Up to this point the first year of the project involved mostly planning and installation issues. These were summarized regularly on the project's web site (www.unh.edu/uv-gw) and in a series of written project updates letters dated October 1, 1997, March 27, 1998, and June 30, 1998. This fourth periodic report is actually the first to contain project data and true research progress. Progress for the fourth period of this research project is summarized in Section III under the following major tasks: A) Project Summary, B) Full Scale Pilot Studies and Data from Other Utilities, and C) Research Obstacles. The report also contains Sections IV and V describing Plans for Next Period and Contract Schedule, respectively. The standard AWWARF budget and progress forms are attached to the end of the report (but not on website version).
The fourth period of the research project had mixed success in starting up the two full-scale UV treatment plants. The initial schedule was for both pilot plants to be operating during the winter of 1997. Indianapolis Water Company (IWC) and South Berwick Water District (SBWD) each encountered installation and set up delays for different reasons, as explained in Section III-B.
In June, UNH-ERG distributed the "Sampling/Monitoring Protocols for Full Scale Implementation of UV in Groundwater Disinfection Systems." Each facility operator, pilot plant staff, and laboratory director (IWC only) reviewed the document. In addition, AWWARF and EPRI project managers reviewed the document. Comments were addressed primarily on the sampling protocol tables, and revised pages were faxed to all recipients. The text of the document was posted on the Full-Scale UV Project web site.
Section III. B. presents up to date findings of the UNH-ERG full scale UV pilot studies at IWC and SBWD. In addition, data from similar pilot studies conducted at the Mannheim Water Treatment Plant in the Municipality of Waterloo, Kitchener, Ontario, Canada are presented. While UNH-ERG is not working at the Mannheim plant, Mr. Brian Pett, the Superintendent of Water Operations has agreed to share data with us.
B. Full Scale Pilot Studies and Data from Other UtilitiesThe IWC medium-pressure UV pilot plant was officially started on August 10, 1998. Since August 10, the IWC process engineer, Dan Moran, has collected operational data and analytical samples. This periodic report is being prepared during the sixth week of operation, and of those six weeks, there were four weeks of successful run time that allowed meaningful data collection. Many operational problems have been encountered. These include (1) incorrect calibration of the intensity sensor during the first week of operation, (2) dose adjustment to account for lower flow than expected, (3) exceedingly frequent cleaning events due to lamp fouling, (4) a cleaning cycle drain failure, and (5) failure of the intensity sensor during the fifth week of running.
The initial calibration of the intensity sensor was set for conditions that were assumed during the design of the UV system. First, the water source identified for use during design of the system was changed; upon startup, the utility decided to use a different well. The water new source had a lower UV absorbance than expected which resulted in a higher delivered dose than expected. Second, a flow of 1 MGD was expected (and incorporated into the design), but the new well was only capable of producing approximately 0.75 MGD. Therefore, the initial calibration was made with some incorrect assumptions. Now, the sensor has been recalibrated, and the lamp power has been adjusted for the lesser flow.
Since start up, IWC has encountered extremely frequent lamp fouling. This presents two problems; one, it causes the UV system to be incapable of running for more than 3-4 hours at a time, and also requires too much operator time. Each time the lamp requires cleaning, the operator must manually shut down the treatment flow, switch on the cleaning cycle, and at the end of the cleaning cycle, switch the system back on to treatment mode. As this is a research pilot study, UNH-ERG relies on the utility for daily operations. The utility is very willing to participate in the study, but cannot dedicate a full time person to run the UV system. Therefore, the number of effective run hours has been limited by the frequent cleaning requirements. Trojan has committed to providing a retrofitted new cleaning technology onto the IWC UV system by October 15, 1998. It is anticipated this will yield increased treatment hours, and far less operator time.
On August 17, 1998, a cleaning cycle drain valve failure caused the system to be shut down for 4 days. The drain valve failed to cycle back to the "neutral" position during the cleaning cycle, and the system was shut down. Dan Moran attempted to install a spare valve that was available at the facility, but was unable to fit the spare valve to the reactor. Dan contacted Trojan, ordered a new drain valve, and installed it on August 21, 1998. This equipment failure meant no analytical or operation data could be collected for the week of August 17, 1998.
On September 8, 1998, the intensity sensor failed. Dan Moran noted that the intensity was dropping from 100% intensity to 4% intensity in a matter of minutes after three consecutive cleaning cycles. Eventually, the sensor reportedly registered values only between 1% and 4%. Dan Moran contacted Trojan and was advised that the shutter on the intensity sensor had probably failed. A new intensity sensor was ordered from Trojan, and delivery is expected on September 23, 1998. This equipment failure meant no analytical or operation data could be collected for the period between September 8 and September 23, 1998.
The problems encountered, service calls to Trojan and resolutions are summarized as follows:
The data collected for the IWC pilot study is attached as Tables 1 and 2, and Figures 1, 2, and 3 (Figures and Tables not available on website version). Both Figures 1 and 2 illustrate cumulative flow versus kilowatt hours, the data were separated into two separate graphs due to problems with calibration of the intensity sensor. Figure 1 shows total flow and total power used, regardless of whether or not we believe the treatment was effective. Figure 2 shows the volume of flow we believe was effectively treated, versus the total power required for effective treatment.
Figure 1 (not available on web page) shows the TOTAL cumulative flow, including periods when perhaps the lamp was fouled and treatment was ineffective. The total flow was 13,144,000 and the total power used (KwH) was 559. Therefore, the power usage was 0.043 KwH/1000 gallons. At a cost of $0.04/KwH, the cost is estimated at $0.002 per 1000 gallons treated.
Figure 2 (not available on web page) shows the cumulative treated gallons versus KwH. The volume of treated water was estimated by summing only the periods when the intensity sensor low level UV alarm was not triggered. The low level UV alarm indicated when lamp fouling had occurred to the point where the water was not receiving the correct dose. During the first five days of run time there were many periods when the UV lamps were left on, and the system left running even after the low level UV alarm had signaled. In some cases, the low level alarm had gone off 12 hours prior to the actual lamp cleaning. This occurred during the first two days when it was expected the system could be left on overnight. However, due to frequent lamp fouling, eventually the actual effective run times were reduced to 3-4 hours. Because the calibration and the intensity sensor were both problematic during the first week or two of operation, we are unable to accurately quantify the dose delivered at those times. Therefore, an estimated value for KwH/1000 gallons from these data averages 0.066KwH/1000 gallons. At a cost of $0.04 per KwH, the cost of treatment averages an estimated $0.003 per 1000 gallons treated.
Figure 3 (not available on web page) shows the ongoing intensity sensor data. During the first 5 days, 100% power was being delivered to the lamp, and the intensity sensor was programmed to a low level UV alarm at 50% intensity. Later, when the power to the lamp was decreased to 58%, the intensity sensor was reprogrammed to a low level alarm at the equivalent 50% intensity. The dose being delivered during the first 5 days was estimated to be in excess of 130 mW-s/cm2 (G. Vanderlaan, personal communication). Trojan determined that by decreasing the power to 58%, the dose would be approximately 50 mW-s/cm2. The extreme irregularity in sensor readings is a clear indication of lamp fouling and/or sensor performance.
The lamp fouling and irregular sensor performance at IWC is suspected to result in part from the combination of iron and manganese concentrations that are relatively high. Pilot studies in Canada (discussed below), where lamp fouling occurred, showed much more stable sensor data and reported excellent disinfection performance. The average influent concentrations of iron and manganese at IWC are 0.36 and 0.23 mg/l respectively. By comparison, the iron and manganese concentrations of the source waters of Canadian pilot studies are 0.004 and 0.005 respectively. So, with concentrations one and two orders of magnitude greater at IWC, the frequent lamp fouling is understandable. There is a marked difference between the sensor data collected at pilot studies done in Canada and this study.
South Berwick Water DistrictIt is anticipated that SBWD's plant will start no later than September 30, 1998. The delay at SBWD continues for various reasons. First, a delay was encountered because the plumbing required reconfiguring to accommodate the UV system. Resolving this required considerable time, as new pipefittings had to be special ordered and installed. Since that time, other delays have been encountered with respect to the electrical system linking the pumps and the UV reactor, the hydraulics between the cleaning tank and the UV reactor, and damaged valves. Each of these problems is being addressed; in some cases, obtaining parts or having a Trojan representative come to the site causes delays of one week or more.
Full Scale UV Data: Regional Municipality of Waterloo, Kitchener, Ontario, CanadaIn addition to the data being generated under this research study, full-scale UV data will be provided to UNH-ERG by Mr. Brian Pett, the Superintendent of Water Operations at the Mannheim Water Treatment Plant located in the Regional Municipality of Waterloo (RMOW), Kitchener, Ontario, Canada. Mr. Pett has agreed to share operation data and some analytical data to augment our study. Two pilot studies were conducted at two of the RMOW's 129 groundwater supplies. A medium pressure UV system and a low pressure UV system were tested.
Testing Medium Pressure UV Disinfection on Groundwater - RMOWA pilot study conducted from March 1997 to September 1997 with groundwater from well G-4. A summary of the G-4 water quality source is provided as Table 3. The UV system was a Trojan Technology UV8000 unit with electromagnetic ballasts and ultrasonic cleaner. During the pilot project, the operational parameters evaluated included: UV intensity, loop failure, power failure, watt-hour meter readings, hours of operation, water flows and pressure. Water quality parameters evaluated during the pilot test included: pH, total hardness, total alkalinity, iron, manganese, total and fecal coliform, and HPC.
The most common problem throughout the study was the frequent fouling of the UV lamp due to iron oxide coating of the quartz sleeve. This caused rapid loss of UV intensity and a decreased delivered dose. The UV system was designed with an ultrasonic cleaner that was meant to prevent the accumulation of hard water scale. The ultrasonic cleaner could not prevent iron oxide from coating the quartz sleeve, and the sleeve then required manual cleaning. Listed here is a summary of service calls to Trojan during the 6-month study:
The problems encountered at RMOW's well G-4 pilot study are similar to the problems being encountered by UNH-ERG at IWC's pilot study. Frequent lamp fouling causing rapid drops in UV intensity are common to both pilot studies. As described previously, the iron (plus manganese) concentrations at IWC are one and two orders of magnitude higher that at RMOW's well G-4. If the RMOW pilot study found lamp fouling to be problematic, this indicates we certainly can expect the same problem at IWC.
Operating and Maintenance Costs (Medium Pressure, G-4) - RMOWThe operating and maintenance costs recorded during the 6-month study were extrapolated to an estimated annual cost. A total of 27.5 hours annually was estimated to be required, which included routine inspections (15 minutes/week), manual cleaning of sleeve (30 minutes every two weeks), and lamp replacement (45 minutes, twice annually).
The medium pressure UV8000 system was equipped with a 3-step ballast that compensated for lamp aging by adjusting the power delivered over time. Accounting for the low, medium and high power levels, the annual power use was estimated at 24,168 KwH. At a cost of $0.075/KwH, the annual cost was estimated at $1,812 (Canadian dollars). In addition, two lamp replacements were assumed at $500 each ($1000 annually in Canadian dollars).
Conclusions from the Medium Pressure UV Pilot Study, RMOWThe medium pressure system was found to require more maintenance than expected due to frequent occurrence of an iron oxide coating on the quartz sleeve of the UV lamp. This adds to labor costs associated with the technology. Still, even with this added labor cost, the findings showed that UV is the most economical disinfection alternative to chlorine. This is especially true in cases where the possible construction of chlorine contact tanks would be needed to allow adequate contact time.
Testing Low Pressure UV Disinfection on Groundwater - RMOWA pilot study was conducted from November 1996 to April 1997 at RMOW's groundwater well W-10. A summary of the W-10 water quality source is provided as Table 4. The UV system was a Trojan Technology UV8000 unit with a manually initiated programmed clean-in-place (CIP) system. During the pilot project, the operational parameters evaluated included: UV intensity, power failure, watt-hour meter readings, hours of operation, water flows and pressure. Water quality parameters evaluated during the pilot test included: pH, total hardness, total alkalinity, iron, manganese, total and fecal coliform, and HPC.
The UV system operated for the 6-month test period without unscheduled interruptions. During the test period, operators spent less than 7 hours monitoring the operation of the system at W-10. The time spent included observing automated cleaning and manual cleaning of the quartz sleeves, and recording data as required for the pilot study.
Operating and Maintenance Costs (Low Pressure, W-10) - RMOWThe operating and maintenance costs recorded during the 6-month study were extrapolated to an estimated annual cost. A total of 15.5 hours annually was estimated to be required, which included routine inspections (15 minutes/week), cleaning of sleeve using the CIP (30 minutes twice annually), and lamp replacement (45 minutes, twice annually). The power costs extrapolated from the test period were 5360 KwH at a cost of $0.075 per KwH, totalling $385 per year (Canadian dollars). Assuming 8 replacement lamps per year at $60 each, the cost of lamps would add $480 annually (Canadian dollars).
Accounting only for the price per KwH, the approximate cost of treatment is $0.007/1000 gallons (Canadian dollars).
The intensity sensor performance was very stable because the lamp did not foul frequently. Attached as Appendix A to this report is a spreadsheet of data collected at W-10 during and after the pilot test. The intensity sensor data on RMOW's well W-10 UV reactor is the same type of data being collected currently at IWC by UNH-ERG. There is a marked difference between the two data sets. The data from IWC is erratic, varying from very low readings to maximum readings over very short periods of time. The data collected at RMOW's pilot study is relatively stable, varying over a small range. The disinfection performance at W-10 is reported to be excellent.
Conclusions from the Low Pressure UV Pilot Study, RMOWThe design minimum dose of 38 mW-s/cm2 (at the end of the lamp life) used in the low pressure UV pilot testing at RMOW's W-10 satisfies the U.S. EPA's dosage requirement for a 3-log inactivation of hepatitis-A virus, with a safety factor of 3.0. The design minimum dosage also provided 12-log inactivation of E. Coli, with no safety factor. UV was found to be capable of effective primary disinfection.
C. Research ObstaclesThis research study has encountered several obstacles, most of which are discussed in Section III B, or in previous reports. The delays in getting started have caused the project to fall about 9 months behind schedule.
It is hoped that South Berwick's problems have been resolved and smooth operation can commence. Cleaning at IWC remains a serious obstacle, as well as an interesting research finding. UNH-ERG is committed to working closely with Trojan to resolve the cleaning difficulties.
During the second period (October 1, 1998 through December 30, 1998), the following activities are planned: (1) We will start-up the full-scale pilot study at SBWD. (2) We will conduct the first MS-2 challenge at IWC during the second or third week of November. (3) We are preparing for a PAC meeting on December 9, 1998 at UNH, and are co-sponsoring a technology transfer meeting with the Maine and New Hampshire Water works Associations on December 10, 1998 at the South Berwick Water District. (4) We are continuing to collect operation and analytical data weekly from each pilot plant.
The requisite AWWARF contract budget and schedule documents are not available on this web page version of the report. The pilot plants are behind schedule, as is discussed above in Section III.
This page is maintained by Susan A. Willoughby
Send comments to
susanw@cisunix.unh.edu
Last updated October 1, 1998