-------
EPA/600/3-88/040
November 1988
INVESTIGATION OF ANALYZER PROBLEMS IN THE
MEASUREMENT OF NOx FROM METHANOL VEHICLES
by
Peter A. Gabele
Emissions Measurement and Characterization Division
Atmospheric Sciences Research Laboratory
Research Triangle Park, N.C. 27711
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
REPRODUCED BY
U.S. DEPARTMENT OF COMMERCE
NATIONAL TECHNICAL INFORMATION SERVICE
SPRINGFIELD. VA. 22161
-------
DISCLAIMER
This report has been reviewed by the Atmospheric Science Research Laboratory,
U.S. Environmental Protection Agency, and approved for publication. Mention
of trade names or commercial products does not constitute endorsement or
recommendation for use.
ii
-------
ABSTRACT
This study was conducted to investigate the extent and source of
irregularities related to the measurement of NOx emissions from methanol cars.
Corrective measures were also explored.
It was observed that NOx chemiluminescent analyzers respond to methanol
and formaldehyde after being exposed to high concentrations of methanol and
formaldehyde over extended periods. This response can cause significant
errors in the measurement of NOx from methanol cars which have inherently
elevated concentrations of exhaust methanol and formaldehyde. The most
effective way of eliminating the spurious response is to clean the analyzer's
reaction chamber regularly when testing vehicles which are being operated on
methanol fuels.
iii
-------
ACKNOWLEDGEMENTS
The author would like to thank Susan Bass, William Ray, and Richard Snow
for their assistance in the completion of this study.
-------
SECTION I
INTRODUCTION
The rapid development of vehicles designed to operate on methanol fuel
has prompted development and evaluation of methods for measuring their
emissions. Classical measurement methods prescribed for gasoline cars are not
always applicable to methanol cars because of the inherently high concen-
trations of methanol and formaldehyde in their emissions. One such exception
applies to the FID (flame ionization detector) measurement of hydrocarbons in
the presence of methanol (1,2,3). Another less publicized exception and the
subject of this report pertains to the chemiluminescent method of measuring
nitrogen oxides (NO ) from methanol cars.
A
The chemiluminescent method is based upon the principle that nitric oxide
(NO) can be reacted with ozone (03) to give about 10 percent electronically
excited N02*. When the electronically excited NOp* transits to its normal
state, a detectable light emission is given off. The intensity of this
emission is directly proportional to the mass flow rate of NO into the
reaction chamber. The light emission is detected and measured by a
photomultiplier tube and the associated electronics process a voltage response
which is proportional to the intensity of light being emitted. To make this
method applicable to NOx (NO + NOp) emissions, the NOp in the sample is
changed to NO in an NOp converter (4).
Problems with chemiluminescent NO measurement from methanol cars were
A
first reported as large variations in NO data and large values of N0? (5).
A ff
Later descriptions identified a sort of residual response which occurred
immediately following sample analysis while the analyzer was being zeroed.
-------
These symptoms were observed with older analyzers which had been exposed to
high formaldehyde and methanol emissions from methanol cars. Because the above
symptoms had not been observed on newer analyzers, an interference problem
caused by contamination of some sort was suspected.
This study was undertaken to determine (1) the source of irregularities
associated with the chemiluminescent measurement of NO emissions from
y\
methanol cars, (2) the severity of the measurement problem, and (3) any
corrective action which might be taken. One of the older analyzers which was
known to have the problem and a brand new analyzer on loan from the manufac-
turer were available for the study. Both were Beckman 951A Chemiluminescent
NO Analyzers.
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
A study was carried out to investigate the extent and source of
irregularities related to the measurement of NOx emissions from methanol cars.
Corrective measures wore also explored. Because the results in this study
were obtained in tests using Beckman 951A Chemiluminescent Analyzers, some of
the conclusions based on those results are relative to the experience with
that particular analyzer. However, there is evidence which suggests that
other types of chemiluminescent analyzers could be similarly affected.
The conclusions of the study are as follow:
1. NOx chemiluminescent analyzers respond to formaldehyde and methanol
gases when exposed to high concentrations of these gases over extended
time periods.
2. The response to formaldehyde and methanol can be severe enough to
cause significant errors in the measurement of NOx emissions from
methanol cars,
3. Analyzer response to methanol accounts for most of the error when
measuring NOx emissions from methanol cars.
4. Reactions involving formaldehyde in the reaction chamber are the
principal cause of spurious analyzer response with samples containing
either methanol or formaldehyde.
-------
5. The most effective way of eliminating the spurious response is to
clean the reaction chamber in accordance with the manufacturer's
instructions.
Further research is recommended to definitively identify the wavelengths
of light emission associated with the interference. Bracketing the range of
those wavelengths might be accomplished through an examination of the effect
of different cut-off filters on response to formaldehyde. Such an approach
might lead to the selection of an improved filter which could eliminate the
interference altogether.
It is also recommended that a study be carried out on other
chemiluminescent analyzers to determine the effect of long term exposure to
high concentrations of methanol and formaldehyde. Thus tar only the Beckman
951A models have been so exposed as a result of emission tests on
malfunctioning methanol cars. Until such studies are completed, one must
assume that all NOx chemiluminescent analyzers could develop problems
associated with extended testing on methanol cars. Therefore, when testing
methanol cars, more than the usual care should be taken by instrument
operators to assure that their instruments are clean and well maintained.
-------
SECTION 3
EXPERIMENTAL PROCEDURES
Upon arrival of the two analyzers involved in the study, a series of
routine checks were completed to assure that instrument performance was within
the manufacturer's specifications. Emphasis was placed on linearity,
stability, and precision. One of the analyzers was new; the other had been in
service for over two years at EPA, Office of Mobile Sources, Ann Arbor,
Michigan. Residual response and large variations in the measurement of NO
A
from methanol cars had been exhibited by the older analyzer, but both
analyzers appeared to be operating in accordance with design specifications.
In order to test for and determine the extent of a suspected interference
problem with formaldehyde and methanol, samples were prepared in the following
manner: Measured quantities of liquid formalin solution (37% formaldehyde by
weight) and methanol were injected through a syringe into a heated (100°C)
reaction chamber. The chamber was swept with zero air flowing at a rate of
5.7 i/m over 10 minute time intervals. The formaldehyde and methanol gases
exiting the reaction chamber were introduced into a Tedlar bag which was
sampled by the analyzers shortly after its preparation.
Some samples were generated using the tailpipe emissions from a 1983
Methanol Escort. The Methanol Escort was being operated on M85 fuel (85%
methanol, 15% gasoline) and its tailpipe emissions were controlled by a
three-way, air-injected catalyst. Analyses of the samples from Bag 1 (cold
transient test phase) of the Federal Test Procedure (FTP) were run because
-------
this test phase contains the highest levels of methanol and formaldehyde. NO
emissions were measured with a TECO chemiluminescent analyzer for comparison
with measurements made using the older Beckman 951A. During these experi-
ments, both analyzers were zeroed and spanned using the same gas cylinders.
Troubleshooting the contamination problem on the older Beckman analyzer
was carried out by systematically exchanging parts from that analyzer with
parts from the uncontaminated new Beckman. The entire photomultiplier
tube-reaction chamber unit was first exchanged, then components within that
unit including the red filters, photomultiplier tubes, and reaction chambers
were individually switched out. After each exchange, the analyzers were
rechecked for zero and span stability, and for response to formaldehyde or
methanol. Frequent linearity checks were made using a flow divider in con-
junction with the span gas.
At one point in the study, a catalyst was devised to scrub the
formaldehyde and methanol at the sample inlet to the analyzer. The catalyst
was 0.5% rhodium on alumina substrate, occupied a volume of about 50ml, and
2
had a surface area of 200 m /g. Temperature of the catalyst was controlled by
varying the current flow through heat tape which had been applied to a glass
tube containing the catalyst material.
Cleaning of the reaction chamber in the older Beckman analyzer was done
by scrubbing with Alconox in deionized water, then by soaking the chamber in
50 percent concentrated HC1 for five minutes, followed by a thorough rinse
with deionized water. These cleansing techniques were done in accordance with
procedures given in the instrument manual.
-------
SECTION 4
RESULTS AND DISCUSSION
Two Beckman 951A Chemiluminescent NO Analyzers were involved in this
A
study. One had been used in testing by EPA, Ann Arbor, the other was a new
instrument on loan from Beckman Industrial. The instrument from Ann Arbor had
been in service for over two years and was exhibiting symptoms of high mea-
surement variability and residual response to methanol car emissions. During
preliminary tests performed in Ann Arbor, analyzer response to formaldehyde
was observed.
Following delivery and check out of the older instrument, tests were
conducted to determine the extent of the reported formaldehyde interference.
The results of these tests are given in Table 1. Response to formaldehyde
tends to increase more rapidly at the lower formaldehyde concentrations, then
tails off at levels above 150 ppm. Response was greater when the analyzer was
being operated in the NO mode (converter by-pass) and was greater in both NO
and NO modes as ozone pressure was reduced from 30 psi to 5 psi. For these
A
tests the converter temperature was maintained at 220°C, its optimum for N02
to NO conversion.
Tests were also run on the older analyzer to determine if it responded to
methanol. After obtaining an appreciable response to a 250 ppm methanol
sample, responses to lower methanol concentrations were measured with the
results given in Table 2. Although the analyzer responded less to methanol
than formaldehyde, NO measurement irregularities experienced with this
-------
analyzer were probably more Influenced by methanol emissions because these
greatly exceed those of formaldehdye from methanol cars.
When the analyzer was operated in the NO mode (sample is by-passed around
a N02 to NO converter), response to methanol decreased sharply (see Table 2).
This was opposite the experience with formaldehyde where the sample bypassed
around the converter resulted in a greater response. These results suggest
that methanol could be reacting in the converter to form formaldehyde which
ultimately causes the response. There is a very strong likelihood of this
occurrence because the converter had been optimized and was operating at a
relatively cool 220°C, a temperature highly conducive to the conversion of
methanol to formaldehyde. Commercial production of formaldehyde takes place
through low temperature conversion of methanol over a catalyst.
A preliminary measure taken to remedy the interference problem was to
combust the formaldehyde and methanol by increasing the converter's tempera-
ture. As converter temperature was increased, response to both formaldehyde
and methanol decreased until essentially no response was detected at 400°C.
For the remedy to work the tests had to be run in the NO mode with the sample
A
passing through the converter; therefore, interference in the NO mode (con-
verter by-pass) remained unimproved. In addition to this drawback, the
heightened converter temperatures resulted in lowered NO^ to NO conversion
efficiencies. This was undesirable because analyzer performance was being
severly compromised.
Use of a catalyst (rhodium on alumina) for scrubbing out methanol and
formaldehyde in the sample gas to the analyzer was also examined. At elevated
-------
temperatures the catalyst was effective in oxidizing both methanol and
formaldehyde and in eliminating the interference problem; however, it was also
effective in reducing the NO in nitrogen used to span the analyzer. When
catalyst temperature was reduced to ambient levels the span problem was
eliminated and both methanol and formaldehyde were being sufficiently
scrubbed. But after prolonged catalyst usage, instrument zero began to drift
upwards suggesting gradual elution of formaldehyde.
The new 951A analyzer from Beckman arrived during the time that the
catalyst scrubber for the old analyzer was being evaluated. The new analyzer
did not respond to either formaldehyde (100 ppm) or methanol (250 ppm). This
finding was consistent with reports indicating that the interference problem
appeared only in analyzers which had been exposed to high concentrations of
methanol car exhaust for extended periods. Because the new analyzer did not
respond to formaldehyde, the task of isolating the component on the older
analyzer responsible for response became simplified. By systematically
exchanging parts between the two analyzers, the responsible component was
found to be the reaction chamber. After the reaction chamber was cleaned,
reassembled, and tested, response to formaldehyde (100 ppm) and methanol (250
ppm) was less than 2 ppm.
NO measurements were made on tailpipe emissions from a Methanol Escort
with the older Beckman 951A before and after its reaction chamber was cleaned.
-------
V
Emissions from Bag 1 of the FTP were measured because these contain relatively
high levels of methanol and formaldehyde. In an earlier study on this vehicle
(6), the Bag 1 methanol and formaldehyde concentrations were about 50 ppm and
3.5 ppm, respectively. FTP emissions results from this study are given in
Table 3. Although the emission rates given for carbon monoxide and total
organics (total hydrocarbons + methanol + formaldehyde) exceed the standard
levels being promulgated for methanol cars (7), they are typical for an older
car such as this one was.
NO measurements on Bag 1 were made using a TECO chemiluminescent analyz-
«
er for comparison with results from the older Beckman 951A. The TECO instru-
ment had not been previously exposed to high concentrations of methanol car
emissions and did not respond to high concentrations of methanol or
formaldehyde. The results shown in Table 4 illustrate the effect of cleaning
the reaction chamber on agreement between measurements made with the TECO and
the older Beckman analyzer. Before cleaning, the Beckman measurement was
about 35 percent higher than the measurement from the TECO. After cleaning,
the difference between measurements was reduced to about 3 percent.
Toward the conclusion of the study, the new Beckman 951A began responding
to formaldehyde. A response between 4 and 7 ppm was obtained with a 100 ppm
formaldehyde sample. No response to methanol occurred. It is estimated that
this first observed response to formaldehyde occurred after subjecting the new
analyzer to three to four 60 liter bags of 100 ppm formaldehyde and two to
three bags of 250 ppm methanol. Unfortunately, time restraints prohibited any
further investigation of the effects of prolonged exposure to formaldehyde on
analyzer performance.
10
-------
REFERENCES
1. H. Menrad, W. Lee, and W. Bernhardt, "Development of a Pure Methanol
Car", SAE Paper 770790, Detroit, MI, Sept. 1977.
2. P. Gabele, et al., "Characterization of Emissions from Vehicles Using
Methanol and Methanol-Gasoline Blended Fuels", JAPCA 35:1168-1175,
November 1985.
3. G.D. Ebersole, F. Manning, "Engine Performance and Exhaust Emissions:
Methanol versus Isooctane", SAE Paper 720692, San Francisco, CA, Aug.
1972.
4. J.E. Sigsby, F.M. Black, T.A. Bellar, D.L. Klosterman, "Chemiluminescent
Method for Analysis of Nitrogen Compounds in Mobile Source Emissions (NO,
N02, and NH^)", Environmental Science & Technology, Vol. 7, January
1973.
5. MEMO: "NOx Measurement with Methanol Fuel" from William Clemmens, TSS,
EPA, Ann Arbor, MI, to Frank Black, Chief, MSERB, EPA, RTP, N.C., April
9, 1985.
6. R. Snow et al., "Characterization of Emissions from a Methanol Fueled
Motor Vehicle", submitted to JAPCA for publication, preprint available
through EPA, MD-46, RTP, NC 27711.
7. Draft "Emissions Standards and Test Procedures for 1990 and Later
Methanol-Fueled Light-Duty Vehicles and Trucks, Heavy-Duty Engines, and
11
-------
Motorcycles", Final Rule submitted for red border review July 1988, EPA,
QMS, Ann Arbor, MI.
12
-------
TABLE 1. ANALYZER RESPONSE TO FORMALDEHYDE
Formaldehyde
Concentration
(ppm)
0
25
50
100
150
200
=============:
NOx Mode
30 psi 03
0.5
10.7
16.1
21.8
25.5
27.0
RESPONSE
NO Mode
30 psi 03
0.6
11.7
17.6
23.5
29.2
••
================:
NOx Mode
5 psi 03
0.5
21.0
28.0
-
32.5
—
=============
NO Mode
5 psi 03
0.5
21.3
29.6
-
34.0
—
-------
TABLE 2. ANALYZER RESPONSE TO METHANOL
Methanol
Concentration
(ppm)
0
31
62
125
:================
NOx Mode
30 psi 03
0.8
8.3
11.2
14.1
================
RESPONSE
NO Mode
30 psi Og
0.8
2.9
3.7
4.7
===========================
NOx Mode NO Mode
5 psi 03 5 psi 03
0.8
15.0 13.4
-
_ _
14
-------
TABLE 3. FTP EMISSION RATES FOR THE METHANOL ESCORT
THC, g/mi 0.48
Methanol, g/mi 0.96
CO, g/mi 6.97
NOX, g/mi 0.64
HCHO, mg/mi 122.30
15
-------
TABLE 4. ANALYZER COMPARISON OF BAG 1 MEASUREMENTS
OF METHANOL ESCORT N0 EMISSIONS
TECO BECKMAN 951A
NOx Mode
(ppm)
13.6
12.3
NOx Mode
(ppm)
18.6*
12.7**
NO Mode
(ppm)
14.5*
_
% Dlff
NOx Mode
+ 36.8%
+ 3.2%
* Before 951A was cleaned
** After 951A was cleaned
16
-------