ML20027C093
| ML20027C093 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 03/31/1982 |
| From: | Ornstein H NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | |
| References | |
| TASK-AE, TASK-E216 AEOD-E216, NUDOCS 8210120406 | |
| Download: ML20027C093 (24) | |
Text
r
. b-[' 4~,
l f
AE00/E216 This is an internal, pre -
1 decisional document not necessarily representing a l
position of AE0D or NRC.
i A Recently Evaluated Preoperational Test Precursor
[
l of the TMI-2 Accident t
i Harold L. Ornstein
. Office for Analysis and Evaluation of Operational Data U.S. Nuclear Regulatory Commission l
l l
t i
l i
For presentation at the American Nuclear Society 1982 Power i
Division Topical Meeting, " Engineering for Nuclear Plant Construction, Operation, and Maintenance," Charleston, South Carolina, March 28-31, 1982 R10120406 BEN)331 i PDR ADOCK 05000320 R __
i Introduction This paper discussas a significant preoperational test event that occurred at Three Mile Island Unit 2 in September 1977, some eighteen months before the March 28, 1979 accident. The event occurred during hot functional testing prior to fuel loading. Had the reactor been fueled and at power when the event occurred, there might have been core uncovery followed by fuel damage similar to that which occurred during the March 28, 1979 accident.
The only known records of the event are piecemeal descriptions which appear in the plant startup logbook. As a result, there is considerable un..ertainty surrounding the actual plant conditions and configurations involved with the apparent steam binding condition, and the subsequent plant recovery. However, the event illustrates the necessity to disseminate, analyze, and evaluate operational data for the benefit of public health and safety.
The information presented in this paper was developed from work does by the author in response to questions raised by the U.S. Congress, Committee on Interior and Insular. Affairs.
On September 5, 1977, some eighteen months before the accident at Three Mile Island Unit 2 (TMI-2), a significant precursor of the accident occurred at the same unit. A seemingly insignificant transient initiated by the condensate polisher system triggered a steam binding event which, during power operating conditions, could have r4 'lted in core uncovery and possible serious damage to the core.
It should be no,ed that the same condensate polisher system caused another, but less significant, transient on October 19, 1977 and was the initiator of the March 28, 1979 accident.
1 1
At the time of the September 1977 event, the reactor was undergoing hot functiorial testing, prior to the NRC's granting of an operating license and
?
initial fuel loading. The startup logbook shows the primary system was close 1/
to normal operating conditions at 2155 psig and 532'F.'-
Primary coolant system pump work and the pressurizer heaters were used as heat sources to achieve these conditions in lieu of fission heat.
As a result of a condensate polisher system malfunction on September 5,1977, i
resins from the condensate polisher system were carried over to the plant's demineralized water system. As shown on Figure 1*, once inside the demineralized water system, the resins migrated to other parts of the plant.
The startup logbook entries of September 7,1977 not-f that polisher resins were found in the domineralized water system in the turbine and auxiliary buildings.
Some of the important systems and components which are connected to the demineralized l
water system are listed in Table 1.
l
--1/
1 Chronology was obtained from GPU Startup Shift Test Engineer Logbook No. 2, September 1977.
4 Figure 1 - Simplified Diagram of the TMI-2 Demineralized Water System was I
obtained from TMI-2 FSAR Figures 9.2-9 and 10.4-1.
i
f 2
Table 1 Partial Listing of Systems and Components Connected to the Demineralized Water System Borated Water Storage Tank Boric Acid Makeup Tank I
Reactor Building Normal Cooling Water System Core Flood Makeup Tank Nuclear Services Closed Cooling Water System l
Intermediate Closed Cooling Water System Decay Heat Closed Cooling Water System
- Reactor Building Spray Pump Suction Header I
- Makeup Fump Suction t
- Seal Return Coolers
- Spent Fuel Cooling System i
- Can be connected to the Demineralized Water System but not usually connected to it.
j l
4
s j The mort significant result of condensate polisher resin migration was the fact that the resins clogged the strainers to all of the Nuclear Services Closed Cooling Water System (NSCCW) pumps, causing them to trip. Tripping of the NSCCW pumps resulted in the loss of the following equipment:
Reactor Coolant Pump Motor Coolers
' Reactor Coolant Pump Motor Bearing 011 Coolers High Pressure Injection (HPCI) or Makeup Pumps and Motor Coolers i
Instrument Air Compressors and Aftercoolers Reactor Building Spray Pump and Motor Coolers 1
Reactor Building Emergency Booster Pump Motor Coolers t
Spent Fuel Coolers t
Subsequent to the loss of the NSCCW pumps, the reactor coolant pumps (RCPs) were tripped. Tripping of the RCPs and shut-off of the pressurizer heaters f
f resulted in depressurization of the primary system, durtng which steam bubbles fomed in the hottest portion of the system and apparently coalesced in the 2/
i hot leg " candy canes" (see Figure 2).
The presence of steam accumulations l
in the candy canes is of concern because of the potential to impede natural circulation flow of the primary coolant in a situation where there is core decay heat to be removed.
Because the loss of the NSCCW pumps precluded cooling of the HPCI/ makeup pumps and motors, capability to use the HPCI/
makeup pumps was lost.
2/
Figure 2 was obtained from the " Report to the United States Senate, Nuclear Accident and Recovery at Three Mile Island, A Special Investi-gation," Subcommittee on Nuclear Regulation for the Senate Committee on Environment and Public Works, U.S. Senate, 96th Congress, Second Session, June 1980, page 34.
l
. -.I
. The net result of the polisher system malfunction, during the September 1977 event, was the potential loss of all primary system heat removal capability, i.e.,
(1) Forced convection using the reactor coolant pumps, (2)
Natural circulation cooling, and (3) " Feed and bleed."*
From the data available (startup logbook), it is not clear at which times the loss of capability for forced convection, natural circulation, and feed and bleed took place. However, it is conceivable that under power operation circumstances all core cooling capability could have been lost concurrently which, if not corrected, would result in an eventual core overheating.
In the case of the September 1977 event, it is not known how long it took the plant operators to provide an alternate cooling source to the HPCI makeup pumps and motor coolers. However, if the reactor were fueled and at power at the time of the event, tre time required to re-establish makeup pump flow to initiate feed and bleed woeld be crucial.
It is estimated ** that for operating Babcock & Wilcox (B&W) plants, under certain conditions, failure to re-establish HPCI makeup would result in core uncovery in about 30 minutes after event initiation.
- " Feed and bleed" involves the introduction of high pressure coolant via the HPCI/ makeup pumps and the discharging of steam and/or water through the relief or safety valves.
. The startup logbook indicated that for more than two days the plant operators unsuccessfully tried to remove the steam bubbles from the hot legs and fill the hot legs with water. As can be seen on a temperature-entropy chart (Figure 3),
the only way to eliminate trapped steam bubbles in the hot leg candy canes with-out venting or forced convection is to remove heat from the steam by either heat removal through the pipe walls, or by the introduction of subcooled liquid.
The subcooled liquid could be introduced at the steam / liquid interface, or could be injected as a spray.
The phenomena of possible steam accumulation in the candy canes was apparently not in the plant's design basis.
In any case, there were evidently no equipment provisions or procedures to eliminate such accumulations.
Eventually, the operators pumped nitrogen into the pressurizer, thereby forcing cooler water from the pressurizer to the hot 1.egs and steam / liquid interface, thereby condensing the steam bubbles.
During the September 1977 event, the startup logbook also indicated that the operators had difficulty reading pressurizer level.
Prior to the introduction of nitrogen, the operators noticed that the pressurizer level increased unexpectedly whenever they vented the pressurizer.
They did not recognize that the water somewhere in the system was flashing and a steam bubble was expanding, thereby forcing water into the pressurizer.
Instead, they surmised that the reference legs of the level instruments had flashed.*
- From startup logbook, September 8,1977, second shift.
r
t l l
While the 1979 accident was in progress, B&W's site representative at TMI-2
(
discussed with B&W management in Lynchburg by telephone some of the details l
of the September 1977 steam binding event. During the conversation B&W's site representative described similarities between the problems being experienced i
during the accident and the problems that had been encountered during the September 1977 event:
i i.e., Cannot fill the whole system!
There is no way to fill it!
The l
system is designed to move water in a LOCA.
We haven't had a LOCA and we can't operate it that way.
What we got is...We have i
been in hot functional test. We had a similar condition where l
we had the hot legs on both loops filled with hotter temperature
[
water than what we had in the system and it took us something like four days to get out of that thing to try and cool it down to where we could get that bubble condition out of there. We've t
got a similar condition here.
The only way we can force that bubble out of that B loop at this point._jo it is to
/
i Conceivably, had information about the September 1977 steam binding event received greater attention and had been evaluated and understood, mitigation of the 1979 accident might have taken a more effective course.
Knowledge and understanding as might have been gained from analysis of the September 1977 event could have provided guidance as to why pressurizer level appeared to increase during pressurizer venting (such as what happened during the accident when the PORV was stuck open); and how the ability to establish and maintain natural circulation flow could be impaired by steam accumulation in the hot legs.
3~/
l
" Reporting of Information Concerning the Accident at Three Mile Island,"
a report prepared by the Majority Staff of the Committee on Interior and Insular Affairs of the U.S. House of Representatives, 97th Congress, First Session, March 1981, page 150.
p i
4 i !
l f
Perhaps the prevailing attitude at the time of the September 1977 event was one i
of unconcern since the reactor was not fueled at the time and there was no core decay heat to be removed, and there were no adverse effects on the plant or the I
general public. Most people involved might have viewed the event as insignificant I
and merely an operational inconvenience.
Perhaps the potential seriousness of l
similiar failures occurring during power operation escaped recognition, i.e.,
if the event had occurred when the reactor was fueled and at power, there might l
have been severe core damage.
5 About one week after the September 1977 event began, a TMI employee reviewed the startup logbook entries covering the event. He wrote his observations in the logbook, including the following:
i f
i Generally this week we had a major unusual occurrence and numberous (sic)
)
things were not entered in the log that should have been; for example,
- a. vacuum pumps were all cleaned of resin. b. did condensate pump 1B on i
9-8 entry have resin in it?
- c. no one logged the fact that we did a complete flush of the demin. water system.
l There is no reason given for how we cot into the problem on pressurizer level. A change to coolcown procedures could be l
4 made if we knew what to do.*
i i
Even though the plant operators apparently recognized that they had steam in f
the hot legs which could not be dissipated for several days, and that there j
\\
were anomalies in the pressurizer level with the possibility of reference leg i
flashing, the plant's startup report made no reference to this event.
j
- Emphasis added.
{
i t
--n-
, - - +. - -,,,, -,. -... - - -
Information about the September 1977 steam binding event was apparently not disseminated to the " nuclear community" until after the 1979 accident.
It appears that the utility management did not identify the event as a significant precursor.
It was not until the summer of 1981, when the author reviewed the TMI-2 startup logbook entries, that the full significance of the September 1977 event was first discovered. The author is unaware of any B&W or Metropolitan
- Edison analysis of the September 1977 event prior to the 1979 accident.
The operators did not have established procedures to handle the recurrence of a steam binding event at the time of the 1979 accident.
Although piecemeal descriptions of the September 1977 steam binding event did appear in the startup logbook, the NRC was not informed of the event until after the 1979 accident. Even now, there is considerable uncertainty surround-ing the actual plant conditions and configurations involved with the event and the subsequent recovery.
This experience clearly illustrates: (1) the importance and need for appropriate analysis and evaluation of preoperational test information as it may relate to the enhancement of plant operational safety; and (2) the potential safety benefits afforded to the nuclear community by sharing such observations of anomalous plant behavior.
I SIMPLIFIED DIAGRAM OF THE TMI-2 DEMINERALIZED WATER SYSTEM
,___.______________1 l
- TURBINE BUILDING s
L=_
g l
==
'TO SERVICE WATER YARD l
' CONDENSATE POLISHER l
DEMIN WATER l
4" l
50,000 gal g
l 6"l l
=i O
6" I
l
( FROM, MIXED BED DEMIN l
l
( REGEN STATION '
~
l l
TO CONDENSATE POLISHER 1 r
'i g
i 10"j
_ _ _-
[_] CONTROL BUILDING _
_,l, l
m i-1
~ l,,,,,,,
l l
CONTROL BUILDING E '"
wA ER
-l 1.000,000 sai l
- AREA, l
l 2" a
'TO VARIOUS LAB 1 r l
TO EMERGENCY y
SUPPLIES AND i
3 L
F _ _S.G. FEED PUMPS
-4 J
CHILLER MAKEUPS I
l AUXILIARY BUILDING l
REACTOR BUILDING,l 3"' la -
i l
SEE TABLE 1 1 TO VARIOUS l
l I'
(WASHDOWNS i
I i
g 4" p q
l D.G. BUILDING i
.I l
1
, 4" 2" l l
) TO DG JACKET i_ _ _ _ _ _ _ _ _ r _ _ r _'f.eoE".s*2"a___J 1
Figure 1
t Ii TMI-2 PRIMARY COOLING SYSTEM UNDER NORMAL CONDITIONS...
-s
. 9. -
i PILOT OPER %TED O PRIMARY WATER,/
t
' RELIEF VALVE IPORV)-,
M SECONDARY WATER 3
l {
8 LOCK VALVE
~
CANDY, CANE CANDY CANE.
8 C '
].
' STEAM GENERATOR 5
l(i.k:
I l
I
(
REACTOR HEACTOR*
TO REACTOR i
7 COOLANT /
l j COCLANT I
COOLANT 4
y.
DRAIN TANK
- .,t
'f,'
PUMP d
, PUMP fi
^
I[
l
'.:;.: l::
' RETURN TO 1
3118158 9 48 l
EAC10R VESSEL
.. f F:j(
l j
i s.
PRESSURIZER j ? !6 l
l l
FROM[
REACTORi.
g, g
1 l
, VESSEL' l
l "U
WJ REACTOR VESSEL LOOP A LOOP B i
t I
l
~
Figure 2 1
i' I
t'.
P~
~
,P-
,P P.
P B
C',
GH
> P C f
' s-
/
/
/
\\
\\
RR RR
' C n
B A'
OO OO BC GH
-> r
.g i
l /
k G
'H
/
\\
k f
\\
f G
M A
i A
H F
E M TA i
,\\.
S SE D T S
E 3
TD M
AE e
s P
P P
A E T EM r
HA u
TA P
P RE SE.
g i
E H PR T T F
E G S GW N
N G
G I
I Z
Z I
N N
I RI RI Z
Z UI UI SR SR SU SU E S ES P
RS RS PE PE P
E R ER P
,P <
DP DP
<,)
P
(
)
I T
^
% Biographical Data i
Harold L. Ornstein, Ph.D.,* is a Lead Systems Engineer at the Nuclear Regulatory l
Commiission's (NRC) Office for Analysis and Evaluation of Operational Data (AE00)
[
where he_ is currently in charge of reviewing event reports and other information i
relating to Babcock & Wilcox type plants.
l l
AE00 is responsible for the analysis and evaluation of operational safety data I
associated with NRC-licensed activities and the feedback of such analyses to j
improve safety. The office provides agency coordination of operational data
{
r collection, storage, and retrieval activities; systematically and independently analyzes and evaluates operational experience; and feeds back the lessons inherent in operational experience to NRC licensing, standards, and inspection activities, and to licensees for all NRC-licensed activities.
I i
Dr. Ornstein has been employed at the NRC since 1975.
His assignments have I
l included assessing the safety margins which were available during the Browns l
\\
Ferry fire and preparing testimony for the Joint Committee on Atomic Energy on the fire.
He also served on the NRC's Special Inquiry Group on the Three
['
t Mile Island Accident (Rogovin Report).
Prior to his employment at the NRC, Dr. Ornstein served as a Senior Analytical 4
Engineer at Pratt and Whitney Aircraft, as a Research Specialist and Instructor at the University of Connecticut, Assistant Director at the New England Research Applications Center (NERAC), and as a Reactor Engineer for the Atomic Energy t
Commission's fast flux test facility project.
)
r
- Ph.D., University of Connecticut, 1971; MSME, Rensselaer Polytechnic Institute,
)
1964; BME, City College of New York, 1961.
l l
i i
l-i 2
)
t i
i f
b f
I I
h
?
b s
l b
}
?
I r
I i
h S
L I
D E
S r
r P
9 e
?
I I
?
\\
t 3
t L
i I
I b
- ,w g
....n---
\\
SIMPLIFIED ~ DIAGRAM OF THE TMI-2 DEMINERALIZED WATER SYSTEM pm_
~
l TURBINE BUILDING s
l l
_ _ =
TO SERVICE WATER l
l YARD l
CONDENSATE POLISHER l
DEMIN WATER 4"
I 50.000 gal g
l 6" l l
"i O
I 6"
I l
c FROM MIXED BED DEMIN l
g l
[
)TO CONDENSATE POLISHER U
l I
( REGEN STATION
'i
,,g g
- l l!
[ CONTROL BUILDING T u i-1 E '"
WATER l
CONTROL BUILDING I
4 1.000.000 sai l
AREA 2"'l
' TO VARIOUS LAB U
a l
TO EMERGENC
SUPPLIES AND l
l i
F _ _S.G. FEED PUMi s
-4 J
4 CHILLER MAKEUPS j
l AUXILIARY BUILDING l
REACTOR BUILDING l
l 3" l l
l SEE TABLE 1 I
a-1 TO VARIOUS l
l I~
(WASHDOWNS i
,q 4"
D.G. BUILDING
-l y
l 3
, 4" 2 l -
)TO DG JACKET l
{
l
{
\\_C_OOLING. WATER 4
Fwms 9
PARTIAL LISTING OF SYSTEMS AND. COMPONENTS CONNECTED TO THE DEMINERALIZED WATER SYSTEM BORATED WATER STORAGE TANK BORIC ACID MAKEUP TANK REACTOR BUILDING NORMAL COOLING WATER SYSTEM CORE FLOOD MAKEUP IANK NUCLEAR SERVICES CLOSED COOLING WATER SYSTEM INTERMEDIATE CLOSED COOLING WATER SYSTEM DECAY HEAT CLOSED COOLING WATER SYSTEM
' REACTOR BUILDING SPRAY PUMP SUCTION HEADER
- MAKEUP PUMP SUCTION
' SEAL RETURN COOLERS
- SPENT FUEL COOLING SYSTEM
'CAN BE CONNECTED TO THE DEMINERALIZED WATER SYSTEM BUT NOT USUALLY CONNECTED TO IT.
a'm-ew --
---,m-ry-w%
e-ei-
+-i
--ww m-
-,we
-rm-g y
e, 3 -
g-y--
t'-e q-me-w----ym 3
ge p-e
--w-p----+-
r--,vy---s
-, -e---
ew--,-----ww-w yva-e-+e
-wwg
SYSTEMS AN) EoulPMENT _0ST AS A RESULT OF THE
_0SS OF THE 6CCW PUMPS 1
l REACTOR COOLANT Pune MOTOR COOLERS j.
REACTOR COOLANT PUMP MOTOR BEARING OIL l
COOLERS HIGH PRESSURE INJECTION (HPI)/ MAKEUP PUMPS ANo MOTOR COOLERS INSTRUMENT AIR COMPRESSORS AND AFTER-COOLERS REACTOR BUILDING SPRAY PUMP AND MOTOR COOLERS REACTOR BUILDING EMERGENCY BOOSTER PUMP MOTOR COOLERS SPENT FUEL COOLERS
TMI-2 l
PRIMARY COOLING SYSTEM UNDER NORMAL CONDITIONS...
i PILOT-OPERATED E PRIMARY WATER
' RELI F VALVE (PORV)
M' SECONDARY WATER f
l BLOCK VALVE
'(
CAND CANE CANDY CANE STEAM OENERATOR STEAM GENERATOR i ;
Y
.;.0; E}::
a
)
l l
. TO REACTOR REACTOR 4
REACTOR k
COOLANT f'
COOLANT
,(y#i W.
COOLANT 1
PUMP
. _U_MP P
,.3.-
.f,-
DRAIN TANK y;
l
^
7:
2 l;
31048804d BETURN TO e.
l EACTOR VESSEL r.;;,
PRESSURIZER
/
,j k
l
- FROM REACTOR l
VESSEL l
"* Wj REA OR VESSEL LOOP A LOOP B j
Figure 2 a
0
1 NET lEStAT OF THE SE'TEMBER 5, 1977 30LISHER SYSTEM M FUNCTION LOSS OF FORCED CONVECTION VIA THE REACTOR COOLANT Punes LOSS OF NATURAL CIRCILATION FLOW PATH LOSS OF FEED AND BLEED CAPABILITY CONCLUSION:
h.L PRIMARY SYSTEM HEAT REMOVAL CAPADILITY WAS LOST
I P>P,
.C f
k C
P, J (
/
P>P, H
H'
/
l l
) k /
A
/
P < P, '
T P-
.F, 7
\\
4 P>P,
' (
\\
\\
1 k
B' P<P, 1 (
p G
G' B
P
G' l PRESSURIZING WET STEAM ,T PH OR F P H' l \\ l l Figure 3 1 i .~. ..~.-- .1. I I l l ) OPERATING EXPERIENCE FROM THE SEPTEMllER 1977 EVENT WHICH WAS DIRECTLY APPLICABLE TO THE 'NI-2 ACCIDENT I 1 PRESSURIZER LEVEL APPEARED TO INCREASE DURING PRESSURIZER VENTING (VS. STUCK OPEN PORV DURING ACCIDENT) 2 ABILITY TO ESTABLISH AND MAINTAIN NATURAL i CIRCULATION FLOW WAS IMPAIRED BY VOIDS IN Ti1E HOT LEGS a. -.. - a u ---a m 2___ a J I 4 p3 b k ua 3 i t4s 5 9 .a t ". > 9 t ( QC < r \\ 1 4' 144 1 'o 1%o, \\ s 0 g b $4 s( 's e: -.1 s } l i l t ,...,..-.-c ,~, i l 0FFICE FOR ANALYSIS AND EVALUATION OF OPERATIONAL DATA FORMED IN 1980 AS AN OUTGROWTH OF TMI-2 ACCIDENT I DIRECTOR: CARL MICHELSON - FORMERLY OF TVA, PRESAGED TMI-2 ACCIDENT 30 STAFF MEMBERS 1 REVIEW OF LERS, PNS, INSPECTION AND ENFORCEMENT DAILY REPORTS, AND OTHER SOURCES OF OPERATING DATA REPORT FINDINGS IN CASE STUDY REPORTS, ENGINEERING EVALUATIONS, POWER REACTOR EVENTS, AND SBNORMAL OCCURRENCE REPORTS MAINTAIN LER FILE MAINTAIN CLOSE LIAISON WITH INP0/NSAC I I .