Resolution of Generic Safety Issues: Issue 144: Scram Without a Turbine/Generator Trip (Rev. 2) ( NUREG-0933, Main Report with Supplements 1–34 )
Before 1989, there were several reported occurrences of the failure of turbine/generators to trip after a reactor scram. These events were caused either by mechanical or electrical problems in the turbine electro-hydraulic control system. In particular, one occurrence at Crystal River-3 was caused by both a defective solenoid spring in the turbine trip circuitry and corrosion in a fuse holder, which produced a reduced DC voltage to the solenoid (LER 88-006). However, in this instance, the turbine also could not be tripped manually. Therefore, the turbine/generator breakers were opened, thereby tripping the turbine on overspeed 14 seconds after the reactor scram. Because of the reported occurrences, the staff recommended that a scram without a turbine/generator trip (SWATT) be evaluated as a new issue to address the potential for a SWATT and its associated effects.1378
One of the possible effects of a SWATT is the overcooling of the primary system which can cause recriticality, pressurized thermal shock (PTS), or both. PTS, which is caused by an overcooling transient followed by or concurrent with high reactor vessel pressure, was previously addressed in Issue A-49. This led to the PTS rule which established new requirements for all PWRs.
Besides possibly causing vessel failure from PTS, an overcooling transient can induce recriticality in the reactor. Recriticality may be caused by an overcooling transient that occurs toward core end-of-life when the boron concentration is low, and thus the moderator temperature reactivity coefficient is most negative. With a rare combination of operating conditions, this could produce a pressure spike in the primary system and lead to the same conditions of a PTS event.
Another possible effect of a SWATT is overspeed of the turbine which can cause turbine missiles and damage the integrity of the onsite power distribution systems by large fluctuations in generator voltage and frequency. Turbine missiles were addressed in Issue A-37 which was determined to have little safety significance. Because of its close relationship with Issue A-37, this issue was used to reexamine the analysis of Issue A-37.
A SWATT is significant to the safety of LWRs because the possible effects from this event can lead to a core-melt. These effects include the following: (1) recriticality and PTS, due to a primary system overcooling event; (2) turbine missiles, from overspeed of the turbine; and (3) loss of core cooling and inventory makeup, due to other events caused by the SWATT, such as SGTRs. In addition, plants without MSIVs do not have the capability to quickly stop the supply of steam to the turbine. MSIVs provide increased protection from the consequences of a SWATT and, thus, plants without them are potentially more vulnerable.
The possible solution involves three activities designed to improve the reliability that a turbine/generator trip occurs following a reactor trip: (1) installation of redundant and diverse emergency trip systems; (2) improved periodic inspection, testing and maintenance of turbine/generator trip systems; and (3) clarification of operating procedures and training for operating personnel.
It was assumed that 71 operating plants would be affected by this issue: 47 PWRs and 24 BWRs with average remaining lives of 27.7 and 25.2 years, respectively. Of these 71 plants, only the three PWRs at Oconee with average remaining lives of 26, 27, and 27 years do not have MSIVs.
The core-melt frequency from a SWATT is related to the failure to isolate the steam supply to the turbine. Isolation of the turbine helps to either prevent or mitigate the consequences of the following: (1) primary system overcooling, which could directly cause core damage through either PTS or recriticality; (2) turbine overspeed, which could indirectly cause core damage through turbine missiles and electrical power fluctuations; and (3) SGTRs, which could directly cause core damage through the effects of a small break LOCA at the steam generator. The events that lead to these consequences and their associated failure probabilities are described below.
The initiating event for the SWATT is a reactor transient, which causes the reactor to trip. The event that leads to this issue, however, is the subsequent turbine trip failure. Using data from the 1990 AEOD Report (NUREG-1272,1472 Vol. 5, No. 1), the average frequency of a reactor trip from 1986 to 1990 was calculated to be 2.8/RY and 2.5/RY for PWRs and BWRs, respectively. The probability of turbine trip failure, given a reactor trip and including contributions from operator error and equipment failure, was estimated1376 to be 1.7 x 10-3 per demand.
Possible responses1377 to a SWATT are as follows: (1) the generator can be tripped, which causes the turbine to trip on an overspeed condition; (2) the turbine can be isolated by closing the MSIVs in those plants that have them; and (3) the turbine can be locally tripped by manual operator action. Each of these responses has failure probabilities due to equipment faults and operator actions. Because of insufficient data, the following conservative assumptions were made:
(1) All equipment faults were assigned a failure probability of 0.05/demand;
(2) The failure probability for an operator to perform a corrective manual action for both the generator trip and the MSIV closure was assigned a value of 0.03/demand;
(3) The failure probability for locally tripping the turbine was assigned a value of 0.1/demand, because this trip is located in the turbine building.
The resulting conditional probabilities for failure to isolate the steam supply to the turbine were calculated by summing each plant state resulting in a non-isolable condition.1378 The resulting event sequences for turbine trip failure, feedwater availability, and core cooling and inventory makeup were then used to investigate the possible effects of a SWATT. These effects include PTS, turbine missiles, and SGTRs. The effects of a SWATT for those reactors without MSIVs (the three units at Oconee) were investigated using the same models by assuming the failure probability for MSIV closure is equal to 1.
An important assumption regarding the previous calculation of the core-melt frequency from PTS was that a SWATT is functionally similar to a large steam line break (LSLB) occurring downstream of the MSIVs. A study1376 by INEL showed that, for B&W reactors, a graph of the reactor vessel downcomer temperature as a function of time for the failed turbine trip scenario was very close to that calculated for a LSLB in developing the PTS rule. The same conclusion would be expected for other reactor types. Therefore, the probability of reactor vessel failure due to a SWATT was estimated to be equal to that calculated for a LSLB. Even if it were assumed that this probability was an order of magnitude higher for a SWATT than for the most limiting steam line break (which is not the largest size break), the core-melt frequencies would still be extremely low.
In addition, to simplify the analysis, recriticality events were neglected from further study because these events were not likely to generate a pressure pulse significant enough to cause reactor vessel failure.1379 In support of this conclusion, others have estimated that the heat addition to the primary coolant system after a recriticality would occur gradually, thus reducing the risk of PTS.1376
The base case core-melt frequencies for PTS, turbine missiles, and SGTRs are summarized as follows:
|PWRs:||Event||With MSIVs||Without MSIVs|
|PTS||8.8 x 10-12/RY||2.9 x 10-10/RY|
|Turbine Missiles||1.1 x 10-9/RY||3.5 x 10-8/RY|
|SGTR||1.4 x 10-8/RY||4.5 x 10-7/RY|
|Total:||1.5 x 10-8/RY||4.9 x 10-7/RY|
|PTS||8.1 x 10-12/RY|
|Turbine Missiles||9.4 x 10-10/RY|
|Total:||9.4 x 10-10/RY|
The effect of the possible solution would be to improve the reliability that the turbine will trip on demand after a reactor trip. The solution is also expected to result in reductions in the probabilities of other events in the turbine trip failure event tree, because of enhanced training and awareness of operators to respond to a SWATT. To reflect the above, it was assumed that the solution will reduce the probability of a failure of the turbine to trip to one-half the base case value. In addition, human error probabilities were assumed to be reduced by 50% as a result of enhanced training and operator awareness. These values were substituted into the turbine trip event tree to calculate revised turbine trip failure frequencies and turbine overspeed trip failure frequencies. No changes were made to the feedwater and core cooling/makeup event trees. The adjusted case turbine and overspeed trip failure frequencies were then substituted into the calculations described previously to calculate the adjusted case core-melt frequencies. These are summarized as follows:
|PWRs:||Event||With MSIVs||Without MSIVs|
|PTS||2.0 x 10-12/RY||1.4 x 10-10/RY|
|Turbine Missiles||2.7 x 10-10/RY||1.8 x 10-8/RY|
|SGTR||3.2 x 10-9/RY||2.2 x 10-7/RY|
|Total:||3.5 x 10-9/RY||2.4 x 10-7/RY|
|PTS||1.9 x 10-12/RY|
|Turbine Missiles||2.4 x 10-10/RY|
|Total:||2.4 x 10-10/RY|
From the base case and adjusted case calculations above, the core-melt frequency reduction was estimated to be 1.2 x 10-8/RY, 2.5 x 10-7/RY, and 7 x 10-10/RY for PWRs with MSIVs, PWRs without MSIVs, and BWRs, respectively.
The next step in this analysis combined the accident sequence frequencies with their corresponding containment failure probabilities. These probabilities were taken from NUREG/CR-280064 for similar accident sequences, where possible. Thus, it was observed that the SGTR sequence is similar to sequence S3H in the Oconee-3 PRA. It was more difficult to match sequences involving PTS and turbine missiles with those for Oconee-3 and Grand Gulf-1. Therefore, a conservative approach was taken that incorporated the most damaging (and highest offsite doses) containment failure modes. The most damaging containment failure mode sequence for Oconee-3 is S3H; the most damaging containment failure mode sequence for Grand Gulf-1 is T23PQE.
The base case and adjusted case affected public risk was determined by combining the accident sequence frequencies, containment failure probabilities, and dose consequences for the release categories corresponding to the appropriate containment failure modes. The dose consequence estimates developed by PNL64 for each release category were used in this analysis. The following is a summary of the potential risk reduction:
PWRs: With MSIVs: 3.2 x 10-2 man-rem/RY
PWRs: Without MSIVs: 6.5 x 10-1 man-rem/RY
BWRs: 2.0 x 10-3/RY
For the 68 affected plants with MSIVs, the total public risk reduction is [(44)(27.7(3.2 x 10-2) + (24)(25.2)(2 x 10-3)] man-rem or 40 man-rem. For the 3 affected plants without MSIVs, the potential risk reduction is [(26 + 27 + 27) (6.5 x 10-1)] man-rem or 52 man-rem.
Industry Cost: The total industry cost was estimated to be $24M and included $16M for implementation of the possible solution and $7.8M for operation and maintenance. For those plants without MSIVs, the total cost was estimated to be $1.2M.1377
NRC Cost: Development and implementation of the solution were estimated to cost $55,000 and $810,000, respectively. Review of operation and maintenance was estimated to cost another $2.6M. Thus, the total cost was estimated to be $3.5M.
Total Cost: For plants with MSIVs, the total cost associated with the possible solution is $(24 + 3.5)M or $27.5M. For plants without MSIVs, the estimated cost was $1.2M.
(1) With MSIVs: Based on a public risk reduction of 40 man-rem and a cost of $27.5M, the value/impact score was given by:
(2) Without MSIVs: Based on a public risk reduction of 52 man-rem and a cost of $1.2M, the value/impact score was given by:
This issue had a greater impact on plants without MSIVs than those with MSIVs. In either case, the potential risk reduction was small and the issue was given a low priority ranking (see Appendix C) in March 1992. Consideration of a 20-year license renewal period did not change the priority of the issue.1564 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 $23,255/man-rem which placed the issue in the DROP category.