ML20136J327
| ML20136J327 | |
| Person / Time | |
|---|---|
| Site: | Indian Point  |
| Issue date: | 07/25/1985 |
| From: | Pal A POWER AUTHORITY OF THE STATE OF NEW YORK (NEW YORK |
| To: | |
| Shared Package | |
| ML20134E988 | List: |
| References | |
| FOIA-85-497 NUDOCS 8508200682 | |
| Download: ML20136J327 (5) | |
Text
..
e SEISMIC / DYNAMIC FRAGILITY AND SYSTEMS INTERACTION STUDY AT INDIAN POINT - 3 Arun C. Pal New York Power Authority 123 Main Street White Plains, NY 10601.
ABSTRACT Systems Interaction (SI) Study for Indian Point Nuclear Power Plant, Unit No. 3 was undertaken per advice of the Advisory Committee on Reactor Safeguards (ACRS) primarily to determine the effect of failure of non safety-related compo-nents and systems on safety-related items and plant safety itself.
In the present paper it is shown that a predominant number (95%+) of " postulated" SI's attributed to a seismic event equivalent to SSE, have extremely low probabilities of oc-currence if seismic / dynamic fragilities of the components /
systems were taken into consideration.
INTRODUCTION Interactions between systems may be intentionally pro-vided for in the design for proper functioning of the plant or unintended. USI A-17 is concerned with the later, speci-fically, non-safety-related-to-safety-related-interactions due to the functional, spatial or induced human error coupling.
l Relevant to present discussion are the spatially coupled sys-tems interactions (SI) caused by external events like 1) earth-quake up to and including SSE, ii) . pipe failure (whip), iii) physical impact (missiles), iv) flooding (tank failure), v) tornado depressurization or overpressurization, iv) LOCA or main steam line break and vii) fire. Of these, earthquake of maximum acceleration equivalent to SSE (0.15g) caused almost all SI's (1). Interestingly, contributions ' to two of the major risks from nuclear power plants, e.g., core-melt and I off-site radioactive release, by seismic events of same magnitude are only 2% and 4% respectively (2). These facts do not contradict each other. Because, SI study was performed on the purely deterministic premises that systems, structures or components not designed seismically shall fail during an SSE 8508200682 PDR FOIA 850725 PEDRD85-497 PDR I
L and so cause postulated SI's, irrespective of seismic safety margins inherent in them. On the contrary, for safety anal-ysis or probabilistic risk assessment analysis (PRA), prob-abilistic estimates of ground motions are coupled with stochastically determined structural reliability of systems and components. Any event with probability of occurrence less than 1 x 10- per reactor year is not considered contributing to overall risk.
METHODOLOGY FOR POSTULATING SI All spatial interactions caused by earthquake are source /
target type impact. Sources can be completely detached from supporting structures and travel some distance in space before collision (missiles) or partially detached - pipe support failure. As a result of the impact the targets "jkil" i.e., become overstressed (beyond yield stress), inoperable etc. Pipes with adjacent unidirectional restraints (hangers) were assumed to fail because displacements of these supports overstressed the pipes beyond code specified allowables.
Interaction influence zones were established by engineering judgement alone.
RESOLUTION OF "FEASIBIBLE" INTERACTIONS The three classical methods of seismically qualifying an equipment or component are 1) Analysis,11) Testing and iii)
Demonstration. For resolution of SI's or " closing" an "open" interaction, analyses and testing or combination of them were utilized. Demonstration or behavior of same (or similar) l components or equipment in past earthquake was prohibited on the ground that it would violate licensing criteria of the plant. Conservatively, the stresses in reanalysis were limited to 0.9 fy. Testing was performed by applying an I
equivalent static load to sources like lighting fixtures, PA system and conduits. Testing was adopted only when unknown material properties made an analysis impossible. 2 difications to many components were done mostly in the form of additional supports to piping systems and redundant-hold-down devices to other sources.
LESSONS LEARNED AND GUIDANCE FOR FUTUEE SI STUDIES Following is a summary of lessons learned and what could be I
done in future SI studies regarding seismically induced spatial interactions.
- 1. Probabilistic not Deterministic Analysis:
Prof. Newmark stated Page II
9 3
It is aidsable to apply the theory ;
of probability and optimization tech- '
niques... . The traditional deter-ministic disguise will do less well in earthquake engineering.
This is so true for the uncertainties like (a) although an earthquake affects the entire plant simultaneously, whether -
a single SSE can cause failure of all non-cat-1 components or equipment is unknown. Also, an SSE is nothing but the maximum rock acceleration (4) with a probability of exceedance of 0.5% in 50 years. This itself calls for probabilistic anal- '
ysis (b) magnitude of rock / ground acceleration required to -
fail a component "C" is not known precisely; because component
~
"C" is not defined completely. In PRA analysis a log normal curve with median (acceleration) and standard deviation ~
Ummdom variation parameter) is given (c) impact between source and target must be determined probabilistically since in the three dimensional space defined to be interaction soundry, source-target contact is not definite (d) perfect mathematical modelling of targets is not possible just as is not possible for sources; probabilistic approach is the only -
way to arrive at a reliable model and (e) role of the failed component or target in the system to which it belongs and ]
that of the system to safety connot be determined definitely also.
Assigning probabilities to items (a) thru (e) , even conser-vatively, leads to total probability of each of the postibited spatial SI's to <<1 x 10-7 Of course, if the argument is that assigning numbers to life safety probability is not an acceptable approach, almost every industrial facility may require to be closed down! -
- 2. Elasto-plaatic not Elastic Analysis:
Failure analyses of sources and targets should not be on the same basis as design analyses of com- ,
ponents. Since material for most sources and targets alike are structural steel, with ductility ,
ratios nearly 20, local yielding must be allowed as long as collapse is not evident. Until further k results are available through extensive analysis or experimental research of the cyclic hysterical behavior of structural steel, the following stress levels are considered useful (5) for SI studies.
Bending: 170 F s (r Axial Compression: 170F g.4 '
Shear: c. 6 Fy Bolts and Welds: 17e A15c AL&owAEES Expansion Anchors:(Pg875.p.(vf Gh 4 J, o Pap III M
'l I
- 3. Utilization of Experience Data:
Behavior of components or equipment in past earth-quakes must be considered as " demonstration"'and utilized on equal footing as testing. This is now acceptable to the ULO for equipment qualification (6) in operating pl,;ets. During reanalysis at Indian Point SI-stu;y most " span evaluations" of piping system failed because of the currently advocated philosophy that restraints must be added until inertia stresses are below allowables.
Experience data show quite contrary (7) . During future spatial interaction study this deserves con-sideration!
The extremely conservative criterion in postulating interactions was to combine SSE and LOCA sinultaneously.
Probabilistic estimates of this extreme load conbination may prove that probability is very negligible. In future SI-studies this consideration should be eliminated.
Note: The technical opinions expressed in this article are those of the Author only and those of his employer.
l i
I Page IV
REFERENCES
- 1. Systems Interaction Study Report - Indian Point 3 Eoasco Services Inc., 1983,
- 2. Indian Point Probabilistic Safety Study Pickard, Lowe and Garrick, 1982.
- 3. Newmark, N.M. and Rosenblenth. E. - Fundamentals of Earthquake Engineering", Prentice Hall, 1971.
- 4. Algermissen, S.T. and Perkins, D.M. "A Probabilistic Estimate of Maximum Acceleration in Rock in the Contignuous United States" 1976 .
- 5. Scismically Induced Systems Interaction Program, Pacific Gas and Electric Company - Diablo Canyon Units 1&2 1984.
- 6. Seismic Qualification Utilities Group - Pilot Program ,
Report Vol. I & II, Prepared by EQE, Inc., 1982. '
- 7. Smith, P.D.; Swan, S.W. and Yanev, P.I. " Experience Data on the Performance of Piping Systems in Earthquakes" ;
1985. l Page V
- ____ _.