ML20198C090
| ML20198C090 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 08/19/1985 |
| From: | Marotta C NRC OFFICE OF NUCLEAR MATERIAL SAFETY & SAFEGUARDS (NMSS) |
| To: | Weller R Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML20198C081 | List: |
| References | |
| NUDOCS 8511110292 | |
| Download: ML20198C090 (5) | |
Text
.
APPENDIX I
- "'%g#
UNITED STATES
-. t NUCLEAR REGULATORY COMMisslON WASHINGTON. D. C. 20555 g, %tJ Af AUG1 9I!ES t
MEMORANDUM FOR:
Richard A. Weller, Leader Safety and Environmental Review Section Three Mile Island Program Office Office of Nuclear Reactor Regulation THRU:
Charles E. MacDonald, Chief
/}
Transportation Certification Branch Division of Fuel Cycle and Material Safety Office of Nuclear Material Safety and Safeguards FROM:
Charles R. Marotta, Senior Criticality and Shielding Engineer Transportation Certification Branch Division of Fuel Cycle and Material Safety Office of Nuclear Material Safety and Safeguards
SUBJECT:
CRITICALITY REVIEW 0F TECHNICAL EVALUATION REPORT (TER) FOR THE TMI-2 DEFUELING CANISTERS AS DOCUMENTED IN REFERENCE A (BELOW)
REFERENCES:
A.
TER (15737-2-G03-il4, Rev. 0) dtd 03/22/85 TMI-2 Division Technical Evaluation Report for Defueling Canisters l
1 B.
Three (3) IBM Computer Listings (B&W property to be returned to B&W):
1.
KENO-IV, 123 Gps, Gen. Geom: Damaged fuel Canister 2.
KENO-IV, 123 Gps, Gen. Geom: Damaged Filter Canister
^
3.
KENO-IV, 123 Gps, Gen. Geom: Damaged Knock-out Canister J,
(delivered to NRC on May 23,1985)
C.
Three (3) B&W fiche copies of the above listings 1
giving nuclear data and geometric details (NRC property; delivered to NRC on May 23, 1985)
I.
Introduction and Conclusions As requested in your memorandum to C. E. MacDonald dated April 24, 1985, a detailed review has been performed of the submitted GPU (Refs. A and B, B&W analyses) criticality Safety Analysis for the loading of canisters in the defueling of the TMI-2 core. Based on this review, we find that the criticality calculational method, physical and geometric assumptions, atomic number densities (giving mass loadings of nuclides per region) and description of canisters analyzed to be accurate and represent the cases intended.
8511110292 851105 fR ADOCK 05000320 PDR
~
Richard A. Weller AUG I 91985 In addition to this detailed review and verification, independent KENO-IV Monte Carlo calculations of the knock-out and fuel canisters were performed. The independent calculations agree with the results obtained by B&W as given in Ref. A.
A comparison of NRC and B&W's keff's under various conditions is given in Table 1.
Since the filter canister contains a similar 2" diameter B C central poison 4
rod as in the knock-out canister and in addition contains about ten times the amount of internal steel of that in the knock-out canister, the filter canister was considered less reactive than the knock-out canister and hence not analyzed by NRC.
We, therefore, recommend acceptance of the criticality analysis portion of Ref. A and concur with the subject submittal that there exists at least a 5% shutdown margin for all three canisters under nomal and assumed accident modes.
i In Table 1, below, and in Ref. A, we note that 'l&W did not report 1
any keff's for 8,C replaced by water or replaced by a void. NRC calculated a sin 31e knock-out canister to have a 4.3% shutdown when the B C is replaced by water; a 3.8% shutdown when the B,C is 3
replaced by water and the remaining steel tubes are defldcted off-center by 1.2 inches. We note that these latter two cases are supercritical for the infinite array calculation as given in Table 1.
Thus, if the above scenarios can be realized in the postulated accident modes, the 5% margin shutdown is compromised. Further, if a void replaces the B C, the shutdown margin is further reduced 4
from 4.3% to 3.4%.
In sumary then, we find:
1.
The B&W calculational methodology (KEN 0 IV-123 Group GamThemos cross-sections) represents one of the best state-of-the-art approaches which has successfully calculated many appropriate benchmark criticals.
In particular, we note that the B&W fuel-water homogenization procedure - fundamental to the B&W approach and results - has been done correctly.
2.
The B&W criticality analyses used the most (neutronically) i reactive fuel / water mixture in representing the core debris in each canister.
3.
Some conservatisms used by B&W were:
(a) Each canister was loaded up to a height of 14 feet (~ an i
extra 3 feet of reactive material).
l (b) The density of B C was taken as 1.35 gm/cy, areal density of B-10 for bora was taken as 0.04 gm/cm 1
~
Richard A. Weller AUG 19 M (c) The minimum amount of steel has been credited to the knock-out canister (~ 1-1/2 volume percent) and the filter-canis ter ( ~ 14 volume percent).
t 4.
NRC independent calculations agree very well with the B&W results for the cases considered.
A brief discussion of the criticality methods used to establish the conservative acceptable parameters fundamental to both B&W's and NRC's follows.
II. Basic Assumptions and Methods Used in Criticality Calculations i
Both B&W and NRC assumed the TMI-2 debris contents for all three canisters to be U(3)0 with a volume fractio $ unciad pellets moderated by unborated H 0 3
of 0.30 of fuel and 0.70 of water. Thit has J
been established via many independent calculations to constitute the most reactive mixture. For a borated water system over the range of 3000 to 5000 ppm boron in water, the most reactive mixture turns out to be a volume fraction of 0.60 of fuel and 0.40 of j
water.
However, for these borated systems, the keff is of the order of 30% less than any corresponding system moderated by unborated water. Thus, the 4 k is of the order of 0.3 and completely controls selection of the most reactive mixture to be fuel moderated by unborated water. All criticality calculations thus use unborated water as moderator.
Both B&W and NRC assume a very conservative density for B C viz 1.35 gm/cc versus 2.43 gm/cc given in the handbooks.
In$ddition, an areal density of 0.04 g/cm for B-10 is assumed for boral.
l Both B&W and NRC use the KENO-IV Monte Carlo computer program with
].
the 123 group Gam Thermos neutron cross-section set adjusting the resonance nuclide (U-238) with the NITAWL program. B&W then homogenizestheU(3)0 and the associated water (30/70 mixture) via an X5DRN cell group-sp,atial weighting into a debris mixture.
t i
Using j
generalized geometry, this homogenized water-fuel mixture occupied j
all space within the boral plates of the fuel canister, all space inside the knock-out canister not occupied by the 5 B C SS clad i
rods and all space inside the 17 filter elements of t$e filter i
canister.
t As a check on the above homogenization procedure NRC's model required that the U(3)0, pellet be described as a discrete cylinder surrounded by the cell 130/70) water. This restricted NRC's canister's geometry to a square-cylinder. The pellet-water constituted a box-type in the KENO-IV geometry, and since the fuel canister possesses a square internal region (surrounded by boral) which will contain the debris, it represents the ideal case to check the homogenization process fundamental to B&W's calculated procedure. Results of Table 1 under Fuel Canister show that the homogenization procedure of B&W and the discrete procedure of NRC to be equivalent - they calculate the same keff for the single fuel canister and for an a rray.
Richard A. Weller AUG 19 I!B5 For the knock-out canister, the square cylinder geometry of NRC maintained the exact masses of UO, H 0, steel and B,,C that exist 7
2 in B&W's cylindrical geometry. Table 1 for the undaihaged single knock-out canister, B C in place shows excellent agreement; for the infinite array, the NkC value of kao is higher by~ 4-1/2% since in this geometry the square box ends come much closer to neighboring boxes whereas the cylinders remain effectively further apart from one another.
The damaged cases for NRC were calculated 'y assuming the B C b
being replaced by water whereas B&W assumed only a displace $ent of singl,C-SS rod. Although NRC's condition is more severe, the the B e damaged knock-out canister is still subcritical, but the infinite array of such damaged canisters is supercritical.
NRC's worth of the B C can be estimated from Table 1:
4 For Single Canister For the Array a k, 0.957-0.887 1.033-0.961
= 7.6%
= 7.2%
k
.926
.997 g/~e?. S~wt__-
Charles R. Marotta
e TABLE 1 C9tPARISON OF B&W AND NRC CALCULATED MAXIMUM
- Keff's
-FOR KNOCK-0UT AND TUEL CANISTERS
(
CONTAINING TMI-2 DEBRIS Using KENO IV with 123 Group GamThemos Neutron Cross-Sections when each canister is loaded with most reactive U(3)0,,/H,0 mixture **
STATUS OF KNOCK-00T CANISTER FUEL CANISTER N
UNDAMAGED DAMAGE 0 UNDAMAGED E1[tIEATED (B C in Place)
(B,C repl by H O)
(B C repl by H.,0 (B C & SS displ (Boral in Place) 4 2
4 4
& SS displ 1.2")
0.75")
NRC B&W NRC B&W NRC B&W NRC B&W SIEE CANISTER 0.887(a) 0.873 0.957(b) 0.962 0.882 0.866 0.857 (H,0 flooded an8 reflected) 0.966(d)
IIF. ARRAY OF CANISTERS 0.961(CI 0.915 1.033 1.041 0.872 0.877 (17.3" c to c l
spacing in H O pool) 2
- Assamed fuel vol/ water vol = 30/70, fuel as pellet in unborated water.
2 All cases usedf(B C) as 1.35 gms/ce; BORAL assumed 0.04 gm B-10/cm 4
(a) NRC Calc. for this case with 3000 ppm boron in H 0; keff = 0.582 - no steel in canister.
(b) NRC Calc. for this case with 3000 ppm boron in H 0; keff = 0.646 - no steel in canister.
(c) NRC calc. for this case with 3000 ppm boron in H 0; keff = 0.618 - no steel in canister.
(d) B C replaced by a void.
4
.