nox measurement errors in ammonia-containing exhaust · 2014-03-10 · nox measurement errors in...
TRANSCRIPT
1
NOx Measurement Errors inAmmonia-Containing Exhaust
August 22,2006Diesel Engine Emissions Reduction Meeting
Detroit, MI
John Hoard, Rachel Snow, Lifeng Xu, Robert Hammerle, Cliff Montreuil
Chemical Engineering DepartmentChristine Gierczak
Physical and Environmental Sciences Department S. Iskander Farooq
Safety Methods Development Department
DEER 2006
2
Introduction
z Diesel NOx aftertreatment can lead to ammonia (NH3) in exhaust – Urea-SCR systems
– LNT systems near end of rich regeneration
– LNT-SCR combinations
z NH3 can cause significant measurement errors with various NOx instruments
z Larger issue for engine-out and mid-bed samples
DEER 2006
3
Conc
entra
tion
(ppm
) Example of Concern
NOx (Introduced as NO2) Measurement on Various Analyzers
120 NH3 off
100
Mass Spec NH380
add Mass Spec NOx
60 30 100 ppm FTIR NH3
ppm NH340 NO2 FTIR NOx
NH3 to Horiba NOx20 100 ppm
0 0 500 1000 1500 2000 2500
time (seconds)
DEER 2006
4
Experimental Set-upheated filter Mass Spec
heated filter FTIR
Empty Catalyst Can (5.66” dia. x 14” long)
DOC heated filter CLD3 (GC)
heated filter CLD2 (GC)
heated filter CLD1 (COM3)
10 lpm MFC 1% NH3 Heated Un-Heated 2.2L
10 lpm MFC 2% NO Diesel Exhaust Engine Analyzer 25 lpm MFC 2% NO2
exha
ust f
low
DEER 2006
5
Standard Addition Test Concept
Y
7
6
5
4 y = 1.0 *x + 3
3
2
1
0
0 0.5 1 1.5 2 2.5 3 3.5
X
DEER 2006
6
Test DesignReplicate Three
MFC Settings Standard Addition to Exhaust Flow rates NO NO2 NH3 NO NO2 NOx NH3 Total NO NO2 NH3 % % % (ppm) (ppm) (ppm) (ppm) L/min L/min L/min L/min
1 100 100 0 75 226 301 0 1330 10 15 0 2 100 100 100 75 224 299 149 1340 10 15 10 3 50 100 100 37 225 262 150 1335 5 15 10 4 100 50 100 75 113 188 150 1332 10 7.5 10 5 50 50 100 38 113 151 151 1327 5 7.5 10 6 50 50 50 38 113 151 76 1322 5 7.5 5 7 50 50 0 38 114 152 0 1317 5 7.5 0 8 50 100 0 38 226 264 0 1325 5 15 0 9 50 100 50 38 226 263 75 1330 5 15 5
10 100 50 50 75 113 188 75 1327 10 7.5 5 11 100 50 0 76 113 189 0 1322 10 7.5 0 12 100 100 50 75 225 300 75 1335 10 15 5
DEER 2006
7
Chemiluminescent Analyzer (CLD)
z Developed at Ford Research in early 1970’s z Most common NOx measurement z Actually measures NO
– NO + O3 Î NO2* + O2
– NO2* Î NO2 + photon
z NO2 reduced to NO in converter; typical compositions – Carbon-molybdenum
– Molybdenum
– Carbon
– Gold
– Ferrous sulfate
– Stainless steel
DEER 2006
8
Carbon-Molybdenum Converter
0
10
20
30
40
50
60
70
80
90
100 C
onve
rsio
n NO2 NH3
300 350 400 450 500 550
Catalyst Temperature °C
Data from Breitenbach and Shelef
DEER 2006
9
Ammonia Reactions
z NOx can be removed by reactions with ammonia – Ammonium nitrate:
• 3NO2 + H2O Î 2HNO3 + NO (1) • NH3 + HNO3 Î NH4NO3 (2)
– SCR reactions:
• 6NO2 + 8NH3 Î 7N2 + 12H2O (3)
z Must keep sample lines above 100°C to avoid ammonium nitrate
z Choose converter that does not promote these reactions
DEER 2006
10
CLD w/ COM3 Converter Main Effect of NH3 on NOx Measurement
NO
x (p
pm)
600
500
400
300
200
100
0 0 25 50 75 100 125 150
NH3 Standard Addition (ppm)
Non-random Run Order:
Memory Effect
1st Replicate CLD1
2nd Replicate CLD1
3rd Replicate CLD1
DEER 2006
11
CLA NOx Response
z Regression equation from second replicate
z CLA with COM3 catalyst
z NOx = 337 + 1.14 NOadd + 0.924 NO2add - 0.164 NH3add
z R2 = 0.978
z All three coefficients are not zero at 95% confidence
z NOadd and NO2add slopes could be 1.0 within experimental confidence limits
DEER 2006
12
95% Confidence Interval for Regression Coefficient(CLD - COM3)
300
320
340
360
380
400
420
440
460
480
500
NO
x M
easu
red
(ppm
)
Regression (m = 1.14)
95% C.I. High (m = 1.47)
95% C.I. Low (m = 0.82)
Expected Value (m = 1.0)
0 20 40 60 80 100
NO Added (ppm)
DEER 2006
13
CLD w/ Glassy Carbon ConverterMain Effect of NH3 on NOx Measurement
NO
x (p
pm)
600
500
400
300
200
100
0 0 25 50 75 100 125 150
NH3 Standard Addition (ppm)
NH3 Interference
but no Memory Effect
1st Replicate CLD2
1st Replicate CLD3
2nd Replicate CLD2
2nd Replicate CLD3
3rd Replicate CLD2
3rd Replicate CLD3
DEER 2006
14
CLA NOx Response
z Regression equation from second replicate
z CLA with Glassy Carbon catalyst
z NOx = 269 + 1.08 NOadd + 0.982 NO2add - 0.728 NH3add
z R2 = 0.986
z All three coefficients are not zero at 90% confidence
z Within experimental error, NO and NO2 slopes could be 1.0
z Large NH3 interference
DEER 2006
15
CLA Summary
z NOx measurement accuracy and repeatability when NH3 is present is poor for the base COM3 converter.
z NOx measurement repeatability is much better with the updated glassy carbon converter, however a large NH3 interference makes measurements inaccurate.
z CLA NOx measurements are not accurate when NH3 is present in large concentrations.
DEER 2006
16
FTIR
z Fourier Transform Infrared Spectroscopy
z Commonly used gas analysis but potential issues – Interference from gases not included in the analysis set.
– Saturation
– Excessive Interference even with gases in the analysis set
DEER 2006
17
Typical Exhaust Spectrum
A b s o r b a n c e
250 ppm NO
Expanded on next
208 ppm NO2 slide
117 ppm NH3
Exhaust
3500 3000 2500 2000 1500 1000
Wavenumbers
DEER 2006
18
Expanded Scale
A b s o r b a n c e
250 ppm NO
208 ppm NO2
117 ppm NH3
Exhaust
1910 1900 1890 1880 1870
Wavenumbers
DEER 2006
19
FTIR NO Response
z Regression equation from second and third replicates
z FTIR
z NO = 57.7 + 0.843 NOadd + 0.0156 NO2add - 0.0279 NH3add
z R2 = 0.943
z NO and NH3 coefficients are not zero at 90% confidence
z NO coefficient range includes 1; could give accurate NO
z NO2 coefficient is not nonzero at 90% confidence
z No spectral saturation was present in the data
DEER 2006
20
FTIR NO2 Response
z Regression equation from second and third replicates
z FTIR
z NO2 = 172 + 0.187 NOadd + 0.522 NO2add - 0.0141 NH3add
z R2 = 0.839
z NO2 coefficient is not zero at 90% confidence
z NO2 coefficient range does not include 1; NO2 readings are inaccurate
z NO and NH3 coefficients are not nonzero at 90% confidence
DEER 2006
21
95% Confidence Interval for Regression Coefficient(FTIR NO2)
0
20
40
60
80
100
120
140
160
180
NO
2 Mea
sure
d (p
pm)
Regression (m = 0.52)
95% C.I. High (m = 0.67)
95% C.I. Low (m = 0.37)
Expected Value (m = 1.0)
0 20 40 60 80 100
NO2 Added (ppm)
DEER 2006
22
FTIR Summary
z NO measurements are not affected by NH3, but may not match the known addition levels accurately
z NO2 measurements are not affected by NH3 level, and also do not match the known addition levels accurately
z Careful modification of the analysis method may well reduce the error in measurement
z Accurate measurement should be possible but requires careful user interaction
DEER 2006
23
Mass Spectrometer Main Effect of NH3 on NOx Measurement
600
500
400
300
200
100
0 0 25 50 75 100 125 150
NO
NO2
NH3 Standard Addition (ppm)
1st Replicate NO
1st Replicate NO2
2nd Replicate NO
2nd Replicate NO2
3rd Replicate NO
3rd Replicate NO2
ppm
Run order effect
DEER 2006
24
Run Order Effects
3530252015105
50
0
-50
Observation Order
Res
idua
l
Residuals Versus the Order of the Data (response is VF_NO2)
non-random run order trend in
residuals
•10 micron orifice? •Other contamination? •Sample lines?
DEER 2006
25
CIMS NO Response
z Regression equation from second replicate
z CIMS
z NO = 105 + 1.21 NOadd + 0.0143 NO2add - 0.0194 NH3add
z R2 = 0.953
z NO coefficient is not zero at 90% confidence
z NO coefficient range includes 1
z NO2 and NH3 coefficients are not nonzero at 90% confidence; small interferences exist
DEER 2006
26
CIMS NO2 Response
z Regression equation from second replicate
z CIMS
z NO2 = 225 - 0.0682 NOadd + 1.13 NO2add - 0.131 NH3add
z R2 = 0.992
z NO2 and NH3 coefficients are not zero at 90% confidence; interferences exist
z NO2 coefficient range includes 1
z NO coefficient is not nonzero at 90% confidence
DEER 2006
27
CIMS Summary
z Accurate measurements require regular instrument maintenance, apparently due to ammonium nitrate formation in the system
z NO measurements are not affected by NH3, but may not match the known addition levels accurately
z NO2 measurements are affected by NH3 level, and additionally may not match the known addition levels accurately
z Careful recalibration and regular cleaning of the unit may well reduce the error in measurement, but is unlikely to remove the NH3 effect.
DEER 2006
28
Estimate of Engine NOxz From Standard Addition Test, regression gives “zero addition” as
estimate of engine out NOx
z Measurements do not agree!
z Note: CLAs are “dry” where FTIR and CIMS are wet; expect 7% higher reading when water is removed
Instrument NO NO2
CLA COM3
CLA GC
FTIR 58 172
CIMS 105 225 330
230
269
337
NOx
DEER 2006
29
Experiment Summary
z Chemiluminescent analyzers do not provide accurate measurement of NOx when NH3 is present in engine exhaust gas.
z FTIR may provide accurate measurements if analysis methods are developed to provide adequately low interference.
z Chemical ionization mass spectroscopy may provide accurate measurements, although the unit tested appears to need significant cleaning and maintenance to do so.
z No presently available measurement provides accurate and robust measurement of NO, NO2, NOx, and NH3 when NH3 is present in large quantity.
z Recommend development of an NH3 scrubber
DEER 2006