ML19308B497
| ML19308B497 | |
| Person / Time | |
|---|---|
| Site: | Oconee, Crane  |
| Issue date: | 05/08/1973 |
| From: | Stello V US ATOMIC ENERGY COMMISSION (AEC) |
| To: | Deyoung R US ATOMIC ENERGY COMMISSION (AEC) |
| References | |
| NUDOCS 8001080707 | |
| Download: ML19308B497 (36) | |
Text
_
l t3Y G G73 h'
R. C. DeYoung, Assistant Director for Pressurized Water Reactors, L ECCS INPUT FOR T11REE MILE ISLAND UNIT 1 SAFETY E Enclosed is a drafC of the ECCS evaluation for the Three The first draft of this Unit 1 (TMI-1) Safety Evaluation Report.
20, 1973) directly to the LPM evaluation was transmitted (AprilOur review and evaluation of the ECCS perf at his request.
i were based extensively on material in Duke Power's Oconee Operat ng
- Also, License application, and the similarity of UtI-1 to Oconee.
1 some of our conclusions are based on information not formallyTo meet your s documented (Oconee or UfI-1) at this time.
needs we have, however, supplied you our evaluation based ou our uderstanding of what will be documented shortly.
The conclusion of our evaluation (without considering the effects ll of fuel densification) is that the ECCS is acceptable ad. wi provide adequate protection for a nominal power level of 2568 MWt.
bridnal 51gned liy Ildor Stdio V. Stello, Assistant Director for Reactor Safety Directorate of Licensing DISTRIBUTION Central File RP Rdg PWR-4 Rdg DKDavis Dross VStello
...L: /CP.-
-L:A ES I.1.
\\
omer >
. SWR-V8 d' sikmf
[.......]'.
/73
../....Y....'.. /.7 3 summt >
I our > /../....../......../. 7. 3...I.l. \\
f" era echt nies-a ess-ers Fone ABC.)le LRev.9 53) AECM 0240 So"' 8 70 7 6
. l, '
t.
a OUTLINE
- I t
6.3 Emergency Core Cooling System 6.3.1 General i
6.3.2 System Description 6.3.3 Performance Evaluation 6.3.3.1 General 6.3.3.2 Analysis of Blowdown Period 6.3.3.3, Analysis of Refill & Reflood Period 6.3.3.4 Results 6.3.3.5 Conclusions 6.3.4 Small Break Analysis 6.3.4.1 General 6.3.4.2 Small Break Model 6.3.4.3 Results and Conclusions 6.3.5 Core Flooding Tank Line Break 6.3.5.1 General 6.3.5.2 ' Applicant's Calculations 6.3.5.3 Staff Calculations 6.3.5.4 Heat Transfer Analysis 6.3.5.5 ' Conclusions 6.3.6 Concit ions
' References i
' l s
e..
r e-l-
Emergency Core Cooling System 6.3 6.3.1 General In 1971 the Atomic Energy Commission reevaluated the theoretical i
and experimental bases for predicting the performance of emergency 4;
i core cooling systems ' (ECCS), including new information obtained from As a result of industry and AEC research programs in this field.
the Commission developed interim acceptance this reevaluation, criteria for emergency core cooling systems for light-water power i
These criteria are described in an Interim Policy Statement reactors.
and published in' the Federal Register on issued on June 25, 1971, June 29, 1971 (36 F.R. 12247).
By letter dated July 9,1971, the Division of Reactor Licensing informed the applicant of the
' additional-information that would be required for our evaluation of d
e the performance of the Three Mile Island Unit 1 ECCS in accor anc The applicant provided a revised _f with the Interim Policy Statement.
1 ECCS performance in report analysis of the Three 1 Ele Island Unit BAW 10034 titled "Multinode Analysis of B&W's 2568-MWt Nuclear Plants During a Loss-of-Coolant Accident" dated October 1971 inclu The analysis was performed using the B&W Revisions 1, 2 and 3.
Evaluation Model in conformance with the Interim Policy Statement, In the analysis it was assumed that a loss-of-Appendix A, Part 4.
l pove r coolant accident occurs during operation at'102% of a nomina I
i i
I 1
1
~
l II 1
6 1
)
O level of 2568 MW thermal as compared to the lower pvicer level of 2535 Ini thermal requested by the applicant.
Subsequent to the staff's review of the analyses presented in BAW-10034, several additional topics associated with ECCS performance were identified during the staff's review of the operating license application for Duke Power. Company's Oconce Unit 1 (Docket 50-269,.
These topics include:
(1) the reflooding analysis ensociated with I
a loss-of-coolant accident; (2) the analysis of small breaks in the primary cooling system; and (3) the analysis of a break in the core flooding tank (CFT) line.
The staff has reviewed the Three Mile Island Unit 1 ECCS and' based on the similarity to the Oconce ECCS find the information and evaluation of the Dconee ECCS performance applicable to Three tule Island Unit 1.
6.3.2
System Description
The Three Mile Island Unit 1 cmergency core cooling system c
iT consists of a high pressure injection system, an injection system employing core flooding tanks, and a low pressure injection system with external recirculation capability outside containment. Various combinations of these systems are employed to assure core cooling for the complete range of break sizes.
1 i
The high pressure injection system includes three pumps, each I
capable of delivering 500 gpm at 600 psig reactor ve ;sel pressure and R
discharges to the reactor coolant inlet lines. One pump will provide h
1
si i
(
o e
the required minimum flow for the high pressure injection system.
The high pressure injection pumps are located in the auxiliary building A concentrated boric acid solution from adjacent to the containment.
the boric acid water storage tank is provided to the suction side of the high pressure pumps during ECCS operation. During normal reactor operation, the high pressure injection system recirculates for purification and for supply of seal' water to the reactor coolant 1
reactor coolant circulation pumps.
The high pressure injection system is actuated at a low reactor coolant system pressure of 1500 psig with a backup signal of 500 psig, or a reactor building pressure of 4 psig.
Automatic actuation switches the system from normal to emergency oper-One of the three high pressure pumps is normally in oper-ating mode.
The system is designed to withstand a single failure of an ation.
active component without a loss of function.
f The two core flooding tanks are located in the containment outside of the secondary shield.
Each accumulator has a total volume 3
3 with a nominal stored borated watcr volume of 1040 ft of 1410 ft Each accumulator is connected pressurized with nitrogen to 600 psig.
to a separate reactor vessel core flooding nozzle by a flooding line Incorporating two check valves and a motor operated, normally open stop valve adjacent to the tank. Therefore, the core flooding tanks
-will-inject water automatically whenever the_ pressure in the primary system is reduced below t'ae core flooding tank pressure of 600 psig.
~
t t !
I
-.1
-J
.1 e
f 8
The low pressure injection system is actuated on low reactor coolant system pressure of 1500 psig or 500 psig, or high reactor rs Mc W To protect the _yrtee from cv2;prc curi--
building pressure of 4 psig.
y M
asesen, the pumps operate in the bypass mode until the reactor coolant system pressure decreases below the pucp discharge pressure.
The low pressure injection system includes two pumps, each capable of delivering 3000 gpm at 100 psig reactor vessel pressure, i
arranged to deliver water to the reactor vessel through two separate injection lines.
One low pressure injection pump is capabic of removing the heat energy generated after a loss-of-coolant accident.
The low pressure injection system pumps initially take their suction from the borated water storage tank and, later, during recirculation from the reactor building emergency sump.
The re-circulation system components are redundant so as to withstand a single failure of an active or passive component without loss of
,7 function at the required flow.
All of the ECCS subsystems can accomplish their function when operating on emergency (onsite) power as well as of fsite power.
If there is a loss of normal power sources, the engineered sa'e-guards power line is connected to the emergency diesel generators which have a startup time of 10 seconds or less.
The pumps and valves of the injection system will be energized at less than 100%
l voltage and frequency to achieve the design injection flow rate within 25~ seconds. '
il
n_
Q:'
6.3.3 Performance Evaluation _
6.3.3.1 General y vm'.sd* *JWe hans developed a set of conservative assumptions and p s.
d to be used in conjunction with the Babcock and Wilcox develope The assumptions and procedures codes to analyze the ECCS. functions.
l d used by B&W in analyzing the performance of the Three Mile Is an Unit 1 ECCS are described in Appendix A, Part 4 of the Interim Policy Statement published in the Federal Renister on December 18,
~
1971 (F.R. Vol. 36, No. 244).
Report BAW-10034 "Multinode Analysis of B&W's 256t, MWt Nuclear Plants During a Loss-of-Coolant Accident October 1971, covers the performance of cores for which the fuel.
rete is 18.15 kW/f t.
pins are pressurized and the peak linear heatn./ val.pd of A Eus mJa msk.;.y AsQ(Duh+ Rr1 so-s)
As a rasvit *+ ak msNe g@, the staff has requested a reanalysis of the reflooding From this analysis,
transient using a more conservative assumption.
cold leg split is the limiting case accident with a 2
E the 8.55 ft For comparison, the peak linear peak clad temperature of 2186*F.
heat rate for Ihree Mile Island Unit 1 is 17.63 kW/f t and the core power is 2535 MWt.
6.3.3.2 Analysis of Blowdown Period The applicant used the CRAFT and THETA 1-B computer codes for Using these the analysis of the blowdown phase of the transient.
codes, and the evaluation model specified in Appendix A, Part 4, t
i
"'M8;
1.-.
~
~
=
-of the Interim Policy Statement, the applicant provided the re-evaluation of the ECCS performance in compliance with the Commis-sion's Interim Policy Statement.
For the blowdown portion of the accident, we have concluded that the applicant's analyses as reported in BAW-10034 conform to the requirements specified in the Commission's Interim Policy State-ment, Appendix A, Part 4.
l Analysis of Refill and Reflood Period 6.3.3.3 The applicant has considered the thermal behavior of the core during the refill and reflood portion of the loss-of-coolant accident, which is explained as follows:
1.
The vessel refill is provided initially by the core flooding tands, and later by the pumping systems, and is assumed to start at the end of the blowdown period. ihe reactor vessel g "'is assumed to be essentially dry at the end of the blowdown e
period a result of the conservative assumption in Appendi (Part4,oftheInterimPolicyStatement stat Yater injected by the core flooding tanks prior to the end of blowdown is ejected from the primary systemy.nd C
o 2.
No heat transfer in the core is assumed until the level of water reaches the bottom of the core, at which time refill is considered complete and the core reflood starts.
4 4
E44%d' o
~
rupture for the f
The end of blowdown is 14.6 seconds a ter l
d refill to the bottom
t cold leg double-ended break, an l
2 The 8.55 ft ds after rupture.
of the core is complete' about 23 secon upture for the 8.55 2
end of blowdown is 18.7 seconds af ter rl te a i
}'
<a cold leg split, and refkums is comp e
(
rupture.
d initially by c rapid The reflood of the core is characterize i
the vessel annulus 3.
liquid level rise both in the core and in d to generate substantial until enough of the core is covere k
The core reflooding rate increases and pea s amounts of steam.
f blowdown at about 11 to 12 in about 10 seconds af ter the end o leveling of f at about l
inches per second, then decreases rapid y, he end of blowdown.
2 inches per second about 30 seconds af ter t&
d in the core t. seat 6 1:1 (tw pdh u Won 4 steam generated
.f 4.
The am" ~
through the vent valves l
s W is assumed to be on ydit is taken for steam flow within the reactor vessel and no cre h
te The steam flow resistance also limit around the loop.
lus water level continues of liquid rise in the core, but the annu Core d level reaches the inlet nozzle.
to increase until the liqui stem water are piped f1 ceding tanks and low pressure injection sy vening reactor directly to the reactor vessel with no inter coolant system piping.
i s
e 1
1
.s 3
\\
~}
i t
.o t
cify The peak temperature reached in the transient for the limiting
-5.
4 8.55 ft cold leg split occurs about 30 seconds af ter the break.
-' l The staff has reviewed the reflooding analysis using a new carryover rate fraction
- correlation developed by B&W during the course of the Q\\
rulemaking hearing (Docket RH50-1) to account for the entrainment of V
reflooding water. The previous reflood analysis performed by B&W (BAW-10034) used an entrainment assumption of 20% of the inlet core i
The 20% entrainment assumption was based on the FLECHT flow rate.
j The staff requested a reanalysis of the reflooding program.
transient using the new CRF correlation in its letter to Duke Power
\\
Company of November 3,1972. Because the new carryover rate fraclion correlation took many FLECHT experimental runs at different conditions into account, the staff views it as a better approach in calculating reflooding races.
The staf f has reviewed the B&W reflood code (REFlOOD) and has e
compared its results with those of the FLOOD 1 code (an ANC/AEC reflood program)'. Reflooding rates predicted by both computer programs agree to within 1%, when the REFLOOD code uses the new carryover rate function to predict the entrainment. If the old entrainment assumption of 20%
is used, the flooding rates calculated by REFLOOD are higher than those
. predicted by FLOOD 1.
of
- The carryover. rate fraction (CRF) is defined as the total core flow rate out
- the top of the core divided by the total mass flow into the bottcm of the core.
a t
.t-
.i 6
')
'. i 6.3.3.4 Results B&W has recalculated the reflooding rates and heat transfer 4
he new coef ficients for several break locations and sizes using t i
The heat transfer coefficients carryover rate fraction correlation.
i d from used in determining the peak clad temperature were determ ne h the new, lower the FLECHT correlation presented in WCAP-7665, wit Peak cladding temperatures calculated by the new I
reflooding rates.
t for f
reflooding rates are higher, and remain at elevated tempera ure However, both the maximum clad temperatures longer time periods.
ithin the limits and the percent metal vater reaction calculated are w forth by the Interim Policy Statement on'ECCS.
set The staf f also requested on November 3,1972 analyses of the i s analyses effect of a higher elevation axial flux peak (the prev ou were donc for an inlet flux peak). The higher elevation peak c
mol;f el (f sine flux peak) resulted in a slightly lower peak cladding The greater metal-temperature, but a greater metal-water reaction.
CS fluid water reaction is due to the extra time required for the EC to advance to the higher elevation.
The following table summarizes the calculational results using correlation at 102% of a the carryover rate fraction entrainment (all cases are for an inlet flux nominal power level of 2568 MWt peak except where'noted):
f s
&twy-
4 i
t l
l t
i Metal-Water Reaction, l
Cold Leg Pipe Breaks __
a Peak Cladding _
Area Type Break Temperature Local Core 8.55 ft D uble Ended 2082 2.11
.075 p
2 2
2186*
2.98
.09 8.55 ft Split (cosine flux 2135 4.2
.24 0
S lit.
8.55 ft P
2 f
Peak) 5.13 f3 Double Ended 2029 1.8
.058 2
3.0 ft Split 1728
.046
.01 2
1560
.22 004 g
0.5 ft split Hot Leg Pipe Breaks _
2 1670
.14
.003 14.1 ft Split
- Limiting Case 6.3.3.5 Conclusions The use of the new carryover, rate fraction correlation provides a than more conservative method of predicting reflood water entrainment the 20% entrainment assumption since the use of this correlation results in lower reflooding rates, higher peak cladding temperatures and greater The staff has also concluded that, based on the' metal water reactions.
present experimental data, the use of this more conservative approach is warranted.
The staff concludes that the ECCS performance analysis using this more conservative approach meets the acceptance criteria,
]s described in the Commission's Interim Policy Statement.
6.3d SMALL EREAK ANALYSIS
/
.'3.4.1 General The Interim Policy Statement concerning emergency core cooling in the event of a loss-of-coolant accident (LOCA) required the analysis of e
e
7 f,.-
i i
(
l The B&W evaluation model in LOCA's over the entire break spectrum.
Part 4 of Appendix A to that statement specified an acceptabic evalua-
- 2 tion model for break sizes from 0.5 f t up to and including the double-
.j 2
ended severance. of the largest pipe of the reactor coolant pressure B&W submitted a Topical Report, boundary (the large break model).
/
BAW-10052, "Multinode Analysis of Small Breaks for 2568 MWt Plants" on September 22, 1972 to present its evaluation model for breaks less The staff has completed the evalvation of this report.
than 0.5 ft in general, small breaks are less limiting accidents than the i
larger design basis breaks.
Also, in addition to the emergency core 7 % A;c..cq-4I cooling system, the specific B&W reactor' design used in tr.c 4 ty-contains internals vent valves which further mitigates the
.cc:
For LOCA -con, sequences, including those caused by small breaks.
cold leg breaks these vent valves prevent a hot leg loop seal from forcing the water level in the core to drop excessively due to steam
.r A low water level in the core could cause a core heatup binding.
transient due to degraded heat transfer.
By venting the reactor the steam generated by upper-plenum to the downcomer annulus, leg depressurization, and by core heat transfer can bypass the hot flow path, if blocked by a water seal, and flow through the vent valves out the cold leg break to the containment.
6.3.4.2 sy;11 Break Model B&W developed a procedure for analyzing the consequences of.
small breaks which differs somewhat from that given in BAW-10034, t
j.
j i
1 l
during
{
Revision 3 "Multinode Analysis of B&W's 2568-MWt Nuclear P ants These methods are similar to these a Loss-of-Coolant Accident."
f:
ccount for used for large breaks but is dif ferent in some aspects to a l
for smaller breaks.
a more tranquil hydrodynamic response of the systems l
break j
Thase differences between the small break model and the large f
d evaluation model have been reviewed and evaluate.
I i
i l te The CRAIT code, as described in BAW-10030, is used to s mu a l break models. The the hydrodynamic response for the large and smal i
I h small break number of nodes representing the primary system for t e h
condary. system model has been reduced to 11, with one node for t e se Additionally, the Redfield and one node for the reactor building.
030 and BAW-10034 was variabic bubble rise model* described in BAW-10 d a zero bubble rise used in all nodes; the large break model assume the upper plenuta ar.d the pump model in the lower head, the core, During a large break this zero bubble rise model would suction nodes.
due to the be more appropriate for those nodes where good mixing occurs rapid depressurization and high flow rr.cs.
hl For the associated heat transfer analysis a THETA model slig lled heat transfer smaller in nodalization is used during the flow-contro For one case examined, this change resulted in only a 7 transient.
break dif ference (in the conservative direction) between i
.j *nrst,*A zero bubbic rise velocity yields a homogeneous node
( : the bubble velocity tends to separate the water phases.
./
1
j m
y.
\\.
. ~
l j i-J I
f c
When core flow f
1NETA model and that used for large break analysis.
drops below 1% of its initial value and flow no longer controls heat I
4 QUENCH has a one transfer, another heat transfer code QUENCH is used.
f axial node,- one clad' node, and one fuel node code; it assumes heat to be transferred by either pool film-boiling or by forced convection Multiple QUENCH runs are made at various axial locations
.I to steam.
Morgan's correlation to obtain the thermal response of the fuel rod.
l for pool film. boiling' is used for that portion of the core covered by I
This correlation is the best available a mixture of steam and water.
It was derived from a for pool film boiling from vertical surfaces.
d to theoretical model of the stable annular flow regime as compare i
the dispersed flow film boiling gegiac, and it is therefore conservat ve The correlation underpredicts the available data for in this. regard.
The pool film boiling from vertical surface for a variety of fluids.
Dittus-Boelter correlatien is used for that part of the core covered in the steam-flow region the average steam flow is conservatively by steum.
calculated for the fuel heatup calculation, with the fluid temperature calculated by hand.
A major dif ference between the small break model and the large brea l
model is' the absence of any arbitrary bypass in the small break mode of core flooding tank (CFT) injection water prior to the end-of-blowdown a
e a
N 1
c..
3 b
I 1,
I The CFT bypass assumption would be unduly conservative in small break
~
analysis since the velocity of fluid in the downcomer is too low to entrain CFT injection water 'and sweep it out the break.
Since the core is never completely uncovered for small breaks,
~
) is not the reflood analysis (which is conducted for larger breaks The reflood analyses and the previously discussed CFT bypass f
done.
A comparison of a 0.5 ft assumption are, however, interrelated.
break analyzed by both the large break model (CFT bypass assumption The two models agree and the small break model was conducted by B&W.
i very well until the CFT bypass assumption. is imposed for the large This results in a calculated peak clad temperature of break model.
1660*F in the reflood transient associated with the large break model, compared, to a peak clad temperature of only 710*F using the more realistic, yet conservative small break model.
c 6.3.4.3 Results and Conclusions
/
The results of B5W's small break analysis for plants at a core power of 2568 MWt are contained in B&W topical report BAW-10052, "Multinode Analysis of Small Breaks for 2568 MWt Plant" and in a' i
letter from A. C. Thies of Duke Power Company to December 19, 1972 A summary of these results is given below:
the AEC.
i i
d
)
. g' e
y e
e-t-p s
4
n I..
s' L
h i
t Long Term 6
Cooling Established *,
Peak Clad Break Size
- F sec Temperature.
and Location s
400 l
O.5 ft2 (pump discharge) 710 I
1100 0.3 ft2 (pump suction) 780 2500 0.1 ft2 (pump suction) 826 3400 0.1 ft (pump discharge) 720 3000 0.04 ft (pump suction) 978 All conditions of the interim acceptance criteria have been met, is little the peak cladding temperature is well below 2300*F, there ble and or no metal-water reaction, the core geometry is still coola On the basis of our evaluation long term cooling can be established.
the emergency core cooling of these analyses, we have determined that h
i ry system will provide adequate protection for small breaks in t e pr ma cooling system.
K ANALYSIS _
r
. CORE FID0 DING TAMK LINE BREA 6.3.5 6.3.5.1 _ceneral in the course of the staff's safety review of several plants with This B&W reactors, a potentially serious accident was identified.
f the two postulated accident involves the double-ended break of one o l
lines which connect a core flooding tank (CFT) to the reactor vesse.
h he core is
- Long term cooling is established in the applicant's opinion w en t kd the pressure covered with mixture,' more water is being supplied than lea e,
f
~
is stabilized and the cladding temperature is f alling.
O 9
9
.4-
}
t t
l
.~
3 to its oconee application dated January 29, 1973 submitted an b
analysis showing the ef fects of the flow limiters for this accident.
A summary of these results is presented in Section 6.3.5.2.
]
t t
~
the staff has In cvaluating the consequences of this accident,
~
conducted independent calculations using the RELAP, T00DEE and SWELL Aerojet Nuclear computer codes, with assistance from our consultant, A summary of these independent calculations is Corporation (ANC).
included in Section 6.3.5.3 of this Safety Evaluation.
I 6.3.5.2 App:.icant's Analysis B&W supplied the results of their anal,ysis of a postulated CFT 39 to the Duke, line break accident for an Oconee reactor by Amendment application dated January 29,-1973.
In conducting this analysis, B&W used the small break evaluation model, BAW-10052, described in Section There were several changes to this small break 6.3.4 of this report.
model for the CFT line break analysis due to the unique break location. g.
The most significant changes involved tuo additional nodes in the downcomer annulus and increasing the size of the core node to include most of the upper plenum volume.
One important parameter in this analysis is the amount of water This determines the remaining in the vessel during this transient.
height of fluid in the core, and therefore, the heat transfer capa-To determine bility' of the core and maximum cladding temperature.
C e
e "f}
Md
~_
l "n.
p L_
b tion (LPI) piping to g
j These lines also connect the low pressure in ec I
Assuming no offsite power and a single active f*
the reactor vessel.
rgency power), the f ailure (such as in one of the buses supplying eme
)
high pressure injection i
ECCS would be degraded to only one CFT and one h
d accident i
Although extremely unlikely, this postulate E,
(llPI) pump.
the requirements of D
h degrades the ECCS to the minimum equipment w en10 CF General Design criterion 35, Appendix A to This postu-f fsite power) are imposed.
- stand a single failure with no o if suf ficient water to lated accident would be particularly severevessel during this cool the core does not remain in.the reaecorin rela-This is due to the low capacity of one HPI pump accident.
nd also to supply sufficient
~
tion to the need to reflood the core aheat and sto make-up cater to compensate for decay primary system.
l during this acci-f In an effort to retain more water in the vesse limiting orifices in the the applicant has installed flow This modification reduces the break siz
to 0.44 ft from 0.72 f t core during this acci-h several more feet of liquid to remain in t e f f's questions on' The applicant's oris.a1 response to the sta u
BAW-10034 Supplement 1, is no dent.
~ this accident, B&W Topical Report, alysis without longer applicable since this report presents an an 39 However, Duke Power Company, in Amendme
. flow limiters installed.
4
\\
y
~,
i.
V l Il I.
I h
I the collapsed core water level (level if there were no heat addition)
B&W used three different CRAFT models to determine the sensitivity of l
the level prediction to noding. The different models provide good agreement, with the lowest quasi-equilibrium liquid level approxi-e mately at the six' foot elevation. When the liquid swell due to core heat generation is' considered, the mixture (water and steam) Icvel covers the core for the most of the accident.
In addition to these CRAFT models which used the Redfield bubble rise model, B&W used a higher bubble rise velocity in one model which would be more con-sistent with the two phase mixture height. predicted by B&W's FOAM code l
(described in Section 6.3.5.4).
This CRAFT model prevented the two phase mixture from being lost through the vent valves and the break.
The liquid level increased to the 9-10 foot core elevation.
B&W's calculations indicate that only the upper part of the core is not covered by mixture during this transient, but sufficient steam _ g is generated by the covered portion to cool this uncovered part.
Since the lower portion of the core is covered with a two-phase mixture, pool film boiling provides sufficient cooling here and the maximum cladding temperature occurs in the upper uncovered portion of the The upper portion when not covered by mixture is cooled by core.
forced convection to steam. To establish the maximum cladding temperature, B&W investigated several. axial power peaking shapes.
i A summary of these results is provided below:
e
Elevation of Peak Elevation of power Cladding Temperature Peak Cladding peak from the. bottom from the bottom of core, Temperature, of dore, ft ft
- F su 5.5 5.5 731 7.8 11,.4 964 10.6 11.4 1199 These cladding temperatures would produce essentially no metal-water reaction and the core geometry would remain unchanged except possibly for some minor clad swelling in the case of the 10.6 f t power peak.
6.3.5.3 staff Calculations h
An independent analysis of B&W core flood tank line br, sk has been performed by the Regulatory staff to aid in the evaluation of this postulated accident.
The analyses have considered both the blowdown hydraulics and the heat transfer phenomena resulting from
{
l the predicted core water level, c
The staff has performed several blowdown analyses using the RELAP computer program.
These analyses included both a modeling study and a determination of the sensitivity of the analyses to the I
bubble rise model. To perform these studies, several system noding models_were developed.
A summary of these models is presented in I
Table 1.
There were three basic models used in the analysis.
The e
first (LARGE MODEL) was a 36 node model' previously used to perform a i
large break analyses.
- This model used excessive computer time for
\\\\
g; 1
. j-i
[
6 small break analysec, but it was used as the basic comparison model between other small-break models. This model had seven heat transfer t
nodes in each steam generator and three core nodes.
Also, it had all cold and hot legs noded separately.
A second model (REDUCED MODEL) was generated to study azimuthal noding in the downcomer region.
It consisted of 2 separate primary loops with the hot legs combined to reduce computer running time.
-l Also the number of heat transfer nodes in the steam generators were j
reduced from 7 to 3.
The third model (SMALL MODEL) was developed to perform downcomer axial noding studies.
The two hot legs and ld legs were com-bined to form a single loop with one steam generator containing two heat transfer nodes.
To insure that each model predicted the same blowdown characteristic, the two smaller models were compared to. the large 36 node model (standard model used for comparison). The pres-sure transients calculated by these three models are presented in Table 2.
This table shows that each model predicts very similar pressure results.
Af ter comparisons had been performed, an axial noding study was made for the downcomer region using the small model.
Investigations into the effect of using a baoble rise assumption (compared with a homogeneous assumption)- and the number of downcomer node was per-
. formed. Also, the effect of bubble rise velocity (V ) n the blow-B down characte,ristics was investigated.
\\
I
w,,
.w:=
-q l
TABLE 1 SU> DIARY OF RELAP COMPUTER MODELS Number of Number of Number of Number of Downcomer Nodes
~ Description'.
Core Nodes Hot Leg Cold Leg Number of Steam Gen Nodes Basic Blowdown ModA1:
del Size 4
1 3
2 7 in each-Used to Perform Radial rge Model.36 Nodes 3
2 2
1 Downconer Noding Study 2 in each
. duced Model 21 Nodes Used to Perform An 1 3
1 1
1 Douncomer Noding Studg 2 (Both Loop Combined) all Model 35 Nodes Homogeneous Dotmcomer -
2 3
1 1
2 Lower Dotmcomar Node '
all Model 1
2 3
1 Homogeneous Bubbic.
2 Rise in Upper tall Model V = 3 'ft/sec 3
Lower Downcomer Node 3
1 1
2 Homogeneous Bub' ale nall'Hodel Rise in Upper 2
V = 5 ft/sec B
All Downcomer Node '
3 1
1 4
llomogeneous 2
nal1 Model Downcomer Node Bubb1G 3
1 1
1 Rise 2
nail Model Break Area = 0.44 ft 1
1 2
3, All Downcocer Node 2
sail Model 3
2 2
4 Homogeneous 2
aduced Model All Downcocer Nodes 3
2 2
8 Homogeneous 2
.zduced Model
+
.e
~_._.
- e-t
=
TABLE 2 VESSEL PRESSURE COMPARISON 4
FOR THREE STANDARD MODELS
- Time-Large Model Reduced Model Small Model Sec 36 Nodes 21 Nodes 15 Nodes 0
1 1597
.06 1604 2
1588 617 1617 5
1583 1637 1637 i
10 1507 1504 1501 i-15 1327 1347 1345 20 1177 1153 1153 30 1101 1060 1060 40 1043 993 933 50 966~
928 928 60 864 849 849 70 734 745 746,
80 587 600 610' 90 409 432 429 100 337 262 269 5
4 m
e
-g_'
t l,
j i
ihede:
I l
9 i
r W
+-t
-"7 EP
y The first effect to be investigated was the assumption of using a bubble rise vs a homogeneous assumption for the downcomer.
Two 1
important differer.ces were noticed when comparing these two models, each having a one-node downcomer, but one having a bubble rise assump-tion (V = 3 f t/sec) and the other using a homogeneous assumption.
B These differences were in the rate of depressurization and amount of water lef t in the vessel. Table 3 shows a comparison of the downcomer pressure vs time.
The effect of using a bubbic rise model is to extend the blowdown time.
One other important diff erence is that the water remaining in the vessel for the homogeneous model during blowdown is reduced.
A comparison of the water level in the vess'el at 200 see showed that the model assuming a homogeneous downcomer predicted 6308 lbs (several f t below the core) of water remaining in the vessel while the bubble rise model predicted 83518 lbs (N 7 ft into the core).
_E There are considerable differences between assuming a homogeneous and bubbic rise models. The bubble rise model inherently assumes that phase separation occurs (separation between the steam and water phases).
The homogeneous model assumes that phase separation does not occur.
In a large break the homogeneous assumption may be closer to reality in the early part of the transient. Analyses performed by the staff (as well as B&W) show that the _CFT-line break leads to a relatively gradual reduction in vessel pressure and low flow rates L'.
.u; c-
~
J I
~*~
TABLE 3 COMPARISON OF VESSEL PRESSURE 1
FOR BUBBLE RISE AND HOMOGENEOUS ASSUMPTION Time' Pressure for Bubble Rise Pressure for Homogeneous Sec Model Assumption Model Assumption 0
2250 2250
'10 1500 1500 e
20 1150 1150 30 1050 1060 40 970 990
-50 890 930 60 800 850 70 720 750 80 620 610 e
90 510 430
/I 100-410 270 110-340 160
-120 280 80 t
-r
>,.. > q
s through the system.
This is especially true after the first 20 sec.
From these analyses the staff believes that phase separation occurs and a bubbic rise model is appropriate for this accident. This assumption leads to a prediction of larger mass of water present during all stages of the transient as compared with the homogeneous assumption, llowever, the homogeneous assumption used throughout the transient is not realistic and leads to a nonrealistic low quiet water level calculation.
This low level would Icad to
'I
-2.
M
- * * ~ ~
. ' O c cr.;-_f,,.,
-a unaccepti le cladding temperatures, fc
_,f Further support for the use of a bubble rise model was given in an Idaho Nuclear Corporation report (Report IN 1444, Dec. 1970).
In this report the RELAP code was used to predict results obtained from a s'emi-scale blowdown experiment.
Figure 10, page 17 of the 1N 1444 report shows that the residual water remaining in a vessel af ter the end-of-blowdown is best predicted by using a bubble rise model. The figure indicates that the density gradient should be between 0.8 and 1.0 with a bubble rise velocity of 3 f t/sec. Based on these results and calculation performed by the staff, it is believed that using a bubble rise model better predicts the actual system response.
One other conclusion drawn by B&W was that the CFT line break never led to an end-of-blowdown (as defined in the B&W evaluation model for a large break). In the downcomer noding studies performed
.\\.
~
r m.
s by the staff it was concluded that end-of-blowdown could be calculated to occur by selecting 4 axial node in the downcomer and using a homo-t The end-of-blowdown occurred at
_ geneous assumption in all nodes.
The end-of-blowdown occurred because the cold core flood s 120 sec.
tank water enters into a node containing steam, which is then con-This e
densed, thus reducing the pressure below containment pressure.
node also contained the broken CFT line such that the reduction in pressure causes the' break flow to go to zero (the definition of end-of-blowdown). This effect was investigated using the REDUCED MODEL with 2 axial nodes in the downcomer.
An end-of-blowdown was not The staff concluded that the REDUCEb predicted using this model.
MODEL was a better representation of the physical system and that end-of-b-lowdown probably would not occur.
The model chosen as the analysis tool to analyze the 0.44 f t CFT line break was the SMALL MODEL using 2 downcomer nodes and bubble.,
rise model. Vessel pressure and quiet water levels predicted by this model were compared with the B&W analysis. Pressure comparisons between the RELAP model and B&W small breck model are presented in i.
Tabic 4.
Quiet water level comparisons were also made which showed i
The staff considers the " quiet good agreement between the two models.
water level" calculated by the B&W rodel to be c. best estimate of residual water Icft in the vessel.
-t
}
1
]
L'
b bypass cri-One assumption used by B6W was that the accumulator B&W gave Flood Tank line break.
s terion should not apply to the Core The first was that the system two reasons for aaking this change.
(A h d the end-of-blowdown pressure for the CFT line breakf never reac e (1/
l id velocity in the down-criterion; the second reason was that the f u During i
periods.
comer was always downward, except for short t me i i s were low (maximum these time periods the calculated veloc t e These low velocities
)
negative velocity was approximately 4 f t/sec.
In d out the break.
should not cause the ECC water to be entraine elocity effect was seen.
the staff independent evaluation the same.v mption, a For a singic node downcomer using a homogeneous assu tely 25 see was obtained.
i maximum velocity of N 4 ft/sec for approx ma er and using a Analysis ' reported by B&W using a three node downcom independent bubble rise model (also calculated by RELAP in the f
ity was approximately _
l analysis) eSowed that the maximum negative ve oc Critical velocity for entrainment from an 100 sec.
4 ft/sce for N 13 f t/sec at 300 psia using the Wallis correlation annuler film is N s.
~
. -gm given below.
-4 EL- /p f/p
" ['j
= 2.46 x 10 "g
8 8
t f pipe
= vapor volumetric flow rate per-unit area o (Critical velocity for entrainment) j 8.
--surface-tension o
= vapor viscosity p
-= density.of the vapor and liquid d
.p,p g h
~0 * $ a v
s
~
'}s
. 7..
TABLE 4 g
COMPARISON OF VESSEL PRESSURE FOR APPLICANT AND STAFF MODEL Sta/f Applicant's Model Time.
Model Sec 2216 i
2216 0
1020 1050 l
50 800 800 530 100 575 412 150
'450 255 200 320 210 300 250 400 170
~
180 500
-4 0
.::..C W
i 4.
.{
5-e' D
1 3
i 9
g 4
d t
4 1,s;[
e s
. ~,
e e
t the accumu-Based on these calculations, the staff has concluded that lator bypass assumption should not be applied to the CFT line break i is AU's'Hed 5 %
2 fh;s concep 3hkm ] of PostHen of he Ryl shy Sh/f'(Dodd RM 56 /].
with a break area of 0.44 f t
'Zonclub,y The boil off rate at N 200 cec is approximately 5110 lb/ min and the one HPI pump is supplying 4078 lb/ min fo*
st of the transient.
The boil of f is more or less matched by the supply.
Since the supply o
rate does not meet the Commission's " abundant emergency core cooling" criterion, the staff believes that the applicant should have a method This of supplying additional water for this postulated accident.
additional water should be supplied at a, rate which would insure that the core could be reflooded at a reasonabic rate.
To supply this additional water, the applicant has modified the low pressure injection (LPI) system piping by adding a cross-connect line and instituted operating procedures to be used in the event of a LOCA.
These procedures will ensure that water from the low pressure injec-c tion system will be delivered into the reactor ves:cl for this accident. The operator will be required to take action to rearran e
- PPW
=-
the _ valving in the LPI system within 15 minutes such that -
one-half the flow rate from the LPI pumps will be injected into the This amount of additional water would assure that reactor vessel.
. an abundant supply of cooling water is availabic to reflood the core and remove stored and decay heat.
s WN g GN v
~~
1 7
The staff has reviewed the makeup of the Three Mile Island Unit 1 operating staff and concludes that the shif t staffing as specified in 4
the Technical Specifications is, sufficient to perform the necessary.
valving to safely mitigate the consequences of this accident.
6.3.5.4 Heat Transfer Analysis B&W's cladding heatup analysis for this accident is basict.lly identical to that described in Section 6.3.4 of this report on small breaks analysis.
Since the primary system never reaches an end-of-f blowdown condition, and water remains in the vessel, the reflooding analysis normally done is replaced by a heatup analysis using the THETA and QUENCH codes with input from the blowdown code, CRAFI and the level swell code, FOAM.
There are two major differences between the methods used for the CFT line accident and those used in the small break model.
- First, the level swell calculation was based on a Wilson bubble rise calcula /
tion in the FOAM code.
The small break model used the mixture level calculated in CRAFT.
Second, the small break model assumed steam generation due to a mixture level 8 ft. into the core, the minimum level for any transient.
In the CFT line break, the mixture level calculated using F0AM was used for the steam generation calculation.
l However the calculation still conservatively assumed the avn. rage f
~
{
assembly steam generation rate and swell level for the maximum heat generation rate assembly.
~t i
1 o:
I.
1 9
L_
' t jv ~j o "
B&W has crop; rcd th2ir FOAM code to thrca cato cf cxp;rimental data, a series of Westinghouse, General Electric and Japanese tests. The Westinghouse
- I test (Pm f -
a-*T was contracted by Duke Power for this explicit purpose.
Of the three tests it utilized the largest nu=ber of simulated fuel rods (490) and the highest pressure (400 psia). The other tests, by GE and the Japanese (20 Tor" P
41, were based on a 49 rod BWR assembly at m
1 100 psia and atmospheric pressure.
Ilowever, neither the number of rods (49 or 490) nor the configuration (PWR vs BWR geometry) significantly affected the applicability of the data for verification of the FOAM code; kub o in fact, the variations in thoseAparameters helped to define the insensitivity of the heat transfer / hydraulics phenomena and FOAM.'s prediction of thesc phenomena to these parameters. The comparisons of FOAM to the data were generally within the experimental uncertainty of the data except for, several Westinghouse data points at 100 psia.
For these data, the FOAM codeover'bredictedthemeasuredswollenlevelbyabout10%. This may be attributed to nonquantified uncertainty in some of the measured parameters, such as the amount of subcooling in the inlet water.
On the whole, the staff concluded the FOAM code predicted well the swollen levels measured in the three tests. These tests were within the range of power levels, pressures and geometric configurations which would exist during the CPI line break accident.
The staff concludes that the use of the FOAM code is appropriate in calculating two-phase mixture heights for this accident.
The results of the application of B6W's FOAM code to the CFT line break accident were presented in Section 6.3.5.2.
The core was predicted to be covered with two-phase mixture during the accid at except for the period between 500 and 700 seconds after the accident. The peak cladding temperature occurred at approximately 700 seconds and reached 1199 F.
t
/ LtG4/
In examinting the swollen 1cvels predicted by FOAM for this accident it is necessarry to Point out a conservatism which may have an exaggerated effect if comp;ared to a more realistic calculation.
The lowest liquid levels predict.ed by CRAIT were used as input to the FOAM code.
This is actually a cont:radiction to fact, since the lowest liquid level CRAFT predictions arm only consistant with the high swollen level (above the top of the core) predicted by CRAFr.
This swollen level (above the top of the core) would not allow any significant cladding heat up.
On the other hand, for the lower stwollen level consistant with the FOAM prediction, CRAFT wou preidet more ILquid left in the vessel and this would result in about four more feet of liquid level in the core (9 feet versus 5 feet).
" a.s calculation would predict the core to be covered with two-phase mixture and there would also be no significant cladding heat up.
Therefore, for a consistant set of Predictions (high swollen level and low liquid level or low swollen level and high liquid level) there would be no significant cladding heat up. The analysis which is reported is the worst combination of both situations and results in an increase in cladding temperature To independently determine the two-phase mixture height in the core the staff and its consultant, Aeroject Nuclear Corporation ha)vG develo xuO a code (SWELL) using the Wilson bubble-rise model and a calculational procedure developed by CE in the Quad-Cities application (Docket 50 The SWELL code uses essentially the same calculational scheme as B&Ws code.
Preliminary calculations from this code have also shown agreement with B&W's FOAM code for the Wes'tinghouse tests.
Since the SWELL code is not presently well indexed against experimental tests, the staff also examined the cladding heat up transient in the 500 to 700 second period where B&W predicts the core may be uncovered
1 o...
Using the TOODEE Cha!.eener-?) heat transfer code, the sensitivity of th level was examined. The swollen 1cvel was cladding temperature to swollen jrt gthis resulted in an increase in peak cladding rch ced by an arbitrary 25%f Although the temperature did
,i temperature, to 1552 F, at 700 seconds.
increase 300 F over the applicant's calculation, the resultant peak cladding temperature would be acceptable even for an arbitrary reduction in swollen level.
6.3.5.5 Conclusions Based on the staff's independent calculations, and B&W's analysis, the staff has concluded that the emergency core cooling system, as i
modified, will provide adequate protection for a break of a CFI line.
O en :
s 6.3.6 Conclusions On the basis of our' evaluation of the additional B&W analyses, i
described above, we conclude that our acceptance criteria, as described in the Commission's Interim Policy Statement have been met:
1.
The maximum calculated fuel element cladding temperature does not exceed 2300*F.
2.
The amount of fuel element cladding that reacts chemically with water or steam does not exceed 1% of the total amount of cladding in the reactor.-
3.
The calculated clad temperature transi,ent is terminated at a time when the core geometry is still amenabic to cooling, and before the cladding is so embrittled as to fail during or after quenching.
4.
The core temperature is reduced and decay heat is removed for an extended period of time, as required by the long lived radioactivityp remaining in the core.
- The results of the applicant's analyses for a loss-of-coolant acci-dent initiated at a core power level of 2568 FNt show that the accept-ance criteria are met on the basis of analyses performed in accordance with an. acceptable evaluatiop model given in the Interim Policy Statement.
On the basis of our evaluation of.the B&W analyses described above, we have determined that the emergency core cooling system is acceptable and will provide adequate " protection for any loss-of-
. coolant accident.
1
=:
_