United States Air and Radiation EPA420-P-02-007
Environmental Protection December 2002
Agency
vxEPA Impacts of Lubrizol's
PuriNOx Water/Diesel
Emulsion on Exhaust
Emissions from
Heavy-Duty Engines
Draft Technical Report
> Printed on Recycled Paper
-------
EPA420-P-02-007
December 2002
of
on
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
This technical report does not necessarily represent final EPA. decisions or positions,
It is intended to present technical analysis of issues using data that are currently available.
The purpose in the release of such reports is to facilitate the exchange of
technical, information and to inform the public of technical developments which
may form the basis for a final EPA decision, position, or regulatory action.
-------
Table of Contents
I. Nature and Purpose of this Technical Report 1
II. Draft Review Process 2
IE. Background 3
IV. Historical research on emulsions 5
A. Studies of water impacts on combustion 5
1. Early efforts to reduce combustion temperatures 5
2. Micro-explosion phenomena 6
3. Determination of amount of water to be used 6
4. Modification of engines 7
5. Combinations of emulsions and aftertreatment 7
B. Problems related to emulsions 8
1. Engine corrosion 8
2. Stability of fuel 8
3. Power loss 9
C. Emission impacts of emulsions 9
1. NOX 9
2. PM 10
3. Other pollutants 11
V. Analysis of PuriNOx emissions data 12
A. Summary of available data on PuriNOx 12
B. Analysis of NOx effects 18
VI. Estimating emission impacts of PuriNOx for the in-use fleet 23
A. Comparison of PuriNOx data to protocol requirements 23
B. Discounted emission effects for use in SIPs 26
References 29
-------
I. Nature and Purpose of this Technical Report
This Report presents a technical analysis of the effect of Lubrizol's PuriNOx diesel/water
emulsion on exhaust emissions from diesel-powered vehicles. It analyzes pre-existing data from
various emissions test programs to investigate these effects. The conclusions drawn in this
Technical Report represent the current understanding of this specific technical issue, and are
subject to re-evaluation at any time.
The purpose of this Technical Report is to provide information to interested parties who
may be evaluating the value, effectiveness, and appropriateness of the use of PuriNOx. This
Report informs any interested party as to the potential air emission impacts of biodiesel. It is
being provided to the public in draft form so that interested parties will have an opportunity to
review the methodology, assumptions, and conclusions. The Agency will also be requesting
independent peer reviews on this draft Technical Report from experts outside the Agency.
This Technical Report is not a rulemaking, and does not establish any legal rights or
obligations for any party. It is not intended to act as a model rule for any State or other party.
This Report is by its nature limited to the technical analysis included, and is not designed to
address the wide variety of additional factors that could be considered by a State when initiating
a fuel control rulemaking. For example, this Report does not consider issues such as air quality
need, cost, cost effectiveness, technical feasibility, fuel distribution and supply impacts, regional
fleet composition, and other potentially relevant factors.
State or local controls on motor vehicle fuels are limited under the Clean Air Act (CAA) -
certain state fuel controls are prohibited under the Clean Air Act, for example where the state
control applies to a fuel characteristic or component that EPA has regulated (see CAA Section
21 l(c)(4)). This prohibition is waived if EPA approves the State fuel control into the State
Implementation Plan (SIP). EPA has issued guidance describing the criteria for SIP approval of
an otherwise preempted fuel control. See "Guidance on the Use of Opt-in to RFG and Low RVP
Requirements in Ozone SIPs," (August, 1997) at: http://www.epa.gov/otaq/volatility.htm.
The SIP approval process, a notice and comment rulemaking, would also consider a
variety of technical and other issues in determining whether to approve the State fuel control and
what emissions credits to allow. An EPA Technical Report like this one can be of value in such
a rulemaking, but the SIP rulemaking would need to consider a variety of factors specific to the
area, such as fleet make-up, refueling patterns, program enforcement and any other relevant
factors. Additional evidence on emissions effects that might be available could also be
considered. The determination of emissions credits would be made when the SIP rulemaking is
concluded, after considering all relevant information. While a Technical Report such as this may
be a factor in such a rulemaking, the Technical Report is not intended to be a determination of
SIP credits for a State fuel program.
-------
II. Draft Review Process
This draft technical report describes a methodology for quantifying the effect of PuriNOx
on exhaust emissions of regulated pollutants using data that has been collected by Lubrizol. We
are making this methodology available to the public in order to take comment before we approve
a set of specific emission benefits that can be attributable to PuriNOx. Following EPA's peer
review guidelines, we are also submitting this draft technical report to outside experts to obtain
independent peer review. After we receive comments from interested parties and our
independent peer reviewers, we will make any modifications to the analysis deemed appropriate
and release a final technical report.
Comments on this draft technical report will be accepted through January 15, 2002.
Comments may be sent to David Korotney at korotney.david@epa.gov, or through regular mail
to:
David Korotney
U.S. EPA National Vehicle and Fuel Emissions Laboratory
2000 Traverwood Drive
Ann Arbor, MI 48105
-------
III. Background
In September of 2001, Lubrizol approached the EPA with a collection of emissions data
on its water/diesel emulsion called PuriNOx. Lubrizol sought formal EPA verification of the
emission benefits of PuriNOx for use in marketing and for emission credit accounting purposes
in SIPs. Lubrizol was already in negotiations with officials at the Texas Natural Resources
Conservation Committee concerning use of PuriNOx in the Houston area to meet their
significant NOx shortfall. We confirmed that Texas was indeed seriously considering the use of
PuriNOx as one element of its overall strategy, and would be looking to the EPA to establish the
exact level of NOx benefits that could be claimed in a SIP. Based on this near-term need, we
agreed that it would be appropriate to assign a specific set of emission benefits to PuriNOx.
Lubrizol had already obtained an emissions verification from the California Air
Resources Board (ARB) in January 2001 under their "Interim Procedure for Verification of
Emission Reductions for Alternative Diesel Fuels." This verification, based on new testing
conducted by ARB, established the NOx and PM reductions associated with PuriNOx as being
14 and 63 percent, respectively. PuriNOx can now be assigned these reductions in any California
credit trading program or grants program as well as their SIP. The emissions data collected by
the ARB added to a pre-existing database assembled by Lubrizol and analyzed by Air
Improvement Resources (AIR) in April 2001. This database contained eight highway and
nonroad engines and nearly 50 separate tests. In analyzing this data, AIR concluded that
PuriNOx produced average NOx and PM reductions of 19 and 54 percent, respectively.
The obvious context in which emission benefits of a fuel or fuel additive should be
verified is the fuels testing protocol being developed by EPA1. This protocol requires a specific
amount of emissions data to be collected on engines falling into several categories under certain
test conditions, and specifies the analytical framework under which benefits should be quantified.
However, at the time that discussions between Lubrizol and the EPA were going on, this protocol
did not exist. The EPA determined that it would be appropriate to formally assign a set of
benefit estimates to PuriNOx for use in SIPs using existing data based on:
1. The substantial volume of existing emissions data on PuriNOx
2. The interest in the use of PuriNOx in both California and Texas
3. The analyses performed and conclusions drawn by other parties about PuriNOx
4. Emulsions had been studied for several decades and the combustion mechanisms
involved in NOx reductions are largely understood
We decided not to require Lubrizol to wait to submit an application under the fuels testing
protocol, but the emerging protocol would provide a reference point for what constitutes a
sufficient dataset during our analysis of existing PuriNOx data.
-------
The data that Lubrizol collected on PuriNOx included a wide variety of equipment types,
test conditions, and contexts. Not all were directly relevant to the estimation of emission benefits
for use in SIPs, though they might be useful in a qualitative fashion. For instance, additional data
that Lubrizol collected but which was not used to quantify the emission benefits of PuriNOx
included:
• Data on versions of PuriNOx containing different amounts of water than the 20%
water/80% diesel blend that Lubrizol ultimately decided to register with the EPA.
• Data on engines that had been recalibrated/repowered for the use of PuriNOx.
Since the primary conditions under which PuriNOx was to be used would involve
engines that had not been specifically recalibrated/repowered for the use of
PuriNOx, this data was not directly relevant.
• Data on chassis running on dynamometers. Test cycles for chassis cannot exactly
duplicate the Federal Test Procedure (FTP) required for engine certification. In
addition, emission inventories are based on engine certification data, not chassis
data.
• Data collected via portable monitors on equipment in the field. The
representativeness of data from these monitors has not yet been demonstrated.
We also determined that any data collected on steady-state cycles would not be appropriate for
determining emission effects of PuriNOx for PM or CO, based on previous investigations of test
cycle impacts on fuel/emission effects. However, even with these exclusions, we had a
substantial set of emissions data on PuriNOx permitting us to quantify its effects on emissions.
-------
IV. Historical research on emulsions
Aqueous emulsions have been studied for at least 30 years as an alternative to
conventional gasoline and diesel fuels. The cooling effect of the presence of water in an engine
has been known to reduce emissions from engines, specifically oxides of nitrogen (NOX) and
particulate matter (PM). Until recently, emulsions have not actually been considered as a viable
option for use in commercial engines due to various outstanding problems (see Section IV.B).
However, manufacturers and researchers are now pursuing emulsions more enthusiastically as
alternative fuel options and are optimizing them for maximum emissions reduction potential.
The idea behind fuel-water emulsions arose from the idea of using water as a cooling
mechanism for engines. Direct injection of water was an idea that investigators studied prior to
emulsions. It was widely known that the cooling effect of water could help to reduce NOX
emissions and the delayed effect that water had on injection could also serve to reduce PM
emissions. However, the feasibility of this idea was questioned in light of the wear and tear on
the engine due to water coming in contact with engine parts. Fuel-water emulsions were
considered a more feasible alternative to direct water injection due to the fact that water could be
mixed with the fuel prior to introducing the fuel into the engine. Unlike water injection, there
would not have to be a separate supply mechanism in the engine to introduce the water.
A. Studies of water impacts on combustion
1. Early efforts to reduce combustion temperatures
The idea that temperature is the dominant effect on NOx emissions has been widely
known. Thus, efforts to decrease peak combustion temperatures, and subsequently NOx, has
been studied by many researchers. Beginning research to achieve reductions in temperature
included methods to dilute the fuel-air mixture with an inert or noncombustible substance. This
idea was then replaced with the idea of injecting an inert or noncombustible liquid-phase
substance into the engine. This substance would have the characteristic of being able to reduce
the temperature through charge dilution without adding a significant loss in power. Because of
its high heat of vaporization, low vapor pressure, and low boiling point, water was chosen as the
ideal cooling substance.
The concept of adding water to engines as a cooling mechanism was first studied in the
form of direct injection of water into the engine, where cooling water is injected into the cylinder
at a predetermined rate. This approach had been studied as far back as the 1930s in gasoline
engines to reduce engine knock and in aircraft engines during World War n. It had been widely
known that water can be injected into the manifold to reduce peak temperatures and thus
decrease emissions of NOx. This idea had not been fully explored due to the cost of providing
auxiliary water tanks and injection systems on vehicles. In addition, direct water injection tended
-------
to produce some engine corrosion problems that led researchers to develop the idea of mixing the
water with the fuel in an emulsion to help decrease the amount of water coming into contact with
engine components and to provide for more thorough mixing of the water and fuel.
2. Micro-explosion phenomena
The reason behind the cooling effect of a fuel emulsion is a result of the micro-explosion
phenomena. Micro-explosions, also called secondary evaporation, help to accelerate the
evaporation of fuel droplets in emulsions and are strong enough to eject fragments of torn
droplets several millimeters away from the spray boundary at high speeds, which can help to
improve the air-fuel mixing.
Micro-explosions are extremely important in macro emulsions in which the emulsifier-
encased water droplets are suspended within the fuel. On emerging from the injector into the
combustion chamber, the lighter components quickly evaporate, breaking the fuel/water droplets
into multiple smaller droplets. This effect is due to the fact that micro-explosions occur in what
is commonly termed as an 'eddy', or a turbulent wave front. When one droplet of emulsified fuel
explodes, then a pressure wave may induce all of the droplets within the same eddy to explode at
once.
Ambient temperature has an important influence on the occurrence and the strength of the
explosions. At an optimum temperature, the water droplets go to a superheated state and
vaporize and explode at the same time (tearing up the droplet and expanding the spray volume).
However, because of the strong intermolecular forces of water, if the temperature is not high
enough, the water in the emulsion droplets will evaporate before being heated, resulting in no
micro-explosion. Likewise, if the temperature is too high, some of the droplets explode
immediately, and some droplets do not reach the superheated state as soon. Both of these
situations can result in weaker explosions due to the fact that all of the explosions are not
occurring simultaneously.
Droplet size is also of importance in micro-explosions. If the initial droplet size is too
small, the available water in the fuel will tend to evaporate before the droplet reaches a
superheated state. Thus, either no explosion will occur, or the explosion is too small to be
observed.
3. Determination of amount of water to be used
Many studies not only looked at the effect of the addition of water to diesel engines, but
also the effect based on the percentage of water added. As each study used different types of
engines and performed the tests over various cycles, no conclusive 'optimum' percentage was
found, however all studies did experience significant problems in performance and a decrease in
-------
emissions benefits (and in some cases, an emissions increase) when the water content was 50%
or greater. It was theorized by scientists that did work on the impact of micro-explosions that
emulsions with a larger concentration of water may actually lose more water, as a large
concentration of water will significantly lower the surface temperature. The ambient temperature
is therefore not high enough to produce many micro-explosions due to the surface tension of the
water (which is greater due to the increase in water).
4. Modification of engines
Early studies suggested that to achieve emission reduction benefits and to reduce
operational problems, significant engine modifications would be necessary before using
emulsions. Specifically, adjustment of the injection timing for optimization of the engine
performance might be necessary. Many believed that the power loss that resulted from the
increased ignition delay negated the PM benefits that emulsions achieved. Further investigation
into the use of water emulsions found that optimization of the water concentration, use of
surfactants and additives, and high shear mixing were actually needed to achieve emissions
benefits to make emulsions a less costly, simple alternative to hardware-based efforts to reduce
emissions in existing diesel engines. Current emulsions may be dispensed directly into an engine
(without the need for engine modifications) without sacrificing emissions performance.
5. Combinations of emulsions and aftertreatment
Many studies did work to look at the effect of using emulsions combined with other
emission reduction technologies, such as exhaust gas recirculation (EGR), oxidation catalysts,
and particulate traps. Use of emulsions with these technologies does not have a significant
negative impact on emissions, and in many cases was found to result in lower emissions when
employed simultaneously.
EGR is best when used at low to middle load ranges. When used alone, EGR has shown
no significant impact on NOX emissions and can increase fuel consumption and particulates when
employed at high loads. However, improvements in combustion and emissions benefits with the
use of emulsified fuels occurs at high load operations. At these conditions, when EGR is used in
conjunction with emulsified fuel, NOX is reduced and the increase in smoke and fuel
consumption are suppressed. Compared to conventional fuel, emulsified fuel with EGR results
in lower NOX concentrations regardless of temperature and engine loading.
In a London bus study2 the exhaust emissions from an engine running with an ultra-low
sulfur diesel fuel were compared to those of the engine with an oxidation catalyst, an oxidation
catalyst with emulsified fuel, a particulate trap, and a particulate trap with emulsified fuel. It was
found that emissions reductions for HC and NOX are slightly better when only the catalyst is
used. However, reductions for all pollutants from the baseline fuel are observed when both the
-------
catalyst and emulsified fuel are used. Similarly, for emulsified fuel with a particulate trap,
emission reductions were observed for all pollutants except CO.
B. Problems related to emulsions
1. Engine corrosion
Early emulsions researchers did not have a complete understanding of the additives and
conditions necessary to keep emulsified fuel from stratifying and there was a tendency for the
water and fuel to separate within days of mixing. This led to problems with corrosion of engine
parts as the water was able to come into direct contact with the engine. The increased density
due to the addition of water also tended to break injector tips in spark ignition engines. Since
this problem was associated primarily with the stability of emulsions, it has been addressed
largely through advances in maintaining fuel storage stability.
2. Stability of fuel
Emulsions are not solutions of water and diesel fuel, but rather are suspensions of (in the
case of PuriNOx) water droplets in fuel. Since water is more dense than diesel fuel, water
droplets have a tendency to settle if the fuel is not agitated periodically. The process generally
begins with agglomeration wherein water droplets collect into groups, followed by coalescence
into larger droplets, and finally settling. The result is that the fuel becomes increasingly
stratified, with a higher concentration of water in the bottom of a tank than at the top of the tank.
This process will occur under any conditions and cannot be eliminated, but it can be slowed
using a number of different approaches.
Emulsion stability can be improved by increasing the surfactant concentration, as
surfactants resist the droplet tendency to coalesce. Droplet size is also an important component
in improving emulsion stability. Smaller droplets settle slower which enhances stability. To
achieve this, high shear mixing is needed to break larger droplets into smaller ones. Stability of
the emulsion can also be improved by increasing the treat rate of the additive package and the
level of anticreamer (which reduce the amount of oil separation and water separation,
respectively). By taking samples at the top, middle and bottom of the fuel sample, some
researchers found that after two months, there was a negligible amount of settling of the larger
droplets.
Finally, fuel storage stability can be maintained by specifying the minimum interval
within which the fuel must be agitated. Most emulsion producers include this type of condition
on the viability of their products.
-------
3. Powerless
The loss in engine power is usually proportional to the concentration of water in the
emulsion, though some emulsion products such as PuriNOx claim a less than strictly
proportional loss in engine power. Although the engine can be "repowered" to ensure that the
same amount of hydrocarbon fuel reaches the combustion chamber as would be the case if no
water was present, many applications of emulsions assume no such engine changes. In these
cases, the engine is effectively derated, i.e. the peak rated power cannot be reached on the
emulsified fuel. Under more typical conditions when the engine is operated below its rated
power level, an operator would generally only notice that it required somewhat greater pressure
on the accelerator to maintain the same level of vehicle performance.
In some experiments, emission reduction results from using water emulsions were
compared with using the emulsion on the same engine that was repowered. The results showed
that the further reductions in emissions due to repowering the engine were only 2-10% greater
than without repowering3. In fact, data from one study4 shows that there was actually a decrease
in emission benefits when an engine was repowered. For this engine, emission benefits for NOx
and PM were 1.6 percent and 2.2 percent greater when the fuel was used as fill-and-go (i.e.
without repowering the engine).
C. Emission impacts of emulsions
The addition of water to diesel fuel has been proven to significantly reduce NOX and PM
emissions. These benefits have been observed both with and without modifications to the test
engine. It was discovered that the effectiveness in lowering the peak combustion temperature is
dependent on the engine timing and decreases as engine timing is advanced. Greater reductions
in both PM and NOX can also be attained with the combined use of emulsions and aftertreatment
technologies such as PM traps and oxidation catalysts.
1. NO
X
NOX emissions increase exponentially with the rise in combustion temperature. The
evaporation of the water in the emulsified fuel helps to lower the peak combustion temperature,
which impedes the formation of NOX. Studies have found that, in emulsions of up to 50% water,
NOX emission reductions have an approximate linear relationship with the percentage of water
added to the fuel.
Some of the earliest studies found that NOX emissions increased when the engine was
operated with low water content (10-20%) at idle and low engine speeds. However, under
normal engine running conditions with emulsions that have been optimized, NOX emissions
decrease.
-------
The cooling effect that water has on combustion temperature may be more beneficial in
terms of NOx reductions if the peak combustion chamber temperature is higher. This result
stems from the fact that the relationship between combustion temperature and NOx formation is
highly nonlinear, so that a fixed reduction in temperature will have a different impact on NOx for
engines with different base combustion temperatures.
2. PM
It is often difficult for an emission reduction technology to reduce NOX without causing
adverse effects on particulate emissions and fuel consumption. Water emulsions have been
shown to eliminate the concern of the NOX-PM tradeoff as benefits have been achieved for both
NOX and particulates simultaneously.
The initial pre-mix burn period of the combustion cycle is fuel rich and leads to the
formation of polycyclic aromatic hydrocarbons, or PAHs, precursors to soot/PM. Soot is formed
from partially burned fuel products of this combustion. While the majority of the soot is burned
with the fuel, this unburned portion becomes an exhaust emission- PM.
The introduction of water alters the quantities of fuel and air during the pre-mix period
and acts as a source of oxygen, thus making it a less fuel rich environment and allowing for less
soot to be oxidized and left as a pollutant. Reductions in particulate emissions tend to be on the
order of two to three times the amount of water added to the fuel.
An early study on water emulsions found that engine timing does not have an effect on
the kinetic chemical changes that result from the water, and thus advancing the timing will not
have a significant effect on PM reduction unless the timing is severely advanced (at which time
the engine may have severe performance problems). However, later studies have found that the
ignition delay with the use of emulsified fuels tends to be 2-5 degrees longer than with
conventional diesel fuel, which results in enhanced combustion. More complete combustion
leads to less products of fuel left in the engine that can form into pollutants.
Previous concerns with water emulsions dealt mainly with the loss in engine power.
Many felt that this was the reason behind the reductions in PM, however a study performed on a
UK mining drill rig found that the impact on PM is not simply the result of the loss in power5.
One study6 looked specifically at the chemical impact that the water had on in-cylinder soot
formation. A single cylinder engine derived from a six-cylinder heavy duty diesel engine was
used in the study. Using this engine, the in-cylinder temperature was reduced to match the heat
release curve of that of a 10% emulsified fuel and results were then compared with those of an
engine running on conventional diesel fuel at a decreased temperature. The study reported a
greater decrease in soot formation in the pre-mix portion of combustion and throughout the entire
combustion cycle. This result indicated that the inclusion of water chemically alters combustion
and inhibits soot formation.
10
-------
3. Other pollutants
For the majority of the work that was performed with regard to water emulsions only NOX
and PM were the pollutants of concern. Some more recent studies made efforts to look at other
pollutants such as CO and HC. A general trend that was found in many of these studies is that
CO and HC tended to increase with the use of emulsified fuels. However, some studies also
noted that while HC did increase, the total ozone precursors (HC+NOX) decreased. Also, even
significant increases in HC on a percentage basis often did not result in an exceedence of the
engine certification standard. The only work that actually showed a decrease in HC emissions
was a study of PuriNOx performed on a London bus in conditions when the emulsified fuel was
used in conjunction with an oxidation catalyst or a particulate trap.
11
-------
V. Analysis of PuriNOx emissions data
PuriNOx is the Lubrizol Corporation's commercially available diesel-water emulsion.
The summer version of the fuel which is the focus of our analysis is comprised of approximately
20% water and small amounts of a proprietary additive package that includes the emulsifier as
well as other additives that assist in storage stability, cold weather operability, etc. PuriNOx is
intended to be primarily a 'fill-and-go' technology, meaning that engine modifications are not
required or even encouraged. Thus no data on repowered engines was included in our analysis.
As described in Section HI, we also excluded data collected on chassis, in-use emission monitors,
and alternative versions of PuriNOx having different concentrations of water. Steady-state
emissions data for PM and CO was also excluded for both highway and nonroad engines.
This Section summarizes our review of the available engine data on PuriNOx. We first
present a summary of the engines tested, the test conditions, and the statistical significance of the
results. We then present a more in-depth analysis of the NOx effects, since the effects of
emulsions on NOx emissions has been studied extensively in the past. Section VI presents a
methodology for applying the estimated emissions effects for PuriNOx to the in-use fleet.
Comments are requested on all aspects of our analysis.
A. Summary of available data on PuriNOx
There were a total of thirteen engines tested on PuriNOx which met our analytical
requirements. Some of these engines were tested in multiple conditions, e.g. with and without an
oxidation catalyst, with two different lubricant oils, or on two different test cycles. Each of these
unique operating conditions was identified as a unique engine in our database. Each engine and
the number of PuriNOx observations made is shown in Table V.A-1.
12
-------
Table V. A-1
Engines tested on PuriNOx
ID
H-a
H-bl
H-b2
H-cl
H-c2
H-d
H-e
H-f
N-al
N-a2
N-a3
N-a4
N-bl
N-b2
N-c
N-d
N-e
N-f
N-g
Engine
'96 DDC Series 50 w/ catalyst
'99 DDC Series 60, lube oil #1
'99 DDC Series 60, lube oil #2
'91 DDC Series 60, lube oil #\a
'91 DDC Series 60, lube oil #2
'00 DDC Series 50 w/ EGR
'94 Caterpillar 3 176
'01 Cummins5.9L
'99 Perkins 1004.4T on high sulfur
'99 Perkins 1004.4T on low sulfur
'99 Perkins 1004.4T on high sulfur
'99 Perkins 1004.4T on low sulfur
'95 DDC 6V92 w/ catalyst
'95 DDC 6V92
'00 Caterpillar 3 508
'90 Caterpillar 3306
'96 Caterpillar 3406B
'85DeutzF8L413
'96DeutzF6L912
Use
Highway
Highway
Highway
Highway
Highway
Highway
Highway
Highway
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Nonroad
Group13
HH
HH
HH
HH
HH
EGR
HH
MH
0 - 100 hp
0 - 100 hp
0 - 100 hp
0 - 100 hp
175-300 hp
175-300 hp
300+ hp
175-300 hp
300+ hp
175-300 hp
100-175 hp
Test
cycle
FTP
FTP
FTP
FTP
FTP
8 mode
8 mode
FTP
Euro trans
Euro trans
8 mode
8 mode
8 mode
8 mode
8 mode
8 mode
8 mode
4 mode
8 mode
PuriNOx
observations
O
3
O
21
O
2
1
1
3
O
3
O
1
1
1
1
1
1
1
a For all test programs in this table, the base fuel was used both as a reference fuel and also to produce PuriNOx. For
engine H-cl, the base fuel met the California specifications for highway diesel fuel while the fuel used to produce
PuriNOx was a commercial California diesel fuel. This disparity is expected to result in a more conservative estimate of
emission benefits.
"Group" is a weight class for highway engines and a horsepower group for nonroad engines, consistent with the fuels
testing protocol. See Section VI for details. LH = light-heavy, MH - medium-heavy, HH = heavy-heavy, EGR = exhaust
gas recirculation
The data collected on engine H-cl was generated under the auspices of the emissions
verification protocol established by the California Air Resources Board. This "Interim Procedure
for Verification of Emission Reductions for Alternative Diesel Fuels" required that the fuel used
as a baseline should meet CARB's specifications for diesel fuel sold in California, including a 10
vol% cap on aromatics content. The conventional diesel fuel used to make PuriNOx, however,
was required to be a commercial diesel fuel. Thus the fuel used to establish a baseline was
different than the fuel used to produce PuriNOx. Although for our analysis we only considered
data for which the baseline fuel was the same as the fuel used to make PuriNOx, the approach
taken by CARB was thought to result in a more conservative estimate of the emission benefits of
alternative diesel fuels tested under their protocol. We therefore chose to leave this substantial
amount of data in our database.
13
-------
Lubrizol also collected baseline data on the commercial diesel fuel that was actually used
to make PuriNOx. Only two emission measurements were made with this commercial diesel
fuel, and this data was not included in the CARB verification nor in our analysis. However, this
data does provide a more direct comparison of PuriNOx with conventional diesel fuel. It may,
therefore, be appropriate to replace the 21 measurements on the CARB specification fuel with the
two measurements on the commercial diesel fuel. Doing so would reduce the amount of data
available for our analysis, but may provide a more apples-to-apples comparison of the emission
effects of PuriNOx. We welcome comment on this alternative approach.
The emissions measurements on PuriNOx for the engines listed in Table V.A-1 showed a
reduction in NOx and PM in every single case, and an increase in HC in almost every single case.
CO effects of PuriNOx were less consistent. Figures V.A-1 through V.A-4 show the measured
emission impacts of PuriNOx for every engine.
14
-------
a
03
"3
fc
"S
O
c
u
O
O
O
ON
.£5
03
-^
C
: emissi
x
O
0
•c
fin
O
cx
^
s
c
1
(
I
1
1
1
U
|
I
1
1
1
|
•
1
|
|
1
1
1
|
1
1
appppppc
O to C4 C4 T- T-
suoissjoia u| uojionpaj %
z
»*J.
z
•o
z
V
z
CM
||
Z
5
z
9
z
to
9
2 g
*— OJ
TO f^-
Z '5)
9
*Y
X
X
X
C4
V
X
V
X
C4
X
X
9
X
-------
>
03
C
•j3
^
"§
O
°03
03
's
(L)
O
K
0
|7j
•c
fin
O
-M
O
1
\
\
\
v° v° ^
\
\
u
3 V0
o o o o
o o o o
m in o
V
B
1
1
^0 ^0 ^0 ^0 ^C
Z
z
z
V
z
CM
i
Z
,_
z
9
z
to
9
^
Z Q
—
m ^
i C
Z '5)
c
9
z
^
•
i
0)
i
^
=
CM
V
V
X
i
5
i
9
3
O O O O O
O O O O O
m o m o m
T- CM CM tO tO
suojssjiua u| uojionpaj %
03
"3
CL>
-M
CL>
Q
CL>
O
O
O
ON
J3
'I
^T c
>\ Q
(D ^3
^«
P-H §
-03
03
's
CL>
0
o
0
•c
fin
O
-M
o
M
c
c
c
(
I
u
I
1
Ss 0s 0s c
3 O O C
i S S e
suojssjuia u
1
1
n
i
1
1
•> q q c
3 O O C
CM ? «
u0!1onp9J %
z
z
z
^
z
V
2
z
5
z
^
9
z
to
9
z y
CM CD
9 .g
Z CD
C
9 ^
*Y
i
¥
i
i
9
V
i
5
i
9
N,
1
i
-------
We conducted t-tests to determine if emission effects for highway engines could be
considered to be different than those for nonroad engines. For NOx, PM, and HC, highway and
nonroad were indeed distinguishable. We therefore calculated the average emission effects
separately for highway and nonroad for these three pollutants. To do this, we used a least-
squares regression of the following form:
In(emissions) = a x PURINOX + £(b; x ENG;)
where
In(emissions) = Natural log of NOx, PM, HC, or CO in g/bhp-hr
PURINOX = Categorical independent variable; 1 for PuriNOx and 0 for base fuel
ENG; = Categorical independent variable; 1 for engine i and 0 for all other engines
a = Regression coefficient representing the effect of PuriNOx
b; = Regression coefficient representing the effect of engine i
The average percent reduction in emissions can then be calculated for all engines from:
% reduction in emissions = [1 - exp(a)] x 100%
We used this least-squares approach rather than simply weighting the engine-specific average
emission effects by the number of observations because this latter approach would not have
permitted us to assess statistical significance. Results are given in Table V.A-2.
17
-------
Table V.A-2
Average % reduction for all engines
Highway engines
Average % reduction
Probability that average is different than zero
98% confidence interval
Lower bound of % reduction
Upper bound of % reduction
Nonroad engines
Average % reduction
Probability that average is different than zero
98% confidence interval
Lower bound of % reduction
Upper bound of % reduction
NOx
13.7
0.9999
12.7
14.8
24.4
0.9999
22.3
26.3
PM
58.0
0.9999
55.6
60.2
27.7
0.9999
16.8
37.1
HC
-87.2
0.9999
-120.2
-59.2
-79.0
0.9999
-100.1
-60.1
coa
22.0
0.9999
13.4
29.7
22.0
0.9999
13.4
29.7
CO calculation was done with highway and nonroad data together. Results are shown to be identical for
highway and nonroad.
As can be seen in Table V.A-2, all emission impact estimates are highly significant. Given the
variety and number of engines tested, these results suggest that PuriNOx will produce significant
reductions in NOx and PM for the in-use fleet.
B. Analysis of NOx effects
We also investigated an alternative approach to estimating the effect that PuriNOx has on
NOx emissions. Based on previous research on emulsions and our understanding of the cooling
effect that water has on combustion temperatures, we hypothesized that PuriNOx would have a
bigger impact on NOx emissions when combustion temperatures were high and a smaller impact
on NOx when combustion temperatures were low. Since combustion temperature is correlated
with NOx emissions, we plotted the average % reduction in NOx for each engine with that
engine's average base fuel NOx emissions. We included both highway and nonroad data in this
analysis. The results are shown in Figure V.B-1. Each "bubble" represents the average effect of
PuriNOx on a single engine, and the size of each bubble is proportional to the amount of data
collected on that engine.
18
-------
Figure V.B-1
Bubble plot of engine-by-engine PuriNOx effects on NOx
40.0%
35.
0.0%
4 Q 8 10 12
Base NOx, g/bhp-hr
14
From this graph it appears that the effects of PuriNOx on NOx emissions are in fact correlated
with the base fuel NOx emissions. This correlation provides some explanation for why highway
and nonroad NOx effects are so dramatically different (13.7% versus 24.4% as shown in Table
V.A-2). Highway engines are subject to more stringent controls than nonroad engines, and as a
result have lower NOx emission rates.
We generated a correlation between percent reduction in NOx due to the use of PuriNOx
and base NOx emissions using least-squares regression analysis. This regression produced the
following equation:
% reduction in NOx = [1 - exp(-0.01052 - 0.03358 x base NOx)] xlOO%
where "base NOx" represents the average NOx emissions in g/bhp-hr when the engine is
operated on conventional diesel fuel. This equation can be used to represent both highway and
nonroad engines, since the regression equation was based on all engines in the database and
because it appears to explain most of the differences between highway and nonroad effects of
PuriNOx on NOx.
We can use the above correlation to predict the impact that PuriNOx would have on NOx
for the in-use fleet. This calculation requires that we predict the impact of PuriNOx on the
19
-------
heavy-duty highway (or nonroad) NOx inventory in each calender year using the model-year
specific NOx emissions. Since NOx standards have been decreasing over time, the fleet-wide
impact of PuriNOx would as well. This trend is shown in Figures V.B-2 and V.B-3 for highway
and nonroad engines, respectively. These graphs also include estimated 98% confidence
intervals'1.
Figure V.B-2
Predicted NOx impact of PuriNOx for in-use heavy-duty highway fleet
20.0%
x
O
15.0%
I 10.0%
o
D
T3
CD
5.0%
0.0%
Upperbobnd
^confidence
Lower bound of 98%
confidence interval
of 98%
interval
2000
2005
2010
2015
2020
The upper and lower bounds actually represent prediction intervals rather than confidence intervals. Whereas
confidence intervals encompass the true population mean, prediction intervals encompass additional, single
measurements. As such, prediction intervals are wider than confidence intervals. The utility of the prediction interval is
described in more detail in Section VI.
20
-------
Figure V.B-3
Predicted NOx impact of PuriNOx for in-use heavy-duty nonroad fleet
30.0%
25.0%
x
T3
CD
2 20.0%
15.0%
10.0%
5.0%
0.0%
Lower bound
confidence
: 98%
Upper bound o
confidence inteh/al
of 98%
interval
2000
2005
2010
2015
2020
For the purpose of accounting for the use of PuriNOx in a SIP, one would need to use
NOx benefit estimates representing the calender year in which attainment was being modeled in
the SIP. For many SIPs this is 2007. Table V.B-1 compares the predicted 2007 NOx benefits of
PuriNOx based on the regression lines shown in Figures V.B-2 and V.B-3 with the average
values presented in Table V.A-2.
21
-------
Table V.B-1
Comparison of % reduction in NOx via two alternative approaches
Highway engines
Average
Low end of 98% confidence interval
High end of 98% confidence interval
Nonroad engines
Average
Low end of 98% confidence interval
High end of 98% confidence interval
Simple average
(Table V.A-2)
13.7
12.7
14.8
24.4
22.3
26.3
Predicted 2007 benefit
based on correlation
with base NOx
12.9
9.0
16.5
20.2
15.7
24.4
22
-------
VI. Estimating emission impacts of PuriNOx for the in-use fleet
As described in Section HI, the fuels testing protocol under which alleged emission
benefits of fuels or additives can be verified was not available during the time that EPA was
assessing the emission benefits of PuriNOx. We therefore decided not to require Lubrizol to wait
to submit an application under the fuels testing protocol.
However, an analytical approach to estimating PuriNOx emission impacts that maintains
some level of conceptual similarity to the protocol might be prudent. To this end we have
compared the available PuriNOx data to the protocol's requirements for minimum number of
engines, minimum number of observations, and engine groupings. In this Section we first
present these elements of the fuels testing protocol and show how the existing PuriNOx data
compares. We then present a methodology for discounting the estimated emission impacts of
PuriNOx for use in SIPs.
A. Comparison of PuriNOx data to protocol requirements
The fuels testing protocol requires that data be collected on engines falling into several
groups. For highway engines, these groups are defined by weight class, with an additional group
for engines equipped with exhaust gas recirculation (EGR). For nonroad engines these groups
are defined by horsepower. Table V.A-1 identified the engine group assignments for every
engine tested on PuriNOx. We have estimated the average emission effects of PuriNOx for each
of these groups using the least-squares regression approach described in Section V.A. For NOx,
rather than use the simple average effects for each engine group, we have applied the predicted
2007 highway and nonroad effects shown in Table V.B-1 to each engine group, since these
benefit estimates represent our best understanding of how PuriNOx affects NOx. The results are
shown in Table VI. A-1.
23
-------
Table VI. A-1
PuriNOx emission impacts by protocol group
Highway engines
Light-heavy duty
Medium-heavy duty
Heavy-heavy-duty
EGR-equipped
Nonroad engines
0- 100 hp
100 - 175 hp
175- 300 hp
300 + hp
% reduction due to the use of PuriNOx
NOx
12.9
12.9
12.9
12.9
20.2
20.2
20.2
20.2
PM
n/a
51.1
58.2
n/a
26.1
n/a
n/a
n/a
HC
n/a
-116.7
-87.8
-63.6
-99.2
-111.1
-53.7
0.0
CO
n/a
-25.2
33.3
n/a
-34.7
n/a
n/a
n/a
Under the requirements of the protocol, emission effects of a given fuel/additive for each
engine group such as those in Table VI. A-1 would be applied to the inventory associated with
that group in the context of a SIP. However, the protocol also requires a minimum number of
engines and minimum number of emission measurements for every group, and the existing
PuriNOx data does not meet these requirements in every case. A summary of these protocol
requirements and the degree to which the existing PuriNOx data fulfills these requirements is
given in Tables VI.A-2 and VI.A-3.
24
-------
Table VI.A-2
Number of engines tested
Highway engines
Light-heavy
Medium-heavy
Heavy-heavy
EGR
Nonroad engines
0 - 100 hp
100- 175 hp
175 - 300 hp
300 + hp
Protocol
requirements
1
2
2
2
2
2
2
1
NOx
0 (0%)
1 (50%)
4 (100%)
1 (50%)
1 (50%)
1 (50%)
3 (100%)
2 (100%)
PM
0 (0%)
1 (50%)
3 (100%)
0 (0%)
1 (50%)
0 (0%)
0 (0%)
0 (0%)
HC
0 (0%)
1 (50%)
4 (100%)
1 (50%)
1 (50%)
1 (50%)
3 (100%)
2 (100%)
CO
0 (0%)
1 (50%)
3 (100%)
0 (0%)
1 (50%)
0 (0%)
0 (0%)
0 (0%)
Table VI.A-3
Number of emission measurements'11
Highway engines
Light-heavy
Medium-heavy
Heavy-heavy
EGR
Nonroad engines
0 - 100 hp
100- 175 hp
175 - 300 hp
300 + hp
Protocol
requirements
12
24
24
24
24
24
24
12
NOx
0 (0%)
2 (8%)
71 (100%)
4 (17%)
39 (100%)
2 (8%)
8 (33%)
4 (33%)
PM
0 (0%)
2 (8%)
69 (100%)
0 (0%)
21 (88%)
0 (0%)
0 (0%)
0 (0%)
HC
0 (0%)
2 (8%)
71 (100%)
4 (17%)
39 (100%)
2 (8%)
8 (33%)
4 (33%)
CO
0 (0%)
2 (8%)
69 (100%)
0 (0%)
21 (88%)
0 (0%)
0 (0%)
0 (0%)
An emission measurement is here defined as one test of a single fuel on a single engine. Two such
tests, one on base fuel and another on PuriNOx, are therefore required to provide a single estimate of
% change in emissions
Based on the above comparison between the protocol requirements and the existing
PuriNOx data, we can recommend additional testing that Lubrizol could conduct. For instance,
there is no data on light-heavy-duty highway engines. There is also very little data on medium-
heavy-duty highway, EGR-equipped highway, and nonroad engines falling in the 100 - 175 hp
group. Additional data from these four groups would go a long way towards rounding out the
25
-------
existing database. Given the more pronounced lack of data for PM and CO (due to the fact that
we disregarded any PM or CO measurements taken on steady-state test cycles), it might be
prudent to take more than the minimum number of emission measurements for any additional
engines that are tested on PuriNOx.
B.
Discounted emission effects for use in SIPs
Although the existing PuriNOx data would not be sufficient were Lubrizol required to
meet all the requirements of the protocol, we believe that it can still be used to estimate emission
impacts of PuriNOx for SIP purposes for several reasons:
1. Emulsions have been studied for several decades and the combustion mechanisms
involved in NOx reductions are largely understood
2. Every single NOx and PM measurement on PuriNOx showed a benefit
3. Chassis and in-use monitor data that we excluded from our analysis supports the
average effects shown in Table V. A-2.
However, when estimating the emission impacts of PuriNOx for SIP purposes, it may be
prudent to take into account the fact that the existing PuriNOx data does not meet all the
requirements of the protocol as shown in Tables VI. A-2 and VI.A-3. We have developed a
methodology for estimating the fleet-wide emission impacts of PuriNOx in a conservative
fashion that "discounts" the existing data based on the comparisons shown in Tables VI.A-2 and
VI.A-3. Our proposed approach permits the average emission effects presented in Table VI.A-1
to be applied to the in-use fleet only to the degree that the existing PuriNOx data meets the
minimum engine and emission measurement requirements of the protocol. Any "missing"
engines or emission measurements would then be assigned the low end of the confidence limit as
calculated in Section V. For each pollutant and engine group, this approach can be represented
mathematically as:
% reduction in emissions
for use in SIPs
(discount factor)^ x (ave % reduction)^
+ (1 - discount factor)^ x (LCF % reduction)^
where
(discount factor)^ =
(ave % reduction)^ =
(LCF % reduction), =
Discount factor for pollutant i and engine group j, derived below
Average % reduction in emissions for pollutant i and engine
group j from Table VI. A-1
Low end of confidence limit for % reduction for pollutant i and
engine group j
26
-------
We first calculated "discount factors" representing the degree to which the existing
PuriNOx data met the requirements of the protocol. To do this, we averaged the % values in
Tables VI.A-2 and VI.A-3 for each pollutant and engine group. The results are shown in Table
VI.B-1.
Table VLB-1
Discount factors
Highway engines
Light-heavy
Medium-heavy
Heavy-heavy
EGR
Nonroad engines
0 - 100 hp
100- 175 hp
175 - 300 hp
300 + hp
NOx
0.0
0.3
1.0
0.3
0.8
0.3
0.7
0.7
PM
0.0
0.3
1.0
0.0
0.7
0.0
0.0
0.0
HC
0.0
0.3
1.0
0.3
0.8
0.3
0.7
0.7
CO
0.0
0.3
1.0
0.0
0.7
0.0
0.0
0.0
We then identified the low end of the confidence limit for each pollutant. Since some
engine groups (e.g. the medium-heavy highway engine) had only a single measurement of the
effect of PuriNOx, we were unable to estimate a confidence interval for every pollutant and
engine group. Instead, we used the 98% confidence intervals for PM, HC, and CO that had been
estimated for highway and nonroad engines, and applied these values to all engine groups. These
values were presented in Table V. A-2.
Given the additional analysis we presented in Section V.B on NOx emission effects of
PuriNOx, we decided not to use the low end of the confidence interval for NOx as presented in
Table V. A-2. Instead, we determined that more realistic lower limits for the in-use NOx benefits
of PuriNOx would be based on the curves shown in Figures V.B-2 and V.B-3. These curves
include a "prediction interval" which establishes the interval within which a single additional
NOx emissions measurement would likely reside. The prediction interval is necessarily broader
than a confidence interval, as the confidence interval establishes the interval within which the
true population mean is likely to reside. Thus the lower end of the prediction interval provides a
very conservative estimate of the NOx benefits of PuriNOx. For our purposes, we chose a
calender year of 2007 since it is for this year that many nonattainment areas conduct their
inventory analyses in support of efforts to reach attainment. The final set of values representing
the low end of the confidence limit for use in calculating the in-use emissions impacts of
PuriNOx are shown in Table VI.B-2.
27
-------
Table VI.B-2
Low end of confidence limits (% reduction in emissions)
Highway engines
Nonroad engines
NOx
9.0
15.7
PM
55.6
16.8
HC
-120.2
-100.1
CO
13.4
13.4
Finally, we combined the discount factors from Table VI.B-1, the average emission effect
values from Table VI.A-1, and the low ends of the confidence limits from Table VI.B-2 to
estimate the emission impacts of PuriNOx that we believe could appropriately be used in SIPs.
The final values are given in Table VI.B-3. Note that for cases in which the lower end of the
confidence limit would actually increase the estimated benefits of PuriNOx when combined with
the average emission effects, we simply used the average effect for a more conservative estimate.
Table VI.B-3
Final emission impacts of PuriNOx (% reduction)
Highway engines
Light-heavy
Medium-heavy
Heavy-heavy
EGR
Nonroad engines
0 - 100 hp
100- 175 hp
175 - 300 hp
300 + hp
NOx
9.0
10.2
12.9
10.2
19.3
17.0
18.8
20.2
PM
55.6
51.1
58.2
55.6
23.3
16.8
16.8
16.8
HC
-120.2
-119.1
-87.8
-103.2
-99.4
-80.1
-72.8
-30.0
CO
13.4
-25.2
33.3
13.4
-34.7
13.4
13.4
13.4
In cases wherein the fleet mix is known, the emission impacts for each engine group given in
Table VI.B-3 could be applied separately. If the fleet mix is not known, the values for each
engine group may need to be weighted together according to some estimate of fleet distribution
appropriate to the area where PuriNOx is being used.
Note that the values for CO in Table VI.B-3 do not indicate a consistent increase or
decrease. This result is not surprising given that CO responses to PuriNOx from engine to
engine as shown in Figure V.A-4 could be either positive or negative. However, the statistically
significant overall CO benefit is heavily influenced by engine H-cl. One possible alternative
approach to generating an estimate of CO effects of PuriNOx is to assume that the lower end of
the confidence limit for CO is zero. Thus we request comment on assigning zero CO benefit for
cases in which the existing PuriNOx data falls short of the protocol's requirements.
28
-------
References
1. "Generic Verification Protocol for Determination of Emissions Reductions Obtained By Use
of Alternative or Reformulated Liquid Fuels, Fuel Additives, Fuel Emulsions, and Lubricants for
Highway and Nonroad Use Diesel Engines and Light Duty Gasoline Engines," EPA Cooperative
Agreement No. CR826152-01-03
2. Barnes, A., D. Duncan, J Marshall, A. Psaila, J. Chadderton, A. Eastlake, "Evaluation of Ester-
Blended Fuels in a City Bus and an Assessment of Performance with Emission Control Devices",
SAE Paper No. 2000-01-1915.
3. Ibid.
4. "Comparative Analysis of Vehicle Emission Using PuriNOx Fuel and Diesel Fuels", Air
Improvement Resource, Inc., April 4, 2001.
5. Duncan, D.A., D.A. Langer, J.C. Marshall, "Emulsion fuels- improving the environment
today", presentation at the SAE Conference, Vienna, Austria, April, 2001.
6. Ibid.
29
------- |