ML20069D670
| ML20069D670 | |
| Person / Time | |
|---|---|
| Site: | Browns Ferry  |
| Issue date: | 03/15/1983 |
| From: | Mills L TENNESSEE VALLEY AUTHORITY |
| To: | Harold Denton Office of Nuclear Reactor Regulation |
| References | |
| NUDOCS 8303180454 | |
| Download: ML20069D670 (9) | |
Text
e TENNESSEE VALLEY AUTHORITY CH ATTA NOCG A.. INNESSEE 374o1 400 Chestnut Street Tower II March 15, 1983 Mr. Harold R. Denton, Director Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington, D.C.
20555
Dear Mr. Denton:
In the Matter of the
)
Docket Nos. 50-259 Tennessee Valley Authority
)
50-260 50-296 In response to verbal requests from your staff, we are providing the enclosed additional information regarding Browns Ferry RETRAN methodology in our reports TVA-TR81-01 and TVA-MDS-553 provides information discussed in a March 2,1983 conference call with your staff and enclosure 2 provides information discussed in a March 8, 1983 conference call. It is our understanding that the enclosed information resolves all NRC concerns on this matter.
As indicated in my [[letter::05000296/LER-1982-067-03, /03L-0:on 821225,coupling Unit on Diesel Generator C Lube Oil Pump Found Broken.Caused by Deterioration of Coupling Spider Due to Normal Svc.Pump Mfg by Viking & Coupling by Lovejoy.Coupling Replaced|January 20, 1983 letter]] to you, your immediate approval of our submittal is requested. Any further delays in your approval beyond our original requested date of November 1, 1982 will result in schedule complications for TVA in preparation of our in-house reload core design for unit 3 cycle 5.
Very truly yours, TENNESSEE VALLEY AUTHORITY L. M. Mills, Manager Nuclear Licensing Subscribe sworn t fore of, me th a day of A GA/ 1983 b
Notary Public Dk My Commission Expires 9
Enclosures cc: See page 2 8303180454 830315 PDR ADOCK 05000259 p
PDR An Equal Opportunity Employer
0 2-Mr. Harold 3. Denton March 15, 1983 cc (Enclosures):
U.S. Nuclear Regulatory Commission Pegion II ATTN: James P. O'Reilly, Regional Administrator 101 Marietta Street, NW, Suite 2900 Atlanta, Georgia 30303 Mr. R. J. Clark Browns Ferry Project Manager U.S. Nuclear Regulatory Commission 7920 Norfolk Avenue Bethesda, Maryland 20814 4
e
_.s-
-._._____,,__-.,__.-.,,-..._m,
_,yy
,y,.-_
1.
-y
ENCLOSURE 1 BROWNS FERRY NUCLEAR PLANT RETRAN MARCP. 2, 1983 CONFERENCE CALL ed 4
9 r
1 6
D I
~
I
'^*y*C
- bA..
.4*%. dmk
-m.
~,
. 1'.."#
' ' ~
e-1 y
01.
Jcotify ths assumption thot ths valeco of th3 random variables in the response surface equation are normally distributed.
A1. The distribution of scram speeds (variable SS in equation 8-2) and response surface fitting errors (variable URS) were tested and no reason to reject the assumption of normality was found. Insufficient data is available to test the distribution of initial steam flow (SF) and model uncertainty (URM), but the use of a normal distribution is a common practice in the absence of contrary information. The distributions of scram speed and model uncertainty have the dominant effect on the statistical adjustment factors.
Since the model uncertainty can be expected to be due to,the combined uncertainty in a large number of independent components the central limit theorem of statistics would indicate that the assumption of a normal distribution for URN is reasonable.
Q2.
The model uncertainty of 25 percent was arrived at primarily from comparisons to the Peach Bottom turbine trip tests.
However, this uncertainty was used not only for the load rejection event but for the feedwater controller failure also. Justify.
A2.
The model uncertainty was set by examining both the Psach Bottom turbine trip test results and the results of sensitivity studies shown in tables 7-1 and 7-2.
The sensitivity study results,for the FWCF generally showed the same trends as for the GLEWOB but with the magnitude of the sensitivity reduced by approximately 1/2 to 1/3.
Since the RCPR for the FWCF.is approximately 1/3 less than for the GLRNOB, the fractional uncertainties are roughly equivalent.
, ][
It should also be noted that a FWCF event evolves into a' turbine trip' fg with operation of the bypass valves and the major change in CPR occurs following the turbine trip. Since the Peach Bottom tests were turbine trips with bypass operation, they are appropriate for evaluating uncertainties in model predictions of FWCF events.
QG. Justify the use of initial steam flow instead of initial power level as a random variable in the response surface equation.
e m
~
A3.
It makes little difference whether initial steam flow or power is selected as the random variable in the response surface equation since they are directly related (for a given operating pressure set point and feedwater heating characteristic). When the initial values are
, expressed as percent of rated values, then both steam flow and power are very nearly equal numerically as shown in f.he table below:
i Initial Steam Initial Power Flow (%NBR)
(% NBR) 90 90.7 95 95.1 100 99.5 105 104.5 110 109.5 A
f It f
1
[
T 4
m ~
.a..*~ T,.;.
y, Asd-Y*'ag.ph &^ g x 3;_
q
4 o-e a,.
2
~
However, for pressurization events resulting from stopping the steam flow to the turbine, the severity of the event is most closely related to the pressurization rate which is, in turn, more directly related to the initial steam flow than to the initial power. Therefore, initial steam flow was selected as she variable for the response surface but results are not appreciably affected by use of initial power.
Q4.
Please provide the A coefficients in equation 8-2 for each response g
sreface.
A4.
The response surface fitting coefficients are '1'isted below for the four response surf aces for which statistical adjustment factors were calculated.
GLRWOB GLRWOB FWCF FWCF s
at BOC at MOC at EOC at MOC A,
1.13970 E-1 6.25919 E-2 9.54929 E-2 6.83487 E-2 A
1.21866 E-3 1.33998 E-3
-4.04006 E-4
-3.42683 E-4 3
A
-1.30147 E-5 1.34697 E-5
-2.16423 E-5 4.79854 E-5 3
A, 2.41622 E-1 1.35571 E-1 8.00299 E-2 1.07703 E-1 A.
1.50950 E-3
- 3.23100 E-3 9.37314 E-4 2.40229 E-3 A,
5.63364 E-3
-5.90817 E-2
-9. 0517 E-2
-1.52479 E-1 4.
s
? L kg r 7
La k[ig,., '.-n.. _. as.a.gg -
,,_y.
Q_,
L
ENCLOSURE 2 BROWNS FERRY NUCLEAR PLANT RETRAN MARCH 8, 1983 CONFERENCE CALL m.*
O e
ep l
e e
1 l
l I
mw::. '.. -
_. = -
-~-2
= i: h--
Q1.
In collapsing from 3D to 1D for various cross sections, either flux or flux-adjoint weighting may be used. Apparently flux weighting is normally used. Is flur-adjoint weighting ever used? If so, under what conditions?
A1.
Our current practice is to use flux-adjoint weighting. However, sensitivity studies have shown little difference in transient results for the licensing basis pressurization events when flux weighting is employed. For exemple, the GLRNOB events described in chapter 6 of i
TVA-TR81-01 yields the results shown below when analyzed identically, except for the radial weighting used in cross section collapsing.
REIRAN Results for GLRTOB Weighting used in 3D to ID Cross Section Collapse Fluz Flux
- Adioint Peak power (%NRB) 378.32 382.42 time (sec) 0.635 0.635 Max core avg. heat flux }% NBR) 119.75 119.85 time (sec)
~
0.850 0. 84 5 Peak done pressure (psia) 1207.18 1207.13 time (sec) 2.425 2.425 Max ACPR 0.221 0.222 4
Q2.
In table 3 of NDS-553, the standard deviation in the percent error in reactivity change is noticeably high for the all rods out configura-tions PB, IT2 and TT3 tests. Why?
A2.
The initial conditions for the Peach Bottom tests employed rod patterns with a large number of partially inserted control rods.
During the transients, the control distribution will vary between the initial distribution and the all control rods fully inserted configuration. The states between the initial distribution and all rods out will not be encountered. Therefore, in developing the RETRAN 1D cross section fits, the initial and all control rods inserted states were more heavily weighted to improve the accuracy of the fits in the range they would actually be evaluated. This results in larger fitting errors in the unschievable all rods out configuration.
Q3.
The 13 percent uncertainty in the void coefficient used to evaluate ARCPR in table 7-1 (page 269) is based on a change of 75 psi in the system pressure. However, a 160 psi change in pressure occurred in the GLRWOB event (figure 6-4).
Should the 13 percent be scaled to account for this or is it assumed that the uncertainty does not depend on the magnitude of the pressure increase?
- ).
j.,
~
w
---.: x-
'I A3.
The results in table 7-1 do not need to be scaled since the cross section polynomials were modified to obtain a 13 percent more negative void coefficient, and thus, the results in table 7-1 reflect the actual pressure change, not an assumed 75 psi change. The magnitude of the uncertainty in void coefficient was assumed to~be independent of the size of the pressure change, but the 13 percent is a conservative estimate of the uncertainty.
Q4.
Certain sources of uncertainties in the scram reactivity, such as the basic cross sections and assembly,modeling uncertainties, are common to both KENO and LKTTICE calculations. Are these considered or are they assumed to be covered by other conservatisms?
A4. There are potentially some sources of uncertainty common to both the KENO and LATTICE calculations; however, these should be smaller in magnitude than the uncertainties due to approximations in the neutron transport solution. Also, the maximum dif ference in control strength between KENO and LATTICE was less than 5 percent while a 10 percent uncertainty in scram reactivity was employed. This was judged to be an adequate allowance for the common uncertainties not identified by the KENO-IATTICE comparisons.
QS.
Neglecting radial distribution changes during transients affects the scram reactivity. How was this treated?
AS.
As discussed in section 9 of TVA-MDS-553, the major shortcoming identified for the ID control representation employed in REERAN was a tendency to underestimate the worth of bank insertion movement for configurations with control rods initially inserted to different levels. The underestimate of bank movement worth is due to the decreased flux perturbation caused by a control rod tip at an axial plane with high void content relative to the perturbation caused at a l,
lower axial plane with lower voids. This tendency to underestimate scram reactivity is conservative for licensing basis analyses with initially partially inserted control rods.
Q6. How were RETRAN modeling uncertainties included ss opposed to input uncertainties?
A6. The sensitivity studies presented in chapter 7 of TVA-TR81-01 examined several sources of uncertainty, including those arising from RETRAN modeling (examples are the subcooled void model and slip models).
These uncertainties were combined into the 0.041 penalty applied to the calculated operating 1 bait CPR (equation 8-1) for option A.
The modeling uncertainties are accounted for in the option B operating limit CPR by the URM (equation 8-2) variable employed in the response surface analysis for the statistical adjustment factors.
j
' i m m#.w ma,_
_m
'/
c =
3 Q1. There appears to be some ambiguity in the treatment of uncertainties.
For example, the 13 percent void reactivity. Uncertainty used. in table 7-1 is apparently a one-signa value (although this is not clear) but is treated as a 95/95 uncertainty in table 8-1 on page 295.
Please clarify.
A7. The 13 percent void reactivity uncertainty is a bounding value; however, the method used to arrive at this value in section 7.1.1.1 of TVA-TR81-01 is inappropriate.for several reasons. The conservatism
~
of the assumed 13 percent uncertainty in void reactivity can be demonstrated by examining the difference in peak excess reactivity inferred by inverse point kinetics from the measured data and calculated by the REIRAN model for the Peach Bottom turbine trip tests. The results in the table b_el_ow indicate a 95 percent._
i confidence upper bonad (from x 8 test) on the standard deviation between measured and calculated peak reactivity of 4.8 percent. Thus, a 95/95 reactivity uncertainty of 9.6 percent is indicated by the Peach Bottom turbine trip tests confirming the conservatism in the assumed 13 percent uncertainty.
}
Peak Excess Reactivity (i)
Test Data
' Calculation
% Difference TT1 0.803 0.7 90 *
-1.62 Tr2 0.793 0.797
+0.50 TT3 0.829 0.821
-0.97 Average
-0.70 Standard Deviation 1.10 95% Confidence Standard Deviation 4.80
- based on calculation driven with measured upper plenum pressure to eliminate reactivity difference due to uncertainties in pressure prediction of 771. 772 and TT3 results are from base calculation (stop and bypass valve position boundary conditions) since no significant differences in measured and calculated apper plenum pressures were observed.
- 08. Are the different perturbations listed in table 2-7 (page 66) uncorrelated?
A8. At many axial planes in the core (especially near the top), the relative perturbations in nodal water density produced by separately perturbing system pressure, power, flow, etc., will be correlated.
l l
+
. ddh
> RLP
.a C., E Na. u ma. __,_ _.;.;.
. g.f..,
-