ML19317H207
| ML19317H207 | |
| Person / Time | |
|---|---|
| Site: | Arkansas Nuclear  |
| Issue date: | 03/04/1977 |
| From: | Ziemann D Office of Nuclear Reactor Regulation |
| To: | Phillips J ARKANSAS POWER & LIGHT CO. |
| References | |
| NUDOCS 8004290612 | |
| Download: ML19317H207 (8) | |
Text
i I
Q (y*,
', 4
-*r
.i DISTRIBtJTION Docket NRC PDR-
~
Local ~PDR ',s.
^
h p #h c*wi
.m
+
... > ; st :
ORB J2 Reading T M. ~
!@'k6ack aa MBd@* g[etlNon;50-31[3MAPN[MWE[y C197 W-blAR
~
Qi I@fEYOf Y M y i y mGol1er gM TM E
- g j. gg Wy M%F14iY-h wfM)TJCarterm ogG;gpNcNr.
b ;^ v -
q v
.x ;gync pp nw
- DLzinna6ns M a, p g pg w-v Arkansas Power & Light Company RMDiggsi#'
4 ATTN: Mr7 J. D. Phillips RPSnaider 4-Senior Vice President RDSilver Production. Transmission and OELD-Engineering.
OI&E-(3)y~..
Sixth and Pine Streets DEisenhut1/p ",
Pine EI'aff, Arkansas 71601 BKGrimesE 1 _
TBAbernathy~
~
3
. Gentlemen:
-JRBuchanan q,p. g ~. y, -
ACRS,-(16): Sq, V,
We have completed our review of your Jidsly 1.-1976 report entitled "Diel Changes in Impingement Levels", warhich was submitted in accordance with Environmental Technical Specificattion 6.3.
We have concluded that the data collected for this program ansd report satisfy the requirement of Technical Specification 6.3.
However, we noted several weaknesses irin your report which are brought to your attention.- A list of comentsris attached.
No reply to these coments is necessary at this time. Hotowever, if in the future it is detennined that present impingement lomsses are too high. or if you propose to reduce the sampling effortPfrom 24-hour samples to 8-hour s
samp1es, we would request that your everaluations include consideration of our comments.
Sincerely.
Driginal signed b9 ~
THIS DOCUMENT CONTAINS M**8 POOR 00AUTY PAGES Dennis L. Ziemann, Chief L'
Operating Reactors Branch #2 y
Division of Operating Reactors
Enclosure:
Coments on Arkansas Nuclear One Report on Diel Changes in Impingement Levels cc w/ enclosure:
See next page
/
/-MG l$l17 D b B #2 D0b[Bk DDR:0f0RB #2
/
JPSnaidcr:.ID Ifr_ -
._.DLZi==a nn
.o....*
5 /77' h $/77 3/W /77
_._3/
r-uc.m a. m acu am 1
., -..,.... ~
8004 290 f / 2 g
(
l Arkansas '8ower & Light Company hWEAR 4 1977
\\
F cc w/ enclosure:
Horace Jewell, Esquire House, Holms & Jewell 1550 Tower Building
?
Little Rock, Arkansas 72201 Phillip K. Lyon, Esquire House, Holms & Jewell i
1550 Tower Building Little Rock, Arkansas 72201 Mr. Donald Rueter Manager, Licensing Arkansas Power & Light Company Post Office Box 551 Little Rock, Arkansas 72201 Arkansas Polytechnic College Russellville, Arkansas 72801 w--
- ~
P,
~'.
(
(
o COMMENTS ON ARKANSAS NUCLEAR ONE REEp0nT ON DIEL CHANGES IN IMP 1NCEMENT LEVEILS 1.
In the description of Arkansas Nuclear One Unit - 1, p. 2, para. 5, it t
is stated that, "There are no fixed screens in firont of the traveling f ails to mention the presence of t trash bars that will screens." It exclude fish unable to pass a three inch opening:g and may thus be con-e sidered a fixed screen.
2.
In the Study Design Section, pp. 5-6, the repor -t states, "The study required subdividing a regular 24-hour impingemenent sample into three 8-hour samples."
'No description was included asas to what a " regular"
- Also, impingement sample is or how many samples are tcto be so treated.
although the estimated number of fish impinged u were to be determined for each species, no mention is made of how the:e estimates were made.
There was no discussion of or rationale for choroosing the 8-hour periods 3.
used. For example, it should be pointed out tir.h a t if impingement were indeed independent of time of day, it would be ; impossible to split the day up into 8-hour. periods and artificially cre:eate a diel impingement i
pattern by manipulating the sampling schedule.
If, (as is the case differences for this study) on the other hand there are sigrgnificant l
in impingement levels during that day, it is pcnossible to mask the differences by arbitrarily placing the end of c one 8-hour period and the beginning of the next 8-hour period on the.e impingement peak, thus
t
(
-2 h
b'I
~
splitting the high impingement level to two time periiods and coming up with nothing more than an " average." There is no n-way to tell from these data to what degree this averaging has occurre&d as a result of arbitrary division of the day, and impingement level. differences indi-cated by this study should be considered low estimatees.
Perhaps the s
sampling periods would have shown greater dif ferencees between them if the divisions had been cued on dawn or dusk or sunrisise and sunset.
4.
Study Design, p. 6, para. 3.
This section states tha2at the April 1976 impingement samples were not collected because the pblant went off line in late March, but that no statistical ef fect occurrwed "because impinge-ment has returned to the anticipated low level of laute spring and sum-mer."
We would like to have seen some values and/orr ranges for these
" low" levels.
5.
Results, p. 6, para. 2.
This section states, "A two-o-way analysis of variance...shows no statistical correlation of fish _ impingement with time of day."
An analysis of variance shows sources.s of variation, not correlation.
6.
Numerous inconsistencies wer, noted between the vari:ious tables and graphs includad at the end of this repo-t.
For examzmple, the weight of fish for the 12 PM to 8 AM period of the 28th of JJanuary was 1923.5 pounds in the Results section, 1928.0 pounds in Tab 121e 18, 1093.5 pounds in Table 31, and approximately 1900 pounds in Figure e 1.
a
(
(
3_
serious deficiencies in the statistical methods used to
-.j 7.
There are The following interrelated points are made in analyze the data.
.1!
regard to this criticism.
I The impingements for individual time periods vary from less than a.
It is most implausibble.
one pound to more than one thousand pounds.
to presume that such impingements (or their residuals) have equal Therefore, the analysis of variances based on the direect variances.
weight of the fish is suspect.
Nonparametric methods such as Friedman's Test (see c. below) are more appropriate. Analyses of-variance on transformed weight, for example by logarithms (see d..
below), are also useful since they are likely to mitigate the problem of unequal variances.
Appropriate transformations will also tend to be in closer conformance with the linear model as-sociated with the analysis of variance. The factors controlling the weight of impinged fish, ould quite plausibly Lave an approxiri--
c l
mately multiplicative rather than additive ef fect.
When the Biomed program IBMDP2V was used to duplicate the analysi-is b.
of variance presented in Table 31 (see attached), quite dif ferentet results were obtained than presented in the report. The reporteacd value of F for time of day was 0.95 with a P value greater than 0.05 (which is not signi~ficant at the 95% confidence level).
Durar duplication of the analysis yielded an F for the time of day of 5.2 with a P value of 0.017 (which is significant at the 98+% coron-fidence 1erel).
c
(
_4 t
The data of Table 31 are strongly suggestive of diel variation in
'j c.
'2 ~~ 3
.I the weight of impinged fish.
For the ten days (dates) for which
.I data were obtained, the 8 AM - 4 PM impingement yielded the lowest value nine times and the 4 PM - 12 M impingement vielded the highest 4
impingement eight times.
Such an outcome on the basis of random ranking is most unlikely as confirmed with Friedman's rank order analysis of variance test (see attached).
This test yielded a P value less than 0.001 (significant at the 99.9+% confidence level).
d.
The Biomed program IBMDP2V was used to analyze the variance of log-transformed weights and yielded results with a much greater level of significance than was obta.ined in b. above (which used untrans-formed weights). Specifically, the analysis yielded an F value of 11 with a P value of 0.001 (which is significant at the 0.999 level).
In summary, and contrary to the above referenced report, we conclude that
. there is strong evidence of d.iel variations in the weight of impinged fish.
9 s
9
(
(
THE FRIEDMAS TEST l
random variables are mutually independent r~
Assumptions:
(1) The b k-variate L.
(The results within one block do not influence the results EE within the other blocks.)
(2) Within each block the observations may be arranged in increasing order according to some criterion of interest.
Hypothesis Each ranking of the random variables within a block is equally likely H:
(i.e., the treatments have identical effects)
H: At least one of the treatments tends to yield larger observed A
values than at least one other treatment.
1 T=
IR -3b(k+1) where T = Friedman Statistic
+1}
b = number of blocks (days) k = number of treatments (8-hour periods) 2 R = sum of rank squares Reject if T> x degrees of freedom = k-1 2
R2 = (28)2+(21)2+(11)2 b = 10 k
=3 x 1346 - (3)(10)(3+1)
(30 (4) 12 x 1346 - 120
=
120 134.6 - 120
=
14.6
=
= 12.8 reject - i.e., the alternate hypothesis is correct.
2 X
(2).001
(
(
2 - WAY ANALYSIS OF VARI ANCE Dependent variable - Weight Prob F degrees of Source Sum of Squarbs freedom Mean Squarc F
Exceeded Mean 1176156.0000 1
1176156.0000 1064.2874 0.000 Time 11484.7148 2
5742.3555 5.1962 0.017 Day 3142027.0000 9
349114.0620 315.9085 0.000 frror 19892.0000 18 1105.1111 Dependent Variables-Log Weight Prob F degrees of Source Sum of Squares freedom Mean Square F
Exceedei Mean 412.4241 1
412.4241 4555.5273 0.000 Time 1.9914 2
0.9957 10.9980 0.001 Day 117.8317 9
13.0924 114.61536 0.000 Error 1.6296 18 0.0905 s
4 y
w 4