ML19309H499
| ML19309H499 | |
| Person / Time | |
|---|---|
| Site: | La Crosse File:Dairyland Power Cooperative icon.png |
| Issue date: | 05/08/1980 |
| From: | Linder F DAIRYLAND POWER COOPERATIVE |
| To: | Ziemann D Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML19309H500 | List: |
| References | |
| TASK-15-13, TASK-RR NUDOCS 8005130405 | |
| Download: ML19309H499 (9) | |
Text
__
8005130yoS
~
D DA/RYLAND COOPERAT/VE eo sox si7 2 sis EAST AV SOUTH
- LA CROSSE. WISCONSIN 960i t308) 7884000 May 8, 1980 In reply, please refer to LAC-6905 DOCKET NO. 50-409 h
Director of Nuclear Reactor Regulation ATTN:
Mr. Dennis L.
Ziemann, Chief Operating Reactors Pranch No. 2 Division of Operating Reac ors U.
S. Nuclear Regulatory Commission Washington, D.
C.
20555
SUBJECT:
DAIRYLAND POWER COOPERATIVE LA CROSSE BOILING WATER REACTOR (LACBWR)
PROVISIONAL OPERATING LICENSE NO. DPR-45 CONTROL ROD DROP ANALYSIS
Reference:
(1)
NRC Letter, Ziemann to Linder, dated February 28, 1980.
(2)
DPC Letter, LAC-6184, Linder to Ziemann, dated March 29, 1979.
Gentlemen:
Please find enclosed as Attachment 1 DPC's response to the NRC staff request (Reference 1) for additional information concerning the Control Rod Drop Analysis submitted by DPC in Reference 2.
Also enclosed as Attachment 2 is Revision 1 of the report NES-81A0033, "LACBWR Rod Drop Probability Study", originally submitted in Reference 2.
If there are any questions concerning this submittal, please contact us.
Very truly yours, DAIRYLAND POWER COOPERATIVE A
Frank Linder, General Manager FL:SJR:af Enclosure cc:
J.
Keppler, Reg. Dir., NRC-DRO III b
o v
e ATTACHMENT 1 TO LAC-6905, MAY 8, 1980
Subject:
LACBWR Rod Drop Probability Study -
DPC Response to NRC Request for Additional Information, dated February 28, 1980.
NRC COMMENT a QUESTION:
Our understanding of the context in which this report is written follous:
A calculation using a three-dimensional kinetics code is used to determine a value for dropped rod worth uhich vill produce peak enthalpy of 280 cal /gm in the LACBWR reactor fuel.
Once this value has been determined, the normal rod vithdraval sequence is followed and potential dropped rod oorths are calculated.
These calculations shou that it is not poscible to achieve the critical rod vorth until the core is at essentially full pover.
The implication here is that a control rod drop accident, RDA, ex-caeding 280 cal /gm is possible without any errors in the rod with-draual sequence having been made during operation.
Please provide responses to the following:
1.
Comment on our undcretanding of the situation.
Provide corrections or explanations where appropriate.
DPC RESPONSE:
The NRC's understanding is essentially correct.
Rods of high worth occur in LACBWR only after reaching a power which produces signifi-cant moderator voiding and a rod pattern in which control rod shadowing is minimal.
At low power, when all 29 rods are banked, the rod density is great and a rod dropping does not expose a large core area.
At higher power, when only a few rods are inserted, if one rod drops, a large portion of the core can become completely unrodded.
In addition to the smaller rod density at high power, another factor of even greater importance causes higher rod worths at power.
This factor is coolant voiding in assemblies at power, with higher powered assemblies having higher steam void content.
These steam voids have a negative reactivity and tend to reduce the reactivity and hence the power density in the upper portion of BWR's.
If a rod should inad-vertently be left inserted in the core, the assemblies around it would have a smaller void density than if the rod were correctly j t 1
l
)
ATTACHMENT 1 - (Cont'd) positioned.
If now the rod were to fall out, the voids and their negative reactivity would still be absent during the very short time period when the rod is falling, leading to a higher rod worth than would occur for voided assemblies.
To determine the potential reactivity worth of a dropped rod, DPC first determines the steady state thermal-hydraulic and neutronic conditions in the approximately critical core with the rod in question inserted.
Then a second calculation is performed with the control rod removed (or withdrawn a desired amount) but with moderator voids, xenon, etc. maintained the same as in the base case.
The calculated keff's for the two cases are compared to obtain the worth of the dropped rod.
By doing the calculation in 4
this way, the reactivity effect of the lower moderator void condition (greater moderator density) in the fuel assemblies surrounding the location of the dropped rod is correctly accounted for as well as the removal of the control rod poison itself.
The effect of the low moderator void condition persisting for a brief period after a control 1
rod is rapidly removed is the major reason for the fact that the re-activity worth of a dropped control rod is much greater at power than DPC has found that if the keffi for equil-at lower or zero power.
ibrium core conditions with and without the control rod,in question are compared, the calculated rod worth is only about 1/2 as great.
NRC QUESTION:
2.A.
What conditions vere assumed for the calculation of the rod vorth required for 280 cals/gm?
DPC RESPONSE:
A typical rod pattern using 9 rods in a 3x3 square array controlled a core made up entirely of fresh Type III fuel of 3.7 w/o U-235, clad in stainless steel.
Both stainless and Zircaloy shrouds were used to make the core approximately critical.
The core was assumed to be hot (561 F) and at 1 MW power with no xenon.
With fresh fuel, no enhancement of the Doppler feedback from Doppler broadening in plutonium exists, so the use of all fresh fuel is conservative.
No moderator feedback was used during the study of the primary power excursion; it was assumed, however, that some mechanism such as moderator feedback or scram would prevent subsequent power peaks.
The core was modeled in three dimensions with the transient analysis code BWKIN.
This code takes as input two group cross sections which were determined for LACBWR as a function of temperature with the NULIF and EXTERMINATOR codes.
Six delayed neutron grcups were used. <
4 ATTACHMENT 1 - (Cont'd)
BWKIN takes into account the movement of control rods, effects on neutron cross sections due to coolant and fuel temperature, heat buildup due to fission and decay heat, heat deposition in fuel (96.7% of fission energy used) clad and coolant, and thermal-hydraulic properties of the core.
A transient analysis begins with a determination by BWKIN of steady state neutron fluxes and a normalization to achieve exact criticality just before reactivity insertion.
Rod drops were per-formed for rod worths of 1.5, 1.6, and 2.5%:
ramp rates were taken as essentially step insertions (1.0 x 10-4 sec. for rod withdrawal) because auxiliary studies had already shown the total energy depos-ition to be nearly independent of ramp rate.
The rod dropped was a corner rod from the 9 rod pattern; calculations with XENOLUX during routine fuel management studies show that corner rods have more worth than side rods or the center rod.
NRC QUESTION:
2B.
Was the initial power assumed to be in the 80-100 percent full pouer range?
DPC RESPONSE:
No, the initial power was conservatively assumed to be 1 MW which is approximately 0.6% power.
Sevnral preliminary BWKIN studies from 2 MW down to very low powers showed very little sensitivity to initial power, and therefore a significant but low power (which simplified the BWKIN run) was used.
The amount of conservatism in this assumption is not expected to be large.
NRC QUESTION:
2C.
Was moderator temperature feedback included in the calculation?
DPC RESPONSE:
No, this was neglected for conservatism.
NRC QUESTION:
2D.
Wac the dropped rod the central rod or uas an off central one aceumed?
DPC RESPONSE:
An off central rod (corner rod of the 9 rod bank) was assumed.,
ATTACHMENT 1 - (Cont'd)
NRC QUESTION:
22.
If both vere calcu12ted, does it make any difference?
DPC RESPONSE:
Only the off central rod was studied and this is. realistic because XENOLUX studies show this rod to have more worth'than the central rod.
NRC OUESTION:
3.
If the acceptance criterion was reduced from 280 cal /gm to 200 cal /gm, vould this make any significant difference in the conclusions of this study?
DPC RESPONSE:
It was determined in the rod drop study that dropping a 1.4% Ak/k rod
- would cause a small portion of the fuel to reach 280 cal /gm; the same study shows that a rod of roughly 1.3 to 1.35% would cause fuel to reach 200 cal /gm.
The only effect this change from 1.4% to 1.3% would have on the probability analysis is in the probability factors called "High Rod Worth" and " Rod Drop Timing" in Table 1 of NES 81A0033, Rev.
1.
1 NRC QUESTION:
<i. A. If rod uithdraual errors (i.e.,
uithdraval of an out-of-sequence rod) vere not considered in the startup and pouer escalation phases of operation, uhat vould be the effect of including such errors?
DPC RESPONSE:
Rod withdrawal errors were considered in the startup and power escal-ation phases.
However, the worth of such an out-of-sequence rod was determined to be less than or approximately equal to the worth of an in-sequence rod at full power.
See answer to question 4.C for the effect of including rod withdrawal errors in the probability analysis.
l l
- See NES 81A0033, Rev. 1.
This rod worth was previously reported as 1.63% Ak/k in NES 81A0033, Rev.
O.
ATTACHMENT 1 - (Cont ' d)
NRC QUESTION:
4.B.
Could an RDA exceeding 280 cal /gm result?
DPC RESPONSE:
Yes, when the worth of the dropped rod exceeds approximately 1.4%.
NRC QUESTION:
- 4. C.
If so, are the probabilities, e.g.,
the nondetection proba-bility, quoted in this study still applicable?
DPC RESPONSE:
Yes, most of the probabilities given in the probability analysis apply for RDA's of greater than 280 cal /gm.
Obviously, since greater RDA's require larger worths of dropped rods, the probabil-ities for "high rod worth" and " rod drop timing" would be decreased but the probabilities for disconnect, sticking, and nondetection remain the same.
Also, another probability factor for operator error in rod selection would be applied to find the total proba-bility for such a RDA.
This additional factor for operator error would be at most no greater than 1.0 x 10-6 since several hundred rod selections and withdrawal movements would be involved (several tens of selections and movements of the rod in question) and several operators and at times more than one crew would have to consistently make the same error without discovery of their mistake.
Therefore, the greater probability of a RDA is found if operator error in rod selection is not included in the analysis.
NRC QUESTION AND COMMENT:
5.A.
Has the effect of normal power level changes been included?
B.
During auch changes, roda might be moved in larger increments than vere considered.
Discuse such changes uith respect to their effect on the statistice.
DPC RESPONSE:
Yes, the effect of normal power level changes has been considered.
Individual control rods are neven withdrawn in large continuous movements during power level changes in the LACBWR.
Operating procedures require that individual control rods be kept within 2
inches of the nominal position of their respective rod bank.
It is standard procedure to not pull any rod more than one inch at a time 4
ATTACHMENT 1 - (Cont'd) i during power changes and in general any one movement is lees than 0.5 inches in order to limit pellet clad interaction and excessive clad stress.
Typical rod movements to compensate for burnup are less than 0.25 inches / day on 4 or 6 rods.
NRC QUESTION:
6.
Provide a brief descriptior. of the methods used to obtain rod uorths and parameters for input to the BWKIN code.
DPC RESPONSE:
The two group cross section sets as a function of temperature for 0
fuel temperature from 561 F to 6000 F required as input to BWKIN were generated with the NULIF and EXTERMINATOR-2 codes.
The input to NULIF was based on the as-built engineering design value for Type III (Exxon) fuel assemblies.
Rod worths used were taken from previously calculated values for input to TRILUX, the fuel management code used for LACBWR.
NRC COMMENT:
7.
Comment on the follouing probability analysis by the staff.
The LACBWR rod drop probabilitiec in Table 1 appear to have been incorrectly calculated and may be greater than the criterion of 10~7 per reactor year.
From the event definitions in Section 4.2 and the multiplication of the probabilities of rod disconnection (RD) and non-detection (ND) in Table 1, it is evident that " rod disconnection" is the event that a rod disconnects, whether or not it is detected.
Houever, the event uhose probability is estimated in Section S.1 appears to be " rod disconnection uith non-detection" (RDND).
The report bases its estimate of the probability of RDND on an observed value of aero occurrences of RDND ("no disconnects uhich have not been detected") and makee no apparent use of eight disconnects uhich have been detected in GE plants.
Based on 2 : 105 vithdravals from Ref.
3, the report calculates an upper 95% confidence limit (incorrectly referred to as a prob-ability) on the probability of RDND of 1. S x 10"5 par withdraval.
It is certainly possible to estimate the LACBWR rod drop prob-ability based on a direct estimate of the probability of RDND rather than by multiplying the probabilities of RD and ND, as uas done in Table 1.
Houever, if this approach is taken, then the factor of 10~8 for the probability of ND in Table 1 must be -.
ATTACHMENT 1 - (Cont ' d) omitted.
The resultant estimate of the LACBWR rod drop probability is 5 x 10~7, a factor of 5 larger than the safety criterion.
The data quoted in Section 5.1 can be used to estimate the probability of RD.
From Ref.
3, the GE plants have experienced 8 disconnects in 2 x 105 vithdravals.
This yields a maximum likelihood estimate of 4.0 x 10~5 and an upper 95% confidence limit of 7.2 x 10~5 for the probability of RD.
(Note that these results are consistent uith the LACBWR experience.
If the probability of RD is 4.0 x 10'5, then the probability of 4 uithdrawala the observed event of no disconnects in 2.5 x 10 is. 3 7.')
If we use maximum likelihood and the GE data from Ref. 3 to estimate the probability of RD, then the estimate of 1.5 = 10~5 must be replaced by 4.0 x 10'S in Table 1.
Furthermore, the estimate of 10'8 for the probability of ND may be too small.
From Section 5.3, this estimate is calculated by assuming that there are 8 independent detection opportunities, each uith a failure probability of 10~l.
Houever, if ND is a common cause failure, so that the failures are dependent, then the probability of ND may be ao large as 10~1 Thus, the estimated probability of LACBWR may be as large as 1.3 x 10^7, somewhat larger than the 10~7 eafety criterion.
i DPC RESPONSE:
DPC agrees with NRC's estimate of rod disconnect (RD) probability of 4 x 10-5, derived from 8 disconnects in 2 x 10-5 withdrawals in GE plants.
The probability of RD in Table 1 should be changed to 4 x 10-5 (See Rev. 1 of NES 81A0033 attached to this submittal).
The probability for non-detection (ND) was calculated to be 10-0 on the basis that a rod is withdrawn in many steps with an estimated probability of ND at each step of 10-1 Hence, (10-1)8 = 10-8 if only 8 steps are assumed.
NRC points out that if ND is due to a common cause failure so that the failures are dependent, the probability of ND may be as large as 10-1, DPC agrees that this could be the case if the detection of rod movement depended on a null measurement because then something as simple as a disconnected meter, always indicating no change, could be a common cause failure leading to 8 successive ND's even though the rod were stuck.
However, the detection of an individual rod movement (proof that rod is not stuck) depends on a positive response from a measuring instru-ment, not a null response.
A single null response would raise a sus-picion of a stuck rod.
ATTACHMENT 1 - (Cont'd)
Therefore, considering the many rod selections and movements.
involved in withdrawing any given control rod in the LACBWR and therefore the many independent opportunities for detection of a suck rod, DPC believes that a probability of 10-8 for ND is reasonable.
1 l
4 l 1
_