ATTACHMENT A TO ICAO MANUAL APPENDIX 5. SPECIFICATION
FOR HC ANALYSER Note 1.- As outlined in 5.2 of Appendix 5, the measuring element in this analyzer is the flame ionization detector (FID) in which the whole or a representative portion of the sample flow is admitted into a hydrogen-fueled flame. With suitably positioned electrodes an ionisation current can be established which is a function of the mass rate of hydrocarbon entering the flame. It is this current which, referred to an appropriate zero, is amplified and ranged to provide the output response as a measure of the hydrocarbon concentration expressed as ppmC equivalent.
Note 2.— See Attachment D for information on calibration and test gases.
1. GENERAL Precautions: The performance specifications in dicated are generally for analyzer full scale.
Errors at part scale may be a significantly greater percentage of reading. The relevance and importance of such increases shall be considered when preparing to make measurements. If better performance is necessary, then appropriate precautions shall be taken.
Supplemental information The performance specifications for these analyzers, given in terms of full scale response, can have a significant and adverse impact on part scale measurements. If this instrument is used for both the afterburning and non afterburning modes, which is likely, then the comments made in Appendix 3, Attachment A apply here as well. In extreme instan ces, concentrations of HC at high power, such as take-off, can differ from concentratio ns at idle by orders of magnitude, e.g., 3,500 ppmC at idle and 10 ppmC at take-off. The FID specification sugges ts the high level can be 5,000 ppmC. For most modern engines the idle concentrations are considerably lower, but still of concern when setting up the FID to maintain suitable accuracy of measurement. In general, where concentrations of species to be measured are known to vary this way it is always good practice to use a multi-range instrument and to adjust ranges such as to keep the measurem ent in the upper 30% of the instrument response range. A measurement made at 20% of full scale coul d result in an error 5 times the error specified as a percent of full scale. This is a general precaution. Some modern instruments with internal electronic ranging and calibration capability can be used over their entire range without penalty.
Calibrations should be performed on each range used as required.
The instrument to be used shall be such as to maintain the temperature of the detector and sample-handling components at a set point temperature within the range 155 ºC to 165 ºC to a stability of ±2°C. The leading specification points shall be as follows, the detector response having been optimized and the instrument generally having stabilized:
Supplemental information Stability is usually defined in terms of a time interval which, because this is a temperature control set point, can be taken as the duration of the test.
a) Total range: 0 to 500 ppmC in appropriate ranges.
Supplemental information The total range of 0 – 500 ppmC reflects its singular use for testing afterburning engines. Since the operational range specified in Appendix 3, Attachment A, covers the non-afterburning range too, an instrument with a total range of 0 – 5,000 ppmC wh ich incorporates a range of 0 – 500 ppmC is likely to be the instrument of choice. As far as possible, measurements should be restricted to the upper 30% of full scale on the range or ranges used except as noted above.
b) Resolution: better than 0.5 per cent of full scale of range used or 0.5 ppmC, whichever is greater.
c) Repeatability: better than ±1 per cent of full scale of range used. or ±0.5 ppmC.
whichever is greater.
d) Stability: better than ±2 per cent of full scale of range used or ±l.0 ppmC, whichever is greater, in a period of 1 hour.
e) Zero drift: less than ±1 per cent of full scale of range used or ±0.5 ppmC, whichever is greater, in a period of 1 hour.
f) Noise: 0.5 Hz and greater, less than ±1 per cent of full scale of range used or ±0.5 ppmC, whichever is greater.
Supplemental information The FID requires fuel and oxidant gases for operation. The fuel gas is typically either a mixture of hydrogen and nitrogen or hydrogen and helium. If the noise specification cannot be met and a hydrogen/nitrogen mixture is being used as the fuel gas, it sometimes helps to change to a hydrogen/helium mixture.
g) Response time: shall not exceed 10 seconds from inlet of the sample to the analysis system, to the achievement of 90 per cent of the final reading.
h) Linearity: response with propane in air shall be linear for each range within ±2 per cent of full scale, otherwise calibration corrections shall be used.
2. SYNERGISTIC EFFECTS Note.- In application there are two aspects of performance which can affect the accuracy of measurement:
a) the oxygen effect (whereby differing proportions of oxygen present in the sample give differing indicated hydrocarbon concentration for constant actual HC concentrations); and
b) the relative hydrocarbon response (whereby there is a different response to the same sample hydrocarbon concentrations expressed as equivalent ppmC, dependent on the class or admixture of classes of hydrocarbon compounds).
The magnitude of the effects noted above shall be determined as follows and limited accordingly.
Oxygen response: measure the response wi th two blends of propane, at approximately 500 ppmC concentration known to a relative accuracy of ±l per cent, as follows:
1) propane in 10 ± 1 per cent O2, balance N2 Page 98
2) propane in 21 ± 1 per cent O2, balance N2 If R1 and R2 are the respective normalized responses then (R 1 - R2) shall be less than 3 per cent of R1.
Supplemental information Unlike the non-afterburning case, the range of oxygen values can vary over a fairly wide range. This will be influenced by the location of the sampling pl ane and the degree of mixing of ambient air with the reacted plume. If needed, the O 2 response can be minimized by adjusting the FID burner fuel/air ratio.
Differential hydrocarbon response: measure the response with four blends of different hydrocarbons in air, at concentrations of approximately 500 ppmC, known to a relative accuracy of ±l per cent, as follows:
a) propane in zero air
b) propylene in zero air
c) toluene in zero air
d) n-hexane in zero air.
If Ra, Rb, Rc and R d are, respectively, the normalized responses (with respect to propane), then (Ra - Rb), (Ra - Rc) and (Ra - Rd) shall each be less than 5 per cent of Ra.
Supplemental information Although assumed to respond as a carbon number counter, for the purposes of testing engine exhaust, the FID response is known to vary some what with the particular hydrocarbon or class of hydrocarbons being measured. For example, three molecules of methane, (CH4), will not necessarily result in the same instrument response as one molecule of propane, (C 3H8). Because of this differential response it is sometimes helpful to think of the FID as responding to an “effective” carbon number. It is important to ensure the instrument responds acceptably to all of the hydrocarbons in the engine exhaust. This group of hydrocarbons – propylene, toluene, and n-hexane, with propane as a reference – was chosen to represent, in terms of instrument response, the range of hydrocarbons expected in the exhaust.
3. OPTIMIZATION OF DETECTOR RESPONSE AND ALIGNMENT
3.1 The manufacturer's instructions for initial setting up procedures and ancillary services and supplies required shall be implemented, and the instrument allowed to stabilize. All setting adjustments shall involve iterative zero checking, and correction as necessary.
Using as sample a mixture of approximately 500 ppmC of propane in air, the response characteristics for variations first in fuel flow and then, near an optimum fuel flow, for variations in dilution air flow to select its optimum shall be determined. The oxygen and differential hydrocarbon responses shall then be determined as indicated above.
Supplemental procedure The FID detector response and alignment can be optimized by adjusting the FID burner fuel and air flow while sampling a mixture containing approximately 500 ppmC propane. Care should be taken when changing fuel flow that the instrument zero re sponse point does not shift. If it does, the instrument zero should be reset. Response curves illustrating this process, shown below in Figures A5-5 and A5-6, were taken from 40 CFR Part 87, Control of Air Pollution from Aircraft and Aircraft Engines, Federal Register, Vol. 43, No. 58 — Friday, March 24, 1978. They were taken, in turn, from Page 99
SAE ARP1256, "Procedure for the Continuous Sampling an d Measurement of Gaseous Emissions from Aircraft Turbine Engines".
Figure A5-5 Typical Fuel Flow Response Curve Figure A5-6 Typical Airflow Response Curve The objective of this procedure is to select operating flow rates that will give near maximum response with least variation for minor changes in fuel flow. It may be necessary to repeat this operation in an iterative fashion:
• Adjust the fuel flow to maximize output;
• Adjust zero if necessary;
• Adjust the air flow to maximize output;
• Readjust the fuel flow, if necessary; and
• Repeat until the burner output is optimized.
3.2 The linearity of each analyzer range shall be checked by applying propane in air samples at concentrations of approximately 30, 60 and 90 per cent of full scale. The maximum response deviation of any of these points from a least squares straight line (fitted to the points and zero) shall not exceed ±2 per cent of full scale value. If it does, a calibration curve shall be prepared for operational use.
Table A5-1 HC ANALYZER PERFORMANCE SPECIFICATIONS Parameter Value Total Range 0 to 500 ppmC (5 000ppmC for non AB) Resolution better than 0.5% full scale range used or 0.5 ppmC whichever is greater Repeatability better than ±1% full scale range used or ±0.5 ppmC whichever is greater Stability better than ±2% full scale range used or ±1 ppmC whichever is greater in a period of 1 hour Zero drift < ±1% full scale range used or ±0.5 ppmC, whichever is greater in a period of 1 hour Noise 0.5 Hz and greater, < than ±1% full scale range used or ±0.5 ppmC, whichever is greater Response time ≤ 10 seconds from inlet of sample to achieve 90% of final reading Linearity ±2% full scale to propane in air, for each range used, or calibration correction required The performance specifications shown in Table A5-1 are typical of those analyzers offered by major analyzer manufacturers.
ATTACHMENT B TO ICAO MANUAL APPENDIX 5.
SPECIFICATION FOR CO AND CO2 ANALYSERS Note 1.- Paragraph 5.3 of Appendix 5 summarizes th e characteristics of the analysis sub-system to be employed for the individual measurements of CO and CO 2 concentrations in the exhaust gas sample. The instruments are based on the princi ple of non-dispersive absorption of infra-red radiation in parallel reference and sample gas cells. The required ranges of sensitivity are obtained by use of stacked sample cells or change s in electronic circuitr y or both. Interferences from gases with overlapping absorption bands may be minimised by gas absorption filters and/or optical filters. preferably the latter.
Note 2.— See attachment D for information on calibration and test gases.
Precautions: The performance specifications indicated are generally for analyzer full scale.
Errors at part scale may be a significantly greater percentage of reading. The relevance and importance of such increases shall be considered when preparing to make measurements. If better performance is necessary, then appropriate precautions shall be taken.
Supplemental information The performance specifications for these analyzers given in terms of full scale response can have a significant and adverse impact on part scale me asurements. This needs to be considered when planning and executing the test and in evaluating t he accuracy of the measurements after the test.
Concentrations of CO, when going from the idle m ode to take-off, can differ by orders of magnitude.
In general, where concentrations of species to be meas ured are known to vary this way, it is always good practice to use a multi-range instrument and to choose ranges such as to keep the measurement in the upper 30% of scale on the ran ge in use. A measurement made at 20% of full scale could result in an error 5 times the error spec ified as a percent of full scale. This is a general precaution. Some modern instruments with inter nal electronic ranging and calibration capability can be used over their entire range without penalty. Calibrations should be performed on each range used as required.
The principal performance specification shall be as follows:
CO Analyzer
b) Total range: 0 to 2 500 ppm in appropriate ranges.
Relative to the precautions mentioned above ranges are chosen that the instrument responds in the upper 30% of scale for the range in use. While this may not always be possible it should be a goal.
b) Resolution: better than 0.5 per cent of full scale of range used or 1 ppm, whichever is greater.
c) Repeatability: better than ±1 per cent of full scale of range used, or ±2 ppm, whichever is greater.
e) Stability: better than ±2 per cent of full scale of range used or ±2 ppm, whichever is greater, in a period of 1 hour.
e) Zero drift: less than ±l per cent of full scale of ra nge used or ±2 ppm . whichever is greater, in a period of 1 hour.
f) Noise: 0.5 Hz and greater, less than ±l per cent of full scale of range used or ±l ppm, whichever is greater.
g) Interferences: to be limited with respect to indicated CO concentration as follows:
1) less than 500 ppm/per cent ethylene concentration Supplemental information It is unlikely that high concentrations of ethylene will be found in gas turbine engine exhaust. The highest concentration of hydrocarbons is found at idle, corresponding to the highest concentrations of CO. If all of the hydrocarbons were ethylene (C 2H4), and the concentration were 100% of the maximum range, 5,000 ppmC, –corresponding to 2,500 ppm ethylene – the allowable interference would be less than 125 ppm, or less than 5% of the highest CO range, 2,500 ppm. Since the interference limit is in absolute terms, the relative error will increase for measurements made at less than full scale. If ethylene is suspected of being present in significant concentrations, then corrections to the data are required.
2) less than 2 ppm/per cent CO2 concentration
4) less than 2 ppm/per cent water vapour.* Supplemental information These two interferents, CO 2 and water vapor, are additive. Being the major products of combustion they increase and decrease together, and are at their highest levels at the highest power.
Unfortunately concentrations of CO tend to be at their lowest concentrations at the highest power.
This can cause significant problems in the accura cy of the measurement even if the interference limits are met. It is not unusual for tests to be conducted with the sample dried before measurement and the interference due to the remaining interferent, CO 2, compensated for through use of gas or optical filters. It does bear mentioning that the c ontribution of high power CO concentrations to the total gross CO emission measured over the LTO cycle is relatively small.
If the interference limitation(s) for CO2 and/or water vapour cannot be met, appropriate correction factors shall be determined, reported and applied.
Note.- It is recommended as consistent with good practice that such correction procedures be adopted in all cases.
* Need not apply where measurements are on a "dry" basis.
CO2 Analyzer
d) Total range: 0 to 10 per cent in appropriate ranges.
Supplemental information Although the total range specified for CO 2 is 0 to 10%, concentrations most often will vary between 1% and 5%. This range is considerably narrower than that for CO. Never the less, good practice dictates using ranges that keep the instrument response in the upper 30% of the meter scale except as noted above.
b) Resolution: better than 0.5 per cent of fu ll scale of range used or 100 ppm, whichever is greater.
c) Repeatability: better than ±1 per cent of full scale of range used or ±100 ppm, whichever is greater.
d) Stability: better than ±2 per cent of full sc ale of range used or ±100 ppm, whichever is Page 103
greater, in a period of 1 hour.
e) Zero drift: less than ±l per cent of full scale of range used or ±l00 ppm, whichever is greater, in a period of 1 hour.
f) Noise: 0.5 Hz and greater, less than ±1 per cent of full scale of range used or ±100 ppm, whichever is greater.
h) The effect of oxygen (O 2) on the CO2 analyzer response shall be checked. For a change from 0 per cent O2 to 21 per cent O2, the response of a given CO2 concentration shall not change by more than 2 per cent of reading. If this limit cannot be met an appropriate correction factor shall be applied.
Supplemental information Gas turbine engines use a considerable amount of internal cooling air that mixes with the combustion products before exiting the engine. This results in an oxygen rich exhaust sample that warrants close attention.
Note.— It is recommended as consistent with good practice that such correction procedures be adopted in all cases.
CO and CO2 Analysers
a) Response time: shall not exceed 10 seconds from inlet of the sample to the analysis system, to the achievement of 90 per cent of the final reading.
b) Sample temperature: the normal mode of operation is for analysis of the sample in its (untreated) "wet" condition. This requires that the sample cell and all other components in contact with the sample in this sub-system be maintained at a temperature of not less than 50°C, with a stability of ±2 °C. The option to measure CO and CO 2 on a dry basis (with suitable water traps) is allowed, in which case unheated analysers are permissible and the interference limits for H2O vapour removed, and subsequent correction for inlet water vapour and water of combustion is required.
Supplemental information Stability is defined in terms of a time interval whic h, because this is a temperature control set point, can be taken as the duration of the test, or one hour to be consistent with the stability limits placed on the detection system.
The temperature quoted for the CO and CO 2 subsystems, 50°C, is on the low end of the sample line specification, 65 °C ±15°C. Good practice would suggest that the subsystem temperature be approximately the same as the sample gas temperature. If the samples are dried and the analyzers unheated it would be reasonable to lower the sample temperature to that of the analyzer. If water is removed prior to analysis corrections must be applied to compensate for the loss of water of combustion and inlet water vapor. Correction proc edures are detailed in Attachment F of this Appendix.
c) Calibration curves:
j) Analysers with a linear signal output characteristic shall be checked on all working ranges using calibration gases at known concentratio ns of approximately 0, 30, 60 and 90 per cent of full scale. The maximum response de viation of any of these points from a least squares straight line, fitted to the points and the zero reading, shall not exceed ±2 per cent of the full scale value. If it does then a calibration curve shall be prepared for operational use.
Supplemental information Straight line fits to sets of linear data can be arrive d at graphically or analytically. If graphically, the results are subject to interpretation, i.e., best estima te by ‘eye’ by individual. If analytically, there is assurance that each case and all data are handled the same way each time. The technique is to perform a linear regression, or ‘least-squares fit’ for a line. The calibration gas values are the “independent” variables and are assumed to be correct, to have negligible error, for the purpose of this analysis. The instrument response values are the dependent variables and are assumed to have errors and that these errors are normally (Gaussian) distributed about the true line. The equation describing the straight line is:
yi = A + B*xi true value for response yi = A (a constant) + B (another constant) * (calibration gas value xi) For instruments which have been adjusted such that zero input results in zero output, and where the variance in known to be proportional to the reading the slope B can be shown to be equal to the ratio of the averages and can be expressed as:
y x Often A and B are not such simple values for intercept and slope but must be calculated as if the variance were not known to be proportional to the instrument response. Again if we set the instrument to read zero for zero gas input then A = 0 which makes the calculation relatively simple. The generalized expression for B can be found in any elementary statistics or error analysis text and is:
Where xi is the calibration gas value and yi is the instrument response and N refers to the number of points used in the analysis.
With N = 4, corresponding to 0, 30, 60 and 90% of full scale, this equation can be rewritten as:
A simple table, for the four sets of values, ca n be used for organizing the information, thereby simplifying the calculation:
Measurement number i x cal. gas value i y response i 2xi xiyi 1 2 3 4 N = 4 Σxi Σyi 2Σxi Σxi yi If the instrument is not set to zero-zero (zero resp onse for zero input), then A must be determined.
The equation for A (for N = 4) is:
The table shown earlier can be used to organize the elements of this equation as well.
Usually the next step would be to calculate the uncertainty in yi about this line. However, instead of controlling uncertainty about the line, ICAO chose to set an absolute limit of ±2% deviation of the full scale value for each point. This should make clear the advantage in using the upper region (top 30%) of the range for all measurements.
ii) Analysers with a non-linear signal output characteristic, and those that do not meet the requirements of linearity given above, shal l have calibration curves prepared for all working ranges using calibration gases at kn own concentrations of approximately 0, 30, 60 and 90 per cent of full scale. Additional mixes shall be used, if necessary, to define the curve shape properly.
Supplemental information For analyzers with a non-linear signal output charac teristic, calibration curves shall be prepared, again using approximately 0, 30, 60 and 90% of full scale calibration gases. If a curve is substantially non-linear in shape, it is recommen ded that additional calibration gases be used with values between the ones specified. These calibrat ion curves can be determined analytically using a least squares fit, but in this case the fit would be to a polynomial or exponential. The equations for doing this can be found in any basic text on statistics or error analysis. It should be noted that for exponential fits it is often convenient to work with the logarithm of the expression, which reduces the problem to a least-squares-fit about a line as is described above. This technique is used in analyzing smoke filters as requi red in 3. of Appendix 2. Al though not stated explicitly, the presumption is that the same ±2% of full scale response deviation is true for non-linear as well as linear instruments.
Table A5-2 summarizes the specifications for CO and CO 2 analyzers. These are typical of those analyzers offered by major analyzer manufacturers.
Table A5-2 NDIR ANALYZER PERFORMANCE SPECIFICATIONS Parameter Value CO2 CO Total Range 0 to 10% in appropriate ranges 0 to 2 500 ppm in appropriate ranges Resolution better than 0.5% fs range used or 100 ppm whichever greater better than 0.5% fs range used or 1 ppm whichever greater Repeatability better than ±1% fs range used or ±100 ppm whichever greater better than ±1% fs range used or ±2 ppm whichever greater Stability better than ±2% fs range used or ±100 ppm whichever greater period of 1 hr better than ±2% fs range used or ±2 ppm whichever greater period of 1 hr Zero Drift < ±1% fs range used or ±100 ppm whichever greater period of 1 hr < ±1% fs range used or ±2 ppm whichever greater period of 1 hr Noise ≥0.5 HZ, <±1% FS RANGE USED OR ±100
PPM, WHICHEVER
GREATER ≥0.5 Hz, <±1% fs range used or ±1 ppm, whichever greater Interference ≤ 2% of reading for O2 between 0 and 21% <500 ppm/% ethylene <2 ppm/% CO2 <2 ppm/% water vapor Response Time ≤10 seconds from instrument inlet to 90% fs ≤10 seconds from instrument inlet to 90% fs Sample Temperature wet samples ≥50°C stability ± 2°C wet samples ≥50°C stability ± 2°C Page 107
ATTACHMENT C TO ICAO MANUAL APPENDIX 5.
SPECIFICATION FOR NOX ANALYSER Note.— See attachment D for information on calibration and test gases.
1. As indicated in 5.4 of Appendix 5, the measurement of the oxides of nitrogen concentration. shall be by the chemiluininescent technique in which radiation emitted by the reaction of NO and O3 is measured. This method is not sensitive to NO2 and therefore the sample shall be passed through a converter in which NO2 is converted to NO before the measurement of total NOX is made. Both the original NO and the total NOX concentrations shall be recorded. Thus by difference, a measure of the NO2 concentration shall be obtained.
2. The instrument to be used shall be complete with all necessary flow control components, such as regulators, valves, flowmeters, etc. Materials in contact with the sample gas shall be restricted to those which are resistant to attack by oxides of nitrogen, such as stainless steel, glass, et c. The temperature of the sample shall everywhere be maintained at values, consistent with the local pressures, which avoid condensation of water.
Precautions: The performance specifications indicated are generally for analyzer full scale.
Errors at part scale may be a significantly greater percentage of reading. The relevance and importance of such increases shall be considered when preparing to make measurements. If better performance is necessary, then appropriate precautions shall be taken.
Supplemental information The performance specifications for these analyzers given in terms of full scale response can have a significant and adverse impact on part scale me asurements. This needs to be considered when planning and executing the test, and in evaluating th e accuracy of the measurements after the test.
Concentrations of NOX, when going from the idle mode to take-o ff, can differ by orders of magnitude.
In general, where concentrations of species to be meas ured are known to vary this way, it is always good practice to use a multi-range instrument, and to choose ranges such as to keep the measurement in the upper 30% of scale on the ran ge in use. A measurement made at 20% of full scale could result in an error 5 times the error spec ified as a percent of full scale. This is a general precaution. Some modern instruments with inter nal electronic ranging and calibration capability can be used over their entire range without penalty. Calibrations should be performed on each range used as required.
3. The principal performance specification, determined for the instrument operated in an ambient temperature stable to within 2 °C, shall be as follows:
a) Total range: 0 to 2 500 ppm in appropriate ranges.
b) Resolution: better than 0.5 per cent of full scale of range used or 1 ppm, whichever is greater.
c) Repeatability: better than ±l per cent of full scale of range used, or ±1 ppm, whichever is greater.
d) Stability: better than ±2 per cent of full scale of range used or ±l ppm, whichever is greater, in a period of 1 hour.