ML20236K962
| ML20236K962 | |
| Person / Time | |
|---|---|
| Site: | Limerick  |
| Issue date: | 10/14/1987 |
| From: | Clark R NRC |
| To: | Butler W NRC |
| Shared Package | |
| ML20236K951 | List: |
| References | |
| NUDOCS 8711100036 | |
| Download: ML20236K962 (9) | |
Text
-_-
l 1
October 14, 198)
NOTE T0: Walter R. Butler FROM:
Richard J. Clark
SUBJECT:
INSPECTION OF LIMERICK EHC SYSTEM l
i
===1.
Background===
In my note to you of September 24, 1987, I discussed a scram that occurred I
at Limerick Unit 1 on September. 19, 1987 due to a rupture in the electrohydraulic control (EHC) oil system.
I advised you that I planned to visit the Limerick site on September 25,'1987 to discuss various.
issues, including the scram.
I suspected that the rupture of the EHC' line at Limerick, the rupture of the same EHC line at NMP-2 on July 11, 1987 and the ruptures of the EHC. piping and the resulting scrams at other plants (e.g., Brunswick) might be related to the GE product improvement' modifice. tion implemented at Limerick during this summer's refueling outage.
2.
Purpose The purpose of this note is to-l 1
1.
Report on my followup of the scram at Limerick.
2.
Advise you of what problems can result from the GE turbine upgrade modification.
3.
Suggest that NRR investigate tne experience of utilities (particularly BWRs) who have implemented the' modification.
a
]
4.
Depending on the above, consider issuance of an information notice so that utilities implementing the modification will be aware of 1
the potential problems, install z oitoring instrumentation and will
.l provide refresher training to operators to cope with pressurization l
3.
Safety Concerns 1
The safety concern with rupture of the EHC oil system is the significant pressurization transient that can result.
In most BWR core reload analyses, turbine trip w/o bypass availability results in the most severe transient.
Loss of EHC pressure closes the turbine control and stop valves. The EHC oil system also controls the bypass valves.
Generally, there are two or three small accumulators on the bypass valves that can be used (if charged) to open the bypass valves for a short period of time. However, the loss of control of the bypass valves eliminates 'the use of the condenser as a heat sink during recovery operations.
The only j
way to avoid repeated actuations of the SRVs is'to quickly lin.e up steam 8711100036 871104-PDR ADOCK 05000352 p
PDR.
l E_______
users (i.e., turbine driven cumps).
The onerators at Limerick quickly took the HPCI and RCIC systems out of the infection mode and placed them
-I in the test recirculation mode (recirculating to and from the condensate storage tank).
They also aligned two of the feedwater pumps in the recirculation mode (back to the condenser). With no steam coming into-the condenser, only a limited amount of steam can be dissipated through
)
Controlling reactor water level with a FW j
pump requires close operator attention to avoid overfilling and shutting off the steam users.
j
.]
So far, the scrams resulting from ruptures of the EHC oil systems have occurred while the reactors were at less than full power.
NMPC-2 was in j
the startup mode with reactor power at approximately 4%. Limerick had
]
discovered the oil leak and was in the process of shutting down when the j
weld let go. Brunswick Unit I was in startup preparing to roll the l
turbine.
Reactor power was approximately 26% with three bypass valves open. The operators had noted "a large high pressure leak"_before the-pipe ruptured and the operators had started inserting control rods and i
reducing power before the EHC oil pressure was lost. Although Brunswick I
has three accumulators on the bypass valves, the rupture apparently depressurized the accumulators so that the bypass valves stayed closed, scramming the reactor on high reactor pressure.
As Niagara Mohawk stated in the LER:
"this is a much more severe transient at full power".
CP&L stated in the LER:
"The event could have been more severe had it occurred at 100% power".
4.
Followup at Limerick Last Friday morning (September 25) at about 10:40 a.m. while I was onsite, the operators noticed another small leak (about 1 1pm) in the EHC oil system.
Following last week's scram, closed circuit television had been set up to observe the EHC system, since this is a high radiation area during operation.
The ieak appeared to be coming from the servo valve (Figure 1) on control valve no. 2.
The servovalve regulates EHC oil to the control valve. The plant manager decided to shutdown; a rapid plant shutdown was commenced at 11:10 a.m.
When the power was below 30%
(within bypass valve capacity),.the main turbine was tripped at 1:54 p.m.
Limerick is unique in that the EHC oil supply to the turbine control /stop valves and the bypass valves can be separately isolated.
Reactor power was maintained at about 22%.
The plant mMager set up a meeting for 3:30 p.m. (which I attended) involving all operations, maintenance and engineering supervisors onsite, representatives of the engineering groups in headquarters who had been involved in the modification and a GE service supervisor. A partial list of attendees is enclosed.
The plant manager opened the meeting by stating that the vibration noted in the EHC system raised a safety concern and that PECo management had decided that the plant would not restart until they were satisfied that it was safe to continue operation.
f The meeting continued until 7:00 p.m. and covered everything known abnut the systen, what additional data would be desirable, what were the potential consequences of the vibration, what could be done to reduce the vibration, etc.
The modification to the turbine control system this summer is refered to as " partial arc emission".
There are four turbine throttle valves, each r# which controls steam to one quadrant.
The normal logic arrangement is for all four valves to open or close as a group to control steam flow to the turbine.
This is referred to as " full arc emission", since steam flow is being uniformly controlled in each quadrant.
In " partial arc emission," the control logic is modified so that two or three of the control valves are operated wide open, and steam flow is regulated with the remaining valve (s).
The Limerick modification was to operate with three throttle valves wide open.
The lower pressure drop results in j
slightly higher steam pressure in the high pressure turbine.
The modification tm supposed to produce 6MW additional power, but since startup on August 26, 1987, the plant has not been able to get more then 2 to 3MW additional output.
From what I have learned so far, the modification can create several potential problems.
(I am chairman of an ASME subcommittee; there are
{
two engineers on the committee from GE's large turbine group in Schenectady.)
In one or two of the initial turbine modifications, there was insufficient clearance between the first stage buckets in the HP turbine and the point of injection.
The higher dynamic stresses wiped out the first stage buckets in the high pressure turbine.
The Limerick j
turbines were designed for partial arc emission by increasing the space i
in front of the first buckets, and, according to GE, eliminates the possible problem of too high stresses on the buckets.' A second major problem stems from steam line resonance.
The changed harmonics feedback into the control circuit, causing the control valves to hunt. The hunting, coupled with the higher volume of EHC oil that needs to be pumped, sets up a vibration in the EHC system, which reflects in the control valve logic and accentuates the vibration.
The GE engineer savs it is a very complex electronics problem that can be dampened by appropriate filters when the basic cause is identified.
PECo engineers were aware of the increased vibration that might result from the turbine modification. Although GE insisted it was not necessary, PECo installed accelerometers on the piping as part of the modification along with several snubbers. A sketch of the EHC oil piping is depicted in Figure 2.
The failure at Limerick on September 19, 1987 was at a reducing tee.
Nine Mile Point Unit 2 (NMP-2) and Brunswick Unit 1 have the same arrangement.
The failures at these two plants were at a similar reducing tee.
In all three cases, metallographic examination attributed the principal cause to be fatigue failure due to vibration.
The stainless steel piping in the EHC systems is primarily assembled using socket weld fittings. At Limerick, the weld failed.
Examination
i
. )
1 disclosed that it was not a full penetration weld on the back side,
?
probably due to the welder not holding the electrode at 45' to the pipe
- s (Figure 3).
The socket welds are not amenable to radiography, so it is not ;assible to determine whether there are other similar marginal welds in the system.
At NMP-2, the weld held; the vibration severed the pipe (Figure 4).
At Brunswick, the weld also held; the failure consisted of two.through-wall circumferential cracks.in the fully annealed piping in j
the socket weld area.
Because of the accelerometers PECo had installed on the EHC piping, they had the deflection, velocity and acceleration measurements to support a stress analysis of the. system.
There was significant visible and measured vibration of the piping.
However, the calculated fatigue load factor was only about 20% to 50% of allowable for 30 years of operation
]
(assuming welds are good). The expansion and thermal growth was about 2" during heat up; the system has to be flexible, which limits the location 1
of snubbers and clamps / stiffeners that can be installed.
PECo plans to l
install special clamps at the four reducing tees (where the failure on l
September 19, 1987 occurred) as soon as they can be obtained and the modification is approval by PORC.
The leak that was observed Friday morning (9/25) was from the
{
electrohydraulic servovalve on the control pac for control valve no. 1.
j The leak apparently was from a 3/4" diameter "0" ring seal (see Figure i
1).
The servovalves on valves 1 and 2 had been replaced during this sunner's refueling outage as part of the preventative maintenance program. The servovalve was replaced with a new spare valve.
1 At the meeting on Friday afternoon (9/25), an extensive test and diagnostic program was outlined.
The. program required preparation and approval of procedures, procuring instrumentation, etc.
In the interim, it was concluded that the plant could be safely operated with close observation and monitoring of the EHC system. One of the accelerometers was relocated to an actuator to measure the extent of vibration on the control pacs.
Power ascension and rolling of the turbine started Friday evening.
Saturday morning, fumes and smoke were noticed on the closed circuit TV; EHC oil from the leak had soaked into the insulation.
The plant manager directed that the unit be shutdown and the entire turbine control area be scrubbed, cable trays washed down, insulation replaced, etc. Another extensive meeting of plant supervision was held Sunday morning to discuss possible safety considerations associated with startup and operation. The plant was started up Monday morning (9/28) with a i
slow, deliberate power ascension.
By Tuesday (9/29), some of the test procedures had been prepared and PORC approved.
It was observed that there was very little vibration in the EHC. system up to about 85% power and that the vibration increased significantly the higher the power level was increased above 87%.
Plant management decided to reduce and hold power to 83% until extensive additional testing could be accomplished on harmonics, electronic feedback, control valve performance, etc.
As of October 8,
1987 (two weeks after the observed leak), the Want is still being held to 83% power while conducting an extensive test program.to find the root cause of the vibration.
To my knowledge, Limerick is the first plant that has devoted the time and effort (at considerabia cost) to try to determine the root cause of this complex problem. Other plants. installed enough snubbers and' dash-pots to dampen the vibration and converted back to full arc emis'sion' during the next refueling outage.
Brunswick 1 and 2, for example, converted from partial arc to full arc emission on December 12,1985 and :
December 4, 1984, respectively.
I understand that GE has sold the partial are modification to several other utilities.
Susquehanna Unit 1 is converting to partial arc emission during the current refueling outage.
I also "nderstand that GE has' tended to minimize any potential problems.
Since th,. EHC system is not a safety-grade system, operating experience with this equipment is not-reported in the INP0 publications unless it causes a scram.
5.
Recommendations 1.
Since the modification which GE is selling has caused pressurization transients and scrams at several BWR's, I recommend that NRR monitor the test programs at Limerick to understand the feedback from the steam line harmonics into the EHC control signals.
l 2.
NRR should investigate the experience at BWRs'and PWRs (e.g.. TMI-1) that converted to partial arc emission.
- 3 3.
Consider some means of alerting utilities to the potential vibration in the EHC system if the turbine is modified for partial arc emission.
N ichar ark
)
cc:
Wayne Hodges, SRXB/NRR
i
/ATR "O'b "
l I dj },
F LS~
87
/
/
.u M
. seip/rnt l
c/ /* /$t Y l f
$bc
~ /t/ G d
{ &.$
/
j
/.C(C Of /-1$/d
.$tincricY keh Nlosnaf er lAl0d ' A//A 8eije_ug,,,sig '
t BcOcww
.. L e s.
- rest: E m.~ e e...... -._..
Yldbstl f.&?4 4f b ea b z sce 9 ri
$Co._ - REa cret gapincEr t
.. PcGc - J'c2-L A
~ i~0 6.. G m.m r >
PEC9
~T? L - 13 u nk, 2"Gic i
a t m,U 1en =sa AHL fLg?&K
,a s :._
QLG~uu..
?e k.: A 4 J - E ce.
J A hphp, PEce - Ann _Emm. _. _.___.
. A. %
fro
- miw m 3 a %-
%e
... wf /k1.L.-..Lv..YLCf..-...
0/LC. c -
EtAinT.- & NG.,_ _.. - -.. -....... -.. - -..
W f..D
.JI?' ~
Y ?!20 Jo k ~ sus s
._ R a
. m e C % :.s +.
1
_ du.ha.7 _. _....____iEG - SpengL L.sr-._.__.
-. ScLil. tJ.LR ~. -..J?m= -..ncspac..e <-
l P y e m. n r a
-.. A co c.w ww..
e
-. O I s le' i A. W e n 7 ci r
/EG -.: A.fi*vy.,4(LL bece.ng....
/ drew /$$ c S E..5;re&c 5'
&+,,f l2' c0. E& y.nar o-fl/ /J 1
km sm,rww pew-nn m m
J tm b.._.msm._c...
___.-.?F.c.,
__Assss,Op;#m J.-g Abw fo raw Pico -Mwr.Depr.-urs tuon anrc
-4 en>en Peco ma~r3er-s'opawa
/O R m - i,,.
x.y sa i:;s i
('n. ;
sic.ven rf b
,%. 1.0;:. p. 1
.:L
.a I
~.
O dt. 5 f 1M V_.
?E6 -
HE l
(oJct0 a
_ q i
l f,.
'd,,
{
3
}
j k
. /]A & &.
,0_._mf4A/s/'.
03_/_W WJ,..-.
.V-i 1
3A nR?.0A' A HA M. _._
ffC&7&E Of [NG8-7
.DAu 3cem W.,-..
A._&.._A )
.I g
pe_.BaysO Muuhos. PLc.5 U~rwusw& J i-_.,
_ o Yg ?'""**
o
^
_ ~ '.
' 4jf f.w,...
p l _
_W_.rh
.h-M k-hkf_-7 i
.i -
'l f I ', j j
i
'I i
i
'1
_ '. _ _. :l i
l l
E 1
1.
', I' dh
.i N
I y
4 i
1 ew w
is. e.e4g ies e_.. s.u.s.e.
.-W pE e.
y gyp 94 h ge.n g en.p.
--g e.i.g e
1 l
a.___.
.I i--..-...-
l
,'.j
.]
I i
.I i
a
s'if Appon$ix $-
- ,,f Revision 0 :.
-l
- Page 6 o f 10 '
g WMD/sp.
"'M" "' noer
,o.T,
\\
I
/'
h_-
3 t
0 (0.-)
h m c L_ o vi O.3
/
/
PRES $URE. PILOT $UPPLY PORT (P)-
PORT (PS)
ToRogoroR__,00 00 s mmv s
s s m
s s
(N
" " ' ~ ~ '
{;fiIS lg]
f l%
' 1%'s?u*Rt k
4 M %i l s F m 5 s ( _K f, a"U'r'u'$f E ~ M" lf,//////////_ f/7Lg h b y / Q } Q'f 4 roact bb'b N-&l// }; / (( secoNo- ) sT4cc sPoot - fpp/ /N Bias r [, [ /,lllll Ill1ll f / Spring l 11111' l ' ' l'l / ,/,,
- o* RING PS E*ealSH
' $ TRAINER acs.,,,, na ~,o-,---. en 9 ~ 2 b'-87 i b u!2 / _l f )
4 'h, f> q ure ' 2. ' Genera / Arrasye men / _ o f f//g g;f Ay,;,. +s denfro / - Va /ves .Conhol' f ca,4,,g ~) ' Con lu e f, - y e Ive f y, ;,, sta<o l. _ su.m' o.a.. [ I" f j ll " o,/ 1 n d' n s,
- s oci,t '
) y re duesng 4ee i I s Fa ilv<ea a I iv m <>- z and l ) Bn ns w ck i s ce vere d i h ,n y;pe p iove ' wild /\\ I' W Limenck or, weld +haf faihd af NQ 9 9 7 ? D.,, was n o/ c: : fu // I peneirabon kveld t b ack s cle / 3 .- j} en 59 '9 y y{ %heI-w:Ided T Redscing ee Frp<es 3 and 4 ~--------.._._.m~-}}