Tuneable diode laser spectroscopy correction factor investigation on ammonia measurement

Current diesel engine aftertreatment systems, such as Selective Catalyst Reduction (SCR) use ammonia (NH3) to reduce Nitrogen Oxides (NOx) into Nitrogen (N2) and water (H2O). However, if the reaction between NH3 and NOx is unbalanced, it can lead either NH3 or NOx being released into the environment. As NH3 is classified as a dangerous compound in the environment, its accurate measurement is essential. Tuneable Diode Laser (TDL) spectroscopy is one of the methods used to measure raw emissions inside engine exhaust pipes, especially NH3. This instrument requires a real-time exhaust temperature, pressure and other interference compounds in order to adjust itself to reduce the error in NH3 readings. Most researchers believed that exhaust temperature and pressure were the most influential factors in TDL when measuring NH3 inside exhaust pipes. The aim of this paper was to quantify these interference effects on TDL when undertaking NH3 measurement. Surprisingly, the results show that pressure was the least influential factor when compared to temperature, H2O, CO2 and O2 when undertaking NH3 measurement using TDL.


Introduction
Diesel engine vehicles have been very popular in the past decade, because of their comparative fuel economy. The ability of diesel engines to inject fuel dynamically in proportion to power demand during transient operations makes them ideal for heavy duty operations. Despite the lower carbon dioxide (CO 2 ) output this efficiency produces however, diesel engines generate more particulate matter (PM) and oxides of nitrogen (NO x ) (Chan et al., 1997;Majewski and Khair, 2006). NO x refers to combined nitric oxide (NO) and nitrogen dioxide (NO 2 ), which are formed at the high temperatures and pressures inside engine cylinders. NOx emissions are now at a level, in many urban areas, that are a risk to human health by causing throat, lung and eye irritations (Hsieh and Wang, 2012;Chenet al, 2012) (Zhang et al., 2013;Boubnovet al, 2014). NO x also contribute to photochemical smog, ozone and acid rain. Therefore, most diesel engine vehicles must be fitted with aftertreatment systems to reduce hazardous exhaust emissions. Selective catalyst reduction (SCR) is the most effective and common treatment method based on reduction of the NO x group (Majewski and Khair, 2006). SCR can reduce NOx produced from a diesel engines by 90% (Chiang et al., 2010). SCR uses ammonia (NH 3 ) to reduce the NOx into nitrogen (N 2 ) and water (H 2 O). NH 3 as a gas is difficult to handle and normally it is generated through decomposition of injected urea-water solution upstream of the SCR. However, if NH 3 is in excess through the SCR, it can lead to hazardous NH 3 being released into the environment, which is referred to as "NH 3 slip" (Suarez-Bertoaet al, 2015). NH 3 has a toxicity limit of 20 mg/m 3 and is very reactive, both with other chemicals and the materials around it (Suarez-Bertoaet al, 2015). Moreover, it contributes to particulate matter (PM) from precipitation, often in the form of ammonium nitrate and ammonium sulphate (Vaittinen et al., 2013;Huai et al., 2005), which is referred to as secondary pollution. NH 3 and particulates, as with other nitrogen compounds, cause respiratory and cardiovascular problems (Huai et al., 2005). Therefore current European standards for EURO V and Stage V engines require NH 3 to be limited to less than 10 ppm for every tailpipe concentration over the test cycle ((EC) No 582/2011) (EU Commission, 2017).
TDL (Tuneable Diode Laser) is one of the methods that can be used to measure raw NH 3 . Most TDLs contain transmitter and receiver units mounted on opposite sides. Some TDL measure gas inside the sample cell which in the analyser. For this paper used type of TDL that fitted inside the exhaust pipe and measure NH 3 directly inside the engine exhaust pipe. Which means that both transmitter and receiver units are The laser light from a TDL is created at a specific wavelength in the near spectral region via a diode crystal, typical wavelength near 1.512 μm for NH 3 (Huai et al., 2005) (Pisano et al., 2009;Choi et al., 2016). Tuned laser conforms to the Beer Lambert law, which means temperature and pressure can affect the spectral absorbance and effect the light intensity of the TDL (Pisano et al., 2009;Choi et al., 2016;Lenaers and Van Poppel, 2005). Therefore, most TDLs require real-time, temperature and pressure correction for the NH 3 spectrum. TDLs are also able to measure other interferences, such as H 2 O, CO 2 and O 2 inside the exhaust to allow further correction of NH 3 measurements. Some TDLs can manually adjust the temperature, pressure and the other interference compounds, which otherwise have a significant effect on NH 3 measurement. Most TDL contain two units (transmitter and receiver units) that can be contaminated by exhaust gas over time which led to reduction of the measurement accuracy of the device, this can be stop by regular cleaning and maintain the units.
Most researchers report that temperature and pressure have the greatest influence when measuring NH 3 using a TDL. However, the literature does not include numerical values to quantify the extent of exhaust temperature, pressure and other interferences on NH 3 readings using TDL directly inside the exhaust pipe, because this require high cost to undertake this type of experimental test. Therefore, this paper investigates the effect of temperature, pressure and other compounds on the TDL correction factor for NH 3 measurement directly inside exhaust pipe by manually adjusting these factors within the TDL software, which means that those are not the real-time values of temperature, pressure, H 2 O, CO 2 , and O 2 within the exhaust. Table 1 represents the specification of the TDL used in this experimental, in which the temperature and pressure were measured in realtime from the exhaust pipe then feeding into the TDL. H 2 O, CO 2 , and O 2 have also been measured by TDL in order to do correction on the NH 3 measurements. The transmitter and receiver units of TDL were mounted on opposite sides inside the exhaust pipe.

Experimental set up
This particular TDL can manually adjust the temperature, pressure and other interferences. Table 2 shows the range of correction factors used in this experiment. The Taguchi Orthogonal Array method was therefore used to design an experiment from those ranges. This is a method that allows the selection of a subset of combination of multiple factors from multiple levels and also balanced to ensure that all the levels of factors are considered equally (Taguchi Orthogonal Array Designs, 2012). Each correction factor was divided into five equal portions of the range as in Table 2; as a result, 25 experimental tests were generated based on the Taguchi L25 array (Table 3). Fig. 1 is a schematic of the test rig and Table 4 show the engine specification used. The engine was run in a steady state test, which was 2200 rpm and 190 N.m for engine speed and load. The aftertreatment system used in this test was the DOC (Diesel Oxidation Catalyst) and SCR. The NH 3 slip from the aftertreatment system was detected in-situ (approximately 50 ppm), with both the receiver unit and transmitter unit located inside the exhaust. The real-time measurements of temperature, pressure and the interference compounds in the exhaust pipe were recorded and fed into the TDL before and after the test. This was applied in order to measure correct NH 3 readings for comparison with NH 3 readings from manually adjusted temperatures, pressure and other interference compounds. The real-time values for temperature, pressure and other interference compounds are presented as start control and end control in Table 3. The tests were conducted as follows: 1. Temperature and pressure were measured using external sensors in the exhaust pipe. H 2 O, CO 2 , and O 2 were recorded using the TDL in the exhaust pipe at the start of the test (Start Control), as shown in Table 3. NH 3 readings were recorded three times with about 1 min intervals between each measurement. 2. The temperature, pressure, H 2 O, CO 2 , and O 2 were then manually applied into the TDL for each experiment according to Table 3 and three NH 3 readings taken at 1 min intervals as before. Table 3 were carried out in a similar way with the triplicate NH 3 values recorded for each experiment. 4. At the end of the experiments, the temperature and pressure were re-measured using an external sensor in the exhaust pipe. H 2 O, CO 2 , and O 2 were also re-measured using the TDL (End Control), as shown in Table 3.

Data analysis and discussion
After the NH 3 readings were recorded three times as show in Table 5, it is clear that temperature, pressure and other interference compounds can have a huge effect on the TDL during NH 3 measurement. The correct NH 3 readings for the start control and end control are   N. Li et al. Atmospheric Environment 172 (2018) 12-15 about 55.20 ppm average. However when the temperature, pressure and other interference compounds been adjusted manually, this lead to NH 3 readings completely changed and differed from the correct NH 3 readings. This is similar to suggestions by Pisano et al. and Lenaers at al (Pisano et al., 2009;Lenaers and Van Poppel, 2005), that at high temperature and pressure, the TDL spectral line shape can be broadened, which can affect the measurement intensity. The same authors also mention that water concentration can affect the TDL on NH 3 readings, so when the amount of water concentration is fed into the TDL, for the instrument to correct itself for NH 3 measurement during water interference by selecting the right spectral line, that because water can have huge absorption on the infrared spectrum. The greatest range calculation method from Taguchi Orthogonal Array was used to determine the most influential factor on a TDL during NH 3 measurement. The method first calculates the SNi (Signal-to-noise ratio) value for each experiment and then averages the SNi value for each factor in order to determine the highest rank and influence on NH 3 measurement using a TDL. The equation below represents the SNi calculation: where: , i i = recorded number u = trail number Ni = number of trials for recording number i Table 5 represents the NH 3 readings, including the SNi value for each experiment. It also contains the correct NH 3 readings at the start control and end of control. By conducting the data analysis using the Taguchi Orthogonal Array show in Table 6, the average SNi for each level of every factor, which found that exhaust temperature has the highest influential correction factor on NH 3 measurement, which was similarly suggested by authors such Pisano et al., Choi at al and Lenaers at al (Pisano et al., 2009;Choi et al., 2016;Lenaers and Van Poppel, 2005). The next most influential factors are O 2 and H 2 O, which are second and third, followed by CO 2 is fourth most influential. However, it is surprising that pressure is shown to be the least influential factor, as it does not have a huge effect on the TDL during NH 3 measurement, despite most previous research believing pressure is the second most influential factor (after temperature) on a TDL during NH 3 measurement. The reason is that the TDL software does not take pressure values into account during the measurement of NH 3 . Those influence factor can affect the spectral absorbance and light intensity of the TDL during NH 3 measurement, but does not shows the amount of been effected, the   only way to see the effect is through the NH 3 reading. Generally, in order to achieve a good NH 3 reading using a TDL, real-time exhaust temperature, and pressure, H 2 O, CO 2 and O 2 must be fed into the TDL.

Conclusion
This main aim of the experimental tests were to determine the extent of the influence from temperature, pressure, H 2 O, CO 2 and O 2 on NH 3 measurements directly inside the exhaust pipe using TDL. Those influence factor can affect the spectral absorbance and light intensity of the TDL during NH 3 measurement.
The tests show a huge effect on the TDL during NH 3 measurement, although those values for temperature, pressure and other interferences compounds are not exactly or not real-time from the exhaust pipe, as show in Table 5.
The tests shows that temperature was the most influential correction factor, followed by O 2 and H 2 O. Meanwhile CO 2 and pressure can be negligible, as both have the lowest influence. These results show that temperature, O 2 and H 2 O can have a huge effect on the TDL spectrum line because the line shape can be broadened which affects the measurement intensity. Most TDL instruments adjust by compensating for these interferences, but if the values for those factors are not taken into account real-time, this still leads to errors in the TDL readings for NH 3 . Overall, in order to achieve better and accurate results for NH 3 readings inside exhausts pipe using a TDL is by feeding in correct or real-time exhaust temperature, pressure, and other interference compounds present inside the exhaust.