- Email: [email protected]
Correlation of musty odor and 2-MIB in two drinking water treatment plants in South Taiwan
The Science of the Total Environment 289 (2002) 225–235
Correlation of musty odor and 2-MIB in two drinking water treatment plants in South Taiwan Tsair-Fuh Lin*, Jiun-Yue Wong, Hsiao-Pin Kao Department of Environmental Engineering, National Cheng Kung University, Tainan City, 70101, Taiwan, ROC Received 23 February 2001; accepted 10 September 2001
Abstract Possible odor groups and intensity, and seasonal effects were elucidated in two representative water treatment plants (WTPs), Feng-Shen and Gun-Shi, in southern Taiwan. The flavor profile analysis (FPA) was employed to determine the odor groups for the source water, while a chemical analysis, solid-phase microextraction (SPME) coupled with a gas chromatograph and mass spectrometric detector (GCyMSD), was used to concentrate and subsequently analyze the corresponding water samples. FPA results show that fishy and musty odors were the two major odor groups in the source water. Results of chemical analysis showed that 2-methyl-isoborneol (2-MIB) was present in the source water. The correlation between 2-MIB concentration and the FPA intensity of musty odor was compared with the dose–response curve generated in the laboratory by the FPA panelists. The experimental data from the two water treatment plants follow the calibration curve closely, indicating that the musty odor of the two source waters were most likely contributed from 2-MIB. In addition, there is good correlation between logarithmic 2-MIB concentration and water temperature, substantiating the importance of seasonal effect. Although approximately 40– 50% of 2-MIB removal was found in the treatment trains for the two WTPs, only an approximately 0.3 FPA intensity scale of reduction was expected. 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Musty odor; Fishy odor; Flavor profile analysis; 2-MIB; Solid-phase microextraction
1. Introduction Drinking water quality in southern Taiwan received complaints from customers for a long time for its unpleasant odors. Previous studies (Chen et al., 1982, 1983; Lin and Fu, 1997) suggested that mustyyearthy and fishy odors were the two major odors in drinking water sources, and algae and actinomycetes were the most likely *Corresponding author. Fax: q886-6-275-2790. E-mail address: [email protected] (T.-F. Lin).
sources for the odors in a raw water source in the area. Although musty and earthy odors were reported to be present in the summer time and the fishy odor was present in winter, the chemicals and algae species responsible for the odors remain unknown. Mustyyearthy odor is a prevalent odor subgroup in drinking water, and has been reported by many researchers (Yagi et al., 1985; Means and McGuire, 1986; Jensen et al., 1994; Bruchet, 1999; Sklenar and Horne, 1999). Five organic chemicals,
0048-9697/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 1 . 0 1 0 4 9 - X
226
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
namely trans-1,10-dimethyl-trans-9-decalol (geosmin), 2-methyl-isoborneol (2-MIB), 2-isopropyl3-methoxy pyrazine (IPMP), 2-isobutyl-3-methoxy pyrazine (IBMP) and 2,3,6trichloroanisole (TCA) are often considered as the responsible mustyyearthy odorants (Lalezary et al., 1986; Suffet and Wable, 1995). The odor thresholds of these compounds are as low as a few ngy l in drinking water. Among these five chemicals, geosmin and 2-MIB are the most common mustyy earthy compounds (Suffet et al., 1999) and are difficult to remove by most treatment processes (Lalezary et al., 1986). Although several advanced treatment technologies are able to remove the mustyyearthy odorants, other methods, such as ozone oxidation (Glaze et al., 1990; Duguet et al., 1995; Atasi et al., 1999; McGuire, 1999) and activated carbon adsorption (Gilloly et al., 1999; Graham et al., 2000), are generally expensive compared to conventional treatment processes. Therefore, in order to develop a cost effective technology for the control of these odorants in drinking water, it is necessary to understand the background information of the odorants and their relation to environmental conditions (Means and McGuire, 1986). Although odor problems have been present in drinking water in southern Taiwan for a long time, no flavor profile analyses (FPA) and chemical analyses were conducted to identify the odor groups and odorants responsible for the odor, and no evaluation of odor removal efficiency for the current processes was performed. In this study, the odor problems in two representative water treatment plants in the area were analyzed for their odor groups and intensity using both sensory and instrumental techniques. The results were correlated with temperature change in the area. In addition, the removal efficiency for different treatment processes at the two treatment plants was also presented. 2. Site description Two water treatment plants (WTPs), Feng-Shen and Gun-Shi, were chosen in this study as both have had odor problems for a long time. As illustrated in Fig. 1, Feng-Shen and Gun-Shi WTPs
are located near Kaohsung City, the largest city in southern Taiwan. Conventional treatment processes are employed in both treatment plants. The following section provides the background information for the two plants. 2.1. Feng-Shen water treatment plant Feng-Shen Reservoir, located in Kaohsung County, Taiwan, is one of the major drinking water and industrial water sources in the Kaohsung Metropolitan area. As an off-channel reservoir, it receives water from the Tongun Creek, which is approximately 10 km away from the reservoir. Surface water from Tongun Creek is pumped into the reservoir through two raw water transfer conduits at a rate of 500 000 m3 per day. The area and storage capacity of the reservoir are equal to approximately 7.5=105 m2 and 4.0=106 m3, respectively. Feng-Shen Water Treatment Plant (FSWTP) is located on the northwestern side of Feng-Shen Reservoir. Important source water quality parameters for the plant in the year 2000 are listed in Table 1. Initially, the production rate of FSWTP was 350 000 m3 per day. As the demand of water supply rate in Kaohsung Metropolitan area increased, the production rate was raised to 500 000 m3 per day. As shown in Fig. 1, a conventional treatment train is used in the plant, with pre-chlorination, coagulation, settling, rapid sand filtration, and post-chlorination processes. Threshold odor number (TON) measurement was conducted in the water treatment plant monthly for both the raw and the finished water by the Taiwan Water Supply Corporation. The measured TON values from January 1996 to January 2000 are illustrated in Fig. 2a. The TON in the raw water was as high as 16, indicating that the odor intensity in the reservoir was strong. Although the TON values of the finished water were all below the national drinking water standard for odor (s 3) (ROCEPA, 1998), there are still common complaints from customers. Since no other sensory tests except TON were conducted at the treatment plant, no description of the odor types was reported. However, the odors were expected to include both fishy and musty, at least. In addition, since
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
227
Fig. 1. Locations and treatment trains of Feng-Shen and Gun-Shi water treatment plants.
free chlorine is present in the finished water and was not accounted for in the TON measurement, it is likely that the odors were impaired by chlorine in the finished water.
2.2. Gun-Shi water treatment plant Gun-Shi water treatment plant (GSWTP) is located right beside Tongun Creek in Pingtong
Table 1 Source water quality of Feng-Shen and Gun-Shi Water Treatment Plantsa Water treatment plant
pH
Alkalinity, mgyl CaCO3
Hardness, mgyl CaCO3
Total dissolved solid, mgyl
Total organic carbon, mgyl
Turbidity, NTU
Feng-Shen Gun-Shi
7.5"0.1 7.4"0.2
176"35 171"35
254"40 248"39
403"76 393"64
2.4"1.0 4.0"1.5
5.1"2.5 16.9 (5–170)b
a b
Monthly monitoring data (January to December of 2000) from Taiwan Water Supply Corporation (mean"S.D.). Monthly monitoring data (January to December of 2000) from Taiwan Water Supply Corporation wmean (range)x.
228
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
odor compound, which is approximately 9–42 ngy l (Watson et al., 2000). 3. Experimental methods
Fig. 2. Historical threshold odor number (TON) for the raw and finished water at (a) Feng-Shen, and (b) Gun-Shi water treatment plants.
County, Taiwan. The raw water of the plant is directly pumped from Tongun Creek, and source water quality parameters for the plant in the year 2000 are listed in Table 1. The water production rate of GSWTP is 26 000 m3 per day, which is distributed to a few nearby towns for domestic water use. Since the ammonia content in the raw water can be as high as 5 mgyl, rotating biological contactors (RBCs) were installed right after the intake of the plant to nitrify the ammonia. An average of 70% ammonia removal was observed in the process. As shown in Fig. 1, a process train including pre-chlorination, coagulation, settling, rapid sand filtration and post-chlorination is used after the RBCs. The chronological TON values for the plant are shown in Fig. 2b. It is noticed that the TON values can be as high as 18. Similar to FSWTP, the TON values in the finished water for GSWTP were all below three. The results of a preliminary analysis (Kao, 1998) indicated that 2-MIB was present in the whole treatment train, from raw water to finished water, and the concentration could be as high as 300 ngyl. This concentration is certainly much higher than the odor threshold for the musty-
Two analytical schemes, including sensory and instrumental methods were employed in this study. A sensory method, flavor profile analysis (FPA) was used to identify the odor group and intensity for the drinking water, while a gas chromatography mass spectrometric detector (GCyMSD) incorporated with solid-phase micro-extraction (SPME) concentration technique was used to analyze the possible odor compounds. A detailed description of training and applications for the FPA method can be found in the Standard Methods for Water and Wastewater (APHA et al., 1995). In this study, more than six non-smokers were trained for the panel, and at least five panelists participated in each test. Seven-point scales of 1–12, as suggested by Philadelphia Water Department of the USA and Lyonnaise des Eaux of France (Mallevialle and Suffet, 1987), were used to describe the intensity of samples based on the consensus of the panel. In addition, water from a Mili-Q water purification system (Millipore Corp., Bedford, MA, USA) was used for all the dilution of samples and standards in the laboratory. The newly developed SPME method (Lloyd et al., 1998; Watson et al., 2000) was chosen and modified as the concentration technique for extracting odorous compounds from water samples. A detailed procedure may be found in Lloyd et al. (1998) and Watson et al. (2000), only the modified part is introduced below. A commercially available fiber coated with a 30y50-mm divinylbenzeneycarboxenypolydimethylsiloxane film (No. 57348-U, Supelco, USA) was used as the adsorbent. The fiber was injected through a Tefloncoated septum and placed into the headspace of a 350-ml extraction vial with 200 ml of water sample. A water-jacketed system was designed for the extraction vial, allowing temperature control for the water sample. During the extraction process, the vial was controlled at temperatures 50"0.5 8C stirred with a Teflon bar. To reduce the impact of water vapor on the adsorption process, 50 g of NaCl (Reagent Grade, Riedel-de Haen,
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
Germany) was added to the sample, as suggested by Lloyd et al. (1998). Preliminary experiments indicated that an exposure of the fiber to the samples at 30 min produced excellent reproducibility and linearity of calibration curves. After exposure, the fiber was introduced into the injection port of an HP 6890 (Hewlett-Packard, USA) gas chromatograph (GC) equipped with an HP 5973 mass spectrometric detector (MSD) for the separation and identification of adsorbed odorous compounds. A 30-m long, 0.25-mm i.d. HP-5 capillary column coated with a 0.25-mm 5% phenyl methyl siloxane film was used in the HP GC. High purity helium ()99.999%, San-Fu Chemical Corp., Taiwan) was used as a carrier gas at a rate of 2 mly min. The injection port was operated at splitless mode with temperature controlled at 250 8C and pressure at 4.9 psi. The oven temperature was held at 60 8C for 1.0 min, followed by an increase to 250 8C at a rate of 8 8Cymin, and then held at 250 8C for 5 min. The analytical standard for the major mustyy earthy odor compound, 2-MIB, was purchased from Sigma (analytical grade, US) at a concentration of 2 mgyl, and that for geosmin was from Wako (analytical grade, Japan) at 0.1 mgyl. The retention times for 2-MIB and geosmin in this study were 9.49 and 13.32 min, with characterized massycharge (myz) ratio equal to 95.1 and 112, respectively. In addition to the retention times, the observed mass spectrum of geosmin and 2-MIB of field samples were compared with the corresponding compounds in the standard database (NBS75K.L, Hewlett-Packard, USA) for confirmation. The calibration curves for these two compounds were established for two sections, one from 5 to 100 ngyl and the other from 100 to 1000 ngy l, all with good linearity and high regression coefficients (R 2)0.995). The detection limits for this study were 8 and 6 ngyl for geosmin and 2MIB, respectively, while analytical results of the duplicates were all within "10% of difference. These detection limits are close to the threshold of the two odorous compounds, which is 9–42 ngyl for 2-MIB and 4–10 ngyl for geosmin (Watson et al., 2000).
229
Fig. 3. Dose–response curves for (a) musty odor and (b) fishy odor for laboratory FPA tests.
4. Results and discussion 4.1. FPA dose–response curves To establish the dose–response curves for the FPA tests, six replicates were conducted in the laboratory at different times for musty and fishy odors. In the tests, 2-MIB at five different concentrations between 10 and 1000 ngyl was used to simulate the field musty odorant at different odor strengths. For the fishy odor, trimethylamine at five concentrations of 0.01–1 mgyl was used. Fig. 3 shows that the FPA odor strength was proportional to the logarithmic concentration of corresponding odorant, which closely follows the dose– response relationship described by the Weber– Fechner model (Suffet and Mallevialle, 1995). Although Watson et al. (2000) reported that the odor threshold of 2-MIB is 9–42 ngyl, a lower threshold was observed in this study. Our observation of odor threshold for 2-MIB (at approx. 1 ngyl) was similar to that shown by Krasner et al. (1985). The high correlation coefficients (both larger than 0.98) indicate that the data points
230
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
the two major odor groups in the source waters, which was similar to the odor groups in the area reported by Lin and Fu (1997). The change of FPA intensity for the two odors, and corresponding temperature and observed 2-MIB concentrations were plotted in Figs. 4 and 5. As illustrated in Figs. 4 and 5, the intensity of the musty odor was strong in the summer months, i.e. in October 1999 and May 2000, and was weak in the winter months, i.e. December 1999. On the other hand, the fishy odor was the strongest in the winter month of February 2000. A similar observation of the seasonal effect on odor groups was observed by Chen et al. (1982, 1983) at a nearby water source, i.e. Cheng-Chin Lake. Chen et al. (1982, 1983) found a presence of fishy odor (up to 10–18 TON) in the wintery
Fig. 4. Seasonal changes of (a) temperature, (b) odor intensity, and (c) MIB concentration at Feng-Shen water treatment plant.
followed the dose–response curves closely, and the small standard deviations (S.D., all within one FPA scale) reflect the good reproducibility of the tested results. Laboratory dose–response curves for the FPA intensity and corresponding odorant concentration were constructed for the two odors tested, as shown in Fig. 3. These curves may be used to analyze the field data obtained later. 4.2. Seasonal change of raw water odors Samples of the raw water were analyzed for the odor in the Feng-Shen and Gun-Shi water treatment plants one to two times each month from September 1999 to May 2000. Both FPA and SPMEyGCyMSD analysis were conducted. FPA results indicated that fishy and musty odors were
Fig. 5. Seasonal changes of (a) temperature, (b) odor intensity, and (c) MIB concentration at Gun-Shi water treatment plant.
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
spring (cold) seasons, and musty odor in the summeryautumn (warm) time. In their study, the number of algae and actinomycetes were also measured. Lin and Fu (1997) analyzed the experimental data from Chen et al. (1982, 1983) to obtain the relationship between odor strength and algaeyactinomycete counts. Strong correlations between TONs and the logarithmic number of actinomycetes in warm seasons, and between TONs and logarithmic diatom numbers in cold seasons were observed. Since no relevant data on algae species were collected in this study, the correlation between algae counts and odor strengthygroups cannot be confirmed. Although the aquatic environment for Cheng-Chin Lake and the two water treatment plants are different, data collected from the Cheng-Chin Lake may be useful information for application to these two source waters. Chemical analysis showed that 2-MIB was present in the source water all year round at both treatment plants. Although geosmin was also observed in the source water, the concentrations were always below the detection limit (8 ngyl) (data not shown in the figure). Additionally, three other common mustyyearthy odor compounds, 2isopropyl-3-methoxy pyrazine (IPMP), 2-isobutyl3-methoxy pyrazine (IBMP), and 2,3,6-trichloroanisole (TCA) were not observed in the mass spectrometric detector. Therefore, only 2MIB dominated the five common musty compounds. As shown in Figs. 4 and 5, 2-MIB concentrations were found to increase as the temperature increased at both water treatment plants. The trend of the water temperature and the logarithmic 2-MIB concentrations is similar at corresponding time. The correlation between temperature and odor in the raw water was analyzed based on the observed 2-MIB concentrations and the corresponding water temperature at sampling time. Although other environmental conditions, such as water quality parameters, may affect the 2-MIB concentrations, Fig. 6 shows that 2-MIB concentrations in the two water sources were closely correlated to the water temperature. This observation suggests that temperature is one of the most
231
Fig. 6. Correlation between water temperature and MIB concentration for Feng-Shen and Gun-Shi water treatment plants.
important factors governing the production of 2MIB in the source waters. Attempts were also made to explore the relationship between fishy odor and water temperature. Although a few flagellated algae (Suffet et al., 1999) and Synura petersenii (Rashash et al., 1996) were identified as being responsible for fishy odor in water resources, we were not able to observe their presence in the source water. Rashash et al. (1996) separated a fishy odor compound, 2trans,4-cis,7-cis-decatrienal, from the Synura petersenii cultures. However, this chemical is not commercially available. Instead, three of other fishy odor related aldehydes, including n-hexanal, n-heptanal, and trans–trans-2,4-heptadienal (Suffet et al., 1999) were analyzed. None of the fishy odorants were observed in this study. The FPA intensity for fishy odor was further analyzed with the corresponding water temperature in the same approach as that for musty odor. However, no simple correlation can be found in the analysis. Indeed, environmental parameters other than water temperature may affect the generation of fishy odor compounds in the source waters. A more rigorous monitoring and analysis may be needed to find the relationship between fishy odor and environmental conditions. 4.3. Comparison between laboratory and field results The correlation of 2-MIB concentration and FPA intensity of musty odor was compared with the dose–response curve generated in the laboratory
232
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
4.4. Odor removal at Feng-Shen and Gun-Shi water treatment plants
Fig. 7. Correlation of FPA intensities and MIB concentrations in laboratory and field samples. (a) Feng-Shen WTP (b) GunShi WTP.
by the FPA panelists. As shown in Fig. 7, the experimental data from the two water treatment plants followed the dose–response curve closely, and all data points fell within the 95% prediction interval of the regression curve. In fact, for all experimental data, only one FPA scale of discrepancy from the calibration curve was found at most which suggested that the musty odor of the two source waters was very likely from 2-MIB. A similar correlation between 2-MIB and FPA intensity in laboratory prepared standards and field samples was observed by Krasner et al. (1985) at the Metropolitan Water District of South California, USA. Although the correlation between FPA and MIB concentrations in laboratory and field samples is promising, the level of this agreement should not be over-expected. For example, at an FPA intensity of 6 in Fig. 7, 2-MIB concentrations range from approximately 12 to 60 ngyl. In fact, this degree of agreement is similar to that observed in Krasner et al. (1985), in which 2-MIB concentrations from approximately 20 to 80 ngyl were observed at the same FPA intensity. Although a more rigorous analysis, including isolation of the odor-producing organisms and associated microbial products, is needed to establish the cause of musty odor in the source water (Suffet et al., 1999), based on the current observation, 2-MIB is very likely to be responsible for the musty odor in the two WTPs.
To understand the effectiveness of odor removal by current treatment processes at the two water treatment plants, water samples were taken and analyzed at four different locations in each plant. The observed FPA intensity for musty odor and corresponding 2-MIB concentrations are illustrated in Fig. 8. For the Gun-Shi plant (Fig. 8b), the raw water samples and the effluent of the first process, RBCs, were both moderate in musty odor. The average FPA scales were 4.5 and 4.4 for the raw water and RBC effluent, respectively. The biological process, RBCs, seemed to have no effect on the odor reduction. The musty odor was found to disappear at the effluent of the sedimentation and filtration treatment processes. A similar trend was found in Feng-Shen plant (Fig. 8a), in which the average musty odor intensity was 6.5 for raw water and zero for the rest of the processes. It may be possible that coagulation and sedimentation processes contributed to the removal of musty odor to a certain extent. However, chlorination was employed right after the RBC process in GSWTP and before coagulation process in FSWTP. It is probable that the odor was masked by the residual chlorine present in the water. Although 2-MIB is expected to be biodegradable (Egashira et al., 1992; Rittmann et al., 1995), as shown in Fig. 8b, the concentration of 2-MIB in GSWTP only changed slightly between raw water and RBC effluent. Since the RBCs in GSWTP are designed for the removal of ammonia, the conditions may not be appropriate for 2-MIB degradation. The slight difference between raw water and RBC effluent for 2-MIB concentrations at GSWTP is similar to that observed in the FPA analysis which confirms the hypothesis that RBCs did not remove musty odor compounds in GSWTP. The 2-MIB removal efficiency at both treatment plants was on the average approximately 40–50%, as illustrated in Fig. 8. Yagi et al. (1985) reported that 2-MIB removal in a water treatment plant in Osaka, Japan, was 22% from raw water to sedimentation and the total 27% measured at the end of the rapid filtration process. Our observation results are similar (approx. 16% on average from
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
233
Fig. 8. Removal of MIB at different processes at (a) Feng-Shen, and (b) Gun-Shi water treatment plants.
raw water to sedimentation and 40% to filtration in this study) to those results by Yagi et al. (1985) for similar treatment processes. According to the laboratory dose–response (calibration) curve developed previously, the equivalent FPA musty odor intensity based on the mean 2-MIB concentration measured for each process as shown in Fig. 8. Due to the fact that odor intensity is proportional to the logarithmic odorant concentration, only an insignificant difference of the equivalent FPA musty intensity (approx. equal to 0.3 odor intensity scale) was found between the raw water and other sampling locations in both water treatment plants. Therefore, it is possible that the musty odor was not removed in treatment processes, and was masked by the residual chlorine instead. To support this hypothesis, three finished water samples, with chlorine residuals between 0.32 and 0.74 mgyl from FSWTP were measured
for their FPA intensities. An overdose (0.1 ml 10% solution) of sodium thiosulfate in pentahydrate form (Reagent Grade, Baker, USA) was then used to reduce the free chlorine in the samples. In all the finished water samples, no mustyyearthy odor was detected by the FPA panel, even though the 2-MIB concentrations were between 62 and 110 ngyl. However, for the dechlorinated finished water, FPA intensities of musty odor between 4 and 8 were observed, indicating that chlorine may mask the musty odor of 2-MIB in the finished water. Therefore, it is reasonable to expect that once the residual chlorine is exhausted in the distribution pipelines, the odor may appear again at the consumers’ taps. 5. Conclusions FPA results show that fishy and musty odors were the two major odor groups in the source
234
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
water of both Gun-Shi and Feng-Shen water treatment plants. The intensity of the two odors has a correlation with temperature. Chemical analysis results showed that both 2-MIB and geosmin were present in the source water. The experimental data of FPA intensity from the two water treatment plants closely followed the laboratory dose– response curve for 2-MIB. A good correlation between laboratory and field FPA results indicates that the musty odor of the two source water were very likely from 2-MIB. In addition, a good correlation was also found between the logarithmic 2-MIB concentration and water temperature, substantiating the importance of seasonal effect. The biological treatment process, RBCs, at Gun-Shi WTP seemed to have no effect on the reduction of musty odor. Although approximately 40–50% of 2-MIB was removed in the conventional treatment processes train at both WTPs on average, only approximately 0.3 FPA intensity scale of reduction was expected. Acknowledgments This study was supported in part by the National Science Council of the Republic of China under Grant Number NSC-89-2211-E-006-115, and in part by Taiwan Water Supply Corporation. The authors thank Professor H.H. Yeh in the Department of Environmental Engineering at National Cheng Kung University, and Professor H.D. Lai in the Department of Environmental Engineering and Health at Chia Nan University of Pharmacy and Science for their support and discussions. The authors also thank Professor C.P. Huang in the Department of Civil and Environmental Engineering at University of Delaware for his review. In addition, we appreciate the help of Mr Y.P. Lai at the Seventh District of Taiwan Water Supply Corporation for providing water quality data of the two source waters. References APHA, AWWA and WPCF. Standard Methods for the Examination of Water and Wastewater, 19th ed., Washington, DC, USA. 1995. Atasi KZ, Chen T, Huddleston JI, Young CC, Suffet IH. Factor screening for ozonating the taste- and odor-causing com-
pounds in source water at Detroit, USA. Water Sci Technol 1999;40(6):115 –122. Bruchet A. Solved and unsolved taste and odor episodes in the files of Inspector Gluzeau. Water Sci Technol 1999;40(6):15 –21. Chen SY, Li CH, Tseng IC. Growth of actinomycetes and phytoplankton of Cheng-Ching Lake in relation to odors of lake water (I). Tainan, Taiwan: Department of Environmental Engineering, National Cheng Kung University, 1982. (In Chinese). Chen SY, Li CH, Tseng IC. Growth of actinomycetes and phytoplanktons of Cheng-Chin Lake in relation to odors of lake water (II). Tainan, Taiwan: Department of Environmental Engineering, National Cheng Kung University, 1983. (In Chinese). Duguet J, Mallevialle J, Ho J, Suffet IH. Oxidation processes: chlorine and chloramines. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association, 1995. Egashira K, Ito K, Yoshiy Y. Removal of musty odor compound in drinking water by biological filter. Water Sci Technol 1992;25:307 –314. Gilloly TET, Snoeyink VL, Newcombe G, Elarde JR. A simplified method determine the powered activated carbon dose required to remove methylisobornel. Water Sci Technol 1999;40(6):59 –64. Glaze WH, Schep R, Chauncey W, Ruth EC, Zarnoch JJ, Aieta EM, Tate CH, McGuire MJ. Evaluating oxidants for the removal of model taste and odor compounds from a municipal water supply. J Am Water Works Assoc 1990;82:79 – 84. Graham MR, Summers RS, Simpson MR, Macleod BW. Modeling equilibrium adsorption of 2-methylisoborneol and geosmin in natural waters. Water Res 2000;34:2291 –2300. Jensen SE, Anders CL, Goatcher LJ, Perley T, Kenefick S, Hrudey SE. Actinomycetes as a factor in odour problems affecting drinking water from the North Saskatchewan River. Water Res 1994;28:1393 –1401. Kao SP. Analysis of odorous compounds in drinking water source, Master Thesis. Tainan City, Taiwan: Department of Environmental Engineering, National Cheng Kung University, 1998. (In Chinese). Krasner SW, McGuire MJ, Ferguson VB. Tastes and odors: the flavor profile analysis methods. J Am Water Works Assoc 1985;77:34 –39. Lalezary S, Pirbazari M, McGuire MJ. Oxidation of five earthy-musty taste and odor compounds. J Am Water Works Assoc 1986;78:62 –69. Lin TF, Fu CY. Origin and control of odors in Cheng-Chin Lake water treatment plant. Proceedings of 3rd International Workshop on Drinking Water Quality Treatment and Management Technology, Taipei, Taiwan. March 1997:3 –5. Lloyd SW, Lea JM, Zimba PV, Grimm CC. Rapid analysis of geosmin and 2-methylisoborneol in water using solid phase micro extraction procedures. Water Res 1998;32:2140 –2146.
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235 Mallevialle J, Suffet IH. Identification of taste and odor compounds in the drinking waters. Denver, Colo., USA: American Water Works Association Research Foundation and Lyonnaise des Eaux, 1987. Means EG III, McGuire MJ. An early warning system for taste and odor control. J Am Water Works Assoc 1986;78:77 –83. McGuire MJ. Advances in treatment processes to solve offflavor problems in drinking water. Water Sci Technol 1999;40(6):153 –163. Rittmann BE, Grantzer CJ, Montiel A. Biological treatment to control taste-and-odor compounds in drinking water treatment. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association, 1995. ROCEPA. Drinking water quality standard. Taiwan, ROC: Environmental Protection Administration, 1998. Rashash DMC, Goehn RC, Dietrich AM, Grizzard TJ, Parker BC. Identification and control of odorous algal metabolites. Denver, Colorado, USA: American Water Works Association Research Foundation, 1996.
235
Sklenar KS, Horne AJ. Horizontal distribution of geosmin in a reservoir before and after copper treatment. Water Sci Technol 1999;40:229 –237. Suffet IH, Khiari D, Bruchet A. The drinking water taste and odor wheel for the millenium: beyond geosmin and 2methylisoborneol. Water Sci Technol 1999;40(6):1 –13. Suffet IH, Mallevialle J. Taste-and-odor problems observed during drinking water treatment. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association Research Foundation, 1995. Suffet IH, Wable O. Removal of taste-and-odor compounds by activated carbon. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association, 1995. Watson SB, Brownlee B, Stachwill T, Hargesheimer EH. Quantitative analysis of trace levels of geosmin and MIB in source and drinking water using headspace SPME. Water Res 2000;34:2818 –2828. Yagi M, Jajino M, Matsuo U, Ashitani K, Kita T, Nakamura T. Odor problems in Lake Biwa. Water Sci Technol 1985;15:311 –321.
Correlation of musty odor and 2-MIB in two drinking water treatment plants in South Taiwan Tsair-Fuh Lin*, Jiun-Yue Wong, Hsiao-Pin Kao Department of Environmental Engineering, National Cheng Kung University, Tainan City, 70101, Taiwan, ROC Received 23 February 2001; accepted 10 September 2001
Abstract Possible odor groups and intensity, and seasonal effects were elucidated in two representative water treatment plants (WTPs), Feng-Shen and Gun-Shi, in southern Taiwan. The flavor profile analysis (FPA) was employed to determine the odor groups for the source water, while a chemical analysis, solid-phase microextraction (SPME) coupled with a gas chromatograph and mass spectrometric detector (GCyMSD), was used to concentrate and subsequently analyze the corresponding water samples. FPA results show that fishy and musty odors were the two major odor groups in the source water. Results of chemical analysis showed that 2-methyl-isoborneol (2-MIB) was present in the source water. The correlation between 2-MIB concentration and the FPA intensity of musty odor was compared with the dose–response curve generated in the laboratory by the FPA panelists. The experimental data from the two water treatment plants follow the calibration curve closely, indicating that the musty odor of the two source waters were most likely contributed from 2-MIB. In addition, there is good correlation between logarithmic 2-MIB concentration and water temperature, substantiating the importance of seasonal effect. Although approximately 40– 50% of 2-MIB removal was found in the treatment trains for the two WTPs, only an approximately 0.3 FPA intensity scale of reduction was expected. 䊚 2002 Elsevier Science B.V. All rights reserved. Keywords: Musty odor; Fishy odor; Flavor profile analysis; 2-MIB; Solid-phase microextraction
1. Introduction Drinking water quality in southern Taiwan received complaints from customers for a long time for its unpleasant odors. Previous studies (Chen et al., 1982, 1983; Lin and Fu, 1997) suggested that mustyyearthy and fishy odors were the two major odors in drinking water sources, and algae and actinomycetes were the most likely *Corresponding author. Fax: q886-6-275-2790. E-mail address: [email protected] (T.-F. Lin).
sources for the odors in a raw water source in the area. Although musty and earthy odors were reported to be present in the summer time and the fishy odor was present in winter, the chemicals and algae species responsible for the odors remain unknown. Mustyyearthy odor is a prevalent odor subgroup in drinking water, and has been reported by many researchers (Yagi et al., 1985; Means and McGuire, 1986; Jensen et al., 1994; Bruchet, 1999; Sklenar and Horne, 1999). Five organic chemicals,
0048-9697/02/$ - see front matter 䊚 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 1 . 0 1 0 4 9 - X
226
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
namely trans-1,10-dimethyl-trans-9-decalol (geosmin), 2-methyl-isoborneol (2-MIB), 2-isopropyl3-methoxy pyrazine (IPMP), 2-isobutyl-3-methoxy pyrazine (IBMP) and 2,3,6trichloroanisole (TCA) are often considered as the responsible mustyyearthy odorants (Lalezary et al., 1986; Suffet and Wable, 1995). The odor thresholds of these compounds are as low as a few ngy l in drinking water. Among these five chemicals, geosmin and 2-MIB are the most common mustyy earthy compounds (Suffet et al., 1999) and are difficult to remove by most treatment processes (Lalezary et al., 1986). Although several advanced treatment technologies are able to remove the mustyyearthy odorants, other methods, such as ozone oxidation (Glaze et al., 1990; Duguet et al., 1995; Atasi et al., 1999; McGuire, 1999) and activated carbon adsorption (Gilloly et al., 1999; Graham et al., 2000), are generally expensive compared to conventional treatment processes. Therefore, in order to develop a cost effective technology for the control of these odorants in drinking water, it is necessary to understand the background information of the odorants and their relation to environmental conditions (Means and McGuire, 1986). Although odor problems have been present in drinking water in southern Taiwan for a long time, no flavor profile analyses (FPA) and chemical analyses were conducted to identify the odor groups and odorants responsible for the odor, and no evaluation of odor removal efficiency for the current processes was performed. In this study, the odor problems in two representative water treatment plants in the area were analyzed for their odor groups and intensity using both sensory and instrumental techniques. The results were correlated with temperature change in the area. In addition, the removal efficiency for different treatment processes at the two treatment plants was also presented. 2. Site description Two water treatment plants (WTPs), Feng-Shen and Gun-Shi, were chosen in this study as both have had odor problems for a long time. As illustrated in Fig. 1, Feng-Shen and Gun-Shi WTPs
are located near Kaohsung City, the largest city in southern Taiwan. Conventional treatment processes are employed in both treatment plants. The following section provides the background information for the two plants. 2.1. Feng-Shen water treatment plant Feng-Shen Reservoir, located in Kaohsung County, Taiwan, is one of the major drinking water and industrial water sources in the Kaohsung Metropolitan area. As an off-channel reservoir, it receives water from the Tongun Creek, which is approximately 10 km away from the reservoir. Surface water from Tongun Creek is pumped into the reservoir through two raw water transfer conduits at a rate of 500 000 m3 per day. The area and storage capacity of the reservoir are equal to approximately 7.5=105 m2 and 4.0=106 m3, respectively. Feng-Shen Water Treatment Plant (FSWTP) is located on the northwestern side of Feng-Shen Reservoir. Important source water quality parameters for the plant in the year 2000 are listed in Table 1. Initially, the production rate of FSWTP was 350 000 m3 per day. As the demand of water supply rate in Kaohsung Metropolitan area increased, the production rate was raised to 500 000 m3 per day. As shown in Fig. 1, a conventional treatment train is used in the plant, with pre-chlorination, coagulation, settling, rapid sand filtration, and post-chlorination processes. Threshold odor number (TON) measurement was conducted in the water treatment plant monthly for both the raw and the finished water by the Taiwan Water Supply Corporation. The measured TON values from January 1996 to January 2000 are illustrated in Fig. 2a. The TON in the raw water was as high as 16, indicating that the odor intensity in the reservoir was strong. Although the TON values of the finished water were all below the national drinking water standard for odor (s 3) (ROCEPA, 1998), there are still common complaints from customers. Since no other sensory tests except TON were conducted at the treatment plant, no description of the odor types was reported. However, the odors were expected to include both fishy and musty, at least. In addition, since
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
227
Fig. 1. Locations and treatment trains of Feng-Shen and Gun-Shi water treatment plants.
free chlorine is present in the finished water and was not accounted for in the TON measurement, it is likely that the odors were impaired by chlorine in the finished water.
2.2. Gun-Shi water treatment plant Gun-Shi water treatment plant (GSWTP) is located right beside Tongun Creek in Pingtong
Table 1 Source water quality of Feng-Shen and Gun-Shi Water Treatment Plantsa Water treatment plant
pH
Alkalinity, mgyl CaCO3
Hardness, mgyl CaCO3
Total dissolved solid, mgyl
Total organic carbon, mgyl
Turbidity, NTU
Feng-Shen Gun-Shi
7.5"0.1 7.4"0.2
176"35 171"35
254"40 248"39
403"76 393"64
2.4"1.0 4.0"1.5
5.1"2.5 16.9 (5–170)b
a b
Monthly monitoring data (January to December of 2000) from Taiwan Water Supply Corporation (mean"S.D.). Monthly monitoring data (January to December of 2000) from Taiwan Water Supply Corporation wmean (range)x.
228
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
odor compound, which is approximately 9–42 ngy l (Watson et al., 2000). 3. Experimental methods
Fig. 2. Historical threshold odor number (TON) for the raw and finished water at (a) Feng-Shen, and (b) Gun-Shi water treatment plants.
County, Taiwan. The raw water of the plant is directly pumped from Tongun Creek, and source water quality parameters for the plant in the year 2000 are listed in Table 1. The water production rate of GSWTP is 26 000 m3 per day, which is distributed to a few nearby towns for domestic water use. Since the ammonia content in the raw water can be as high as 5 mgyl, rotating biological contactors (RBCs) were installed right after the intake of the plant to nitrify the ammonia. An average of 70% ammonia removal was observed in the process. As shown in Fig. 1, a process train including pre-chlorination, coagulation, settling, rapid sand filtration and post-chlorination is used after the RBCs. The chronological TON values for the plant are shown in Fig. 2b. It is noticed that the TON values can be as high as 18. Similar to FSWTP, the TON values in the finished water for GSWTP were all below three. The results of a preliminary analysis (Kao, 1998) indicated that 2-MIB was present in the whole treatment train, from raw water to finished water, and the concentration could be as high as 300 ngyl. This concentration is certainly much higher than the odor threshold for the musty-
Two analytical schemes, including sensory and instrumental methods were employed in this study. A sensory method, flavor profile analysis (FPA) was used to identify the odor group and intensity for the drinking water, while a gas chromatography mass spectrometric detector (GCyMSD) incorporated with solid-phase micro-extraction (SPME) concentration technique was used to analyze the possible odor compounds. A detailed description of training and applications for the FPA method can be found in the Standard Methods for Water and Wastewater (APHA et al., 1995). In this study, more than six non-smokers were trained for the panel, and at least five panelists participated in each test. Seven-point scales of 1–12, as suggested by Philadelphia Water Department of the USA and Lyonnaise des Eaux of France (Mallevialle and Suffet, 1987), were used to describe the intensity of samples based on the consensus of the panel. In addition, water from a Mili-Q water purification system (Millipore Corp., Bedford, MA, USA) was used for all the dilution of samples and standards in the laboratory. The newly developed SPME method (Lloyd et al., 1998; Watson et al., 2000) was chosen and modified as the concentration technique for extracting odorous compounds from water samples. A detailed procedure may be found in Lloyd et al. (1998) and Watson et al. (2000), only the modified part is introduced below. A commercially available fiber coated with a 30y50-mm divinylbenzeneycarboxenypolydimethylsiloxane film (No. 57348-U, Supelco, USA) was used as the adsorbent. The fiber was injected through a Tefloncoated septum and placed into the headspace of a 350-ml extraction vial with 200 ml of water sample. A water-jacketed system was designed for the extraction vial, allowing temperature control for the water sample. During the extraction process, the vial was controlled at temperatures 50"0.5 8C stirred with a Teflon bar. To reduce the impact of water vapor on the adsorption process, 50 g of NaCl (Reagent Grade, Riedel-de Haen,
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
Germany) was added to the sample, as suggested by Lloyd et al. (1998). Preliminary experiments indicated that an exposure of the fiber to the samples at 30 min produced excellent reproducibility and linearity of calibration curves. After exposure, the fiber was introduced into the injection port of an HP 6890 (Hewlett-Packard, USA) gas chromatograph (GC) equipped with an HP 5973 mass spectrometric detector (MSD) for the separation and identification of adsorbed odorous compounds. A 30-m long, 0.25-mm i.d. HP-5 capillary column coated with a 0.25-mm 5% phenyl methyl siloxane film was used in the HP GC. High purity helium ()99.999%, San-Fu Chemical Corp., Taiwan) was used as a carrier gas at a rate of 2 mly min. The injection port was operated at splitless mode with temperature controlled at 250 8C and pressure at 4.9 psi. The oven temperature was held at 60 8C for 1.0 min, followed by an increase to 250 8C at a rate of 8 8Cymin, and then held at 250 8C for 5 min. The analytical standard for the major mustyy earthy odor compound, 2-MIB, was purchased from Sigma (analytical grade, US) at a concentration of 2 mgyl, and that for geosmin was from Wako (analytical grade, Japan) at 0.1 mgyl. The retention times for 2-MIB and geosmin in this study were 9.49 and 13.32 min, with characterized massycharge (myz) ratio equal to 95.1 and 112, respectively. In addition to the retention times, the observed mass spectrum of geosmin and 2-MIB of field samples were compared with the corresponding compounds in the standard database (NBS75K.L, Hewlett-Packard, USA) for confirmation. The calibration curves for these two compounds were established for two sections, one from 5 to 100 ngyl and the other from 100 to 1000 ngy l, all with good linearity and high regression coefficients (R 2)0.995). The detection limits for this study were 8 and 6 ngyl for geosmin and 2MIB, respectively, while analytical results of the duplicates were all within "10% of difference. These detection limits are close to the threshold of the two odorous compounds, which is 9–42 ngyl for 2-MIB and 4–10 ngyl for geosmin (Watson et al., 2000).
229
Fig. 3. Dose–response curves for (a) musty odor and (b) fishy odor for laboratory FPA tests.
4. Results and discussion 4.1. FPA dose–response curves To establish the dose–response curves for the FPA tests, six replicates were conducted in the laboratory at different times for musty and fishy odors. In the tests, 2-MIB at five different concentrations between 10 and 1000 ngyl was used to simulate the field musty odorant at different odor strengths. For the fishy odor, trimethylamine at five concentrations of 0.01–1 mgyl was used. Fig. 3 shows that the FPA odor strength was proportional to the logarithmic concentration of corresponding odorant, which closely follows the dose– response relationship described by the Weber– Fechner model (Suffet and Mallevialle, 1995). Although Watson et al. (2000) reported that the odor threshold of 2-MIB is 9–42 ngyl, a lower threshold was observed in this study. Our observation of odor threshold for 2-MIB (at approx. 1 ngyl) was similar to that shown by Krasner et al. (1985). The high correlation coefficients (both larger than 0.98) indicate that the data points
230
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
the two major odor groups in the source waters, which was similar to the odor groups in the area reported by Lin and Fu (1997). The change of FPA intensity for the two odors, and corresponding temperature and observed 2-MIB concentrations were plotted in Figs. 4 and 5. As illustrated in Figs. 4 and 5, the intensity of the musty odor was strong in the summer months, i.e. in October 1999 and May 2000, and was weak in the winter months, i.e. December 1999. On the other hand, the fishy odor was the strongest in the winter month of February 2000. A similar observation of the seasonal effect on odor groups was observed by Chen et al. (1982, 1983) at a nearby water source, i.e. Cheng-Chin Lake. Chen et al. (1982, 1983) found a presence of fishy odor (up to 10–18 TON) in the wintery
Fig. 4. Seasonal changes of (a) temperature, (b) odor intensity, and (c) MIB concentration at Feng-Shen water treatment plant.
followed the dose–response curves closely, and the small standard deviations (S.D., all within one FPA scale) reflect the good reproducibility of the tested results. Laboratory dose–response curves for the FPA intensity and corresponding odorant concentration were constructed for the two odors tested, as shown in Fig. 3. These curves may be used to analyze the field data obtained later. 4.2. Seasonal change of raw water odors Samples of the raw water were analyzed for the odor in the Feng-Shen and Gun-Shi water treatment plants one to two times each month from September 1999 to May 2000. Both FPA and SPMEyGCyMSD analysis were conducted. FPA results indicated that fishy and musty odors were
Fig. 5. Seasonal changes of (a) temperature, (b) odor intensity, and (c) MIB concentration at Gun-Shi water treatment plant.
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
spring (cold) seasons, and musty odor in the summeryautumn (warm) time. In their study, the number of algae and actinomycetes were also measured. Lin and Fu (1997) analyzed the experimental data from Chen et al. (1982, 1983) to obtain the relationship between odor strength and algaeyactinomycete counts. Strong correlations between TONs and the logarithmic number of actinomycetes in warm seasons, and between TONs and logarithmic diatom numbers in cold seasons were observed. Since no relevant data on algae species were collected in this study, the correlation between algae counts and odor strengthygroups cannot be confirmed. Although the aquatic environment for Cheng-Chin Lake and the two water treatment plants are different, data collected from the Cheng-Chin Lake may be useful information for application to these two source waters. Chemical analysis showed that 2-MIB was present in the source water all year round at both treatment plants. Although geosmin was also observed in the source water, the concentrations were always below the detection limit (8 ngyl) (data not shown in the figure). Additionally, three other common mustyyearthy odor compounds, 2isopropyl-3-methoxy pyrazine (IPMP), 2-isobutyl3-methoxy pyrazine (IBMP), and 2,3,6-trichloroanisole (TCA) were not observed in the mass spectrometric detector. Therefore, only 2MIB dominated the five common musty compounds. As shown in Figs. 4 and 5, 2-MIB concentrations were found to increase as the temperature increased at both water treatment plants. The trend of the water temperature and the logarithmic 2-MIB concentrations is similar at corresponding time. The correlation between temperature and odor in the raw water was analyzed based on the observed 2-MIB concentrations and the corresponding water temperature at sampling time. Although other environmental conditions, such as water quality parameters, may affect the 2-MIB concentrations, Fig. 6 shows that 2-MIB concentrations in the two water sources were closely correlated to the water temperature. This observation suggests that temperature is one of the most
231
Fig. 6. Correlation between water temperature and MIB concentration for Feng-Shen and Gun-Shi water treatment plants.
important factors governing the production of 2MIB in the source waters. Attempts were also made to explore the relationship between fishy odor and water temperature. Although a few flagellated algae (Suffet et al., 1999) and Synura petersenii (Rashash et al., 1996) were identified as being responsible for fishy odor in water resources, we were not able to observe their presence in the source water. Rashash et al. (1996) separated a fishy odor compound, 2trans,4-cis,7-cis-decatrienal, from the Synura petersenii cultures. However, this chemical is not commercially available. Instead, three of other fishy odor related aldehydes, including n-hexanal, n-heptanal, and trans–trans-2,4-heptadienal (Suffet et al., 1999) were analyzed. None of the fishy odorants were observed in this study. The FPA intensity for fishy odor was further analyzed with the corresponding water temperature in the same approach as that for musty odor. However, no simple correlation can be found in the analysis. Indeed, environmental parameters other than water temperature may affect the generation of fishy odor compounds in the source waters. A more rigorous monitoring and analysis may be needed to find the relationship between fishy odor and environmental conditions. 4.3. Comparison between laboratory and field results The correlation of 2-MIB concentration and FPA intensity of musty odor was compared with the dose–response curve generated in the laboratory
232
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
4.4. Odor removal at Feng-Shen and Gun-Shi water treatment plants
Fig. 7. Correlation of FPA intensities and MIB concentrations in laboratory and field samples. (a) Feng-Shen WTP (b) GunShi WTP.
by the FPA panelists. As shown in Fig. 7, the experimental data from the two water treatment plants followed the dose–response curve closely, and all data points fell within the 95% prediction interval of the regression curve. In fact, for all experimental data, only one FPA scale of discrepancy from the calibration curve was found at most which suggested that the musty odor of the two source waters was very likely from 2-MIB. A similar correlation between 2-MIB and FPA intensity in laboratory prepared standards and field samples was observed by Krasner et al. (1985) at the Metropolitan Water District of South California, USA. Although the correlation between FPA and MIB concentrations in laboratory and field samples is promising, the level of this agreement should not be over-expected. For example, at an FPA intensity of 6 in Fig. 7, 2-MIB concentrations range from approximately 12 to 60 ngyl. In fact, this degree of agreement is similar to that observed in Krasner et al. (1985), in which 2-MIB concentrations from approximately 20 to 80 ngyl were observed at the same FPA intensity. Although a more rigorous analysis, including isolation of the odor-producing organisms and associated microbial products, is needed to establish the cause of musty odor in the source water (Suffet et al., 1999), based on the current observation, 2-MIB is very likely to be responsible for the musty odor in the two WTPs.
To understand the effectiveness of odor removal by current treatment processes at the two water treatment plants, water samples were taken and analyzed at four different locations in each plant. The observed FPA intensity for musty odor and corresponding 2-MIB concentrations are illustrated in Fig. 8. For the Gun-Shi plant (Fig. 8b), the raw water samples and the effluent of the first process, RBCs, were both moderate in musty odor. The average FPA scales were 4.5 and 4.4 for the raw water and RBC effluent, respectively. The biological process, RBCs, seemed to have no effect on the odor reduction. The musty odor was found to disappear at the effluent of the sedimentation and filtration treatment processes. A similar trend was found in Feng-Shen plant (Fig. 8a), in which the average musty odor intensity was 6.5 for raw water and zero for the rest of the processes. It may be possible that coagulation and sedimentation processes contributed to the removal of musty odor to a certain extent. However, chlorination was employed right after the RBC process in GSWTP and before coagulation process in FSWTP. It is probable that the odor was masked by the residual chlorine present in the water. Although 2-MIB is expected to be biodegradable (Egashira et al., 1992; Rittmann et al., 1995), as shown in Fig. 8b, the concentration of 2-MIB in GSWTP only changed slightly between raw water and RBC effluent. Since the RBCs in GSWTP are designed for the removal of ammonia, the conditions may not be appropriate for 2-MIB degradation. The slight difference between raw water and RBC effluent for 2-MIB concentrations at GSWTP is similar to that observed in the FPA analysis which confirms the hypothesis that RBCs did not remove musty odor compounds in GSWTP. The 2-MIB removal efficiency at both treatment plants was on the average approximately 40–50%, as illustrated in Fig. 8. Yagi et al. (1985) reported that 2-MIB removal in a water treatment plant in Osaka, Japan, was 22% from raw water to sedimentation and the total 27% measured at the end of the rapid filtration process. Our observation results are similar (approx. 16% on average from
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
233
Fig. 8. Removal of MIB at different processes at (a) Feng-Shen, and (b) Gun-Shi water treatment plants.
raw water to sedimentation and 40% to filtration in this study) to those results by Yagi et al. (1985) for similar treatment processes. According to the laboratory dose–response (calibration) curve developed previously, the equivalent FPA musty odor intensity based on the mean 2-MIB concentration measured for each process as shown in Fig. 8. Due to the fact that odor intensity is proportional to the logarithmic odorant concentration, only an insignificant difference of the equivalent FPA musty intensity (approx. equal to 0.3 odor intensity scale) was found between the raw water and other sampling locations in both water treatment plants. Therefore, it is possible that the musty odor was not removed in treatment processes, and was masked by the residual chlorine instead. To support this hypothesis, three finished water samples, with chlorine residuals between 0.32 and 0.74 mgyl from FSWTP were measured
for their FPA intensities. An overdose (0.1 ml 10% solution) of sodium thiosulfate in pentahydrate form (Reagent Grade, Baker, USA) was then used to reduce the free chlorine in the samples. In all the finished water samples, no mustyyearthy odor was detected by the FPA panel, even though the 2-MIB concentrations were between 62 and 110 ngyl. However, for the dechlorinated finished water, FPA intensities of musty odor between 4 and 8 were observed, indicating that chlorine may mask the musty odor of 2-MIB in the finished water. Therefore, it is reasonable to expect that once the residual chlorine is exhausted in the distribution pipelines, the odor may appear again at the consumers’ taps. 5. Conclusions FPA results show that fishy and musty odors were the two major odor groups in the source
234
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235
water of both Gun-Shi and Feng-Shen water treatment plants. The intensity of the two odors has a correlation with temperature. Chemical analysis results showed that both 2-MIB and geosmin were present in the source water. The experimental data of FPA intensity from the two water treatment plants closely followed the laboratory dose– response curve for 2-MIB. A good correlation between laboratory and field FPA results indicates that the musty odor of the two source water were very likely from 2-MIB. In addition, a good correlation was also found between the logarithmic 2-MIB concentration and water temperature, substantiating the importance of seasonal effect. The biological treatment process, RBCs, at Gun-Shi WTP seemed to have no effect on the reduction of musty odor. Although approximately 40–50% of 2-MIB was removed in the conventional treatment processes train at both WTPs on average, only approximately 0.3 FPA intensity scale of reduction was expected. Acknowledgments This study was supported in part by the National Science Council of the Republic of China under Grant Number NSC-89-2211-E-006-115, and in part by Taiwan Water Supply Corporation. The authors thank Professor H.H. Yeh in the Department of Environmental Engineering at National Cheng Kung University, and Professor H.D. Lai in the Department of Environmental Engineering and Health at Chia Nan University of Pharmacy and Science for their support and discussions. The authors also thank Professor C.P. Huang in the Department of Civil and Environmental Engineering at University of Delaware for his review. In addition, we appreciate the help of Mr Y.P. Lai at the Seventh District of Taiwan Water Supply Corporation for providing water quality data of the two source waters. References APHA, AWWA and WPCF. Standard Methods for the Examination of Water and Wastewater, 19th ed., Washington, DC, USA. 1995. Atasi KZ, Chen T, Huddleston JI, Young CC, Suffet IH. Factor screening for ozonating the taste- and odor-causing com-
pounds in source water at Detroit, USA. Water Sci Technol 1999;40(6):115 –122. Bruchet A. Solved and unsolved taste and odor episodes in the files of Inspector Gluzeau. Water Sci Technol 1999;40(6):15 –21. Chen SY, Li CH, Tseng IC. Growth of actinomycetes and phytoplankton of Cheng-Ching Lake in relation to odors of lake water (I). Tainan, Taiwan: Department of Environmental Engineering, National Cheng Kung University, 1982. (In Chinese). Chen SY, Li CH, Tseng IC. Growth of actinomycetes and phytoplanktons of Cheng-Chin Lake in relation to odors of lake water (II). Tainan, Taiwan: Department of Environmental Engineering, National Cheng Kung University, 1983. (In Chinese). Duguet J, Mallevialle J, Ho J, Suffet IH. Oxidation processes: chlorine and chloramines. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association, 1995. Egashira K, Ito K, Yoshiy Y. Removal of musty odor compound in drinking water by biological filter. Water Sci Technol 1992;25:307 –314. Gilloly TET, Snoeyink VL, Newcombe G, Elarde JR. A simplified method determine the powered activated carbon dose required to remove methylisobornel. Water Sci Technol 1999;40(6):59 –64. Glaze WH, Schep R, Chauncey W, Ruth EC, Zarnoch JJ, Aieta EM, Tate CH, McGuire MJ. Evaluating oxidants for the removal of model taste and odor compounds from a municipal water supply. J Am Water Works Assoc 1990;82:79 – 84. Graham MR, Summers RS, Simpson MR, Macleod BW. Modeling equilibrium adsorption of 2-methylisoborneol and geosmin in natural waters. Water Res 2000;34:2291 –2300. Jensen SE, Anders CL, Goatcher LJ, Perley T, Kenefick S, Hrudey SE. Actinomycetes as a factor in odour problems affecting drinking water from the North Saskatchewan River. Water Res 1994;28:1393 –1401. Kao SP. Analysis of odorous compounds in drinking water source, Master Thesis. Tainan City, Taiwan: Department of Environmental Engineering, National Cheng Kung University, 1998. (In Chinese). Krasner SW, McGuire MJ, Ferguson VB. Tastes and odors: the flavor profile analysis methods. J Am Water Works Assoc 1985;77:34 –39. Lalezary S, Pirbazari M, McGuire MJ. Oxidation of five earthy-musty taste and odor compounds. J Am Water Works Assoc 1986;78:62 –69. Lin TF, Fu CY. Origin and control of odors in Cheng-Chin Lake water treatment plant. Proceedings of 3rd International Workshop on Drinking Water Quality Treatment and Management Technology, Taipei, Taiwan. March 1997:3 –5. Lloyd SW, Lea JM, Zimba PV, Grimm CC. Rapid analysis of geosmin and 2-methylisoborneol in water using solid phase micro extraction procedures. Water Res 1998;32:2140 –2146.
T.-F. Lin et al. / The Science of the Total Environment 289 (2002) 225–235 Mallevialle J, Suffet IH. Identification of taste and odor compounds in the drinking waters. Denver, Colo., USA: American Water Works Association Research Foundation and Lyonnaise des Eaux, 1987. Means EG III, McGuire MJ. An early warning system for taste and odor control. J Am Water Works Assoc 1986;78:77 –83. McGuire MJ. Advances in treatment processes to solve offflavor problems in drinking water. Water Sci Technol 1999;40(6):153 –163. Rittmann BE, Grantzer CJ, Montiel A. Biological treatment to control taste-and-odor compounds in drinking water treatment. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association, 1995. ROCEPA. Drinking water quality standard. Taiwan, ROC: Environmental Protection Administration, 1998. Rashash DMC, Goehn RC, Dietrich AM, Grizzard TJ, Parker BC. Identification and control of odorous algal metabolites. Denver, Colorado, USA: American Water Works Association Research Foundation, 1996.
235
Sklenar KS, Horne AJ. Horizontal distribution of geosmin in a reservoir before and after copper treatment. Water Sci Technol 1999;40:229 –237. Suffet IH, Khiari D, Bruchet A. The drinking water taste and odor wheel for the millenium: beyond geosmin and 2methylisoborneol. Water Sci Technol 1999;40(6):1 –13. Suffet IH, Mallevialle J. Taste-and-odor problems observed during drinking water treatment. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association Research Foundation, 1995. Suffet IH, Wable O. Removal of taste-and-odor compounds by activated carbon. In: Suffet IH, Mallevialle J, Kawczynski E, editors. Advances in taste-and-odor treatment and control. Denver, Coloroda, USA: American Water Works Association, 1995. Watson SB, Brownlee B, Stachwill T, Hargesheimer EH. Quantitative analysis of trace levels of geosmin and MIB in source and drinking water using headspace SPME. Water Res 2000;34:2818 –2828. Yagi M, Jajino M, Matsuo U, Ashitani K, Kita T, Nakamura T. Odor problems in Lake Biwa. Water Sci Technol 1985;15:311 –321.