Evaluation of Evaporative Leaks using RSD dI t I li ti and Inventory Implications David Hawkins 1 , Constance Hart 1 , Carl Fulper 1 , David Brzezinski 1 James Warila 1 James Sidebottom 2 James Brzezinski , James Warila , James Sidebottom , James Kemper 2 , Sandeep Kishan 3 , Timothy H. DeFries 3 , 1 U.S. Environmental Protection Agency, Office of Transportation and Air Quality, 2000 Traverwood Drive, Ann Arbor, MI 48105 2 Colorado Department of Public Health and Environment (CDHPE), 4300 Cherry Creek Drive South, Denver, Colorado 80246 h ( ) ld i i 3 Eastern Research Group (ERG), 3508 Far West Blvd, Suite 210, Austin,TX 78731 1
Transcript
Evaluation of Evaporative Leaks using RSD d I t I li tiand Inventory Implications
David Hawkins1, Constance Hart1, Carl Fulper1, David Brzezinski1 James Warila1 James Sidebottom2 James Brzezinski , James Warila , James Sidebottom , James Kemper2, Sandeep Kishan3, Timothy H. DeFries3, 1 U.S. Environmental Protection Agency, Office of Transportation and Air Quality, 2000 Traverwood Drive, Ann Arbor, MI 481052 Colorado Department of Public Health and Environment (CDHPE), 4300 Cherry Creek Drive South, Denver, Colorado 80246
h ( ) l d i i3 Eastern Research Group (ERG), 3508 Far West Blvd, Suite 210, Austin,TX 78731
1
B k dBackgroundMOVES evaporative rates in need of pupdate
New FuelsNew Control Systems
CRC E-77 suite of programs has dd d th d f l b d t addressed the need for lab data on
aging enhanced-evap emission vehicles with different fuels and vehicles with different fuels and implanted leaksMissing piece is how many “high” Missing piece is how many high evap vehicles exist in the fleet
2
P j t G lProject GoalsDevelop cost-effective non-intrusive Develop cost effective, non intrusive, reliable measurement tools for large volume of vehicles in field
RSD as screening methodPortable SHED (PSHED)Identify leak source(s)?Identify leak source(s)?
Better characterize the evaporative emissions inventoryemissions inventory
Fraction of “High-Evap” in LD fleetLiquid vs. Vapor Leaksq pApplication to other RSD datasetsUpdate MOVES 3
Found a Partner with Same Mission
fDiscovered Colorado Department of Public Health and Environment (CDPHE) was already working on (CDPHE) was already working on same issues with RSD’s evap detectiondetectionDeveloped Cooperative Research And Development Agreement And Development Agreement (CRADA) to partner together in project developmentp j p
4
P j t D iProject Design
RSD Screening DevelopmentHow to recruit?
RSD Measurement DevelopmentPretesting at known speeds, leak rates, and exhaust HC emissionsand exhaust HC emissions
Testing Methodology
5
RSD S i D l tRSD Screening Development
Statistical sampling Sampling is biased towards recruiting vehicles identified in higher RSD binsvehicles identified in higher RSD bins
Evaporative Index 23 (EI23)Algorithm developed for recruitmentAlgorithm developed for recruitment“Work in progress”
Handheld HC “Sniffer” CheckPSHED for 15 min hot soak
8
/RSD Testing in I/M Lane
9
PSHED Recovery/Retentiony
Average Recovery 97 6%Average Recovery 97.6%Recovery Std. Dev. 3.3% Average Retention 95.7% R t ti Std D 2 3%Retention Std. Dev. 2.3%
1010
Participation Results at Denver IM Lanes
Lipan 2008 Ken Caryl 2009
L th f T ti 4 k 8 kLength of Testing 4 weeks 8 weeksRSD Data Points (vehicle throughput) 7,011 13,027( g )
Recruited Vehicles 196 558Participants with PSHED evaluation 87 175
Acceptance Rate 44% 31%
11
Field Testing Analysis Parameters
Ultimate goal is to relate RSD scores to evap leaks
ll l f d /Initially classified .3g/15min PSHED as a “high” evap vehicle
Evap standard 2g Hot Soak + DiurnalEvap standard = 2g = Hot Soak + DiurnalAssume 20% (.4g) attributed to Hot SoakAssume 75% of that (.3g) is in first 15 min
RSD scores stratified into 8 bins to describe emissions spread
12
PSHED vs. EI23 for the Stratified Set
EI23 is one algorithm of many in development process; and
is particularly weak at is particularly weak at detecting the influence of high
exhaust HC
13
K I ith RSDKnown Issues with RSD
f fConfounding factorsNoise (variability)E h HCExhaust HCVehicle SpeedEnvironmentEnvironment
14
PSHED vs E-77 (Lab Hot Soak)
10.000
1.0005min
) LAB DATA
0.100
0 1 2 3 4 5 6 7 8 9 10
ED
(g/1
5
MeasurementRSD
0.010
PS
HE E77NoLeak
Gas Cap LeakTop of Tank/Canister Inlet Leak
0.001
Standard = .3 g/15minMedian
15Note: Vehicle condition not reconciled with data shown (ie not checked
against surveys on recent refueling, oil changes, fluid fills, or source of leaks).
Measurement RSD Bin (1-7)
E-77-3 (PSHED) vs E-77 (Hot Soak) 1996 & later
100.000M RSDPSHED and RSD agree that
PSHED vs E-77 (Hot Soak) 1996 & later
10.000
MeasurementRSDE77NoLeakE77LeakE77cLeak.3 g/15minM di
PSHED and RSD agree that vehicle is a high (>.3g) evaporative emitter.
10.000
1.000
D (g
/15m
in) Median
1.0000 1 2 3 4 5 6 7 8 9 1015
min
) LAB DATA
0.100
0 1 2 3 4 5 6 7 8 9 10
ated
PSH
ED
0.100
0 1 2 3 4 5 6 7 8 9 10
ED
(g/1
Measurement RSD Ken Caryl
E77NoLeak
0.010
Estim
a
0.010
PS
H o ea
Gas Cap LeakTop of Tank/Canister Inlet Leak
Standard = 3 g/15min
0.001 LAB DATA0.001
M t RSD Bi (1 7)
Standard .3 g/15min
Median
16
Measurement RSD Bin (1-7)
Note: Vehicle condition not reconciled with data shown (ie not checked against surveys on recent refueling, oil changes, fluid fills, or source of leaks).
Measurement RSD Bin (1-7)
Pilot Analysis & Comparison to MOVESDraftMOVES
Rates for leaking/non-leaking
Historical CRC diurnal testing% f l ki / l ki i fl t% of leaking/non-leaking in fleet
Historical IM database OBD failure codes
Preliminary comparison of E-77 to DraftMOVESRates for leaking/non-leaking
E 77 2 programs diurnal testingE-77-2 programs diurnal testing% of leaking/non-leaking in fleet
Preliminary analysis of Colorado Program
17
y y gReconstructed population from PSHED participantsReal population of fleet driving by RSD (EI23)
How Vapor Leaks are modeled in MOVESTVV)
Average “Leaking” Vehicle
Ven
ted (
T
MOVES input is a weighted
k Vap
or
V MOVES input is a weighted average based on Leak/NoLeak %
Tan
Average “Non-Leaking” Vehicle
18
Tank Vapor Generated (TVG)
Function of Fuel RVP and Temperature
Probability of "HighEvap" Summer2009 Ken Caryl IMStation y
RSD to PSHED model
70
80of
se
t EI23 basis (RSD)
50
60
70
ract
ion
m D
atas
EI23 basisPSHED basis
( )
40
50
Evap
Fr
Ran
dom
(%) Discrepancy will improve with
development of EI and/or higher“High Evap” definition
20
30
Hig
h-E
Car
yl R
0
10
0.1 1 10.3g used in
19
0.1 1 10
Definition of High PSHED (grams)analysis
First 12-hour Heating Cycle of 3-Day Diurnal(10psi fuels removed; consistent gross canister breakthrough)25
MOVES 99-03 Age 8-9 Weighted Emission RateE77 Weighted Emission Rate (PSHED)E77 No Leak (19 tests)
15
20
nted
(g)
g g
E77 Weighted Emission Rate (PSHED)E77 No Leak (19 Tests)
y 3.2007x15
apor
Ven
ted
This linear best-fit represents ALL E77 leaks from model years 1996 and later.
10
15
Vapo
r Ven
5
10
Tank
V
RSD5
10
Tank
V
RSD
0
5 RSD
MOVES
PSHED
0
5 RSD
MOVESPSHED
21
0 1 2 3 4 5 6
Tank Vapor Generated (g/gal)
0 0 1 2 3 4 5 6
Tank Vapor Generated (g/gal)
C l d S 2010Colorado Summer 2010
fDevelopment of z-score indexVehicles have multiple RSD readingsC RSD k hi h d !Can use RSDs taken at high speeds!
But, lower speed RSDs have more influence
Vehicle TestingPSHED at tech centerLSHED/6-hour Diurnal at Aurora Lab
Measure real world leak behavior in diurnal!
22
MOVES SMOVES Summary MOVES is being developed along with MOVES is being developed along with RSD methodologyMOVES inputsMOVES inputs
Curve for “non-leaking” vehiclesCalculated from Diurnal test data
Curve for “leaking” vehiclesCalculated from Diurnal test data
Fraction of “leaking” vehiclesWill be used to weigh both curves togetherAs RSD indices improve this fraction can be
23
As RSD indices improve, this fraction can be measured with RSD data
How Vapor Leaks are modeled in MOVESTVV)
Average “Leaking” Vehicle
Ven
ted (
T
MOVES input is a weighted
k Vap
or
V MOVES input is a weighted average based on Leak/NoLeak %
Tan
Average “Non-Leaking” Vehicle
24
Tank Vapor Generated (TVG)
Function of Fuel RVP and Temperature
N t StNext StepsPredictive Analysis Tools, Further y ,Analysis:
Use RSD data sets to predict fraction of high levap in a metropolitan area
Do candidate RSD data sets exist within the limitations of the measurement capabilities?limitations of the measurement capabilities?How do the PSHED results relate to OBD evap codes and can we make a link between RSD and OBD from this?
CDPHE Summer 2010 High Evap Testing
25
Develop z-score RSD indexAnalyze PSHED/LSHED data for use in MOVES
