7.7.2. Engine Load and Emission Correlation

7.7.2.1. Emission Correlation Methods

Establishment of an aircraft emissions inventory for a given flight traffic scenario requires a knowledge of the engine's emissions over the aircraft's total flight mission. This inventory is relatively straightforward for CO2, H2O, and SOxO (in total) emissions because of their direct link with mission fuel burn. For several other emissions that depend on engine power setting, combustor design, and flight condition, correlation methods have been developed to calculate the exhaust emissions for a specific engine type using measured data taken during the engine certification process. This correlation is carried out primarily with species of emissions such as NOx, CO, HC, or soot by means of a semi-empirical correlation between emissions and principal combustion parameters, using measurement programs involving combustor rigs and engine systems together with theoretical considerations of the main combustion processes. Exhaust emissions production processes generally are complex because they involve unsteady physical processes as well as non-equilibrium chemical processes. A fundamental element in the development of formulas to correlate measured and predicted emission indices is a model of the process based on the relationship of the emission index, chemical kinetic rates, and residence time in the reaction zone. This model is of great value to manufacturers wishing to predict the emissions performance of a new or development combustor. For an existing design, a reference correlation method is used to predict emissions on the basis of measured data. Because combustor inlet pressure (p3) and temperature (T3) are the main parameters involved. For EI(NOx), the "p3/T3" method leads to a relationship that, in its simplest form, is as follows:

EI(NOx)/EI(NOx)ref = (p3/p3ref)ⁿ f(Humidity) for T3 = T3ref where n is an exponent in a range between about 0.3 and 0.6 derived from engine and combustor rig testing.

The parameters p3 and T3 come either from measurements or from computational engine simulation. Measurement programs, especially for NOx emission indices, reveal agreement (compared with testing) of better than 5% (AERONOx, 1995; Brasseur et al., 1998). Emissions of HC and CO depend on the completeness or efficiency of the combustion process. Test data are well correlated using a combustor loading parameter, which takes account of the residence time of the burning products and reaction time in the combustor reference volume. As an example, Figure 7-30 shows typical functions of emission indices of NOx and CO versus engine load for a turbofan engine at different altitudes on the basis of a thermodynamic engine simulation. Unburned hydrocarbons will follow the same trend as CO, but on a much lower level. Both are products of incomplete combustion; thus, they are controlled by the same mechanism.

Despite the intrinsic complexity of the chemical processes associated with soot production, a semi-empirical correlation between calculated and measured SN has been developed with a 40% standard deviation (De Champlain et al., 1997). By also considering reaction kinetics, Döpelheuer (1997) developed a correlation method based mainly on the concentration of carbon and oxygen in the combustion zone. The concentration of soot is considered to be proportional to the equivalence ratio, to pressure at the combustor inlet (both with exponents to be determined), and to an exponential function controlling the reaction rate. To apply the method as a reference type of correlation, soot density as a function of measured SN (Finch and Eyl, 1976; Whyte, 1982; Champagne, 1988; Hurley, 1993) has been approximated as a first step. Up to a 20% deviation exists between different correlations of soot loading and SN.

Because of the limitations of the standard measuring method for SN (i.e., by collecting soot on a white filter paper and evaluating reflected light intensity) (ICAO, 1993), only the major particles of soot are measured; the remaining wide spectrum of very small particles are not measured (see Section 7.5.3).

7.7.2.2. Simplified Alternate Correlation Methods

The methods described above to correlate engine emissions for different operating conditions all require engine internal gas path data. Such data are normally sensitive from an engine manufacturer's point of view. Alternate correlation methods that do not expose proprietary or sensitive engine data have therefore been sought. These methods are based mainly on emission correlation with the fuel flow of an engine. Thus, SLS engine test data from certification testing, together with relevant methods of correlating emissions with engine performance in flight, is favored as one of the principal sources for the development of aircraft emission inventories. These methods are designed to use unrestricted data from the ICAO engine emission databank coupled with fuel flow data, which can easily be acquired from aircraft missions.

Several methods based on the above principles have been developed and are undergoing continuing improvements. The key element of these simplified methods is the assumption that emission indices at different engine inlet conditions might be correctable to a reference day condition, thus collapsing into a single function of the corrected fuel flow (Lecht and Deidewig, 1994; Martin et al., 1994, 1995; Deidewig et al., 1996). An example of the effectiveness of such a method developed for NOx emissions is outlined in Figures 7-31 and 7-32. Figure 7-31 shows actual NOx emission indices according to ambient flight conditions and calculated with a complex correlation formula; Figure 7-32 gives the result of the same values but corrected for ISA SLS conditions. For comparison, ICAO LTO cycle test data are marked separately.

This method allows ISA SLS measurements to act as a reference function; this function simply has to be re-corrected for actual in-flight engine inlet conditions. These simplified emission correlation methods are highly relevant in terms of their value for the further development of aircraft emission inventories (see Chapter 9). There is, as yet, no fuel flow-based correlation that can be used to predict engine soot for the purposes of inventory preparation because of the use of SN data in the ICAO emissions databank, which cannot readily be converted to soot loading (Döpelheuer and Lecht, 1999).

7.7.3. Validation of Emission Prediction Methods

7.7.3.1. Altitude Chamber Testing

Cruise-level emission index prediction methods need validation by measurement. Such validation can be carried out at ground-level test chambers in which pressure and temperature can be varied to simulate a wide range of engine operation conditions in flight. Within the AERONOx project, exhaust emissions of two selected engines (Rolls Royce RB211 and Pratt & Whitney PW305) have been analyzed and compared with predictions (Lister et al., 1995). This comparison shows that emission prediction methods developed by engine manufacturers and research institutes can predict the NOx emission at a flight condition within an error band of 5-10% for a modern high bypass engine. These experiments found a tolerance for fuel-based methods that was nearly as good. Measurements from the AERONOx project also showed that for highest accuracy, the prediction equations must be adapted for specific engine type. A kind of indirect validation has been undertaken by comparing NOx emissions of fuel flow-based methods with the p3/T3 method. Such an evaluation showed an agreement within 13% maximum and 6% standard deviation for a variety of aircraft and flight missions (ICAO, 1995c).

7.7.3.2. Validation by In-Flight Measurement

An alternate suitable method is to carry out measurements in the freshly emitted exhaust plume of an aircraft in flight. Several such cruise altitude emissions measurement programs have been conducted. Cruise altitude in situ chemical probing was carried out for the first time in December 1991, when the DLR research aircraft "Falcon" flew through the plume of a commercial DC-8 airliner (Arnold et al., 1992). In 1993, the first in-flight measurements of emission indices were accomplished, again using the "Falcon" (Schulte and Schlager, 1996). This measurement involved measuring CO2 simultaneously with other species of interest (in terms of volume mixing ratios). A detailed discussion of these assumptions and the emission index determination from the measured ratios D[X]/D[CO2] appears in Schulte and Schlager (1996).

In recent years, several in situ EI measurements have been carried out. Fahey et al. (1995a,b) determined emission indices for NOx, CO, and N2O in the exhaust plumes of the NASA ER-2 research aircraft and the Concorde in the lower stratosphere. Haschberger and Lindermeir (1996, 1997) analyzed the exhaust plume of the DLR experimental Advanced Technology Testing Aircraft System of DLR, a two-engine jet aircraft of type VFW 614 (ATTAS) with onboard instruments and derived emission indices for NOx, CO, and H2O. The subsonic long-range jet aircraft types that dominate global air traffic were investigated by Schulte et al. (1997) to derive NOx emission indices. Figure 7-33 shows a comparison of all available in situ measured EI(NOx) values with corresponding predicted values. Note that the measurement details were different for each case.

In summary, these measurements revealed good agreement between predicted and measured NOx emission indices. Special attention was paid to predictions based on fuel flow methods (Lecht and Deidewig, 1994; Deidewig et al., 1996; Martin et al., 1996; Schulte and Schlager, 1996; Schulte et al., 1997) because they are the methods of choice for the building of aircraft NOx emission inventories. These methods seem to underestimate EI(NOx) by about 12% on average. However, this 12% deviation is within the uncertainties-indicated as error bars-of the measurements and the predictions.

7.7.4. Engine Emission Variations

Engine emissions may vary for a number of reasons. Manufacturing differences, aging characteristics for individual engines, operational and atmospheric conditions, and changes in fuel contents all have parts to play. Data from new engine certification testing reveals a standard deviation for EI(NOx) of 1-7%. Fuel type, fuel content, and sampling methods have only a slight influence on NOx emission variation (Lyon et al., 1980; Lukachko and Waitz, 1997). Another major source of variability may be engine deterioration over a long period of time. This deterioration is evident in measurements of engine parameters such as exhaust gas temperatures and fuel consumption rates. Economic considerations recently led to allowable SFC limits of between 2 and 4% because of engine deterioration; exceedances lead to engine overhaul. Measurements of overhauled engines revealed a similar standard deviation for NOx based on the LTO cycle (Lister and Wedlock, 1978), which implies that in emissions terms their performance is similar to new engines. Lukachko and Waitz (1997) investigated the influence of an ongoing degradation process resulting from aging. In a combined sensitivity study, they found that a 3% SFC increase from deterioration led to a -1 to +4% change of NOx emission efflux, depending on the engine part mostly affected by the deterioration. This study was conducted under cruise operating conditions. New consideration is being given by ICAO/CAEP to the development of a more appropriate certification methodology in terms of emissions variability over the entire flight cycle.

IPCC Homepage