IRRADIATION EMBRITTLEMENT MITIGATION

Size: px
Start display at page:

Download "IRRADIATION EMBRITTLEMENT MITIGATION"

Transcription

1 AMES Report No. 1 EUR EN IRRADIATION EMBRITTLEMENT MITIGATION Tapio Planman Reijo Pelli Kari Törrönen European Network on Ageing Materials Evaluation and Studies Espoo, September 1994 VTT Manufacturing Technology P.O. Box 1704, FIN VTT, Finland Tel , Telefax

2 ABSTRACT Neutron irradiation affects the material properties, and hence the structural integrity, of reactor pressure vessels (RPVs) in nuclear power plants. Mitigation of irradiation damage is one of the major issues within nuclear plant life management. An overview is given of proposed and utilised mitigation methods. Low-leakage loading schemes are commonly used in pressurised water reactors (PWRs) to mitigate consequences of RPV embrittlement. Dummy assemblies have been installed in VVER 440 -type and in some old western power plants where exceptionally fast embrittlement has been encountered. Shielding techniques for RPVs have been developed, but these are not in common use. Prestressing of the RPV has been proposed as a potential method for preventing RPV failures due to pressurised thermal shock (PTS) loading, but applicability of the method to nuclear pressure vessels has not yet been demonstrated. The large number of annealing treatments performed in VVER 440 -type reactors, and the intensive research and development work done on the methods and benefits of annealing treatments during the recent years, suggest that more applications can be expected in the near future also in western PWRs. The emergency core cooling systems have been modified in VVER 440 -type reactors in connection with other mitigation measures, and in some western PWRs. Efforts to extend the service life of RPVs further emphasise the role of plant specific surveillance programmes. The embrittlement management of PWRs should include evaluation of all realistic mitigation alternatives, and generation of additional material data whenever possible, before final decisions on life extension measures are made. The use of decision tools is recommended in dealing with the complex PTS issue. 1

3 PREFACE This report has been compiled at the VTT Manufacturing Technology as part of the state-of-the-art review on irradiation embrittlement, surveillance and mitigation methods carried out by the European Network for Ageing Materials Evaluation and Studies (AMES). The work is financed by the CEC DG XI through the contract "Centre d'etudes de Saclay Gif Sur Yvette Cedex, Ref B018750, M. Soulat". The Finnish Centre provided additional financing for Radiation and Nuclear Safety (STUK) and VTT. Associated with this subject, reports are to be compiled within AMES also on thermal annealing and on the scientific basis of Russian and European approaches for evaluating irradiation effects in reactor pressure vessels. The authors greatly acknowledge Acad. Myrddin Davies and Dr. Colin English for conducting a peer review for this report. 2

4 CONTENTS ABSTRACT...1 PREFACE INTRODUCTION FLUENCE RATE REDUCTION TECHNIQUES Core modification and/or reduction Description of methods Calculated cases for different cores Special fuel designs Shielding of the pressure vessel Factors affecting fluence rate reduction Secondary consequences of reduced fluence rate MODIFICATION OF EMERGENCY CORE COOLING AND OTHER SYSTEMS Pressurised thermal shocks Normal heatup and cooldown conditions Low-temperature overpressure transients High-temperature transients ANNEALING Methods Accomplished annealings Recovery and re-embrittlement of western RPV steels Recovery and re-embrittlement of VVER 440 RPVs OTHER MITIGATION METHODS Prestressing Warm prestressing Weld replacement Replacement of the RPV Power reduction APPLICATIONS AT SOME NUCLEAR POWER PLANTS EVALUATION OF MITIGATION METHODS IN PWRs Applicability and use of different methods Strategies and methods for managing irradiation embrittlement On research needs and experience Recommendations for utilities Research needs CONCLUSIONS...81 ACKNOWLEDGMENTS...82 REFERENCES

5 1 INTRODUCTION Reactor pressure vessel (RPV) integrity is assured by several arrangements including the setting of operational limits for normal and various transient conditions, surveillance of irradiation embrittlement of beltline materials, and in-service inspections and other maintenance procedures. The critical loading conditions for RPVs are typically associated with different postulated emergency core cooling, i.e. pressurised thermal shock (PTS) events, during which the pressure vessel is subjected both to thermal stresses and to those caused by repressurisation. The temperature-pressure limits (operating windows) set for heatup and cooldown stages may require operational restrictions due to the RPV embrittlement. Many investigations of RPV failure probability have been put forward since 1982, when surveillance results from some operating plants showed that embrittlement of a vessel material was faster than predicted. Of special concern were certain old pressurised water reactors (PWRs), where predictions made by U.S. Nuclear Regulatory Commission (NRC) showed that failure during a PTS transient could be possible after only a few years of operation (Smock, 1982). Recently, irradiation embrittlement mitigation has come under closer scrutiny as the operating licenses of the oldest plants are due to expire within the next few decades and plant life extension needs to be considered. If replacement of RPVs is excluded, remedial measures such as thermal annealing of the RPV may be needed for life extension in many old PWRs, besides possible preventive measures already implemented. At any rate, the most important measures are those assuring RPV integrity during operating and possible accident conditions. Irradiation embrittlement is normally the most severe degradation mechanism in RPVs, although other mechanisms also exist (Gerard, 1990). RPV embrittlement is caused mainly by the fast neutron flux from the core. Embrittlement depends also on the impurity contents of a RPV material. Different guides have been developed to evaluate the fluence dependence of materials with certain impurity contents. Federal Regulation 10CFR50.61 provides rules for evaluating the nil-ductility transition temperature (RT NDT ) and upper shelf energy for different steels and fluences (USNRC, 1988 and Shah et al., 1989). In NRC Regulatory Guide 1.99 Revision 2, the transition temperature (at the CVN 41 J energy level) is presented as a function of Cu and Ni contents of the steel and the neutron fluence. The guide differentiates between base and weld materials. The upper shelf energy, which is used to characterise RPV materials in operating conditions, is calculated as a function of the Cu content and fluence. Guide 1.99 Rev. 2 has been compiled on the basis of surveillance test results from commercial U.S. power reactors, and has been found to give conservative estimates (Mager, 1993). It is noteworthy that this guide excludes the effect of P, which is also a detrimental element. Predictive equations based upon irradiation data from test reactors and surveillance capsules have been provided also by Framatome (Mager, 4

6 1993). However, it is recommended that RPV embrittlement should be followed by plant-specific surveillance programmes. Measures to mitigate irradiation embrittlement and reduce failure probability of the RPV have typically been directed towards following factors: 1) the material, 2) the environment, and 3) the stress state in the most severe loading situations and the probability of such events. 1. In new RPVs the steel composition has been optimised (Leitz & Koban, 1989), meaning that the contents of Cu, P and some other impurities have been minimised. A high Ni content may also enhance irradiation embrittlement if impurity contents (P, Cu) are high enough. Thermal annealing has been used for old RPVs to recover irradiation defects and mechanical properties. 2. Irradiation embrittlement can be mitigated efficiently in the early stages of RPV service life by reducing the neutron fluence (hence the fluence rate) to the pressure vessel, without the need to reduce core power. Low-leakage fuel management, at least in some form, is already applied in most PWRs, either for economic reasons or to mitigate irradiation embrittlement (Bagnal et al., 1984). Irradiation temperature and fluence rate also affect the embrittlement rate. 3. Stress concentrations in the RPV during a postulated PTS event, i.e. the severity of PTS, should be minimised to reduce the RPV failure probability. Thermal stresses can be decreased by raising the emergency core cooling water temperature and/or increasing mixing. Prestressing of the RPV, which has been suggested as a method for preventing PTS failures (detailed description given later), would provide one way to reduce stresses. The probability of severe transient conditions should also be minimised. As a consequence of advanced steel and RPV manufacturing techniques, irradiation embrittlement of the RPV is not expected to limit the service life of modern PWRs (Leitz & Koban, 1989). These have typically only a few circumferential welds in the RPV (welds are usually most susceptible to embrittlement due to their chemical composition), low impurity contents in welds and base materials, and a low neutron fluence rate at the RPV due to the large water gap between the core and the RPV wall (Figs. 1 and 2). The effect of some of these factors on the 41 J transition temperature shift is shown in Fig. 3. 5

7 Fig. 1. Development of RPV dimensions and core configurations (KWU) (Leitz & Koban, 1989). Fig. 2. Development of VVER-type RPV dimensions and core configurations (Štepánek & Šaroch, 1983 and Dragunov & Tyulpin, 1993). 6

8 Fig. 3. Effect of fluence reduction and material improvement on the transition temperature shift (Leitz & Koban, 1989). Some old PWR vessels (built in ) are particularly susceptible to embrittlement (Leitz & Koban, 1989). Typical reasons are the high design end-oflife (EoL) fluence of the RPV, an unfavourable steel composition, or welds (circumferential and/or longitudinal) located in the beltline area. In older RPVs, high neutron fluence peaks often exist due to the typically small size of the core (low symmetry) and the relatively large size of fuel elements. Excessive embrittlement has also been caused by weld materials with high impurity contents. In U.S. reactors, irradiation problems are associated both with longitudinal welds of the RPVs manufactured from plates instead of ring forging, and with high Cu content of the weld materials (Fig. 4). In West-European power plants, the impure weld material is often given as the major reason for fast embrittlement. In many western RPVs, the embrittlement rate is also enhanced by a high Ni content. In some old PWRs, embrittlement mitigation measures have had to be implemented to achieve the original design service life of the RPV (Franklin & Marston, 1983). 7

9 Fig. 4. Critical weld locations in a typical U.S. PWR (Bagnal et al., 1984). High embrittlement rates of the VVER 440/213 and 440/230 -type RPVs have been attributed to high P and Cu contents of the welds and a high fluence rate at the RPV. In these RPVs, high Cu contents originate either from Cu plated electrodes used for welding or the residual Cu of the filler material, or both (IAEA-TECDOC-659, 1992). However, P is possibly a more important element due to its relatively high contents. As the most critical site in VVER 440 RPVs has been considered the circumferential weld locating in the beltline area (Fig. 2), but sites prone to significant degradation have been assessed to be also outlet/inlet and instrumentation nozzles and flange closure studs, subjected both to irradiation and mechanical and thermal loads (IAEA-TECDOC-659, 1992). Due to the exceptional design basis of the VVER 440 RPV, i.e. the requirement set for transporting of vessels, the distance between the core and the RPV wall had to be made very small (Fig. 5), resulting in a high fluence rate at the RPV. As a consequence, only a few VVER 440 plants are operated at full core (see Table 9, p. 66). On the other hand, the RPVs have been welded from ring forgings (of Cr-Mo-V steel) without longitudinal welds, which simplifies the annealing treatment. The Ni contents of the RPV steels are also relatively low. Uncertainties about the material properties and chemical composition of the beltline weld are regarded as a concern in some cladded VVER 440 RPVs where the weld has been covered by protective surface layers with different compositions (WWER-SC-081, Rev. July 1994). Samples taken near the (outer) surface of such a RPV weld are not always representative of the inner weld material. 8

10 Fig. 5. Cross section of the VVER 440 core (reduced core) (Bärs & Serén, 1993). When differences in the chemical composition (P, Cu, Ni) of steels and in irradiation conditions (fluence, fluence rate, neutron spectrum and irradiation temperature) are taken into account, irradiation responses of western and VVERtype RPV steels have in general been rather similar. In principle, all mitigation methods applicable for western RPVs are effective also for VVER 440 -type RPVs and inversely. Differences in embrittlement rates naturally affect the timing and required effect of mitigation measures. Measures may also have different effects on steels of various types. The aim of irradiation embrittlement management is to find and implement measures which are necessary, either to ensure that embrittlement of the RPV will not reduce the design service life of the power plant, or to make plant life extension possible. Extreme loading conditions for RPVs are typically associated with different transients, which fall into the following groups (Throm, 1989): 1. A PTS is generally regarded as the most critical, though highly improbable, loading situation for RPVs. During a PTS the RPV is first cooled rapidly by the emergency cooling water and then pressurised, which together create high tensile stresses in the embrittled and cooled inside wall of the RPV. Both the likelihood and severity of possible PTS cooling events are normally reduced by several protection systems. 9

11 2. Normal heatup and cooldown stages, during which temperature and pressure changes are restricted by the operating windows specified for the RPV. The size of operating windows and hence the operability of the plant is affected by the condition of the RPV beltline region. 3. Low-temperature overpressure transients, during which the temperature-pressure limits are exceeded temporarily. The likelihood of these events is reduced by various protection systems, which are discussed in greater detail later. An overpressure transient may induce crack growth in the embrittled inside wall of the RPV. 4. High temperature transients may become critical for RPVs where the upper shelf toughness of the material has been significantly reduced by neutron irradiation. In one accident scenario, thermal transients are created in the RPV due to cooling of the outside surface when cold water floods into the containment and comes into contact with the RPV (Laaksonen, 1994). Irradiation embrittlement management should always be considered as part of the life management of the whole plant. This means that the necessity, costs, and optimum timing of various embrittlement mitigation methods will also be affected by the degradation and operation of other components such as steam generators. Some factors incorporated in irradiation embrittlement management are given in Fig. 6. The significance of uncertainties associated with determination of material properties and embrittlement rates, such as the chemical composition, irradiation dose, and conservatism of Charpy-V impact tests, is emphasised but a detailed study on this subject is outside the scope of this work. Approaches used for integrity analyses of RPVs in different countries are discussed by Griesbach (1993). In Finland, safety of RPVs is evaluated from the failure probabilities in certain postulated accident situations using only measured material properties, e.g. the elastic-plastic K Jc. The measures already in use or proposed for extension of RPV technical service life consist of operations reducing the fluence rate at the RPV, recovery annealing of the RPV material, modification of the emergency core cooling and related systems, as well as other (proposed) operations. The use and applicability of these measures are reviewed in this report. 10

12 11

13 2 FLUENCE RATE REDUCTION TECHNIQUES Two principles are available for reducing fluence accumulation in the RPV. The core, i.e. the irradiation source, can be modified or reduced to give a lower fluence rate. Another way is to place irradiation shields or reflectors between the core and the RPV. 2.1 CORE MODIFICATION AND/OR REDUCTION Description of methods For any core configuration there are generally numerous possible loading schemes available. In general, to maximise core power the radial power distribution should be as even as possible. This can be achieved by following the OUT-IN scheme, where the fresh and most highly enriched fuel is placed for its first cycle on the core periphery and the exposed fuel in the interior (Bagnal et al., 1984). These procedures increase fission power density on the core periphery, where it is reduced by neutron leakage. Unfortunately, all actions to increase fission power density on the core periphery increase neutron leakage from the core and hence the fluence rate at the RPV. The OUT-IN scheme was previously the standard loading scheme in PWRs. The disadvantages were poor neutron economy and fast embrittlement of RPVs (Franklin & Marston, 1983 and Bagnal et al., 1984). However, in some instances low-leakage fuel management has been followed for reasons of fuel cycle economics (Anderson & Whitmarsh, 1984). The terms "low-leakage" and "low-fluence (or fluence rate) loading schemes" are used in the literature in different contexts. Low-leakage schemes are applied typically to minimise the overall neutron leakage from the core. The main objective of low-fluence schemes is to modify the circumferential neutron fluence rate distribution in such a way that the fluence rate at the critical locations of the RPV is reduced. The critical locations, i.e. those restricting the RPV service life, are generally welds due to their chemical composition and subsequent fast embrittlement, but also the base material may become life limiting. In RPVs without longitudinal welds, fluence reduction measures are applied for making circumferential fluence rate distribution more even, as in small cores this is typically very uneven with no modifying actions. The circumferential fluence distribution in the RPV of a 900 MWe reactor is shown in Fig

14 Fig. 7. Fluence distribution in a 900 MWe reactor pressure vessel (Gerard, 1990). The situation is more complicated for RPVs with longitudinal welds, as these are not usually (fortunately) located at circumferential fluence peak areas. This means that the fluence rate distribution should be modified so that significant reduction in the fluence rate is achieved at a certain fixed location instead of only flattening the fluence rate profile. If "low fluence rate assemblies" are to be implemented for achieving this kind of reduction, the fuel assemblies to be replaced are not necessarily those nearest the RPV wall. A consequence is generally that more fuel assemblies on the core periphery must be replaced by low fluence rate assemblies to achieve the desired effect on the fluence rate distribution (especially if the core symmetry is desired to be maintained). In such cases some other location also becomes sensitively critical. Hence, fluence rate reduction at a fixed circumferential location (e.g. longitudinal weld) is generally more difficult to achieve than the same reduction in a circumferential fluence rate peak (Stucker et al., 1983). The critical location of the RPV is often just on the inside surface of a longitudinal weld, if this exists. Besides the location of welds in the RPV, also the size and configuration of the core affect the applicability and possible benefits of different low-leakage (or lowfluence) schemes. The increase in power peaking due to a given low-leakage scheme is larger for small cores. In addition, in a large core more low-leakage loading schemes are available than in a small one. If there are only circumferential welds in the RPV, the fluence rate at the vessel can be reduced most efficiently by reducing power in the critical peripheral fuel assemblies, i.e. in those nearest the RPV wall. Roughly 85% of the fluence to the RPV is estimated to come from the core peripheral assemblies (Carew & Lois, 1991). The contribution of some adjacent assemblies to the fluence of two longitudinal welds is shown in Fig

15 Fig. 8. Contribution (%) of some fuel assemblies to the fluence of welds at positions 105 o and 345 o (Anderson & Whitmarsh, 1984). The following procedures are applicable for reducing the fluence rate (Franklin & Marston, 1983; Bagnal et al., 1984; Moylan & Balkey, 1987; Moylan et al., 1987; Leitz & Koban, 1989): 1. Low-leakage fuel management. Some or all of the peripheral fresh fuel assemblies are replaced by low reactivity fuel assemblies, i.e. those having spent one to three cycles in the reactor. 2. Some of the peripheral fuel assemblies are replaced by dummy assemblies, which contain stainless steel or zirconium rods instead of UO 2 pellets. Either partially or fully replaced assemblies can be used. Typically 5-10% of the fuel assemblies need replacing to maintain circumferential symmetry. 3. Installation of neutron absorbing materials on the core periphery. For example, peripheral control rods or burnable absorber rods placed at critical locations can be used to reduce the fluence rate. When power is reduced at the core periphery, power derating can be avoided only if the power of remaining assemblies is increased. Generally this means an increase in power peaking and, if power is not reduced, a decrease in thermal margin. In some cases sufficient reduction in the fluence rate for achieving the design service life of the RPV could not be reached without reducing core power (Franklin & Marston, 1983; Bagnal et al., 1984). More detailed data on possible fuel management alternatives are given in Table 1. 14

16 Table 1. Fuel management fluence reduction alternatives (Meyer et al., 1990). Group Options A) Loading pattern 1) Annual fuel cycles modification 2) Low leakage loading pattern (L 3 P) 3) Multi-enrichment regions 4) High discharge burn-up 5) Low-low leakage loading pattern (L 4 P) B) Poisons in guide tubes 1) Peripheral poisons 2) Peripheral burnable poisons C) Modified assembly designs 1) Radial blanket rods 2) Variable enrichment assemblies 3) Stainless steel rods or cells D) Radial assembly designs 1) Dummy assemblies 2) Radial half assemblies 3) Peripheral burnable poisons /large water holes E) Other 1) Reconstitutable assembly The standard loading scheme (OUT-IN) leads to an even core power distribution, i.e. minimum power peaking, and hence to maximum core power. Usually it is not necessary to use burnable absorber fuel to even out the power distribution at the start of the cycle (Franklin & Marston, 1983; Bagnal et al., 1984). The simplest way to reduce the fluence rate locally would be to replace only the adjacent fuel assemblies with assemblies with high burn-up (e.g. two cycles exposed). This (low-fluence) scheme can be performed without a marked increase in power peaking, if the number of replaced assemblies is small (Bagnal et al., 1984). However, reduction of the overall neutron leakage remains small. The effect on RPV lifetime may also be smaller than expected if some other location becomes critical. A fluence rate reduction factor of up to 2 with little or no increase in power peaking (and without reducing power) seems to be achievable locally for most PWRs, when fresh fuel is replaced by two cycles exposed fuel adjacent to the critical locations. The overall neutron leakage reduction is slight (Bagnal et al., 1984). 15

17 Previously the main objective of low-leakage schemes was to minimise the overall neutron leakage from the core, with the motivation of improving the fuel economy. Present low-leakage loading schemes are often modified to minimise the fluence rate, particularly at the critical location(s) of the RPV for mitigating irradiation embrittlement, although this may increase somewhat the overall neutron leakage from the core and raise fuel costs. Some 30-40% local reduction in the fluence rate at the RPV and 1.2% reduction in the overall neutron leakage (compared to the OUT-IN scheme) can be achieved with only a slight increase in power peaking (less than 3%) by following a modified low-leakage fuel management scheme, where one and two cycles exposed assemblies are loaded in selected peripheral locations, while the power of certain other assemblies is increased to avoid a reduction in core power (CE 217 assembly core) (Bagnal et al., 1984). As a consequence, changes in assembly enrichments and the use of burnable absorber fuel are required. Examples of various low-leakage loading schemes and their use are given in Table 2. Table 2. Fuel vendor low-leakage management schemes (Franklin & Marston, 1983 and Anderson & Whitmarsh, 1984). VENDOR NAME PATTERN TYPICAL FLUX TYPE REDUCTIONS Babcock & Wilcox LBP (1) IN-OUT-IN 30-40% IN-IN-OUT 50% locally Combustion SAV-FUEL IN-OUT-IN (4) 20% Engineering IN-IN-OUT Exxon LRL (2) Mixed 50% locally Westinghouse L 3 P or IN-OUT-IN 10% to 50% LLLP (3) (1) LBP: Lumped Burnable Poison (2) LRL: Low Radial Leakage (3) LLLP: Low-Leakage Loading Pattern (4) SAV-FUEL was initially IN-OUT-IN but as IN-IN-OUT has become attractive, it has been used as a general name for CE low-leakage plans. (5) These schemes are intended to improve fuel cycle economics. CE estimates that a scheme designed to improve economics and reduce fluence rate at vessel welds would reduce neutron fluence rate at the vessel by 20-50% A fluence rate reduction factor of up to about 3-5 can generally be achieved (without the need to reduce power) by applying low-leakage fuel management, if also part of the remaining peripheral assemblies at selected locations are replaced with dummy stainless steel assemblies (with stainless steel rods instead of UO 2 ), which not only reduce neutron production, but also to some extent reflect neutrons back to the core interior (Guthrie et al., 1982 and Bagnal et al., 1984). When dummy assemblies are used the need for burnable absorber fuel and an increase in power peaking is obvious (Todosow et al., 1983 and Bagnal et al., 1984). Maximum achievable fluence rate reductions for different loading schemes are given in Table 3. 16

18 Table 3. Decrease in the fast neutron fluence rate at the RPV due to changes in fuel management. Comparison is made with the OUT-IN-IN scheme (Franklin & Marston, 1983). LOADING PATTERN MAXIMUM FLUENCE RATE REDUCTION (%) OUT-IN-IN (REFERENCE) 0 IN-OUT-IN 30% IN-IN-OUT 40% IN-IN-OUT: Max. local 50% OUT-IN-IN: 4-cycle fuel at selected locations 60% locally IN-IN-OUT: 4-cycle fuel at selected locations 70% locally IN-IN-OUT: 4-cycle fuel at welds and control rods in the assemblies at the welds 90% at weld Dummy peripheral assemblies 90-95% The calculated effect of fluence rate reduction in a case where the implementation occurs after 7 full power years (EFPY) is shown in Fig. 9. After this operation time most of the expected transition temperature shift has already occurred. The difference in EoL RT NDT for the 10:1 fluence rate reduction scheme is about 45 o C. The horizontal lines show the NRC screening criteria for longitudinal welds (132 o C) and circumferential welds (149 o C). Most B&W reactors are using IN-OUT-IN cores (Table 2). The IN-IN-OUT loading scheme causes only a 2-3% increase in power peaking (177 fuel ass. B&W core) and has no significant effect on mechanical or thermal-hydraulic conditions. The resulting reduction in required 235 U-enrichment may reduce fuel cycle costs. On the other hand, burnable absorber fuel is required to decrease power peaking (Anderson & Whitmarsh, 1984). In some plants it has been possible to reduce the fluence rate at the RPV even by a factor of 10, when both low-leakage fuel management and dummy assemblies were applied (Franklin & Marston, 1983 and Leitz & Koban, 1989). Generally a fluence rate reduction exceeding a factor of 5 is not possible without reducing power (Bagnal et al., 1984). In general, the maximum achievable and realistic reduction in fluence rate depends on thermal margins. 17

19 Fig. 9. RT NDT shift over time for a range of fluence rate reduction schemes (Franklin & Marston, 1983). The employed low-fluence loading management techniques (for U.S. reactors), where two or three cycles exposed fuel or shield assemblies are placed on the core periphery, are expected to be used also for plant life extension (Carew & Lois, 1991) Calculated cases for different cores Various optimised combinations calculated for the CE 217 assembly core and the RPV with three longitudinal welds (see Fig. 4) are compared in Table 4. 18

20 Table 4. Calculated reduction in fluence rate and corresponding increase in power peaking for CE 217 assembly core (longitudinal welds at 120 o intervals, 0 o weld critical) (Bagnal et al., 1984). FUELLING SCHEME FL. RATE REDUCTION factor or (%) POWER PEAKING INCREASE (%) OVERALL (% ρ) LOCAL AT 0 o /30 o OUT-IN 0 1.0/ cycles burned fuel at critical location (low-fluence scheme) Low-leakage (LL) loading scheme LL + control rods at selected peripheral locations LL + dummy assemblies at selected peripheral locations LL + dummy ass. + control rods at selected peripheral locations /1.0 (40/0%) /1.5 (40/33%) /1.2 (54/17%) / * / * *) True fluence rate reduction factor (limiting fluence rate peak changed to 30 o weld). The calculations performed by Aronson et al. (1983) for different core geometries, i.e. Ft. Calhoun-1 (CE), H. B. Robinson-2 (Westinghouse) and Oconee-1 (B&W), showed that up to a 45% reduction in the fluence rate at the RPV can be achieved (locally) when the power of the peripheral assemblies is reduced 50% by lowleakage loading schemes. The local increase in power peaking was 20% when no steps were taken to flatten the power distribution. Reduction of (local) fluence rate by a factor of up to 9-18 was shown to be possible when peripheral assemblies were replaced with dummy ones. The corresponding increase in local power peaking was as large as 30-40% without flattening. In general, greater reductions in the fluence rate are possible for RPVs where the limiting locations are along the symmetry boundaries of the core, i.e. on a circumferential weld or base metal. If longitudinal welds exist, a practical limit on fluence rate reductions appear to be around 3 (control rods at the core periphery, 18- month cycle, Westinghouse design) (Stucker et al., 1983). Fig. 10 shows a case in which a fluence rate reduction factor of 7 was achieved at a circumferential weld location by using part-length shield assemblies on the core periphery. 19

21 Fig. 10. Pressure vessel fluence reduction using part-length shield assemblies (PLSA) (Carew & Lois, 1991) Special fuel designs Selection of fuel designs for a low-leakage scheme is based on the evaluation of various parameters, including cost, safety and impact on plant operations (Twitchell, 1991). The number of fluence rate reduction assemblies depends on the number and location of critical welds. The number of these assemblies will be larger if core symmetry is to be maintained. Assembly types used in fluence reduction programs supported or analysed by the Advanced Nuclear Fuel Co. are listed below (Twitchell, 1991): 1. Assemblies with high burn-up (usually three cycles exposed). 2. Low enriched assemblies in which the bottom third of all fuel rods contain stainless steel. 3. Reconstituted assemblies with high burn-up and multiple rows of stainless steel rods. 4. Low enriched assemblies with multiple rows of stainless steel rods. 5. Assemblies with high burn-up and in which Hf inserts are placed in the guide tubes. 6. Low enriched assemblies in which Hf inserts are placed in the guide tubes. Implementation of a low-leakage loading scheme is not always a simple and cheap way of reducing the fluence rate at the RPV, although no structural modifications of core components are needed. Besides long transfer times, low-leakage loading schemes mean totally different fuel management compared with OUT-IN schemes. For example, changes in the used enrichments, possibly enrichment zoning, and 20

22 extended use of burnable absorber fuel are often required (Bagnal et al., 1984). Although a change to lower enrichments usually means lower fuel costs, using several enrichments and burnable absorber fuels tend to increase costs. At any rate, use of burnable absorber fuels decreases reactivity and thus any economic gain achieved with lower neutron leakage. A detailed analysis of different loading schemes, including neutron transport calculations to predict the effect of loading schemes on the fluence rate at the RPV, should thus be performed before any ranking of schemes is possible. The low-leakage scheme calculated for the CE 217 assembly core, which led to a 40% reduction in fluence rate at the critical weld position (compare Table 4), increased the cycle length by 1.5 MWd/kgU and decreased the required fresh fuel enrichment by 0.25 wt% besides a reduction in overall neutron leakage (-1.2%) (Bagnal et al., 1984). The tendency towards longer cycle lengths (and higher discharge burn-ups) makes it more difficult to decrease neutron leakage as fuels with higher initial enrichments are required (Franklin & Marston, 1983). If fresh fuel with high enrichment is placed on the core periphery, both neutron leakage and the fluence rate at the RPV are increased. If fuel with low enrichment is used on the core periphery, an increase in power peaking follows and hence the need to further increase the number of burnable absorber fuel assemblies in the core interior. However, a 24 month cycle length together with extended burn-up and low-leakage fuel management is pursued by some U.S. utilities (Strasser et al., 1991). One can conclude that the extended use of low-leakage schemes has only been possible due to developments in absorber fuels, which are needed especially at the start of cycles when reactivity is inherently highest. There are different ways to make the core power distribution more even. These include: - Placing burnable absorber fuel in the core interior. One example is the Westinghouse ZrB 2 integral fuel (first irradiation in 1987) (Secker & Erwin, 1990 and Fecteau, 1991). Burnable absorber fuels are used to reduce reactivity at the start of operating cycles. - Using so-called inert rod cluster assemblies (proposed by Fragema), which include either stainless steel or Zircaloy pins, in the core interior to reduce reactivity (Quinaux et al., 1986). This fuel type provides a time-independent decrease in neutron production. 2.2 SHIELDING OF THE PRESSURE VESSEL Neutron flux to the RPV can be reduced by fitting new materials between the outer fuel elements and the RPV, as neutron moderation and diffusion depends on the detailed neutron scattering and absorption cross-sections of the materials. In fact, the use of dummy assemblies can also be regarded as shielding. One possibility is to modify the core support barrel or core shroud so that e.g. stainless steel shields (patches) can be attached to selected locations (Pat , Federal Republic of 21

23 Germany; Bagnal et al., 1984). Materials like tungsten or some metal hydrides have also been considered (Dragonajtys et al., 1991). These materials are more efficient than stainless steel in reflecting fast neutrons back to the core, reducing the number of fast neutrons reaching the RPV (and improving the neutron economy of the core) (Moylan et al., 1987). For example, a 50 mm thick stainless steel patch is estimated to reduce the fluence rate at the RPV by a factor of around 1.5 (Bagnal et al., 1984). Various technical solutions to make shielding possible have been developed. One fixing solution is presented in Fig. 11. Shielding enables relatively high fluence rate reduction factors to be achieved, i.e It does not normally reduce flexibility in loading, but the achievable benefit is decreasing rapidly as the fluence of the RPV is increasing. Besides, the achievable fluence rate reduction depends on the shield thickness, which is limited by the space between the core support barrel and the RPV. A disadvantage is also that the temperature of the shield material increases due to gamma heating (Schwirian et al., 1986 and Ayres & Siegel, 1994a). Aronson et al. (1983) calculated (compare p. 19) the shielding effect of peripheral assemblies with a stainless steel/water volume fraction of 0.4 or 0.7, equipped either with nominal or maximum size stainless steel rods, respectively. With the 0.4 volume fraction EoL fluence (1-8 x n/cm 2, E > 1 MeV) reductions of 9-12% were achieved from initial fluence values of x n/cm 2 (E > 1 MeV). With the 0.7 volume fraction, a total EoL fluence (1 x n/cm 2, E > 1 MeV) reduction of 17% was achieved from an initial fluence value of 0.2x10 19 n/cm 2 (E > 1 MeV). One concept according to which neutron reflectors are placed between the core baffle plate and the core barrel is shown in Fig. 12a (Schwirian et al., 1986 and Moylan et al., 1987). When a neutron shield panel is used (instead of a thermal shield) it can be bolted to the core barrel as shown in Fig. 12b. Neutron transport calculations performed for the Point Beach NPP showed that reduction factors of 2-3 could be achieved in the peak fluence rate, if the combination of a heavy metal reflector and a shield panel made of metal hydrides or a high density metal alloy was applied. It was estimated that a fuel cycle cost benefit of about 2% could be achieved for any design with a reflector (Moylan et al., 1987). Preliminary vibration, flow and seismic investigations also showed that the design criteria were likely to be satisfied for each design (Figs. 12a-b). 22

24 Fig. 11. A shielding solution to reduce fluence rate at a 180 o axial weld location (Ayres & Siegel, 1994a). 23

25 a) b) Fig. 12. Neutron reflector (a) and shield panel (b) constructions considered for reducing RPV fluence in the Point Beach NPP (Moylan et al., 1987). Significantly higher fluence rate reduction factors than those presented above can be achieved by increasing the volume fraction of steel between the core and the RPV. Fig. 13 shows a solution where shields (reflectors) between the baffle and the core barrel take up about 90% of the volume, which is reported to give a fluence rate reduction factor of 6.83 (900 MWe PWR). The reflector has been assembled from stainless steel blocks locked axially with threaded rods. Columns are equipped with vertical holes which allow water circulation for cooling. The structure also allows free expansion and mutual alignment of the columns with respect to each other. The reflector has been estimated to reduce cycle costs by about 2.5% due to the reflecting effect (Vrillon & Luneville, 1991). The modification cost of the core (including loss in power production) due to shielding may be significant. A comprehensive coolant flow analysis is also necessary (Bagnal et al., 1984). Issues that must be addressed include thermalhydraulic requirements, mechanical-fluids interactions, disposal, surveillance, inservice inspection and future maintenance (Server et al., 1993). Shielding has probably not been applied in commercial PWRs needing considerable structural modifications. 24

26 Fig. 13. Neutron reflector assembled from stainless steel blocks (Vrillon & Luneville, 1991). 2.3 FACTORS AFFECTING FLUENCE RATE REDUCTION In general, the aim of fluence reduction procedures is to reduce the fluence rate at critical locations of the RPV without limiting too much the operational flexibility of the reactor and, if possible, without reducing power. It is clear that power reduction is not a problem when it must anyway be performed for degradation of other components (Fenech, 1985). The minimum reduction in fluence rate required for a certain design service life depends on - the design EoL fluence of the RPV; - the circumferential fluence rate distribution and the initial value of the fluence rate; - composition of the RPV base metal and welds, i.e. the irradiation embrittlement sensitivity; - location and number of welds in the RPV; - years of operation before the intended fluence reduction measures. The restricting boundary conditions in applying different fluence reduction schemes are, for example, - the availability of thermal margins and the resulting possibility to increase power peaking without operational restrictions and without reducing power; - the operational margins (pressure-temperature windows); 25

27 - reactivity margin, especially when long cycle lengths (18 moths) are used; - core configuration and size. Evidently the applicability of different fluence rate reduction methods is highly plant-specific. 2.4 SECONDARY CONSEQUENCES OF REDUCED FLUENCE RATE Besides the benefits, a reduced fluence rate causes some consequences which must be taken into account. Reducing the fluence rate at the RPV decreases the signal-to-noise ratio in power distribution monitoring. This may be a potential problem when large reductions in the fluence rate are pursued (Perrin, 1993). Implemented and intended fluence reduction measures have also to be taken into account in scheduling the withdrawals of surveillance capsules. The fraction of Pu in the fissile material rises when fuel burn-up is increased (Table 5). Due to the higher relative neutron production in Pu fission, the decrease in fast neutron leakage from the core remains somewhat smaller than could be concluded from the burn-up when high burn-up fuel is placed on the core periphery (Anderson & Whitmarsh, 1984). Table 5. Effect of Pu build-up on ex-core fluence (Anderson & Whitmarsh, 1984). Number of Cyles Mid Cycle Burn-up (MWd/kgU) Fraction of Pu Fissions 23% 45% 59% Neutrons per Watt-Second 7.83 x x x Special Effect a) a) Fast neutron fluence at the RPV inner diameter relative to 100% 235 U fissions at the same reactor power. 26

28 3 MODIFICATION OF EMERGENCY CORE COOLING AND OTHER SYSTEMS Different protection systems are used to protect RPVs against pressure and/or thermal transients, i.e. pressurised thermal shocks (PTS), low temperature overpressure transients and high temperature transients. 3.1 PRESSURIZED THERMAL SHOCKS The primary measures to mitigate RPV embrittlement in operating power plants are those reducing the fluence rate at the RPV. As the most severe expectable loading situation for RPVs is considered to be a PTS, an extra safety margin can be achieved also by modifying the emergency cooling system so that the maximum loading in the RPV during such events is reduced. A PTS is a serious loading condition for the RPV, because - high thermal and mechanical stresses due to pressure are created near the inside surface of the RPV; - the fracture toughness of the RPV material near the inside surface is reduced due to both the rather low temperatures associated with a PTS and the high neutron fluence. The screening criteria for PTS are set by the NRC in 10CFR50.61 as follows: - RT PTS = 132 o C for plates, forgings and axial welds. - RT PTS = 149 o C for circumferential welds. when RT NDT + irradiation shift in RT NDT on the inside surface of the RPV material shall be lower than RT PTS minus the required margin. These criteria are based on failure probability analysis performed for certain reactor configurations. The PTS criteria is reported to have a major impact on plant operability and safety when the criteria are exceeded (Gamble, 1994). By 1982 eight significant PTS events (in U.S. PWRs) had been identified by the NRC (Table 6). 27

29 Table 6. Significant pressurised thermal shock events in U.S. PWRs (Chexal et al., 1983). Plant/Vendor Date Initiating Event H. B. Robinson/Westinghouse 4/28/70 Steam line break H. B. Robinson/Westinghouse 11/5/72 Stuck open steam generator relief valve H. B. Robinson/Westinghouse 5/1/75 Reactor coolant pump seal leak Rancho Seco/Babcock & Wilcox 3/20/78 Excessive feedwater transient Three Mile Island 2/Babcock & Wilcox 3/28/79 Stuck open relief valve on pressuriser Prairie Island/Westinghouse 10/2/79 Steam generator tube rupture Crystal River 3/Babcock & Wilcox 2/26/80 Inadvertent opening of a power operated relief valve R. E. Ginna/Westinghouse 1/25/82 Steam generator tube rupture As a most severe PTS is generally considered to be emergency core cooling due to a small leak associated with the operation pressure of the safety relief valve, i.e. pressure x operation pressure. In general, RPV failure risk can be reduced - by minimising the probability of abnormal events such as PTS; - by minimising the maximum expectable stress concentration in the RPV during possible abnormal events. Emergency core cooling systems can be modified in order to reduce stresses during PTS - by increasing coolant temperature and/or mixing (for example the location of coolant inlet(s) can be changed) in order to reduce thermal stresses; - by limiting the maximum pressure increase. The significance of thermal-hydraulic parameters in assessing the PTS risk has been emphasised by the NRC. The cooldown rate, heat transfer coefficient and steadystate temperature adjacent to the vessel wall are regarded as key parameters (Chexal et al., 1984). Thermal-hydraulic calculations and miniature model simulations have been used to assess the temperature and flow conditions associated with different PTS events. The flow chart of the integrated approach developed by EPRI for performing plantspecific PTS analyses is given in Fig. 14. The code calculates the stress intensity versus time from the results of 3-dimensional fluid mixing and thermal stress analysis, and finally compares the stress intensity and fracture toughness profiles. 28

30 Fig. 14. EPRI integrated approach on RPV integrity analysis (Chexal et al., 1983). In minimising temperature differences in the coolant and between the coolant and the RPV, mixing of the cold high-pressure safety injection water and the hot recirculation water should be as complete as possible before the water reaches the RPV. Mixing of these waters in the downcomer and cold legs is a primary concern for PTS scenarios (Chexal et al., 1983). Mixing processes at different locations are described schematically in Fig. 15. Effective mixing occurs at the junction of the injection and cold leg pipes and at the junction of the cold leg pipe and the downcomer. Between these locations no significant mixing occurs due to stratification of the cold injection water and the hot recirculation water (volume part 3 in Fig. 15). Mixing below the cold leg is often inadequate, which means that the local narrow cooler zone (temperature of zone 5 less than that of zone 4) will induce additional axial thermal stresses. The significance of a cold leg has been verified experimentally in the HDR tests. 29

31 Fig. 15. Schematic diagram of governing mixing mechanisms in the cold leg and downcomer without loop flow (upper figs., HPI = high-pressure injection). Lower figs. show specification of control volumes used for modelling (Chexal et al., 1984). In Doel 1 and 2 (Belgian 390 MWe PWRs) the PTS issue was solved in 1992 by modifying the high-pressure safety injection (Gerard, 1993). In the original design cooling water was injected directly into the downcomer, which resulted in insufficient mixing and hence in a large difference between the average temperature in the downcomer and the temperature of the cold safety injection stream. A typical temperature evolution for a small break LOCA (loss of coolant accident) and formation of this rather large temperature difference is shown in Fig. 16. In the new configuration the injection was done directly into the upper plenum above the core as shown in Fig. 17. The K Ic, K Ia and K I curves before and after the modification (for a small break LOCA) is presented in Fig. 18. It should be noted that this modification was performed only for one particular PTS transient, which otherwise would have required a probabilistic failure analysis as recommended by Regulatory Guide Besides the modification described above, the temperature of the safety injection water in storage tanks had been increased to 40 o C. 30

32 Fig. 16. Temperature evolution during small break LOCA transient (Doel 1/2) (Gerard,1993). Fig. 17. Downcomer geometry and safety injection modification (Doel 1/2) (Gerard, 1993). Fig. 18. Fracture toughness (K1c, K1a) and stress intensity (K1) in small break LOCA transient at 1500 s before and after downcomer modification (Doel 1/2) (Gerard, 1993). Increasing the safety injection water temperature has a twofold effect: the stress intensity in the RPV wall falls due to lower thermal stresses, and the fracture toughness of the RPV material is increased. This is illustrated in Fig. 19, which shows the results of two hypothetical PTS analyses made for the four-loop Westinghouse H. B. Robinson unit (Dickson & Simonen, 1992). The fluence in the most embrittled axial weld (assumed Cu = 0.13%, Ni = 0.8%) was predicted to be 3.15x10 19 n/cm 2 at 32 EFPY. In transient 1, fracture is predicted to occur 80 min after the onset of the transient due to repressurisation, whereas for transient 2 no fracture is expectable. Transients 1 and 2 were identical, except for the coolant temperature, which was 8.3 o C (15 o F) higher in transient 2. 31