Teachin g Notes
Operationa l Reacto r Safet y Course
Lecture : 1 1 – Probabilisti c Safet y A nalysis
Objective :
The objectiv e of this lecture it to gain an app r eciation of p r o b abilistic ris k asses sm e nts in design and operation. It is not intended for students to becom e PRA experts but to understand what PRA is in term s of event an d fault trees. NRC us e of risk inform ed decision m a king is discussed as are safety goals.
Ke y Point s t o Brin g Out:
Slid e num be r Point s
3-4 Contrast determ inistic and probabi listic analysis - key points are that PRAs do not assume system s function with 100% reliability as in dete rm inistic an alys is and include s hum an erro r. Determ inistic analysis attempts to bound system hum an failures.
5 The purpose of showing this s lide is to point out the com plex nature of nuclear plants. PRAs n eed to m odel all system s that have any influence on the core or heat removal.
6 Review PSA applications noting th at the trend in regulatory and design applications - explain wh at risk inform ed perform a nce based regulation m eans.
7 Review history of PRA and why it is a good tool for decision m a king and design to increase unde rstanding of plant system s and their failure modes.
8-10 Review the basis of the "Farm e r Paper" as a m e thodology for judging acceptable risk results by using credible and incredible acciden t s.
11-13 Review what goes into a PSA - event and fault trees and how calculated. Summ arize hi storical studies - technology has been in use since 19 75 with the Ras m ussen Reactor Safety Study.
14-15 Review results of the Reactor Safety Study - W a s h 1400 - key is that de term inistic analy s is m i ssed more lik ely an d signif i can t risks.
16-17 Begin explaining how a PRA is perf orm e d with an explanation of sequence of events diagram s -(judgem e nt of wha t initiating events m a y occur and what m i ght happen) - event and fault trees.
17 Explain the different types of PRAs - Levels and their functions. Be sure to include a discussion of uncertainties. Re m i nd students that th e cons equence ana l ysis us es de term inistic a n alysis to pr edict results under varying co nditions po s t ulated. One can not do a PRA without a great deal of determ inis tic analysis of plant performance to which probabilities of com ponent failures are assigned for a judgem e nt of the likeliho od of the event sequence.
19-29 Go over a simple exam ple of a pum ping system to develop system event trees and fault trees.
30 Point out where data comes from fo r failure rates of com pone nts - industry and NRC maintain data ba ses for com p onent failures - plant experience data is very im portant for credible PRAs.
31 One of the key questions about th e use of PRAs is uncertainty. Most PRAs carry uncertainties thro ugh the analysis that addresses distribution of failure rates and uncer tainties in o u tcom es of system perform a nce . Key point is uncer tain ty m ay dom i nate re sults. Confidence levels are ty pical ly given based on sam pling m ethodologies.
32-36 Explain how the risk is calculated by com bining the frequency, tim e interval and conseq uences that are summ ed to asses s ov erall risk of the f a cility. For a com p lete risk assessm ent (all th ree levels ), one must integr a t e the p l an t, containm ent and site m odel to assess risk to the public.
37 An i m portant slide that shows the co mplete plan t risk m odel - all three levels.
38-39 Shown is an exam ple of an ATW S e vent tree with quantification of frequency and assessm ent of conse quence. Im portant to go over to have students appreciate the tool . Note the inputs needed to perform the analysis as shown in slide 39. Hum a n perform ance response is controversial - use of Swain' s handbook is the standard reference based on actions and tim e needed.
40-44 Explain definitions of core dam a ge fr equency and large early
release frequency which are key indicators of saf e ty of the plant. Point out typical risk contributors in a PW R noting external events are as im por tant as internal (com ponent) failures. Discuss def i nition of condition al f ailure prob ability.
45-47 Review shutdown risk -which ar e not negligible particularly during refueling. T h is m a y be surpri sing to students - 43% of the total r i s k .
48-54 These slid es review the NRC' s view of risk assessm ents post the Ras m ussen report - key is they walked away from the technology for m a ny years even though the i ndustry continued to perform PRAs on their plants - review results presented to observe variability in core dam a ge fr equency. Mention that NRC did eventually require Individual Plan t Evaluations - Level 1 PRAs on all plants M ention th e use of m ean vs. Median results of PRA and confidence lim its.
55-58 Review NRC safety goals and base s of subsidiary quantitative safety goals.
59 A good summ ary of quantitative safe ty goals as applied to th e three levels of PRAs. Don' t forg et to include uncertainties as a factor.
60-64 Using the United Kingdom' s risk tolerance curve, the US evolution of the use of PRA in decision m a king should be explained. Slide 64 shows how the current Comm ission is applying risk in regulation.
MIT OpenCourseWare http://ocw.mit.edu
22.091 / 22.903 Nuclear Reactor Safety
Spring 200 8
For informati on about citing these mate rials or our Terms of Use, visit: http://ocw.mit.edu/term s .