CALGON CARBON CORPORATION
|
|
|
|
A Case Study of Geosmin in Philadelphia's Water |
||
|
Two episodes of unacceptable tastes and odors, which corresponded with levels of geosmin in the water that were much higher than the background level of 20 ng/L or less, occurred in Philadelphia during 1985. The source of one episode was found to be a localized bed of algae in the Schuylkill River. An existing taste- and odor-control program, which utilizes instrumental and sensory analyses, was largely responsible for the effective management of the episodes. Hydraulic strategies were used to reduce the treatment plant's intake of geosmin, and powdered activated carbon in the treatment train further reduced the geosmin level. Dilution of this treated water with other finished waters in the distribution system also diminished the impact of geosmin on the taste and odor of Philadelphia's water. The occurrence during 1985 of geosmin in the Schuylkill River provided the Philadelphia, PA, Water Department with a historically unique experience in taste and odor control. The geosmin occurred at levels that greatly affected the quality of the water and that consumers found disagreeable. Steps were taken to locate its source and mitigate its impact. Although the Schuylkill River and Philadelphia's drinking water may have, at times, a weak earthy or musty character, the contribution of geosmin to that characteristic was not previously known. The earthy or musty taste and odor problem caused by geosmin for water treatment utilities is well documented in the literature.1-6 However, it was not until the mid-1960s, with the use of gas chromatography, 7,8 that the occurrence of geosmin could be understood. Geosmin is found in surface waters worldwide and is associated with aquatic populations of actinomycetes and blue-green algae. 1,2,6-10 Adsorption with activated carbon, 2,5,8,11 either in the powdered form or by use in filtration, has been considered the most effective treatment for reducing geosmin levels. The ability of the Philadelphia Water Department to control geosmin and mitigate its impact on the taste and odor quality of the finished water was determined by two factors. First, the availability of information in the literature was a prerequisite to the selection of a treatment and to the confirmation of the source of geosmin. Second, the presence within the department of a taste- and odor-control program with instrumental and sensory analysis capabilities was essential for making decisions, confirming observations, and determining the source of the geosmin. This article describes the occurrence of geosmin in the Schuylkill River, its impact on the taste and odor of Philadelphia's water, and the management of its control. The authors also discuss the implications of their experiences. The Belmont Plant The Philadelphia Water Department's Belmont treatment plant, originally constructed in the early 1900s, was upgraded most recently during 1965. It is located about 0.9 mi (1.5 km) west of its source water intake on the Schuylkill River. The Belmont plant supplies approximately 20 percent of the water used in Philadelphia and has an average daily output of 63 mgd (239 ML/d). There are two other treatment plant s in Philadelphia, which are located on the Schuylkill and Delaware rivers. The Schuylkill River begins in the Appalachian Mountains of eastern Pennsylvania and flows southeast through coal mining regions, agricultural lands, and localized urban areas to its confluence with the Delaware River in Philadelphia. In 1978, a significant stretch of the river was designated a component of the Pennsylvania Scenic River System. The heaviest usage is along its final 24 mi (39 km), where it passes through Norristown, Conshohocken, and Philadelphia. In Philadelphia the Schuylkill River has average flows of 969 to 3230 mgd (3668 to 12 226 ML/d) and has flows of less than 645 mgd (2445 ML/d) on occasion.
A localized bed of blue-green algae in the Schuylkill River was the source of unacceptable tastes and odors. The Belmont plant's intake is situated on the west bank of the Schuylkill River in Philadelphia. A 140-mgd (531-ML/d) pumping station lifts the river water approximately 249 ft (76m) throughout the day to two raw water storage basins, which have a combined capacity of 80 mil gal (303 ML). The water then flows by gravity through the plant, where it is conventionally treated using flocculation with alum, settling, and rapid sand filtration. chlorine is added for algae control at the influent to the raw water basins, for process control at the rapid mix, and for breakpoint chlorination at the sand filter influent, with post-ammoniation for a chloramine residual of 2-25 mg/L. The plant has the capability of applying chlorine dioxide or powdered activated carbon to the raw water basins and rapid mix. Finished-water storage of 40 mil gal (152 ML) and the raw water storage previously mentioned allow the plant intake to be closed for up to 24 hours. Interconnections with the other treatment plants can also be used to reduce demand on Belmont or to dilute its water with other waters. However, this can only take place in a portion of the Belmont distribution system. The plant's storage and the distribution system interconnections allow considerable flexibility in the management of short-term river quality problems, such as chemical spills.
Pennsylvania's Schuylkill River is subjected to uses ranging from sailing to discharges from coal mines and agricultural lands before it flows into the Delaware River in downtown Philadelphia.
Taste and Odor Control The Philadelphia Water Department has a taste- and odor-control program that includes instrumental and sensory analyses of source and treated waters. The program was established during 1982 for the purpose of determining the nature and extent of tastes and odors in Philadelphia's water. It also involves the development of the analytical methods. As a part of the program, known taste- and odor-producing compounds, such as geosmin, have been tested. This effort has been coordinated, in part, with an AWWA Research Foundation project on the treatment of tastes and odors in water.* Instrumental analysis incorporates closed-loop stripping (CLSA)4,12,13 for the concentration of volatile and semi-volatile organics from 0.26-gal (1-L) samples of water. Detection involves capillary gas chromatography and flame ionization detection. Gas chromatography (GC) and mass spectrometry (MS) provide confirmation of individual compounds that have low detection limits (2-5 ng/L for geosmin). Sensory analysis uses flavor profile analysis (FPA), a method that has been used for more than 30 years in the food and beverage industry but was more recently adapted for water by the Metropolitan Water District of Southern California. 12,13 The analysis is performed by a four-member panel of trained flavor and aroma testers, who determine the various types and intensities of tastes and odors in a water sample by panel consensus. Taste testing is done on treated-water samples only. Reference standards are used to train the panelists to recognize tastes and odors. Geosmin has an earthy taste and odor and can be detected at 10 ng/L or less in distilled, carbon-filtered water. Flavor profile analysis is being researched in detail under the AWWARF project. The panel using FPA in Philadelphia has found the drinking water to have a typical, weak, chlorous and musty characteristic. This is an improvement over the source waters' array of odors, which are characteristic of decaying and fresh vegetation, as well as grassy, fishy, septic, and earthy. The possible contribution of geosmin to the quality of Philadelphia's water had not been determined prior to 1985. Occurrence of Geosmin in the Schuylkill River Geosmin has been detected at background levels of 20 ng/L or less in the Schuylkill and Delaware rivers at the intakes to the three treatment plants in Philadelphia and throughout the Schuylkill River's watershed. Although the same levels are found in the plant's treated waters, indicating that conventional treatment has no significant effect on geosmin, the typical taste and odor quality of the water does not appear to be adversely affected. Two separate events of much higher levels of geosmin occurred in April and September 1985 (Figure 1), and because of the earthy characteristics of geosmin, these episodes caused unacceptable tastes and odors at the Belmont plant. The first episode was unexpected, lasted only 10 days, and left many questions unanswered. The FPA and CLSA-GC data included many anomalies. The second episode lasted 20 days; by the twelfth day, however, the situation was under control and the source of the geosmin had been determined.
Figure 1. Geosmin
levels and water flows in the Schuylkill River during 1985 The high levels of geosmin that occurred during April were found throughout the Schuylkill River's watershed. Geosmin was found at 108 and 57 ng/L at 13 and 36 mi (21 and 58 km), respectively, above the Belmont plant's intake. Other treatment plants, including the Queen Lane plant in Philadelphia, which is located 1.8 mi (3 km) upstream from the Belmont plant, were affected. The Belmont plant's treated water contained more than 70 ng/L of geosmin for five days. Levels then dropped below 50 ng/L, and a rainfall that increased river flows by 2325 mgd (8800 ML/d) ended the episode. It is interesting to not that these high levels of geosmin occurred simultaneously with a diatom (Cyclotella) bloom and were preceded by a period of relatively high levels of the diatom throughout the watershed. Diatoms have not been shown to directly produce geosmin, nor do the authors have evidence to suggest that any direct correlation exists. Nevertheless, although the two events appeared to be only coincidental, it is possible that an indirect relationship existed. 15
Figure 2. Geosmin levels in the Belmont treatment plant's source and treated waters during September 1985 The high levels of geosmin that occurred during September changed from day to day (Figure 2) and from hour to hour (Figure 3). The geosmin level reached 70 ng/L in the Schuylkill River during early September for about two days and then fell to around 40 ng/L. Seven days later it rose to more than 100 ng/L for about two days. Six days later the level rose again to 80-100 ng/L and remained high for five days until high river flows of 6460 mgd (24 450 ML/d), which were caused by Hurricane Gloria, ended the episode.
Figure 3. Hourly changes in the geosmin levels in the Schuylkill River during September 1985 During the September episode, high levels of geosmin were not found throughout the watershed but only within 0.6 mi (1km) upstream of the Belmont plant's intake. These high levels coincided with a bed of algae around Peter's Island the the west bank of the river (Figure 4).
Figure 4. Occurrence of blue-green algae and high levels of geosmin in the Schuylkill River during Spetember 1985 The algae found around Peter's Island were attached to the sandy-silty river bottom, and colonies of algae could be observed in the shallow water. The heavy growth produced floating masses, which began to appear at noon every day, intensified by 3:00 p.m., and floated downstream overnight. Geosmin levels upstream of the floating masses were found to be background levels - around 10 ng/L. The highest levels of geosmin were found between 3:00 and 5:00 p.m., when the floating masses were most intense (Figure 3). The authors believe that this fluctuation was representative of the event. Lower levels of geosmin were found (4.0 ft 1.2m) below the water surface, where only scattered segments of algae could be collected. A sample of the algae in river water was tested by CLSA-GC and was found to contain geosmin at the parts-per-billion level and to give off the indicative earthy odor. Subsequent, unialgal cultures of a blue-green alga that made up a major portion of the floating masses were isolated and found to produce an earthy odor. Further investigations indicated that the alga does indeed produce geosmin. These findings and the occurrence of high levels of geosmin in the river water containing the floating algae confirmed that a blue-green alga was the cause of the episode in September. The blue-green alga that made up a large portion of the floating masses and was associated with geosmin was a member of the Oscillatoriaceae family. According to Drouet's classification, 16,17 the alga may be Schizothrix mexicana. Traditional water supply manuals18 appear to identify it as Oscillatoria (possibly princeps). Both Oscillatoria and floating masses of Schizothrix have been associated with geosmin. 1,2,6,8,10 Floating masses containing this same alga have also been observed in a canal in Philadelphia upstream from the Belmont intake and fed by the Schuylkill River. The occurrence of high levels of geosmin during April and September coincided with drought conditions. On April 26, 1985, the governor of Pennsylvania proclaimed a drought emergency for 16 counties in the eastern part of the state. The Delaware River Basin Commission subsequently declared a drought emergency, and restrictions on water use were implemented in the four states of the basin. Water flows in the Schuylkill River reached a low of around 387 mgd (1467 ML/d) during the last week in April when the first geosmin episode occurred (Figure 1). A low flow of around 258 mgd (978 ML/d) at the end of July had no geosmin episode associated with it. A third low flow of 258 mgd (978 ML/d) during September coincided with the second geosmin episode. At this time, geosmin levels in the river were as high as 100 ng/L. If the river flows had been more typical for September, at 1098 mgd (4157 ML/d), then the 100 ng/L might have been only 25 ng/L. Hurricane Gloria and subsequent rainfall permitted drought restrictions to be withdrawn by November. The authors have developed a possible scenario for the geosmin episode in September. As a result of the lower than normal river flows that had persisted for several months, a population of the blue-green alga was able to colonize a section of the Schuylkill River. The water elevation in that area was less than 4.9 ft (1.5 m), and the area was away from the main flow. According to the literature, 19,20 blue-green algae can produce intracellular gas vacuoles or trap air bubbles within their mats. Either vacuoles or oxygen bubbles may play a role in the vertical movement of algae, which occurs during the day, along with photosynthesis. Portions of algal growth were carried to the surface of the river and accumulated daily into floating masses. The highest levels of geosmin were associated with the floating masses in the river. Changes in river currents could have compounded the problem by creating more disturbance. In fact, the authors have some evidence to suggest that three conservative releases from an upstream reservoir, which preceded each peak in the geosmin level during September, may have created such a disturbance. Very high river flows, however, serve to scour the river's sandy-silty bottom and prevent population of aquatic life from permanently establishing itself. Control of the Geosmin Episodes Some type of action was needed to mitigate the impact of the high levels of geosmin on the taste and odor quality of the water. Customer complaints described the water as rusty, muddy, and dirty. When the geosmin level was greater than 45 ng/L, there were many customer complaints (Table 1). When the level was below 30 ng/L, there were only one or two, if any, complaints. The indirect relationship between customer complaints and the geosmin level on any given day made it difficult to determine what an acceptable level of geosmin would be. Customers respond to changes in taste and odor, to the intensity of a taste and odor, or to the persistence of a taste and odor. In addition, because of deviations in travel time from the plant and the mixing of waters in the distribution system, customers in different areas receive different waters.
The difficulty in determining an acceptable level of geosmin was further compounded by its sensory characteristics. Geosmin causes sensory fatique more readily than other odorants, and its intensities in river and treated water were lower than expected (Table 1). Although flavor profile analysis easily identified geosmin as the cause of both of the taste and odor episodes, the FPA results could not be used to direct management decisions. Since removal of the geosmin would be costly - perhaps impossible - it was essential that an acceptable level of geosmin in treated water be determined. This target level was eventually set at 30 ng/L. Treatment then depended on how far above this level the geosmin level actually was. it is interesting to note that other utilities have a much lower target level for geosmin.4 Analysis be CLSA-GC was utilized to located the source of the high levels of geosmin and to direct treatment decisions. A normal working day for an analyst resulted in three analyses. Therefore, the priority given to samples was critical and changed almost daily as new developments unfolded. Because geosmin levels fluctuated throughout the day, composite sampling was necessary. Offset sampling was used to obtain removals of geosmin throughout the treatment plant. These sampling protocols were developed as a result of anomalies in the data for the April episode that made that data difficult to interpret. Three control strategies were considered to maintain geosmin levels in the treated water below 30 ng/L. The treatment strategy involved using powdered activated carbon (PAC). The hydraulic strategies involved changes in intake pumping an dilution of water treated at the Belmont plant with other waters in the distribution system. Source control was a strategy aimed at the bed of algae in the Schuylkill River. Powdered activated carbon adsorption was used immediately during both episodes, since the literature5,11 indicated that PAC would be more effective than chlorine or chlorine dioxide. Doses of PAC ranging from 6 to 42 mg/L were fed at the rapid mix. The application of chlorine was moved forward in the treatment process, since it could have reduced the adsorption capacity of the PAC.11 The data for the September episode (Table 2) indicated that an average of less than 43-percent removal of geosmin was observed with 12 mg PAC/L, 50-percent removal with 30 mg PAC/L, and 73-percent removal with 42 mg PAC/L. These removals can be expected to vary from time to time and with different waters.11 Other studies found that lower amounts of PAC were just as effective.2,11 The cost of PAC treatment at the Belmont plant during September was $ 53 400.
The hydraulic strategies were developed during the September episode. Because the highest levels of geosmin were found during the afternoon, hydraulic changes could be made to ensure that water containing only the lowest possible levels of geosmin would be used (Figure 3). The Belmont plant's intake was closed down between 8:00 a.m. and 10:00 p.m. on four consecutive days. The combined effects of storage and interconnection of the distribution system with those of the other plants permitted this pumping strategy. This appeared to reduce geosmin levels in the raw water basins below the river levels by 10-20 percent, and with and average 73-percent removal using 42 mg PAC/L, the plant was able to obtain treated water with geosmin levels as much as 84 percent below levels in the river. Dilution of the Belmont plant's treated water with other waters in the distribution system helped to control the impact of geosmin, and customer complaints were almost non-existent 12 days into the September episode. Strategies for source control, including copper sulfate treatment and physical removal of the algae, were discussed. Copper sulfate did not appear to be a good alternative for the Schuylkill River. Physical removal, which involved draging the sandy bottom prior to an increase in river flow, appeared to be feasible. However, a source control strategy became unnecessary when the hurricane raised the river flow 2500 percent in two days and scoured the river of algae. Conclusion The background level of geosmin in Philadelphia's water was determined to be 20 ng/L or less. However, for the first documented time in history, levels rose more than 100 ng/L, resulting in two episodes of unacceptable taste and odor. As a result of these episodes, the authors established that geosmin concentrations exceeding 30 ng/L should be avoided in Philadelphia's drinking water. The contribution of a background level of geosmin to the typical tastes and odors of treated water is unknown and will be further investigated. An ongoing taste and odor program was responsible for identifying geosmin as the cause of the episodes and for directing the strategies used to mitigate its impact on taste and odor quality. The essential analytical tool was CLSA-GC, and its results determined the source of the geosmin in the river and the success of the control strategies. More than 60 CLSA-GC and 20 FPA analyses were made during the September episode. Since it can test up to four times more samples, more economically, than CLSA-GC, FPA would have been the preferred analytical tool. However, more work is needed before the sensory characteristics of geosmin and its possible interaction with other odorants in the water can be understood and reproducible sensory method for the detection of geosmin can be developed. The geosmin episode at the Belmont plant in September was associated with a blue-green alga in the Schuylkill River. It represents the first documented, localized occurrence of unusual tastes and odors in Philadelphia. The drought conditions of the region may have provided the opportunity for such a source of geosmin to develop. The lower than normal river flows undoubtedly resulted in less dilution and in higher levels of geosmin. A heavy growth of attached algae produced floating masses of the blue-green alga only during the afternoon. This highlights the importance of establishing a watershed monitoring program that utilizes more than planktonic algae sampling. Geosmin levels were reduced by using a treatment strategy and hydraulic changes. Changes in intake pumping resulted in lower levels of geosmin in the raw water basins than in the river. Subsequently, 42 mg PAC/L further reduced the geosmin by about 73 percent. Interconnection with the other treatment plants in Philadelphia resulted in some dilution of the geosmin concentration during distribution and in a lower demand on the Belmont plant. This reduced overall treatment costs and reduced the number of customers exposure to the higher concentrations of geosmin. Since geosmin has been documented as an important constituent of Philadelphia's source waters and watersheds, it will be monitored routinely. Future studies will examine further the sources, characteristics, and treatability of geosmin in Philadelphia's water. Acknowledgments The authors acknowledge the assistance of James Santo (supervisor of the organics testing laboratory), Marilyn Hinton (coordinator of CLSA-GC analysis), Irene Taylor (coordinator of FPA), Roy Maddrey (coordinator's assistant), and Richard Rogers (aquatic biologist). Assistance in treating the episode and locating its source was provided by Michael Pickel (Belmont plant engineer), George Izaguirre (microbiologist with The Metropolitan Water District of Southern California) isolated the blue-green alga and examined it for geosmin production. Frank Acker (biologist with the Academy of Natural Sciences in Philadelphia) identified the alga. References
About the authors: Gary a Burlingame is a project biologist for the Bureau of Laboratory Services, Philadelphia Water Department, 4290 Ford Rd., Philadelphia, PA 19131. A member of AWWA, he has been involved in the cooperative study of tastes and odors in water conducted by the AWWA Research Foundation. Since 1980, Roger M. Dann has been manager of the Belmont Treatment Plant, Philadelphia Water Department, 4300 Ford Rd., Philadelphia, PA 19131. Geoffrey L. Brock is acting director of the Bureau of Laboratory Services, Philadelphia Water Department. MARCH 1986 G.A.BURLINGAME ET AL Reprinted
and copyrighted as part of |
