MEMORANDUM

Subject:	Evaporative Emissions Modeling in MOVES: Technical Document for
RFS NPRM Docket

From:		Prashanth Gururaja

		Assessment and Standards Division

To:		Docket EPA-HQ-OAR-2005-0161



Evaporative Emissions Modeling in MOVES

Technical Document

Prashanth Gururaja

May 11, 2009

Table of Contents

  TOC \o "1-4" \h \z \u    HYPERLINK \l "_Toc229836105"  1 Background	 
PAGEREF _Toc229836105 \h  4  

  HYPERLINK \l "_Toc229836106"  2 Data Sources	  PAGEREF _Toc229836106
\h  4  

  HYPERLINK \l "_Toc229836107"  3 Design and Analysis	  PAGEREF
_Toc229836107 \h  4  

  HYPERLINK \l "_Toc229836108"  3.1 Fuel Tank Temperature Generator	 
PAGEREF _Toc229836108 \h  5  

  HYPERLINK \l "_Toc229836109"  3.1.1 Input parameters	  PAGEREF
_Toc229836109 \h  5  

  HYPERLINK \l "_Toc229836110"  3.1.2 General steps	  PAGEREF
_Toc229836110 \h  5  

  HYPERLINK \l "_Toc229836111"  3.1.2.1 Calculating soak temperatures
(as a function of ambient temperature)	  PAGEREF _Toc229836111 \h  5  

  HYPERLINK \l "_Toc229836112"  3.1.2.2 Calculating fuel tank
temperatures during operation	  PAGEREF _Toc229836112 \h  6  

  HYPERLINK \l "_Toc229836113"  3.2	Permeation	  PAGEREF _Toc229836113
\h  7  

  HYPERLINK \l "_Toc229836114"  3.2.1 Base Rates	  PAGEREF _Toc229836114
\h  7  

  HYPERLINK \l "_Toc229836115"  3.2.2  Temperature adjustment	  PAGEREF
_Toc229836115 \h  8  

  HYPERLINK \l "_Toc229836116"  3.2.3 Fuel Adjustment	  PAGEREF
_Toc229836116 \h  8  

  HYPERLINK \l "_Toc229836117"  3.3 Tank Vapor Venting	  PAGEREF
_Toc229836117 \h  9  

  HYPERLINK \l "_Toc229836118"  3.3.1 Cold Soak	  PAGEREF _Toc229836118
\h  9  

  HYPERLINK \l "_Toc229836119"  3.3.2 Hot Soak	  PAGEREF _Toc229836119
\h  12  

  HYPERLINK \l "_Toc229836120"  3.3.3 Inspection/Maintenance (I/M)
Program effects	  PAGEREF _Toc229836120 \h  13  

  HYPERLINK \l "_Toc229836121"  3.4 Liquid Leaks	  PAGEREF _Toc229836121
\h  17  

 

1 Background

Evaporative emissions account for a significant portion of the total
gaseous hydrocarbon inventory.  Its processes are unique and require a
unique modeling approach.  For a long time, evaporative emissions were
thought of being quantifiable in three distinct modes and subsequent
test procedures: running loss (during vehicle operation),
diurnal/resting loss (stabilized parked emissions), and hot soak (parked
emissions immediately after vehicle operation).  However, it has become
evident that different factors, such as ambient temperature and fuel
type for example, affect evaporative emissions more noticeably in the
emissions processes, rather than the aforementioned three modes. 
Evaporative emissions can be broken up into three main processes:

Permeation – the migration of hydrocarbons through elastomers in a
vehicle’s fuel system

Tank Vapor Venting – expulsion into the atmosphere of fuel vapor
generated from evaporation of fuel in the fuel system

Liquid Leaks – fuel, in liquid form, leaking from the fuel tank or
fuel system, which then evaporates into the atmosphere

These three processes occur and can be addressed in each mode. 
Therefore, we can measure and/or calculate permeation, tank vapor
venting, and liquid leaks in each of the three testing regimes prevalent
in major evaporative emissions test programs.  Then, we can relate the
emissions from each of the processes to different factors that occur
independently of modes.  This makes for easier, more accurate modeling
of scenarios that do not perfectly replicate the test procedures.

The factors that affect permeation, vapor venting, and leaks that we
considered were:

Ambient temperature

Fuel tank temperature

Model year

Age

Vehicle class

Fuel (ethanol %, RVP)

Failure modes

Presence of I/M

2 Data Sources

CRC E-9 – Measurement of Diurnal Emissions from In-Use Vehicles

CRC E-35 – Measurement of Running Loss Emissions in In-Use Vehicles

CRC E-41 – Evaporative Emissions from Late-Model In-Use Vehicles

CRC E-65 – Fuel Permeation from Automotive Systems

CRC E-65-3 – Fuel Permeation from Automotive Systems: E0, E6, E10, and
E85

BAR Gas Cap Study

API Gas Cap Study

EPA Compliance Testing

Appendix A has a summary of most of the test programs mentioned above.

3 Design and Analysis

We found fuel tank temperature to be the driver of the two transient
emissions processes, permeation and vapor venting.  Determining fuel
tank temperature was critical in predicting emissions in each of these
Operating Modes.  Fuel tank temperature is dependent on the daily
ambient temperature profile, times that the vehicle is operating, and
the model year of the vehicle.  Then, we can use the calculated fuel
tank temperature profile to calculate permeation and vapor venting. 
Other factors were included as needed.  Liquid leaks were not dependent
on temperature.  This section will first describe the Fuel Tank
Temperature Generator, and then explain how we used the fuel tank
temperature to determine emission rates for each of the emissions
processes.

3.1 Fuel Tank Temperature Generator

This section explains how to generate fuel tank temperature through time
for a given day’s ambient temperature profile and a vehicle’s trip
times.  Generating fuel tank temperature allows for the calculation of
permeation and vapor venting, two major fuel-related evaporative
emissions processes.  As a result, this algorithm is instrumental in
modeling evaporative emissions in MOVES.

3.1.1 Input parameters

Hourly ambient temperature profile

Key on and key off times 

HourDayID (day and hour) of first KeyON

Vehicle Type (LDT/LDV)

Pre-enhanced or enhanced evaporative emissions control system

MOVES defines these input parameters via the sampleVehicleTrip and
zoneMonthHour tables and the sourceBinID variable.

3.1.2 General steps

Input parameters must be defined.

Fuel tank temperature is computed up to the start of the first trip,
assuming that the vehicle has been parked for a long time (overnight). 
This is done through the block diagram in   REF _Ref229825273 \h  \*
MERGEFORMAT  

Figure 1  below, which represents the differential equation in equation
1.  All soaks (hot and cold) are calculated using this portion of the
algorithm.

Next, for each trip, the fuel tank temperature is computed for the
operation period and the corresponding soak period after the key off for
that trip.  It computes fuel tank temperature until the start of the
next trip (next key on), at which point this step is repeated, or until
the end of the day.  The fuel tank temperature during operation is
calculated using equations (3) and (4).  The fuel tank temperature
during hot soak is calculated as for cold soak (equation 2b), but with
the initial temperature (Ti) changed to the temperature at the end of
each trip, and the time interval modified to accommodate the key on/off
times.

3.1.2.1 Calculating soak temperatures (as a function of ambient
temperature)

The following equation was used to model tank temperature as a function
of ambient temperature.  This was used for hot and cold soaks.

 ,                      (1)     

where Ttank is the fuel tank temperature, Tair is the ambient
temperature, and k is a constant proportionality factor (k = 1.4).  The
value of k was verified by trial and error against EPA compliance data. 
There was no distinction made between hot soak and cold soak
calculations.  We assumed that during either soak, the only factor
affecting fuel tank temperature was the ambient temperature profile and
the fuel tank temperature at the start of the soak.  The block diagram
below simplifies the equation into several mathematical steps, which are
explained below.  

Figure   SEQ Figure \* ARABIC  1  – Simulink® block diagram of the
relation between ambient temperature and fuel tank temperature

The time periods for which this part of the algorithm is used depends on
the key on and key off times.  Since this equation can be used only for
cold soaks and hot soaks (all parked conditions), it applies for the
following time intervals only:

from the start of the day to the first trip,

from all key off to key on times, and

from the last key off to the end of the day.

Mathematical steps

At time t0 = 0 or KeyOFF (start of soak), Ttank = Ti.  This value will
either be the ambient temperature (at the very start of the model) or
the fuel tank temperature at the end of a trip.

Then, for all t > 0 or KeyOFF, the next tank temperature is calculated
in this manner:

      	(2a)	or

 		(2b)

(Tair – Ttank) is a function of time.  Since analytical integration is
too complicated (the input ambient temperature data is in a table),
numerical integration should be used to perform this step.  The method
of numerical integration varies based on the accuracy desired.  The
above method represents the Euler method, one of the simplest methods of
integration.  The less accurate the method, the smaller the time step
Δt should be, to improve the solution.  MOVES uses a Δt of 15 minutes,
which is accurate enough for our modeling purposes.

3.1.2.2 Calculating fuel tank temperatures during operation

Operation periods (trips) are relatively short compared to the length of
the day or modeling period.  Therefore, even though the fuel tank
temperature profile during operation is not exactly linear, assuming a
linear increase in temperature makes calculations easier without
compromising accuracy.  However, the increase in temperature ΔTtank
depends on the temperature at the start of operation.  It also depends
on vehicle type.  The convention used in this algorithm is that ΔTtank
applies over a 4300 second period, which is the length of the running
loss test performed by manufacturers for certification.  To find
ΔTtank, we must first find ΔTtank95, the average increase in tank
temperature at a standard 4300 second @ 95F running loss test.

If the vehicle is evap-enhanced, then ΔTtank95 = 24F  

If the vehicle is pre-enhanced, the vehicle type affects ΔTtank95.  
NOTEREF _Ref229571412 \h  \* MERGEFORMAT  2 

If LDV, then ΔTtank95 = 35F.

If LDT, then ΔTtank95 = 29F.

We can use these values for ΔTtank95 for 95F to calculate the ΔTtank
for other starting fuel tank temperatures (other trips) using the
following equation:

   (3)  

Since this gives us the increase in tank temperature, we can create a
simple linear function that models fuel tank temperature for each trip.

     (4)  NOTEREF _Ref229569972 \h  \* MERGEFORMAT  9   

The 4300/3600 appears, as the running loss test done by manufacturers is
4300 seconds long, and we convert that to hours maintain consistency in
the algorithm.

Assumptions:

The first trip is assumed to start halfway into the hour stated in the
first trip’s HourDayID.  

We assumed the effect of the ambient temperature or change in ambient
temperature during a trip was negligible compared to the effect of
operation.  

The KeyON tank temperatures will be known by way of the calculations of
the tank temperatures from the previous soak.

Permeation

3.2.1 Base Rates

We first determined base rates for permeation.  We define these rates as
the non-leak hydrocarbon gram-per-hour emission rate during the last six
hours of a 72-96°F diurnal test (also known as resting loss).  In these
six hours, the emissions rate and the ambient and fuel tank temperatures
are relatively stable or constant, leading us to believe that the
constant permeation process is the only emissions process occurring.  We
stratified these rates by model year group and age group.  The base
permeation rates are in   REF _Ref229458734 \h  \* MERGEFORMAT  Table 1
.  Separate inputs were created from model years 1996-1998 to account
for the 20/40/90% phase-in of enhanced evaporative emissions standards.

Table   SEQ Table \* ARABIC  1  – Base permeation rates at 72 F

Model year group	Age group	Base permeation rate [g/hr]

1971-1977	10-14	0.192

	15-19	0.229

	20+	0.311

1978-1995

	0-3	0.0554

	4-5	0.0554

	6-7	0.0913

	8-9	0.0913

	10-14	0.124

	15-19	0.148

	20+	0.201

1996	0-3	0.046

	4-5	0.046

	6-7	0.075

	8-9	0.075

	10-14	0.101

	15-19	0.120

	20+	0.163

1997	0-3	0.037

	4-5	0.037

	6-7	0.059

	8-9	0.059

	10-14	0.079

	15-19	0.093

	20+	0.125

1998	0-3	0.015

	4-5	0.015

	6-7	0.018

	8-9	0.018

	10-14	0.022

	15-19	0.024

	20+	0.029

1999-2003

	0-3	0.0102

	4-5	0.0102

	6-7	0.0102

	8-9	0.0102

	10-14	0.0102

	15-19	0.0102

	20+	0.0102

3.2.2  Temperature adjustment

Use following equation for temperature-adjusted permeation rate for each
hour not in the last six hours of a diurnal:

       (5)  NOTEREF _Ref229735672 \h  \* MERGEFORMAT  6 

where Pbase is the base permeation rate calculated by averaging the last
six hours of emissions, Ttank is the tank temperature, and Tbase is the
base temperature for a given temperature cycle (e.g. 72 for a 72-86-72
diurnal test).

This is derived from the E-65 permeation study  NOTEREF _Ref229735672 \h
 \* MERGEFORMAT  6  which found that permeation rate doubles for every
18 degrees F.  Normalizing the base permeation rates about 72 F
generates the following equation:

        	(6)

3.2.3 Fuel Adjustment

E10 affects evaporative emissions from gasoline vehicles due to the
increased volatility of E10 blends, the increased permeation of fuel
vapors through tanks and hoses, and the increased vapor emissions due to
the lower molecular weight of E10.  Each of these effects were modeled
using the draft MOVES model, which separates permeation emissions from
vapor venting emissions to allow better accounting for these different
processes.   

Permeation effects were developed from CRC’s E-65 and E-65-3 programs,
which measured evaporative emissions from ten fuel systems that were
removed from the vehicles on E0, E5.7, and E10 fuels; fuel systems were
removed to ensure that all evaporative emissions measured were from
permeation of the fuel through the different components of the fuel
system.  For this analysis, we separated the evaporative enhanced
vehicles from the pre-enhanced vehicles.  Enhanced evaporative vehicles
began being phased in from 1996 through 1999 and needed to meet a 2.0 g
standard over a 24-hour diurnal test.  Pre-enhanced vehicles needed to
meet 2.0 g over a 1-hour simulated diurnal.  We estimated the ethanol
effect by calculating the percent increase in average emissions over the
65-105-65 deg F diurnal cycle from each of the two groups of vehicles. 
To determine the effect of ethanol blend, we first averaged the E5.7 and
E10 results (where both fuels were tested) for each vehicle to obtain
its mean ethanol permeation rate.  We then averaged each vehicle’s
mean permeation rate on E0.  The percent difference between the ethanol
rate and the E0 rate was input into MOVES as the fuel adjustment.  Due
to the phase in from 1996 to 1999 (20/40/90/100 %), the two fuel
adjustments must be properly weighted for those model years.  The fuel
adjustment in MOVES is based on a variable called fuelModelYearID.   
REF _Ref229456510 \h  \* MERGEFORMAT  Table 2  shows the fuel
adjustments used for E5 through E10 for the fuelModelYearID’s used in
MOVES.

Table   SEQ Table \* ARABIC  2  – Increase in emissions due to ethanol
levels of 5 to 10% compared to E0 (gasoline)

Model years

(via fuelModelYearID in MOVES)	Percent increase due to ethanol (E5
through E10)

1995 and earlier	37.3

1996	69.4

1997-2000	175

2001 and later	198



We plan to revisit our permeation emissions estimates with the release
CRC E-77 and E-77-2b studies.  

3.3 Tank Vapor Venting

The following explains how vapor venting rates were calculated for each
of the operating modes.  For cold soak, MOVES first finds the amount of
vapor generated in the tank as a function of fuel tank temperature and
RVP.  Then, it determines how much of this vapor is released into the
atmosphere based on several criteria, such as model year and fill pipe
pressure test result.  The temperatures will have been generated by the
fuel tank temperature generator, and the RVP will have been generated by
the MOVES tank fuel generator.  This cannot apply for when vapor is not
generated (when fuel tank temperature is not increasing), such as during
a hot soak, but is released.  For these situations, we have aggregated
TVV rates after subtracting out permeation and leaks from the test
results.  Also, due to the availability of test data for running loss
and the short length of trips, we determined TVV rates during operation
the same way we did for hot soak.  

3.3.1 Cold Soak

We calculated fuel tank temperature at each hour using the fuel tank
temperature algorithm.  

After calculating base permeation rate for each vehicle (average of last
six hours of HC evaporative emissions), we used the fuel tank
temperature adjustment with the temperatures calculated in step 1 to
calculate the permeation for each hour.  The fuel tank temperature is
determined through the fuel tank temperature algorithm for MOVES
diurnals.

We filtered/reduced data set such that each test met the following
requirements:

Non-leakers

“As received” vehicles (no retests)

Hours where tank temperature increased from previous hour

Pressure test result must be pass, fail, or blank only (no dashes,
slashes, “I”, etc.)

We subtracted permeation from HC for each hour to get tank vapor venting
(TVV) rate

We summed TVV from beginning of diurnal to each hour to get Cumulative
TVV.

We then determined Tank Vapor Generated (TVG) from hour 1 to hour x for
each hour that the fuel tank temperature is rising.

  [grams/gal]  (7)

where A, B, and C are constants listed below in   REF _Ref229458884 \h 
Table 3 , and Tx is the temperature at hour x.

Table   SEQ Table \* ARABIC  3  – TVG constants for equation 7.  
NOTEREF _Ref229822816 \h  \* MERGEFORMAT  10 

	Gasoline	E10

Constant	Sea Level	Denver alt.	Sea Level	Denver alt.

A	0.00817	0.00518	0.00875	0.00665

B	0.2357	0.2649	0.2056	0.2228

C	0.0409	0.0461	0.0430	0.0474



TVG is the amount of vapor generated in the tank.  We will establish a
relationship between Cumulative TVV and TVG for inputs into MOVES.

We constructed quadratic curves (zero intercept) of CumTVV vs. TVG,
stratifying by model year group, age group, and pressure test result.  

                   (8)

Having the zero-intercept ensures that the (0, 0) is a point on the
quadratic curve.  In other words, it implies that at 0 TVG, there is no
tank vapor venting, which is an accurate physical assumption.

Curves were generated for model year groups 1971-1977 (ages 15+),
1978-1995 (ages 0-19), and 1996-2003 (ages 0-9).  We also stratified by
pressure test result.  In failing vehicles, more of the vapor that is
generated in the fuel system will be vented than in passing vehicles,
where the evaporative emission controls should be functioning properly. 
The remainder of the coefficients was found by extrapolation or
previously determined relationships.  The coefficients of variation
(CVs) were calculated by dividing the standard error of the sample
(calculated by SPSS) by the mean for each coefficient in the quadratic
equation.

After failure frequencies (F) were generated from pressure, gas cap, and
OBD test results from the Phoenix I/M program (see section   REF
_Ref229733363 \h  \* MERGEFORMAT  3.3.3 Inspection/Maintenance (I/M)
Program effects ), aggregate coefficients were calculated:

 ,             (9)   

where x = 1 or 2, corresponding to quadratic equation 8.

Since the aggregate coefficients were determined using the failure
rates, which are essentially weighting factors, the standard errors of
the aggregate coefficients are calculated:

        (10)

	As a result, the CV’s for the aggregate coefficients are calculated:

    (11)

  REF _Ref229567195 \h  \* MERGEFORMAT  Table 4  shows the I/M
coefficients resulting from the analysis.  Ratios between age groups
were used to extrapolate for the 10-14 age group in the 1971-1977 model
year groups, and older age groups where data did not exist were forced
to have the same coefficients as their preceding age groups.  The
passing coefficients for the 2004 and later model year group were
reduced by 32% from the 1999-2003 model year group, which reflects a
reduction in the evaporative emissions standard from enhanced-evap to
Tier 2/LEV II.  Separate model year groups are created for 1996 through
1998 due to the phasing of enhanced evaporative standards.  These three
groups are only different weightings of the 1978-1995 and 1999-2003
model year groups based on the 20/40/90% phase-in for 1996/1997/1998. 
Similarly, though not shown, is a table that was developed for non-I/M
vehicles using non-I/M failure frequencies calculated from the Phoenix
I/M data set.

Table   SEQ Table \* ARABIC  4  – I/M coefficients for equation 8. 
The aggregate columns are the inputs in the MOVES model for I/M
coefficients in the cumTVVCoeffs table.

model year group	age group	a1	a2



pass	fail	aggregate	pass	fail	aggregate

1971-1977	10-14	1.227	11.314	1.941	2.175	0.402	2.049

	15-19	5.406	9.254	5.835	2.331	3.117	2.419

	20+	5.406	9.254	6.127	2.331	3.117	2.479

1978-1995	0-3	1.578	3.073	1.589	0.440	1.338	0.446

	4-5	1.578	3.073	1.604	0.440	1.338	0.455

	6-7	1.578	3.073	1.610	0.440	1.338	0.459

	8-9	1.578	3.073	1.623	0.440	1.338	0.466

	10-14	0.849	11.314	1.283	2.095	0.402	2.025

	15-19	3.743	9.254	4.120	2.246	3.117	2.305

	20+	3.743	9.254	4.376	2.246	3.117	2.346

1996	0-3	1.339	3.073	1.354	0.344	1.338	0.352

	4-5	1.339	3.073	1.362	0.344	1.338	0.357

	6-7	1.339	3.073	1.376	0.344	1.338	0.365

	8-9	1.339	3.073	1.392	0.344	1.338	0.374

	10-14	0.756	9.666	1.124	1.668	0.589	1.624

	15-19	3.071	8.017	3.399	1.789	2.762	1.853

	20+	3.071	8.017	3.530	1.789	2.762	1.879

1997	0-3	1.100	3.073	1.120	0.248	1.338	0.259

	4-5	1.100	3.073	1.129	0.248	1.338	0.264

	6-7	1.100	3.073	1.146	0.248	1.338	0.273

	8-9	1.100	3.073	1.163	0.248	1.338	0.283

	10-14	0.663	8.018	0.976	1.241	0.776	1.222

	15-19	2.399	6.781	2.686	1.332	2.406	1.402

	20+	2.399	6.781	2.791	1.332	2.406	1.428

1998	0-3	0.502	3.073	0.538	0.009	1.338	0.027

	4-5	0.502	3.073	0.553	0.009	1.338	0.035

	6-7	0.502	3.073	0.575	0.009	1.338	0.046

	8-9	0.502	3.073	0.596	0.009	1.338	0.057

	10-14	0.429	3.897	0.589	0.174	1.244	0.223

	15-19	0.719	3.691	0.907	0.189	1.516	0.273

	20+	0.719	3.691	0.959	0.189	1.516	0.296

1999-2003	0-3	0.383	3.073	0.422	-0.039	1.338	-0.019

	4-5	0.383	3.073	0.438	-0.039	1.338	-0.011

	6-7	0.383	3.073	0.461	-0.039	1.338	0.001

	8-9	0.383	3.073	0.483	-0.039	1.338	0.012

	10-14	0.383	3.073	0.508	-0.039	1.338	0.025

	15-19	0.383	3.073	0.552	-0.039	1.338	0.048

	20+	0.383	3.073	0.595	-0.039	1.338	0.070

2004 and later	0-3	0.124	3.073	0.151	-0.013	1.338	-0.001

	4-5	0.124	3.073	0.161	-0.013	1.338	0.004

	6-7	0.124	3.073	0.175	-0.013	1.338	0.010

	8-9	0.124	3.073	0.187	-0.013	1.338	0.016

	10-14	0.124	3.073	0.203	-0.013	1.338	0.023

	15-19	0.124	3.073	0.229	-0.013	1.338	0.035

	20+	0.124	3.073	0.255	-0.013	1.338	0.047



3.3.2 Hot Soak

First we found the temperature at the start of the soak.  This is done
by adding on the temperature increase experienced during an LA-4 running
loss test cycle (1372 seconds), since the vehicle is put through this
test before entering the soak chamber.  These temperature rises depend
on the fuel tank temperature at the start of the LA-4 test (ambient). 
To calculate hot soak start temperature Tstart, see section   REF
_Ref229733262 \h  \* MERGEFORMAT  3.1.2.2 Calculating fuel tank
temperatures during operation .

We then found the average temperature in that hour:

      (12)

This is derived from the average value of a function over an interval
(in this case, between 0 and 1 hour after the start of the hot soak). 
As stated in the Fuel Temperature algorithm, k = 1.4.

We used this average temperature to determine the average permeation
rate during the hot soak via the permeation temperature adjustment using
the 72F base permeation rates determine by model year group and age
group.

We filtered/reduced the data set such that each test met the following
requirements:

Non-leakers (emissions less than 10.0 grams; taken from M6.EVP.009_2.4;
Since hot soak emissions are measured after one hour, the total
emissions is “equal” to its g/hr rate)

“As received” vehicles (no retests)

Pressure test result must be pass, fail, or blank only (no dashes,
slashes, “I”, etc.)

We subtracted permeation from HC for each hour to get tank vapor venting
(TVV) rate

We averaged TVV rates by model year group, age group, and pressure test
result, shown in

Table   SEQ Table \* ARABIC  5  – Average hot soak tank vapor venting
rates in g/hr by model year group, age group, and pressure test result.

Model year group	Age group	Pressure test result	TVV rate 

1971-1977	20+	F	6.17

	20+	P	2

1978-1995

	0-5	Both	1.25

	0-5	P	0.56

	0-9	F	2.37

	6-9	Both	1.75

	6-9	P	1.38

	10-14	Both	5.13

	10-14	F	3.41

	10-14	P	1.76

	15-19	F	4.51

	15-19	P	2.99

1996-2003	all	Both	0.1073



Like with cold soak, aggregate rates were found using failure rates
involving pressure, gas cap, and OBD tests for non-I/M and I/M.    REF
_Ref229824606 \h  \* MERGEFORMAT  Table 6  below does not reflect the
most updated I/M analysis explained in section   REF _Ref229733363 \h 
\* MERGEFORMAT  3.3.3 Inspection/Maintenance (I/M) Program effects 
(unlike the cold soak coefficients in   REF _Ref229567195 \h  \*
MERGEFORMAT  Table 4 ) or the enhanced evaporative phase-in, so these
numbers will be updated for the final version of MOVES.

Table   SEQ Table \* ARABIC  6  – Example of non-I/M hot soak tank
vapor venting rates

 Model year group	Age group	TVV rate [g/hr]

1971-1977

	10-14	3.099

	15-19	5.149

	20+	5.455

1978-1995

	0-3	0.627

	4-5	0.627

	6-7	1.451

	8-9	1.471

	10-14	2.082

	15-19	3.492

	20+	3.817

1996-2003

	0-3	0.124

	4-5	0.124

	6-7	0.150

	8-9	0.168

	10-14	0.250

	15-19	0.383

	20+	0.611

2004 and later	0-3	0.060

	4-5	0.060

	6-7	0.086

	8-9	0.105

	10-14	0.187

	15-19	0.323

	20+	0.553

3.3.3 Inspection/Maintenance (I/M) Program effects

Our assumption in MOVES is that tank vapor venting is the only
evaporative process where the effects of I/M are realized.  The types of
evaporative tests performed in I/M programs (gas cap test, fill pipe
pressure test, OBD scans) do not affect permeation or liquid leaks.  

In order to develop I/M and non-I/M tank vapor venting rates, we used
available data from I/M programs to determine the failure frequencies of
evaporative control systems.  These frequencies were then used to
combine the cumulative tank vapor venting coefficients for failing
vehicles and those for passing vehicles (determined from the TVV
analysis).  Details of each of the four programs are in   REF
_Ref229458604 \h  \* MERGEFORMAT  Table 7  below.

Table   SEQ Table \* ARABIC  7 - Description of evaporative
characteristics of I/M programs available for analysis 

	Gas cap test	OBD	Pressure test	Frequency	Network	Calendar years

Colorado	Y	Advisory only	N	Biennial	Hybrid	2003-2006

N. Carolina	N	Y	N	Annual	Decentralized	2002-2006

Phoenix	Y	Y	Y	Biennial	Centralized	2002-2006

Tucson	Y	Y	N	Annual	Centralized	2002-2006



Since the Phoenix program contained the most extensive amount of data,
we used it to develop reference I/M evaporative failure frequency.  The
Tucson, Colorado, and North Carolina data were used to adjust the
Phoenix numbers for differences in I/M programs.  

The Phoenix evaporative I/M program used gas cap tests on all vehicles,
OBD scans on OBD-equipped vehicles, and fill pipe pressure tests on
pre-OBD vehicles.  The OBD codes used to determine evaporative failures
were P0440, P0442, P0445, P0446, and P0447 for all vehicle makes and
additionally P1456 and P1457 for Honda and Acura vehicles.  Vehicles
that had one or more of these faults were flagged as failing vehicles,
analogous to pre-OBD vehicles that failed the pressure test.  Very few
vehicles failed both the gas cap test and the pressure/OBD test. 
Therefore, our total number of failures was the sum of gas cap and
pressure/OBD failures.  

To determine failure frequencies for I/M areas, from the Phoenix data,
we looked at the initial and final results for each vehicle in a given
I/M cycle.  For passing vehicles, the initial test and the final tests
were one and the same.  We averaged the initial and final failure
frequencies (weighted equally) to calculate an overall I/M failure
frequency by model year group and age group.  Using the initial failure
frequencies alone would neglect the effect of repair that most failing
vehicles would be required to undergo, and using the final failure
frequencies alone would neglect the existence of the failing vehicles
driving around in the fleet in the first place.  To determine non-I/M
failure frequencies, we restricted our sample in the Phoenix data to
those vehicles with license plates from states that do not have an I/M
program anywhere.    REF _Ref229483467 \h  \* MERGEFORMAT  Figure 2 
gives an example of how failure frequencies increase with age.  Shown
are frequencies for model years 1978-1995, where data was extrapolated
for the youngest age groups.

Figure   SEQ Figure \* ARABIC  2  – Evaporative failure frequencies
for I/M and non-I/M vehicles in the Phoenix area.  This figure shows
model years 1978 to 1995.

The Tucson data was used to determine the effect of program frequency
(annual vs. biennial).  For OBD vehicles, Tucson performs gas cap and
OBD tests annually, while Phoenix performs them biennially.  Therefore,
we were able to develop failure frequencies for annual programs by
analyzing the Tucson data.  We applied the ratio between Tucson and
Phoenix to determine the failure frequencies for where we did not have
data (e.g. pre-OBD vehicles).

 

The North Carolina data was used to determine non-I/M failure
frequencies for OBD tests.  In North Carolina, expansion of I/M program
has led to counties where many vehicles were tested under the I/M
program for the first time.  Vehicles were flagged as non-I/M tests if
they were tested:

before the official start of the I/M program,

in a new I/M county and were registered in that same county, or

in a new I/M county and were registered in a non-I/M county or a county
that did not start I/M within the last year.

We compared those failure frequencies to those for vehicles tested in
older I/M areas, where vehicles were previously tested.  From the North
Carolina data, the average ratio of non-I/M to I/M OBD failure
frequencies is 1.6.  This ratio was then applied to the Phoenix OBD and
pressure test failure frequencies to determine non-I/M failure
frequencies.  

The Colorado data was used to determine non-I/M failure frequencies for
gas cap tests.  In Colorado, the I/M data comes mostly from the Denver
and Boulder metropolitan areas.  Many residents are new to this area,
with many having moved in from non-I/M areas in Colorado or non-I/M
states.  Vehicles were flagged as non-I/M tests if their registration
state was a 100% non-I/M state, or if the registration county was a
non-I/M county in Colorado.  We compared the failure rates of the
flagged vehicles to those of the full tested fleet.  The ratio of these
two frequencies was then applied to the Phoenix gas cap failure
frequencies to determine non-I/M failure frequencies.  Colorado OBD data
was not used, since OBD in Colorado is only advisory, and does not pass
or fail a vehicle.

From the Colorado data, the average ratio of non-I/M to I/M gas cap
failure frequencies is 2.2.  This ratio was then applied to the gas cap
failure frequencies to determine non-I/M failure frequencies.  

IM factor

The IM factor lets MOVES interpolate and extrapolate the non-I/M
emission rates and the I/M emission rates depending on the
characteristics of the I/M program in each county.  Our reference
program, Phoenix, was given an IM factor of 1.  The non-I/M rates were
given an IM factor of 0.  For each model year group and age group
stratification, we used the failure frequencies determined from the
analysis described above to calculate IM factors for the diverse types
of evaporative I/M programs.    REF _Ref229573030 \h  \* MERGEFORMAT 
Figure 3  illustrates how the I/M factor is influenced by the types of
evaporative tests conducted in I/M programs.  We modeled the estimated
failure frequency linearly with the I/M factor, with Phoenix, our
reference program, always receiving a value of 1, and our non-I/M
failure frequency always receiving a value of 0.  Different programs
move along the line, as determined by the analysis from above, based on
which evaporative tests they choose to use.  The figure is an example
using model year group 1999-2003 and age group 4-5.  For these vehicles,
Tucson’s OBD and gas cap tests are annual, compared to Phoenix’s
biennial requirement, which gives Tucson a lower failure frequency and a
higher I/M factor.  Colorado’s frequency is biennial, like
Phoenix’s, but its OBD test is non-enforcing.  As a result, their data
shows a higher failure frequency, which results in a lower I/M factor.

Figure   SEQ Figure \* ARABIC  3  – Example of how we calculated the
I/M adjustment factor.  This figure applies for model years 1999-2003
ages 4-5.

3.3.4 Running Loss

For each vehicle, we calculated fuel tank temperature at the end of the
running loss test using the fuel tank temperature algorithm (see section
  REF _Ref229733262 \h  \* MERGEFORMAT  3.1.2.2 Calculating fuel tank
temperatures during operation ).  The running loss test performed was
the typical 4375-second LA-4 – NYCC – NYCC – LA-4 sequence, with
two minute idle periods following the first LA-4, the second NYCC, and
the final LA-4 (CRC E-41).

We found the average temperature during the test by assuming a linear
increase in temperature during the test.  Thus, the average was
calculated by averaging the start temperature of the test and the final
temperature of the test found in step 1.

We used this average temperature to determine the average permeation
rate during the hot soak via the permeation temperature adjustment using
the 72F base permeation rates determine by model year, age.

We calculated gram/hour rates by dividing total emissions by the
duration of the running loss test (4300 seconds)

We filtered/reduced the data set such that each test met the following
requirements:

Non-leakers (emissions less than 137.2 g/hour  NOTEREF _Ref229733514 \h 
\* MERGEFORMAT  11 ; taken from M6.EVP.009_2.4)

“As received” vehicles (no retests)

Pressure test result must be pass, fail, or blank only (no dashes,
slashes, “I”, etc.)

Subtract permeation from HC for each hour to get tank vapor venting
(TVV) rate

After analysis of TVV data, we found that the best way to stratify
running loss TVV was by model year only.    REF _Ref229733661 \h  \*
MERGEFORMAT  Table 8  shows the results of the analysis.

Table   SEQ Table \* ARABIC  8  – Final running loss tank vapor
venting emission rates.

Model year group	TVV mean [g/hr]

1971-1977	12.59

1978-1995	11.6

1996-2003	0.72

2004 and later	0.23



Since model year group is the only stratification, the running loss TVV
rates are not affected by the failure rates.  Therefore, the I/M and
non-I/M rates are the same and equal to the table above. 

3.4 Liquid Leaks

Liquid leaks are the final evaporative emissions process discussed in
this document.  To calculate the average leaking rate, we used the
leaking vehicles excluded from the previous analysis for tank vapor
venting.  We estimated permeation and tank vapor venting on these
vehicles using the methods described above.  We assumed the remainder of
emissions to be caused by liquid leaks.  We averaged these emissions by
the three different modes, shown in   REF _Ref229738080 \h  Table 9 .  

Table   SEQ Table \* ARABIC  9  – Emission rates for liquid leakers by
mode.

Mode	Liquid leak rate

Cold Soak	9.85 g/hr

Hot Soak	19.0 g/hr

Operating	178 g/hr



The rates in   REF _Ref229738080 \h  \* MERGEFORMAT  Table 9  must be
multiplied by the percentage of leakers in the fleet to get an average
liquid leaking emission rate.  For this we relied on the studies by BAR
and API.  Our estimates of the percentage of liquid leakers are shown in
  REF _Ref229739294 \h  \* MERGEFORMAT  Table 10 .  On average, we
assume that most leaks do not occur until vehicles are 15 years or
older.

Table   SEQ Table \* ARABIC  10  – Percentage of liquid leakers by
age.

Age group	Percentage of leakers in fleet

0-9	0.09 %

10-14	0.25 %

15-19	0.77 %

20+	2.38 %



Combining   REF _Ref229738080 \h  \* MERGEFORMAT  Table 9  and   REF
_Ref229834759 \h  \* MERGEFORMAT  Table 10 , we get   REF _Ref229739946
\h  \* MERGEFORMAT  Table 11 , which shows the liquid leaking rate of
the entire fleet.  We do not assume that this rate changes with model
year or is affected by I/M.

Table   SEQ Table \* ARABIC  11  – Final liquid leak rates in g/hr by
age group and mode.

Age group	Cold soak	Hot soak	Operating

0-9	0.009	0.017	0.158

10-14	0.025	0.048	0.450

15-19	0.075	0.145	1.36

20+	0.235	0.452	4.23

	

Appendix A - Notes on Evaporative Emission Data

Parameters: 	Vehicle Numbers, Test #, Ambient Temperature, RVP, Model
Year, Fuel System, Purge, Pressure, Canister, Gram HC, Retest

E-41 CRC Late Model In-Use Evap. Emission Hot Soak Study (1998): 

50 vehicles (30 passenger cars and 20 light duty trucks)

Age: 1992 to 1997 model year

Average RVP: 6.5 psi

Diurnal Temperature: 72 to 96F

Fuel System: Port Fuel Injection, Throttle Body Injection

Vehicle fuel tank drained and refilled to 40% of capacity with Federal
Evaporative Emission Test Fuel 

Driving schedule will be a full LA-4-NYCC-NYCC-LA4 sequence, with two
minute idle periods following the first LA-4, the second NYCC, and the
final LA-4.

	Hydrocarbon readings will be taken continuously throughout the running
loss test.

	Cumulative mass emissions will be reported at one minute intervals.

Ambient Temperature in running loss enclosure: 95F

E-9 CRC Real Time Diurnal Study (1996):

151 vehicles (51 vehicles from MY 1971 through 1977, 50 vehicles from 

MY1980 through 1985, 50 vehicles from MY 1986 through 1991)

Odometers range from 39,000 to 439,000 miles

Fuel tank volume was 15% of the rated capacity

RVP: 6.62 psi (average sum of 47 vehicles)

Diurnal temperature: 72 to 96F

	Fuel System: Port Fuel Injection, Carburetor, Throttle Body Injection

CRC E-35 Running Loss Study (1997)

	150 vehicles

	Ambient Temperature in running loss enclosure: 95F

	RVP: 6.8 psi

	Fuel System: Port Fuel Injection, Carburetor, Throttle Body Injection

EPA Compliance Data:

2-Day Test

Length of the hot soak: 1 hour

77 vehicles

RVP: average 8.81 psi

Ambient Temperature:

Federal Standard (72 to 96 F) Diurnal

Cal. (65 to 105 F) Diurnal

Hot Soak: 81.67F

Fuel System: Port Fuel Injection

Unleaded Cert Fuel

	CARB Phase II Fuel





MSOD (Mobile Source Observation Database):

	Hot Soak: 1 hour hot soak evaporative test

	FTP: Federal test procedure (19.53 mph), also referred to as the UDDP
schedule

	NYCC: New York City Cycle Test (7.04 mph)

	BL_1A: 1 hour Breathing Loss Evap. Test – Gas Cap left “On”

	BL_1B: 1 hour Breathing Loss Evap. Test – Canister as recd.

	ST01: Engine Start cycle test

	4HD: 4 hour Diurnal test

	24RTD: 24 Hour Real Time Diurnal

	33RTD: 33 Hour Real Time Diurnal

	72RTD: 72 Hour Real Time Diurnal

	3Rest: 3 Hour Resting Loss Evap. Emission Test (follows 1 HR Hot Soak)

	CY6084: Real time diurnal temperature pattern: range 60 to 84 F

	CY7296: Real time diurnal temperature pattern: range 72 to 96 F

	CY8210: Real time diurnal temperature pattern: range 82 to 102 F

	DIURBL: Standard temperature rise for 1 hour diurnal or breathing loss
evaporative emission test

	F505: Bag 1 of federal test procedure (25.55 mph)

	ASM: Acceleration Simulation Mode Test Procedure

	ATD: Ambient Temperature diurnal evaporative Test, shed temp constant,
vehicle begins 24 degree cooler



 CRC Report No. 610, Project E-9.  Measurement of Diurnal Emissions from
In-Use Vehicles.  September 1998. 
http://www.crcao.com/reports/emission/e9.htm.

 CRC Report No. 611.  Project E-35.  Measurement of Running Loss
Emissions from In-Use Vehicles.  Automotive Testing Laboratories. 
Februrary 1998.  http://www.crcao.com/reports/emission/e35.htm.

 CRC Report No. 612.  Project E-35.  Running Loss Emissions from In-Use
Vehicles.  Harold Haskew and Associates, Inc.  Februrary 1999. 
http://www.crcao.com/reports/emission/e35.htm.

 CRC Report No. 622.  Project E-41-1.  Real World Evaporative Testing of
Late-Model In-Use Vehicles.  October 1999. 
http://www.crcao.com/reports/emission/e41.htm.

 CRC Report No. 622.  Project E-41-2.  Evaporative Emissions from
Late-Model In-Use Vehicles.  October 1999. 
http://www.crcao.com/reports/emission/e41.htm.

 Haskew, Harold M., Thomas F. Liberty and Dennis McClement, “Fuel
Permeation from Automotive Systems,” Final Report, for the
Coordinating Research Council and the California Air Resources Board,
CRC Project E-65, September 2004. 
http://www.crcao.com/reports/recentstudies2004/E65%20Final%20Report%209%
202%2004.pdf.

Available in Docket EPA-HQ-OAR-2005-0161.  

 h°s

 h°s

h°s

h°s

-

t

%

-



 

<

=

>

?

R

S

T

n

o

p

q

r

s

t

u

v

’

“

 “

”

•

á

â

ã

ý

þ

ÿ



 

!

"

#

$

%

&

'

C

D

"D

E

F

c

d

e



€

‚

ƒ

„

…

†

‡

£

¤

¥

¦

»

¼

½

×

Ø

Ù

Ú

Û

Ü

Ý

Þ

ß

û

ü

j;

 %

…

Ý

̤̀☊଀Ɇ愀̤摧䵆

똁ਡᜀü

ý

þ

j/

j¸

j¬

&

 h{

gd{

 h{

gd{

 h{

gd{

F

 h{

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

ñÿ

	

h

gd{

Æ

Æ

Æ

 h{

h

‡

‡

‡

‡

‡

‡

‡

‡

‡

‡

␃༁梄币梄愁Ĥ摧憼s

Î

Ï

ഀࣆȀǂȜ@摧憼s

␃ഁ׆Āːཀ킄市킄愂Ĥ摧ざL

j;

 h{

j¸

×

 h{

h{

gd{

 h{

&

F

F

␃ਃ&䘋

Â

Â

Â

Â

␃ਃ&䘋

Â

 h°s

  h°s

  h°s

- h°s

  h°s

  h°s

 h°s

 h°s

  h°s

  h°s

  h°s

- h°s

  h°s

 h°s

  h°s

 h°s

”ÿô

”ÿô

”ÿô

”ÿô

”ÿô

>	™

>	™

>	™

•

 Haskew, Harold M., Thomas F. Liberty and Dennis McClement, “Fuel
Permeation from Automotive Systems: E0, E6, E10, E20, and E85” Final
Report, for the Coordinating Research Council and the California Air
Resources Board, CRC Project E-65-3, December 2006.    HYPERLINK
"http://www.crcao.com/reports/recentstudies2006/E-65-3/CRC%20E-65-3%20Fi
nal%20Report.pdf" 
http://www.crcao.com/reports/recentstudies2006/E-65-3/CRC%20E-65-3%20Fin
al%20Report.pdf 

 Certification fuel tank temperature profiles of top selling passenger
cars and light-duty trucks provided by manufacturers.

 T. Cam, K. Cullen, and S. L. Baldus, Running Loss Temperature Profile,
SAE. 930078, Society of Automotive Engineers, Warrendale, Pa., 1993;

 R.S. Reddy, Prediction of fuel vapour generation from a vehicle fuel
tank as a function of fuel RVP and temperature, SAE 892089, 1989.

 Landman, Larry C.  U.S. EPA MOBILE6 Technical Document M6.EVP.009. 
Evaporative Emissions of Gross Liquid Leakers in MOBILE6
(EPA420-R-01-024).  April 2001. 
http://www.epa.gov/otaq/models/mobile6/r01024.pdf.

 Sierra Research Report No. SR2005-12-03.  United States Motor Vehicle
Inspection and Maintenance (I/M) Programs.  December 2005.

Phoenix (reference I/M)

Non-I/M

Tucson

Colorado

