electrodeionization in power plant applications - SUEZ in power plant applications Authors: Brian P. Hernon, R. Hilda Zanapalidou, Li Zhang ... Power plants have stringent water quality

Water Technologies & Solutions technical paper

Find a contact near you by visiting www.suezwatertechnologies.com and clicking on “Contact Us.” *Trademark of SUEZ; may be registered in one or more countries. ©2017 SUEZ. All rights reserved. TP1073EN.docx Mar-10

electrodeionization in power plant applications Authors: Brian P. Hernon, R. Hilda Zanapalidou, Li Zhang, Keith J. Sims and Linda R. Siwak, Ionics

Presented at the 8th Annual Ultrapure Water Expo ‘94, Philadelphia, Pennsylvania, May 9-11, 1994. Reprinted with the permission of Tall Oaks Publishing, Inc.

Note: SUEZ purchased Ionics in 2005.

introduction

Electrodeionization (EDI) use is growing in the commercial-scale production of ultrapure water. In EDI, ion-exchange resins, ion-exchange membranes and a direct current (DC) electrical field are combined to provide very high levels of demineralization.

Essentially the EDI process is a modification of conventional electrodialysis (ED) systems with ion-exchange resins installed in the ED stacks. The resin reduces the electrical resistance of the unit, especially for dilute solutions. As the product becomes more pure, the DC voltage splits water to hydrogen and hydroxyl ions which continuously regenerate a portion of the resin. The main advantages of the EDI process include continuous operation, stable product quality, and the ability to produce high-purity water without the need of chemical regeneration.

One area where EDI technology is gaining momentum is the production of ultrapure makeup water in nuclear and fossil-fuel power plants. Power plants have stringent water quality requirements to reduce corrosion and scaling and the associated expensive down-times. In pressurized water reactor (PWR) nuclear plants, high-purity makeup water is most important in reducing corrosion in steam generators. In boiling water reactor (BWR) nuclear plants, high-purity water is most important in maintaining water quality in the nuclear reactor.

Traditionally, makeup-water systems have been ion exchange-based. To improve makeup-water quality, to counter rising chemical regeneration cost and to extend regeneration cycles, membrane-based pretreatment systems have been installed to reduce the ionic load to the ion-exchange beds. Now the EDI process makes it possible to eliminate ion-exchange beds except for portable polishers.

Portable vessels of ion-exchange resins are economical because the very low ionic load to the resins corresponds to long bed life, often exceeding six months. Also portable ion exchange often produces higher quality water than large in-situ ion-exchange systems since smaller resin volumes are involved and regeneration occurs in a dedicated facility where the regeneration of resin can be tightly monitored and controlled.

In this paper, we present data from the operation of EDI units installed in water treatment systems that deliver ultrapure water at three power stations. These water treatment systems employ EDI installed downstream of reverse osmosis (RO) units in an RO/EDI combination. We also compare the overall performance of such and RO/EDI system to that of a double-pass RO (RO/RO) system. In both configurations, ion exchange is required as a polishing step only.

EDI performance in power plants

EDI Performance at Grand Gulf Nuclear Station

At Grand Gulf Nuclear Station (GGNS) in Port Gibson, MS, a Triple Membrane Trailer (TMT) followed by portable mixed-bed ion exchange has been used since 1988 to provide 50 gpm (0.2 m3/h) of ultrapure water to the Nuclear Station1. Raw water is supplied to the water treatment system by the Mississippi River. The feedwater characteristics include a conductivity of 400-500 µs/cm, pH of 7.2 to 7.6, about 20 ppm (mg/l) of silica, and temperature varying from

Page 2 TP1073EN.docx

55°F to 81°F (13°C to 27°C). The makeup-water specifications require a conductivity of less than 0.06 µs/cm, TOC of less than 50 ppb, less than 1 ppb ions, and less than 5 ppb silica. The TMT system, comprising ultrafiltration (UF), electrodialysis reversal (EDR), and RO, has been described elsewhere2.

In September 1991, a 50-gpm (0.2 m3/h) EDI unit was installed downstream of the RO to increase ion-exchange-bottle life. The overall system now consists of multimedia filtration, activated-carbon filtration UF, EDR, RO, and EDI, followed by portable demineralizers used as a polishing step. No chemical softening, sometimes used before EDI to prevent scaling, is needed in this system.

The EDI unit consists of a single membrane stack,3,4 a power supply and a hydraulic skid. Two pumps are required, one to recycle the concentrate stream and one to repressurize the EDI product supplied to the ion exchangers downstream.

The TMT/EDI system has been operated on a demand basis, on average about 50% of the time. After a period of about 7,000 operating hours, the EDI stack installed in 1991 was removed for inspection and replaced by a new, slightly redesigned 50-gpm stack in July 1993. The original EDI stack accomplished conductivity removal in the range of 95-98% and silica removal of 96%. This reduction in the ionic load prolonged the demineralizer regeneration cycles by a factor of 20 compared to operation of the TMT system without EDI.5

Since July 1993, the new EDI stack has operated for over 1,800 hours at an average power consumption of about 1.3 kwh/kgal. To monitor the stack performance, EDI feed conductivity and product resistivity, have been determined on-line using a Thornton Dot Two analyzer. Silica measurements were taken on line using a Hach, Series 5000 Silica Analyzer, programmed for alternating EDI feed and product sampling.

Figure 1 shows the time-averaged conductivity removal, where 100% removal corresponds to the minimum conductivity achievable due to normal water dissociation. Throughout most of the operating period, the average conductivity removal was over 99.5%. Between hours 780 and 900, upsets occurred because alternative feedwater sources were used during a power plant refueling outage.

Figure 1: EDI Conductivity Removal, GGNS (Power Plant Refueling Outage: 780-900 hours)

Figure 2: EDI Product Quality, GGNS

Figure 2 is a plot of the time-averaged EDI product resistivity, corrected to 77°F (25°C). For the first 600 operating hours, the average product resitivity was about 15.5 megohm-cm. Since then, the performance improved to an average product resistivity of about 17.5 megohm-cm. The increase in product resistivity can be attributed to an improved EDI feed quality, as indicated by the decreasing silica level shown in Figure 3.

Figure 3: EDI Feed Silica, GGNS

Figure 4 shows the time-averaged EDI-product silica level during the operating period. The silica level in the EDI product ranged from 10 ppb to 0.5 ppb, the detection limit of the on-line silica analyzer. For the last 600 hours, the silica level was consistently below 2 ppb, corresponding to silica removal of 99.8% (Figure 5), a remarkable performance for an EDI unit without any pH adjustment of the feedwater.

TP1073EN.docx Page 3

Figure 4: EDI Product Silica, GGNS

Figure 5: EDI Silica Removal, GGNS

In addition to the on-line silica measurements, EDI feed and product samples were analyzed for specific-ion content using capillary electrophoresis. Table 1 lists typical EDI feed and product compositions. The content of all ions in the EDI product was found to be below the detection limit of the analytical method. This TMT system comprises two trailer-mounted 100-gpm (0.4 m3/h) trains.

Table 1: Typical EDI Feed and Product Compositions at GGNS

EDI Feed (ppb) EDI Product (ppb)

Na+ 750 < 1

Ca++ 70 < 1

Mg++ 18 < 1

Cl- 32 < 2

SO4- 44 < 4

Raw water is obtained from the Russellville municipal supply, which is clarified and filtered surface water normally taken from the Illinois Bayou. Total dissolved solids (TDS) are typically in the 45 to 85 ppm (mg/L) range but can exceed 200 ppm (mg/l) during periods of high water demand. Silica averages 6.0 ppm (mg/L), and TOC ranges from 1 to over 4 ppm (mg/l). Media and activated-carbon filtration are used as pretreatment steps to the TMT system.

In April 1993, a 200-gpm (0.8 m3/h) EDI trailer was installed at ANO. The EDI trailer has two identical EDI

units, each with 100-gpm (0.4 m3/h) capacity. Similar to GGNS, the EDI units were installed between the RO units and the final ion-exchange step (Figure 6). However at ANO, each EDI unit consists of two parallel membrane stacks, a power supply and a hydraulic skid with the pumps needed for the concentrate stream recycle and the product repressurization. As at GGNS, no chemical softening is used before EDI, and the system operates on a demand basis.

The EDI units at ANO have been operating at 95% water recovery and have required no pH adjustment. Since April 1993, both EDI units at ANO have been running for over 3,200 hours. Table 2 shows the typical performance of one EDI unit at ANO for individual ion removal. EDI feed and product samples were analyzed with ion chromatography by Balazs Analytical Laboratory, Sunnyvale, CA.

Figure 6: Makeup Water Treatment at Arkansas Nuclear One

The time-averaged EDI product resistivity of a single EDI unit from April 1993 to February 1994 is shown in Figure 7. The product resistivity ranged from 12 megohm-cm to 17.6 megohm-cm, which reflects the changes of feedwater quality and operating conditions. After optimizing the operating conditions, product resistivity has stabilized in the 17 megohm-cm range. In Figure 8, the time-averaged conductivity removal (corrected for conductivity due to normal water dissociation) is plotted for the same EDI unit. Conductivity removal ranged from 97.0% to 99.8%.


Table 2: Typical EDI Feed and Product Compositions and Specific Ion Removal at ANO

EDI Feed (ppb)

EDI Product (ppb)

Rejection (%)

Na+ 595 4.4 99.26

K+ 80 0.31 99.61

Cl- 40 0.85 97.88

SO4- 11 < 0.1 99.09

NO3- 32 < 0.1 99.69

The change of feedwater temperature over the period is given in Figure 9. Silica removal of the EDI unit is dependent upon operating conditions and, in particular, the EDI feed temperature. Figure 10 shows the effect of feedwater temperature on silica removal. The temperature dependence is most pronounced at lower temperatures. Nevertheless, a silica-removal performance better than 95% was obtained as long as the feedwater temperature was above 61°F (16°C). The average EDI stack power consumption at ANO has been about 0.8 kwh/kgal.

Figure 7: EDI Product Quality, ANO

Figure 8: EDI Conductivity Removal, ANO

Figure 9: EDI Feed Temperature, ANO

Figure 10: Temperature Effect on EDI Silica Removal, ANO

The main benefit of using EDI in the makeup water treatment systems is the prolongation of ion-exchange-bottle life. EDI can extend ion-exchange bottle life more than 20-fold.5 For example, at GGNS, no bottle replacement has been required in the seven months since the installation of the new stack. The reduced frequency of bottle changeout not only reduces operating cost and bottle traffic, but is also very important for reliably producing high-quality water since it reduces the risk of downstream contamination by resin fines or chemicals from newly regenerated ion-exchange beds. The EDI units have required virtually no operator attention or maintenance.

comparison of RO/EDI to double-pass RO

Membrane processes, especially RO and EDR, have been widely used to reduce the ionic load and thus the frequency of chemical regeneration of ion-exchange systems. Neither of these two processes can replace the ion-exchange step. Two double-membrane processes which are considered potential replacements of ion-exchange systems for the supply of ultrapure water are RO/RO and RO/EDI.

Double-Pass Reverse Osmosis

RO/RO has been in use since the mid-1980s. In this process, the permeate from the first RO unit is used as feed to the second RO unit. This double-membrane process has been successfully used to produce high-quality water for many applications.6,7 The RO/RO process was originally projected to replace conventional ion-exchange systems on a wide scale.7 However, RO/RO has some significant limitations. Due to the nature of RO membranes, the rejection of salts is usually considerably lower in the second-pass RO unit than in the first unit.8 Thus, the benefits obtainable from combining RO steps are not as great as initially expected.

The difference between first-stage and second-stage rejection characteristics can be seen (Table 3) in data


from a 600-gpm (2.3 m3/h) ultrapure water system owned and operated by Ionics (now SUEZ). In this system, an RO/RO unit is incorporated in a multi-step purification process that uses UF and EDR upstream of the RO/RO, which is then followed by vacuum degasification for oxygen removal and ion exchange. The specific-ion removal of the RO/RO system is excellent; however the actual rejections in the second-pass RO are, as expected, significantly lower than in the first pass.

These data are similar to data reported by Lesan, et. al.8 in their investigation of ion removal in two-stage RO systems and with low TDS feedwaters (Table 4).

Reverse Osmosis/ Electrodeionization

A relatively new combination of membrane processes uses RO and EDI processes together in the production of ultrapure water. The RO unit provides excellent pretreatment, for EDI on most feed waters. RO removes contaminants in feed water that could

cause problems in an EDI unit. Particulate matter, which is difficult to remove from an EDI stack without removing the resin from the stacks, is completely removed by the RO process. Calcium hardness, which at high enough levels necessitates chemical addition to control calcium scale in the EDI stack, will be reduced by over 95% by the RO process. Organics in the feedwater, which can foul the EDI resins, will also be removed by RO. RO followed by EDI, in many cases, is thus able to continuously produce water that is comparable to mixed bed ion-exchange-treated water. In all of SUEZ’s industrial-scale plants that now incorporate the EDI process for ultrapure water production, RO is used immediately in front of the EDI unit.

SUEZ’s industrial-scale plants that now incorporate the EDI process for ultrapure water production, RO is used immediately in front of the EDI unit.

Table 3: Double-Pass RO Performance, Diablo Canyon

Table 4: Double-Pass RO Performance

Source: ref. 8, Lesan et. al.

Table 5: RO/EDI Performance, Grand Gulf and Turkey Point

First Stage RO Second Stage RO Overall

Feed (ppb) Product (ppb) Rejection (%) Product (ppb) Rejection (%) Rejection (%)

Na+ 19,500 561 97.1 153 72.2 99.22

Ca+ 568 17 97.0 8 52.9 98.59

Mg++ 368 4 98.9 2 50.0 99.46

Cl- 24,400 155 99.3 3 98.1 99.99

SO4- 750 9 98.8 7 22.2 99.07

SiO2 3,300 35 98.9 4 88.6 99.88

First Stage RO Second Stage RO Overall

Feed (ppb) Product (ppb) Rejection (%) Product (ppb) Rejection (%) Rejection (%)

Na+ 8,600 500 94.1 70 86.0 99.18

Ca+ 14,100 220 98.4 40 81.8 99.72

SiO2 4,900 190 96.1 < 10 > 95.0 > 99.80

Grand Gulf Turkey Point

RO Feed (ppb) EDI Product (ppb)

Overall Rejection (%)

RO Feed (ppb) EDI Product (ppb)

Overall Rejection (%)


SUEZ currently operates five systems which utilize the RO/EDI combination. As discussed previously, two systems are installed at ANO and one at GGNS. These installations employ EDR upstream of the RO/EDI system. In September 1993, SUEZ installed two 100-gpm (0.4 m3/h) RO/EDI units at the Turkey Point Fossil Power Station, just south of Miami, Florida. These two plants do not use EDR ahead of the RO/EDI process (flow diagram in Figure 11, installation photograph Figure 12). Table 5 summarizes specific ion removal by RO/EDI at two of the sites. RO feed and EDI product were analyzed by ion chromatography and capillary electrophoresis, respectively.

At GGNS, the feedwater to the RO unit ranges from 80 to 160 µs/cm. EDI product water at GGNS averages 16.5 megohm-cm (Figure 2). At Turkey Point the feedwater to the RO unit ranges from 850-900 µs/cm. EDI product at Turkey Point ranges between 6-10 megohm-cm. The data from GGNS show that essentially semiconductor-grade ultrapure water can be produced in the RO/EDI system. The Turkey Point data show that RO/EDI product quality in the ppb and sub-ppb range is produced from a relatively high 890 µs/cm feed water.

Figure 11: Makeup Water Treatment at Turkey Point

Figure 12: SUEZ 240 GPM Triple Membrane Trailer

Both RO/RO and RO/EDI systems can produce high-purity water. A major advantage of RO/EDI is that much higher ionic-purity levels are achieved when compared to RO/RO, particularly for monovalent ions such as sodium and chloride. Although overall ionic rejection in a RO/RO system is excellent, the specific ion levels remaining in a RO/RO product stream can be an order of magnitude higher than the specific ion levels in an RO/EDI product.

conclusions

The performance of EDI units in use at GGNS and ANO has been excellent. EDI Product water is consistently over 10 megohm-cm at both sites. Long-term data show no degradation of the EDI system. Data from the more recent installation of EDI at Turkey Point also show consistently high product quality. These EDI systems with a combined capacity of 450 gpm, are integral in the production of ultrapure water for commercial power plants. The success of EDI at these sites demonstrate the viability of EDI as a demineralization process in the industrial production of ultrapure water.

The combination of RO and EDI processes produces water quality that is at or near the standard definition of ultrapure water. The use of RO/EDI as a

Na+ 7,000 < 1 > 99.99 26,000 11 99.96

Ca+ 4,500 < 1 > 99.98 75,000 < 3 > 99.99

Mg++ 1,300 < 1 > 99.92 2,000 < 0.5 > 99.99

Cl- 900 < 2 > 98.78 48,000 4 99.99

SO4- 7,600 < 4 > 99.95 7,000 6 99.99

SiO2 12,000 < 5 > 99.96 3,300 12 99.64


replacement for conventional ion exchange systems should increase over the next few years. The safety and environmental advantages are compelling: no need for large quantities of acidic and caustic chemicals for regeneration of spent resins; no regenerant waste disposal; and the continuous nature of the RO/EDI process, producing consistent-quality water reducing risks of downstream contamination.

We anticipate that further improvements of RO/EDI systems will eventually make it possible in many cases to completely eliminate the need for an ion-exchange step in the production of ultrapure water.

references

1. Smith, G.O.; Wilson, K.S., “Makeup Water Treatment Utilizing Triple Membrane Demineralizers at Entergy Operations, Inc.’s Grand Gulf Nuclear Station”, Proceedings of Ultrapure Water Expo ‘90 West, Conference on High Purity Water, San Jose, CA, November 1990.

2. Valcour, H.C., “Triple Membrane Water Treatment at Four Nuclear Power Plants”, Proceedings of International Water Conference, Pittsburgh, PA, October 1991.

3. Katz, W.E.; Elyanow, I.D.; Sims, K.J., U.S. Patent 5,026,465, June 1991.

4. Parsi, E.J.; Sims, K.J.; Elyanow, I.D.; Prato, T.A., U.S. Patent 5,066,375, November 1991.

5. Coker, G.; Sims, K.J.; Zhang, L., Elyanow, I.D.; Williamson, T., “Makeup Water Treatment Incorporating Electrodeionization Along with Triple Membrane Treatment at Grand Gulf Nuclear Station”, Proceedings of International Water Conference, Pittsburgh, PA, October 1992.

6. Comb, L.; Schneekloth, P., “High Purity Water Using Two-Pass Reverse Osmosis”, Ultrapure Water, April 1990, p. 49.

7. Pittner, G.A., “Unique Double-Pass Reverse Osmosis System Eliminates the Need for Many Deionization Applications”, Ultrapure Water, September/October 1986, p. 23.

8. Lesan, R., “A Comparison of Different Classes of Spiral-Wound Membrane Elements at Low Concentration Feeds”, Ultrapure Water, April 1990, p. 18.

electrodeionization in power plant applications - SUEZ in power plant applications Authors: Brian P. Hernon, R. Hilda Zanapalidou, Li Zhang ... Power plants have stringent water quality

Documents