Experimental Study of the Inﬂuence of Natural Gas Constituents on CO Emission from Chinese Gas Cooker

: In China, it has become a more common practice to introduce natural gases from di ﬀ erent sources into the same distribution system to improve supply security and reliability. Variable gas constituents may cause a negative impact on the performance of domestic gas appliances. This paper aims to study the CO emission of a Chinese gas cooker under di ﬀ erent constituents of natural gas. A typical Chinese gas cooker with two burners, each of which has a nominal heat input of 3.8 kW, was selected. One of the burners was modiﬁed to a forced-mixed mode to replace primary air injection. Within operational ranges corresponding to the permissible Wobbe index—namely, primary air coe ﬃ cients and heat inputs—the equivalence between original gas and the CH 4 / C 3 H 8 / N 2 three-component mixture in terms of CO emission was experimentally validated. Then, di ﬀ erent three-component mixtures were input into the other unmodiﬁed burner, which operates under injected primary air, to investigate how the CO emission changed with di ﬀ erent gas constituents. It was found that the CO emission of a natural gas and a CH 4 / C 3 H 8 / N 2 three-component mixture, in terms of CO emission, were equivalent. The combination of the two indexes, W and PN , can describe the CO emission from a gas cooker accurately. By means of a three-component mixture, the empirical formula, which can correlate CO and the gas property parameters, was proposed. A set of equal-CO lines was revealed for a given initial primary air adjustment. Finally, a feasible approach to manage gas quality management in China was put forward, and the conclusion can help control the CO emission of gas cookers and improve indoor air quality.


Introduction
The Chinese gas industry witnessed an unprecedented rapid increase over the past decade, with annual consumption soaring to 237.3 billion cubic meters (BCM) in 2017 in comparison with 60 BCM in 2007. It was estimated that annual consumption would account for up to 600 BCM by 2025 [1]. In order to improve supply security and reliability, more and more provincial and cities' network operators began to introduce gases from different sources, including pipeline natural gas (PNG), offshore gas, liquefied natural gas (LNG), coal-based synthesized natural gas, or even bio-natural gas into the same distribution network [2]. However, variable gas constituents may have a negative impact upon end-users' equipment.
The domestic gas cooker ranks first in terms of popularity, and annual production was maintained to be 30 million in recent years [3]. CO emission from a gas cooker when fueled with a specific gas depends upon two kinds of factors: structural parameters such as port shape, burner clearance the influence of W and PN upon CO emission, the input mixture was deliberately designed to remain constant W or constant PN. It is concluded that the combination of the two indexes W and PN can calculate the CO emission from a gas cooker. By means of a three-component mixture, the empirical formula that can correlate CO and parameters describing gas properties, namely W and PN, was established. Finally, a set of equivalent CO curves related to gas quality indexes was derived and experimentally validated. A method of gas quality management in China based upon the W-PN diagram was discussed on this basis. The conclusion can be used to predict the CO emission of a gas cooker under different gas constituents and to control the quality of indoor environments.

Dutton's Approach
Dutton put forward a three-component mixture approach to simplify experiments for interchangeability research. A gas consisting of CH 4 , C 2 H 6 , C 3 H 8 , C 4 H 10 , N 2 , and CO 2 was equivalently reduced to a mixture of CH 4 , C 3 H 8 , and N 2 only, which was termed the "equivalent mixture". He set up a diagram with the PN number (volumetric percentage of C 3 H 8 plus N 2 ) as the abscissa and the Wobbe index (W) as the vertical coordinate. The equivalent mixture was input into some gas appliances to determine the boundary limits beyond which unstable combustion phenomena such as lift, sooting, CO emission, etc. could occur. Therefore, Dutton established a CO limit for water heaters, sooting limit for infrared radiators, and lift limit for on-top cookers through elaborately designed experiments. The gas quality system following his approach remains valid in Britain today [20]. The principle for equivalent mixture was different for different unstable combustion phenomena. For the CO emission of a water heater, two conditions should be met. First, the ratio of C/H atoms for a "three-component mixture" should be equal to that of the original gas. For example, C 2 H 6 = 0.5 CH 4 + 0.5 C 3 H 8 , and C 4 H 10 = 1.5 C 3 H 8 − 0.5 CH4. Second, the mixture should have the same Wobbe index as that of the original gas. Dutton's approach significantly reduced the number of experiments and much was easier to follow in comparison with the Delbourg diagram, AGA indexes, etc. [21][22][23].
For an original gas containing CH 4 , C 2 H 6 , C 3 H 8 , C 4 H 10 , N 2 , and CO 2 , the volumetric percentage of which is expressed as r 1 , r 2 , r 3 , r 4 , r 5 , and r 6 , respectively, the percentage of CH 4 , C 3 H 8 , and N 2 can be expressed as x 1 , x 2 , and x 3 . Mathematically x 1 , x 2 , and x 3 can be determined by solving Equations (1)-(3): x 1 x 2 = r 1 + 0.5r 2 − 0.5r 4 0.5r 2 + r 3 + 1.5r 4 (1) where H hi is the higher calorific value for individual constituent, MJ/m 3 ; and s i is the specific gravity of an individual constituent. For a water heater, Dutton measured CO emissions changing with heat inputs when operated at 100-120% nominal heat input. He found that the slopes for different gases tend to be the same, implying a group of parallel lines on the ln(CO)-Q diagram. Therefore, he defined the ICF (incomplete combustion factor) [21], shown in Equation (4), as the average distance between a specific gas and 12T-0. This index can be used to predict the CO emission of a water heat with the change of gas constituents, as shown in Equation (5): where ICF is the incomplete combustion factor; W is the Wobbe index, MJ/m 3 ; PN is the volumetric percentage of C 3 H 8 plus N 2 of the three-component mixture; E CO is CO emission under a certain gas, ppm; and E CO-CH4 is CO emission under reference gas, methane, ppm. Some obvious differences exist between water heaters and gas cookers, if measured in terms of structure and combustion method involved. For gas cookers, the flame always directly contacts the cold surface being heated, while for water heaters, a combustion chamber is usually provided for the combustion process to fulfill. More importantly, initial adjustment is essential for a gas cooker to provide suitable primary air and flame length accordingly. Therefore, a gas cooker may operate under different heat input and primary air coefficients, while primary air variation is seldom considered when predicting the performance of water heater operation under different gas constituents. Whether Dutton's approach could be adopted for Chinese gas cookers today depends heavily upon whether an equivalent mixture corresponding to a specific original gas constituent can reproduce the same CO emission.

Test Rig
The gas appliance to be tested was a typical embedded-type cooker with two burners, each of which has a nominal heat input of 3.8 kW. The numbers, diameters, and arrangement of ports for the two burners were the same. The atmospheric burner on the left side remained unchanged, and primary air was delivered by an injector, as shown in Figure 1a. Corresponding to a specific initial primary air, the injector would automatically adjust to another primary air when operated under another gas constituent. The CO emission characteristics were experimentally studied on the left-side burner. The atmospheric burner on the right side was modified to a forced-mixed mode, as illustrated in Figure 1b. The primary air to the burner was supplied by compressed air. The gas flow rate and air flow rate can be independently controlled and metered. The measurement to determine "equivalence" between the original gas and the three-component mixture was done on the modified burner. where ICF is the incomplete combustion factor; W is the Wobbe index, MJ/m 3 ; PN is the volumetric percentage of C3H8 plus N2 of the three-component mixture; ECO is CO emission under a certain gas, ppm; and ECO-CH4 is CO emission under reference gas, methane, ppm. Some obvious differences exist between water heaters and gas cookers, if measured in terms of structure and combustion method involved. For gas cookers, the flame always directly contacts the cold surface being heated, while for water heaters, a combustion chamber is usually provided for the combustion process to fulfill. More importantly, initial adjustment is essential for a gas cooker to provide suitable primary air and flame length accordingly. Therefore, a gas cooker may operate under different heat input and primary air coefficients, while primary air variation is seldom considered when predicting the performance of water heater operation under different gas constituents. Whether Dutton's approach could be adopted for Chinese gas cookers today depends heavily upon whether an equivalent mixture corresponding to a specific original gas constituent can reproduce the same CO emission.

Test Rig
The gas appliance to be tested was a typical embedded-type cooker with two burners, each of which has a nominal heat input of 3.8 kW. The numbers, diameters, and arrangement of ports for the two burners were the same. The atmospheric burner on the left side remained unchanged, and primary air was delivered by an injector, as shown in Figure 1a. Corresponding to a specific initial primary air, the injector would automatically adjust to another primary air when operated under another gas constituent. The CO emission characteristics were experimentally studied on the left-side burner. The atmospheric burner on the right side was modified to a forced-mixed mode, as illustrated in Figure 1b. The primary air to the burner was supplied by compressed air. The gas flow rate and air flow rate can be independently controlled and metered. The measurement to determine "equivalence" between the original gas and the three-component mixture was done on the modified burner. Compared to the left-side original burner, only the primary air supply was changed. The resulting flame shape and resistance of the downstream burner were not influenced. Through modification, the primary air and the heat input can be adjusted artificially to simulate the possible working condition of an injector. However, the combustion process would not be affected.
The test rig was designed strictly according to Chinese National Standard GB 16410-2007 [24]. As shown in Figure 2, three sub-systems were included: a continuous gas-blending sub-system, flue gas analysis sub-system, and gas cooker performance measurement sub-system.
The continuous gas-blending sub-system consists of five sets of mass flow rate controllers (MFC), four of which are fed by high-pressure gas cylinders, and one of which relates to pipeline Compared to the left-side original burner, only the primary air supply was changed. The resulting flame shape and resistance of the downstream burner were not influenced. Through modification, the primary air and the heat input can be adjusted artificially to simulate the possible working condition of an injector. However, the combustion process would not be affected.
The test rig was designed strictly according to Chinese National Standard GB 16410-2007 [24]. As shown in Figure 2, three sub-systems were included: a continuous gas-blending sub-system, flue gas analysis sub-system, and gas cooker performance measurement sub-system. accurately measure the gas flow rates of the outer ring and the inner ring of the gas cooker. Compressed air was introduced to provide the primary air. It can also be controlled and metered through MFC and wet-type gas meters. The tested cookers with a round-port burner represent the typical structures of cookers in the Chinese market. There was a vacuum pump inside the flue gas analyzer. The flue gas was sampled through the sampling ring. The concentrations of the combustion product components are measured by the flue gas analyzer. The tested data are recorded by the computer including O2, CO2, and CO. The instrumentation involved in the test is listed in Table 1.

Test Gas
Listed in Table 2 are tested original gas constituents and their equivalent mixtures, where the former are coded as E and N, and the latter are coded as ET and NT. The continuous gas-blending sub-system consists of five sets of mass flow rate controllers (MFC), four of which are fed by high-pressure gas cylinders, and one of which relates to pipeline natural gas (PNG), which is available in the lab. The gas constituents to be tested and their flow rates can be preset and controlled through the computer. Two wet-type gas meters were added to accurately measure the gas flow rates of the outer ring and the inner ring of the gas cooker. Compressed air was introduced to provide the primary air. It can also be controlled and metered through MFC and wet-type gas meters. The tested cookers with a round-port burner represent the typical structures of cookers in the Chinese market. There was a vacuum pump inside the flue gas analyzer. The flue gas was sampled through the sampling ring. The concentrations of the combustion product components are measured by the flue gas analyzer. The tested data are recorded by the computer including O 2 , CO 2 , and CO. The instrumentation involved in the test is listed in Table 1.

Test Gas
Listed in Table 2 are tested original gas constituents and their equivalent mixtures, where the former are coded as E and N, and the latter are coded as ET and NT.

Test Procedure
First, measurements were made to determine the "equivalence" between original gases and three-component mixtures.
The CO emission of a gas cooker changes with the heat input and primary air coefficient. According to GB/T13611-2006 [25], the permissible W fluctuates between 90%-110% that of CH 4 . Therefore, the heat input range to be tested was set to be 90%-110% of nominal heat input, and the primary air coefficient range to be tested was 0.4-0.6. Apparently, such a range was broad enough to cover all the possible operation points of a gas cooker.
For a specific primary air, heat input of a gas-blending sub-system was preset through a computer. The flow rates of primary air and gas were metered by MFC and a wet-type gas meter, respectively. The precise primary air coefficient was determined by comparing detailed constituents of a gas-air mixture with that of gas by means of gas chromatography. When combustion became stable, a flue gas analyzer began to sample for 10 min at a frequency of 1 Hz. Then, CO emission was repeatedly measured for another primary air. Afterwards, the above-mentioned procedures were repeated for other heat inputs, and CO changing with primary air and heat input can be determined.
Secondly equivalent CO lines were measured on the left-side atmospheric burner. 12T-0 was input into the burner to make an initial adjustment, as follows: To maintain gas pressure at the cooker inlet of 2 kPa so that the burner operated under a nominal heat input (3.8 kW); and to adjust the air shutter so that a suitable flame shape can be achieved. Then, we recorded the flue gas components. Afterwards, different three-component mixtures were successively input into the burner, and the flue gas was analyzed.

Uncertainty Analysis
The uncertainties come from two respects: uncertainty of instruments, and uncertainty of measured data. As to the calculation of instrument uncertainty, the precision of the instruments involved is listed in Table 1. The uncertainties come from the flow-rate measurement of gas and primary air by the wet flow meter, the temperature and pressure measurement, and the CO measurement by the flue gas analyzer. As for the calculation of measured data uncertainty, the data were measured several times, and the standard deviations were derived to calculate the uncertainty [26].
The relative uncertainties can be calculated according to equation given by Ref. [25], and the results are shown in Table 3.

Validation of Dutton's ICF Index
When gas constituents change, the operation of gas cookers, namely the primary air coefficient α and heat input Q, will be changed from (α ,Q) 1 to (α ,Q) 2 . When investigating if a three-component mixture can reproduce the same amount of CO as that fueled with original gas containing CH 4 , C 2 H 6 , C 3 H 8 , C 4 H 10 , C 5 H 12 , CO 2 , and N 2 , the "equivalence" should be evaluated for all possible primary air coefficients and heat inputs corresponding to gas constituents' variations. Hence, the relationship between CO emission and heat input, together with the primary air coefficient of a gas, was studied first.
For different gas constituents, CO emission was found to change in an exponential pattern. Figure 3 shows the relationship between ln(CO) emissions from E13/ET13 and reference gas CH 4 under different primary air coefficients as an example. The "equivalence" of some groups of E13/ET13 test points is not satisfactory. Inevitably, there are some errors existing in all the measured data. To minimize the influence of these errors, it is more reasonable to build up a function and check the trend of CO changing with α and Q than to compare measured COs directly. When comparing CO curves of E13/ET13, it can be found that the two functions tend to overlap each other. Both E13/ET13 have similar emissions under the same working conditions (α ,Q). Intuitively, the distance between the E13/ET13 function and CH 4 function increases gradually with the increasing primary air coefficient. Due to the difference in the combustion equipment investigated, Dutton did not need to consider the effect of the primary air coefficient. For the water heaters in Dutton's research, a unique ICF number could be calculated for each three-component mixture and used in the gas quality management. However, for the Chinese cooker, a different ICF number can be calculated under a different primary air coefficient. Table 4 shows the comparison of Dutton's ICF index with the calculated ICF based on experimental data. Hence, the ICF index cannot be applied to the Chinese gas cooker discussed herein. measurement by the flue gas analyzer. As for the calculation of measured data uncertainty, the data were measured several times, and the standard deviations were derived to calculate the uncertainty [26]. The relative uncertainties can be calculated according to equation given by Ref. [25], and the results are shown in Table 3.

Validation of Dutton's ICF Index
When gas constituents change, the operation of gas cookers, namely the primary air coefficient α' and heat input Q, will be changed from (α',Q)1 to (α',Q)2. When investigating if a three-component mixture can reproduce the same amount of CO as that fueled with original gas containing CH4, C2H6, C3H8, C4H10, C5H12, CO2, and N2, the "equivalence" should be evaluated for all possible primary air coefficients and heat inputs corresponding to gas constituents' variations. Hence, the relationship between CO emission and heat input, together with the primary air coefficient of a gas, was studied first.
For different gas constituents, CO emission was found to change in an exponential pattern. Figure 3 shows the relationship between ln(CO) emissions from E13/ET13 and reference gas CH4 under different primary air coefficients as an example. The "equivalence" of some groups of E13/ET13 test points is not satisfactory. Inevitably, there are some errors existing in all the measured data. To minimize the influence of these errors, it is more reasonable to build up a function and check the trend of CO changing with α' and Q than to compare measured COs directly. When comparing CO curves of E13/ET13, it can be found that the two functions tend to overlap each other. Both E13/ET13 have similar emissions under the same working conditions (α',Q). Intuitively, the distance between the E13/ET13 function and CH4 function increases gradually with the increasing primary air coefficient. Due to the difference in the combustion equipment investigated, Dutton did not need to consider the effect of the primary air coefficient. For the water heaters in Dutton's research, a unique ICF number could be calculated for each three-component mixture and used in the gas quality management. However, for the Chinese cooker, a different ICF number can be calculated under a different primary air coefficient. Table 4 shows the comparison of Dutton's ICF index with the calculated ICF based on experimental data. Hence, the ICF index cannot be applied to the Chinese gas cooker discussed herein.

Validation of Three-Component Mixture
Within the operation range (heat input Q and primary air coefficient α') investigated, the CO emission increases exponentially with heat input Q and decreases with the primary air coefficient α'. Three-component mixtures tend to give similar CO emissions. Furthermore, comparison between CO fueled by the other five sets of original gases and their three-component mixtures can lead to similar conclusions. To quantitatively evaluate the difference between CO emission fueled by an original gas and that by a three-component mixture under different heat input and primary air, a characteristic curved surface can be configured as shown in Equation (6). Given a test gas, such a surface can be fitted by 30 measured data. For different gas constituents, the coefficient matrix A is different. Figure  4 is a three-dimensional view of CO under the three-component mixture ET8 and CO under the original gas E8. It can be found that the maximum CO appears at highest heat input and minimum primary air, while the minimum CO appears at the lowest heat input and maximum primary air. Meanwhile, it can be found that the CO surface for original gas and its corresponding threecomponent mixture are comparatively close to each other.
where CO is the CO emission of the gas cooker, ppm; Q is the heat input of the cooker, kW; and α' is the primary air coefficient.

Validation of Three-Component Mixture
Within the operation range (heat input Q and primary air coefficient α ) investigated, the CO emission increases exponentially with heat input Q and decreases with the primary air coefficient α . Three-component mixtures tend to give similar CO emissions. Furthermore, comparison between CO fueled by the other five sets of original gases and their three-component mixtures can lead to similar conclusions. To quantitatively evaluate the difference between CO emission fueled by an original gas and that by a three-component mixture under different heat input and primary air, a characteristic curved surface can be configured as shown in Equation (6). Given a test gas, such a surface can be fitted by 30 measured data. For different gas constituents, the coefficient matrix A is different. Figure 4 is a three-dimensional view of CO under the three-component mixture ET8 and CO under the original gas E8. It can be found that the maximum CO appears at highest heat input and minimum primary air, while the minimum CO appears at the lowest heat input and maximum primary air. Meanwhile, it can be found that the CO surface for original gas and its corresponding three-component mixture are comparatively close to each other.
where CO is the CO emission of the gas cooker, ppm; Q is the heat input of the cooker, kW; and α is the primary air coefficient.
To quantitatively assess how "equivalently" CO emission from an original gas can be reproduced by a three-component mixture, a parameter I can be defined as the mathematical expectation of the absolute value of difference between two functions, as shown in Equation (7). Within the range of (α ,Q) measured experimentally, namely α = 0.4-0.6, Q = 3.42-4.18, several points of (α ,Q) are selected. The value of |CO O -CO T | on each point of (α ,Q) are calculated. The parameter I equals the arithmetic mean value of |CO O -CO T | on each point. Apparently, its physical meaning is the average distance between two surfaces.
where CO O is the CO emission function of original gas, ppm; and CO T is the CO emission function of the three-component gas, ppm.  To quantitatively assess how "equivalently" CO emission from an original gas can be reproduced by a three-component mixture, a parameter I can be defined as the mathematical expectation of the absolute value of difference between two functions, as shown in Equation (7). Within the range of (α',Q) measured experimentally, namely α' = 0.4-0.6, Q = 3.42-4.18, several points of (α',Q) are selected. The value of |COO-COT| on each point of (α',Q) are calculated. The parameter I equals the arithmetic mean value of |COO-COT| on each point. Apparently, its physical meaning is the average distance between two surfaces. where COO is the CO emission function of original gas, ppm; and COT is the CO emission function of the three-component gas, ppm; If I < σ + σT, it can be concluded that a three-component mixture can give "equivalent" CO as that from original gas. Table 5 lists the calculated values of I, σ, and σT for six sets of original gases and their three-component mixtures. For all sets of gases, the calculated I's are small enough. It can be concluded that the selected three-component mixtures can equivalently represent their original gases, viz., if a certain gas is input into a gas cooker, the CO emission would be the same as that from its three-component mixture operating at the same heat input Q and primary air coefficient α'.

Influence of Gas Constituents on CO Emission
All gases falling into the category 12T in Chinese standard GB 13611-2007 could be reduced to a three-component mixture. It was found that all natural gas fell within an area: W = 45-55 and PN = 0-15. Compared to the range within which Dutton established his prediction method (W = 45-55, PN = 0-100) [21], the exploration range in this paper was narrowed down a lot.
In order to input 12T-0 into the left-side burner of tested cooker, first, we adjusted the primary air shutter until a satisfactory flame appeared. Figure 5 shows flame shapes under different primary air coefficients. Figure 5a is a result of smaller primary air, which is usually called a "soft" flame, and would lead to excessive CO emission. Figure 5d corresponds to a higher primary air coefficient. The If I < σ + σ T , it can be concluded that a three-component mixture can give "equivalent" CO as that from original gas. Table 5 lists the calculated values of I, σ, and σ T for six sets of original gases and their three-component mixtures. For all sets of gases, the calculated I's are small enough. It can be concluded that the selected three-component mixtures can equivalently represent their original gases, viz., if a certain gas is input into a gas cooker, the CO emission would be the same as that from its three-component mixture operating at the same heat input Q and primary air coefficient α .

Influence of Gas Constituents on CO Emission
All gases falling into the category 12T in Chinese standard GB 13611-2007 could be reduced to a three-component mixture. It was found that all natural gas fell within an area: W = 45-55 and PN = 0-15. Compared to the range within which Dutton established his prediction method (W = 45-55, PN = 0-100) [21], the exploration range in this paper was narrowed down a lot.
In order to input 12T-0 into the left-side burner of tested cooker, first, we adjusted the primary air shutter until a satisfactory flame appeared. Figure 5 shows flame shapes under different primary air coefficients. Figure 5a is a result of smaller primary air, which is usually called a "soft" flame, and would lead to excessive CO emission. Figure 5d corresponds to a higher primary air coefficient. The flame is "hard" and tends to lift when substituted by other gas rather than adjustment gas. Figure 5b,c show a satisfactory flame. The primary air coefficient is properly moderate and results in a quite acceptable flexibility to changing gas constituents. The adjustment of the flame above was set according to the report published by the Gas Research Institute (GRI) [27]. Unstable combustion phenomena, such as lift and excessive CO emission, will not occur when gas constituents fluctuate. The underlying principle for initial adjustment when fueled with 12T-0 is that the satisfactory flame shape in Figure 5b,c appears to ensure maximum flexibility. The flame shape gave an intuitionistic determination of the initial primary air coefficient. However, it could not be quantitatively analyzed. In practice, the gas-air mixture was extracted from the gas separator of the cooker by an injector and sent into gas chromatography to analyze the molar volume fraction of various components. Then, the primary air coefficient could be calculated. Therefore, it was possible to make the same condition for combustion under different gases. All the combustion cases were measured under the same initial air coefficient. shape in Figure 5b,c appears to ensure maximum flexibility. The flame shape gave an intuitionistic determination of the initial primary air coefficient. However, it could not be quantitatively analyzed. In practice, the gas-air mixture was extracted from the gas separator of the cooker by an injector and sent into gas chromatography to analyze the molar volume fraction of various components. Then, the primary air coefficient could be calculated. Therefore, it was possible to make the same condition for combustion under different gases. All the combustion cases were measured under the same initial air coefficient. After the primary air shutter was fixed, we let the burner that injected primary air operate at a nominal heat input of 3.8 kW. Then, the burner remained unchanged, and different three-component mixtures were input into the burner to record CO emissions. To systematically investigate the influence of W and PN upon CO emission, the input mixture was deliberately designed to remain constant W or constant PN. Figure 6 shows the measurement result, in which all blue figures donate CO emission in ppm.  After the primary air shutter was fixed, we let the burner that injected primary air operate at a nominal heat input of 3.8 kW. Then, the burner remained unchanged, and different three-component mixtures were input into the burner to record CO emissions. To systematically investigate the influence of W and PN upon CO emission, the input mixture was deliberately designed to remain constant W or constant PN. Figure 6 shows the measurement result, in which all blue figures donate CO emission in ppm.
The underlying principle for initial adjustment when fueled with 12T-0 is that the satisfactory flame shape in Figure 5b,c appears to ensure maximum flexibility. The flame shape gave an intuitionistic determination of the initial primary air coefficient. However, it could not be quantitatively analyzed. In practice, the gas-air mixture was extracted from the gas separator of the cooker by an injector and sent into gas chromatography to analyze the molar volume fraction of various components. Then, the primary air coefficient could be calculated. Therefore, it was possible to make the same condition for combustion under different gases. All the combustion cases were measured under the same initial air coefficient. After the primary air shutter was fixed, we let the burner that injected primary air operate at a nominal heat input of 3.8 kW. Then, the burner remained unchanged, and different three-component mixtures were input into the burner to record CO emissions. To systematically investigate the influence of W and PN upon CO emission, the input mixture was deliberately designed to remain constant W or constant PN. Figure 6 shows the measurement result, in which all blue figures donate CO emission in ppm.   Figure 7 shows the CO emission changing with PN for a constant W index. When the Wobbe index was kept unchanged, the observed increase of CO emission could be attributed to an increasing tendency for incomplete combustion resulting from increasing propane, and to an increase of contact time between the flame and cold bottom being heated because of increasing N 2 . Meanwhile, when the PN index increased, the port intensity (kW per square meter of burner port area) decreased because the heating value decreased, leading to a decreasing of secondary air. All these factors cause the CO emission to increase with increasing PN, in a linear manner.
investigate the influence of the Wobbe index on the formation of CO. Figure 8 shows the CO emission changing with W under a constant PN number. Due to the change of the Wobbe index, the operation point of the cooker changed accordingly. Given a fixed PN number, increasing the W index means a higher heat input and lower primary air. Both factors enhanced the CO formation in an exponentially increasing pattern. Therefore, CO tends to increase exponentially with W. The change of the Wobbe index affects the formation of CO in two aspects: the gas properties and the operation condition of the burner.   To obtain the relationship between CO emission and gas quality parameter, it is necessary to investigate the influence of the Wobbe index on the formation of CO. Figure 8 shows the CO emission changing with W under a constant PN number. Due to the change of the Wobbe index, the operation point of the cooker changed accordingly. Given a fixed PN number, increasing the W index means a higher heat input and lower primary air. Both factors enhanced the CO formation in an exponentially increasing pattern. Therefore, CO tends to increase exponentially with W. The change of the Wobbe index affects the formation of CO in two aspects: the gas properties and the operation condition of the burner.
index was kept unchanged, the observed increase of CO emission could be attributed to an increasing tendency for incomplete combustion resulting from increasing propane, and to an increase of contact time between the flame and cold bottom being heated because of increasing N2. Meanwhile, when the PN index increased, the port intensity (kW per square meter of burner port area) decreased because the heating value decreased, leading to a decreasing of secondary air. All these factors cause the CO emission to increase with increasing PN, in a linear manner.
To obtain the relationship between CO emission and gas quality parameter, it is necessary to investigate the influence of the Wobbe index on the formation of CO. Figure 8 shows the CO emission changing with W under a constant PN number. Due to the change of the Wobbe index, the operation point of the cooker changed accordingly. Given a fixed PN number, increasing the W index means a higher heat input and lower primary air. Both factors enhanced the CO formation in an exponentially increasing pattern. Therefore, CO tends to increase exponentially with W. The change of the Wobbe index affects the formation of CO in two aspects: the gas properties and the operation condition of the burner.   CO emission was found to be increasing monotonically with both W and PN. Equation (8) can describe the relationship between CO and W, PN quite accurately. CO = (w 1 × PN + w 2 )e w 3 ×W = (3.797 × 10 −4 × PN + 0.02)e 0.2045×W (8) where CO is the CO emission of the gas cooker, ppm; w 1 , w 2 , and w 3 are the coefficients; W is the Wobbe index, MJ/m 3 ; and PN is the volumetric percentage of C 3 H 8 plus N 2 of the three-component mixture. Let CO be equal to a constant value, e.g., 500; then, an equation depicting CO = 500 can be derived, as shown in Equation (9). In similar manner, a set of equal-CO lines for the tested cooker corresponding to initial primary air can be derived, as shown in Figure 9. W = −4.890 × ln 3.797 × 10 −4 × PN + 0.02 + 30.39 (9) mixture. Let CO be equal to a constant value, e.g. 500; then, an equation depicting CO = 500 can be derived, as shown in Equation (9). In similar manner, a set of equal-CO lines for the tested cooker corresponding to initial primary air can be derived, as shown in Figure9. ( ) To examine the precision of Equation (8), some supplemental measurements were made to compare the predicted CO emission and measured data. Several deliberately designed threecomponent mixtures, as listed in Table 6, were input into the burner. Measured CO emission were given as well (in carmine square), as shown in Figure 9. From the difference between the predicted equal-CO lines and measured values, the derived equation can be precise enough. In addition, for a gas cooker that was initially adjusted under 12T-0, its CO emission can be predicted by the formula CO = (w1 × PN + w2)e w3×W .

Figure 9.
Equal-CO lines of tested cooker, the thickening line is a 500-ppm equal-CO line, and the Chinese National Standard [24] stipulates that CO emission from cooker should not exceed 500 ppm.  Equal-CO lines of tested cooker, the thickening line is a 500-ppm equal-CO line, and the Chinese National Standard [24] stipulates that CO emission from cooker should not exceed 500 ppm.
To examine the precision of Equation (8), some supplemental measurements were made to compare the predicted CO emission and measured data. Several deliberately designed three-component mixtures, as listed in Table 6, were input into the burner. Measured CO emission were given as well (in carmine square), as shown in Figure 9. From the difference between the predicted equal-CO lines and measured values, the derived equation can be precise enough. In addition, for a gas cooker that was initially adjusted under 12T-0, its CO emission can be predicted by the formula CO = (w 1 × PN + w 2 )e w3×W . It is stipulated in Chinese National Standard GB16410-2007 [24] that the CO emission from a cooker under test gas should not exceed 500 ppm. Figure 9 shows that the CO emission of the tested cooker will not exceed 500 ppm when fueled with three-component mixtures that fall below a 500 ppm equal-CO line (thickening line). It can be found that the range of gas to ensure a qualified emission is