MUPRA: a Few Observations and Technical Challenges

MUPRA: A Few Observations and Technical Challenges

N. Siu and D. Hudson Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission

Presented at Special Session #2: MUPRA Advances, Issues, Impediments and Promise

ANS International Topical Meeting on Probabilistic Safety Assessment (PSA 2017) Pittsburgh, PA; September 24-28, 2017 Multi-Unit Precursor Events Date Plant Description Date Plant Description 6/22/82 Quad Cities LOOP, Maintenance 8/14/03 6 Sites LOOP, NE Blackout

8/11/83 Salem LOOP, Clogged screens 6/14/04 Palo Verde LOOP, Offsite fault

7/26/84 Susquehanna SBO, Bkr mis-aligned 9/25/04 St. Lucie LOOP, Hurricane

5/17/85 Turkey Point LOOP, Brush fires 5/20/06 Catawba LOOP, Switchyard fault 7/23/87 Calvert Cliffs LOOP, Offsite tree 3/26/09 Sequoyah LOOP, Bus fault 3/20/90 Vogtle LOOP, Truck hit support 4/16/11 Surry LOOP, Tornado 8/24/92 Turkey Point LOOP, Hurricane 4/27/11 Browns Ferry LOOP, Winds/tornadoes 12/31/92 Sequoyah LOOP, Switchyard fault 8/23/11 North Anna LOOP, Earthquake 10/12/93 Beaver Valley LOOP, Offsite fault 3/31/13 ANO LOOP/Trip, Load drop Trip, Foreign material in 6/28/96 LaSalle SW Tunnel 4/17/13 LaSalle LOOP, Lightning 6/29/96 Prairie Island LOOP, High winds 5/25/14 Millstone LOOP, Offsite fault

2 Some Other Old But Illustrative Events Event Notes Event Notes Large TB explosion and fire, smoke Multi-unit cable fire; both units Browns Ferry Narora => U1/U2 MCR abandonment; U1 tripped. (BWR, 1975) (PHWR, 1993) SBO, U2 cold shutdown. Electrical cable fire; U1 station Extensive, extended flooding. Roads Greifswald blackout (SBO), U2 tripped. Cooper (one evacuation route) blocked. (VVER, 1975) Shutdown used x-tie with U2 (BWR, 1993) Leakage through penetrations seals. U3 maintenance activity leads to Large storm causes LOOP (U2/U4), Oconee large lake flood in common Turbine Blayais flooding of U1/U2, affected site (PWR, 1976) Building (TB); lost TD-EFW. (PWR, 1999) access. Large steam release in TB; also Large seawater flood in common Doel Mihama entered portions of control and U1/U2 TB; both units tripped. (PWR, 1979) (PWR, 2004) intermediate buildings. Electrical cable fire (multiple Large water leak lifts U2 TB floor, Armenia locations), smoke spread to U1 Nogent also floods U1 TB, enters ESW (VVER, 1982) MCR, secondary explosions and (PWR, 2006) gallery through penetrations, CCW fire; U1 SBO, U2 tripped. pump room through drains Heavy rainfall floods Unit 1 TB and Large TB explosion and fire, partial Chernobyl Surry U2 ESGR through manhole and roof collapse. Only U2 affected. (RBMK, 1991) (PWR, 2006) conduit penetrations

3 3 Operational Experience Observations

• Many events caused by spatial (internal and external) hazards • Some events – involve single units, and/or – were terminated before serious plant impact, but mechanisms are relevant to MUPRA • All observed mechanisms and scenarios are relevant to (and should, in principle, be captured by) single-unit PRA • Assumed defenses (e.g., seals, equipment qualification) and other screening factors may need re-examination

4 L3PRA Overview

• Project objectives – Develop state-of-practice Level 3 PRA – Extract new insights – Enhance NRC PRA staff capability and expertise – Demonstrate technical feasibility and cost of new Level 3 PRAs • Planned project scope – All hazards – All operational modes – All major sources • Plant: Vogtle 1&2 (special thanks to SNC for support)

5 L3PRA Integrated Site Risk Approach*

• Logical combination of important single source scenarios (assumed symmetry) • Inter-source dependencies (5 types) • Parallel, independent search for scenarios

6 *See ML17255A580 for details. New Tools to Aid Searches

• Operational experience reports and other sources can spur (and temper) imagination • Database is massive and unstructured; how to efficiently identify, access, and analyze data? • Recently completed scoping study (long-term research) – Currently available technology (content analytics) can efficiently identify important target documents – Increased subject matter expert involvement or more powerful machine learning tools would help in development of “industrial strength” applications

7 ADDITIONAL SLIDES Multi-Unit Precursor Events Date Plant Type CCDP Date Plant Type CCDP 6/22/82 Quad Cities LOOP 1E-4 6 Sites (NE 4E-6 to 8/14/03 LOOP Blackout) 3E-5 8/11/83 Salem LOOP 1E-4 7/26/84 Susquehanna SBO 2E-4 6/14/04 Palo Verde LOOP 9E-6/4E-5 5/17/85 Turkey Point LOOP 4E-5 9/25/04 St. Lucie LOOP 1E-5 7/23/87 Calvert Cliffs LOOP 5E-4 5/20/06 Catawba LOOP 9E-5/6E-5 3/20/90 Vogtle LOOP 9E-4 3/26/09 Sequoyah LOOP 4E-6 8/24/92 Turkey Point LOOP 2E-4 4/16/11 Surry LOOP 9E-5/7E-5

12/31/92 Sequoyah LOOP 2E-4 4/27/11 Browns Ferry LOOP 1E-5

10/12/93 Beaver Valley LOOP 6E-5 8/23/11 North Anna LOOP 3E-4/6E-5

6/28/96 LaSalle Trip 7E-6 3/31/13 ANO LOOP/Trip Yellow 6/29/96 Prairie Island LOOP 5E-5 4/17/13 LaSalle LOOP 1E-5

5/25/14 Millstone LOOP 2E-5/1E-5

9 Inter-Source Dependencies