound Ship of the 21st Century,”
NATO Industrial Advisory Group, Brussels,
Belgium, 1996.
Port of Los Angeles, Port of Long Beach, Dames &
Moore, and Morrison and Foerster, “Control of
Ship Emission in the South Coast Air Basin:
Assessment of the Proposed Federal
Implementation Plan Ship Fee Emission Fee
Program,” Port of Los Angeles, Los Angeles, CA,
1994.
Venkatesh, S., “Reduction of NOx and PM from
Navy Diesel Engines: A Feasibility Analysis,”
Acurex Environmental Corp., Mountain View, CA,
1996.
Wartsila NSD, Wartsila Diesel Group Customer
Journal: Marine News, Wartsila NSD Corporation,
Annapolis, MD, 1994.
Rivenbark, T., Financial and Operating Statements
for Norfolk Naval Shipyard: Facts Derived from
Year Ending {FY 1996}, Norfolk Naval Shipyard,
1996.
Wartsila NSD, Wartsila 32: The Performer,
Wartsila NSD Corporation, Annapolis, MD, 1997.
Sierra Research, “Regulatory Strategies for
Reducing Emissions from Marine Vessels in
California Waters,” California Air Resources
Board, Sacramento, CA, 1991.
Wartsila NSD, Wartsila 32: Technology Review,
Wartsila NSD Corporation, Annapolis, MD, 1998.
Woodyard, D., “Designers Clean Up Diesel
Exhausts,” in Jane’s Speed At Sea, pp. 46-48, 1998.
Swedish Port and Stevedores Association,
“Differential Charging for Environmental
DR. JAMES J. CORBETT, P.E. is an assistant professor of Marine Policy in the Graduate College of
Marine Studies at the University of Delaware, in Newark, Delaware. His research evaluates air emissions
from international and domestic shipping and develops feasible technology-policy measures to reduce
their impacts on climate change and air quality. He has authored more than 10 peer-reviewed articles
related to the environmental impacts and technology-policy mitigation alternatives of air pollution from
ships. Dr. Corbett is a graduate of the California Maritime Academy, with a masters degree in mechanical
engineering and masters and doctoral degrees in engineering and public policy from Carnegie Mellon
University. He formerly held a commission in the Naval Reserve as an Engineering Duty officer and is a
member of ASNE, SNAME, ASME, among other professional organizations.
DR. PAUL FISCHBECK is an Associate Professor in the Departments of Engineering and Public Policy
and Social and Decision Sciences at Carnegie Mellon University, in Pittsburgh, Pennsylvania. He holds a
Ph.D. from Stanford University. His research focuses on the quantification and communication of uncer-
tainty, including both theoretical improvements to decision analysis and numerous applied real world
problems. Using a systems approach, he and Corbett have quantified the pollution caused by interna-
tional shipping and proposed several engineering and economic solutions. He is currently studying per-
formance-based regulations and how they can be used to improve fire-safety systems and oil-tanker
design. He is a retired Navy captain.
1 0 6 ■ WINTER 2002
NAVAL ENGINEERS JOURNAL