運用開徑式傅立葉轉換紅外光光譜儀及空氣擴散模式定位逸散源

Integrating Open Path Fourier Transform Infrared Spectrometry and an Air Dispersion Model for Locating Emission Sources

葉政凱(Cheng-Kai Yeh)

指導教授 ： 吳章甫

國立臺灣大學/公共衛生學院/職業醫學與工業衛生研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.01445

摘要

倒轉大氣擴散模式(Inverse dispersion model)來定位汙染源是許多定位方法中的一種。 在過去的研究中，藉由定點採樣的方式與逆大氣擴散模式的慨念，結合最佳化演算法，其回推汙染源位置的結果與真實值的距離小於1公尺。 然而，以定點採樣來取得濃度資料可能需要大量的定點採樣器(例如採樣鋼瓶)，這可能會花費許多的成本；且在收集與分析的過程中，也可能也會花費較多的時間。 因此，在本研究我們利用光學遙測技術(ORS)之儀器與多道測線來獲得下風處濃度資料。藉由電腦模擬或實際量測的path-integrated concentration (PIC)資料，並利用最佳化演算法可得出符合的預測PIC資料，藉由逆大氣擴散模式的方法來完成回推與重建汙染源。
在本研究中包含了兩個主要部分:電腦模擬實驗與實地實驗。在電腦模擬實驗中，我們除了探討PIC資料做為輸入資料外，也進一步比較PIC資料與採樣點資料在利用逆大氣擴散模式的方法，其回推汙染源結果的差異。在此模擬實驗中，我們使用高斯擴散模式模擬下風處測線所得到的煙流分布情形，在長寬各為201公尺的面積下，我們設立了兩道互相垂直的測線，每一道測線均架設三面反射鏡(67公尺、134公尺與201公尺)，用來提供不同線段的濃度資料。我們模擬了181道不同角度風向(從315°到135°)在流率為1.5g/s下，其下風處測線的濃度分布情形。因此，每一個模擬的汙染源皆有181個回推的結果。在(100, 100)至(200, 200)間隔距離為10m的121個模擬的汙染源，我們發現使用PIC資料做為輸入資料下,有66個模擬的汙染源其回推的中位數距離小於10公尺(54.5%)，而有69個模擬的汙染源(57%)其回推的中位數流率介於1.2g/s~1.8g/s之間，相對於使用採樣點資料的5% (n=6)與36.4% (n=44)，PIC資訊的回推的結果是比較佳的。另外，我們也發項當擁有6道有效量測的光徑數時，其上述兩項數值會提高至100%。然而，使用採樣點資料的情況下，僅有111個模擬的汙染源擁有3點有效量測的資料，其中位數距離小於10公尺與中位數流率介於1.2g/s~1.8g/s之間的這兩項百分比僅能提高至51%與93%。
在實地實驗中，我們使用開徑式傅立葉轉換紅外線光譜儀(OP-FTIR) 做為偵測儀器。除了以1道風向(6道PIC)來回推汙染源外，我們也利用兩道風向(12道PIC)並搭配Moving Average PIC資料來改善其回推結果。在第一次實地實驗中，其污染源位於座標(150,100)且流率為0.43g/s，最佳的回推的結果是利用兩道風向其下風處各有6道有效量測的光徑數來回推，其中位數距離與中位數流率為31.3公尺與0.48g/s。而在第二次實地實驗中，污染源位於座標(100,100)，流率同樣也是0.43g/s，最佳的回推的結果是利用兩道風向其下風處各有3道有效量測的資料，搭配組數為10的Moving Average PIC資料來回推，其中位數距離與中位數流率為23.8公尺與0.43g/s。

關鍵字

開徑式傅立葉轉換紅外線光譜儀 ； 輻射形光徑煙流重建法 ； 定位 ； 逆大氣擴散模式 ； 光學遙測

並列摘要

Inversing dispersion model was one of the approaches to locate emission sources. In one previous study, the point sampling data were combined with an optimization algorithm to inverse the dispersion model. The result showed that the distance between the true and reconstructed source location was less than 1m. Nevertheless, using large amounts of point samplers was expensive and the processes for sample collection and analysis were time consuming. In our study, we proposed using an optical remote sensing (ORS) instrument with multiple beam paths to collect the downwind concentration data. By fitting the predicted path-integrated concentration to the observed path-integrated concentration, the inverse dispersion process could be implemented to reconstruct the source locations.
In the computational simulation study, we compared the differences between the reconstructed results by using PICs data and by point sampling data. The beam geometry of the experiment was composed of two perpendicular monitoring lines in a 201m×201m area. For each monitoring line, three retroreflectors were located equally to provide segmenting information to reconstruct the source location. The emission sources with the emission rate of 1.5g/s were simulated as gas leakage at the 121 different locations (from the source locations at (100, 100) to (200, 200) with the interval of 10m). The wind direction varies from 315° to 135° (n=181). For the reconstructed result with input PICs data, the proportion of having DISTmedian less than 10m of each reconstructed source was 54.5%. The proportion of RATEmedian between 1.2 g/s and 1.8 g/s of each reconstructed source was 57.0%. For the reconstructed results from using point sampling data, the two proportions were 5% and 36.4%, respectively. It was also observed that using 6 effective input PICs improved the proportions to 100%. On the contrary, the proportions improved to only 51% and 93% when using the input data of 3 effective point sampling data.
In the field study, the open-path Fourier Transform Infrared (OP-FTIR) spectrometer was used as the monitoring instrument. In the first field campaign, the real source location was at (150,100) and the release rate of the tracer gas was 0.43g/s. The result showed that the median distance between the real and reconstructed source (DISTmedian) was 29.2m and the and the median reconstructed emission rate (RATEmedian) was 0.49g/s by using 12 effective field PICs data. In the second campaign, the tracer gas was set at (100,100). By using 6 effective PICMAvg10 (a moving average with a group size of 10) data, the DISTmedian and the RATEmedian were 25.5m and 0.43g/s, respectively.

並列關鍵字

OP-FTIR ； RPM ； Source localization ； Inverse dispersion model ； Optical Remote Sensing

參考文獻

1.Piccot, S. D., S. S. Masemore, E. S. Ringler, S. Srinivasan, 1994. Validation of a method for estimating pollution emission rates from area sources using open-path FTIR spectroscopy and dispersion modeling techniques. Air & waste Management Association., 44, 271-279.

2.Brook, Robert D., 2008. Cardiovascular effects of air pollution. Clinical Science, 115(6), 175.

3.Thomson, L., B. Hirst, G. Gibson, S. Gillespie, P. Jonathan, K. Skeldon, M. Padgett, 2007. An improved algorithm for locating a gas source using inverse methods. Atmospheric Environment, 41(6), 1128-1134.

4.Allen, C., G. Young, and S. Haupt, 2007. Improving pollutant source characterization by better estimating wind direction with a genetic algorithm. Atmospheric Environment, 41(11), 2283-2289.

5.Lushi, E., J. M. Stockie, 2010. An inverse Gaussian plume approach for estimating atmospheric pollutant emissions from multiple point sources. Atmospheric Environment, 44(8), 1097-1107.