ML20212B214
| ML20212B214 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 12/17/1986 |
| From: | Papanic G YANKEE ATOMIC ELECTRIC CO. |
| To: | Mckenna E Office of Nuclear Reactor Regulation |
| References | |
| TASK-03-02, TASK-03-04.A, TASK-3-2, TASK-3-4.A, TASK-RR DCC-86-206, FYR-86-120, NUDOCS 8612290198 | |
| Download: ML20212B214 (23) | |
Text
.
YANKEE ATOMIC ELECTRIC COMPANY Tpho"'(6 ")*' -oo TWX 710-380-7619 a
d ' ~ f
^i.,
}
1671 Worcester Road, Framingham, Massachusetts 01701 DCC 86-206
,' I/ u Y K EE.'
~....~. /
December 17, 1986 FYR 86-120
' United States Nuclear Regulatory Commission Washington, DC 20555 Attention:
Ms. Eileen M. McKenna, Project Manager PWR Project Directorate No. 1 Division of PWR Licensing - A
References:
(a) License No. DPR-3 (Docket No. 50-29)
(b) Letter, USNRC to YAEC, dated November 3, 1986 (c) Letter, YAEC to USNRC, dated July 9, 1984 (d) Letter, YAEC to USNRC, dated September 5, 1986 (e) Letter, USNRC to YAEC, dated November 25, 1986
Subject:
Responte to Requests for Additional Information, SEP Topics III-2 and III-4.A
Dear Ms. McKenna:
Enclosed please find Yankee Atomic Electric Company's responses to the requests for additional information presented in Reference (b) and Reference (e).
The responses incorporate the comments and additional information requested by the NRC during the November 21 meeting and telephone conversation. The responses to Reference (b) are provided in Enclosure 1.
The responses to Reference (e) are provided in Enclosure 2.
As a result of the evaluation performed for this work, the west wall of Diesel Cenerator Cubicle No. 3 (designated as wall D11053) will be upgraded for a 134 mph straight wind /121 mph tornado.
Yankee Atomic Electric Company's commitment to perform the plant modifications described herein, as well as those in Reference (d), are contingent upon acceptance of the enclosed material and that previously provided (References (c) and (d)).
8612290198 861217 PDR ADOCK 05000029
\\
PDR P
h
c United States Nuclear Regulatory Commission December 17, 1986 Attention:
Ms. Eileen M. McKenna Page 2 FYR 06-120 We believe that the enclosed information, along with that provided in References (c) and (d), is sufficient to resolve SEP Topics III-2 and III-4.A.
Very truly yours, YANKEE ATOMIC ELECTRIC COMPANY G
G. Papa e, Jr.
Senior Project Engineer Licensing GP/jmk Enclosures
i ENCLOSURE 1 Page 1 of 10 l
RESPONSE TO LETTER, E. M. MCKENNA (NRC) TO G. PAPANIC, JR. (YAEC), DATED NOVEMBER 3, 1986 (NYR 86-242)
QUESTION 1
~With respect to the vent stack analysis, provide a discussion of the mathematical model and analytical procedures. Also indicate tae controlling stresses and allowables for the vent stack components listed in the response to Question 2 of Enclosure 1 in Reference 1.
RESPONSE
The mathematical model and analytical procedures used in the vent stack wind analysis are discussed in detail in Reference 1.
Further details, including This controlling stresses and allowables are provided in th'e calculation.
calculation was made available for review by the NRC during the November 21 meeting.
QUESTION 2 Provide technical justification for not expecting dynamic effects of tornado wind on the vent stack at wind velocities over 60 mph. Also, provide assurance that resonance due to the dynamic effects of the wind does not occur and control at wind speeds below the 10-5 (165 mph) wind speed.
RESPONSE
Dynamic effects of winds on circular cross-sectional structures in addition to the dynamic pressure, can include ficxural vibration and vortex shedding.
The vent stack is subject to two types of wind loadings: tornado and straight Typically, the incoming wind flow which may lead to flexural vibration wind.
and vortex shedding is assumed to be laminar (smooth and uniform).
In the case of a tornado, the wind field is of a rapidly varied nature in both the 5134R
ENCLOSURE 1 Page 2 of 10 direction and magnitude and of short duration. Therefore, it is concluded that for tornadoes consideration of flexural vibration and vortex shedding may be neglected. Therefore, this discussion will be limited to straight winds.
As previously noted, the incoming flow capable of producing resonant flexural vibration or vortex shedding is assumed to be laminar. The physical location of the vent stack with respect to other plant structures and the surrounding terrain preclude laminar flow. Also, the flow at wind speeds of concern are turbulent in nature. This turbulent flow would tend to inhibit any resonance effects on the stack. Additionally, an analysis with the conservative assumption of laminar flow is presented below:
Vortex Shedding For vortex shedding to be of concern, the vortex shedding frequency must be in resonance with the natural frequency of the stack. Typically, under the action of vortices shed in the wake of the stack, the stack will be driven periodically, but this driving will elicit only small responses unless the Strouhal frequency of the alternating pressure loading approaches the stack frequency. The natural frequency of the vent stack was determined to be 4.6 Hz.
Using information from References 11 and 12, the critical wind speed for resonant vibration can be determined as a function of the stack frequency, diameter of the stack and the Strouhal number. Reference 12 provides a relationship between the Reynolds and Strouhal numbers.
The critical wind speed for resonant vibration of the stack was determined 6
to be 63 mph with associated Reynolds and Strouhal numbers of 2.9 x 10 and 0.25, respectively.
Per Reference 12, vortex shedding at Reynolds 6
numbers between 200,000 and 4 x 10 is of a random nature. Therefore, it is concluded that a critical resonant condition due to vortex shedding is unlikely for the vent stack.
5134R
ENCLOSURE 1 Page 3 of 10 This conclusion is supported by Reference 11 which notes that experience indicates that periodic vortex shedding is inhibited by the natural turbulence of the airstream at the higher wind velocities, and that it is not likely to occur at velocities greater than 60 mph.
Flexural Vibration In addition to transverse swaying oscillations produced by vortex shedding, a steel stack may also be subject to flexural vibration in the circular cross-sectional plane as a result of vortex shedding.
The frequency of the lowest mode of ovaling vibration in the circular stack was determined to be 6.8 Hz per Reference 11.
The critical wind velocity for ovaling resonant vibration of the stack was determined to be about 22 mph per Reference 12 based upon the 6.8 Hz value.
The compatible Reynolds 6
and Strouhal numbers were 1 x 10 and 0.52, respectively. At this low critical velocity any stresses on the stack due to ovaling would be insignificant.
Reference 11 also suggests that, to guard against ovaling vibrations, the thickness of the stack should not be less than 1/250 (.004) of the stack diameter. The stack satisfies this criteria with a thickness of.25" and a diameter of 5'-0".
Based on the above information, flexural vibrations of the vent stack are not a concern.
A review of meteorological data collected from 1977 to the present was performed to determine the upper range of wind speeds observed at the plant.
The data base consists of a 15-minute average wind speed for the first 15-minute period of each hour. This data is maintained on computer tape and periodically updated.
51342
ENCLOSURE 1 Page 4 of 10 There were a total of 18 occurrences when this value exceeded 22.5 mph, with the maximum observed 15-minute average of 30.1 mph.
The original strip charts were then reviewed at these occurrences. The strip charts provide a permanent, continuous record of peak winds. The corresponding peak winds ranged from approximately 34 mph to 59 mph. Wind speed records before 1977 were not reviewed since the information is not on the data base computer tape.
Since the 15-minute average values were used as the indicator for strip chart review, it is possible that higher peak winds are contained in some 15-minute averages below 22.5 mph. Also, since only the first 15-minute period of each hour is in the data base, only one-out-of-four periods per hour has been sampled. Given these above points, it is concluded 'that the vent stack has experienced peak winds in the vicinity of 60 mph. This is consistent with information in Reference 13 which gives an expected fastest mile wind speed of
~
61 mph at an annual probability of 5 x 10 (20 years mean recurren:e inte rval).
The vent stack has not experienced any degradation in structural integrity since the start of plant operation in 1960.
Also, the stresses in the vent stack are low when subjected to a 165 mph to rnado. Stresses in the stack when subjected to the 165 mph tornado are at 0.6% and 17% of the allowables for compression and bending, respectively.
Based on the above discussions regarding vortex shedding, flexural vibration, meteorological data, and state of stress, the following can be concluded:
1.
Resonance of the vent stack due to vortex shedding is not expected.
2.
Flexural vibrations (ovaling) are not a concern.
3.
The vent stack has been subjected to wind speeds in the range which could cause resonant conditions, and no degradation in the integrity of the stack has been observed.
5134R
F-i ENCLOSURE 1 Page 5 of_10 4.
The analysis of the stack and supporting elements shows that there is considerable margin to allowable limita when subjected to a 165 mph tornado.
Therefore, resonance due to. dynamic wind effects will not significantly affect the primary vent stack.
QUESTION 3 Regarding the sample calculations for masonry walls, discuss, and justify a flexural stress allowable higher than the design specification (23 psi vs.
14 psi).
RESPONSE
For the purposes of the tornado cost / benefit analysis, an ultimate capacity of the masonry walls was required. This was calculated by increasing the design allowable for tension normal to the bed joint (14 psi) by a factor of 1.67.
This factor is as recommended in " Recommended Guidelines for Reassessment of Safety-Related Concrete Masonry Walls" (Reference 2).
The increased value has a factor of safety of 2.8 to failure. The above values are based on the specified mortar strength (m ) f 750 psi.
0 Testing of existing block and mortar has been performed. The test results show a mortar strength per ACI 531-79 of 1,386 psi.' Using a 1.3 increase in allowable mortar tension-(per SRP 3.8.4 for extreme environmental loadings) results in an allowable of 24 psi.
Based on the above discussion, 23 psi is a conservative value for determining the ultimate capacities of the masonry walls.
The ultimate strength for tension parallel to the bed joint was calculated in a similar manner. The design value (41 psi) was based on specified mortar strength with an increase factor of 1.5 per SRP 3.8.4.
The ultimate strength 5134R
ENCLOSURE 1 Page 6 of 10 used in the cost / benefit analysis (43 psi) is based on an increase factor of 1.67 per the " Recommended Guidelines" (Reference 2).
The 43 psi value has a safety factor of 3.8.
Using the test results of the existing block and mortar and the 1.5 increase factor per SRP 3.8.4 results in an allowable mortar tension parallel to the bed joint of 55.8 psi.
Based on the above discussion, 43 psi is a conservative value for use in determining the ultimate capacity for the masonry walls.
QUESTION 4 The Licensee must provide a quantitative assessment of the successive failures of the Diesel Generator Building's interior walls D11043 and D11054. Provide the geometric characteristics of any equipment or obstruction that ;ay shield the walls from direct wind loading.
RESPONSE
As discussed in the response to Question 10 of Enclosure 2 of Reference 1, Walls D11043 and D11054 are not expected to fail during a wind / tornado event.
To provide further assurance that Diesel Generators 1 and 2 will remain operational until the failure of Diesel Generator Building (DGB) Wall DlX1, the west wall of Diesel Generator Cubicle No. 3 (Wall D11053) will be upgraded.
The failure of wall DlX1 is assumed to disable all power and control cables to and from Emergency 480 V ac Buses 1, 2, and 3 and their associated loads.
This includes power from Emergency Diesels 1, 2, and 3.
Any modifications to Wall D11053 will be designed in accordance with DCD-648-6-1 (Reference 3) for a straight wind speed of 134 mph or tornado wind speed of 121 mph. These are the failure speeds for DGB Wall D1XI. Any modifications required for Wall D11053 will be implemented in 1989.
5134R
q ENCLOSURE 1 Page 7 of 10 Figure 1 shows a plan view of the DGB with walls indicated.
QUESTION 5 Indicate whether all modifications to the PAB walls have been implemented.
RESPONSE
Modifications to the PAB north wall have been completed. Modifications to the upper level PAB west wall are currently scheduled to be implemented in 1989.
QUESTION 6 With respect to the response to Question 4 of Enclosure 2 in Reference 1, provide the wind speed capacity of the west wall of the west staircase in the Turbine Building.
RESPONSE
The west wall of the Turbine Building southwest staircase is a 12" thick reinforced concrete shear wall. This wall is designed to act as part of the Turbine Building seismic load path.
The wind capacity of the wall is greater than 200 mph.
QUESTION 7 For the tornado missile investigation, please provide the following information:
a.
Indicate whether the Fujita-scale tornado classification was used in the analysis.
If not, provide justification.
It is noted that the EPRI Report proposed a modified tornado classification to the Fujita scale, but insufficient documentation was provided to support this proposal [2].
5134R
ENCLOSURE 1 Page 8 of 10 b.
Discuss the sensitivity of the possible variation of the tornado wind speed near the ground.
It is noted that the reductions in tornado wind speed near the ground due to surface friction effects are not sufficiently documented in the EPRI Study [2].
RESPONSE
a.
In the tornado missile impact probability analysis (Reference 4), the missile impact and entrance probabilities from three separate analyses were adjusted for differences in the number of missiles and target areas between the analyses and the Yankee plant. The three analyses utilized were EPRI NP-768 (Reference 5), EPRI NP-2005 (Reference 6), and a site-specific study for Seabrock Station (Reference 7).
In the two EPRI studies, a modified version of the Fujita F-scale was used in the case study calculations. Table 2.3 from Reference 8 (see ) gives a comparison between the two tornado intensity scales: F (Fujita) and F' (EPRI). The differences in the two scales are graphically shown in Figure 2.
The two scales are similar through the lower intensities, with the F'-scale giving lower wind speeds at the higher tornado intensities. The 10' upper 95% confidence level tornado wind speed for the Yankee analysis of 165 mph is within the intensity 3 category for both scales, a
A review of EPRI NP-768 shows that about 75% of' the total tornado missile risk is from F'2 and F'3 tornadoes. Since the differences in the two scales (F and F') are not as dramatic at these intensities as at higher intensities, the impact of the tornado scale differences on the missile impact analysis is not significant.
A similar review of EPRI-2005 showed that about 90% of the total tornado missile risk is also f rom the F'2 and F'3 tornadoes.
5134R
r ENCLOSURE 1 Page 9 of 10 Also attached is a copy of a letter (Reference 9) from the author of the EPRI Reports to Dr. Simiu, the NRC technical reviewer of the reports (see ).
This letter contains additional information which further c.upports the modified F' scale.
In the site-specific Seabrook Station study (Reference 7) the Fujita F-scale was used in the calculations.
b.
In both EPRI studies (References 5 and 6) and in the Seabrook Station study (Reference 7), the tornado wind speed at ground level was reduced to 75% of the reference level speed (33 feet above grade).
Due to boundary layer effects', it is reasonable to assume that there is some reduction in tornado wind speed below the reference level.
In Appendix J to the Seabrook Station SSER 3 (Reference 7), the NRC consultant. Dr. Simiu, states that even though there is no data in the literature to support the reduction, that the assumptions used in the report appear nevertheless to be reasonable.
Also, under Item 3 in the attached letter to Dr. Simiu (Attachment 2), the EPRI Report author gives an opinion that the impact of the near ground wind speed reduction would not be significant due to the dominant role of the tornado strike probability.
REFERENCES 1.
Letter, G. Papanic, Jr. (YAEC) to E. McKenna (NRC (FYR 86-084), Response to Requests for Additional Information, SEP Topics III-2 and III-4.A, dated September 5, 1986.
2.
Recommended Guidelines for the Reassessment of Safety-Related Concrete Masonry Walls, prepared by Owners and Engineering Firms Informal Group on Concrete Masonry Walls, October 6, 1980.
5134R
ENCLOSURE 1 Page 10 of 10 3.
Charles T. Main, Inc., Document No. DCD-648-6-1, Revision 0, Structural Design Criteria for Evaluation and Modification of Existing Masonry Block Walls.
4.
Cost-Benefit Evaluation for SEP Topic III-2, Wind and Tornado loadings and SEP Topic III-4A, Tornado Missiles for the Yankee Nuclear Power Station, Revision 1, December 1984.
5.
Electric Power Research Institute, Tornado Missile Risk Analysis, EPRI NP-768, May 1978.
6.
Electric Power Research Institute, Tornado Missile Simulation and Design Methodology, Volume 1, EPRI NP-2005, August 1981.
7.
Seabrook Nuclear Power Plant. Tornado Missile Analysis, Applied Research Associates, Inc., Fina) Report C569, September 1983.
8.
Technical Evaluation of Report, " Tornado Missile Simulation and Design Methodology (EPRI NP-2005) by Emil Simiu, April 1983.
9.
Letter from Twisdale to Simiu, November 16, 1983.
- 10. Safety Evaluation Report Related to the Operation of Seabrook Station, USNRC, NUREG-0896, Supplement No. 3, July 1985.
- 11. Structural Engineering Handbook Second Edition, Section 26, Chimneys, Zor and Chu, McGraw-Hill, 1979.
- 12. Wind Effects on Structures, Second Edition, Simiu and Scanlon, Wiley &
Sons, 1985.
- 13. Enclosure 2 of Letter, Crutchfield (USNRC) to Kay (YAEC), Yankee Rowe -
SEP Topic II-2.A. dated December 17, 1980.
5134R
e J
.--.-.e.
= =...
==.-.... -.
l o
s ao' o as.c l
g g.. g.
...,,,..g...-..
~h-
=
. *- -Q-~~~v-.
\\'
T E.f * & COWC.
LOLufv'N$'@ AS
/
w stanAGr mios s,
.-D g.: s s, 4.: y
...,,.7..y. ABCvt) 3
\\i a
).
,.6a3c gre att.gr tp.ger n,;
~ 4
_L g,. s a t.T.. *:. u.L 5
- / i.
i
- a.. s..-. 3' r.,.,
r. e
..re %
s a
f
..n
.= iQl-.l.,-
t.
. =~ ~. - - a,
.. Suor - d.
'
- n.c. -t* a c;,/.;..'a.-.
. 3:
, s ms l
fj\\
1 q
l==.-l fl :I;8 P'I.
n-
- , o..tr
';y,_--c con cr.
-I l
g,
'[r e
it i-
'e
,y i
.?
- m i
-4 pi :.
=-
. -i
.2.
se
/
t i
- I l
' I g: i:
! 1.6l.
[
cTl' (
+
G*,'
I r4 eec *gr!
s--- M =
.k i-v, e -% ?
~
p
.=*7
~.
2-i 1
\\. l!
W ** *"Piy4 Ce'9,& -
1 l
en
'. m y s t u"*
v ss ',.
r-l uut. :s i
T
! A..Q. S A O 4,,.-
l; 4 i
T
- d o.'
3-
- m.. - !.;.
,,> i L,
9 I.a i y
E--
El i
[1 h
e,e 2D p _v:
a j
i
,e w 3 t.
f
,,,,,, nx,:
... g
,o m,, ~ m. i... urm mana sr.n.e.
m c
- i. ;;r
/ n:-, y
- **' 7.L
.o g
-m q"
p Eg 3s-ij Mi
". y
? m,.r g r.ra !!l J
.e OfI045 h s-.-q M&;! e DIIO53 I
i k*A 4dj 4
1 3 p p*
x t
011o52 ie E r ;
-,..-u
~~
i 1
I
) (
34 T 1 a' ; A i lt I
1
-c
\\
.q' ;_czy 2
, -en~.:gIa W**s I s=r t
Iw s
> fnf:
w= ~
p
- g..,
a n e
,~,
g v
g,.
DIV5 DtV2 ogy g J4 1st t
FIGURE 1 l
DGB Plan View I
I
___-.._.o......
I I
l
.I j
' l
}-
.{
1 t i
y.
i
! l l
l 6
6 t
4 i
Y l
{
4 300
_..d __--
o r
I t
i l
[ I t
.O I
EPRI I
1 4
I A
I
( g 2:
O i
a.
T a
i
___.__IG, l
FUJITA W
W 4
4 8A f
t i
I
<m l
zo I
l9 a: z I
os s :s i
o.
oi l
t 200
-q -. -
i 1
l
.i I
i i
()
ID l
9 e
I l
FIGURE 2 0
TORNADO l, SCALE COMPAR. ISON I
i
.o I
I FUJITA vs. EPRIi 4
i 100 d>
s ---- -
il 1 - --
-- - - t -
-- - - 4 1
1
.O l
.O ii
.,I
- +
p i
1 i
f 4D I
t t
i j
l O
L--
F0 F1 F2 F3 F4 F5 TORNADO SCALE i
i
Attzenmtnt 1
~
Pags 1 of 1 Wind Velocity Ranges for Various Tornado Intensities 7able 2.3 (in miles per hour)
Ranges Proposed Ranges Originally Tornado in the Study Proposed by Fujita (T'-scale)
(F-scale)
Intensity
<72 PO 40-73 73-112 73-102 T1 113-157 103-134 T2 156-206 T3 135-167 207-260 T4 166-208 261-318 F3 209-276*
T6 277-300**
319-380 Ter regions A, E, and C.
For region D, the upper limit is 225 mph.
For region C the upper limit
- For regions A and E.
is 285 mph.
e 0
3
~
hC Page 1 of 6 q
A
, InC.
Engineering and AppliedScience N
i 16 November 1983 s
s Dr. Emil Simiu U.S. Department of Commerce v'\\
National Bureau of Standards
\\
Washington, DC 20234
\\
Dear EmiI:
I have had a chance to review your comments on EPRI NP-2005, " Tornado Missile Simulation and Design Methodology."
Since your comments and discussions raised some very interesting and useful questions, I thoajht I would share our ideas with you on a few of them:
1.
Corrected and Prior Occurrence Rates, p. 7-10 Your comment (p. 9) that the corrections for direct, random encounter, and path length intensity variation are not significant for several F-scale categories is correct.
However, in view of the indirect nature of the data base, we believe that each of these corrections are importar.t concepts and should remain distinct and separate.
For example, in site specific studies, one would 1
introduce random encounter corrections that reflect actualland-use within the subregion. The motivation for filtering the data is simply that the recorded data are not reflective of true tornado occurrence rates or F-scale intensity distributions.
Also, one should realize that the prior rates in your Table 2.2 (wtiich is Table I-20 in NP 2005) are actually reported rates that have been adjusted for reporting trend and unreported events.
2.
Wind Velocities of F-Rated Tornadoes, p.10-12 Your comments on the data supporting the proposed F' windspeeds neglected the data we used to update Fujita's original windspeeds.
First, we used Fujita's windspeeds as prior judgmental evidence (in a Bayesian scheme), consistent with the role his expert opinion played in the development of these windspeeds.
Second, we updated these windspeeds by considering new data, which was' ahilable since -
the 1965-1970 development of the F-scale classification methodology.
This updating is based principally on the Dames and Moore engineering analysis of windspeeds required to damage buildings to certain f ailure levels.
The Bayesian analysis was done with two assumptions on upperbound windspeed ranges, uniform and linear.
In order to assess which, if either, of these updated windspeed distributions correlated best with available windspeed damage data, additional comparisons were made.
These included:
hOUTHEAST DIVISION E.t417 OROFESSIONAL COURT a RALEIGH, NORTH CAROLINA 27609 E W19) 876M118
h Attach = nt 2 y
Pags 2 of 6 Dr. Emil Simiu 3
16 November 1983 Page 2 s
t \\,
\\.
s(
(1) Eagleman's wind tunnel tests a s k
h (2) Mehta's et a1. ' evaluations of the Lubbock tornado
(
p s-
")
(3) Mehta's analysis of the Xenia tornado
[
(4) Photograrrmetric analysis of the Xenia tornado i
s
\\
t i
Ecch of these data sources tended to support the linear updated F' windspeeds. Your coments concentrated on the quality of this data and/or,our analysis of it.
Specifically, our comments on these 4 data sdurces are: {
(
(1) Therearecleprlyanumberofduestionsregardingthesemodel studies.
However, the wind tunnel results seem reasonably consistent with windsoeedf-damage information from other more
'A recent data.
w e
(2) Your interpretation of near ground as 10 f t above grade is g-solely your own and seemsatoo low fort tne structures Mehta i
\\
ev aluated. A 30 f t average height seems more reasonable.
The O
f act that Lubbock was classified F4 has nothing to do with s
s' actual windspeeds that may have b w present in that storm.
(Further, are you aware that Fujita's w'ginal photo of F5 damage used for F-scale classification was actually taken from i
Lubbock, a storm which he had classified as F4.
What does this say about the need for analysis of F-scale classification
,5 d
errors?)
t (3) We disagree with your conclusion that one of our arguments for F' winds rests solely on difference of opinion of F2 or F3 i
damage. ' First of all, Abbey's figure does not distinguish F2 from F3 damage but merely' lumped the two for purposes of y
mapping. By comparing damage and F-scale classification photos, we concluded that F3 is consistent with the integrated damage of Xenia high school.
(One should also not.e from Abbey's map, that there was F4-F5 damage adjacent to the school; hence we concluded F3 was probably the best choice.}
A y' -
Regarding the use of the 133 mph chimney f ailure, we felt '. hat the assumptions in Mehta's analysis gave more credibility to the lower bound estimate.
In any case, if one uses the 167 mph val,ue, the updated mean computed according to Eq. 7 in Ref. 6 is 160 inTh (v,s 151 mph), which is still well within the updated F3 windspeed rangNf 135 to 168 mph.
Hence, we reject your
(
conclusion of blas.in o,ur analysis.
's s-(4), The coment here 'was a ' lack of justification that the windspeeds analyzed independently by Golden and Fujita were in an F5 damage zone.
The Boyd footage was taken as the tornade was moving northeastward through the Arrowhead subdivision,
)
\\
I
)l i~
s
,g
+
Attachm:nt 2 Page 3 of 6 Dr. Emil Simiu
- s 16 November 1983 Page 3 where many houses were destroyed or literally blown astay.
In the film the huge vortex was characterized by an evolving series of 2-6 suction vortices.
Both, Golden's peak speed of 220 mph and Fujita's peak ' speed of 25$ mph occurred in or on a revolving suction vortex.
In addition, the Abbey map shows the Arrowhead area with houses blown away as FS, which is consistent with the Fujita classification photos.
Hence, we assumed t. hat the photogrannetric analysis was within an F5 damage area.
In summary, the proposed updating of F-scale windspeeds was performed using Fujita's scale as a prior source of information, a series of engineering calculations, and checks with observed damage and photogrammetric evidence.
In our opinion, this provides a much firmer basis for windspeed correlation than a set of windspeeds proposed in 1969 by a meteorologist with no such data.
Further, one only has to look at damage caused by measured winds of 100-150 mph k
and see the inconsistencies in the F-scale.
In f act, F5 type damage can easily be caused by winaspeeds in the 200 mph range.1 I realize that the simple study that we did 6 years ago is not the final answer to tornado-windspeed damage correlation, but it seems that the available evidence does justify its use.
One only has to turn the question around and ask what evidence justifies the original F-scale windspeeds.2 3.
Windfield Modeling The main comment here concerns the potential for unconservatism in the vertical profile of the rotational velocity component. Yet, no reason is offered as to why the profile selected is not reasonable.
Hence, there is not much of a basis for discussion, except that this profile is consistent when compared with other tornado wind profiles.
For example, it is comparable to the TRW model, which solves the slip and no slip boundary condition equations, as well as 1 For example, Ramey's analysis (JSD, Sept. 1980) of windspeeds in the Birmingham tornado in F5 damage zones also found our F' scale windspeeds to be much more consistent with the F-scale damage descriptions.
2 Recognizing the need for a separate research project on this issue, I have tried to get NSF funding to calibrate the F-scale, compiling all the data collected over the past 10-15 years, performing calculations, and developing maximum likelihood probability distributions of F-scale winds 4
for engineering use.
However, the politics of the situation are too much to fight as Fujita is invariably a reviewer and apparently doesn't want any one to tamper with "his" scale.
.~
.,~
Attachmznt 2 Page 4 of 6 Dr. Emil Simf u 16 November 1983 Page 4 the empirical Fujita model. Further, it is very conservative when compared to the profile used for ordinary winds (as illustrated in attached figure). As f ar as impact on results of even a straight profile, I don't thinig' it would be too significant for risk-type studies, because of the dominant role of the tornado strike probability.
4.
Missile Aerodynamics We did not coin the term " cross-flow theory," as you imply. For example, see Horner, Fluid Dynamic Drag, for information on -the application and use of the cross flow principle. -
5.
Equations of Motion We believe your comments that our statement on the " lack of complete aerodynamic data is a hindrance to the use of 6-D models needs
' qualification" has missed a main point in the trajectory modeling approach. First, one should carefully read the paragraph in which the sentence occurs and previous material in the section on missile transport. The proposed missile spectrum for probabilistic analysis of tornado missile risk contains 26 shapes, of which the 6 NRC missiles are a subset. The JPL aerodynamic study addressed only the 6 NRC shapes and.hence there is not complete aerodynamic data for a general. set of missiles. Further, one should realize that 6-0 calculations are very time consuming and are not suited for probabilistic codes. Finally, and most importantly, we demonstrated the power of the R0-60 model to simulate the mean and' variance of 60 transport and also pointed out the potential inadequacy of 30 transport models, which is what most studies were using at that l
time.
I 6.
Comparison to Palo Verde UHS PRA l
You comment tht the Palo Verde procedure is more conservative than the' EPRI approach in several areas.
Yet I have found it virtually impossible to assess this document because of the tortuous mathematics and method of presentation that obscure the fundamental assumptions, most of which are never stated directly.
In my opinion, the entire document seems an exercise in mathematical frivolity based on weak assumptions or nonexistent data.
Also, and,
more importantly, the data sources used and quoted (particularly my i
J own work) are of ten used incorrectly.
For example, the calcula-tional procedure is based on a local surf ace density of potential missiles.
However, the authors have no consistent data base to I
estimate this density.
They use Twisdale [4] survey coupled data 7 ft2 that with the area of a totally hypothetical plant (the 2.5x10 i
they refer to). This is a luotcrous bit of logic, which indicates
Page 5 of 6 Dr. Dnil Simiu 16 November 1983 Page 5 some of the sloppiness in the details of the work. Alst, in the injection model, they appear to use only aerodynamics for a cylin-drical missile in these calt ulations.
This is very unconservative since cylinders are well know to have poor aerodynamics.
- Further, the authors offer no evidence vis a vis observations of missile transport to justify their combined injection and transport model.
Finally, and most importantly, tne bottom line results in Table I l
are ridiculous, in my opinion. Do the authors really believe that the median probability of damaoing only 10 nozzle sets is 2x10-23 per year and that the "95th percentile" is 6x10-10 per year? All I can say is that there are serious problems with this approach and data they used. My estimate is that the difference between 1 percent and 10 percent of nozzles being damaged is a factor of perhaps 5 to 20,. not 3x104 as seen from Table I.
Hence of nozzle damage might be about 10-5 rather than 7x10-9,10 percent Also, the uncertainty intervals for 50th to 95th percentiles raise serious questions about the statistics they generated.
I would like to see what their 97 or 99 percen.ile numbers would be.
In summary, although I was impressed with the mathematical rigor used in certain pieces of the work, the inconsistencies in the approach, weak assumptions, data misuse, and poor documentation would lead me to reject this approach totally in its present form as a viable tool for nuclear power plant safety assessment.
In my opinion, this approach lies at the opposite end of the spectrum of a direct, straightforward and well documented methodology for nuclear power plant safety analysis.
Needless to say, I would like to have more time to review this document and to do some independent calculations on Palo Verde.
I hope that these comments add toward the objective of improved I
understanding and communication in this research area.
Please feel free to call me so that we can discuss these and other areas of mutual research interest.
Sincerely, L wrence
. Twisdale, Ph.D., P.E.
i Princip Engineer and Manager South Division 6
LAT:Ir
'1
---mmmmm----
ee
e-Page 6 of 6 100 I
I I
80 1
I Tornado Winds "
T l
?
g o
=
o r-I e
a l
a 80 T
l s-l!
8 A
A h l
?
' l I
.8 1
4 I
y 40 a
3 Anemometer
j---------
- 33 Height
/
/
/
20 Ordinari Winds
/
(Logarithmic)
/
/
/
/
I I
I I
I 0
50 75 100 125 150 200 250 300 Windspeeds (mph)
Figure 7. Windspeed variation with height.
M
,e
ENCLOSURE 2 Page 1 of 3 C PONSE TO LETTER, E. M. MCKENNA (NRC) TO G. PAPANIC, JR. (YAEC), DATED NOVEMBER 25, 1986 (NYR 86-262)
QUESTION 1 The qualitative statements regarding the limited impact of failing wall D11053 have merits but may not be sufficient to resolve the staff's concern in this
- matter, e.g., that fail'ne of'D11053 could lead to failure of all three diesel generaters.
RESPONSE
In order to resolve the staff's concern, YAEC will upgrade the west wall of Diesel Generator Cubicle No. 3 (Wall D11053). See also the response to Question 4, Enclosure 1.
QUESTION 2 The development of top event LE (see Page 5 of 21 in 9-5-86 YAEC letter) needs further clarification. The staff notes that OERCMOV can be used to reduce the unavailability of the recirculation valves, but its position in the equation reduces the unavailability of the diesel generators as well. Please explain.
As part of the response, also clarify the time for operator action between reaching the point when recirculation should be established per procedure and when it must be established to prevent pump damage.
RESPONSE
The term OERCMOV accounts for the operator taking manual action to initiate opening of a' Safety Injection Recirculation (recirc.) Motor-Operated Valve (MOV) manually at the valve (locally). If the expression for RC is expanded, OERCMOV is ANDed with the other terms in RC.
This reduces the unavailability of the recire. The reduction process corrects the Diesel Generators when the Top Events LD and LE are combined in the development of the LOCA portion of the analysis. The Diesel Generator Random Failure will dominate the 5177R
ENCLOSURE 2 Page 2 of 3 expression at the lower wind speeds due to the conservative failure probability of D.1 for failure to start and run for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> (Section 6.4.1, Reference 1).
At the higher wind speeds, location failure will dominate and the manual recovery, OERCMOV, will become important.
As discussed in the telephone conversation during the November 21, 1986 YAEC-NRC meting, the expression for RC includes the fact that the operator has already recognized the need to transfer to recirculation.
The analysis considered the probability that the operator fails to take action to initiate the transfer to recire. Note that the expression for RC takes into account the fact that the recire. MOVs are readily accessible to the operator for local operation.
Since the operator and his supervisors have been following the progress of Safety Injection and plant response for many hours, the probability that the operator would fail to take the action to switch to recirculation AND his supervisors would fail to direct the action is a negligible contributor to the failure of RC.
Also, the nature of the event must be considered.
The LOCA being addressed is small (i.e., stuck open PORV) so the time to reach the recirculation switchover point would not be short and the time period available for the operator action between reaching the point when recirculation may be established and when it must be established to prevent pump damage would be long. The failure rates included in the expression RC are conservatively assumed to be relatively large (on the order of 0.1), even though the action is part of a rule-based procedure and is well trained.
Additionally, the commitment to upgrade Wall D11053 makes the overall recirculation failure model, RC, even more conservative in that the No. 3 Diesel West Wall will now be designed to withstand a much higher wind speed thereby reducing the unavailability of the Number 1 and Number 3 diesels due to location failure.
The time available for the operator to take action to transfer from injection to recirculation is two hours, minimum, assuming the maximum allowable flow from the ECC System.
5177R
ENCLOSURE 2 Page 3 of 3 QUESTION 3 On Page 4 of 21 (Question 2 response), the statement is made, "The failure of the nonreturn valve platform is not an important contributor to failure as discussed below for Question 11."
How does Question 11 (main steam / feed support structure) relate to the NRV platform?
RESPONSE
The NRV platform is located on and supported by the main steam /feedwater support structure.
QUESTION 4 With regard to exterior wall TlJ2 (Page 15 or 21), what equipment important to safety is located outside the Auxiliary Boiler Room that could be affected if interior walls are exposed to the wind load after TlJ2 fails?
RESPONSE
The only equipment in the Auxiliary Boiler Room (ABR) credited in this analysis is the steam-driven emergency feed pump and associated piping. The failure of Wall T1J2 is assumed to fail this equipment.
There is no other equipment important to safe shutdown located on or near the north, east, or west walls of the ABR.
See Figure 1.
REFERENCES 1.
Tornado Cost-Benefit Analysis for Proposed Backfits at Yankee Nuclear Power Station, YAEC-1428, September 1984.
5177R
EACLOSURE 2 f/GURE 1 I.
__I _.~..,
E,.7.__. _ ____.
L.,
'f - r-
.,_w g
I- - d e,r e an c esT EC.
tm. a m,.ua =\\
,l T.as g*.A.3,..n...
/
WW L DWC, ir I
s e c c.a.
g
-. p. _.,
. __ )..
sn...
rf.t2 4
/
u n
f.,.tm.. a t
/\\
_ " A ' -fa _. /
am
-.c, c a,. u w.7-Jr/
r
.rtu.7 4
(=cr titaar..-
m
..g
-9
,! _p I c.s.: Ic+z l L'*~
I 6;plt:a O -(.
y i
i,rgaA..
i 7..
- g. s. c.cg".
ca
... l +
g etty 4
L.,
cm a n== r.
.c..
,4 I
_ce_,,cet _
- 3,,,. - - _ _ _ _ a 1,,
ra,,,,
- v..
s e net t
c po
.g.'gi
\\ 5,,
s.
,,cm.
[
dit';',',
c v.
i a rout = = =
i.c t,
.rueg.,. t i
su ar.. 7. _ /.
=.r.. _ e
, c.,.
co:r., cm.
m,.,m,.
m
,,, _, N t-
- a..ms a ;-
i a-
-, :.m.
q 5
<==
3 C.,
pt r
o ceri,.s opa s r e rwecir,s.. i, I
f em s a r
,a,.r m
.s l.. w.1 -h l' ar i
g 4
s s
N.s 3
c,-La h_
H, l. a >
, _t.
3 m i
- Yd l 'c., r,_.1._Af./L.BollER
- L.e,,.
h.m l'l
.._ _ _ __ i i a __,.a Y.F K
k,
-i ii.
m
! 9 iigoog_.f t.a.n_a_u.:
N -
-- l.
i ccm.vuoi,. msa s s.
s
%.- l-ocam m.owen i CC'8 N WN.
.?
t, - - - - - - -t -
i wo,3 sgargy l
i; i
ie cwTsa. c.c**
1 y :s cecten
{.
Ar-p s., r
.,............r.
\\
7 '0' I:
i T*Mc 1
r' 4-C'.
i sg _y.-...
- 4.
[
6 T
. _._ _ i.a..n r
1
..... h w
.l..__. TI *
.MT _[,r]g
,9 I
/
T
,,l A
4(Cab c
' aSo~
- o p
g i
%.l l
/Eh 4.j si
!,_(urar:
his q =' gw,,N'",.,*j % j/T.
- w a
a
=
_,\\
H.
=tr 9 i
r i
w r.u.s. ;. as. -
i.s. t n
u
} Les os Tw4
'i'
.i s a t s o'sce.eu eta.. p. mort.org.rg.
- j ;
c_
g gggg3gfgg j
L i.,,- - M-
' - ----,j gro-g, e g.y,N l ?.
I f
[' i Y
\\
'd=8-..I' ( l ;, ' "e, d.
- _44 l
g /l t ervrrm w 7 MA
.e i
j.
f J'.
l l l g -*""" i I
M l
%{
7tT g 4
untar cc.
t l1 Ni j
/"'*
+
l I
p2 4
. !',,.,l (,,
i, s ;
i 4g i
_=t. ] d].,
,..e 4'.
UP,[
~
is r" -__.i
_.J=
l 1
T"T*' ]
I.
I J' 'I
--T '
l Ql; Ihl8 l'
2
'e[ '
l l
t
.m, M.gew 'u*
e Lu'.I CRL E2:rs 4
NEATE2 l
gl i.'T" EAN 8
I.
/
,' LfCTRO.R/U j
i
[; q'
.% 'I "n'~sremt
. '#'-df -l s
6.
i r--
4 Crk Tasen.
op ^
Il i
l !
' 7**! C Q
. +
r 4 ;oy 3:ri,
- -* a rvo a.
Tr.
Al uAsucou tNSER e
s
.v-.
4 e
.- e wn.'..se,.M f
ae, ChT.L Pan.
y a
_n
>\\r a mme.a
.u..at w
l l
}
r< ss:s W
s ls t, ws c.on
~~-
snes
=.
i r.aq.
l 3
. n-i rmr c~.m.
i 1.- r1 7
- e. 3
_. -fi p-_y,.,=.I,
,.m.
T
- i. [._ e l _-
s l
O s
' "p
,l l..' y g
C, r-C M sete vaenn s g
P* *' a @ 'vas"s f
1 t'=L y,
,I i
.c :
t-i-
i i
-. l. ~.
.I /
.yg f~4- ~~; v,~.- = G = =.. a.*;f-cott n.4.5 ri rt acce t
...-f, f, 7'
t
.:~.= : _.-. -
- - - - : _ L -.,- -,~ r t
_. ] - _ s-
. g
-_-__s.
o s
g __
'N i l
l
'/ "
- lt Cootnstmd.
3 3".g,*
.. 7 l'
I A
g w o'.
L.
. ;.r.c
.w o-n' o' I
.. r. c-i y
, r}
a
\\
G3,1
..'z' 5
s,.
_ r
\\