The city of Rochester, N.Y., takes its water supply from Hemlock Lake, a small city-owned lake located about 30 miles south of Rochester, on the western edge of New York’s wine country. Unfiltered until 1993, the city has for the past 13 years operated a 37 million gal per day direct filtration plant on the lake’s northern shore. Given the city’s ownership and control of a large portion of the 60-sq mile watershed, including a large setback of undeveloped wooded acreage around the entire perimeter of the lake, the high quality of this near-pristine source has long been a bragging right of local citizens.
To the surprise of the city’s water treatment personnel, in the summer of 2004, Hemlock Lake developed its first noteworthy taste and odor (T&O) problem. Prior to then, occurrences of off-flavor water from Hemlock Lake were very rare, and when they did occur, they were of low intensity, short duration, and generally not noticed by customers. The 2004 event, however, persisted for about 10 weeks, with the water being characterized by numerous customers as having a musty and/or earthy flavor.
Two naturally occurring organic compounds, geosmin and 2-methylisoborneol (MIB), are the most cited perpetrators of musty/earthy flavor. At Hemlock Lake, limited water sampling during the 2004 T&O event indicated that geosmin levels were as high as 16 parts per billion, far above the taste threshold for this compound. MIB levels were negligible.
Customer reaction to the 2004 T&O event presented a wake-up call to the Hemlock filtration plant staff. Never before had they confronted a treatment problem of this nagging persistence. As investigations into the possible cause(s) of the event proceeded, so did the fears that musty/earthy flavors could become an annually recurring event. On that basis, upper management support was soon granted to find a treatment solution.
According to the AWWA, musty/earthy flavors cause more problems in public water supplies than any other type of T&O problem, in part because the responsible compounds are among the most difficult to treat. There are three strategies to consider for dealing with this problem.
Strategy A: Treat the source water. Techniques such as aeration, chemical precipitation, application of an algaecide and artificial circulation are sometimes used to treat source water in order to reduce the growth of algae and macrophytes, thus reducing production of off-flavor compounds. Each of these techniques has limitations, and none guarantee the desired affect. As a result, none of these four methods were implemented at Hemlock Lake.
Strategy B: Destroy T&O compounds using oxidation. Chemical oxidants such as chlorine, potassium permanganate and ozone can convert complex organic T&O compounds into simpler, more-inert byproducts. The effectiveness of oxidation generally depends on the type of oxidant(s) used, the rate of application and the contact time allowed. Several factors limit how much oxidant can be applied, including concerns about the formation of undesirable chemical byproducts, upper-end limits on residuals, and costs associated with storing (or generating) and applying these powerful oxidizing agents. In the case of potassium permanganate, excessive feed rates can also result in a pink colorization of the treated water.
Both chlorine and potassium permanganate have been used throughout the years as pretreatment additives at Hemlock Lake, with either compound added to the flow at the point-of-intake in the lake. The primary intent of preoxidation has been to enhance particle removal in the filtration process, and more recently, to protect against zebra mussel colonization within the intake piping. Upon the onset of the T&O problem, it was expected that one or both compounds might effectively control the flavor problem. It was soon determined, however, that neither chlorine nor permanganate was effective at oxidizing the T&O compounds, at least at the upper limit of application rates that were employed.
Strategy C: Adsorption by activated carbon. Adsorption by activated carbon has been widely demonstrated to remove organic T&O compounds. Activated carbon, in both powdered and granular form, is an extremely high-surface-area product that is created when pure carbon sources such as bituminous coal or coconut shells are exposed to extremely high temperatures. As treatment progresses, the activated carbon adsorbs organic compounds from the water that is being treated.
As a general rule of thumb, powered activated carbon (PAC) -based systems are most appropriate for systems requiring PAC doses below 12,000 lbs of PAC per year per million gallons per day of water treated. For applications requiring PAC loading higher than that, adsorption with granular activated carbon (GAC) tends to be the more cost-effective option.
At Hemlock Lake, PAC use was ruled out for several reasons. As a direct filtration plant with no sedimentation step available, the contact time was considered insufficient for PAC to work well. Offsetting the contact time limitation with increased dosage was not a viable solution because the added solids load on the filters was certain to plug them prematurely. Finally, the facility had no storage space or feed facilities for the powdered carbon, and the costs and logistics of adding these only added to the impracticality.
Unlike PAC, adsorption with GAC is carried out in a dynamic, online process, using a down-flow (either a gravity flow filter or a pressure vessel) fixed bed adsorption system that has been loaded with the granular media. When the GAC is spent, the system is taken offline and the media is removed for disposal or reactivation.
When GAC is used as a filtration media, it is generally substituted for existing media such as anthracite coal. The coal is simply removed and replaced by a layer of GAC. The GAC is then removed and replaced in-kind whenever its adsorption capacity is consumed. The change-out interval for GAC depends on the total organic load in the water being filtered, the flow rate through the filter and the quantity (depth) of activated carbon in the filter. All other conditions being equal, the deeper the carbon bed, the longer the interval between change-out. A three- to four-year service life seems typical for GAC use in upstate New York waters.
GAC use at Hemlock Lake
In July 2005, the 3-ft layer of anthracite coal within each of the eight gravity-fed filters at Hemlock Lake was replaced with a 4-ft layer of GAC. By converting one or two filters at a time, the entire process took about one week, with no interruption in water production. In order to maintain the desired flow rate through the filter beds without excessive pressure drop, a somewhat coarse GAC was chosen for use. GAC is generally available in a variety of grain sizes, with 8 x 16, 8 x 20, 8 x 30 and 12 x 40 mesh sizes being the most common for water treatment. The 8 x 20 mesh was most similar in effective size to the anthracite layer it was replacing.
The conversion to GAC required about 23,000 cu ft of media. At the time of this writing, it appears that use of GAC will cost about $13.70 per million gallons of water treated. Because this assumes a three-year service life before change-out, the actual cost will not be known until the service life is empirically determined. To provide perspective to this number, the overall cost of chemical treatment at Hemlock Lake is roughly $18 per million gallons of water treated, and the cost of electricity use at the facility equates to about $22 per million gallons of water treated. GAC is not an inexpensive treatment, but its cost paled in comparison to the other treatment alternatives that were considered. Both ozonation and UV/peroxide treatments would have required a very large capital investment. In addition, their associated operating costs would have been as much or more than GAC.
To reduce the upfront material investment, Rochester opted for a service contract with Calgon Carbon Corp. By this contract, the activated carbon is leased for use at the Hemlock filtration plant. The company retains ownership of the product, and Rochester pays a negotiated service fee on a quarterly basis. This approach minimized upfront investment and affords maximum operating flexibility. At the end of the service contract, there is an option to have Calgon Carbon either remove, replace or regenerate the carbon for reuse. One more advantage of the lease arrangement is that it places the full burden for any legally required record keeping associated with the regeneration and/or disposal of GAC on the vendor.
As it turned out, the GAC installation at Hemlock Lake was completed just in time, as the T&O problem returned within two days of the filter conversions, and once again, lasted for about 10 weeks. Through that period, when both geosmin and MIB were present in the raw water, GAC adsorption proved a very effective treatment for their removal. There were no customer complaints of musty or earthy flavor. Because of the conversion to GAC, particle removal in the filters has remained as good as ever, and uniform filter run volume (run-time) has not been depreciated. For the first several months of use, GAC provided an added benefit in the removal of disinfection byproduct (DBP) precursors (as measured by UV 254). As might be expected, that benefit has since diminished. Fortunately, DBPs in treated Hemlock Lake water are already adequately low, so this is not an issue.
Compared to the other T&O removal options that were considered, GAC implementation was quick and easy. The service contract with the supplier helps minimize risk and keeps the cost affordable.