The ADS-IDAC Dynamic PSA Platform with Dynamically Linked System Fault Trees

Transcription

1 The ADS-IDAC Dynamic PSA Platform with Dynamically Linked System Fault Trees Mihai Diaconeasa Center for Reliability and Resilience Engineering The B. John Garrick Institute for the Risk Sciences University of California, Los Angeles Presented at ANS PSA September 26 th, 2017

2 ANS PSA 2017 Outline Dynamic PRA ADS-IDAC Platform Architecture HCL methodology Dynamically linked FTs Example: PWR plant trip with loss of reactor coolant pump seal injection and cooling due to electrical fault 2

3 Discrete Dynamic Event Tree 3

4 History of ADS-IDAC COSIMO (JRC Ispra, Italy) Cacciabue, et al IDA Cognitive Model & Error Taxonomy (UMD) Smidts, Shen Model Based HRA (UMD) Chang IDAC Model (UMD) Chang, Mosleh ADS-IDAC 3.0 (UMD) Li Today ADS-IDAC 2.0 (UMD) Coyne ADS RELAP (UMD) Chang ADS-IDAC 4.0 (UCLA) Accident Dynamics Simulator (UMD) Hsueh Dynamic Event Tree Analysis Method (MIT) Acosta, Siu 4

5 Introduction to ADS-IDAC Objective Predicting operators performance under accident situations. Initiating Event System and Crew responses Coyne, K. A. (2009). A Predictive Model of Nuclear Power Plant Crew Decision-Making And Performance In A Dynamic Simulation Environment. University of Maryland. 5

6 Modules of the ADS-IDAC Platform User User Interface Module Scheduler Module Crew Module Indicator Module System Module (RELAP5/MOD 3.3) System Reliability Module 6

7 Thermal-Hydraulic Model: RELAP5/MOD3.3 RELAP5 Thermal-Hydraulic Engine Recognized thermal hydraulic analysis tool Existing RELAP plant models can be readily adopted to the ADS-IDAC environment Plant models require some modifications Interactive controls and instrumentation Realistic representation of plant systems, protective features and controls The current three-loop PWR plant model includes over: 250 hydraulic components 100 heat structures 1500 control systems Coyne, K. A. (2009). A Predictive Model of Nuclear Power Plant Crew Decision-Making And Performance In A Dynamic Simulation Environment. University of Maryland. 7

8 High Level View of the IDAC Model Top-down attention control Reasoning Module Mental State Information pre-processing Decision-making Influence Input System, Other Crew Members, Other External Sources Action execution Li, Y. (2013). Modeling and Simulation of Operator Knowledge-Based Behavior. University of Maryland. 8

9 Team Model of Individual Operators Consultant (Shift Technical Advisor) Decision Maker (Shift Supervisor) System, Other External Sources Action Taker (Reactor Operator, Field Operator) 9

10 Cognitive Infrastructure of the Reasoning Module I:Basic Concept Unit II: Composed Concept Unit III: Semantic Sentence Steam_Generator SG_A Loop Loop_A SG_A_pressure A SG_A_pressure decrease pressure RCS Tave decrease Power < Load Load increase RCS borate Control rods in Turbine Load increase Steam dump increase SG PORV open SG Safety Valve open decrease Knowledge Base Cold coolant injection Safety Injection on Main steam line break Semantic Base I: Basic Concept Unit II: Composed Concept Unit III: Semantic Sentence Reasoning Mental Representation Mental Representation Thought Threads pool Control Panel Item (Parameter/Component State/Alarm) Control Panel ID--Ontology Concept ID Effective situational statement pointer Readings: ->Recent reading ->History reading: -reading n -reading 2 -reading 1 Situational Statements Examples n: PRZ level is low 3: AFW_Pump turn_on 2: MS_Flow Increase 1: RCS_Tavgdecrease Investigation item 1: PZR pressure decreases Investigation item 3: Load > Power Thought Threads Pool Investigation item 2: Investigation item 5 RCS Temp decreases Investigation item 4 Investigation item 6: Turbine Load Increases Investigation item 7 Investigation item 8 Li, Y. (2013). Modeling and Simulation of Operator Knowledge-Based Behavior. University of Maryland. 10

11 Surrogates-PSFs-Manifestations propagation paths PSF Quantitative Assessment Mechanism Modeling Dynamic/Static List Attention X D Problem-solving style X D Prioritization X D Information use X D Time load X D Task load X D Expertise X X S Passive information load X D Information load X D System criticality X D Task complexity X D Stress X D Fatigue X D Li, Y. (2013). Modeling and Simulation of Operator Knowledge-Based Behavior. University of Maryland. 11

12 Hybrid Causal Logic Methodology C. WANG, Hybrid Causal Logic Methodology for Risk Assessment. 2007, University of Maryland: College Park, Md. 12

13 Binary Decision Diagram representations of FTs C. WANG, Hybrid Causal Logic Methodology for Risk Assessment. 2007, University of Maryland: College Park, Md. 13

14 Modules of the ADS-IDAC 4.0 Platform RELAP5/MOD3.3 Crew Module System Module Scheduler Module System Reliability Module Control Panel Module HCL Module 14

15 Dynamic Linking of FTs for Frontline and Support Systems 15

16 Robinson PWR LOMFW due to Electrical Fault Most dominant sequence, CCDP = 2.5x10-4 (64% of total internal events CCDP): Reduced power to RCP B Reactor trip succeeds AFW succeeds PORVs close Loss of RCP seal cooling/injection Operators fail to trip the RCPs Subsequent RCP seal LOCA Automatic SI succeeds Operators successfully cooldown the RCS Operators fail to initiate shutdown cooling mode of the residual heat removal system High-/low-pressure recirculation fails Final Precursor Analysis of H.B. Robinson Electrical Fault Causes Fire and Subsequent Reactor Trip with a Loss of Rcp Seal Injection and Cooling. 2010, US NRC 16

17 Modified Robinson Loss of Seal Cooling FT 17

18 Modified Robinson RCP B Seal Injection FT 18

19 Modified Robinson Emergency Bus E2 FT 19

20 Generated Discrete Dynamic Event Tree 20

21 Pressurizer pressure vs time for ES 4 and ES 5 Loss of off-site power to vital Bus E2 EDG B fails on demand Operator s manual actuation of the ECCS based on the action step in the EOPs 21

22 Thank You This work was funded through a Research Grant (NRC Grant HQ G-0013) by the U.S. Nuclear Regulatory Commission.