Resolution of Generic Safety Issues: Item A-22: PWR Main Steam Line Break - Core, Reactor Vessel, and Containment Building Response ( NUREG-0933, Main Report with Supplements 1–34 )
Several aspects of the MSLB analyses as currently provided by applicants and accepted by the NRC have been questioned. The concerns derive principally from Issues 1 and 15b of NUREG-0138.32 Issue 1 in NUREG-013832 questioned credit for the operation of nonsafety-grade equipment as a backup for single failures in safety-grade equipment following an MSLB. NUREG-03712 states that the consequences of this issue should be determined by the results of sensitivity studies. These analyses are to consider the action or inaction of nonsafety-grade equipment and operator action during the course of the accident. Issue 15b in NUREG-013832 was concerned with the mechanical response of the pressure vessel following an MSLB. Specifically, vessel integrity following the temperature transient that an MSLB would produce was questioned.
If a major steam line break were to occur, a blowdown of some fraction of the secondary system water inventory would result. In evaluating this accident, the staff considered the effects on the core, on primary system components, on safety-related components in the vicinity of the steam lines, and on the containment structure. In addition, the staff evaluated the radiological consequences associated with such a postulated break.
The evaluation of the consequences of major steam line break, either inside or outside of the containment, is based on the continuous rapid depressurization of one steam generator and the isolation of the main feedwater system when necessary. As part of this evaluation, it is assumed that a single active failure occurs in the systems required to mitigate the consequences of such events.
In the event of a steam line break inside containment, it is necessary to isolate the main feedwater to the steam generator associated with the failed line to preclude overpressurizing the containment and to limit the reactivity transient. If the single active failure postulated for this accident is the failure of the appropriate safety-grade main feedwater isolation valve to function, then credit is taken for closing the nonsafety-grade main feedwater control valve or tripping the feedwater pump in that line. Issue 1 in NUREG-013832 centered on the reliability of such nonsafety-grade equipment.
Issue 15b in NUREG-013832 centered on the pressure and temperature conditions of the reactor pressure vessel following an MSLB. In the event of such a major steam line break, the initial rapid depressurization of one or more steam generators would result in cooling and depressurizing the primary system. An engineered safety features actuation signal generated by this event would initiate the isolation of all the steam generators in addition to other safety actions, including initiation of safety injection, which will bring primary pressure back up. Depending on the break location and the single active failure assumed in evaluating this event, there could be continued cooldown of the primary system because of the blowdown of a nonisolable steam generator. Because reactor vessel material toughness decreases with increased total irradiation, this leads to the concern that at some future time pressurization at low temperatures could result in failure of the reactor vessel. This issue is now included in USI A-49, "Pressurized Thermal Shock," which considers the integrity of reactor pressure vessels following all types of overcooling transients.
Since in a PWR the main steam lines inside containment are seismically qualified, the probability that an earthquake would cause a break in the portion inside containment is very small. An earthquake-induced failure of a steamline in a nuclear power plant has not occurred and we estimate the frequency to be less than 10-4/RY. The frequency of steam line ruptures not induced by earthquakes is higher, 10-3/RY.
To exceed the design bases, the feedwater isolation valve associated with the steam generator which is blowing down must fail and the feedwater must continue flowing. There are six feedwater isolation valve closure failures in the NPRDS files. These valves are tested at least annually. Thus, since every PWR has at least two steam generators, in the current 340 PWR years of experience there have been at least 680 valve tests. We thus estimate the isolation valve failure to close probability to be at most 10-2/demand. This is probably quite conservative, even though not all failures may have been reported to NPRDS, since the actual number of the valve tests is almost certainly in excess of the assumption used here. Since these valves are seismically qualified, we will assume that the failure probability is the same with and without an earthquake.
The safety injection signal (generated by low pressurizer pressure, high differential pressure between steam lines, high containment pressure, etc.) generally will close the feedwater control valves, close the feedwater pump discharge valves, and trip the main feed pumps. Since these actions are diverse from feedwater isolation valve closure but are not safety grade, we will assume that these "backup" actions are at least 90% reliable (i.e., have a failure probability of less than 0.1 per demand) if not subject to an earthquake. However, operator action can be effective in shutting off feedwater if the automatic trip fails.
Improvements in the response of operators are being made as specified in TMI Action Plan Item I.C.1.48 Licensees have been asked to perform analyses that consider the occurrences of multiple failures, consequential failures, and operator errors which, if unmitigated, could lead to inadequate core cooling. In addition, these analyses are being carried out far enough in time to assure that all relevant thermal/hydraulic/neutronic phenomena are identified and to address possible failures and operator errors during the long-term cooling phase. These analyses are being performed as part of TMI Action Plan Item I.C.1 and will serve as the bases for Emergency Procedure Guidelines for Transients and Accidents including MSLB. These emergency procedure guidelines will be used as a basis for the development of plant-specific emergency procedures. We estimate that operator action would be successful in halting feedwater flow at least 90% of the time, i.e., failure of operator action is 0.1.
The result of a failure to shut off feedwater is a high pressure in the containment. We will assume that the containment fails (although this is by no means certain). The effect on the core would be to fail some fuel cladding, which
as a limiting condition we assume is bounded by the failures resulting from a design basis (i.e., mitigated) LOCA. In the case of earthquakes, the proper operation of the nonseismically qualified equipment is less assured. However, we will assume that the probability of not stopping the feedwater flow either through automatic closure of the valves or tripping of the feedwater pumps is 0.1. Tripping of the pumps could occur automatically as a result of the earthquake or through operator action.
Thus, the frequency (F) of exceeding the containment design limits following a main steam line break is the sum of the frequencies with and without an earthquake.
F < [(10-4)(10-2)(0.1) + (10-3)(10-2)(0.1)(0.1)]/RY
< 2 x 10-7/RY
The release can be bounded by a PWR-8 release which is a design-basis LOCA with failure of the containment to isolate.
R < 7.5 x 104 man-rem
The proposed solution is to add backup safety-grade feedwater isolation valves. The cost to the licensee has been estimated to be $203,000 for a forward-fit or $318,000 for a backfit. There are 43 operating PWRs with 350 RY of experience and 36 PWRs under construction. Total initial cost to the industry is then about $21M.
NRC costs are estimated to be 6 staff-months of generic work plus 2 staff-weeks per plant or about $350,000.
Therefore, the total cost for the solution to this issue is ($21M + $0.35M) or $21.35M.
If we assume a 40-year plant life, the total remaining operating life of 79 PWRs is 2,810 years. Thus, the total risk reduction for this issue is less than 42 man-rem and the value/impact score is given by:
The value/impact score above assumes a 90% probability that the nonsafety-grade closure of the feedwater control valves, feedwater pump discharge valves, or trip of the main feedwater pumps will stop main feedwater flow. On some plants, the feedwater pump discharge valves do not close automatically on a safety injection signal. Nevertheless, the 95% figure should still conservatively bound the probability of at least one of the remaining actions shutting off feedwater.
More importantly, some other PWRs do not have separate feedwater isolation valves for which the value/impact score above assumed a 10-2/demand failure rate. These plants, however, invariably have feedwater pump discharge valves which close automatically on safety injection signal. Thus, our previous assumptions of a 10% combined failure rate of all of these actions is overly conservative for such plants. If instead we assume a 10% maximum failure rate for the control valve closure and a 10% failure rate for the feedwater pump discharge valve closure, the recalculated value/impact score upper limit rises to 20 man-rem/$M.
Based on the discussion above, this issue need not be continued as a separate generic issue. In addition, the improvements in procedures as a result of Item I.C.1 of NUREG-073798 will reduce the risk further. Therefore, the first of the two concerns associated with this issue (the failure of containment following an MSLB) is of such low safety significance that it need not be considered further. The second concern (overcooling) will be thoroughly addressed by USI A-49, "Pressurized Thermal Shock." Thus, this issue has been placed in the DROP category.