EPA-AA-IMG-84-3
Investigations Into The Emissions
Effects of Vehicle Misfueling
Jane A. Armstrong
Project Manager, I/M Group
Office of Mobile Sources
U.S. Environmental Protection Agency
Presented To:
The Tenth North American Motor Vehicle
Emission Control Conference
State and Territorial Air Pollution
Program Administrators
New York, New York
April 2, 1984
-------
-2-
Investigations into the Emissions Effects
of Vehicle Misfueling
So far this afternoon, we have heard about the extent of
tampering and misfueling around the country. These
statistics are dismaying when one envisions the loss in
emission control that results. Our role, in the Emission
Control Technology Division, is to quantify that loss of
control, and our most recent projects have focused on the
effects of vehicle misfueling.
In our test programs, we are investigating how both the
amount and the frequency of misfueling affect emissions.
There are four different programs, three of which have been
completed and one which is still under way. The first
program, which we call "ATL Continuous," subjected five 1981
and 1982 model year vehicles to ten successive tanks of
leaded fuel. The program was conducted by Automotive Testing
Laboratories in Liberty, Ohio, and mileage was accumulated
rapidly on the vehicles at the ATL Ohio test track. The
second, which we call "ATL Frequent," subjected six 1981 and
1982 model year vehicles to 12 tanks of leaded fuel in cycles
where one out of two tanks was leaded. The last two test
programs, called "ATL Intermittent" and "EPA Intermittent"
-------
_ "3 _
focused on the effects of using one tank of leaded fuel out
of every four to five tanks of fuel consumed. At ATL,
mileage was accumulated quickly on the test track, but the
cars were operated on a variable driving cycle intended to
simulate real world driving, averaging approximately 30 miles
per hour. In the EPA program, mileage was accumulated
through normal driving practices. We accomplished this by
utilizing the fleet of leaner vehicles given to owners who
participate in the emission factor program at the Motor
Vehicle Emission Laboratory in Ann Arbor. The EPA program is
still under way.
This slide shows the effect of misfueling on the vehicles in
the ATL Continuous study. The first bar of each histogram
shows the baseline, the second the emission level after ten
tanks, and the third the level with the catalyst removed.
Although you do not see it in these histograms, emissions
were measured after each two tanks of leaded fuel were
consumed.
HC emissions increased fairly steadily for all five vehicles,
although not to the same degree. In this program, there was
a 76% loss of HC conversion efficiency after ten tanks of
leaded fuel. Also the HC increases are more dramatic than
those for NOx and CO, similar to observations in previous
-------
-4-
testing programs. Catalysts are generally more negatively
affected by misfueling in their HC conversion efficiency.
Later, when we compare the effects of continuous vs. frequent
vs. intermittent misfueling we will look only at the HC
comparisons.
For the four vehicles with three-way catalysts, NOx emissions
rose to nearly double the baseline levels. Once again the
rate of increase varied from vehicle to vehicle. After ten
tanks, NOx conversion efficiency was reduced by 29%.
The increase in CO emissions is not so dramatic as that for
HC, but still the average CO emissions after ten tanks are
2.8 times the baseline levels. Conversion efficiency loss is
approximately 45%. The degree of emission increase varied
considerably from vehicle to vehicle, however. For example,
the CO emission increases of the five vehicles ranged from
10% to 300%. CO emissions increased significantly for only
four out of the five vehicles, and often leveled off after a
few tanks of leaded fuel, rather than continually increasing.
We noticed a similar, but not so dramatic leveling off of the
increase in HC emissions. Hydrocarbon levels increased
rapidly with the first four tanks and then rose more slowly.
We believe that the emission levels would continue to rise
with subsequent misfueling, eventually reaching the catalyst
-------
-5-
removed levels. However, these levels were not reached after
ten tanks of leaded fuel.
An additional element of the ATL Continuous study was the
replacement of the oxygen sensor on the three closed-loop
vehicles. This slide shows the HC levels at five different
test conditions: baseline, after ten tanks, with the catalyst
removed, with a new catalyst installed, and with a new oxygen
sensor. It was observed that the main impact of misfueling-
was on catalyst poisoning, since the emission levels returned
to near baseline when new catalysts were installed. With new
oxygen sensors, HC and CO decreased further on all three
vehicles, but NOx increased on two vehicles. This indicates
the poisoned oxygen sensors were sending incorrect signals
indicating a richer air-fuel mixture was needed, when it
really was not.
With ATL Continuous serving as our focal point, we then
varied the testing conditions for comparison purposes. First
the five poisoned catalysts and oxygen sensors from ATL were
sent to the Colorado Department of Health for testing on
matching vehicles at high altitude. The lab in Colorado
attempted to obtain vehicles which matched the ATL vehicles
in terms of model year, manufacturer, transmission, engine,
catalyst, and mileage accumulation. In one case a match was
-------
-6-
not possible, so the high altitude comparison is based on
only four vehicles. These vehicles were all equipped with
three-way catalysts, and three of the four were closed-loop.
The next three slides illustrate the comparison between low
and high altitude effects of continuous misfueling. The
histograms appear to indicate similar effects at low and high
altitude, but a calculation and comparison of emission
increases and conversion efficiency losses would indicate a
significantly lower effect at high altitude. Frankly, we can
think of no reason why this should be the case and are
inclined to attribute the difference to car to car
variability and small sample sizes. We are, however, looking
more closely at the individual test data to see if there is a
better explanation. We are comfortable at this point saying,
at least, that misfueling at high altitude has no greater
adverse effect than at low altitude.
Next we turned to the frequency question. We can envision
circumstances where a vehicle would not be habitually
misfueled. For instance, there may be only one member of a
household who misfuels, and this member may drive the vehicle
infrequently. We were also interested in finding out whether
operating the vehicle on unleaded fuel for a period of time
might restore some or all of the catalyst's conversion
efficiency.
-------
-7-
As I explained before, we have been conducting four different
test programs in which vehicles are misfueled either
continuously or one out of two tanks or one out of five
tanks. The two intermittent programs are the ones in which
the vehicles received the most normal mileage accumulation.
This slide shows the first comparison of the four test
programs. Here we are plotting FTP HC emission levels
against the number of leaded tanks. A preliminary conclusion
might be that the frequency of misfueling is not a factor in
the rate of emission increase. We also see a somewhat
surprising result in the fact that the vehicles in the
frequent misfueling study reach higher HC emission levels
than those in the continuous study.
We then decided to look at the comparison by plotting HC
emission levels against grams of lead consumed. The grams of
lead measure may also be thought of in terms of gallons,
since the fuel we used contained very close to one gram of
lead per gallon, which is the average of commercially
available leaded fuels.
Now we begin to see some difference in the effects of casual
vs. habitual misfueling. The continuous and frequent study
vehicles reach similar levels and so do those in the two
-------
-8-
intermittent studies. The rate of emission increase,
however, is definitely lower in the intermittent misfueling
programs.
One final measurement of interest is the increase in methane
versus non-methane hydrocarbons. As you can see from this
slide, there is an increase in methane emissions, about 38%.
Non-methane HC emissions, however, increase by 321%. Thus,
the effect of misfueling is to cause a sharp rise in reactive
hydrocarbons in the atmosphere.
To summarize the conclusions from our ongoing investigation
into the emission effects of misfueling catalyst vehicles:
1. Emission levels steadily increase with misfueling such
that after ten tanks, HC emissions are over four times
the baseline levels; CO emissions nearly three times;
NOx emissions nearly twice the baseline levels.
2. Most catalyst deactivation occurs within four tanks of
leaded fuel. HC and CO emissions continue to increase
with further misfueling, but not to the same degree.
After ten tanks, catalysts are not completely
deactivated, but only about one-fourth of the original
HC control, half the original CO control, and three
quarters of the original NOx control remain.
-------
-9-
3. There is, as expected, car to car variability in the
effect of misfueling, particularly for CO.
4. The primary reason for the emission increase is that the
catalyst is poisoned. The oxygen sensor is also
affected, but is only responsible for a small portion of
the increase.
5. Misfueling at high altitude does not appear to have a
greater emission impact than at low altitude.
6. Continuous or habitual misfueling causes emissions to
rise at a higher rate than intermittent or casual
misfueling.
7. The increase in total hydrocarbons is comprised mostly
of non-methane HC.
Before I close this afternoon, I would like to give you a
preview of coming attractions. There is a growing interest
in finding a short test for catalyst function that can be
used in inspection programs which conduct misfueling checks.
It has always been our custom to perform the whole battery of
-------
-10-
short emission tests along with the FTP at each test cycle.
In the EPA Intermittent test program which is still running
we are also exploring the potential of other tests to
accurately identify vehicles which exceed FTP emission
standards due to misfueling. There is no time left today,
and it is a bit premature to draw conclusions, but I just
want to give a feeling for the options being investigated.
These include:
Skin temperature measurements at the inlet and outlet of
the catalyst to associate catalyst activity with a
minimum temperature rise;
Idle and loaded mode tests with a spark plug
disconnected to discover if the additional HC burden
will cause a misfueled vehicle to fail a short test.
Higher speeds and loads to improve the loaded short
test's ability to identify misfueled vehicles;
A short transient test consisting of the first two
cycles of Bag 2 of the FTP; and
Gamma ray measurements to detect lead accumulation in
the catalysts.
That is the conclusion of my presentation. I would like to
thank you for the opportunity to present the results of our
studies, and I would be happy to answer any questions you may
have.
-------
MISFUELING TEST PROGRAMS
ATL CONTINUOUS I:I
ATL FREQUENT 1:2
ATL INTERMITTENT 1:5
EPA INTERMITTENT 1:5
-------
FIVE CONTINUOUSLY MISFUELED CARS
CO
c
g
*co
CO
LLJ
X
O
z:
Q
2
<
O
2.6-
2-
1.6-
1-
0.6-
Legend
EZl BASELINE
ID 10 TANKS
ESD NO CATALYST
HC
NOx
30
-26
-20
-16
co
c
O
"co
co
LJJ
O
O
-10
-6
-------
2.
MISFUELED HC VS NO-CRTRLYST
2. 0
RVERRGE EMISSIONS - CRTRLYST REMOVED
in
•ZL
O
t—«
cn
CO
•—i
21
UJ
1 . 6
1 . 2
o_ 0 . 8
H-
U_
0. i|
CONTINUOUSLY MISFUELED (N=U)
o. o
_L
_L
0 20
40 60 80
GRRMS OF LERD
100 120
-------
NEN CRTflLTST flND NEN OXYGEN SENSOR
RVERflGE HC EMISSIONS
2. 2
2. 0
1 R
1 . O
1 . 6
i u
1 . 4
t\
£l.2
A 1 r\
u— 1 . U
i__
1 1
0.8
OR
. D
0. LI
0 2
0. 0
.
0. 30
\\
N. \
\ \
1 c; u
1 . D 4
1 . 99
i "^ Q
j . j j
//,D.20
//
//
/ /
,\ RflSFI I NF
1 1 1
.ill 10 TflNKS LERD
Ml) iMTH YnT
IN u u> n i n i_ i j i
Z~
, NFN CflTfll YST
IIIIIIH M c LI r Q T x M P LI no
UJ1IJJJ NLW LHI fi. INLW U c.
3 CLOSED-LOOP VEHICLES
-------
2. 4
2. 0
1 . 6
1 . 2
Q_
I—
U_
0. 8
0. 4
0. 0
RLTITUDE COMPflRISON - HC
2. 04
1 . 70
0. 37
LOW flLTITUDE
1 . 88
1 . 29
0. 48
HIGH flLTITUDE
BflSELINE
MISFUELED (10)
NO CflTRLYST
-------
fl'LTITUDE COMPflRISON - CO
30
25
? 0
o
LJ
1 C
1 O
Q_
h-
U_
10
5
0
-
LI. 7
i
1
5
»
<-
)
20 o
c: . c:
8. 4
v\
\\
1
L|
q
I
25. 9
A
II
5] BflSELINE
MISFUELED (10)
1 NO CRTflLYST
LON flLTITUDE
HIGH RLTITUDE
-------
. 0
3. 5
3. 0
2. 5
x
o
> 0
Q_
1 . 5
1 . 0
0. 5
0. 0
fl L TIT UDE COMPRRISON- NOX
2. 87
1 . 32
0.70
LOW RLTITUDE
1 . 76
0. 89
HIGH RLTITUDE
BflSELJNE
MJSFUELED (10)
NO CflTflLYST
-------
2.U
Q_
0. 0
HC VS TRNKS OF LEflD
CONTINUOUS
FREQUENT
- INTERMIT (flTL)
- INTERMIT (EPfl)
6 8 10
TflNKS OF LEflD
-------
HC vs G'RRMS OF LERD
LU
_J
•—i
21
cn
2.
2. 0 .
en i . 6
a:
1 . 2
cn
in 0.8
UJ
o_ O.M
I'-
LL
0. 0
CONTINUOUS
oaa FREQUENT
*• INTERMIT (flTL)
- INTERMIT (EPfl)
0
50 75 100 125
GRRMS OF LEflD.
150
175
-------
METHflNE VS NON-METHflNE
I .2
1 . 0
UJ
0. 8
IT)
21
CE
DC
0. 6
00. 11
Q_
0.2
0.0
38X
METHRNE
321X
NON-METHflNE
EH BflSEL JNE (N=l0)
El 4 TflNKS LEflD
-------
ADDITIONAL CATALYST CHECKS
CATALYST SKIN TEMPERATURE MEASUREMENTS
SHORT TESTS WITH SPARK PLUG DISCONNECTED
LOADED SHORT TESTS AT VARIOUS SPEEDS
TRANSIENT LOADED SHORT TEST
GAMMA RAY LEAD MEASUREMENTS
------- |