effect of fuel composition on pm · pdf file• the fuel effect on pm is also not addressed...
TRANSCRIPT
Effect of Fuel Composition on PM Emissions
LEV III WorkshopLEV III Workshop May 18th, 2010
J. Jetter, Honda R&D Americas, Inc.
LEV III Workshop May 18th, 2010 2 Overview
• Particulate Matter (PM) is a significant concern.
• Recent efforts to reduce PM focus on engines• Recent efforts to reduce PM focus on engines, not fuel. – Current LEV III proposal: 10mg 6mg 3mg/miCurrent LEV III proposal: 10mg 6mg 3mg/mi
• Honda theorized that better fuel specifications co ld red ce PMcould reduce PM. – This was confirmed by research and testing.
• Honda’s PM Index can lead to fuel specifications which reduce previously unaccounted variations in PM measurement
LEV III Workshop May 18th, 2010 3 Background
F l f l ti h f d ff t t il i i i f• Fuel formulation can have a profound effect upon tailpipe emissions of particulate matter (PM).
• This effect is not directly addressed in CaRFG3 market fuel specifications or in the proposed CaRFG3 certification fuel specifications, although some regulated properties (e.g. T90) do affect PM emissions.
• The fuel effect on PM is also not addressed in the current CaRFG3The fuel effect on PM is also not addressed in the current CaRFG3 predictive model.
Proposed CaRFG3 Certification Fuel Specifications
Pollutant Predictions Units
Oxides of Nitrogen (NOx) gm/mile
Exhaust Hydrocarbons (HC) gm/mile
Pollutants Included in Predictive Model
Evaporative Hydrocarbons (HC) Percent Change (Candidate Fuel Relative to Reference Fuel)
Exhaust Potency-Weighted Toxics (PWT) mg/mile
Evaporative Benzene mg/mile
Exhaust CO (Adjustment Factor for O )
gm/mile Oxygen)
Source: CARB’s “Preliminary Discussion Paper,” February 2010 Source: “California Procedures for Evaluating Alternative Specifications for Phase 3 Reformulated Gasoline using the California Predictive Model,” August 2008
LEV III Workshop May 18th, 2010 4 “PM Index” Concept
• Over the last year, Honda has been working on a model that can predict trends y , g pin tailpipe PM emission levels based on fuel properties. We call this model the “PM Index.”
• Overview of the development work:Overview of the development work:
• Pure hydrocarbons were added to an Indolene base fuel at nominally 10% by weight; i.e., each test fuel consisted of the base fuel spiked with one hydrocarbon.
PM emissions were measured over the NEDC cycle for each fuel PM number (in place• PM emissions were measured over the NEDC cycle for each fuel. PM number (in place of PM mass) was used as the metric to improve accuracy at lower PM emission levels.
• PM emissions were determined to primarily be a function of the DBE* and vapor f th f l tpressure of the fuel components:
C10 DBE 6
C10 DBE 7C10 DBE 5
C10 DBE 2
1.20E+12
1.60E+12
2.00E+12
m 9
0EM
) [#]
Alkene & AlkaneAromaNaphthalenes
∑ ⎟⎟⎠
⎞⎜⎜⎝
⎛×
+=
n
ii Wt
KPVDBE
IndexPM)443(
1
• This research, and subsequent test results, will be presented at SAE in October C12 DBE 0
C9 DBE 0
C9 DBE 5
C8 DBE 4
C9 DBE 4
0 00E+00
4.00E+11
8.00E+11P
N (I
ncre
ase
from
∑=
⎟⎠
⎜⎝i iKPV1 )443(.
will be presented at SAE in October.(2 papers)
C9 DBE 00.00E+00300 350 400 450 500 550
Chemical boiling point [K]
*DBE (Double Bond Equivalent) is essentially an indication of the number of double bonds or rings present in the molecule, and can be easily calculated from the molecular formula.
Example of test fuels, in this case arranged by BP, DBE, and PM number increase from the base fuel.
LEV III Workshop May 18th, 2010 5 Confirmatory Tests Performed by Honda
• EUDC cycle • Test fuels and market fuels
• 120 km/h steady-state • Test fuels and market fuels
RDX NEDC
2E+12
2.5E+12RDX 120km/h steady state
1.4E+13
1.6E+13
Test fuels and market fuels• Turbo-charged PFI vehicle
• Test fuels and market fuels• Turbo-charged PFI vehicle
5E+11
1E+12
1.5E+12
2E+12
PN [#]
4E+12
6E+12
8E+12
1E+13
1.2E+13
PN [#]
0
5E+11
0.00000 0.50000 1.00000 1.50000 2.00000PM INDEX
0
2E+12
0.00000 0.50000 1.00000 1.50000 2.00000PM INDEX
25.00
]
• Engine bench, 3000rpm @ full load • Market fuels • Normally-aspirated PFI engine
• PM Index correlated well with PM
R2 0 977410.00
15.00
20.00
conc
entra
tion
[mg/
m3
number and mass.
• The correlation held regardless of engine technology and cycle.
R = 0.9774
0.00
5.00
0 1 2 3 4 5
PM INDEX
Car
bon
g gy y
LEV III Workshop May 18th, 2010 6 Testing Performed at SwRI – Background
• Three test fuels were chosen, based on the range of PM Indices calculated for US k t f l Th F i b k d L V f l t i d f thUS market fuels. The Fairbanks and Las Vegas fuels were retrieved from the market in drum quantities.
EPA Certification Las Vegas Fairbanks
90% 100%
y Indolene
test fuel
test fuel (no PM Index spec)*
(top 5% of US PM Indices)
(bottom 5% of US PM Indices)
60% 70% 80%
Pro
babi
lit test fuel (sample, not
spec)
• The PM Index was calculated for >400 US
30% 40% 50%
umul
ativ
e
The PM Index was calculated for 400 US fuels, based on DHA analyses performed by Southwest Research Institute for Honda.
PM Index of CA market fuels:
0% 10% 20%
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Cu
Los Angeles San Francisco
PM Index of CA market fuels:
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5PM Index More tailpipe particulates (predicted)More tailpipe particulates (predicted)
* Since there is no PM Index specification, certification fuel from different suppliers, or a different batch from the same supplier, will have different characteristics and a different PM Index.
LEV III Workshop May 18th, 2010 7
• PM Index of the test fuels compared to the distribution of PM Indices worldwide.
Testing Performed at SwRI – Background
• Note that the US has the widest range of PM Indices, and also some of the highest values.
EPA Certification* Las Vegas Fairbanks
80%90%
100%
bilit
y
USA
50%60% 70% 80%
ive
Pro
bab
World wideIndia Japan EU
USA
30% 40% 50%
Cum
ulat
USA China
Russia Brazil
0% 10% 20%
5.00.0 1.0 2.0 3.0 4.0PM Index
* Since there is no PM Index specification, certification fuel from different suppliers, or a different batch from the same supplier, will have different characteristics and a different PM Index.
LEV III Workshop May 18th, 2010 8
• Factor of 4 difference in FTP PM mass emissions between Fairbanks
SwRI Testing – PM Mass Results
Factor of 4 difference in FTP PM mass emissions between Fairbanks fuel and Las Vegas fuel.
• Variations in the PM Index of the cert fuel can make the difference between a pass and a fail.between a pass and a fail.
910 LEV II (Current)
67 8 9
mg/
mi) Fairbanks EPA Cert Las Vegas
LEV III (2014) d
34 5 6
Mas
s (m LEV III (2014) proposed
LEV III (2017)
01 2 3
PM LEV III (2017) proposed
• Test vehicle: 2009 2.0L GDI, LEV2 ULEV (not a Honda product)
0 FTP
LEV III Workshop May 18th, 2010 9
• FTP PM number emissions from Las Vegas fuel >3x higher than
SwRI Testing – PM Number Results
FTP PM number emissions from Las Vegas fuel >3x higher than those from Fairbanks fuel.
• Variations in the PM Index of the cert fuel can make the difference between a pass and a fail.between a pass and a fail.
910
12)
Fairbanks EPA Cert Las Vegas
67 8 9
rt./m
i x 1
01
LEV III (2014) d
34 5 6
mbe
r (p
a LEV III (2014) proposed
LEV III (2017)
01 2 3
PM N
um LEV III (2017) proposed
0 FTP
• Test vehicle: 2009 2.0L GDI, LEV2 ULEV (not a Honda product)
LEV III Workshop May 18th, 2010 10 SwRI Testing – Correlation with PM Index
PM Mass
3.0
m)
PM Mass Measurement vs. PM Index
Total PM from filter
R² = 0.9988 2.0
2.5
ons
(mg/
km
Total PM from filter(40 CFR Part 1065)
Solid PM only (from AVL MSS)
R² = 0.9998 1.0
1.5
M E
mis
sio ( )
0.0
0.5
0 1 2 3
FTP
PM
Fairbanks fuel
Las Vegas fuel
EPA Cert fuel
PM Index
( )• Excellent correlation between PM Index and actual (measured) PM mass emissions.
LEV III Workshop May 18th, 2010 11 SwRI Testing – Correlation with PM Index
PM Number
12
12)
PM Number Measurement vs. PM Index
US06
R² = 0.99848
10
rt./k
m x
101
R 0.9984
4
6
ssio
ns (p
ar FTP
• Results shown are for solid PM > 23nm. R² = 0.9795
0
2
PM E
mis (EU PMP-type measurement.)
Fairbanks fuel
Las Vegas fuel
EPA Cert fuel0
0 1 2 3 PN Index
• Good correlation also achieved between PM Index and actual (measured) PM number emissions.
LEV III Workshop May 18th, 2010 12 Summary & Recommendations
• Fuel formulation can have a significant effect upon tailpipe emissions of particulate matter (PM).
• The “PM Index” provides a method by which tailpipe PM emission trends can be predicted, based on fuel composition.
Recommendations:
• Control the Certification Fuel: • The CaRFG3 certification fuel specification should include the
PM Index (or similar).
• Control the Market Fuel:• Control the Market Fuel:• Market fuel specifications (including the CaRFG3 predictive
model) should also include the PM Index. Improvements to market fuel specs will affect real-world PM emissionsmarket fuel specs will affect real-world PM emissions.
14
X9V (GDI PM)AR Report
March 9th, 2010 PM Mass – Cumulative Emissions FTP
Las Vegas fuel
Fairbanks fuel
EPA Cert fuel
15
X9V (GDI PM)AR Report
March 9th, 2010 Solid PM Number – Cumulative Emissions FTP
LEV III proposal (2014) Las Vegas fuel
EPA Cert fuel
Fairbanks fuelFairbanks fuel
LEV III proposal (2017)
• Control of PM emissions throughout Phase 1 of the FTP (not just the cold-start period) is critical.
LEV III Workshop May 18th, 2010 16 Other Test Fuel Properties
FairbanksDistillation Curve Carbon Number DistributionEPA CertLas Vegas
• DHA analysis indicated th t th t j it fthat the vast majority of HCs >C9 were aromatic.
• The T90 point does not predict the fact that the PM emissions
C ti l i d f PM i i Th L V f l
were relatively close for the Fairbanks and EPA Cert fuels.
Chemical Composition • Conventional wisdom for PM emissions: The Las Vegas fuel
would be relatively low (10% ethanol content and moderate aromatic level) and the Fairbanks fuel relatively high (high aromatics). Actual results were the complete opposite.
• It is difficult to judge the PM-producing potential based only on these traditional properties; assumptions are conflictingproperties; assumptions are conflicting.
• Honda’s “PM Index” is a better predictor.