A real-time chemistry monitoring system is being used at the Waterford 3 nuclear plant to “virtually” verify instrument data.
Effective water treatment at power plants depends on accurate sample analysis and instrumentation. At many plants, weekly manual instrument verification is performed to ensure such accuracy. Software solutions are now available, however, that can simplify this process, provide increased monitoring capabilities, eliminate manual theoretical calculations, and significantly reduce manual data entry.
Due to staff downsizing in the power industry, plant personnel are assuming responsibility for a broader range and volume of tasks. Individuals are often challenged to find enough time to accomplish tasks assigned to them. As part of the quality control program at Entergy Corp.’s 1,075 MW Waterford 3 Nuclear Power Station in Taft, La., for example, plant personnel had performed manual instrument verification on a weekly basis using theoretical calculations with SysChem and AminMod.1
Smart Chemistry
To achieve “work destruction,” Waterford began using SMART ChemWorks (SCW),2-3 an EPRI-developed real-time chemistry monitoring and advisory system currently being implemented at numerous nuclear power plants. SCW can benefit power plants by automatically verifying instrument data and by transferring instrument data automatically into the plant’s chemistry database. Using measured plant data as inputs, SCW continuously runs a simulator and performs calculations for simulated virtual sensors, which are used for instrument verification.
Most of the secondary instrumentation at Waterford is routed to the plant computer, which connects to the network, permitting instrument observation on the plant Satellite Display System (SDS), PI Process Book and SCW. By transferring the instrument data automatically into the plant’s Windows-based Chemistry Data Management System (WinCDMS), SCW enabled Waterford to eliminate manual logging of acceptable data, previously performed on each shift. In addition to the online instrument data, data from the virtual sensors simulated by SCW are transferred into the database at specified frequencies.
To prove that SCW could be used reliably for virtual instrument verification and in-line data transfer, the project team performed detailed monitoring4-6 of key parameters at Waterford. Specifically, the team compared data values generated by SCW-simulated virtual sensors for ammonia, ETA (ethanolamine), pH, and conductivity with the values obtained from bench analysis and calculated values using SysChem and AminMod. The team also validated electronic signal transfer from online meters to a plant computer.
Instrument Verification
For instrument verification, SCW runs a plant chemistry simulator and calculates pH, conductivity and impurity concentrations throughout the secondary steam cycle using known impurity concentrations at feedwater or blowdown. A Dionex DX-320 Ion Chromatography (IC) instrument analyzed feedwater ETA and ammonia on a weekly basis. Based on the measured feedwater impurity concentrations, SCW predicted virtual sensor measurements for ammonia, ETA, pH and conductivity values at other locations in the secondary cycle. The virtual numbers were compared with measured values and theoretically calculated values using the normal calculation method.
pH
To verify SCW’s virtual pH values, the plant monitored the pH values obtained from the instruments, the theoretical calculations using SysChem, and SCW for several months. The results shown in Figure 1 illustrate an excellent agreement between the virtual pH values by SCW and those from the instruments and SysChem calculations, indicating that SCW can be used to calculate a theoretical pH value instead of SysChem. This resulted in a more efficient process for determining theoretical pH and reduced required work load by eliminating manual SysChem calculations.
 |
As a two-loop reactor, Waterford has two steam generators, designated SG-1 and SG-2. Since SCW does not calculate split system data, theoretical data is not calculated for SG-2. However, a comparison of the calculated theoretical pH values for SG-1 and SG-2 revealed no significant difference between the two generators. This suggested that the SCW-predicted pH data can be used to verify both SG-1 and SG-2 pH meter readings. SG blowdown, feedwater and condensate pH meters are verified by virtual pH values simulated by SCW on a weekly basis.
Conductivity
Waterford currently uses bench sampling with a separate in-line sampler to verify measurements from conductivity meters on a quarterly basis. To verify SCW’s virtual conductivity values, the plant studied the differences between the conductivity values obtained from the instruments, the theoretical calculations using SysChem, and SCW. As with pH, an excellent agreement was observed between the virtual conductivities predicted by SCW, the instrument data and the theoretical values calculated by SysChem.
Ammonia and ETA
 |
Waterford uses the DX-320 for ETA and ammonia analysis to monitor secondary pH control additives. Since the DX-320 is qualified daily prior to use by a second source quality control standard, it is not necessary to verify this data. However, since SCW uses feedwater ammonia and ETA concentrations as inputs, monitoring the difference in theoretical ammonia and ETA for steam generator blowdown and condensate would provide an indication of the accuracy of SCW modeling. Correlations of ammonia and ETA values calculated by SCW and those from SysChem and DX-320 analyses show excellent agreement (see Figure 2 for the ammonia comparison). However, the SCW virtual values appeared to be slightly underestimated compared to those from other methods at steam generator blowdown. At the end of July 2003, therefore, the SCW model was fine-tuned and the correlation improved significantly.
pH(t), pH(n), and Bench pH
Waterford had used AminMod to calculate pH values at temperature (pH(t)) in the secondary cycle using feedwater ETA and ammonia. Since neither SCW nor AminMod computer simulators are dynamic enough to compensate for actual system temperatures, a constant temperature is used for a given system. Because AminMod is a simplified model compared to the SCW simulator, there might be a small difference in pH(t) between the two models. However, this bias is not deemed to be significant since pH(t) is used as a diagnostic trending tool.
Neutral pH (pH(n)) values calculated by SCW and AminMod were compared, but little or no differences existed (typical differences were less than 0.02 pH units). Therefore, SCW can be used to calculate pH(t) and pH(n) values. The use of SCW increased the number of monitored parameters including pH(25C), pH(n), and pH(t) in main steam, heater drains and moisture separators. These virtual pH values are directly transferred into the WinCDMS on a specified frequency.
In-Line Data Transfer
SCW receives a digital signal from the plant computer (SDS) and then sends it digitally to WinCDMS. Transmitting signals from the in-line meter to the plant computer, however, incorporates analog-to-digital conversion and creates an opportunity for a breakdown in communication. Direct data transfer from SCW to WinCDMS could reduce error and lessen data entry requirements. To research this opportunity, Waterford tracked the SDS points against the actual online readings for several instruments including pH, conductivity, cation conductivity, dissolved oxygen and hydrazine.
 |
Small differences are expected in analog-to-digital conversion; large differences, however, indicate suspect data accuracy. Data transfer accuracy depends on several factors, including compression limits and how often the plant computer updates. As shown in Figure 3, all electronic points are within an acceptable deviation of the respective online reading. Control charts were established and maintained for electronic verification.
Once SCW receives data from SDS, SCW displays the data on Web pages via the Internet. Selected data, including both measured data and virtual sensor data, are sent to the plant WinCDMS at a preset schedule. Automatic in-line data transfer using SCW allows for the elimination of about half of the secondary manual log data entry performed each shift and provides a reliable, efficient and accurate way to record secondary data.
Conclusion
SCW has proven to be a reliable tool to verify instrument readings at Waterford. Its use has eliminated manual theoretical calculations performed with SysChem and AminMod on a weekly basis. This has not only reduced bench sampling but also increased the number of chemistry data points without adding any extra tasks for personnel. Automatic in-line data transfer using SCW also allowed for the elimination of about half of the secondary manual log data entry performed on each shift. The work reduction realized, as a result of the implementation of SCW at Waterford, was estimated at approximately 250 man-hours per year. p
References
1 EPRI ChemWorks - AminMod Version 4.0, EPRI 109560-P6, August 2001.
2 EPRI SMART ChemWorks - Volume 1: System Description and Specifications, EPRI TR-108739-V1, June 1998.
3 EPRI SMART ChemWorks - Volume 2: Implementation Roadmap, EPRI TR-108739-V2, June 1999.
4 Bourgeois, J., “SMART ChemWorks Improvement Plan for Waterford 3 Nuclear Power Plant,” Waterford internal documentation, May 2003.
5 Bourgeois, J., “SMART ChemWorks Instrument Verification Plan for pH & Ammonia at Waterford 3,” Waterford internal documentation, September 2003.
6 Bourgeois, J., “SMART ChemWorks to Secondary In-Line Transfer Plan,” Waterford internal documentation, October 2003.
Authors -
Samuel Choi is a Senior Applications Engineer at EPRISolutions, where he is responsible for continuous development of EPRI ChemWorks and implementation of SMART ChemWorks. Prior to joining EPRISolutions, Choi worked as a Senior Consultant at NWT Corporation, where he gained 19 years of experience relative to impurity transport and corrosion in nuclear systems. Choi holds degrees in chemical engineering from the University of California at Berkeley (BS) and the San Jose State University (MS).
Jeff Bourgeois is currently the Chemistry Instructor at Entergy Corp.’s Waterford 3 nuclear facility. He has seven years of experience at Entergy as a Chemistry Technician, an Instrumentation Specialist and a Work Control Coordinator. Bourgeois holds a bachelor’s degree in chemistry from Nicholls State University.
Membrane Methods for Makeup Pretreatment
By Brad Buecker, Contributing Editor
Makeup water production techniques for steam generating facilities continue to evolve in somewhat surprising fashion. A case in point is increasing interest in the membrane technologies of microfiltration and ultrafiltration (MF and UF) for suspended solids removal from makeup system inlet streams.
During the heyday of coal plant construction from the 1950s through the 1970s, the most popular technique for raw water suspended solids removal was clarification followed by media filtration, typically with graded sand. While clarification/filtration is a proven technique, the clarifiers of earlier generations were large and required substantial feed of coagulants and flocculants to effect proper solids removal. At today’s prices, annual coagulant and flocculant costs for an older unit may exceed $50,000.
As is true with many other technologies, clarifier design has improved significantly over the last half decade, and a number of water treatment firms offer advanced clarifier/filter combinations that perform quite well with low chemical dosages. However, membrane techniques are beginning to make inroads into the pretreatment market, and from this author’s observations, deservedly so.
Many readers are no doubt aware of, or perhaps even participated in, retrofit projects to place a reverse osmosis (RO) unit ahead of an ion exchange unit to improve demineralizer run times and reduce regenerant chemical costs. The typical RO system utilizes a spiral-wound membrane structure in which a flat membrane, along with spacer and support material, is wrapped many times around a perforated central core and enclosed in a pressure-tight element. The particular subtleties of membrane fabrication produce membranes with pores of only angstrom-unit diameters that screen all but some of the smallest ions. A problem with RO membranes is that the material is rather delicate and can be damaged or fouled by mechanical stress, scale buildups, microbiological growth, or chemical attack. Proper pretreatment of RO inlet feed is critical for long membrane life and top performance. Enter MF and UF.
Before discussing the merits of MF and UF, it might be worthwhile to examine the relative pore size of each. In the scheme of membrane structures, RO has the smallest pore size, followed by nanofilters, which can still remove divalent ions, followed by ultrafiltration, and then microfiltration. Osmonics, now a part of General Electric, once published a piece that outlined the relative pore sizes of RO, UF, and MF: RO pore size is similar to a dime compared to the Pacific Ocean, UF pore sizes equate to the size of a quarter, and MF pore sizes correspond to the size of a half dollar. Small indeed!
Micro or ultrafiltration pretreatment has become popular at potable water production plants due to the ability of the membranes to filter out pathological microorganisms. The cryptosporidium outbreak in Milwaukee in the mid-1990s comes quickly to mind. If these membrane techniques can remove bacteria and viruses from water, then they ought to be suitable for removing the suspended solids that would otherwise plague RO units and demineralizers. Because microfilter and ultrafilter pore sizes are much greater, relatively, than those in RO membranes, the pressure drop through MF and UF membranes is substantially lower. This allows different production techniques with much more durable materials. For example, a common material for MF membranes is the strong and chemically resistant plastic, PVDF.
Most if not all MF and UF membrane design is based on the hollow-fiber configuration. When raw water enters the pressure vessel, it passes along thousands of spaghetti-sized hollow fiber tubes. Some designs utilize an outside-in flow pattern, wherein the suspended solids flow along the outside of the membrane while the filtered water passes through the membrane walls and is collected at the opposite end of the pressure vessel. Other systems utilize an inside-out pattern. Constant discharge of the high-concentrate stream minimizes membrane fouling, but the membranes still collect solids. Virtually all systems are equipped with automatic membrane cleaning devices that reverse-flush the membranes, often in conjunction with an air agitation step to loosen solids. The fact that the membranes are constructed of durable material allows them to withstand the scrubbing mechanism.
Another benefit of strong material regards treatment for microbiological fouling. Many RO users are aware that membranes can serve as host sites for microorganism attachment and growth. MF and UF membranes are no different. However, PVDF is resistant to oxidizing biocides, so continuous feed of sodium hypochlorite or similar compounds is very beneficial for microbiological fouling control.
Practical results from MF and UF systems indicate 95 percent water recovery with effluent turbidities well below 0.1 nephelometric turbidity units (NTU). Also, with a bit of continuous chemical feed to the inlet, systems can literally run for months with little or no maintenance. The only major O&M cost is power for the feed pump. Manufacturers of microfiltration and/or ultrafiltration systems include Pall, USFilter and Zenon.
The next step, and one that some utilities are considering, is membrane processing of sewage plant effluent as the first step in makeup water production. Not only is this raw water typically very inexpensive, but the power plant would receive credit and accolades for water conservation.
In a future article I will outline my own personal, and very positive, experience with a microfilter for RO pretreatment. I had the opportunity to assist with startup and subsequently monitor a system that has operated virtually maintenance free while producing water with a turbidity of less than 0.05 NTU.