Resolution of Generic Safety Issues: Task I.G: Preoperational and Low-Power Testing (Rev. 3) ( NUREG-0933, Main Report with Supplements 1–34 )
The objectives of this task were to: (1) increase the capability of shift crews to operate facilities in a safe and competent manner by assuring that training for plant changes and off-normal events was conducted; and (2) review the comprehensiveness of test programs. Near-term operating license facilities were required to develop and implement intensified training exercises during the low-power testing programs
ITEM I.G.1: TRAINING REQUIREMENTS DESCRIPTION
This TMI Action Plan48 item called for new OLs to conduct a set of low power tests to achieve the objectives of Task I.G. These tests were to be determined on a case-by-case basis.
This item was clarified in NUREG-073798 and requirements were issued.
ITEM I.G.2: SCOPE OF TEST PROGRAM
The major thrust of TMI Action Plan48 Task I.G was to use the preoperational and startup test programs as a training exercise for operating crews. In contrast to this, Item I.G.2 called for a more comprehensive test program to search for anomalies in a plant's response to a transient. This issue was suggested independently by the Kemeny Commission,175 the Rogovin Commission,161 the ACRS,176 and the TMI Operations Team.177
The safety significance of this issue was in the early discovery of anomalies or unanticipated plant behavior. The TMI-2 accident was the most well-known example, but other less severe examples, such as the core-annulus water level decoupling at Oyster Creek, have taken place.
When a plant responds to a transient in an anomalous or unanticipated manner, the result may be an accident caused directly by the new phenomena, or by the surprise or confusion on the part of the operators. The latter is probably the more likely of the two.
The nature of the solution to this issue was implicit in its definition - an augmented test program. However, relatively little had been written concerning the nature and extent of this program. NUREG-O66O48 merely called for the NRC to develop a program. Recommendations177 made by an OIE team investigating TMI-2 were more specific: detailed review of all unscheduled transients during the first year as well as review of the preoperational and startup tests.
There was a spectrum of possible test programs ranging from the existing program to programs that would take years to implement. Morever, it might not have been necessary for each plant to perform each test. In addition, there was a large amount of operating experience data from which information could be gathered.
At the time this issue was initially evaluated, transients occurred at an approximate rate of 10/RY; however, most of these were relatively routine (e.g., turbine trip) and were thus unlikely to produce unpleasant surprises. In any case, existing startup programs were expected to cover them adequately. Therefore, attention was focused on transients that were rare but nevertheless frequent enough to be considered "anticipated operational occurrences." EPRI NP-801178 contains a history of the transients that were actually experienced during operation. Based on judgment, transients that were candidates for suspicion of anomalous behavior were selected.
|PWR Transients||Frequency (RY-1)|
|Hi/Lo Pressurizer Pressure||0.10|
|Pressurizer Safety or Relief-Valve Opening||0.02|
|Loss of RCS Flow||0.04|
|Close All MSIVs||0.05|
|Sudden Opening of Secondary Relief Valves||0.06|
|Loss of Component Cooling||0.01|
|Loss of Service Water System||0.01|
|BWR Transients||Frequency (RY-1)|
|Pressure Regulator Fails Open||0.29|
|Pressure Regulator Fails Closed||0.14|
|Inadvertent Opening of S/RV||0.20|
|Trip One Recirculation Pump||0.02|
|Trip All Recirculation Pumps||0.06|
At the time of this evaluation, reactor experience totaled 565 RY: 225 BWR-years and 340 PWR-years.179 Thus, it was estimated that about 270 of the listed transients had occurred, some of which had indeed illustrated the need for corrective measures. Unfortunately, it was not practical to use the computerized data banks to search for "anomalous behavior." Once again, judgment was used.
At least four transients with anomalous response had occurred (Davis-Besse, Three Mile Island, Oyster Creek, Pilgrim) and were widely known. If a more thorough review of operating experience had been made, more would have been discovered. It was estimated that perhaps 10 transients had shown some sort of unanticipated phenomenon. However, the number of interest was the number of phenomena left to be discovered. With a history of about 270 transients of interest, anomalous events were not expected to be very common. Moreover, the discoveries that had been made led to measures intended to prevent future problems.
It was estimated that anomalous or unanticipated behavior could be expected at a rate of about 5 events in 565 RY (i.e., half the estimated historical rate) or about 10-2/RY. This number was an "educated guess" that the actual number of events that had occurred was higher than the four events listed, but would be lower in the future because the experience had been used to correct the problems.
Most anomalous transients have no consequences in the sense of releasing radioactivity. Based on the experience of TMI (one event in perhaps 10), it was assumed that one event in 10 will result in core damage (extensive cladding failure) and one event in 100 will result in a core-melt with a significant release. The former was approximated with a PWR-9 or BWR-5 Category event and the latter with a PWR-7 or BWR-4 event.
It was assumed that an augmented startup program would be 50% effective in discovering and correcting problems. The total risk reduction associated with this issue was 2.58 x 104 man-rem, based on 252 man-rem for 36 PWRs and 2.56 x 104 man-rem for 21 BWRs.
Industry Cost: As stated previously, there was a spectrum of possible test programs. It was assumed that the test program would average 2 weeks/plant. At $300,000/day for replacement power (which would dominate the cost), this was $4.2M/plant. The 2-week average estimate was based on the assumption that not every plant would perform every test. In many cases, the first product of a given product line would be subjected to a great deal of testing that would apply to all plants of the same design, or testing could be shared within a product line by some other plan. Therefore, the total industry cost was $239.4M.
NRC Cost: It was assumed that 5 staff-years would be required to develop guidelines and approve generic plans with one staff-month/plant of post-test review. With 57 OLs, (36 PWRs and 21 BWRs), this cost was about $1M.
Total Cost: The total industry and NRC cost associated with the possible solution was $(239.4 + 1)M or approximately $240M.
Based on an estimated public risk reduction of 2.58 x 104 man-rem and a cost of $240M for a possible solution, the value/impact score was given by:
The frequency estimates did not rest upon firm bases. This was not surprising because, like any other program where the goal is discovery, if good bases were available for estimates of effectiveness, the tests would not be necessary. Nevertheless, an attempt was made to put bounds on the estimates. The frequency of core damage was not likely to be uncertain by more than a factor of 10. If the true frequency were a factor of 10 higher, about 6 core-damage accidents would have occurred. If it were a factor of 10 lower, the TMI-2 accident would have a probability on the order of 0.05.
However, the core-melt frequency was subject to more uncertainty. It was assumed that the frequency of core-melt was one-tenth of that for core-damage. It was assumed that this figure could be either a factor of 5 higher (every second TMI-like event a core-melt) or a factor of 5 lower (one core-melt in 50 core-damage events).
Assuming that the public dose estimates were uncertain to a factor of 5 and the costs to a factor of 5, then the value/impact score would have a range from 3 x 100 to 4 x 103 man-rem/$M.
The averted costs of cleanup were not considered in the value/impact score. If such costs ($0.25M/RY) were included, the value/impact score would be significantly higher, but not enough to justify a higher priority.
Based on the value/impact score and the associated public risk reduction, this issue was given a medium priority ranking. With revisions to SRP11 Section 14 and the OIE Manual, the issue was RESOLVED with new requirements.654