Case Study TOC Reduction in Steam Generator Feedwater Utilizing a Combination of Treatment Processes to Improve the Feedwater Quality at Kerncentrale Doel, Belgium Authors: Raphael Philippe, Manager Plant Chemistry Doel 3-4, Electrabel and Mark Loonen, Applications GE Water & Process Technologies Table 1: Analysis of pond water Presented at: Ultrapure water 5, 7th December, 1994, SCI, London, England Introduction Kerncentrale Doel (KCD) is a PWR nuclear power plant north of Antwerp in Belgium, operated by Electrabel. It consists of four reactors, Doel 1, 2, 3, & 4, which were completed in 1975, 1975, 1982, and 1985 respectively. Doel 4 is a 1010 MW reactor, designed and built by Traction-Electricite and using three Westinghouse steam generators. The holding pond is chlorinated to control algae growth. The water is then treated via sand filtration, activated carbon, Weak Acid Cation resin, Strong Acid Cation resin, Weak Base Anion resin, Degas, Strong Base Anion resin and mixed bed resin, as shown below: KCD replaced all three steam generators in Doel 3 in September 93, and the three steam generators in Doel 4 are due to be replaced in 1996. This means that the production lifetime for the steam generators has been only 11 years (the normal expected production lifetime is supposed to be more than 30 years). The steam generators are being replaced mainly due to problems associated with corrosion. Doel 4 has recently had a shutdown to repair some steam generator tubes, and is currently operating with many of its steam generator tubes plugged. Feed Water Source The feed water for the KCD water treatment system comes from a large holding pond which is fed with Antwerp city water, which itself originates from the river Rhine via the Prins-Albert-Kanal. Figure 1: Holding pond process to control algae growth Organics In order to ensure that the replacement steam generators do not experience similar corrosion problems, KCD conducted a study to determine the possible sources of the corrosion. The Total Organic Carbon (TOC) in the steam generator make-up water was determined to be a possible contributing Find a contact near you by visiting www.ge.com/water and clicking on Contact Us. * Trademark of General Electric Company; may be registered in one or more countries. 2011, General Electric Company. All rights reserved. CS1089EN.doc Aug-11
factor to the corrosion. At the high temperatures and pressures found in steam generators, the TOC breaks down to lower molecular weight organic acids which cause corrosion of the steam generator tubes. TOC is a general term, which covers a wide variety of organic compounds. Largely, the organic entities present in raw or potable waters are of vegetable origins, generically described as humic (>2000 MW) and fulvic (500 ~ 2000 MW) acids. The feedwater TOC is typically in one of three forms; 1. Naturally occurring weak organic acids - These are not completely removed by the existing water treatment at Doel. At the high temperatures found in the steam generators, these break down to form low molecular weight acids such as acetic and formic acid. The new steam generators are made of metals resistant to acetic/formic acids. 2. Chlorinated organics - This type of TOC is a result of chlorinating the feed pond at Doel. The chlorinated organic compounds formed by the action of chlorine on the humic and fulvic acids can break down in the steam generators to form HCI, resulting in chloride stress corrosion of the steam generator tubes. 3. Colloidal organics - Ions such as chlorides or sulphates may hide within the matrix structure of colloidal organics and be released in the steam generator. It has been observed at other PWR plants, that there is a direct correlation between make-up water TOC levels and high chloride and sulphate levels in the steam generators. Colloidal organics pass easily through ion exchange resins, but are removed 100% by reverse osmosis membranes. year, but organic fouling of the resins has not been a problem. The TOC content of the steam generator make-up water produced by the mixed bed at Doel can be up to 450 ppb in the summer. A typical TOC profile across the water treatment system is shown in the following graph: Figure 2: Typical TOC (ppb) Levels on the Outlet From the Different Stages in the KCO Water Treatment System The INPO (Institute of Nuclear Power Operators) guidelines list 100 ppb of TOC as an achievable value. Plant-specific action levels should be established to prevent excessive inputs of TOC. If the influent TOC routinely is above 100 ppb, actions should be established to prevent excessive input of TOC. If the effluent TOC routinely is above 100 ppb, actions should be considered to reduce TOC. INPO 88-021 Rev 01 Institute of Nuclear Power Operators. Table 2: INPO Achievable Values for Make-up Water Treatment Plant Effluent TOC removal The existing water treatment plant reduces the towns water TOC from ~2000 ppb after filtration, to less than 450 ppb at the effluent of the mixed bed. The activated carbon beds in KCD s water treatment system remove approximately 50% of the organics, and the anion resin beds remove around 75% of the remaining organics. The resins in the make-up water Dl system are to be replaced this. Typically, steam generator manufacturers set their make-up water warranty specification to 100 ppb, and increasingly, in the USA, nuclear power plants have lowered their TOC specifications in line with the currently achievable minimums. Page 2 CS1089EN
After analysing the current data, KCD decided to set a TOC limit of 50 ppb for the make-up water to the steam generators. Their existing water treatment plant typically had effluent TOC values of < 450 ppb, so they decided to put out a tender for a plant to reduce their make-up water TOC content down to a specification of 50 ppb, with an aliowance of up to 100 ppb if the demineralised feedwater TOC was unusually high. It has been found that, while no single removal technique totally removes the wide variety of organic matter that constitutes TOC, reverse osmosis and mixed bed ion exchange typically remove 90% of the TOC. It was therefore decided that reverse osmosis would be the best technique to be able to meet the specification running between 20-50 ppb. Since TOC removal is the primary objective, and the feedwater to the system is demineralised quality (so TDS reduction is not necessary), the RO system was designed to operate at 90% recovery. As the RO is in the post-di position, scaling is not a concern because the reject is essentially demineralised water with an increased TOC content. The reject from the system is routed to the front end of the make-up system where it is blended with city water. The Reverse Osmosis system The feedwater to the reverse osmosis (RO) system is supplied by a number of sources. Most of the water supplied to the reverse osmosis system comes from the demineraliser system described previously, although additional make-up water is also supplied from the two demineralisers at Doel 1 & 2, as well as from a 10,600 gpm (40 m 3 /h) flash evaporator. If used, the water from the flash evaporator is blended to prevent feed water to the RO system from rising above 95 F (35 C) which would damage the RO membranes. Table 3: Typical feedwater quality to the RO The RO system which was designed, supplied and is operated by GE Water & Process Technologies needed to assure in specification make-up water TOC levels at all times and accommodate the different types of TOC that may be found in the demineralised feedwater. The system was also designed to allow for stand by, low and high demand water use. The system was primarily designed to remove organics, but also has the added advantage of reducing the TSS of the make-up water. which was Figure 3: Contents of System 2 x Booster pumps to pressurize water from the feed storage tank to adequate pressure for the RO high pressure pumps. One pump is on stand by while the other pump is in operation. 3 x 5 µ filter cartridge housings to prevent suspended solids from fouling the RO membrane. The cartridge filters are rinsed with DI water prior to use to prevent them from contributing TOC. The filters can be changed out while the RO is running. 1 x Reverse Osmosis unit consisting of two independent trains rated at 6600 gpm (25 m 3 /h) at 77 F (25 C) each. Each train consists of a high pressure multistage centrifugal pump and a 3-2-1 array of stainless steel membrane housings. Each of the 12 housings hold 6 DOW filmtec BW 8040 membranes. 1 x UV light to oxidize any trace TOC which has not been removed by the RO. The UV light may also help prevent bacterial problems during low demand periods. 1 x Activated carbon filter for use downstream of the RO to remove trihalomethanes CS1089EN Page 3
that would have passed through the RO membranes. 1 x Mixed Bed polisher to remove TOC ionised by the UV light and any salts leached from the activated carbon filter. 1 x 5 µm resin trap to prevent mixed bed resin fines from entering the product water. The entire system is housed in a 45 foot insulated trailer which enabled the system to be fabricated off-site, driven on-site and hooked up with the minimum of effort. The RO system arrived on site on 27th September 1993, was hooked up, the membrane preservative rinsed out of the system, rinsed into quality, and was operational by the contractual start-up date of 1st October. During routine maintenance outages at KCD, the water demand decreases considerably. Occasionally large quantities of Dl water are required for rinsing, but many days require no Dl water. It is not good practice to leave an RO unit idle for days, due to biological build-up on the membranes, so during periods of low demand, the RO units run 5 minutes per day to flush the membrane surface. Operating Data (Normal operation) The TOC in the product water has always been below the 50 ppb specification and is typically around 3-7 ppb. KCD analyses grab samples of the product water and measures < 50 ppb as well. System Peripherals The system pumps water from a 317,000 gallon (1200 m 3 ) tank, treats it and transfers it to another 317,000 gallon (1200 m 3 ) tank. The RO operates at 90% recovery, and the concentrate from the RO system is returned back to KCD s feedwater pond. The TOC and conductivity are measured at the outlet of the trailer using Anatel A100 (TOC) and Thornton 770 PC (conductivity) meters. If the TOC or conductivity are too high, the system will recirculate product water internally until the water is in specification. Figure 4: System Peripherals Level controllers control the operation of the RO. Both RO trains run if the product tank is low. One train runs at mid capacity and both are off if the product tank is full. If the feed tank is empty, the RO will turn off. The system is also protected with High TOC, high/low flow, high permeate pressure, low feed pressure and high temperature alarms. Figure 5: Samples of Product Water After operating the system for 4 months, GE and KCD determined that the mixed bed polisher and the activated carbon tanks were not required to maintain a TOC of less than 50 ppb in the product water (see previous graph). The reverse osmosis system produced satisfactorily low levels of TOC by itself, so the activated carbon and mixed bed tanks were taken off line in February 1994. The activated carbon and mixed bed are currently on a standby mode in case they are needed during the summer months with higher TOC loading. With the mixed bed on line, the conductivity of the product water is close to the theoretical 0.055 µs/cm. As can be seen from the next graph, with the mixed bed off line, the conductivity of the product water is normally ~ 0.3 µs/cm but has occasionally peaked above 2 µs/cm, and is mainly due to CO 2 or ammonia. Figure 6: Mixed Bed Off Line Page 4 CS1089EN
The feed TOC is probably mostly colloidal, since it is rejected nearly 100% by the RO system, and there was only a 3 ppb increase in the product water TOC with the activated carbon and mixed bed tanks off line. Figure 9: Feed Water Tank and Product Water Tank TOC Levels (ppb) Figure 7: Product Water TOC Levels (ppb) Utilizing Reverse Osmosis Only A few days after the system was installed, problems were experienced with the UV system, which, due to an internal part failure, would overheat and automatically switch itself off. Conclusion The RO system at Doel has significantly reduced the levels of TOC in the make up water to levels well below the specified limits. Utilizing reverse osmosis, UV, activated carbon and mixed bed resins for demineralised water TOC reduction, we have managed to consistently produce levels of <10 ppb TOC. At the time of this paper, the effects of the lower TOC on rates of corrosion had not yet been determined, but KCD has seen an improvement in the steam generator water chemistry. Figure 8: TOC (ppb) With the UV Lamps Cycling Off and On The limited TOC data we obtained while the UV lamps were cycling on and off indicated that they produced approximately a 33% reduction in the TOC reading (from ~13 ppb to ~10 ppb TOC). With the TOC readings being well below specification anyway, it is debatable whether the cost of the UV system justifies the ~3 ppb drop in TOC. The UV was however repaired, and has operated perfectly ever since. All of the above TOC data was measured using an Anatel A100 analyser which samples the product water from the trailer. KCD takes their own samples from the feed water and product water tanks for laboratory analysis. The impact of the improved water chemistry on corrosion will not be known for three to five years. KCD are going to replace the three steam generators in Doel 4 in 1996, and we anticipate results showing an extended steam generator lifetime with the improved feedwater quality. Figure 10: Outlet Conductivity vs. Time CS1089EN Page 5
Figure 11: TOC vs. Time References 1. Miller, W.S. et al., TOC removed from make-up water at Millstone Nuclear Generating Station, International Water conference, October 1986, Pittsburgh, PA. 2. INPO Guidelines 88-021, pages 12-13. 3. Meltzer, T.H. High-purity Water Preparation. 4. O Brien, M.J. Applicable Processes for Organic Removal in Make-up and High Purity Water Systems, Ultrapure water 4(9), pp 331-334 (1987) 5. Nuclear News. Figure 12: System Recovery vs. Time Page 6 CS1089EN