- Email: [email protected]
BWR pipe crack control using hydrogen water chemistry: Status report on Dresden-2 program
Nuclear Engineering and Design 89 (1985) 505-512 North-Holland, Amsterdam
B W R PIPE C R A C K C O N T R O L ON DRESDEN-2 PROGRAM
505
USING HYDROGEN WATER CHEMISTRY:
J.T. A d r i a n R O B E R T S , R o b i n L. J O N E S , M i c h a e l N A U G H T O N
STATUS REPORT
a n d A l b e r t J. M A C H I E L S
Nuclear Power Division, Electric Power Research Institute, Palo Alto, California, USA
Received 22 February 1985
One of the proposed remedies for intergranular stress corrosion cracking of stainless steel piping in BWRs is an alternative water chemistry called hydrogen water chemistry (H2WC) that involves suppression of reactor water dissolved oxygen to _< 20 ppb via hydrogen injection to the feedwater in conjunction with control of conductivity to < 0.3 /tmho/cm. A long-term verification program, over two or three 18 month fuel cycles, was started at Commonwealth Edison's Dresden-2 reactor in April 1983 (Cycle 9). This paper describes the results of the water chemistry changes, structural material and fuel evaluations, and plant radiation level changes during Cycle 9, which ended in October 1984. To date the results of the verification program are very encouraging. They indicate that the alternative water chemistry, based on hydrogen additions to the feedwater to suppress oxygen and low conductivity, can be maintained in a large operating BWR, and that it does mitigate IGSCC in stainless steel recirculation piping. Monitoring of fuel and plant materials will continue in Dresden-2 at least through Cycle 10 to confirm the absence of any unusual side effects of this remedy for IGSCC.
!. Introduction
2. Basis for an alternative BWR water chemistry
A major environment-related materials performance problem encountered in the reactor coolant system of BWRs has been intergranular stress corrosion cracking (IGSCC) of sensitized austenitic stainless steel. I G S C C of sensitized material adjacent to welds in type-304 and type-316 stainless steel piping systems has been responsible for more than 400 cases of pipe cracking over the last ten years [1]. Athough these cracks are not thought to pose a major safety concern, inspections and repairs associated with pipe cracking have proved costly to the utilities and substantial R & D programs have been undertaken to understand the I G S C C phenomenon and develop remedial measures [1-3]. Much of the early remedy-development work focused on alternative materials or local stress reduction, but recently, the effects of water chemistry parameters on the I G S C C process have received increasing attention in work funded by EPRI, the B W R Owner's Group, and the U S N R C . A complete understanding of the interrelationship between B W R water chemistry variables and I G S C C of sensitized stainless steels has not yet emerged but some important features have been identified.
Water itself is relatively innocuous towards stainless steels even at the relatively high temperatures involved in its use as a working fluid in electric generation plants. However, impurities that enter the water can make it an aggressive environment towards sensitized stainless steel. Oxygen is generated as a result of radiation in the core of a BWR during normal operation and this leads to a steady level of - 2 0 0 ppb dissolved oxygen in B W R water. Ionic impurities (salts) enter the water from sources such as makeup water and condenser leaks. These are controlled by condensate and reactor water cleanup systems which, however, can be costly to operate. The objective of the overall EPRI program on impurities in BWR water is to determine the effects of impurities on I G S C C so that guidelines for water quality to minimize pipe cracking can be formulated on a quantitative basis. Impurities that exacerbate pipe cracking fall into two classes depending on their mode of action. The oxidizers, represented by oxygen dissolved in the water, increase the chemical driving force for the corrosion reactions; and, because the reactions are electrochemical in nature, ionic impurities, such as chloride ions, increase their rates by increasing the electrical conductivity of the
0 0 2 9 - 5 4 9 3 / 8 5 / $ 0 3 . 3 0 © E l s e v i e r S c i e n c e P u b l i s h e r s B.V. (North-Holland Physics Publishing Division)
506
J.T.A. Roberts et al. / B W R pipe crack control
I Oxygenco~letllOt J corrosionpolent]al
Bw.r,no,LI
aWRH2WE;Z range L
iOr Impurity concenttal~on J conducl~'vTly Fig. 1. Schematic summary of the results of laboratory studies of effect of impurities on SCC of stainless steel.
water. Fig. 1 represents current thinking about the synergism between the effects of the two classes of impurities on IGSCC [4]. It is apparent that the contents of both classes of impurities must simultaneously be minimized to minimize the likelihood of IGSCC. This concept underlies an alternative BWR water chemistry, known as hydrogen water chemistry (H2WC), which entails oxygen suppression via hydrogen addition to the feedwater in combination with conductivity control via optimized plant operational procedures. The remainder of this paper presents a status report on the development and in-plant verification of H2WC. 2.1. B a c k g r o u n d
In 1977, the US Department of Energy sponsored a program designed to identify additives suitable for oxygen suppression in the BWR, to determine their possible impact on various plant systems, and to select the best additive and test bed. The additive selected was hydrogen, and the demonstration plant selected was Commonwealth Edison Company's Dresden-2 reactor in Morris, Illinois. In 1982, General Electric, DOE, Commonwealth Edison, and EPRI participated in a one-month hydrogen addition feasibility demonstration in Dresden-2 [5].
This one-month demonstration included in-reactor stress corrosion cracking (SCC) tests on furnace-sensitized stainless steel and low alloy steel specimens and electrochemical potential measurements on various BWR structural materials. The results showed that the coolant oxygen levels in the GE designed BWR can be suppressed to below 20 ppb during power operation by adding practical amounts of hydrogen to the feedwater, on the order of 1.5 ppm. Reactor water having an oxygen content of 20 ppb and a conductivity of 0.30 p, m h o / c m was shown to be insufficiently aggressive to promote either IGSCC in sensitized stainless steel or transgranular stress corrosion cracking of pressure vessel steel in short-term, slow strain-rate tests. At low oxygen levels, the quantity of N-16 in the steam increased as expected, but the consequent four to fivefold increase in steam line gamma radiation was not found to be a major problem in Dresden-2. No other significant adverse effects of H2WC were identified in the relatively short-term demonstration. The next step in the H2WC program, a long-term verification over two or three 18 month fuel cycles started at Dresden-2 in April of 1983 (Cycle 9) under EPRI funding. It involves extensive monitoring of fuel and core materials behavior and continued evaluation of plant structural materials behavior in longer term tests.
J.T. A. Roberts et al. / B WR pipe crack control
507
TURBINE
REACTOR PRESSURE VESSEL DEMINERALIZER
®
FEEDWATER SYSTEM HYDROGEN INJECTION PURIFICATION SYSTEM
POTENTIAL MEASUREMENTS
CONSTANT EXTENSION RATE TEST
Q
CONDENSATEDEMINERALIZER EFFLUENT (CDE)
Q
FINALFEEDWATER (FFW)
Q
COREEXIT WATER (RCX) (PLANNED BUT NEVER INSTALLEDI
Q
RECIRCULATIONLOOP (RRC)
Q
CLEANUPINLET (RCI}
Q
CLEANUPOUTLET (RCE}
Q
MAIN STEAM (MST)
t
Fig. 2. Hydrogen addition flowsheet at Dresden-2.
The water chemistry goals are < 20 ppb oxygen and a reactor water conductivity of < 0 . 3 / ~ m h o / c m . The following paragraphs summarize the results to date on this program, problems encountered and their resolution, and system upgrade activities currently under way. This is an update of an earlier paper on this program presented at the 1984 American Power Conference [6].
2.2. Plant operations
The hydrogen addition flowsheet is shown in fig. 2. High purity (99.999%) hydrogen is purchased as gas which is delivered in tank trucks carrying 115000 or 50000 scf at 2400 psi. These tanks supply hydrogen at about 200 psi through a ~" carbon steel line into a flow control panel in the condensate pump room.
50g
J. 7~ A. Roberts et al. / B WR pipe crack control
This panel varies the hydrogen flow in proportion to feedwater flow to give a constant but adjustable hydrogen concentration in the reactor feedwater. Maintenance of feedwater hydrogen at 1.5 ppm results in dissolved oxygen concentrations of 20 ppb or less throughout the primary coolant system. Oxygen from a liquid oxygen storage trailer provides 60 psi oxygen gas to add into the off-gas system at the first stage steam jet air ejector (SJAE). Oxygen flow is controlled manually based on the reading of oxygen meters sampling the recombiner outlet gas. A steadystate oxygen flow is set to maintain the oxygen concentration at the recombiner outlet between 8 and 12 vol%. When hydrogen injection rate increases are to be made, oxygen flow is increased first to be sure there is always excess oxygen. The hydrogen addition system is not operated continuously. It is shut off for a variety of reasons, including maintenance activities on the hydrogen addition system itself, maintenance in other areas of the plant where there are high N-16 radiation levels due to hydrogen addition, and during the extinguishing of fires in the off-gas treatment system. In addition, during reactor startups and shutdowns, hydrogen addition is not used below 220 MW (e) (i.e., 25% power). When the injection system is operating, it significantly reduces the oxygen concentration in the reactor water. Overall, the injection system availability has been about 87%, During 1984 the injection system was available for about 90% of the time and the oxygen was controlled below 20 ppb for 86% of the time (fig. 3). It is worthwhile to investigate the nature of the time the reactor water was greater than 20 ppb oxygen because of the existence of a corrosion potential " m e m o r y effect" discussed later in the paper. As a result of this effect, the corrosion potential stays below the critical
99.999 99.990 z ~ 99.900 ~>. 99.000
--
~ 90.000
--
o
60.000 40.000 20.000 10.000
2.3. Plant water chemistry
Control of ionic impurities has been particularly good in Dresden-2 for the first 18 month cycle of H2WC. Conductivity has been maintained at less than 0.2 b t m h o / c m for 98% of the time (fig. 4), and pH has been maintained in the 7 to 7.5 range. Ion chromatography studies of the coolant at Dresden-2 have shown that under continuous operation conditions, no more than 30 ppb carbonate, 5 ppb chloride and 5 ppb sulfate were present. The balance of the conductivity is made
99.9999 99.9990
4
2 ~0 o r,, "0
-I
1.00o ~__ o.100
g O0
O.OLO --=l IIIl~lll IIIII1[1[ IIIIIIIII o.001 1.0
10.0 Oxygen (ppb)
100.0
---3
[
II IIII
-4
1000.0
Fig. 3. Probability plot of all oxygen data during Cycle 9.
J
99.9900 2 99.9000
3
z z -SS
o.1
value for IGSCC for at least 10 h after shutdown of the hydrogen injection system. The statistical data base shows that greater than 70% of the cumulative time the reactor water was above 20 ppb for periods of less than 10 h in any given 24 h period. This implies that over 93% of the total cumulative operating time since startup of Dresden-2 in April 1983 has been in the corrosion potential region that is " i m m u n e to I G S C C " for conductivity levels _< 0.3 / z m h o / c m . The major problem causing shutdown of the H2WC system has been off-gas fires. An extensive evaluation of the reasons for the off-gas fires revealed they were not directly related to H2WC but, rather, to the design of the off-gas piping system. The probable ignition source is migrated catalyst in the off-gas trains, and flow or pressure changes cause ignition by fluffing the catalyst. Since the fires are located in the SJAE, after the condenser, the recommendations for eliminating the fires include elimination of a bypass line in the A B off-gas trains (which is unique to the Dresden and Quad Cities Units), followed by poisoning or removal of the migrated catalyst in the remaining lines. Also, a change in the oxygen addition point to the steam diluted downstream region is being considered.
-- 4
99.0000 z ~ 90.0000o 60.0000 ---
- - 2 ._o g
o_ 40.0000 20.0O00-
m.oooo 5 1.oooo E= 0.1000 _ _ o
o.0100
0.0010 0,0001
0.01
-3 03
I [ II
IIIITT -4 0.10
1.00
Conductivity (microsiemenslcm)
Fig. 4. Probability plot of all conductivity data during Cycle 9.
J.T.A. Roberts et al. / B WR pipe crack control 5O
509
~
0.05
elements that were analysed all show similar patterns, with no element showing an upward trend with time. The concentration of the other insoluble metals is generally less that 0.1 ppb. In the reactor water, the dominant activation product is 6°Co, but no clear trend, either upward or downward, was observed. Values between 0.1 and 0.2 i.tCi/I are consistent with data from other BWRs. Of the soluble species that were monitored in the feedwater, cobalt is the element of greatest concern because of its activation to 6°Co in the reactor core and accompanying potential for radiation buildup. Fig. 6 shows the concentration of soluble cobalt in the feedwater as a function of time. The downward and stable trend of cobalt is readily apparent. This trend pattern is also observed for the other elements that were analyzed. In contrast, there was no definitive upward or downward trend in the cobalt data in the reactor water; average values of 40 ppt for soluble Co agree with measurements from other BWRs. Because of the operational practice changes to the condensate treatment system that were implemented during Cycle 9, it cannot be explicitly concluded that these trends are a result of hydrogen water chemistry. Nonetheless, there appear to be no adverse consequences of hydrogen water chemistry for soluble species in either the reactor water or feedwater. Summarizing the corrosion product data, we have not seen any indication that hydrogen water chemistry has caused any detrimental change in the soluble and insoluble corrosion product transport in the feedwater or the reactor water. In the feedwater, insoluble spikes are common to all BWRs, and clearly will propagate to the reactor water. The downward trend in feedwater solubles we feel is the result of a gradual implementation of good operational practices centered around the management of the condensate treatment system. The concentrations of the impurities in the feedwater and reactor water are consistent with data from other deep bed plants.
8
0.04
2.4. Radiation levels
0.03
An important concern associated with H2WC is the impact on radiation levels in the turbine building, plant environs and off-site. The earlier 30 day test had indicated an increase on the order of a factor of 4 to 5 in N-16 in the steam. Detailed measurements have now been made in and around the plant, indicating that although the impact of H2WC is significant and measurable in the vicinity of the turbine building, it diminishes rapidly away from the plant and is negligible
O.
40 30
e-
20 C)
o
10 0 0
40
80 120 160 200 Days Since July 15, 1983
240
280
Fig. 5. Trend in feedwater insoluble iron during Cycle 9.
up of hydronium and hydroxyl ions. Thus, the species making up the conductivity at Dresden-2 are less aggressive than the sodium sulfate used to assess conductivity effects in the laboratory studies [7]. In addition to monitoring conductivity, pH, and oxygen, the H2WC verification program also includes the measurement of various metals and isotopes in the feedwater and reactor water. Soluble and insoluble corrosion products have been collected continuously, with samples changed at roughly three-day intervals. Each sample fraction has been analyzed at Vallecitos Nuclear Center for Fe, Cu, Ni, Co, Zn, Cr, and Mn. The dominant impurity in the Dresden-2 feedwater is insoluble iron. Fig. 5 shows the concentration of insoluble Fe as a function of time for a period since July 15, 1983. The 20 ppb spikes are the result of several condensate demineralizer changeouts over a short interval; the 40 ppb spike encompasses an orderly shutdown for pipe crack inspection and the restartup of the reactor. No long-term adverse consequences of hydrogen addition are evident. The time-base plots for other
0.06
8
0
0.02 0.01
I
0
0
40
80 120 160 200 Days Since July 15, 1983
I 240
Fig. 6. Trend in feedwater soluble cobalt during Cycle 9.
280
510
J.T.A. Roberts et aL / B W R pipe crack control
0
280
-100-
~
-2o0-
H2 - 0 n ,Gsoc
-
-300 -400
--
-
240 200 160
-,
120
"(~l -500 --1
Zk. 02 (ppb) rn Gold
-600 : ~
O Type 304 SS
-700 - ~ 2 ° ' 1 1200 1600 2-5-84
[
80 40
1 I 2000 Time
o 2400 400 600 2-6-84
Fig. 8. Example of electrochemical potential memory effect during hydrogen water chemistry.
Fig, 7. Net contribution of hydrogen water chemistry to Dresden-2 environs dose rate (#R/h).
off-site. Fig. 7 indicates the contribution of H2WC to the total dose rate on a series of contours drawn from the turbine building. Addition of Hydrogen Water Chemistry at Dresden-2 has been found to add about 10 man rem to the 2000 rein currently incurred each year. However, in terms of exposure for IGSCC-related repairs, H2WC is A L A R A effective. For example, pipe replacement programs typically lead to exposure of 1500 to 2500 man rem. It must be noted, that the Dresden-2 plant layout and site location are favorable for reduced radiological effects of the H2WC system. Other plants might not be so favorably constructed or situated and implementation of H2WC might require costly shielding or changes to operation and maintenance procedures.
2.5. Fuel and plant materials characterization Slow strain rate, crack growth, and constant load tests on a variety of plant structural materials are being undertaken in-plant to supplement the extensive laboratory data already available [7]. These in-reactor tests are conducted in special autoclaves using reactor water fed from a header in the recirculation system. The results of stress corrosion and electrochemical potential studies provide strong evidence that stainless steel stress corrosion cracking activity at Dresden-2 has been arrested. A series of in situ stress corrosion tests were conducted on furnace-sensitized stainless steel with the tests ongoing
over about 2000 h. No cracking was observed in either smooth or I G S C C precracked specimens tested while the plant was operated with reactor water at 20 ppb O 2 or less and reactor water conductivity at 0 . 3 / ~ m h o / c m or less. Electrochemical potential measurements on stainless steel made at the plant for over 3000 h revealed potentials well below - 3 5 0 mV (SHE) over 95% of the time. This potential is considered to be a threshold for I G S C C for the conductivity values listed above. Coolant oxygen content excursions up to 200 ppb for limited periods of time for up to about 15 h during the constant extension rate tests did not result in IGSCC. Electrochemical potentials during these short-term excursions did not increase to the values observed under normal operation at 200 pbb oxygen indicating the presence of a memory effect (fig. 8). Perhaps the most revealing test result is the constant extension rate test conducted on a furnace-sensitized and laboratory I G S C C precracked specimen that had seven hours of in-plant test time with 200 ppb oxygen. The balance of the in-plant test time was within the H2WC regime. No I G S C C in addition to that in the precracked phase was noted. This result is consistent with in-service inspection observations. The recirculation piping was inspected in November 1983 and December 1984 in response to an N R C directive since I G S C C flaws had been detected earlier in a large-diameter line. The inspections revealed no further growth of the existing flaws. Additional in-plant and laboratory studies have shown that stress corrosion a n d / o r corrosion fatigue of the other major structural alloys including Inconel 600, carbon steel, and low alloy steel are effectively eliminated by application of HzWC. No effects were found for martensitic stainless steel. Some increase in carbon steel general corrosion rates can be expected,
J.T.A. Roberts et al. / B W R pipe crack control
but resultant corrosion is well within design tolerances [7]. The concern over the long-term behavior of fuel, particularly with respect to hydrogen pickup and hydriding, prompted a plan for extensive examination of fuel rods and bundle hardware. For this purpose, four leading test assemblies (LTAs) of carefully characterized fuel components have been inserted at the beginning of the first cycle of hydrogen addition. Zircaloys with a known, precharacterized range of corrosion behavior representative of the range usually found with cladding batches used in BWR fuel fabrication have been selected for these bundles. In addition, discharged fuel bundles having been exposed to various combinations of normal BWR water chemistry and hydrogen water chemistry cycles are also included in the test matrix. A combination of site and hot cell examinations will provide required information on corrosion and hydriding characteristics of Zircaloy-2 and -4 fuel bundle components, and on crud deposition characteristics of the reactor system. Components examined will be urania fuel rods, gadolinia-urania fuel rods, spacers, and water rods. The first poolside, nondestructive fuel inspection was succesfully accomplished in October 1984. Preliminary results for the fuel bundles having experienced either three cycles of normal water chemistry (bundle 3/0, Cycles 6 through 8), two cycles of normal water chemistry followed by one cycle of H2WC (bundle 2/1, Cycles 7 through 9), or one cycle of H2WC (LTA 0/1, Cycle 9) indicate that: (1) the visual appearance and the oxide thickness for the bundles 2/1 and 3 / 0 are not significantly different; (2) the visual appearance of LTA 0 / 1 is very good. Moreover, two water rods and seven spacers from a leading test assembly with a burnup of 7.5 G W d / M T were shipped to the Vallecitos Nuclear Center for hot cell examination. Initial hot vacuum extraction measurements for the hydrogen content of the materials likely to have experienced the highest hydrogen pickup gave values less than, or equal to, 33 ppm hydrogen. Archive samples contained between 12 and 20 ppm hydrogen. This result indicates that hydrogen pickup during the first cycle of operation in hydrogen water chemistry was quite small. Based on the data at hand, no adverse effects due to hydrogen injection are anticipated, and it has been recommended that use of H2WC be continued through the next reactor cycle. Further analysis of the data obtained at poolside is in progress, hot cell examination of bundle hardware is continuing; fuel components will be destructively examined in 1985.
511
2.6. Future plans Based on experience to date, several areas for improvement of the hydrogen water chemistry system at Dresden-2 have been identified. To reduce costs, the plant will convert to liquid hydrogen storage and supply for the next cycle. Also, better system control will be achieved through control room panel modifications to bring the oxygen and hydrogen supply monitors to a single observation point so that any changes to the hydrogen injection rate can be rapidly reflected by changes to the oxygen supply to the SJAE. In addition, all H2WC data will be recorded by the process computer so that an integrated printout of plant operations including hydrogen water chemistry parameters can be produced. Finally, another development which we believe will be important in verifying the mitigating effects of H 2WC on IGSCC is the installation of a continuous UT monitor on the 28" flawed pipe mentioned earlier. Commonwealth Edison is installing a high-temperature UT sensor package developed by Amdata Systems under EPRI funding during this outage. 3. Conclusions To date, the results of the combined laboratory development program and in-plant verification of hydrogen water chemistry are very encouraging and indicate that an alternative BWR water chemistry which will mitigate IGSCC in stainless steel recirculation piping during power operation can be specified. Experience shows that the maintenance of _< 20 ppb oxygen by continuous addition of hydrogen and < 0.2 /~mho/cm conductivity, through efficient operation of condensate and reactor water demineralizers, can be achieved in a commercial power plant for > 90% of the time at power with minimum system impact. The off-gas fires are not attributable to H2WC but, rather, may be aggravated by the unique off-gas piping system at Dresden-2 and the engineering solution appears to be straightforward. Radiation level increases due to H 2WC, although not a concern at Dresden-2, could pose a more significant problem at other plants. This must be evaluated, and recommendations on shielding requirements, etc., provided to other BWR owners. No serious detrimental effects of H2WC on the performance of BWR structural materials and fuel assemblies have been identified in laboratory and in-reactor testing. The verification project will continue at Dresden-2 for Cycle 10 to confirm that H2WC is generally a less aggressive environment than normal oxygenated BWR water.
512
J.T. A. Roberts et al. / B W R pipe crack control
Finally, the estimated cost for the installation of H2WC is about $3 million, with subsequent operation costing from $300K to $500K per year. The operating costs are strongly influenced by the cost of hydrogen. It appears that for many plants the use of liquid hydrogen as a delivery/storage system would be optimum. Installation of a hydrogen recovery/recycle system may also be cost-effective at some plants. In some cases lower hydrogen costs may be achievable through the use of advanced electrolytic generation on site. These options are covered in comprehensive detail in a recently published EPRI report [8].
References [1] J.C. Danko, Recent observations of cracks in large-diameter
[2] [3] [4]
Acknowledgements The authors gratefully acknowledge the contributions made by T. Wojnulewicz and J. Almer of the Dresden-2 Reactor; Commonwealth Edison Project Managers, E. Zebus and E. Rowley; General Electric Project Managers, R. Cowan, B. Gordon and R. Adamson and L. Anstine of APT, without whom this project would not be possible.
[5]
[6]
[7]
[8]
BWR piping: analysis and remedial actions, Proc. 1st Int. Symp. on Environmental Degradation of Material in Nuclear Power Systems: Water Reactors, Myrtle Beach, So. Car., (August 22-25, 1983) National Association of Corrosion Engineers. Seminar on Countermeasures for BWR Pipe Cracking, EPRI WS-79-174, Vols. 1-4 (May 1980). EPRI Seminar on Countermeasures for BWR Pipe Cracking (November 1983) Palo Alto, Calif. R.L. Jones, A. Machiels, M. Naughton and J.T.A. Roberts, Controlling stress corrosion cracking in BWR piping by water chemistry modification, in: Corrosion effects, Events and Control in the Nuclear Power Industry, Corrosion '84, NACE, New Orleans (April 1984). E.L. Burley, Oxygen suppression in boiling water reactors Phase 2, Final Report, DOE/ET/34203-47, NEDC-23856-7 (October 1982). J.T.A. Roberts, R.L. Jones and M. Naughton, Mitigation of BWR Pipe Cracking Through Water Chemistry Changes, American Power Conference, Chicago (April 24-26, 1984). B.M. Gordon et al., Hydrogen water chemistry for BWRs, EPRI RP1930-1 lnterim Report (January 1, 1981-April 1, 1983). D.L. Roberts et al., Options for hydrogen use in BWRs, EPRI NP-3282 (November 1983).
B W R PIPE C R A C K C O N T R O L ON DRESDEN-2 PROGRAM
505
USING HYDROGEN WATER CHEMISTRY:
J.T. A d r i a n R O B E R T S , R o b i n L. J O N E S , M i c h a e l N A U G H T O N
STATUS REPORT
a n d A l b e r t J. M A C H I E L S
Nuclear Power Division, Electric Power Research Institute, Palo Alto, California, USA
Received 22 February 1985
One of the proposed remedies for intergranular stress corrosion cracking of stainless steel piping in BWRs is an alternative water chemistry called hydrogen water chemistry (H2WC) that involves suppression of reactor water dissolved oxygen to _< 20 ppb via hydrogen injection to the feedwater in conjunction with control of conductivity to < 0.3 /tmho/cm. A long-term verification program, over two or three 18 month fuel cycles, was started at Commonwealth Edison's Dresden-2 reactor in April 1983 (Cycle 9). This paper describes the results of the water chemistry changes, structural material and fuel evaluations, and plant radiation level changes during Cycle 9, which ended in October 1984. To date the results of the verification program are very encouraging. They indicate that the alternative water chemistry, based on hydrogen additions to the feedwater to suppress oxygen and low conductivity, can be maintained in a large operating BWR, and that it does mitigate IGSCC in stainless steel recirculation piping. Monitoring of fuel and plant materials will continue in Dresden-2 at least through Cycle 10 to confirm the absence of any unusual side effects of this remedy for IGSCC.
!. Introduction
2. Basis for an alternative BWR water chemistry
A major environment-related materials performance problem encountered in the reactor coolant system of BWRs has been intergranular stress corrosion cracking (IGSCC) of sensitized austenitic stainless steel. I G S C C of sensitized material adjacent to welds in type-304 and type-316 stainless steel piping systems has been responsible for more than 400 cases of pipe cracking over the last ten years [1]. Athough these cracks are not thought to pose a major safety concern, inspections and repairs associated with pipe cracking have proved costly to the utilities and substantial R & D programs have been undertaken to understand the I G S C C phenomenon and develop remedial measures [1-3]. Much of the early remedy-development work focused on alternative materials or local stress reduction, but recently, the effects of water chemistry parameters on the I G S C C process have received increasing attention in work funded by EPRI, the B W R Owner's Group, and the U S N R C . A complete understanding of the interrelationship between B W R water chemistry variables and I G S C C of sensitized stainless steels has not yet emerged but some important features have been identified.
Water itself is relatively innocuous towards stainless steels even at the relatively high temperatures involved in its use as a working fluid in electric generation plants. However, impurities that enter the water can make it an aggressive environment towards sensitized stainless steel. Oxygen is generated as a result of radiation in the core of a BWR during normal operation and this leads to a steady level of - 2 0 0 ppb dissolved oxygen in B W R water. Ionic impurities (salts) enter the water from sources such as makeup water and condenser leaks. These are controlled by condensate and reactor water cleanup systems which, however, can be costly to operate. The objective of the overall EPRI program on impurities in BWR water is to determine the effects of impurities on I G S C C so that guidelines for water quality to minimize pipe cracking can be formulated on a quantitative basis. Impurities that exacerbate pipe cracking fall into two classes depending on their mode of action. The oxidizers, represented by oxygen dissolved in the water, increase the chemical driving force for the corrosion reactions; and, because the reactions are electrochemical in nature, ionic impurities, such as chloride ions, increase their rates by increasing the electrical conductivity of the
0 0 2 9 - 5 4 9 3 / 8 5 / $ 0 3 . 3 0 © E l s e v i e r S c i e n c e P u b l i s h e r s B.V. (North-Holland Physics Publishing Division)
506
J.T.A. Roberts et al. / B W R pipe crack control
I Oxygenco~letllOt J corrosionpolent]al
Bw.r,no,LI
aWRH2WE;Z range L
iOr Impurity concenttal~on J conducl~'vTly Fig. 1. Schematic summary of the results of laboratory studies of effect of impurities on SCC of stainless steel.
water. Fig. 1 represents current thinking about the synergism between the effects of the two classes of impurities on IGSCC [4]. It is apparent that the contents of both classes of impurities must simultaneously be minimized to minimize the likelihood of IGSCC. This concept underlies an alternative BWR water chemistry, known as hydrogen water chemistry (H2WC), which entails oxygen suppression via hydrogen addition to the feedwater in combination with conductivity control via optimized plant operational procedures. The remainder of this paper presents a status report on the development and in-plant verification of H2WC. 2.1. B a c k g r o u n d
In 1977, the US Department of Energy sponsored a program designed to identify additives suitable for oxygen suppression in the BWR, to determine their possible impact on various plant systems, and to select the best additive and test bed. The additive selected was hydrogen, and the demonstration plant selected was Commonwealth Edison Company's Dresden-2 reactor in Morris, Illinois. In 1982, General Electric, DOE, Commonwealth Edison, and EPRI participated in a one-month hydrogen addition feasibility demonstration in Dresden-2 [5].
This one-month demonstration included in-reactor stress corrosion cracking (SCC) tests on furnace-sensitized stainless steel and low alloy steel specimens and electrochemical potential measurements on various BWR structural materials. The results showed that the coolant oxygen levels in the GE designed BWR can be suppressed to below 20 ppb during power operation by adding practical amounts of hydrogen to the feedwater, on the order of 1.5 ppm. Reactor water having an oxygen content of 20 ppb and a conductivity of 0.30 p, m h o / c m was shown to be insufficiently aggressive to promote either IGSCC in sensitized stainless steel or transgranular stress corrosion cracking of pressure vessel steel in short-term, slow strain-rate tests. At low oxygen levels, the quantity of N-16 in the steam increased as expected, but the consequent four to fivefold increase in steam line gamma radiation was not found to be a major problem in Dresden-2. No other significant adverse effects of H2WC were identified in the relatively short-term demonstration. The next step in the H2WC program, a long-term verification over two or three 18 month fuel cycles started at Dresden-2 in April of 1983 (Cycle 9) under EPRI funding. It involves extensive monitoring of fuel and core materials behavior and continued evaluation of plant structural materials behavior in longer term tests.
J.T. A. Roberts et al. / B WR pipe crack control
507
TURBINE
REACTOR PRESSURE VESSEL DEMINERALIZER
®
FEEDWATER SYSTEM HYDROGEN INJECTION PURIFICATION SYSTEM
POTENTIAL MEASUREMENTS
CONSTANT EXTENSION RATE TEST
Q
CONDENSATEDEMINERALIZER EFFLUENT (CDE)
Q
FINALFEEDWATER (FFW)
Q
COREEXIT WATER (RCX) (PLANNED BUT NEVER INSTALLEDI
Q
RECIRCULATIONLOOP (RRC)
Q
CLEANUPINLET (RCI}
Q
CLEANUPOUTLET (RCE}
Q
MAIN STEAM (MST)
t
Fig. 2. Hydrogen addition flowsheet at Dresden-2.
The water chemistry goals are < 20 ppb oxygen and a reactor water conductivity of < 0 . 3 / ~ m h o / c m . The following paragraphs summarize the results to date on this program, problems encountered and their resolution, and system upgrade activities currently under way. This is an update of an earlier paper on this program presented at the 1984 American Power Conference [6].
2.2. Plant operations
The hydrogen addition flowsheet is shown in fig. 2. High purity (99.999%) hydrogen is purchased as gas which is delivered in tank trucks carrying 115000 or 50000 scf at 2400 psi. These tanks supply hydrogen at about 200 psi through a ~" carbon steel line into a flow control panel in the condensate pump room.
50g
J. 7~ A. Roberts et al. / B WR pipe crack control
This panel varies the hydrogen flow in proportion to feedwater flow to give a constant but adjustable hydrogen concentration in the reactor feedwater. Maintenance of feedwater hydrogen at 1.5 ppm results in dissolved oxygen concentrations of 20 ppb or less throughout the primary coolant system. Oxygen from a liquid oxygen storage trailer provides 60 psi oxygen gas to add into the off-gas system at the first stage steam jet air ejector (SJAE). Oxygen flow is controlled manually based on the reading of oxygen meters sampling the recombiner outlet gas. A steadystate oxygen flow is set to maintain the oxygen concentration at the recombiner outlet between 8 and 12 vol%. When hydrogen injection rate increases are to be made, oxygen flow is increased first to be sure there is always excess oxygen. The hydrogen addition system is not operated continuously. It is shut off for a variety of reasons, including maintenance activities on the hydrogen addition system itself, maintenance in other areas of the plant where there are high N-16 radiation levels due to hydrogen addition, and during the extinguishing of fires in the off-gas treatment system. In addition, during reactor startups and shutdowns, hydrogen addition is not used below 220 MW (e) (i.e., 25% power). When the injection system is operating, it significantly reduces the oxygen concentration in the reactor water. Overall, the injection system availability has been about 87%, During 1984 the injection system was available for about 90% of the time and the oxygen was controlled below 20 ppb for 86% of the time (fig. 3). It is worthwhile to investigate the nature of the time the reactor water was greater than 20 ppb oxygen because of the existence of a corrosion potential " m e m o r y effect" discussed later in the paper. As a result of this effect, the corrosion potential stays below the critical
99.999 99.990 z ~ 99.900 ~>. 99.000
--
~ 90.000
--
o
60.000 40.000 20.000 10.000
2.3. Plant water chemistry
Control of ionic impurities has been particularly good in Dresden-2 for the first 18 month cycle of H2WC. Conductivity has been maintained at less than 0.2 b t m h o / c m for 98% of the time (fig. 4), and pH has been maintained in the 7 to 7.5 range. Ion chromatography studies of the coolant at Dresden-2 have shown that under continuous operation conditions, no more than 30 ppb carbonate, 5 ppb chloride and 5 ppb sulfate were present. The balance of the conductivity is made
99.9999 99.9990
4
2 ~0 o r,, "0
-I
1.00o ~__ o.100
g O0
O.OLO --=l IIIl~lll IIIII1[1[ IIIIIIIII o.001 1.0
10.0 Oxygen (ppb)
100.0
---3
[
II IIII
-4
1000.0
Fig. 3. Probability plot of all oxygen data during Cycle 9.
J
99.9900 2 99.9000
3
z z -SS
o.1
value for IGSCC for at least 10 h after shutdown of the hydrogen injection system. The statistical data base shows that greater than 70% of the cumulative time the reactor water was above 20 ppb for periods of less than 10 h in any given 24 h period. This implies that over 93% of the total cumulative operating time since startup of Dresden-2 in April 1983 has been in the corrosion potential region that is " i m m u n e to I G S C C " for conductivity levels _< 0.3 / z m h o / c m . The major problem causing shutdown of the H2WC system has been off-gas fires. An extensive evaluation of the reasons for the off-gas fires revealed they were not directly related to H2WC but, rather, to the design of the off-gas piping system. The probable ignition source is migrated catalyst in the off-gas trains, and flow or pressure changes cause ignition by fluffing the catalyst. Since the fires are located in the SJAE, after the condenser, the recommendations for eliminating the fires include elimination of a bypass line in the A B off-gas trains (which is unique to the Dresden and Quad Cities Units), followed by poisoning or removal of the migrated catalyst in the remaining lines. Also, a change in the oxygen addition point to the steam diluted downstream region is being considered.
-- 4
99.0000 z ~ 90.0000o 60.0000 ---
- - 2 ._o g
o_ 40.0000 20.0O00-
m.oooo 5 1.oooo E= 0.1000 _ _ o
o.0100
0.0010 0,0001
0.01
-3 03
I [ II
IIIITT -4 0.10
1.00
Conductivity (microsiemenslcm)
Fig. 4. Probability plot of all conductivity data during Cycle 9.
J.T.A. Roberts et al. / B WR pipe crack control 5O
509
~
0.05
elements that were analysed all show similar patterns, with no element showing an upward trend with time. The concentration of the other insoluble metals is generally less that 0.1 ppb. In the reactor water, the dominant activation product is 6°Co, but no clear trend, either upward or downward, was observed. Values between 0.1 and 0.2 i.tCi/I are consistent with data from other BWRs. Of the soluble species that were monitored in the feedwater, cobalt is the element of greatest concern because of its activation to 6°Co in the reactor core and accompanying potential for radiation buildup. Fig. 6 shows the concentration of soluble cobalt in the feedwater as a function of time. The downward and stable trend of cobalt is readily apparent. This trend pattern is also observed for the other elements that were analyzed. In contrast, there was no definitive upward or downward trend in the cobalt data in the reactor water; average values of 40 ppt for soluble Co agree with measurements from other BWRs. Because of the operational practice changes to the condensate treatment system that were implemented during Cycle 9, it cannot be explicitly concluded that these trends are a result of hydrogen water chemistry. Nonetheless, there appear to be no adverse consequences of hydrogen water chemistry for soluble species in either the reactor water or feedwater. Summarizing the corrosion product data, we have not seen any indication that hydrogen water chemistry has caused any detrimental change in the soluble and insoluble corrosion product transport in the feedwater or the reactor water. In the feedwater, insoluble spikes are common to all BWRs, and clearly will propagate to the reactor water. The downward trend in feedwater solubles we feel is the result of a gradual implementation of good operational practices centered around the management of the condensate treatment system. The concentrations of the impurities in the feedwater and reactor water are consistent with data from other deep bed plants.
8
0.04
2.4. Radiation levels
0.03
An important concern associated with H2WC is the impact on radiation levels in the turbine building, plant environs and off-site. The earlier 30 day test had indicated an increase on the order of a factor of 4 to 5 in N-16 in the steam. Detailed measurements have now been made in and around the plant, indicating that although the impact of H2WC is significant and measurable in the vicinity of the turbine building, it diminishes rapidly away from the plant and is negligible
O.
40 30
e-
20 C)
o
10 0 0
40
80 120 160 200 Days Since July 15, 1983
240
280
Fig. 5. Trend in feedwater insoluble iron during Cycle 9.
up of hydronium and hydroxyl ions. Thus, the species making up the conductivity at Dresden-2 are less aggressive than the sodium sulfate used to assess conductivity effects in the laboratory studies [7]. In addition to monitoring conductivity, pH, and oxygen, the H2WC verification program also includes the measurement of various metals and isotopes in the feedwater and reactor water. Soluble and insoluble corrosion products have been collected continuously, with samples changed at roughly three-day intervals. Each sample fraction has been analyzed at Vallecitos Nuclear Center for Fe, Cu, Ni, Co, Zn, Cr, and Mn. The dominant impurity in the Dresden-2 feedwater is insoluble iron. Fig. 5 shows the concentration of insoluble Fe as a function of time for a period since July 15, 1983. The 20 ppb spikes are the result of several condensate demineralizer changeouts over a short interval; the 40 ppb spike encompasses an orderly shutdown for pipe crack inspection and the restartup of the reactor. No long-term adverse consequences of hydrogen addition are evident. The time-base plots for other
0.06
8
0
0.02 0.01
I
0
0
40
80 120 160 200 Days Since July 15, 1983
I 240
Fig. 6. Trend in feedwater soluble cobalt during Cycle 9.
280
510
J.T.A. Roberts et aL / B W R pipe crack control
0
280
-100-
~
-2o0-
H2 - 0 n ,Gsoc
-
-300 -400
--
-
240 200 160
-,
120
"(~l -500 --1
Zk. 02 (ppb) rn Gold
-600 : ~
O Type 304 SS
-700 - ~ 2 ° ' 1 1200 1600 2-5-84
[
80 40
1 I 2000 Time
o 2400 400 600 2-6-84
Fig. 8. Example of electrochemical potential memory effect during hydrogen water chemistry.
Fig, 7. Net contribution of hydrogen water chemistry to Dresden-2 environs dose rate (#R/h).
off-site. Fig. 7 indicates the contribution of H2WC to the total dose rate on a series of contours drawn from the turbine building. Addition of Hydrogen Water Chemistry at Dresden-2 has been found to add about 10 man rem to the 2000 rein currently incurred each year. However, in terms of exposure for IGSCC-related repairs, H2WC is A L A R A effective. For example, pipe replacement programs typically lead to exposure of 1500 to 2500 man rem. It must be noted, that the Dresden-2 plant layout and site location are favorable for reduced radiological effects of the H2WC system. Other plants might not be so favorably constructed or situated and implementation of H2WC might require costly shielding or changes to operation and maintenance procedures.
2.5. Fuel and plant materials characterization Slow strain rate, crack growth, and constant load tests on a variety of plant structural materials are being undertaken in-plant to supplement the extensive laboratory data already available [7]. These in-reactor tests are conducted in special autoclaves using reactor water fed from a header in the recirculation system. The results of stress corrosion and electrochemical potential studies provide strong evidence that stainless steel stress corrosion cracking activity at Dresden-2 has been arrested. A series of in situ stress corrosion tests were conducted on furnace-sensitized stainless steel with the tests ongoing
over about 2000 h. No cracking was observed in either smooth or I G S C C precracked specimens tested while the plant was operated with reactor water at 20 ppb O 2 or less and reactor water conductivity at 0 . 3 / ~ m h o / c m or less. Electrochemical potential measurements on stainless steel made at the plant for over 3000 h revealed potentials well below - 3 5 0 mV (SHE) over 95% of the time. This potential is considered to be a threshold for I G S C C for the conductivity values listed above. Coolant oxygen content excursions up to 200 ppb for limited periods of time for up to about 15 h during the constant extension rate tests did not result in IGSCC. Electrochemical potentials during these short-term excursions did not increase to the values observed under normal operation at 200 pbb oxygen indicating the presence of a memory effect (fig. 8). Perhaps the most revealing test result is the constant extension rate test conducted on a furnace-sensitized and laboratory I G S C C precracked specimen that had seven hours of in-plant test time with 200 ppb oxygen. The balance of the in-plant test time was within the H2WC regime. No I G S C C in addition to that in the precracked phase was noted. This result is consistent with in-service inspection observations. The recirculation piping was inspected in November 1983 and December 1984 in response to an N R C directive since I G S C C flaws had been detected earlier in a large-diameter line. The inspections revealed no further growth of the existing flaws. Additional in-plant and laboratory studies have shown that stress corrosion a n d / o r corrosion fatigue of the other major structural alloys including Inconel 600, carbon steel, and low alloy steel are effectively eliminated by application of HzWC. No effects were found for martensitic stainless steel. Some increase in carbon steel general corrosion rates can be expected,
J.T.A. Roberts et al. / B W R pipe crack control
but resultant corrosion is well within design tolerances [7]. The concern over the long-term behavior of fuel, particularly with respect to hydrogen pickup and hydriding, prompted a plan for extensive examination of fuel rods and bundle hardware. For this purpose, four leading test assemblies (LTAs) of carefully characterized fuel components have been inserted at the beginning of the first cycle of hydrogen addition. Zircaloys with a known, precharacterized range of corrosion behavior representative of the range usually found with cladding batches used in BWR fuel fabrication have been selected for these bundles. In addition, discharged fuel bundles having been exposed to various combinations of normal BWR water chemistry and hydrogen water chemistry cycles are also included in the test matrix. A combination of site and hot cell examinations will provide required information on corrosion and hydriding characteristics of Zircaloy-2 and -4 fuel bundle components, and on crud deposition characteristics of the reactor system. Components examined will be urania fuel rods, gadolinia-urania fuel rods, spacers, and water rods. The first poolside, nondestructive fuel inspection was succesfully accomplished in October 1984. Preliminary results for the fuel bundles having experienced either three cycles of normal water chemistry (bundle 3/0, Cycles 6 through 8), two cycles of normal water chemistry followed by one cycle of H2WC (bundle 2/1, Cycles 7 through 9), or one cycle of H2WC (LTA 0/1, Cycle 9) indicate that: (1) the visual appearance and the oxide thickness for the bundles 2/1 and 3 / 0 are not significantly different; (2) the visual appearance of LTA 0 / 1 is very good. Moreover, two water rods and seven spacers from a leading test assembly with a burnup of 7.5 G W d / M T were shipped to the Vallecitos Nuclear Center for hot cell examination. Initial hot vacuum extraction measurements for the hydrogen content of the materials likely to have experienced the highest hydrogen pickup gave values less than, or equal to, 33 ppm hydrogen. Archive samples contained between 12 and 20 ppm hydrogen. This result indicates that hydrogen pickup during the first cycle of operation in hydrogen water chemistry was quite small. Based on the data at hand, no adverse effects due to hydrogen injection are anticipated, and it has been recommended that use of H2WC be continued through the next reactor cycle. Further analysis of the data obtained at poolside is in progress, hot cell examination of bundle hardware is continuing; fuel components will be destructively examined in 1985.
511
2.6. Future plans Based on experience to date, several areas for improvement of the hydrogen water chemistry system at Dresden-2 have been identified. To reduce costs, the plant will convert to liquid hydrogen storage and supply for the next cycle. Also, better system control will be achieved through control room panel modifications to bring the oxygen and hydrogen supply monitors to a single observation point so that any changes to the hydrogen injection rate can be rapidly reflected by changes to the oxygen supply to the SJAE. In addition, all H2WC data will be recorded by the process computer so that an integrated printout of plant operations including hydrogen water chemistry parameters can be produced. Finally, another development which we believe will be important in verifying the mitigating effects of H 2WC on IGSCC is the installation of a continuous UT monitor on the 28" flawed pipe mentioned earlier. Commonwealth Edison is installing a high-temperature UT sensor package developed by Amdata Systems under EPRI funding during this outage. 3. Conclusions To date, the results of the combined laboratory development program and in-plant verification of hydrogen water chemistry are very encouraging and indicate that an alternative BWR water chemistry which will mitigate IGSCC in stainless steel recirculation piping during power operation can be specified. Experience shows that the maintenance of _< 20 ppb oxygen by continuous addition of hydrogen and < 0.2 /~mho/cm conductivity, through efficient operation of condensate and reactor water demineralizers, can be achieved in a commercial power plant for > 90% of the time at power with minimum system impact. The off-gas fires are not attributable to H2WC but, rather, may be aggravated by the unique off-gas piping system at Dresden-2 and the engineering solution appears to be straightforward. Radiation level increases due to H 2WC, although not a concern at Dresden-2, could pose a more significant problem at other plants. This must be evaluated, and recommendations on shielding requirements, etc., provided to other BWR owners. No serious detrimental effects of H2WC on the performance of BWR structural materials and fuel assemblies have been identified in laboratory and in-reactor testing. The verification project will continue at Dresden-2 for Cycle 10 to confirm that H2WC is generally a less aggressive environment than normal oxygenated BWR water.
512
J.T. A. Roberts et al. / B W R pipe crack control
Finally, the estimated cost for the installation of H2WC is about $3 million, with subsequent operation costing from $300K to $500K per year. The operating costs are strongly influenced by the cost of hydrogen. It appears that for many plants the use of liquid hydrogen as a delivery/storage system would be optimum. Installation of a hydrogen recovery/recycle system may also be cost-effective at some plants. In some cases lower hydrogen costs may be achievable through the use of advanced electrolytic generation on site. These options are covered in comprehensive detail in a recently published EPRI report [8].
References [1] J.C. Danko, Recent observations of cracks in large-diameter
[2] [3] [4]
Acknowledgements The authors gratefully acknowledge the contributions made by T. Wojnulewicz and J. Almer of the Dresden-2 Reactor; Commonwealth Edison Project Managers, E. Zebus and E. Rowley; General Electric Project Managers, R. Cowan, B. Gordon and R. Adamson and L. Anstine of APT, without whom this project would not be possible.
[5]
[6]
[7]
[8]
BWR piping: analysis and remedial actions, Proc. 1st Int. Symp. on Environmental Degradation of Material in Nuclear Power Systems: Water Reactors, Myrtle Beach, So. Car., (August 22-25, 1983) National Association of Corrosion Engineers. Seminar on Countermeasures for BWR Pipe Cracking, EPRI WS-79-174, Vols. 1-4 (May 1980). EPRI Seminar on Countermeasures for BWR Pipe Cracking (November 1983) Palo Alto, Calif. R.L. Jones, A. Machiels, M. Naughton and J.T.A. Roberts, Controlling stress corrosion cracking in BWR piping by water chemistry modification, in: Corrosion effects, Events and Control in the Nuclear Power Industry, Corrosion '84, NACE, New Orleans (April 1984). E.L. Burley, Oxygen suppression in boiling water reactors Phase 2, Final Report, DOE/ET/34203-47, NEDC-23856-7 (October 1982). J.T.A. Roberts, R.L. Jones and M. Naughton, Mitigation of BWR Pipe Cracking Through Water Chemistry Changes, American Power Conference, Chicago (April 24-26, 1984). B.M. Gordon et al., Hydrogen water chemistry for BWRs, EPRI RP1930-1 lnterim Report (January 1, 1981-April 1, 1983). D.L. Roberts et al., Options for hydrogen use in BWRs, EPRI NP-3282 (November 1983).