Resolution of Generic Safety Issues: Issue 15: Radiation Effects on Reactor Vessel Supports (Rev. 3) ( NUREG-0933, Main Report with Supplements 1–34 )
This issue addressed the potential problem of radiation embrittlement of reactor vessel support structures (RVSS). It was originally identified as a Candidate USI in NUREG-070544 where it was recommended for further study before a judgment was made on its designation as a USI. In the initial prioritization of the issue in November 1983, it was concluded that the ORE associated with resolving the issue far outweighed the potential decrease in public risk. As a result, the issue was assigned a low priority until additional data on the problem became available that would warrant a reevaluation of the issue. In April 1988, data developed by ORNL1253, 1254 suggested that the potential embrittlement of the RVSS, as result of neutron irradiation damage, could be significantly greater than was previously anticipated. Based on this new information, MEB/RES requested a reevaluation of the issue in September 1988.1252
Neutron damage of structural materials causes embrittlement that may increase the potential for propagation of flaws that might exist in the materials. The potential for brittle fracture of these materials is typically measured in terms of the material's nil ductility transition temperature (NDTT), which is the lowest temperature at which the material would not be susceptible to failure by brittle fracture. As long as the operating environment in which the materials are used has a higher temperature than the materials' NDTT, no failure by brittle fracture would be expected. Many materials, when subjected to neutron irradiation, experience an upward shift in the NDTT, i.e., they become more susceptible to brittle fracture at the operating temperatures of interest. This effect should have been accounted for in the design and fabrication of RVSS. However, the ORNL research indicated that the upward shift in NDTT with increased exposure to neutron irradiation had been underestimated. The loss in fracture toughness could result in failure of the RVSS and consequent movement of the reactor vessel, given the occurrence of a transient stress or shock such as would be experienced in an earthquake.
ORNL surveyed RVSS designs at LWRs and categorized each plant into one of five categories or types of RVSS: (1) skirt; (2) long-column; (3) shield-tank; (4) short-column; and (5) suspension. Skirt supports are located away from the core with a large volume of intervening metal and water; radiation embrittlement of skirt RVSS was not anticipated. Long-column supports are located in a zone of potentially high neutron fluence and are thus susceptible to radiation damage. Shield-tank supports are also located in a potentially high radiation damage zone. Short-column supports include several subcategories that are located in various regions relative to the reactor core and have a wide variability in susceptibility to radiation damage. Many plants with this type of support have special designs for heat dissipation, including natural convection, forced convection, and water/cooling-coil designs. Suspension supports were employed at only one plant and, although these supports are located in a region of potentially high irradiation damage, the temperature may be high enough to preclude brittle fracture. However, for this analysis, plants employing the long-column, shield-tank, short-column, and suspension supports were assumed to be susceptible to irradiation damage.
A large seismic event can cause failure of auxiliary piping which can cause an embrittled RVSS to fracture thereby allowing the reactor vessel to move. Such movement can then worsen the resultant LOCA from the rupture of auxiliary piping by rupturing other piping attached to the primary coolant loop and instrument tubing attached to the bottom head of the reactor vessel.
The proposed resolution for some plants involved the application of local heaters and insulation for the RVSS to maintain operating temperatures well above the NDTT of the potentially embrittled support. This resolution would only involve those plants that employed long-column and shield-tank supports. Short-column and suspension supports were in a higher temperature environment and thus heaters were not necessary to maintain the temperatures above the NDTT. However, minor design and equipment changes would be needed to control the amount of heat dissipation applied to the short-column and suspension supports to ensure the NDTT of the structural materials did not exceed the environmental temperature. In all cases, appropriate safeguards must be installed to prevent overheating of the concrete around and in contact with the supports.
The number of potentially susceptible plants (78) was determined from the results of the ORNL survey and are summarized below:
|Plant Type||RVSS Type||Operating||Under Construction|
The ORNL report also provided the basis for estimates of the length of time a plant could potentially operate in a vulnerable condition, i.e., with embrittled reactor vessel supports. The radiation embrittlement of RVSS materials from two operating LWRs (Turkey Point and Trojan) was investigated and data on the change in NDTT over time were developed. The approximate time when the RVSS material was believed to become susceptible to brittle fracture occurred after 23 effective full power years. Therefore, the potential susceptibility of the RVSS to brittle fracture exists for 7 years at the end of a reactor's operating life, assuming an average operating life of 30 effective full power years. Data from the Oconee 3 and Grand Gulf 1 RSSMAP studies were used in this analysis to determine the estimated risk for PWRs and BWRs, respectively.
The assumed accident scenario was the occurrence of a seismic event of sufficient magnitude to cause fracture of an embrittled RVSS, subsequent movement of the reactor vessel, and a corresponding LOCA as attached piping ruptures. The analogous accident sequences were those involving LOCA initiators S1, S2, and S3 for Oconee (different initiator frequencies for three pipe diameters) and S for Grand Gulf; these are the corresponding LOCA initiators for pipe ruptures. However, this issue was concerned with only seismically-induced pipe ruptures, which were not addressed in the original Oconee and Grand Gulf studies. As a result, seismically-induced LOCAs were defined here and incorporated into the base case.
The base case frequencies of seismically-induced LOCA initiators SS1, SS2, SS3, and SS were assumed to be equal, i.e., the conditional probabilities of fracturing different sizes of pipe, given an earthquake, were assumed to be equal. Their base case frequencies were estimated as follows:
f(SS1) = f(SS2) = f(SS3) = f(S) = [f(PGA > 0.2g)][p(NDTT)][p(PR)]
|where,||f(PGA > 0.2g)||= frequency of a seismic event with peak ground acceleration greater than or equal to 0.2g; frequency = 7 x 10-4/yr.16|
|p(NDTT)||= conditional probability that a RVSS is susceptible to radiation damage and fails as a result of reactor vessel movement (this value is derived below)|
|p(PR)||= conditional probability of pipe rupture given movement of the reactor vessel [assumed to be accounted for in estimate of p(NDTT); effectively 1 for pipes of all diameters]|
The conditional probability of failure of an embrittled RVSS as a result of a seismic event [p(NDTT)] is a function of the NDTT at the time the seismic event occurs, the number and size of preexisting flaws in the support material, and the safety factor built into the design of the supports and selection of the material. As discussed above, the RVSS materials at some plants may exceed operating temperatures during the last 7 years of reactor operation. Assuming that this occurs, the safety factor built into the RVSS may not exceed 1 whereas, using previous predictions of radiation damage, this safety factor may be as much as 20. Using a correlation1255 between safety factor and failure probability, PNL determined64 that the conditional probability of failure leading to reactor core damage for a safety factor of 1 was 0.5. Using this value, the frequency of seismically-induced LOCAs was:
f(SS1) = f(SS2) = f(SS3) = f(S) =
|(7 x 10-4/RY)(0.5)(1)|
|3.5 x 10-4/RY|
PNL derived the base case frequencies by substituting the above frequency of the seismically-induced initiators into the minimal cut sets given in NUREG/CR-2800.64 The results were as follows:
|SS3H -||(PWR-3) = 1.4 x 10-6/RY|
|(PWR-5) = 2.0 x 10-8/RY|
|(PWR-7) = 1.4 x 10-6/RY|
|SS1D -||(PWR-1) = 2.4 x 10-7/RY|
|(PWR-3) = 4.8 x 10-6/RY|
|(PWR-5) = 1.8 x 10-7/RY|
|(PWR-7) = 1.9 x 10-5/RY|
|SS3FH -||(PWR-2) = 6.0 x 10-7/RY|
|(PWR-4) = 8.8 x 10-9/RY|
|(PWR-6) = 6.0 x 10-7/RY|
|SS2FH -||(PWR-1) = 1.1 x 10-8/RY|
|(PWR-4) = 8.0 x 10-9/RY|
|(PWR-6) = 8.8 x 10-7/RY|
|SS2D -||(PWR-1) = 1.8 x 10-8/RY|
|(PWR-3) = 3.6 x 10-7/RY|
|(PWR-5) = 1.3 x 10-8/RY|
|(PWR-7) = 1.4 x 10-6/RY|
|SS3D -||(PWR-3) = 1.9 x 10-7/RY|
|(PWR-5) = 2.7 x 10-9/RY|
|(PWR-7) = 1.9 x 10-7/RY|
|S1 -||(BWR-1) = 1.2 x 10-8/RY|
|(BWR-2) = 1.2 x 10-6/RY|
Summing the base case frequencies for the affected release categories, the following were obtained:
|PWR-1 = 2.7 x 10-7/RY||BWR-1 = 1.2 x 10-8/RY|
|PWR-2 = 6.0 x 10-7/RY||BWR-2 = 1.2 x 10-6/RY|
|PWR-3 = 6.8 x 10-6/RY|
|PWR-4 = 1.7 x 10-8/RY|
|PWR-5 = 2.2 x 10-7/RY|
|PWR-6 = 1.5 x 10-6/RY|
|PWR-7 = 2.2 x 10-5/RY|
Based on the above, the affected base case core-melt frequency was 3.1 x 10-5/RY for PWRs and 1.2 x 10-6/RY for BWRs. The possible solutions were assumed to eliminate the potential for radiation embrittlement of RVSS materials. Thus, the adjusted case core-melt frequency was zero and the potential reduction in core-melt frequency was 3.1 x 10-5/RY for PWRs and 1.2 x 10-6/RY for BWRs.
In order to obtain the consequences associated with this issue, the CRAC Code64 was used. An average population density of 340 persons/square-mile was assumed (the average for U.S. domestic sites) from an exclusion area one-half mile about the reactor out to a 50-mile radius. A typical Midwest site meteorology was also assumed. Based on these assumptions, the risk for each Release Category was stated in Appendix D of NUREG/CR-2800.64 Using the frequency estimates derived above, the total estimated risk from the base case was 41.6 man-rem/RY for PWRs and 8.6 man-rem/RY for BWRS. Since the possible solutions were assumed to eliminate the potential for radiation embrittlement of RVSS materials, the adjusted case risk was essentially zero. The risk reduction associated with this issue was as follows:
PWRs: (41.6 man-rem/RY)(77 reactors)(7 years) = 22,400 man-rem
BWRs: (8.6 man-rem/RY)(1 reactor)(7 years) = 60 man-rem
Therefore, the total potential risk reduction for all affected reactors was 22,460 man-rem.
Industry Cost: At operating plants, the solution consisted of controlling the temperature of the RVSS, either through application of local heaters and insulation or through controlling cooling systems that were already in place, to ensure that the temperatures of the structural materials did not fall below the materials' NDTT after irradiation embrittlement. At future plants, the use of non-susceptible materials was the proposed resolution. Since this could be accommodated during the design and construction stages of a plant, no additional costs were foreseen beyond those normally incurred during design and construction.
Affected backfit plants were assumed to implement the resolution after about ten years of reactor operation. It was further assumed that only plants with long-column and shield-tank supports would install and operate local heaters and insulation on their RVSS. The plants with suspension and short-column supports were assumed to implement measures to control or limit cooling of the RVSS. Affected forward-fit plants would implement the solution before fuel was loaded into the core. The following was the breakdown of the solutions at the 78 affected plants:
For plants with long-column and shield-tank supports, it was assumed that heaters would be attached to four reactor vessel support columns and that mounting hardware, metal-sheathed heating cables, switchgear, transformers, and a power controller would be installed. It was also assumed that the equipment would be installed during scheduled reactor outages. Therefore, no additional replacement power costs would be necessary. It was further assumed that access to the reactor cavity was possible for heater installation. PNL estimated the equipment cost to be $52,000/plant; labor associated with installation of this equipment was estimated to be 105 man-weeks/plant. At a cost $2,270/man-week, the installation cost for heaters was (105 man-weeks/plant)($2,270/man-week) or $245,000/plant. An additional cost of $26,000/plant was estimated for a Class V amendment. Therefore, the total implementation cost for plants that used heaters was $320,000/plant.
For plants with short-column and suspension supports that would utilize cooling methods, it was assumed that equipment and labor requirements were 10% of that estimated for application of local heaters and insulation. In this case, PNL estimated the equipment cost to be $5,200/plant; labor associated with installation of this equipment was estimated to be 10.5 man-weeks/plant. At a cost of $2,270/man-week, the installation cost for cooling was (10.5 man-weeks/plant)($2,270/man-week) or $25,000/plant. The Class V license amendment fee of $26,000/plant would also be applicable. Therefore, the total implementation cost for plants that used cooling was $56,000/plant.
Therefore, the total industry implementation cost was given by:
(18 plants)($320,000/plant) + (46 plants)($56,000/plant) = $8.34M
PNL calculated that operation and maintenance costs would be $130,000/RY for those plants that use heaters and $7,100/RY for those that use cooling. Therefore, the total operation and maintenance cost over the 7-year vulnerability period for the affected reactors was given by:
(18 plants)(7 years)($130,000/RY) + (46 plants)(7 years)($7,100/RY) = $18.7M
The total industry cost for implementation, operation, and maintenance of the possible solutions was $(8.34 + 18.7)M or $27M.
NRC Cost: PNL estimated that it would require 16 man-weeks of staff effort to develop the possible solutions. At a rate of $2,270/man-week, this amounted to $36,000; contractor support was expected to cost an additional $500,000. Therefore, the total NRC development cost was estimated to be $536,000.
NRC effort to support industry implementation of the solutions was estimated to be 15 man-weeks/plant for those with heaters and 2 man-weeks/plant for those with cooling. Assuming a rate of $2,270/man-week, the total NRC implementation costs were:
$2,270[(18 plants)(15 man-wk/plant) + (46 plants)(2 man-wk/plant)] = $822,000
NRC review time for operation and maintenance was estimated to be 1 man-week/RY for all affected plants. At a cost of $2,270/man-week, the total NRC cost for review of operation and maintenance of the possible solutions over the 7-year vulnerability period was given by:
(64 plants)(7 years)($2,270/RY) = $1.02M
Therefore, the total NRC cost for development, implementation, operation, and maintenance of the possible solutions was given by:
$(536,000 + 822,000 + 1,020,000) = $2.4M
Total Cost: The total industry and NRC cost associated with the possible solutions was $(27 + 2.4) or $29.4M.
Based on an estimated public risk reduction of 22,460 man-rem and a cost of $29.4M for a possible solution, the value/impact score was given by:
No occupational dose would be incurred during implementation, operation, and maintenance of the solutions at forward-fit plants. Based on a radiation field of 100 millirem/hr in the vicinity of the reactor vessel, PNL estimated the total occupational dose increase of the 64 backfit plants to be 1,880 man-rem. Operation and maintenance of the solutions at these plants were estimated to result in an additional risk of 5,100 man-rem. Thus, the total occupational dose increase from implementation, operation, and maintenance of the possible solutions was estimated to be 7,000 man-rem.
Occupational dose reduction due to accident avoidance would be realized at the forward-fit plants, as well as at backfit plants, over the last 7 years of reactor operation. The occupational dose reduction due to accident avoidance was calculated to be 330 man-rem for all 78 affected plants.
Based on the potential public risk reduction and value/impact score, the issue would have been given a medium priority ranking. Consideration of the net occupational dose increase associated with the possible solutions would not have changed this conclusion. However, because the change in core-melt frequency from implementation of the proposed solutions was estimated to be 3.1 x 10-5/RY for 99% of the affected plants (PWRs), the issue was given a HIGH priority ranking.
Work completed by the staff in resolving the issue led to the preliminary conclusion that the potential problem did not pose an immediate threat to public health and safety; this conclusion was reported to the Commission in SECY-89-1801256 in June 1989.
Preliminary results from a new theoretical model for determining damage by low energy neutrons showed that the large NDTT shifts previously observed may have been a function of the particular neutron energy spectrum to which the steel samples were exposed. Data from the High Flux Isotope Reactor (HFIR) surveillance program and from the neutron shield tank of the decommissioned Shippingport reactor were normalized to the same trend curve established from samples irradiated in materials test reactors. Extrapolation along the trend curve of NDTT change against exposure suggested that end-of-life values for RVSS would be on the order of one-third or one-fourth the value obtained by extrapolation of the HFIR data when plotted against traditional measures of neutron exposure. Substantiation of these results would allow the staff to conclude that the predicted degradation in RVSS toughness (fracture resistance) would be insufficient to cause concern during the 40-year license life of each plant.
Additionally, preliminary results from the analysis of the Trojan plant indicated that RVSS failure would not cause failure of the reactor coolant system piping or reactor vessel in the event of an SSE or guillotine break in the pressurizer surge line. However, the ability of the control rods to scram and the integrity of instrument lines connected to the bottom of the reactor vessel under these conditions were to be confirmed. The Trojan RVSS was studied because of its configuration and certain significant design and fabrication details which the staff considered to be among the most vulnerable to failure under accident loads.
The above tentative results indicated that plant safety could be maintained despite RVSS radiation damage; however, extensive confirmatory analyses needed to be performed to support this preliminary conclusion.
The staff found that there was significant variability in the effect of radiation on RVSS among plants because of the variety of RPV support designs, material properties, and fuel management procedures that affected the neutron flux and spectrum in the cavity. In order to encompass the uncertainties in the various analyses and provide an overall conservative assessment, several structural analyses conducted demonstrated the following:
(1) Postulating that one of four RPV supports was broken in a typical PWR, the remaining supports would carry the reactor vessel load even under SSE seismic loads;
(2) If all supports were assumed to be totally removed (i.e., broken), the short span of piping between the vessel and the shield wall would support the load of the vessel.
The results of the analyses virtually eliminated the concern for both radiation embrittlement and significant structural damage from a postulated RPV failure. A study of the neutron spectra at different HFIR pressure vessel surveillance locations and the staff's technical findings were published in NUREG/CR-6117632 and NUREG-1509,408 respectively. Based on the staff's regulatory analysis,672 the issue was RESOLVED with no new requirements.923 Consideration of a license renewal period of 20 years did not change this conclusion.