Outsourced Water Treatment Plant for Boiler Make-Up Water: Case Study RUSSELL HOURIGAN, GABRIEL NICOLAIDES, Ontario Power Generation, Pickering, Ontario and MICHAEL DEJAK, Eco-Tec Inc., Pickering, Ontario IWC-05-78 ABSTRACT: An outsourced Water Treatment Plant featuring down flow micro media filters, reverse osmosis units and short bed demineralizers designed to produce 66 l/s of makeup water for eight (8) functioning reactor-turbine units has been running for almost 4 years. Individual equipment and overall plant performance is assessed and plant optimization discussed. INTRODUCTION Ontario Power Generation s Pickering Nuclear Generating Station (PNGS) is an eight-unit nuclear station located near the shore of Lake Ontario, in the City of Pickering, Ontario, Canada. The original set of four (A-side) units were placed in commercial service in the early 1970 s, the second set of four units (B-side) were placed in commercial service in the mid-1980 s. The generating units are rated at a Net Electrical Output of 515 MW each. The original PNGS s water treatment plant (WTP) was based on conventional water treatment process technology and included clarification, sand filtration, activated carbon and ion exchange. The original water treatment plant was aging, required a high degree of maintenance, and was limited in its ability to produce demineralized water of sufficiently high quality. In 1999, Ontario Power Generation (OPG) conducted a study of the available alternatives, and decided to build a new outsourced WTP. This was OPG s first experience at contracting out WTP operation. OPG awarded a 10-year build, own, operate, and maintain WTP contract to a team consisting of a Prime Contractor and Eco-Tec Inc (hereafter referred to as the O&M Company/Team). OPG supplied the building; the Prime Contractor owns the WTP equipment, and Eco-Tec Inc., a local, Pickering-based company, supplied the equipment and is operating the WTP under contract to the Prime Contractor. The new WTP was designed, built, installed and commissioned in eight (8) months, and was available for service in October 2001. The following aspects of the WTP project will be discussed in this paper: Site selection Inlet water characteristics Demineralized water specifications WTP process description Plant waste management System automation and monitoring Operating experience Summary Positive Experiences/Lessons Learned SITE SELECTION OPG considered the following raw water sources as possible influents to the WTP [1]: Condenser cooling water return (CCWR) Service water Domestic water Of these three options, the quality of service water and CCWR are approximately the same, i.e. they are both very similar to lake water, with the difference that both of these streams have somewhat higher temperatures (i.e. 5 10ºC) higher than lake water. The CCWR stream was the preferred choice since it represented the least potential for contamination. The potential for CCWR contamination with oil from a lube oil system leak and its impact on the WTP raw water quality are rather small. Based on the very large dilution factor and the large CCWR flow rate of 190,000 l/s per unit, more than 100 l/hr of oil must be spilled into the CCWR in order to reach the maximum allowable concentration of oil in the RO feed water of 1 mg/l. The likelihood of this situation occurring was deemed to be insignificant. A similar case could also be made for the impact of boiler blow-down, as
boiler blow-down volumes are also very small relative to the CCWR flow-rate. It should also be noted that the higher temperature of the CCWR stream represents a major advantage over the use of raw lake water, since it translates into > 25% reduction in the overall WTP plant size and its capital cost. The use of domestic water as feed to the new WTP was considered, but was not pursued by the PNGS s Design Team. The main deterrents to using domestic water were: Estimated additional operating cost of $600,000/year to purchase domestic water from the municipality Additional infrastructure would have to be installed to supply the domestic water Concerns regarding the presence of trihalomethanes in the domestic water and ability of the WTP process to remove them OPG decided to locate the WTP outside of the security perimeter on the east side of the station, over the outfall of the PNGS s-b Condenser Cooling Water Discharge Channel. The WTP utilizes CCWR -water from the discharge ducts of the PNGS s Units 7 and 8. This site was selected because it offered the following advantages: Reliable, low-cost source of water Warm raw water for improved RO efficiency (i.e. up to 10ºC higher temperature) Close proximity to power, sewer, and demineralized water header tie-ins Location allows vendor ready access and permits equipment installation and retrofits as commercial modifications Use of existing CCWR outfall structure for raw water supply pumps The location of the PNGS s WTP is as depicted in Figure 1 and 2. The complete system including all equipment, storage tanks, piping, instrumentation and controls is housed inside a 620 m 2 (6,700 ft 2 ) building. This is quite compact for a system of 66 l/s, (1050 gpm) capacity. By contrast, the original water treatment plant occupied approximately 1,279 m 2 (13,770 ft 2 ). Figure 1: Aerial View of the Pickering Nuclear Generating Station Water Treatment Plant Water Treatment Plant Figure 2: Water Treatment Plant INLET WATER CHARACTERISTICS OPG provided the following Raw Water Analysis and the Lake Ontario Water Temperature Data in Table I as a basis for design of the water treatment plant [2]: Source: Lake Ontario Water at the PNGS. Characteristics: ph 7.95 8.07 Turbidity (NTU) 0.08-120 Conductivity (ms/m) 30.6-34 OH Alkalinity (ppm) < 5.0 CO 3 Alkalinity (ppm) < 5.0 HCO 3 Alkalinity (ppm) 98-129 Calcium (ppm) 35 39 Magnesium (ppm) 7.9 8.4 Sodium (ppm) 10.1 13.0 Potassium (ppm) 1.42 1.80 Sulphate (ppm) 23.8 26.8 Chloride (ppm) 20.8 24.8 Nitrate (ppm) 1.7 2.0 Total Silica (ppm SiO 2 ) 0.75 27.4 Reactive Silica (ppm SiO 2 ) 0.6 1.3 TIC (ppm) 16.9 17.5 TOC (mg/l) 1.9 2.2 Silt Not quantified (variable quantity) Table I: Lake Ontario Pickering Nuclear Generating Station s Mean Monthly Water Temperatures 1970 1988 [3] Temperature o C Surface Ambient 8 m Depth 12 m Depth January 1.6 1.9 2.1 February 1.2 1.8 1.8 March 2.3 2.4 2.3 April 5.4 4.0 3.9 May 7.9 6.3 5.9 June 10.3 8.2 7.2 July 12.6 10.0 8.6 August 17.3 14.8 13.5 September 14.3 13.1 12.0 October 9.9 9.1 8.5 November 6.3 6.0 6.2 December 2.9 3.4 3.6
DEMINERALIZED WATER SPECIFICATIONS The demand of demineralized water is dependent on the operating mode of the eight generating units in the station, i.e. unit shut down, unit start-up, unit on power, etc. The average flow was estimated to be between 35 l/s to 53 l/s. The water treatment plant has been designed to produce the following [4]: 1. Flow: Net product during normal operation 35-53 l/s Net product during peak operation 66 l/s 2. Quality Specific conductivity (at 25ºC) 0.06 µs/cm Total silica (reactive and colloidal) 2.0 µg/kg Sodium 0.2 µg/kg Total organic carbon 10 µg/kg Total dissolved and suspended solids 100 µg/kg Total chloride (per ASTM D5542) 0.8 µg/kg Total sulphate (per ASTM D5542) 1.8 µg/kg WTP PROCESS DESCRIPTION The water treatment process is illustrated in Figure 3. The basic process involves treating feed water by: 1. Direct filtration for suspended solids removal using a set of micro media backwashable filters. 2. Reverse osmosis for bulk ion removal and the separation of organic compounds. 3. Ion exchange polishing using short, compressed bed ion exchangers for final polishing of the water to the specified product water quality. DIRECT FILTRATION Feed water from the PNGS s condenser cooling return ducts is pumped to a set of three Micro Media Filters (see Figure 4 on the following page). Prior to the filters, coagulant (poly-aluminum chloride) is injected and mixed inline. The coagulant dosage is automatically controlled based on flow and feed turbidity, and is typically in the range of 1 5 ppm. The Micro Media Filters operate similar to a multimedia filter. Water flows downward through two layers of media in a pressure vessel. The upper layer of media is coarse anthracite while the lower layer is a proprietary fine media, which is much finer and denser than media commonly used in multimedia filters. Due to the very fine size of these media particles, the filter is able to remove very small-suspended solids with a very high efficiency. Technical details and operating parameters for this type of filter have been reported previously at this conference [5]. Filtrate quality from the filters is continuously monitored on-line for turbidity and typically falls in the range of 0.05 0.12 NTU (see Figures 5 and 6 on the following page). Periodic manual sampling for silt density index (SDI) has been conducted and SDI values in the range of 2 5 have been obtained. The filtrate quality is maintained despite significant fluctuations in the feed water turbidity (3 100 NTU) caused by varying weather conditions over the lake. Using a patented operating feature known as backslip, the filters are able to maintain extended runs between backwashes even at relatively high turbidity conditions (50 100 NTU). The filter backwash sequence is automatically initiated based on flow and pressure drop across a filter. Backwash is accomplished using filtered water directly from the remaining, in service, filters. During infrequent occasions when the demineralized water demand is high and the feed water turbidity is also very high, City Water is used to augment the filter-backwash. A cartridge filter is installed downstream of the Micro Media filters as a trap in the event of a filter lateral failure. The cartridge filters are rated at 10 microns and run for 4 6 months before replacement. Figure 3: The PNGS Water Treatment Process Schematic
Figure 4: Micro Media Filters Figure 5: Feed Water Turbidity Levels for OPG s Micro Media Filters for the months of January and February 2003 Figure 6: Filtrate Turbidity Levels for OPG s Micro Media Filters for the months of January and February 2003 REVERSE OSMOSIS - Prior to entering the reverse osmosis (RO) unit (see Figure 7), filtered water passes through an ultraviolet sterilization unit and acid is injected and mixed in-line. Ultraviolet light was selected as a sterilization technology instead of the more conventional method of chlorination because of evidence that chlorination would form tri-halomethanes (THM), which are not completely removed by reverse osmosis or ion exchange. This would compromise the system s ability to meet the 10 ppb TOC limit. Even if the TOC was within the 10 ppb limit, a significant portion of the TOC would be in the form of THM s and this would result in the release of chloride when the THM is decomposed in the boilers. Acid is injected into the filtered water to lower the ph and prevent scaling of the RO membranes. The reduced ph results in a shift of the carbonatebicarbonate-carbon dioxide equilibrium, increasing the concentration of carbon dioxide. Carbon dioxide is not rejected by the RO membranes, producing a permeate with high levels of carbon dioxide gas. The increased carbon dioxide levels result in an increased caustic consumption of the polishing demineralizers. However, a higher CO 2 level was deemed acceptable because the polishers could still produce the required water quality, and the overall chemical operating costs would be lower than the alternative approach of maintaining a more neutral water ph (thereby maintaining a lower carbon dioxide concentration) and using a polymeric antiscalant to control carbonate scale formation. The filtered and pretreated water then passes through a RO system arranged in a three parallel bank, single pass, two-stage configuration. The RO elements are thin film composite type. A clean-inplace system is used to clean one bank of elements at a time. Figure 7: Reverse Osmosis Unit
ION EXCHANGE - The RO permeate is collected in a storage tank and then drawn by one of two trains of ion exchange polishers. Each train consists of two polishers in series and is capable of treating 33 l/s (50%) of full flow on a net basis (see Figure 8). These polishers are not the traditional type consisting of mixed cation and anion resins. The polishers use short (15 cm tall), separate cation and anion resin beds with fine mesh resins packed under slight compression in a fixed column with no freeboard. Each polisher is a self-contained skid complete with resin bed, regeneration system, and controls. Regeneration consists of on-skid pumps drawing 93% sulphuric acid and 50% caustic soda directly from storage tanks and injecting them into DI water a few immediately before the inlet to the appropriate resin bed. Since the regeneration times are short (complete regeneration and return to service in 7 minutes), additional trains were not required. This type of ion exchange system and the principles of its operation have been described previously at this conference [6,7]. Compared with a traditional mixed bed polisher, this type of ion exchanger has a number of benefits: 1. Since the resins are fixed in separate beds, the equipment design and system operation is very simple. This allows for full system automation including unattended regeneration. By contrast, mixed bed systems have a more complex design and experience operational challenges related to the need for excellent separation of resins before each regeneration and thorough uniform mixing of the resins after each regeneration. While this may not be a challenge initially, as resins age, and their characteristics change, this can become increasingly more difficult. 2. Since the resins are fixed in the vessels with no freeboard, the possibility of flow channeling is eliminated and it is very simple to carry out counter-current regeneration. This results in a much more chemically efficient regeneration than that of mixed beds with the added advantage of a significantly lower regenerant chemical consumption. 3. It is possible to operate the resins in a column without freeboard due to the relatively low loading (about 10% of resin capacity), which in turns results in minimal shrinking and swelling of the resins due to changes in ionic form. The minimal shrinking and swelling of the resin beads also reduces the mechanical fatigue of the resin particles, which prevents breakage of resin beads and subsequent attrition losses and flow distribution problems. 4. The small resin volume in each vessel and low resin loading also result in very small waste volumes during regeneration. This feature enables the neutralization process to be carried out in small tanks. 5. The use of fine mesh resins results in a very high surface area despite the small volume of resin. This means that a greater proportion of exchange sites are at the surface of the resin particles. Since the rate of ion exchange is dependant on the rate of diffusion of ions to exchange sites within the porous structure of the resin particle, a greater proportion of surface exchange sites means that proportionately more exchange sites can be accessed rapidly through diffusion of ion through the liquid film around the resin particles rather than through the much slower process of intra-particle diffusion to access internal resin sites. Consequently, contact time required in the resin bed is short. This allows flow velocities of 30-50 gpm/ft 2 and regeneration times as short as 40 seconds for cation bed regeneration and 1.5 minutes for anion bed regeneration. The rapid ion exchange kinetics also allow for very complete ion exchange to take place resulting in the ability to produce very high purity water without mixed beds. Pilot testing prior to the system design demonstrated that a single cation-anion polisher was capable of producing the specified water quality (specific conductivity < 0.06 µs/cm). Two were used in series in this system for redundancy. Another technology that has been more recently applied for polishing after RO is electrodeionization. While this technology has the benefit of eliminating the need for regenerant chemicals, experience reported at this conference has indicated that it may not be sufficiently robust to be used after a single pass RO system without risking irreversible fouling. The added cost and complexity of the pretreatment system required for operation of a reliable electrodeionization system eliminated this technology from consideration in this case. Figure 8: Polisher with Ion Exchange Technology
WASTE TREATMENT Wastes from the system include filter backwashes, RO reject and cleaning wastes, and ion exchange polisher regenerant wastes. The majority of the wastes are collected, ph adjusted, auto-sampled and discharged back to the lake. When the turbidity of the lake water feeding the WTP is above 6 NTU, the suspended solids concentration of the filter backwash water exceeds the discharge regulations. In order to ensure compliance under any condition, half of the filter backwash (source of suspended solids) is automatically discharged to the municipal sewer system rather than being discharged to the lake. All discharge data; including analysis of auto samples is recorded and reported to the environmental regulatory authority to confirm compliance with discharge regulations. SYSTEM AUTOMATION AND MONITORING The WTP is totally automated and is controlled by redundant programmable logic controllers (PLCs). This includes all filter backwashing initiation and sequences, RO flow control, IX polisher regeneration initiation and sequence, and waste neutralization. The O&M Company s personnel are on-site approximately 15 hours per week. The WTP can be operated unattended via a SCADA system which links all system sensors to a central control station within the water treatment plant building. This control terminal can be accessed remotely via a secure VPN (virtually private network) by modem. Through this connection, the O&M Company s staff has access to view and control the plant from their off-site office location, which is located a few minutes away. In the event of any alarm condition, the alarm is displayed at the central control system which also initiates a phone call to alert the O&M Company s on-call service engineers. The service engineer, once alerted, can access the plant control system via laptop from any modem connection and confirm the alarm condition, review any related process trends, take remote action (start or stop units, open or close valves), and do a site inspection via three web cams (with zoom and pan capability) located in the water treatment building. In the event that the condition requires on site presence, a technician can be on site within 1-2 hours. The O&M Company does an on site walk around 4-5 days per week and performs scheduled instrument calibration, and preventative maintenance tasks (e.g. cleaning of RO membranes). OPERATING EXPERIENCE Since the WTP was commissioned in October 2001, it has continuously supplied the normal water requirements of the nuclear power plant without the need for supplemental demineralized water from other sources. During the first few months of operation, the WTP experienced some teething pains. For example, the TOC levels in the product water were slightly above the 10 ppb criteria. This problem was found to be related to incomplete sealing of the RO membranes. Once this was corrected, the membrane performance improved and the TOC has remained consistently below the specified criteria. Also, during the first few months of operation, the WTP required frequent operator attention to address various system alarms. The frequent alarms were addressed by progressively tuning the automation system until the occurrence of the alarms became rare. The system then became fully automated, requiring minimal operator attention, as originally intended. During the first year of operation of the WTP, it was found that the feed water turbidity excursions, along with the presence of silt and sand were considerably higher, more frequent, and prolonged than was originally anticipated. For the most part, the Micro Media filters responded well in terms of their ability to produce a high quality filtrate despite turbidities exceeding 100 NTU on a number of occasions. However, this resulted in more frequent backwashing of the Micro Media filters, with a resultant decline in net flow capacity. This situation was corrected by modifying the filter operating sequence to introduce a very short flow reversal, rather than a complete backwash. This increased the run length between backwashes by more than 500% with the improved net capacity despite turbidity upsets. Subsequent investigation into the mechanism of this flow reversal (subsequently named backslip ), and its refinement has led to a patent application for the process by the Equipment supplier. Both the O&M Company and OPG s personnel have monitored the performance of the reverse osmosis system closely. At the time of writing this paper, nearly four years since the plant start-up, two of the three RO membrane skids are still operating with the original membrane elements in place. After the first two years of service, the normalized flow through the membranes had declined to an unacceptable level and in-situ cleaning was ineffective in restoring the permeate flux. It was decided to remove the membranes and send them off-site for cleaning. The external cleaning was successful in cleaning the membranes and removing the microbiological
fouling material that had caused the permeate flux decline. In order to keep the WTP in operation during the external cleaning, one set of new membranes was purchased and installed on one of the three skids. Then as the membranes for one skid were cleaned, they replaced the membranes from the next skid to be sent for external cleaning. At the end of the cleaning process, two skids contained virtually all cleaned membranes, one skid contained virtually all new membranes, and a set of cleaned membranes were set aside for future potential replacement. At the time of writing, after almost four years of operation, no further membrane replacements have been required. In-situ cleaning of the membranes is required primarily to control microbiological growth, which is particularly acute during summer months when the feed water temperature exceeds 15-20 o C. Water quality data is continuously recorded and the average values for the key chemistry parameters are summarized below in Table II. The WTP comfortably and consistently meets all of the quality criteria. Despite brief excursions in quality, usually due to slight increases in the influent TOC levels, the WTP still meets the upper acceptance criteria for the station (viz. 50 ppb for TOC). Table II: Overall Averages Item Avg. Value Quality Criteria Flow (l/s) 27.3 66.0 design flow rate Resistivity (megohm-cm) 18.0 > 16.7 Sodium (ppb) 0.05 < 0.2 Silica as SiO 2 (ppb) 0.36 < 2.0 Total Organic Carbon (ppb) 5.1 < 10.0 % of Volume Meeting Specified Quality * 97.8 N/A Water quality criteria was also established for total sulphates and total chlorides, although it was recognized that these parameters could not be readily monitored on-line at the low levels specified (< 1.8 ppb for total sulphates, and < 0.8 ppb for total chlorides). It was assumed that if the conductivity met the specification that the total sulphate and total chloride levels would also be within the specification. * Note: For the remainder of volume (2.2%) water did not meet highest quality for which contractor is paid for water, but met acceptable quality standard for acceptance by OPG. This assumption was to be tested and confirmed periodically by carrying out sulphate and chloride testing programs using sensitive instrumentation (on-line ion chromatography), the results of which are summarized in Table III. The results indicate that the total sulphate and total chloride levels indeed meet the quality criteria. The low value of total chloride is an indication that minimal amounts of trihalomethanes are passing through the WTP. Table III: Total Sulphate and Total Chloride Test Results for 2004 Item 2001 Value 2004 Value Quality Criteria Total Organic Carbon (ppb) 8-9 <10 < 10 Total Organically Not Bound <1 <1 specified Heteroatoms (ppb) Total Chloride ** (ppb) <1 <0.3 < 0.8 Chloride (ppb) <0.5 <0.1 Total Sulphate *** (ppb) Not specified <0.05 0.2 <0.5 < 1.8 Sulphate (ppb) <0.05 <0.05 Not specified By contrast, the old WTP at the PNGS, which consisted of conventional technology (i.e. clarifier, sand filter, carbon filter and conventional ion exchange), produced demineralized water having a TOC of about 200 ppb, with significant associated organically bound sulphate and chloride. The improved TOC values produced by the new WTP are attributable to the use of RO membranes to minimize the passing of TOCs from the feed, and the use of the short bed ion exchange polishers which minimize resin volume and resulting throw of organic impurities from the resin that is encountered in conventional deep bed systems. In 2003, it was determined that the operating costs of the plant could be reduced by an alternate chemical pretreatment scheme. The original design used acid to reduce the RO feedwater ph to prevent calcium carbonate scale formation on the membranes. This results in the conversion of carbonate to bi-carbonate and carbon dioxide, which passes through the RO membranes and is removed ** Total Chloride = Ionic Chloride + organically bound chlorine as chloride *** Total Sulphate = Ionic Sulphate + organically bound sulphur as sulphate
by the anion exchange polishers. However, this requires the consumption of caustic soda by the polishers for their regeneration. It was decided to use an anti-scalant chemical to pretreat the RO feedwater at the more neutral ph of 7.5 8.0. This would reduce carbon dioxide formation, polisher loading, and resultant caustic consumption. While this approach initially proved successful, after a few months, it was determined that the performance of the RO membranes had declined significantly, and that in-situ chemical cleaning was unable to successfully restore performance. Upon investigation of various operational parameters, it was determined that the decline in RO system performance coincided with the introduction of the anti-scalant dosing. Upon restoring the pretreatment to the original design (i.e. ph adjust only), RO system performance returned to previous levels, and membrane cleaning was reduced to the original frequency. The root cause of the RO membrane fouling with the anti-scalant based pretreatment could not be identified, but it was suspected that microbiological fouling played a significant role. SUMMARY As expected, some teething pains were experienced during the first year of WTP operation. OPG and the O&M Team s staff worked closely together during this period to identify and overcome these problems. Operation and maintenance of the WTP has now become routine and requires less involvement from OPG s personnel on a day-to-day basis. The following is a brief summary of some of the positive experiences and lessons learned from this project. POSITIVE EXPERIENCES - WTP produces excellent product water quality; exceeds the demineralized water quality specification for the key parameters by a fairly wide margin Short implementation time 8 months from contract signing to delivery of first volumes of demineralized water (includes building construction, system design and manufacture, installation, and commissioning) Cost to produce water is fixed for 10 year contract term Highly automated plant can be operated remotely; staffed by one full-time operations coordinator The O&M Company s office is close by; additional support staff are just minutes away WTP building and equipment, as well as provision for back-up emergency water treatment trailers. In this case, the WTP building is surrounded on all sides (i.e. by water, roadways, buildings, and equipment). REFERENCES 1. R. Al-Samadi, Technical Memorandum, Proposed PNGS WTP Siting, September 21, 2000 2. OPG Specification No.: P-TS-71600-10002, Technical Specification for Outsource Water Treatment Plant, Rev. 1 Section 2B5, DCP No. 395, March 5, 2001 3. PNGS B Safety Report, Volume I Plant Description, 1997 Edition, Table 2.5.2-5 4. Proposal from Prime Contractor for PNGS WTP, Section 3.14.3, January 15, 2001 5. Mike Sheedy, Eco-Tec Inc., Peter G. Kutzora, WE Energies. The Use of Short Bed Ion Exchange Technology for the Production of High Purity Water at WE Energies Pleasant Prairie Power Plant. Technical Paper presented at International Water Conference 2004 6. Ed Ritchotte, Commonwealth Electric Co., Wayne Rezendes, Canal Electric Co., Bradley Smith, Tod Wilson, Eco-Tec Inc. Water Treatment System Retrofit at Canal Electric Power Station. Technical Paper Presented at PowerGen 98 conference in Orlando Florida, December 1998. 7. Doug Jackson, Bryce Heartwell, TransAlta Energy Corp., Michael Dejak, Cathy Fletcher, Eco-Tec Inc., Short Bed Demineralizer Technology at TransAlta Energy. Technical Paper Presented at International Water Conference, Pittsburgh, October 1994. LESSONS LEARNED - If possible, the WTP location should be selected to allow sufficient space to accommodate the