ML19325D329

Comm Ed Rept of LaSalle Unit 2 Trip,890826.

ML19325D329
Person / Time
Site:	LaSalle
Issue date:	10/02/1989
From:	COMMONWEALTH EDISON CO.
To:
Shared Package
ML19325D328	List: ML19325D327 ML19325D329 ML19325D332
References
NUDOCS 8910230102
Download: ML19325D329 (35)

Text

.

Commonwealth Edison Report of LaSalle Unit 2 Trip August 26, 3989 l

{

8910230102 gDR ADOCK 050008910o$74 FDC

. _____-__s

E TABLE OF CONTENTS EAGE I. INTRODUCTION 2 II. DESCRIPTION OF EVENT A. Initial Conditions 3 B. RPS Actuation 4 i

C. Equipment Performance 6 l III. EVALUATION OF PLANT TRANSIENT A. Information available for evaluation 7 (

B. Potential sources of trip signal 7 C. Evaluation of Control Rod movement 11 f IV. EVALUATION OF ' RPS EQUIPMENT ,

A. Potential causes of Observed RPS operation 17 B. In-plant testing performed on RPS 19 C. Laboratory Testing results 22 V. SAFETY EVALUATION OF EVENT 26 [

VI. COMMITMENT LETTER 27' VII. CORRECTIVE ACTIONS 32 VIII. CONCLUSIONS 33 ,

IX. APPENDICIES APPENDIX A GENERAL ELECTRIC EVALUATION OF EVENT r

APPENDIX B OAD CONTACTOR TEST PLAN / REPORT i i

APPENDIX C PRINTOUTS, RECORDER TRACES s

APPENDIX D ADDITIONAL REPORTS AND DOCUMENTS l

ZCADTS Page 1

~

f)

. I. INTRODUCTION On August 26, 1989 an unexpected Reactor Protection System (RPS) actuation occurred at LaSalle County Station, Unit 2 during valve testing of the Main Turbine Stop Valves. Although all control rods were automatically inserted into the core, one of the primary automatic shutdown channels did not appear to operate correctly. The backup scram system did work as designed.

This report describes the. event and investigation into its cause and the operation of the RPS channels. The report concludes that the initiation of the event could not have been a valid scram signal. The studies done for the report show that the observed operation of the RPS system was not the result of mechanical binding but could be mimicked by applying trip' signals for short (8 to 12 milliseconds) intervals. Such signals were observed to cause only one of the two relays in the RPS automatic scram channel to actuate (de-energize and stay de-energized). .

CECO believes that the most plausible explanation of this event is that -

it was initiated by a spurious signal having a duration of 8 to 12 milliseconds. Because not all plant monitoring equipment was available at the time, the exact source of this spurious signal cannot be  ;

determined.

c t

j i

l l

i l

t l

i l

l l

l l l l

1 ZCADTS Page 2

.i

. II. DESCRIPTION OF EVENT y A. Initial Conditions. I I

1. LaSalle Unit 1 LaSalle Unit 1 was operating at 99% power during the event on Unit 2. The Unit 2 event did not affect Unit 1 operation.

g 2. LaSalle Unit 2 t Unit 2 was operating in the RUN mode at approximataly 10%

reactor power, with the main generator of f-line. Turbine valve tightness testing was in progress, in accordance with i LaSalle Operating Surveillance procedure LOS-TG-SA2. Both t annunciator alarms for "Turb Cont /Stop Viv Trip Bypass" were  ;

up.

5 I

t

(

5 i

l i t

(

( l l

l l

1 l

U I

i I

l t 1 l

i 1

ZCADTS Page 3 b

1 f

=.

J I

B. RPS ACTUATION On August 26, 1989 a controlled shutdown was in progress'on Unit 2 -

in preparation for a planned maintenance outage. The time line on the following page summarizes the event sequence. At 0336 hours0.00389 days <br />0.0933 hours <br />5.555556e-4 weeks <br />1.27848e-4 months <br />, l LOS-TG-SA?, " Turbine Valve Tightness Test," was begun.

After successful completion of the Turbine Control Valve leak tightness portion of LOS-TG-SA2 at approximately 0412 hours0.00477 days <br />0.114 hours <br />6.812169e-4 weeks <br />1.56766e-4 months <br />, an L Equipment Operator (EO) installed a jumper on the #2 Main Turbine Stop Valve (MSV 2) pre-amp function board in the EHC cabinet. This causes the Number 2 MSV to close slowly, after reaching approximately 90% open valve position,.the other MSV's (1, 3, and

4) lose their open permissive and begin a controlled closure (approximately 20 seconds close time). At 0413:06, MSV's-1, 3, and ,

4 reached the full closed position. ,

Approximately 23 seconds after MSV's 1, 3, and 4 reached the fully closed position, an indication of a reactor scram was received by .

the plant operators. The Unit 2 Nuclear Station Operator (NS0) heard the RPS actuation and control board alarms. While walking towards the reactor control panel, he noticed that 2 of the 8 RPS rod scram group monitoring lights (A2 and A3) were energized. He instructed an additional NSO who was closest to the manual scram >

buttons to manually ~ scram the unit. This was done immediately and resulted in the A2 and A3 scram lights de-energiz!ng, 12 seconds after the initial actuation. The Unit 2 NSO then proceeded to manually select control rods to verify position indications. All control rods indicated full in.

Approximately 3 seconds after the manual scram, another NSO in the control room placed the reactor MODE switch in the Shutdown position, in accordance with LGP 3-2, " Reactor Scram."

At 0414, the process computer printed out the results of a rod position scan which had been automatically initiated at the same time as the initial actuation. This printout appears to indicate that all control rods were in motion, apparently as a result of the backup scram actuation, prior to the manual scram signal. At 0415, a subsequent manually demanded printout showed all control rods to be full in.

Post scram equipment performance was normal, with no significant problems encountered.

Page 4 ZCADTS 0253T:8

• EVENT SEQUENCE LaSalle Unit 2, August 26, 1989 IIME EVENT / OBSERVATION 0412 (approx) Operator pinced jumper in Main Stop Valve (MSV) 42 control circuit per test procedure. ,

0413 06 MSV 1, 3, 4 full closed.

(AS'LXPECTED) 0413 29 Reactor SCRAM. Backup Scram Channel "A" actuates.

0413 30 Rod DRITT alarms.

Unit Operator noticed A2 and A3 rod group SCRAM solenoid '

lights still lit.

Unit Operator instructed extra operator at feedwater controls to " ARM and DEPRESS" manual SCRAM.

f 0413 41 Manual SCRAM. A2 and A3 lights de-energize. Backup SCRAM channel "B" actuates.

0413 44 Reactor MODE switch in SHUTDOWN.

0414 Automatically triggered control rod position scan printed (scan started with auto scram signal).

0415 Second control rod scan (manually demanded) printed shows all 1 rods fully inserted.

l l

l l ,

l l

l l

r i

I I

l ZCADTS Page 5

• C. EQUIPMENT PERFORM &tiCE 2

The operation of plant equipment was reviewed to evaluate anomalies which may have existed in the following areas

1. The source of the RPS actuation signal.

2. The time that all control rods began moving, and the point i where all control rods fully inserted. .

3. The cause of the A2 and A3 rod group scram lights remaining lit until the manual scram.

4. Separate indications received by the operators during the performance of the turbine surveillance, showed that MSV's 2 .

and 3 appeared to have faulty position indication.

t i

t 4

1 I

I l

l ZCADTS Page 6 i

. U

1. .

• l.

III. LyALUAll M OF PLANT'TRANSIENI. I I A. Information Available for Evaluation l In the normal review of plant transients, a key source of information.is the Hathaway Sequence of Events Recorder (SER).

This computer records most of the information related to RPS via alarm relay actuations, including.the subchannels which tripped-and I

the time that each tripped. Because the SER typer had been turned ,

off shortly before this event, the analysis must rely on the  !

reconstruction of data from other indications. Some data cannot be i obtained by any means. Thus, the evaluation of the event requires some assumptions based on knowledge of the physical processes in the plant at the time of the trip.

The primary sources of recorded data for this event are the plant  ;

I process computer (CX), and the Startrec computer. The process computer prints out certain digital points if they change state for

~

at least I second (its digital point scan rate). It also printed a "NSSS post trip log" which prints a historical log of 10 Analog sensor readings (stored at 5 second intervals) retrieved from 5  !

minutes' prior to its initiation untti it is manually stopped after i the trip. This log was initiated by the receipt of the "A" backup scram signal which accompanied the initial automatic scram. The Startec recording was initiated directly by the automatic RPS signal, and records selected signals from .5 seconds before the trip to 1 minute after the trip. It has a scan rate of 20 milliseconds per scan..

Attachment C contains copies of the available computer and chart recorder printouts.

B. Potential Sources of the Trip Signal To evaluate the cause of the scram, a list of all possible scram .

signals was generated. Available information (including Control Room chart recorder data) was used to eliminate any parameters which it could be shown had not exceeded their scram setpoints (LSSS value) at any time. Table 111-1 summarizes this review.  ;

The result of this evaluation shows that no LSSS variable actually  :

reached a required scram setpoint nor were any variables approaching a LSSS setpoint.. Therefore, any RPS channels which received trip l signals were receiving only false trip indications not reflective of the true process conditions, j To further evaluate the possible source of a trip signal which is known not to indicate an actual process condition, the following ]

L availabit information is used: l l l l 1. The "A" and "B" RPS channels received.their trip signals very 1 close in time, as indicated on a print of the Startrec data. l The "B" 1/2 scram is present on one Startrec scan, and the "A" l 1/2 scram is present 20 milliseconds later. Since the "A" I signal is scanned prior to "B", the signals could have I occurred less than 20 milliseconds apart, but they could not i be more than 40 milliseconds apart.

l ZCADTS Page 7 l l

l

. t t

2. Ccnfirming thi cperctsra cbs3rvatien, th3 "A" chann31 b ckup scram occurred simultaneously with the initial actuation. The l "B" backup scram channel cid hot actuate until the manual scram pushbuttons were pressed.

L

3. There was no Recirculation Pump Trip (RPT) signal initiated. [

l l l

4. None of the possible confirmatory indications for the t existence of a scram signal were found. These would be t l

[

signals such ast  ;

Process Computer digital points accompanying Reactor Water .

Level 3 trip, or PCIS isolation Groups VI, and VII.

• Main Steam Line Radiation monitor drawer actuation trip l

indicator. .

MSIV isolation signal (s) - no MSIV isolation occurred, and the I

( Startrec channel monitoring the MSL hi-rad trip signal did not show any actuation.

The operators confirmed that they did not see the, Turbine 1st. .

stage pressure scram bypass alarm (s) clear (they would reflash '

if cycled, and would be found in the fast flash condition instead of solidly lit). +

No APRM panel trip indicators (latching lights) lit at the APRM back panel. This was' observed several hours after the l event.

The reactor MODE switch computer digital inputs change to "not-RUN" 3 seconds af ter the manual scram, confirming that [

the mode switch was in RUN at the time of the actuation. ',

1 No Scram Discharge Volume high level rod blocks were present, l which would be present prior to reaching the scram setpoint.  !

L t Item 1 above indicates that the trip signal originated from a single  ;

event which affected an "A" and a "B" RPS subchannel at the same time.

A review was conducted to determine where the two RPS' channels have.

common components, wiring, or sensing piping. This review foundt No sensors are common to "A" and "B" RPS-(3ther than the MODE switch).

l No sensors from "A" and "B" RPS have wiring routed together (same cable trays) between the plant and the control room RPS panels.

l Although they have separate power supplies, tae APRM panel bey has ,

L one APRM drawer feeding "A" RPS (APRM A) and Jne drawer feeding "B" l l RPS (APRM B). Each of these APRMs feeds only one subchannel in its l'

RPS channel. ]

l l

1 l

ZCADTS Page 8 .

• - Thr00 RPS inputs havo instruments from both "A" cnd "B" RPS which share a common sensing line. (Refer to Figure III-1). These are

1) Turbine first stage Pressure, 2) Narrow Rang- Reactor Water Level, and 3) Scram Discharge Volume High Water level trip. In each case, only one RPS subchannel in each RPS channel is serviced by instruments which share a sensing line. For example, the pressure switches for Turbine 1st stage pressure which bypass MSV trips for subchannels Al and B1 share a sensing line, and A2 and B2 share another line. In no case are both subchannels (say Al and A2) serviced by instruments which share a common sensing line.

Testing was conducted to determine if a trip could be induced on an instrument using one of the common sensing lines. This was done by dropping a rubber mallet onto the sensing line. No trips were introduced by this method. However, because there are other mechanisms which could potentially " spike" the sensing line internally (including process noise and instrument deflection), it is still considered possible that the original trip affected one of these lines in some manner. Part of the evaluation process involved comparing possible trip combinations to attempt to determine which subchannels actuated (i.e. Al and B1, A2 and B2, Al and B2, etc). None of these evaluation checks provides positive information, however, some of them do indicate a lower probability. These combinations are not listed in complete detail here, but the more significant ones are discussed below Scram .jpass removal by Turbine First Stage Pressure Switches:

If a removal of the First Stage Pressure Scram bypass occurred, it would be accompanied by a Recirculation Pump Trip (RPT), as long as the logic combination was for channels Al and B1 or A2 and B2. No RPT occurred. It is not likely that the scram bypass for Al and B2 occurred simultaneously (<40 msec) because these channels have separate sensing lines.

l l

Scram bypass removal plus Stop Valve 42 reaching RPS limit switch.

l i

One postulated condition was that due to MSV 2 slowly closing, it hadn't reached its RPS limit switch until the trip. This requires an A1/B2 combination of pressure switch scram bypass removal j

because MSV2 feeds channels Al and B2'. Also, this combination would not enable either RPT channel. This combination is not considered likely for the following reasons: 1) Two earlier l

turbine trips did not result in any 1/2 scrams, indicating proper I scram bypass operation earlier, 2) Instruments on separate sensing

, lines would not be considered likely to falsely clear together, and

3) Startrec indication of MSV2 position indication of 80% has been l

verified to be accurate, and is well below the 95% RPS limit switch sotting (indicating that the trip would have occurred earlier).

l l

t i

f 1

l l

i l

l l

l ZCADTS Page 9 l i

i

\

V L.

4

. \

C. EVALUATION OF CONTROL ROD MOVEMENT ]

Since actuations from both the "A" and B" RPS channels occurred, I automatic control rod motion for all control rods would normally be '

expected. This can be seen from the summary of the RPS logic shown  !

in Figures III-2 and III-3. The observation by the operator that i the A2 and A3 scram solenoids were still energized indicates that the control rods in those' groups did not initially receive a scram signal, and would not necessarily be moving immediately after the initial actuation. His action of manually scramming the Unit ensured that the final 2 rod groups were inserted, although the first 2 rod groups (Groups 1 and 4) were already full in at the time of the manual scram.

The process computer has digital inputs for the backup scram l channels A and B. Because of the logic to the backup scram, i actuation of either channel indicat9s a " full scram" condition, This initiates several computer routines. One is the initiation of j the NSSS post trip log, and another is a recently added automatic  !

control rod scan program which will continue to perform full core '

i scans (each one takes 30-60 seconds) until it detects that all rods are full in. The first automatic scan printed approximately 45 seconds to 1 minute after the initial actuation. From.the scan sequence used by the computer, some of the rods in Groups 2 and 3 which are read early in the scan can be seen to have " bad" position indication, indicating that they are in motion prior to the manual scram at 12 seconds into the event. (i.e. that the rods were  ;

ir.oving or had recently moved and not settled into a notch position l by the time they were scanned). This is the normal indication  !

shortly after a scram. The process computer / rod position interface is not designed to provide valid transient. data during scram l conditions, which Ilmits the usefulness of these Indicated " bad" l l positions. However, this first printout does :Upport the conclusion l that the G2 and G3 rods had moved some distance into the core at I the time of the manual scram. The rods which are shown at position 1 00 (full-in), were full in prior to the event. l l

The operation of the backup scram channel is designed to vent off J the scram air header, causing all rods to scram. Since 1/2 of the rods (G1 and G4) received automatic scram signals, their

" inventory" of stored air was vented immediately by their local I scram solenolds. Also, since the largest portion of the air volume I at the HCU is between the scram solenoids, and the "B" scram j

! solenolds vent this air off, additional venting for all rods l l~ occurred because even the G2 and G3 rods received the "B" channel ,

1/2. scram. This appears to have allowed the backup scram actuation l to vent the air header more effectively than if it had to vent the l l entire air inventory. This would be expected to allow those rods to begin moving in shortly after the initial actuation. These rods ;

I would be expected to start moving somewhat later than the l automatically scrammed rod, and they may have started moving more i

slowly than normal scram speeds. This is because the scram valves

will throttle slightly until the air pressure drops enough to allow j l

ZCADTS Page 11 l

0253T:15 l l

i Proosuro pulso en roacter water lovol instrument eensing lino, transmitted to reference legs of two instrument racks.

- Since the 2 wide range level reference legs feed instruments at all  ;

4 instrument racks, RPS divisions can be " crossed" by pulses ,

transmitted through instruments. This possibility is not I considered plausible because pulses originating on the reference leg side of an instrument are inverted on the variable leg side j

'(1.e. the pulse could drive the first instrument downscale, but ,

transmission through tha instrument to the variable leg drives the I other instruments on that variable leg upscale). This inversion  ;

prevents getting simultaneous trips even though the second trip may i occur when the instrument " rebounds".

I Although the exact mechanism could not be determined, the information above is useful in determining that only one RPS subchannel in "A" and one subchannel in "B" could receive a trip signal from a single source external to the process itself. Based on this, the observed operation of the RPS scram solenoid lights indicates that only one contactor did not operate when demanded.

i b

i i

l i

[

P l

i i

1 4

ZCADTS Page 10 j i

i

the scram valves to pop fully open. Since the air header was ,

apparently depressurizing rather quickly, this time of slower than 1 normal motion is probably not very long and in any cate, the rods received a normal scram at 12 seconds due the operators manual action. l for Unit 2, all control rods have scram insertion times from fully l' withdrawn of Isss than 3 seconds (per most recent control rod sram timings 6/89).

Therefore, the rod insertion times can be bounded to conclude that  !

half of the control rods were full in 3 seconds after the initial actuation, and the other half reached full in no later than 3 seconds after the manual scram. Many of the second half of the rods may have been full in by the time of the renual trip due to starting positions of less than full out, and some prior movement i due to the backup scram. The APRM traces included in Appendix C show that the flux level read zero for approximately 7 seconds prior to the manual scram. ,

l ZCADTS Page 12 i l

l 0253T:16 L _ __ - __ _ ___. _. _

. TABLE III-1 -

CHECK OF LSSS PARAMETERS ,

TRIP DATA SOURCE / INDICATION 1

1. IR:4 HI N/A - Mode switch in RUN i

2. APRM HI (15% in Startup N/A) -

Flow Blased CR recorders, Startrec Plot APRMs <15%, no spikes  :

Fixed (118%) NSSS post trip log 10%, stable

3. Rx Pressure CR recorder 940-950 psig  ;

(1043 psig) NSSS post trip log 933

• Startrec 940 psig

4. Rx Water L3 NSSS post trip log 32" (12.5" NR) Process Computer -no L3 alarms from K6-X relays Startrec NR 37", stable

5. MSIV Closure CR indication, MSIVS open I Startrec - all MSIVs open l

6. MSL rad CR indication (recorder), stable Startrec monitors trip signals (K7), none trippe6

7. Primary Cont CR recorder <.5 psi -

Hi Press No PCIS isol.

(1.69 psig) Startrec .3 psig, stable

8. Scram Disch. No rod block present l Hi level <

9. Stop Valves Stop Valves closed. Scram bypass alarms up, Not Full Open Previous turbine trip no scran/ 1/2 scrams -

1 l 10. TCV Fast Close Startree traces no TCV fast close demand TCV's open at time of scram, close when Turbine Trip

11. Mode Switch Process Computer. Mode switch moved out.of run at  !

In S/D 04:13:44 l

12. Manual Scram Startrec, manual scram 12 see af ter Auto  :

I

's

13. CRD low chrg N/A, Mode Switch in RUN water header ,

press. ,

l

)

ZCADTS Page 13

l '!

. l 1

I 1

l u

)

1 FIGURE III-1 <

)

EXAMPLE, RPS SHARED SENSING LEG i

(TURBINE ist STAGE PRESSURE) l 1

i i

DOC?Lt. DNO m e. . g el e PS -

= a - , ,- a .% 1185 t.00C tur.,i-in D, IW-- Du pg g

@ - wamnen 3g .e W#W1 l g y.

. - -e, a h,, , . .._ mr.-

h.

88 c . , ,, a 3

W W s 8 1 m -- a

, u - ,u , y > == . ,nw--

46 00 ,

V==mm L. "" ,, * ,,

2C11 ~~

2 Cit e g

% .,. , - - -,2 m . - ra,, '

1. tCSLOOC aurisntiI .

j W s'u AC14 I . ==e. g iH'F1 AAl r>

D601LS DM0 ,

l DET AIL" A'

• l 111. STAGE PRESSURE l

l 1

l 1

l l

FIGURE III-2 REACTOR PROTECTION SVSim l i

J i

i RPS CHMWEL A RPS CHMWEL 3 3 rh E E )

I"9 3l',l.-." ,

l ,

c, ,, H M14 R g'2:as

' ,,8 m, ,,F D,, l 1 D <w @ @ @ @ @ @

i runcti <w @ @ @ @ @ @ @

@ @ @ ~@ @ @ @@  ;

AUTOMATIC SCRAM: .

- De-energi.e to .otuate 1/2 Twice losio

- Actuates assigned rod Groups N

n n n .

==A ntactor' f= E

- ==G ==C

==C X14 --- ==G =E ==A 45-G1 RODS 45-G2 RODS 47-G3 RODS 48-G4 RODS

% 4F hel 4

• ~

L 'N!, lL( '

NR , .,

= =D. , = =H, , ==F ==B

==B. -- F . '

TD SCRAM

=f-H '

T O

, y , I,I ,

AIR

y g ,19)

.

Header M14 bA Q C

,  : O  ;

F H BACKUP SCRAM: <

Backup "A" -] : -Energize to actuate l [. - - -

Backup "B'

< -Driven by Auto scram )

B 3 -Sorens all rods E G 125 VDC 125 VDC

FIGURE III-3

. RPS LOGIC DWIPLE LaSalle SUSCHRfGEL fRTl W 120 URC g n g 0 Turb. Stop Ulv Not Open i j = = Turb ist Cont Ulv Fast Close o \

a "StePress g i

u

[.

Scran Disch Uolume Hi u

\" "'B/P su _

o 3

MSIUs not Full Open u m n\= ='Suitch "

Rx Mode ..o -

t n-./ . d.

1 gL HI DAl Pressure .o l HI Rx. Pressure a a J

K WWI  !

L0 Rx. Water Lvl MSL HI Rad n d ,

i HI Neutron Flux " u

( IIM/ RPRM -RUM) H Manual Soran n -----------~~~! Kt40 Kt4C Contactor 14E  !

r -' q, Kte H-s' (Kt4_)

1 i..l L 9'

6. s X I 1

\

s f 'M I

i X15 X19_

IWRIAL SCRAM SCRAM RESET I

l SCRAM RESET LOGIC SW. Roset Resets B/U Position Relag Contactor Lockout X1 g g g Gi+G4 X19 Xi G2+G3

= * < llll l X190 gg g

-i

__ __. _______a - _ _ _ _ _ _ _ _ - - * -*--_u- + = _ _ _ _ - -

• _m -

IV. EVALUATIN OF RPS EQE12tG2[T A. POTENTIAL CAUSES OF OBSERVED RPS OPERATION Because of the design configuration of the RPS (see Figures III-2 and III-3), there would normally be no valid trip condition which would result in 2 of the 4 scram solenoid groups remaining energised. This is because each logic string (subchannel) actuates 2 contactors. One of these contactors actuates rod groups 1 and 4, the other actuates groups 2 and 3. Figure III-2 shows that in order for the A2 and A3 scram solenoid group lights to remain lit, contactors K14E and K14G must remain energized. The following discussion refers to the apparent actuation of one RPS subchannel (either Al or A2). Since both contactors in the subchannel see the same logic input, the condition where the A2 and A3 solenoid groups remain energized implies one of the following possibilities:

1. Contactor K14E or G is failed in the energized state.,

2. K14E or G takes longer to de-energise and open circuit than the duration of the trip signal.

3. K14E or G had a closed circuit around its seal-in path so that it de-energized only for-the duration of the trip signal, then .

re-energised.

4. Some kind of voltage pulse was introduced on the neutral side of the power for a "B" RPS subchannel (or the whole channel),

and its energy dissipated prior to reaching the "A" RPS channel, such that it raised the ground potential only enough to drop one contactor.

Possibility 91 was eliminated shortly after the event by several functional checks, all of which showed proper operation of the contactors. This was also confirmed by laboratory testing.

Possibility 82 was the subject of extensive testing, described 3 below.

Possibility 93 was the focus of the initial functional tests, which ,

were intended to show proper bi-stable operation of the contactors  !

and proper operation of the reset-circuitry. No indications of an unexpected reset path have been received. There are two ways.that the circuit can provide the reset path: 1) the reset relay is failed in the energized state (or it has a stuck closed "a" contact), or 2) the reset switch is continuously holding the reset relay energized. Tho operator depressed the A2 (and B2) manual j ceram button and the A2/A3 lights did not re-energize when it was released. If the partial actuation of RPS A was caused by.the reset path being present, then either this condition cleared prior  !

to the manual scram, or the problem was affecting subchannel A1  !

only.  !

s l

l I

i ZCADTS Page 17 l 1

i

b Ev:n with o p3rm:n;nt rc0st pith, th3 initial pulsa wculd htv3 to '

be short (less than approximately 50 milliseconds), because the backup scram channel A relay forces a lockout of the scram reset to all the "A" RPS contactors (see rigure III-3).- Since the A backup channel actuated, this possible reset path would have been removed ,

quickly. The reset pathway is not considered likel- ' o be the cause of the A2 and A3 solenoid lights remaining 13.  !

Possibility 04 was not specifically tested for. During the numerous tests performed on the RPS system, recorders were connected to the bus power and various points in the circuit. No' unusual " noise" conditions were seen on either the "A" or "B" RPS channels. If this condition took place, it was not present during i subsequent testing and has' not repeated (no other spurious trips I

have occurred - these would be observed during plant shutdown as

.well as during operation),

.l i

l l

1

-l I 1

i l

1 i

i l

t I

j

)

i ZCADTS Page 18 l

)

s

-(

'- 6. In-plant testing performed on RPS.  :

Several tests were conducted on the installed RPS components prior-to and after removal of the K14E'and K14G contactors for laboratory .

l inspection and testing. These tests are summarized in Table IV-1.

All of the tests'showed normal performance. However, on two )

separate (non-test) occasions, anomalous indications were received which appeared to have some similarity to the original RPS i actuation.

On August 27, 1989 while preparing connections for response time

tests on RPS subchannel AI, the technician reported that the test lead fell off of the terminal block. This caused a 1/2 scram condition (as it should). However, another technician at the main control panel and the licensed operator both reported that the A2 l and A3 scram lights appeared to de-energize slightly later than the i Al and A4 lights. Additionally. the technician in the back panel ,

performing the connections heard two relay actuations instead of l the usual single " thud" of the contactors dropping out in unison.

Because the test t.onnections'were still in progress, the recorder i was not running and no hard copy data was recorded. The cause of ,

the observed behavior was not determined. However, it was considered possible that the " wait" time since the last (surveillance induced) trip signal.was related to the behavior.

Therefore, the test was delayed at varying intervals to attempt to duplicate the interval between.the most recent surveillance trip i and the " slow drop out" observation event. No repeat of the -

! observed behavior was possible. It was also considered that the test lead may have made many-momentary " bounces" as it fell off, one of them being sufficiently short to cause a single contactor to

drop out, and the final state of the circuit (open circuit),

l dropping out the second contactor. Attempts were.also made to t duplicate or force a " bouncing" lead to cause such spikes. Although rpikes could be induced that caused'short interruption of the contactor power source, the mintrum spike was about 30 milliseconds (this duration is also uncontrol.lable). No further attempts to duplicate the observed behavior were made.

On September 8, 1989 an instrument technician was returning the "D"

l. Intermediate Range Neutron Monitoring (IRM) drawer from the " trip

test" mode to " operate". In the off-normal condition.for this l method of returning the drawer to service (no 1/2 scram was up as it usually is during this portion of the surveillance), the l technician told the operator to " expect a 1/2 scram". When he '

turned the mode switch to operate, a trip signal occurred to the B2 RPS channel, which is fed by the "D" IRM. However, only the B2 and B3 lights went out, with B1 and B4 staying lit. This'was a stable condition (the lights stayed that way), even after the technician  ;

released the IRM drawer. All activity was stopped and investigation u was initiated. After taking contact resistance readings, inspecting the K140 contactor, and consulting with GE and CECO Engineering, it was decided to: 1) Perform special testing on the scram reset I switch to look for potential sticking contacts, and 2) reset the trip condition and devise a special test to attempt to duplicate the short pulse event.

ZCADTS Page 19 0253T:20

' i

\

l Subsequently, a test was devised to insert short pulse trips from the "D" IRM drawer. This test used the installed relays to l transmit short trip demands (inserted into the IRH logic) to the 1 RPS. .The test inserted 470 trip signals.into the IRM drawer. Most 1 of these were too short to cause any RPS' relays to operate. l although most of them did cause the IRM trip relay (HFA) armature  !

to begin to drop out, and then re-engage without having open 1 circuited any of its "a" contacts (and without closing any of the "b" contacts. The tests caused a total of 59 trips (1/2 scrams). l Of.these, 30 were estimated to open the logic contacts to the  !

K14D/H contactors (pulse widths) between 8-12 milliseconds. From '

preliminary laboratory test results received at the time of the IRM drawer test, it was expected that about 1 in 8 pulses inside.the 8-12 millisecond window would result in only 1 contactor in the pair dropping out. Later, this number was observed to be much smaller, approximately 1 in 50 to 1 in 70 (more extensive *

. laboratory data obtained after the IRM testing). Therefore, it was concluded that the'30 samples of the IRM drawer test would not necessarily have been expected to reproduce the single contactor'  ;

dropout condition.

This event (on 9-8-89) was the same type of narrow pulse-single  :

contactor dropout condition as had been demonstrated in the laboratory testing. Since the in-plant. testing verified that the t Hathaway alarms were consistent (or explainable) with acceptable conditions which are known to cause a single contactor dropout, the K140 contactor is considered normal and was left in service.

The IRM testing did show two results which assist in understanding the sequence during the Instrument surveillance on 9-8-89. These are that 1) The narrow pulses (8-12 milliseconds open circult to  !

the K14 contactors) always result in the HFA relay not completely I dropping out. That is, the alarm point.(See figure IV-1) which is monitored by a "b" contact never occurs when the pulse width is inside the 8-12 msec window, and 2) The Hathaway point input circuit which monitors the "B" neutron monitoring. alarm will'not change state faster than 27 milliseconds,.even when the contacts are closed for only 2 milliseconds. This appears to be a i particular condition of the input circuit for that channel, and  ;

will be investigated as a separate question. These results showed l that the 9-8-89 event is completely consistent with the other  !

information which suggests that the single contactor dropout is 1 only caused by short periods of open circuit in the logic (short  ;

pulses). l l

ZCADTS Page 20 0253T:21

r l

i l -

The observations from the 8-27-89 occurrence during preparations for response time testing of the Al RPS subchannel do not lead to a clear indication of the condition of the K14E contactor at that l

time. This is due to the fact that no hard copy data was obtained,  !

I and the observations are somewhat subjective. It is not possible a to make good estimates how the test connection open circuited - I whether cleanly or with significant bouncing of the circuit connection in the precess. The time djfference between the contactors dropping out is clso subjective. If a significant time. j difference between the contactors dropping out was in fact present, and the test lead had cleanly open circuited at the onset, then ,

some physical binding of the contactor would be indicated.

However, it the potential error in the personnel estimates of the time difference between the lights is sufficient to cause estimates of "I to 2 seco Js" to a light indication that is delayed by 30 ,

milliseconds, and the actual condition of the test lead falling off is to cause " rattling" of the contactors before they finally trip l

open, then the event most closely represents the verified laboratory pnenomenon of a short interval interruption of the logic string.

Because of subsequent inspections and testing of the K14E contactor, the question of physical binding appears to be resolved  !

as unlikely. If for some reason, the actuation (s) subsequent to the event had caused the contactor to change such that any physical binding was cleared away (masked), then the replacement of the contactor will have eliminated the possibility of it being a problem in the future.

~

l 1

o i

l l

ZCADTS Page 21

C. LABORtTORY TESTIOG RESULTS i

The initial evaluations of the plant transient were not able to determine which "A" RPS subchannel (Al or A2) was challenged. [

Therefore, it was not possible to determine whether the K14E or the }

t K14G contactor was suspect, and the decision was made to replace both contactors and inspect and test the removed ones. A test plan  ;

was developed by the CECO Engineering Department and Operational  ;

Analysis Department (OAD), with assistance from General Electrie. f The contactors were removed and sent for implementation of the inspection and test plan, under observation by vendor and NRC {

Personnel. j The test report is included as Attachment B. The inspection i results did not find any degradation of the removed contactors which would hava affected their performance during this event. The teating program determined that the single contactor dropout f condition was repeatable if the relay contact which controls the ,

contactor actuation are open for 8 to 12 milliseconds. This condition appears to result in a single contactor in the pair  !

dropping out approximately one time in 50 demands (if the demand is in the 8-12 millisecond time window). {

These results are consistent with the observed behavior during the l initial actuation and during the event on 9 8-89 (IRM "D" trip).  ;

i t

i i

i

)

l j

l I

I j

l i

ZCADTS Page 22 l l

)

t

• TABLE IV-1  !

IN-PLANT RPS TESTING l t

DATE IES.T._1t04 DESCRIPTICav/RESULTS 8-26 99 LST 09-090 VISUAL & TUNCTIONAL TEST Or U2 CONTACTORD Contactors in "A" RPS were inspected in place  !

(visual, mating surfaces inspected with boroscope).

Manual trips were performed to verify proper trip and reset logic. Tound loose ausiliary contact ,

plate (still functioning) in K14G. Loose wire  !

check, tightened several connections, none  ;

~

significant.

8-26 LST 89-091 RESPONSE TIME TEST OF SCRAM CONTAC70RS 8-27 Channel drop-out tests to compare drop out times of  !

contactors in each pair. All ("A" and "B")

subchannels tested. Each contactor in pair had drop out times within 5-6 milliseconds of each other. i Technician noticed R14E/G difference while hooking up (not recording). See test. ,

8-29 LST 89-094 RESPONSE TIME TEST Or "A" RPS CONTACTORS Similar to 89-091 with difference in recorder monitoring points to obtain coil voltages. Same results.

8-29 LST 89-096 SCRAN CONTACTOR ELECTRICAL DATA Measurements of coil operating voltage, resistance, contact resistances, varistor currents. All i measurements nominal.

8-31 LST 89-200 INSTRUNENT SENSING LINE SENSITIVITT Test to see if impact on sensing line(s) at various .

points along line will cause trips. No trips 7 initiated (Rm Level, Turbine 1st stage press, SDV i level sensing lines tested). '

8-31 LST 89-101 CONTACTOR ELECTRICAL SPECIAL TESTS Obtained operating temperature, resistance, coil ,

shorted turns test, cold coil resistance, temperature rise daca. All contactors normal.

9-4 Contactors K14E and K14G removed and replaced. Old contactors sent to CECO Operational Analysis for performance of test plan. New contactors installed / tested per work package.

ZCADTS Page 23

TABLE IV-1

\

IN-PLANT RPS TESTING DATE TEET._RQ4 DEECalPTICnUREsuLTs 9-8 LST 89-108 SCRAM RES3T SWITCH TEST No contact or switch position " sticking" found.

Roset switch and relays appeared to operate properly.

at all positions.

9-15 LST 89-116 IRN/RPS trip testing.

Very short IRN drawer spikes " induced", 59 trip signals to RPS subchannel 32, 30 of the tri' i

estimated to be 8-12 maec wide. No "1/4 ses xs ' " .'

f l

l l

l 1

i I

l l

l l

l i

ZCADTS Pege 24 l

t

?

FIGURE IV-1 IRN/RPS ALAIN CONFIGURATIDIE AND TEST SETUP LST 89-116, 15 SEPT 1989 ERM Dract st.xx INo!

KES - - - - - - - -

-N ' ** W** * * * !"

s3 s .

(-

cmen ,

- e c ,. A e

ref. I kit _ _ __ _ _ 4_1

• Yen) *I

~-

I "a * <... t. < t WI. -

i i

"\ gg h -- -h_ _ _ _-4 g ! db1 'cM S1 4 4=70 scu i i < -> Z f _

y , T T Pulse Gen.

L __n kn ---- b

- 4 , , . - . . - . ,-

t- - - -j k" m _ _ _1 234, ---

k14 - - - _- .- _ - -

._ ; E A % 't 'IRm yr"

V. SAFETY EVALUATION OF EVENT +

I The event which occurred on August 26, 1989 at'0413 did not pose any significant safety concerns for the following reasons:

1. There was no plant transient taking place which required an RPS actuation. The RPS actuation which occurred was a spurious  ;

actuation which resulted in the reactor being shut down and placed in a more conservative configuration.

2. The event did not present a valid challenge to the RPS. Therefore,  !

the system design requirement which necessitates that a scram j signal go to completion was not applicable. Had there been a valid ,

scram condition present, the redundant RPS subchannel would have  !

completed the automatic actions (as actually happened when the -

manual scram was initiated). l

3. The Reactor Protection System is not designed to carry out trip ,

signals which are significantly shorter than the minimum response time requirements of the channels. Because the false trip signal appears to only have been present for a time interval of approximately 8 to 12 milliseconds and no RPS channel response time '

requirement is less than 60 milliseconds, no actuation is required by the system design.

4. The scram of 2 rod groups is sufficient to bring the reactor to hot shutdown. The operator actions to complete the rod insertions was  !

prompt and correct. Even if the operator had not manually scrammed -

the unit, all control rods would have been fully inserted a short time later because of the actuation of the backup scram channel. .

5. The potential for slower rod insertion during part of this event (for the Group 2 and Group 3 rods), did not represent any -

i undesirable conditions with respect to the fuel, thermal limits, or i power distribution due to the very low local power levels at 107. l l core power. The operation of these rod groups did not result in a ,

significant risk of an unanalyzed rod drop accident because: 1) '

the probability of a rod drop accident occurring is unaffected by ,

the rod motion, 2) the total time that the rod configuration was abnormal with respect to Bank Position Hithdrawal System (BPHSs) rules is insignificant, and 3) all " fully withdrawn" rods had been verified coupled to their drive mechanisms during the previous t weekly surveillance.

Because the RPS operated as designed when tested under verifieble ,

conditions, there is no potential that exists for an undetected failure ,

to be present which would make operation unacceptable. The actions .

l specified under section VII, " CORRECTIVE ACTIONS" are intended to i increase the understanding of the observed actions or record useful data ,

l should a similar event re-occur.

ZCADTS Page 26 02S3T:27 ,

I

VI. CCMdMMENT LETTER In order to ensure that the NRC was properly apprised of the states of the investigations into the plant performance, CECO committed to specific actions in a letter on August 30, 1989 from D.P. Galle (Ceco) to A. Bert Davis (USNRC). A copy of this letter is attached.

Items 1 and 2 of this letter (cause of RPS response and analysis of initiating event) are covered in the above discussions relating to the plant and RPS evaluations. In addition to the station investigation, separa':e evaluations were conducted by Ger,eral Electric and by CECO's Safety Assessment Department, using trained investigators supplied by INPO. These independent evaluations did not reveal any information not included in this report.

Item 3 of this letter (laboratory inspection of K14E contactor) is covered in the attached Appendix B. This inspection did not find any physical condition inside the contactor which would affect its safety function.

Item 4 of the commitment letter addressed espected service life and preventive maintenance practices associated with the scram contactors.

General Electric's recommendations are included in Appendix A. Although GE identified the contactors as being rated for in escess of 50,000 cycles, it was also noted that earlier malfunction of the contactors could occur due to contact wear or coil service lifetime considerations. Fa!?vres due to these causes would be in the conservative (tripped) condition. In order to preclude spurious trips due to coil life considerations, LaSalle station intends to replace the contactorc before they escoed ?M years of service life. In addition, in accordance with GE't recommendation, the contactor enclosures will be opened and inspected each refueling outage. This inspection will include tightness of the terminals and visual checks of the enclosure i and accessible parts of the contactor for cleanliness. Prior to this event there were no recommendations for preventive maintenance from the  !

supplier, and no history of any coil or scram contact failures at ['

LaSalle.

item 5 (Turbine 1st stage pressure switch drift). Early reports of f large deviations between the turbine 1st stage pressure switch as-found i data and the Technical Specifications were in error because the head )

correction for two of the switches was not taken into account in the reported data. Using the proper head corrections, two of the pressure switches were outside of their Technical Specification listed pressure {

limit by 2.0 and 4.7 psig (Technical Specification limit 151.8 psig).  :

The cause of this deviation was evalutted and seen to be normal  !

instrument drift. The calibration history of these switches (included )

in Attachment D) shows no unusual performance trends. Further, the l Technical Specification basis requires a setpoint to ensure that the scram bypass operation occurs at less than 30% reactor power. Because the original setpoint determination conservatively assumed no feedwater heating, the actual switch performance even with the above mentioned setpoint drift was well below 30% reactor power. The instrument ,

i.iaintenance department will monitor the performance of these switches after the next calibration. ,

l i

ZCADTS page 27

]

I i

Item 6 (Turbine stop valve indication problems). The operators j initially believed that two turbine stop valves exhibited indication 1 problems. This was because stop valve #2 was closing so slowly that the l operators believed it may have an indication delay. As a result of the i test procedure review, it was identlfted that the test procedure was designed to cause a very slow close ramp on the #2 stop valve, and its j operation appears to be normal. The position indication on this valve  :

was subsequently verified to be accurate. Stop valve #3 indicated full {

open during the event. The operators had verified prior to the scram  !

that the valve had actually closed as expected (by use of the process  ;

computer printer). This was further confirmed after the event. The ]

position indication for stop valve #3 was repaired under a maintenance  ;

work request and is operating correctly.

Item 7 (Hathaway computer availability) is discussed below: l At approximately 0300, the sequential memory of the alarm printer was i being saturated due to chattering relays from the (shutdown) Turbine -

Driven feedwater Pump Seal Injection Pressure switch, and Off-Gas  !

Pro-Treatment Radiation Monitor Low Sample Flow detector. Both of these i alarms are " normal" for the shutdown condition, and were chattering because the process signals were slowly passing through the alarm setpoints. Authorization to bypass these alarms had just been completed j at the time of the event, but the associated bypass jumpers had not been j i installed.

At about 0338, the Hathaway Sequential Events Recorder (SER) alarm i printer was turned off because the typer was filling up with printed

messages caused by nuisance alarms from the Off-Gas Pretreatment Sample

Flow High-flow, feedwater Pump Seal Injection Pressure Low, and 28  ;

Reactor Recirculation Pump Seal Leakage High alarms. This had no effect j on the operation of the visual or audible alarms in the Control Room. i However, the Hathaway typer is located just behind the operator when he l is standing at the turbine controls, as for this test procedure. Due to its proximity and continuous typing, the typer provided a more  !

distracting condition than for work in other parts of the control room. [

Because of its operation throughout the evening, the operator did not believe that the typer would provide any useful information, and was  :

distracting him from the performance of the test procedure with l managements concurrence, the typer was shut off, j Procedure changes have been made to ensure that data logging equipment. l especially the Hathaway typer are operable whenever possible.  !

Alternatives to shutting off the typer have been provided to the l operators. These include consideration of stopping significant ,

evolutions untti distractions are eliminated without loss of recording  !

equipment. Controls have been established to require Shift Supervisor ,

approval prior to turning off the typer, i i

P ZCADTS Page 28 -

l i

0253T:29  :

i

m.

Item 8 (Turbine first stage prOO0ero r c3rder icking). Th] rOctrd?r traces were reviewed and seen to indicate that the turbine first stage l pressure recorder stopped inking when the turbine was taken off-line approximately 1.25 hours2.893519e-4 days <br />0.00694 hours <br />4.133598e-5 weeks <br />9.5125e-6 months <br /> prior to the trip. Prior to that time, no )

anomalous indications are apparent on the recorder. The operating ,

procedures specify a daily check of recorder paper and ink. This is )

normally done close to midnight, and had been completed as required i during the midnight shift. The t-urbine steam flow recorder is not an important parameter after the turbine has been taken off-line. LaSalle operators have a good awareness of instrument indications and maintain high instrument availability for control room instruments.

P e

t t

i l

I ZCADTS Page 29

\ Commonweteth Edison

. l

.=== 1 v_ / 7:ww 6 E7iiE Asem. nest av  !

t- cnceso mnaseosso.0767 l 1

August 30, 1989 l t

i t

i Mr. A. Dort Davis i Regional Administrator U.S. Nuclear Regulatory Commission l t

Region !!! i 799 Roo3evelt Road j Glen Ellyn. IL 60137 l

Subject:

LaSalle County Station Unit 2 l August 26, 1989 Unit Scram Mac w kat No.10-174  !

I

Dear Mr. Davis:

This letter identifies the actions Commonwealth Edison is undertaking i to address the event analysis and corrective actions for the subject Unit 2 Scram and the concerns identified with the response of the reactor protection ,

system. These actions will provide Commonwealth Edison and the NRC with  !

assurance that the same problem is unlikely to occur again. Following is a i summary list of our actions:

1. Determine the cause of the reactor protection system (RPS) observed [

response.

2. Conduct an event analysis to show that.the cause determined in )

Item 1 above is the only likely cause that describes the observed I RPS response, j

3. Perform a laboratory inspection of the K14E contactor. l t

4. Provide the following information for the RPS K14 contactor:  ;

i

a. The expected service life, and ,

i

b. The recommended preventive maintenance /inspectior: requirements. j Identify the actions taken prior to startup consistent with these i l requirements. Further, indicate the maintenance program in place  :

prior to the event for these contactors. l i  !

1

02697

1 l l

1..

August 30, 1989

. A.S. Davis I

Determine the cause of the two of the four first stage pressure sensors' {

! 5.

i settings being above the Technical Specification limit and identify corrective actions. l i

6. Determine the cause of the lndication problems of the turbine stop l

valves and identify corrective actions.

Determine the cause of the Hathaway computer inability to collect event j l 7.

i data and actions to be taken to maximize Nathaway availability.

1

+

l 8. Determine the conditions that resulted in the turbine first stage pressure chart recorder being without ink and the programs which are l in place to assure the availability of this and other control room 1 l chart recorder. ,

9. Submit a formal report to NRC Region !!! by October 1,1989, detailing the '

findings and conclusions.

The startup of Unit 2 will not occur without the concurrence of the  :

Regional Administrator or his designee. He expect to have responses to items I through 8 above for discussion with your staff prior to the startup. ,

' If there are any further questions regarding this matter, please ,

contact this office.

Very truly yours,

t. & %

O. P. Galle 00 0)

Vice President BWR Operations psk/Im cc: Region !!! Inspector - LSCS P.C. Shenanski - Project Manager, NRR s

31 0269T:2

i

(

l CORRECIlyLACIl0NS l Corrective actions have been taken to correct specific equipment  !

operational problems. These have been done under LaSalle Work Requests,  ;

and will be reviewed in accordance with a special On-Site Review for l plant startup.

The RPS specific corrective actions are as follows: l

1. Scram contactors K14E and K14G were replaced. *

2. The turbine valve leak tightness testing will be repeated on Unit 2 startup to verify that the event is not repeated as a result of '

testing. ,

3. Recording instrumentation will be installed to monitor the RPS {

channels during the subsequent startup of Unit 2, until after performance of the Turbine valve leak tightness testing. The configuration for this monitoring will be determintd by Toch Staff '

with assistance from General Electric.

4. Procedure changes have been implemented to prevent future i i circumstances where significant plant evolutions are in progress i without the Hathaway event recorder being operational.

5. Instructions for responding to partial RPS actuations will be i

provided to all licensed operators, j t

l I

e i

t ZCADTS Page 32 l

i 0253T:31  ;

l L

i i

r -

e V1110 CottCWE10 tis The specific cause of the RPS actuation has not been possible to determine. The only plausible hypothesis consistent with the circumstances as known is that the actuation was spurious, and that the RPS actuation was an unnecessary shutdown and resulted in placing the plant in a safe configuration.

All available evidence indicates that no RPS equipment malfunctioned during the event. The evaluations and test data support the conclusion that the RPS is functioning properly and will continue to function properly after Unit 2 restart.

Some additional soonitoring will be conducted in order to collect data to either verify or modify the conclusions in this report. This monitoring is not necessary to assure safe operation, but will be used to ensure that this event has been correctly interpreted. Future actuations of this nature are not anticipated, however, and if routine operation demonstrates that the event will not be repeated, the monitoring equipment will be removed.

Investigative efforts include those described above and additional investigstions by Ceco off-site Safety Assessment group and additional technical reviews by General Electric. These reviews have not resulted in any information which would affect the conclusions reached above regarding the root cause evaluation or the safety consequences of this event.

2CADTS Page 33

i 1

i i

l l

APPENDIX A l t

GENERAL ELECTRIC REPORT ;

LASALLE UNIT 2 SPURIOUS SCRAN l AUGUST 26, 1989 1

i t

t f

r i

e t

4 L

i I

i i

i i

i 1

i ECADTS Page 34 i

!