Resolution of Generic Safety Issues: Issue 107: Main Transformer Failures (Rev. 3) ( NUREG-0933, Main Report with Supplements 1–34 )
This issue was identified in a DL/NRR memorandum1183 which called for an assessment of the high failure frequency of main transformers and the resultant safety implications. Concern for this issue arose when the North Anna Power Station had seven main transformer failures in 26 months; five of these resulted in reactor trips. Of the seven failures, three included rupture of a transformer tank that resulted in two fires. One of the fires spread beyond the transformer bay to the turbine bay. In a report1184 prepared for the NRC by LLNL, it was concluded that there was a possibility of generic implications arising out of the plant-specific failures reported for the North Anna units.
The potential generic concerns identified in the LLNL report1184 included the fire protection system, overhead conductor/buses, cable trays, storage of flammable materials, and oil-filled transformers in general. In addition, certain secondary aspects of the transformer failures were identified which included cascading effects, extensive electrical/mechanical damage, and missiles/explosions, although the LLNL report noted that these latter items appeared to be either indirectly or remotely related to specific safety-significant concerns. Existing NRC regulations and guidance pertaining to fire protection and some of the generic concerns raised in the LLNL report1184 are embodied in 10 CFR 50 Appendix R, the SRP,11 and Regulatory Guide 1.120.1185 In this analysis, the need for additional actions by the licensees to prevent main transformer failures and to reduce the resultant risk were evaluated.
Safety-related loads in nuclear power plants are supplied from buses that can be supplied from any one of the following sources: (1) the unit auxiliary (main) transformer; (2) the startup transformer (or reserve auxiliary transformer); or (3) the emergency onsite power supply (i.e., diesel generators). A main transformer failure will result in a loss of load or unbalanced load on the main generator. This would lead to turbine/generator trip and power would not be available to the unit transformers for the station power; however, station power can be obtained from the grid through the startup transformer or from emergency onsite power sources. Switchyards have redundant systems to provide sufficient relaying and circuit breakers so a transformer failure is not expected to cause a loss of offsite power.
Other generic concerns associated with this issue included: (1) oil from a ruptured transformer could float on the water delivered to extinguish the fire by the fire protection system such that the fire will move in the direction of drainage; (2) the fire may propagate to overhead cables and buses and create the need for access to adjacent locations (such as building roofs) by fire-fighting crews.
Resolution of this issue could involve the following actions:
(1) Evaluation of main transformer design and arrangements by licensees to ensure that the supply of offsite power is protected against transformer fires and smoke. Design requirements should be established for routing and separation of offsite power source feeds to protect against power loss due to a transformer fire.
(2) Review of fire protection system features for the main transformers for adequacy and revision, as necessary, to ensure that a potential fire is prevented from spreading to other plant areas. The review should address the deluge system, drainage system, fire barriers, and fire-fighting equipment and procedures.
(3) Review of maintenance and operating procedures for the main transformers for adequacy and revision, as necessary.
(4) Modification of drainage systems, if necessary, to provide drains for each transformer so that liquids flow away from the turbine building, power lines, and safety-related cables to the reactor and related safety equipment. Modifications could include adding drains, building dikes, and sloping the transformer yard away from buildings and other transformers.
(5) Modification of fire-fighting equipment and procedures, if necessary. This could include longer hoses, increased ease of access to building roofs, mobility of fire-fighting equipment, and training for personnel.
(6) Relocation of power lines to the safety-related buses, if necessary, so that they would not be affected by a fire in the transformer bay.
To establish the priority of this issue, the potential reduction in core-melt frequency as a result of improved main transformer reliability due to implementation of the proposed solutions was quantified. It was believed that improved reliability of main transformers would reduce the frequency of transients induced due to main transformer failures, thus leading to enhanced plant safety.
In the representative plant PRAs (Oconee-3 for PWRs and Grand Gulf-1 for BWRs), main transformer failures are integrated into a category of transients that result from loss of network load. The affected PRA parameters are transients other than loss of offsite power requiring or resulting in a reactor shutdown, i.e. T2 (frequency of 3/RY) and T23 (frequency of 7/RY) for Oconee-3 and Grand Gulf-1, respectively. It was assumed that implementation of the possible solutions would enhance the reliability of main transformers and thus reduce the frequency of the resultant transients.
Data in NUREG/CR-38621186 on a specific transient category, characterized as a loss of incoming power to a plant as a result of onsite failure (such as main transformer failure), suggest that the transient frequency associated with this category is 0.02 event/RY. In addition, the IEEE reliability data for liquid-filled transformers (347 to 550 KVA) at nuclear power plants indicate that the main transformer failure rate due to all causes was 2.67/million-hours. This corresponded to an annual frequency of 0.023 failure/year for main power generator or unit transformers. This value was used as the base case for the failure frequency of main transformers. The second aspect of the main transformer failure, the risk from resulting fire, was determined to be insignificant and was not analyzed further. This conclusion was based on the findings of the Oconee-3 PRA which included the analysis of fires and their potential for causing failures of redundant safety-related components. Also, no particular sensitivity to main transformer fires was identified in NUREG/CR-5088.1211
It was assumed that implementation of the possible solutions (i.e., no design improvements to the transformer but improved maintenance and mitigative designs/procedures) would increase the reliability of main transformers by 50%. Therefore, the adjusted case main transformer failure frequency was estimated to be 0.01 event/RY. In addition, the adjusted case frequencies of the resultant transients (T2 and T23) were estimated as follows:
|T2||= (3 - 0.01)/RY|
|T23||= (7 - 0.01)/RY|
Incorporating these values in the Oconee-3 and Grand Gulf -1 PRAs provide reductions in core-melt frequency estimates of 1.4 x 10-7/RY for PWRs and 3.6 x 10-8/RY for BWRs.
This issue was assumed to be pertinent to all LWRs and thus had an affected population of 90 PWRs and 44 BWRs with average remaining lives of 28.8 years and 27.4 years, respectively. Based on the Oconee-3 and Grand Gulf-1 PRAs, the associated public risk reduction was estimated to be 0.38 man-rem/RY and 0.25 man-rem/RY for PWRs and BWRs, respectively. Thus, the average public risk reduction associated with this issue was 9.6 man-rem/plant.
Industry Cost: Implementation of the possible solutions at the affected plants would require review of existing systems and procedures and hardware changes. It was estimated that the review of the existing systems and procedures would require 15 man-weeks/plant at $2,270/man-week. These efforts would include evaluation of the fire protection systems, review of protective circuitry, review of operating and maintenance procedures, revision of operating and maintenance procedures, and revision of staff training. It was also assumed that, as a result of these reviews, about 10% of all affected plants would require hardware changes, modifications to fire protection systems, and re-routing of cables around the main transformer areas. It was estimated that 9 man-weeks would be required to prepare the design modifications and acceptance testing plan, install and test hardware changes, and revise procedures. Hardware and labor were estimated to cost $48,000/plant to provide the following: additional drains, gravel, and concrete to slope the area around the transformers and construct dikes; additional power lines to route power to the buildings; additional breakers to protect equipment connected to the auxiliary transformers; and longer fire hoses. The cost was itemized as follows:
Dike (250 ft. long, 4 ft. high) =$ 3,750
Concrete and Gravel = 15,800
Power lines (1,000 ft) = 5,000
Breakers (2 at $2500 each) = 5,000
Fire Hose/Storage Cabinet (110 ft) = 500
Note: An escalation factor of 1.8 was used by PNL to convert 1982 dollar values to 1988. Therefore, the cost to implement the possible solutions at 90% of the plants was about $30,000/ plant; for the remaining 10%, the cost was estimated to be $100,000/plant. The average cost for the affected population was approximately $40,800/plant.
For the affected plants, periodic review of main transformer procedures, operations, and maintenance was estimated to require 0.2 man-week/RY. At a cost of $2,270/man-week, this amounted to $450/RY. In addition, those plants requiring hardware modifications (10% of affected plants as discussed above) require 1 man-week/RY (or $2,270/RY) for periodic maintenance/inspection of drains and new diked areas, removal of trash from drains, etc. Plant maintenance and operation costs are recurring costs and were adjusted for present worth at a 5% discount rate over the 28.3-year average remaining plant life for the 134 affected plants. This resulted in an average plant cost (present worth) of $11,200/plant.
It was believed that improvements to the reliability of main transformers and improvements to fire protection systems could potentially result in: (1) avoided costs of replacing a transformer damaged by fire (3 out of 14 transformer failures resulted in fire, or 0.002 main transformer failure/RY); and (2) avoided replacement power costs associated with reducing the number of reactor trips caused by main transformer failures.
NRC Cost: NRC costs consisted of initial regulatory development and the resources required in support of the regulatory implementation. The initial regulatory development cost could involve the issuance of a generic letter or bulletin to the licensees, review of licensee responses, other related activities (i.e., revised design guidance, assessment of differences in plant design related to transformers, development of potential implementation measures), and the required technical, legal, and administrative staff labor. This portion of resource requirements was estimated to require 40 man-weeks ($90,000) in addition to potential outside contractor support (estimated to cost $50,000) for a total of approximately $140,000. Averaging this over the 134 affected plants resulted in an approximate NRC cost of $1,000/plant.
The implementation resource requirements consist of NRC labor to review utility plans to comply with revised guidance and additional inspection and monitoring of transformer maintenance/testing programs during the routine NRC plant inspections. This was estimated to require $4.1M over the life of all affected plants. These costs are also recurring costs and when adjusted for present worth, as indicated above, resulted in an average NRC cost (present worth) of $17,000/plant.
Total Cost: The total industry and NRC cost associated with the possible solution was estimated to be $70,000/plant.
Based on a potential public risk reduction of 9.6 man-rem/reactor and an estimated cost of $70,000/reactor for a possible solution, the value/impact score was given by:
(1) Implementation of the possible solutions was assumed not to involve any labor in radiation zones because the main transformers are not located in a building in which radioactive materials are used or stored and thus the radiation dose rates are zero.
(2) The core-melt frequency reductions of 1.4 x 10-7/RY for PWRs and 3.6 x 10-8 /RY for BWRs results in ORE avoidance associated with core-melt cleanup operations of 20,000 man-rem/core-melt.64 The accident avoidance over the remaining plant life was [(28.8)(90)(1.4 x 10-7/RY) + (27.4)(44)(3.6 x 10-8/RY)] (20,000)/134 or 0.06 man-rem/plant. The present worth cost of a core-melt accident was estimated to be $1.65 billion considering cleanup and replacement power cost over a ten-year period.64 The present worth of accident avoidance at each plant was estimated to be [(28.8)(1.4 x 10-7/RY)(90) + (27.4)(3.6 x 10-8 /RY)(44)]($1,650M)/134 or $5,000.
(3) Existing designs of operating nuclear power plants incorporate various independent means of supplying loads so that main transformer failures would not cause a total loss of offsite power. In addition, the promulgation of the station blackout rule (10 CFR 50.63) should further reduce the risk from loss of AC power from that considered in the Oconee-3 and Grand Gulf-1 PRAs.
(4) It was believed that implementation of the possible solutions could be accomplished during normal plant outages and would not require design modifications or work in radiation zones. The relatively high failure frequency of the main transformers at the North Anna plant highlighted a possible need for plant-specific evaluations by some licensees to review their main transformers and to implement an appropriate combination of the alternatives proposed in order to enhance safety.
Based on the above value/impact score, the issue was on the borderline between a low and medium priority for existing plants. However, it was believed that the risk estimates were high (because the effect of the station blackout rule was not included in the Oconee-3 and Grand Gulf -1 PRAs). Therefore, the issue was given a low priority ranking (see Appendix C) for existing plants.
Following a periodic review of low priority issues, NRR provided new information1749 on transformer failures that required a reevaluation of the issue. Further prioritization, using the conversion factor of $2,000/man-rem approved1689 by the Commission in September 1995, resulted in an impact/value ratio (R) of $11,565/man-rem which placed the issue in the DROP category.1750