Downwind concentration of methane

A total of 9 continuous measurements of methane concentration at a fixed point downwind were made at 5 LNG fuel stations. Each continuous measurement time ranged from 2 to 7 hours. The measured methane concentration is shown in Figure 2. The methane concentration for each station varies over time, and the measured baseline methane concentration varies between 1.9 and 2.5 ppm. The maximum concentrations of the five stations vary greatly. Since the distance between the meter and the station is almost the same, the broadcast rate from each station will be completely different.

Figure 2
Figure 2

Measuring methane concentrations downwind of LNG supply stations (measurement conditions are given in Table 2).

Figure 3 shows the statistical distribution of the measured gas concentration enhancement. We divided methane concentration enhancement into three categories: 0; 0-0.2ppm; >0.2 ppm. It can be seen that the majority (>50%) of the measured methane concentration enhancement falls into the 0-0.2 ppm category. The ratio >0.2 ppm represents about 20% of the total number in most stations, while the amount of data without methane concentration boosting accounts for the lowest percentage, which is less than 5%.

Figure 3
Figure 3

Statistical distribution of the measured methane concentration enhancement.

methane emission rate

The Gaussian point source method, US Environmental Protection Agency OTM33A, is used to calculate the methane emission rate. It usually takes 20 minutes to measure the concentration to obtain an estimate of the emission rate12. In our study, the measurement time is between 2 and 7 hours. Hence, in our analysis, we divide the entire measurement period into several consecutive 20-min sections, and the data within each section are analyzed to obtain an average emission rate within that section. A series of pseudo-instantaneous broadcast rates can then be derived to detect the temporal change of broadcasts for the total time periods measured. The starting point for the first 20 minute section may have some effects on each average emission rate. With the data obtained for Station 1 on September 16, 2020 as an example, Figure 4 shows that when we chose the starting points for the first 20 min section as 0, 5, 10 and 15 min after the concentration measurement was launched, the shape of the mean emission rates changes slightly. When we calculated the average total emission rate over the 5-hour measurement period, the numbers we obtained were 0.0271, 0.0200, 0.0193, 0.0233 kg/h, respectively, for these four treatment groups. He points out that the choice of the emission rate estimation interval does not affect the total average emission rate, but within a limited range.

Figure 4
Figure 4

Effect of starting point on quantified 20 min emission rate (1 .).# station on September 16, 2020).

One can imagine that two sources contribute to emissions: continuous emission such as a pipeline or equipment leak, and intermittent emission from processes such as LNG refueling, degassing or even venting a leaky gas (BOG). In this study, we define continuous emission because it occurs at time periods when the measured methane concentration is less than 2.5 ppm, while the residual emission is defined as intermittent emission.

For the five stations measured, Table 3 shows the average total emission rates, their contribution from continuous and intermittent broadcasts, as well as the loss rate, where it is defined as

$${\text{Loss}}\,{\text{rate}}={\text{CH}}_{{4}}{\text{emission}}\,{\text{rate}}\times {24}/({\text{Daily}}\,{\text{LNG}}\,{\text{sales}})$$


Table 3 Average methane emissions from LNG fueling stations.

It can be seen from Table 3 that the standard deviation of broadcast rates per station is very high compared to the estimated average value, which is an indication that broadcast rates per 20 minutes quantitatively are very scattered. This is consistent with the emission pattern shown in Figure 4: While the emission is very low most of the time, there are sporadic emission peaks.

Another source of high standard deviations may come from measurement uncertainties. OTM 33A itself has its own uncertainties. OTM 33A Inherent uncertainty due to computational model assumptions influenced by different time span lengths, data rates, and wind filters12, 20, 21. In addition, there are uncertainties in the measurement process, where the main factors will be the downwind distance and altitude differences between the emitting source, or possible multiple sources, and the measurement point. Experiments by Rachel et al.17 showed that source height had little effect on the OTM 33A method. However, the presence of potential emitting sources may cause a certain deviation in the direction of the wind. The main emission sources for conventional LNG fuel stations come from the refueling area, storage tank area and unloading area, among which there is a refueling area with distributors of about 18m x 20m; Refueling area with 4 dispensers is approximately 40m x 20m; Unloading area 20m x 6m; The tank area is 15m x 20m. We assumed the extreme case of the measured station that when the main potential emission sources are present, the downwind distance is determined to be 28 m with the mean and netted distance between the measurement point and each emission source, which differs by 5 m from the downwind distance used in our original calculation. In this case, the emission flux changes by 37.9%. Since there were sixteen 20-min measurement events at our measurement sites, the increasing uncertainty in the average emission rate during each 20-min period was ~9.475%, which means that repeated measurements can reduce the uncertainty in randomly distributed measurements.

The methane emission rates of the five stations can be seen to vary widely, in terms of size, the average methane emission rate 1# The gas station is the smallest, at 10 . in size–2 kg / h, while emissions from 2 . plants#3#5# be on the order of 10-1 kg/h, emission rate 4# The gas station is the highest, reaching 4.5 kg / h. From the perspective of the contribution ratio, it can be seen that the intermittent emission of gas stations is the main source of emissions. Loss rates at the five refueling stations also differ significantly. The lowest loss rate is only 0.004% for 1# LNG fuel station, the highest loss rate comes from 4# Refueling station 0.257%.

As previously discussed, methane emission from LNG filling stations shows a strong time dependence. Hence it would introduce significant bias if one took a short period of time to measure and tried to use this result to represent the overall average emission. Based on the data collected by this study, the error from any 20-minute measurement is estimated, by comparing the 20-minute measurement value with the mean value over the entire period of 2-7 hours, and the result is shown in Figure 5. Most of the relative errors are less than zero and concentrated Between -1 and 0, which may make the average measurement method in the short period less than the measurement average of the long period, which leads to an underestimation of methane emissions. Even in some time periods, the error can be as high as 10, which leads to a dangerous overestimation of methane emissions from gas stations. Therefore, a comprehensive estimation of methane emissions from gas stations may require a longer continuous measurement period.

Figure 5
Figure 5

Relative error for different measurement periods.

Figure 6 shows the effect of different measurement periods on the relative error probability. When applying a time-length measurement of 40 min, 60 min, 1 h, and 6 h, we assume that the measured emission rates are the mean values ​​from continuous 20 min measurements during this time scale. As the length of the measurement time increases, the probability of underestimation decreases. In fact, when using the measurement time of 2 hours, the probabilities of underestimating and overestimating each other appear to be underestimated.

Figure 6
Figure 6

Relative error in the period of different data analysis.

Emission event distribution

If emission every 20 minutes is considered as an individual event, then the population distribution of these emission events and the total emission as a function of emission volume is shown in Figure 7, it can be seen that 71% of emission events are at the 0.01 kg/h level, but contributes to only 2% of the total emissions, while 76% of methane emissions come from 1% of events with an emission rate of 10 kg/hr. This is a typical fat distribution.

Figure 7
Figure 7

Population distribution of emission events and total emission as a function of emission volume.

In addition, there is also a fat tail distribution of methane emission rates among gas stations. It can be seen from Table 3 that out of the 5 stations, 4# The gas station has a very high emission rate of 4.5 kg/h, more than five times the total emission rate of the other four stations, that is, a small portion of poorly functioning gas stations contribute to a large portion of the total emissions. This result is consistent with the general distribution of emissions in natural gas infrastructure, as described by Zavala-Araiza et al.22,23.

Emission Classification

The Matlab-based EPA OTM 33A implementation requires the quantification of Gaussian emissions from skilled scientists or engineers to perform tedious computational efforts, which may not be feasible for routine measurement and monitoring of industrial emissions. In the real world, if at all possible, it might be useful to classify an emission degree for a particular station by measuring the methane concentration downwind, without calculating the emission rate figure. By observing the emission rates measured for the five LNG plants, with reference to the urban natural gas leakage classification scheme from Weller et al.24we decided to classify methane emission rates from LNG plants into three levels: low, medium and high, on the order of magnitude of the emission rate, as shown in Table 4.

Table 4 LNG plant emission rating classification.

By analyzing the measured methane concentrations approximately 23 m downwind of the nearest distributor, we found that there is a certain correlation between the average measured methane concentration enhancement and methane emission rate, as shown in Fig. 8. We determine the average methane concentration enhancement (ca(as the measured average methane concentration)cFrom(minus background methane concentration)cbg):

$${\varvec{C}}_{{{\text{ae}}}={\varvec{C}}_{{\text{av}}}}-{\varvec{C} } _ {{{\text {bg}}} $$


Figure 8
Figure 8

Correlation between average increase in methane concentration and methane emission rate and emission rating rating.

It can be seen from Figure 8 that the emission rate increases approximately proportionally to the average measured methane concentration enhancement, with the dotted line a regression lined for emission rate and methane enhancement. The concentration improvements that define low/medium and medium/high emission are about 0.22ppm and 0.94ppm, respectively. However, it is worth noting the following:

  1. (1)

    The current methane concentration was measured about 23 m downwind from the nearest distributor. If the measurement location changes, the current evaluation criteria should also vary.

  2. (2)

    In the current study, only one plant falls into the large emission category. Further measurements are needed to improve the assessment of the current classification.

Leave a Reply