Objectives

Complaints from various sources of unpleasant odors have become a serious concern in sensitive locations downwind of industrial establishments in Taiwan. Compared with non-polar VOCs, odorous compounds not only are more polar but more reactive. Toxicity is also a common concern. In an effort to tackle malodor issues, this study is to aimed at developing an environmental forensic method to link odor to specific industrial activities. There were two major parts in this study: 1. to develop a novel H2S event-triggering system; 2. in-lab analyses for VOC fingerprints and forensic evidence. The triggering system was placed at the borderline of a could-be source of malodor targeting H2S. When the H2S concentration exceeded a pre-determined trigger level, the canister array unit was instantly triggered to collect air sample to “hold evidence”. The canister samples were then analyzed for 110 VOCs by GC-MS/FID in-lab to obtain VOC fingerprints of events. Meanwhile, a monitoring station measured C2-C12 NMHCs by in-situ GC-FID and dimethyl sulfide (DMS) by in-situ GC-PFPD further downwind of the suspected plant. Composition is useful in forensics, because malodor usually accompanies with other high levels of pollutants. Meteorology and VOC ratio techniques were employed to help establish the link between the suspected plant and the receptor sites – a useful way for forensic investigation.

Objectives

Identify malodor sources by event triggering and fingerprint analysis異臭味污染物指紋鑑定技術以石化廠為例 廖偉呈 a 、 張志忠 b 、劉文治 a 、王振興 c 、巫月春 c 、王家麟 a*, a 國立中央大學 化學系 , 中壢 320, 台灣

b 中央研究院 環境變遷中心 , 南港 115, 台灣c 行政院環保署 環境檢驗所 , 中壢 320, 台灣

Experimental I. The study domain

N

H2S monitoring site

Petrochemical plant

Downwind receptor site

Fig. 1. The suspected malodor source – a petrochemical plant and its surroundings. There were two monitoring sites established in this study. The red spot is the location of the H2S event-triggering system. The blue spot is the downwind site of in-situ GC-FID and GC-PFPD.

II. H2S event-triggering system & in-lab sample analysis

Fig. 2. The H2S event-triggering system with a canister array unit (left). GC-FID (for C2-C12 NMHCs) and GC-PFPD (for DMS) (right). Sampling probe was fixed at the roof-top of a four story building. The monitoring study started from September 1, 2011 to November 11, 2011

In-situ GC-FID In-situ GC-PFPD

Canister sampler inlet

H2S inlet

H2S event-triggering system

Data logger

H2S

Fig. 3. On the left, schematics of H2S event-triggering system: H2S analyzer (TELEDYNE, MODEL 101E); Data logger (CHINO, KR2000); canister array (Entect, 1800/1816). Triggering line is connected to trigger output of the data logger to activate ports of the canister array. The trigger signal is controlled by the logic algorithm within the data logger. Sampling will be activated when H2S value exceeds a pre-determined threshold value (8.5 ppbv). On the right, schematics of GC-MS/FID (Varian, CP3800/Saturn 2200) and the canisters introducer (Entech, 7016A). The event samples were analyzed in-lab for 110 VOCs including alkanes, alkenes, alkyne, aromatics, halocarbons, ester, ether, aldehyde, ketone.

Results & discussion

Event sample H2S (ppbv) Event date status

1 10.9 2011/10/24 19:41 Automated

2 12.7 2011/10/24 19:47 Automated

3 18.5 2011/10/24 19:53 Automated

4 11 2011/10/24 21:01 Automated

5 19.6 2011/10/24 21:07 Automated

6 11.5 2011/10/24 21:13 Automated

7 16.9 2011/10/24 21:21 Automated

8 9.5 2011/10/24 21:43 Automated

9 8.9 2011/10/26 1:05 Automated

10 11.5 2011/10/26 1:11 Automated

Fig. 4. H2S and wind speed (left); H2S and wind direction (right). Ten events occurred which triggered 10 air samples (left), as detailed in Table 1.

10 Events !!

Baseline sample H2S (ppbv) Event date status

1 0.6 2011/09/26 14:15 manual

2 0.3 2011/10/07 10:50 manual

3 0.7 2011/10/19 15:15 manual

Table 1. Three background samples (BG) were also collected to pose a contrast to the 10 event (Alarm) samples.

Fig. 5. The breakdowns of all 13 canister samples. AL = Alarm sample; BG = background sample. The total 110 VOCs were classified into six groups (alkanes, alkenes, alkyne, aromatics, halocarbons and oxygenated) and the sum of total VOCs is labeled on the top of each column. The pattern of BG is noticeably different from that of AL, and all AL patterns were very similar to each other.

Fig. 6. The composition of the10 AL samples vs. that of traffic samples: 10 AL samples (left); PAMS site ( 萬華光化站 ) (right). Light-weight hydrocarbons (C2- C5) were the major component in all AL samples, which is different from the traffic dominant samples.

Fig. 7. Time-series data of light-weight hydrocarbons and DMS (ethane, propane, butane, pentane, hexane) at the downwind site: alkanes (left); DMS (right). Interestingly, high alkane levels were also observed during the high H2S period (10/24 17:00 ~ 10/24 19:00). In addition to H2S, the intensity of DMS also increased abruptly (right). The high content of light alkanes may suggest gasoline storage tank leakage, being transported to downwind site. The ratio technique (Fig. 8) can better illustrate the differences of air masses.

Events !!

Acknowledgements1. NIEA of EPA (EPA-100-E3S2-02-01)2. Research center of environmental changes, Academia Sinica3. CPC corporation in Taoyuan, Taiwan

Fig. 8. Ratios of ln (propane/ethane) to ln (n-butane/ethane) were plotted to distinguish air samples of different origins: 1. A PAMS site ( 萬華站 ) to represent traffic fingerprint (red) 2. Downwind receptor dataset (purple)4. Alarm samples ( blue)5.Samples taken from the plant (green). The urban source (traffic) tend to cluster to lower left, the AL samples tend to move to the upper right. The downwind samples lie between urban and AL samples, implying partial influence from the plant.