ML17139C679

Testing

was

expanded to include all

185 rods.

Final results

were

4 rods

failed to scram

and

9 rods hesitated

before

scramming.

10/07

The four failures were determined to be caused

by the scram

pilot solenoid

valve

(SPSV).

These

four valves

were

replaced.

10/07

Licensee

formed a task force to investigate

SPSV failures.

One

valve was sent to

GE and one to Franklin Institute for analysis.

10/12

GE informed PPKL that the SPSV fai lures were due to the disc

holder subassembly

adhering to the valve seat.

The disc material

was

polyurethane.

GE

recommended

that the polyurethane

be re-

placed with Viton-A.

10/12

Unit 1 commenced

a reactor

shutdown at 9: 10 p.m.

since all other

SPSVs

also

had polyurethane

discs.

Unit 2

commenced

a reactor

shutdown at 9:50

p.m.

Ninety-three of the

185 Unit 2 SPSVs

had

polyurethane

disc material.

ENS calls were properly made

on the

above.

10/13

Unit 1 in Hot Shutdown at 4:00 a.m., Unit 2 in Hot Shutdown at

5:12 a.m.

10/14-15

New disc holder subassemblies

with the Viton-A material were

installed in all

185

SPSVs

on Unit

1

~

10/16

Meeting in Bethesda with PP&L, GE, Region I and

NRR to discuss

SPSY problem.

10/17

Unit 1 reactor critical at 12:34 a.m.

Unit 2 reactor critical

at

9:45

p.m.

following replacement

(and/or

inspection)

of all

SPSV disc holder

subassemblies.

NRC issued

Confirmatory Action

Letter (CAL).

10/18

Unit 1 conducting individual rod scram testing of all rods.

The licensee

discovered

that the

18 month surveillance

interval

on the

scram discharge

volume

(SDV) vent

and drain valve timing

was past

due.

At 9:21 p.m.,

Unit

1

was manually

scrammed

from

55 percent

power to perform the surveillance.

The

SDV vent valve

closed in 32.4 seconds

exceeding

the Technical limit of 30 seconds.

10/19-20

The pilot solenoid valve for the Unit 1 SDV vent valve'was

replaced

with

a pilot solenoid

valve

manufactured

by Valcor.

Unit 2 conducting

rod testing.

0

a

d

1

10/21

Unit 1 returned to criticality at 3:52 a.m.

and was scrammed at

1:29

p.m.

to 'meet

the

SDV vent

and drain

valve

surveillance

requirement following the pilot valve replacement.

10/22

Unit returned to criticality at 12:35 a.m.

and increased

power

to continue

rod testing.

2.0

Descri tion of Scram Pilot Solenoid Valves

The

Scram Pilot Solenoid Valves

(SPSV) are

a three-way,

solenoid-operated,

normally energized

valve.

During normal

reactor

operations,

each of the

two channels

of the

Reactor

Protection

System

(RPS)

energizes

one of the

pilot valve solenoids.

When energized,

the pilot valve allows instrument

air to be supplied to the diaphragm actuators

of the inlet and outlet scram

valves,

holding these

valves shut.

Upon initiation of a scram,

both chan-

nels

of the

RPS

de-energize

and

both pilot valve solenoids

de-energize,

rapidly venting air pressure

from the

scram inlet and outlet valves,

and

allowing them to open to scram the control rod.

To prevent

inadvertent

of a single rod, both pilot valve solenoids

must

be de-energized

before air is vented

from the

scram valves.

De-ener-

gizing only

one of the

two solenoids will not result

in

a

because

the remaining coi 1 is sufficient to maintain

the valve in

a configuration

that allows air to continue to be supplied to the

scram valves through the

pilot valve.

There are

185 scram pilot solenoid

valves installed

on

each

unit (1 per control rod.)

For added reliability, the Control

Rod Drive (CRD) Instrument Air System

has

two

DC solenoid operated,

three-way air valves installed

on the supply

header

called

backup

scram valves.

The air supplied to the Hydraulic Con-

trol Units

(HCUs)

and

the

discharge

volume

vent

and

drain

valves

passes

through

these

two valves.

As

a

backup to the individual

SPSV and

scram discharge

volume vent and 'drain valves,

the

backup

valves

are

energized

by the

Reactor

Protection

System

(RPS)

on

a

and vent the

entire

CRD Instrument Air System.

Units

1

and

2 utilize an Automatic Switch

Company

(ASCO) three-way, dia-

phragm type,

redundant piloted solenoid valve

(ASCO Part

No. HV-176-816-1,

GE Part

No. 9220138) for the

SPSV.

The valves

are of brass

construction.

Figure

1

shows internal

components

of the valve.

When the valve is ener-

gized (both solenoids),

the inlet port from the external

air

supply (in-

strument air) is open to the

scram inlet and outlet valves,

and the exhaust

port is closed.

The disc holder

subassembly

contains

a polyurethane

disc

which covers

the valve exhaust

port

on the 'B'olenoid side.

When both

solenoids

de-energize,

the disc

holder

subassembly

moves

away

from the

exhaust port and the core assembly of the

A solenoid

closes

the air inlet

port, this action allows air to bleed off the

scram inlet and outlet valves

thus causing control rod insertion.

/

0

3.0

Licensee Actions

3.1

October

6

1984 Surveillance Test

and Followu

Actions

On

October

6,

1984,

the

licensee

commenced

quarterly individual rod

scram testing

on ten percent of the Unit 1 rods

as

required

by Tech-

nical Specification (T.S.)

4. 1.3.2.

At approximately

9:05 a.m.

con-

trol rod 42-23 failed to insert after 'the test

switches

were placed

in the test

position.

The test

was

repeated

three

times

and

each

time the rod failed to scram.

Instrument

and Controls (I&C) technicians

investigated

the problem

and

when they tapped

the solenoid valve the

rod

scrammed.

The

rod

was

then full withdrawn

and

retested.

The

retest

at

9:40

a.m.

was

satisfactory.

When

rod testing

continued,

rod 42-39 also initially failed to insert at

10:30

a.m.

and when

an

operator

tapped

the

solenoid

valve,

the

rod

inserted

as

designed,

The licensee

then

decided

to individually scram test all

185 Unit

1

rods.

Two additional

rods (i.e.

58-31

and

38-39) failed to

during this testing

and

9 other

rods hesitated initially when tested.

Rods 38-39

and 58-31 subsequently

scrammed

when operators

tapped their

solenoid

valves

and

then

passed

a retest

when

withdrawn

again.

Of

the rods that hesitated,

none exceeded

the required

maximum insertion

time of seven

seconds.

During the testing the appropriate

T.S. Limiting Condition for Opera-

tion

(LCO) was properly entered

when individual rods were determined

to be inoperable.

The

LCO',s were cleared following successful

retests.

On October 7, the

scram pilot solenoid valves

(SPSV) were replaced

on

the four rods which did not

The licensee

set

up

a task force

consisting of plant staff and Nuclear Plant Engineering

(NPE) engineers

to investigate

the

SPSV failures.

One of the replaced

SPSVs

was

sent

to GE,

San Jose,

California for analysis of the failure mechanism

and

another

was sent to Franklin Institute.

3.2

GE Anal sis Results

and Plant Shutdowns

On October

12,

1984,

GE informed PP5L that it was GE's judgement that

the

SPSV failed due to the disc holder

subassembly

disc sticking to

the seat

on the valve body.

The sticking was

due to

a degradation

of

the polyurethane

disc material.

GE,

as

a product upgrade

changed

the

disc material

to Viton-A.

The Viton-A material

was

subsequently

in-

corporated

into

ASCO spare

part kits for these

valves,

beginning

in

1982, although

an immediate

replacement for existing valves apparently

was not

deemed

necessary.

Viton-A was

an

environmentally qualified

material.

The polyurethane

disc

was designed

for temperatures

up to

220'.

However,

based

on

the

service

temperatures

of the

valves,

the

polyurethane

material

apparently

degrades

at

temperatures

near

160 F.

Licensee testing

found that the skin temperature

of the

SPSV

when energized

was approximately

140~

F.

Based

on this information

and the determination that most if not all

of the

SPSVs

on

both units

were affected,

the

licensee

decided

to

shutdown

both Units

1

and

2.

Unit

1 shutdown

was

commenced at about

9: 10 p.m.

on October

12 and the unit was in Hot Shutdown at 4:00 a.m.

October

13.

Unit

2

shutdown

was

commenced

at

about

9:50

p.m.

on

October

12

and

Hot

Shutdown

was

achieved

at

5: 17

a.m.

October

13.

Unit 2 subsequently

went to Cold Shutdown at about 5:00 p.m. to con-

duct unrelated

maintenance.

3.3

Com onent

Re lacement

ASCO spare part kits containing the Viton-A disc were obtained

by the

licensee

and

the disc

holder

subassemblies

were extracted

from the

spare part kits and installed in all

185

SPSVs

on Unit

1 during October

13 thru October

15.

These

valves were functionally tested

by observ-

ing the

scram inlet

and outlet valves

stroking

when the individual

rod test

switches

were actuated.

Additionally, individual rod testing

was

performed while shutdown.

The

backup

scram valves

and the

discharge

volume

(SDV) vent

and drain pilot valves

on Unit

1

were

also rebuilt with

new Viton-A discs.

Subsequently,

on

October

17,

Unit

1 returned

to

power

and individual rod testing

on all rods

was

conducted at approximately

50 percent

power.

On

Unit 2,

the

licensee

determined

that

93 of the

SPSVs

had

been

previously rebuilt in April 1983 using

ASCO spare part kits containing

the Viton-A disc.

The licensee installed the

new disc subassembly

in

the remaining

SPSVs

and also

inspected

the ones that were previously

rebuilt to

ensure

that

they

contained

the

Viton-A disc.

New disc

holder

subassemblies

were also installed in the backup

scram valves.

The Unit 2 scram discharge

volume vent

and drain pilot valves,

manu-

factured

by Valcor were not affected.

Inspector

review of these activities identified no unacceptable

con-

ditions.

4.0

NRC Res

onse

Re ion I and

Head uarters

NRC Headquarters

was notified, via

ENS call

on October

12, of the defective

SPSVs

and of the licensee's

decision to shutdown Units

1 and 2.

On October

15, at the request of Region. I, the licensee

committed to remain

below

5%

power pending

the results

of

a meeting

in Bethesda,

Haryland

on the fol-

lowing day to discuss

the

SPSV problem.

At the meeting,

the licensee

com-

mitted to the following actions:

scram-time

test all

185

rods,

on

each

unit,

when

a

50 -

60% power

level is reached

develop

a surveillance

procedure

to unambiguously

assess

scram pilot

valve operability,

to

be

submitted

to and

approved

by

NRC prior to

implementation,

and performed every

4 to 6 weeks

trend

and report immediately to NRC, via the

ENS network,

any failures

or anomalies

found during scram

solenoid

valve operability tests,

or

individual control

rod

time testing

(normally performed for

a

10% rod sample every

4 months)

provide

the failure analysis

results

from Franklin

Research

Center

and

General

Electric

testing

on

the

original

valves

.which failed

On October

17,

Region I issued

a Confirmatory Action Letter confirming the

above

commitments.

These actions will be reviewed in subsequent

inspections.

(387/84-35-01)

5.0

NRC Followu

Review

Resident

Onsite

Inspector

review of the

SPSV failures

focused

on the following aspects:

(1) monitoring licensee

actions to replace

the

SPSV disc holder

subassem-

blies,

(2) determining

the

reportabi lity aspects

of the

October

6

SPSV

failures, (3) review of maintenance

history of SPSVs

(4) review of scram

timing history,

and (5) post-maintenance

testing.

Item (1) was discussed

in section 3.0.

The remaining

items are discussed

below.

Following discovery of the

SPSV failures

on October

6, the licensee

prepared

Significant

Operating

Occurrence

Report

(SOOR)

1-84-370.

The

SOOR did not indicate that this occurrence

was reportable

by

10 CFR 50.72

and

no

ENS call was made.

The inspector

reviewed the 50.72

reportability criteria and di scussed

reportabi lity with the licensee,

Based

on

the inspector's

review,

no report of this

occurrence

was

required

per 50.72.

10 CFR 50.72

reports

were

made

on

October

12,

informing the

NRC of initiation of plant shutdown.

5.2

Maintenance

Histor

The

inspector

reviewed

maintenance

records

of

SPSV

maintenance

on

Units

1

and

2 to determine if previous

indications of

SPSV sticking

existed.

In August

1981, prior to licensing,

new

SPSVs

were installed in all

Unit

1

HCUs because

of a powdered granular

substance

(apparently rust)

found in four of the valves.

In June

and July 1982, just prior to

Unit

1 licensing,

new SPSVs were installed

on half of the Unit

1

HCUs

and the other half were obtained

from Unit 2, rebuilt and installed

on Unit

1.

This action

was

taken

because

of

a concern with Buna-N

parts.

GE Service Information Letter (SIL) 128 recommended

rebuilding

these

valves periodically to ensure

Buna-N parts are not used in excess

of seven

years.

The rebuild kits did not contain

the

Viton-A disc

material.

Other

maintenance

on

SPSVs

since that time

has

consisted

of replacement

of SPSVs

on several

HCUs due to air leaks

and

burned

up solenoids.

No maintenance

due to sticking

SPSVs

was

noted

in the

document review on Unit 1.

I

1f

r

With respect

to Unit 2,

as noted in Section 3.3, only one half of the

SPSVs

had polyurethane

discs.

In April 1983,

18C personnel

had re-

built about

one half of the Unit 2

SPSVs for replacement

of

Buna-N

parts.

The

spare

part kits

used

for these

valves

contained

discs

made of Viton-A material.

The remainder

of the Unit

2 valves

were

replaced

with new

SPSVs

which apparently did not contain the Viton-A

disc.

These actions

were taken

because

of recommendations

in SIL No.

128.

Other

maintenance

involved replacement

of SPSVs

on several

HCUs

due

to air leaks.

In April 1984,

during

rod testing

conducted

during

initial fuel loading,

two rods (i.e., 38-03

and 58-23) failed to in-

sert

when the test switches

were replaced

in test.

The

SPSVs for these

HCUs were replaced.

The inspector

examined

the

removed valves.

These

valves

had

small

crimps

in

one

solenoid

shaft

which

may

have

been

caused

by

someone

stepping

on

the

valves

during construction.

The

polyurethane disc in the valves

was not degraded.

Hence, this valve

problem is not considered

a precursor

to the October

6

SPSV problem.

Another rod (rod 38-15)

had

a

slow scram

time during

the April rod

testing.

Valve stroking of scram valves

on this

HCU was checked with

no problems

found.

5.3

Time Histor

5.3. 1

Surveillance Testin

The inspector

reviewed

the licensee's

surveillance

testing

program utilized to measure

the

maximum

and

average

insertion times of each control rod in accordance

with Tech-

nical Specifications

4. 1.3.2,

4. 1.3.3,

and

4. 1.3.4,

which

delineate

the

timing

requirements

for the

control

rods.

Surveillance

Requirement

4. 1.3.2- states

that

the

insertion

time of the control

rods

shall

be demonstrated

through measurement,

with reactor coolant

pressure

greater

than

or

equal

to

950 psig for: (1) all control

rods prior

to exceeding

40 percent

thermal

power after reactor

shutdowns

greater

than

120 days;

(2) specifically affected individual

control rods following maintenance

or modification; and (3)

at least

10% of the control 'rods,

on

a rotating basis,

at

least

once

per

120

days

of operation.

The

maximum

insertion time is measured

from the de-energization

of the

scram pilot valve solenoids

as time zero, with the rod fully

withdrawn.

The following three

surveillance

procedures

were

reviewed

to ascertain

whether the procedure

was adequate

to meet the

Technical Specification requirements:

SR-155-001,

Revision 0, Scram Time Measurement of All

Operable

Control

Rods

SR-155-002,

Revision 0,

Scram Time Measurement

of Rods

Following Maintenance

or Modification

SR-155-003,

Revision 0,

Scram Time Measurement

of Rods

Every 120 Days

Due to Susquehanna's

unique configuration,

the

scram times

can

be obtained

by two different methods.

One method util-

izes

a recorder

plugged into the Control

Rod Test Instrument

Panel

in the control

room.

The panel

senses

when the

pilot solenoids

are de-energized,

and records

the

odd (i.e.,

45,

39, etc.) notch positions of the rod as it inserts into

the core.

The insertion

times

are calculated

by analyzing

the

strip

charts

from

the

recorder.

The

second

method

utilizes the

General

Electric Transient Analysis Recording

System

(GETARS) computer,

which monitors

sensor

inputs

and

stores

the applicable data,

even during an unscheduled full

reactor

This

method

must

be corrected

due

to the

GETARS

scan

rate,

and

because

of the time response

of the

relays that the

computer

senses.

This time correction

is

properly reflected in the procedure.

The inspector

reviewed

the

completed

documentation

for the

last

three

Unit

1 -surveillance

tests

which

were

used

to

meet

the

Technical

Specification

requirements

performed

prior to October 6, 1984.

On

June

13,

1984,

Unit

1

scrammed

from full power

due to

the loss of startup transformer T-10.

During the unscheduled

trip,

GETARS obtained

time data for the

147 control

rods that were fully withdrawn.

On June

25, this data

was

utilized to

meet

Technical

Specification

surveillance

re-

quirements

4. 1.3.2,

4. 1.3.3,

and 4. 1.3.4.

The surveillance

was

signed off as

being

complete

and

meeting all of the

acceptance

criteria

on July 26 by the individual performing

the surveillance

and his supervisor.

NRC review of the completed surveillance

data

on October

18

determined that

one of the

acceptance

criteria of the sur-

veillance

was

not

satisfied.

Technical

Specification

3. 1.3.4 establishes

maximum

average

insertion

times

from the fully withdrawn position

to four specific

notch

positions for the three fastest

control

rods

in each

group

of four control rods arranged

in a two-by-two array.

During

'he

scram of June

13, the four rod array containing control rods 38-39,

38-43,

42-39

and

42-43

exceeded

the

allowable

average

scram insertion time from the fully w'ithdrawn posi-

tion (notch

48) to notch

45.

The

Technical .Specification

requires

the

average

insertion

time

not

to

exceed

0.45

seconds,

but the average

scram insertion

time of the three

fastest

rods in this array was 0.462

seconds.

The following

table

shows the applicable data;

Position Inserted

from Fully With-

dr awn

45

39

25

05

TS Limit on

Average

Insertion

Time

(Sec)

0.45

0.92

2.05

3.70

Average

Insertion

Time

During June

13

for

4

Rod

Array (Sec)

0.462

0.795

1.561

2.641

These data indicated

a hesitation of the rods at the initia-

tion of the

but the

rods fully inserted within the

required

time.

The

two slowest

rods

(38-39

and

42-39) of

the

four rod array later failed to insert during the

rod

testing

on

October

6,

1984.

The

anomaly

in

the

June

13

times

appears

to

be

a

precursor

to

the

October

6 event.

The calculations

which determined

the

scram insertion data

were

performed

by

computer.

One

page

of

the

computer

printout specifically indicated that this rod array exceeded

the required

average

scram insertion

times.

The four rods

and the average

time were listed under

a page

heading which

stated:

"Four

Rod Array with Average

Three

Fastest

Rods

Exceeding

T.S.

3. 1.3.4 Limit".

On the test data

sheet for

the surveillance,

the individual who performed the surveil-

lance,

was required to indicate (by circling YES and initi-

aling)

that

the

acceptance

criteria

had

been

satisfied.

The fact that the four rod array did not meet the Technical

Specification

was not noted

by either the

individual per-

forming the surveillance

nor

the applicable

supervisor

who

reviewed

the

completed

surveillance.

In discussions

held

with

the

individual

who

performed

the

surveillance,

he

indicated that

he

had assumed

that since the rods

had satis-

factorilyy

met the

maximum and average

scram insertion times

that the two-by-two array

numbers

would also

be satisfied.

This surveillance test

was

the first time the individual,

and the work group (i.e.

Reactor

Engineering)

had performed

the test.

Previously it had

been

performed

by the plant

engineering staff.

The Technical Specifications

required that the control

rods

with the

slower

than

average

insertion

times

be

declared

inoperable

unti 1

an

analysis

was

performed

to

determine

that the required

scram reactivity

remained

for

the

slow

four

control

rod

group

and

to

increase

the

surveillance

frequency

to

at

least

once

per

60

days.

Otherwise,

the plant

was

to

be

in at least

Hot

Shutdown

within the next

12 hours1.388889e-4 days 0.00333 hours 1.984127e-5 weeks 4.566e-6 months .

K

10

During the

scram of June

13,

1984, five rods were signifi-

cantly

slower

in reaching

the

notch

45 position

than

the

average

of all of the rods.

The

average

time for all of

the

rods to reach

notch

45 was

301 milliseconds,

but these

five rods

exceeded

500 milliseconds.

Of these

five rods,

four of

them

were

the

same

four that failed to

and

one

was

one of the nine that hesitated

during the surveil-

lance testing

on October 6.

The plant restarted

on June

14, but the

scram timing data

was not analyzed

until

June

25

when

the

surveillance

was

performed.

Therefore,

the

inoperability

of

the

four rod

array was not identified prior to the startup.

A thorough

review of the

time data

is normally not

performed

after each

nor is it required.

On June

25

when the

surveillance

was completed,

the appropriate Action Statement

was not entered

since the inoperability was not identified.

The plant

operated

continuously until another

unscheduled

occurred

on July

3,

1984;

Inspector

review of rod

insertion

data for that

indicated that all rod scram

insertion times met the Technical Specification

requirements'ata

was also

obtained

on

a July

15

which indicated

that

all

Technical

Specification

requirements

were

met.

The data

from the July

15

was also

used

to meet

the

surveillance

requirement.

With the capabilities of the

GETARS, all of the fully with-

drawn

rods

are

timed

on

each

and far

more

data

is

available

than

required

by Technical Specifications'or

example,

on the June

13,

1984 scram,

data for 147

rods

was

input into

the

computer

for analysis,

but

the

Technical

Specifications

only require

19 rods to

be tested

every

120

days.

If the

normal quarterly surveillance

had

been

per-

formed instead

of using all of the

scram data, it is very

probable

that

the

Technical

Specification

nonconformance

would not

have

been

detected.

During the October

6,

1984

surveillance,

the

problem again

may not have

been detected

if two of the four failed rods

had not

been

in the

20 rod

sample

selected.

It should

be

noted

that

the

data

reviewed indicates

that

the pilot valve hesitation

and sticking phenomenon

occurred

during

the June

13

after

an

82 day continuous

run,

and during the October 6,

1984 testing, after

a 78 day con-

,

t ~

tinuous

run.

The

mechanism

appears

to

be

time

dependent

and

primarily

occurs

after

a

long

period

of operation.

5.3.2

Scram Data Review

The inspector

reviewed the

GETARS scram insertion time data

for eleven

on Unit

1 which. occurred in the

18 months.

prior to the October 6,

1984 surveillance testing.

Several

of the

were utilized to

meet the surveillance

re-

quirements

as

noted

in section

5'. 1.

The

data

not

utilized for the surveillance

was

reviewed to determine if

any evidence of hesitation

occurred during previous

scrams.

Additionally, the computer generated

control rod scram time

history file for each

individual control

rod was

reviewed.

The

average

time of all the control rods to reach

notch

45

from the fully withdrawn position

was approximately

260 to

270 milliseconds

throughout

the

18 month

t'ime period.

The

average

scram insertion time required

by Technical Specifi-

cations for the

same position is 430 milliseconds.

Several

occurrences

of rod insertion times greater

than

430

milliseconds

were noted during the data review:

DATE

Narch 22,

1983

October 31,

1983

June

13,

1984

July 3,

1984

July 15,

1984

ROD 58-31

42-27

42-23

54-47

58-31

42-39

38-39

58-31

54-47

TINE NS

626

432

584

752

975

891

668

507

556

Although the

above

rods

apparently

hesitated,

all of the

rods

met

the

Technical

Specification

maximum

and

average

insertion

times

(except

for the

four rod

array

on

June

13 noted in section 5.3. 1).

Of the six rods that hes-

0

'L

12

itated,

four of them failed to

and

two hesitated

on

October 6,

1984.

5.4

Post Maintenance

Test Witnessin

Between

October

18

and

24,

1984,

the

licensee

conducted

individual

rod scram testing

on both Units

1 and

2 from power levels between

50%

and

60%.

The testing

was

conducted

in accordance

with Surveillance

Procedures

SR-155-002

and SR255-002,

"Scram

Time Measurement

of Rods

Following Maintenance

or Modification", Revision

0.

The inspectors

witnessed

portions of the testing.

No

unacceptable

conditions

were

noted.

The

inspectors will review the test results

when available.

6.0

Scram Dischar

e Volume Vent and Drain Pilot Valve Ino erabilit

At 2: 15 p.m.

on October

18,

1984,

the licensee

discovered that surveillance

procedure

S0-155-003,

SDV Vent and Drain Valve Eighteen

Month Operability,

was

overdue

by approximately

15 months.

A documentation

review identified

that the surveillance,

which is required

by Technical Specification

4. 1.3. 1.4

every

18 months,

was last performed

on January

23,

1982.

At some

unknown

time during

1983, it was identified that the original surveillance

docu-

mentation

was misplaced,

and

on June

2,

1983,

a

new surveillance

author-

ization cover sheet,

with the associated

test data attached

was generated

and

signed.

Due to

an administrative

error,

the

date

the

form was

re-

produced (i.e.

June

2,

1983)

was entered

into the surveillance

tracking

system

as

the completion

date of the surveillance.

Therefore,

according

to the tracking

system,

the

surveillance

was

not

due until

December

2,

1984 instead of July 23,

1984.

Technical

Specification

Surveillance

Requirement '4. 1.3. 1.4.a

states

that

the

discharge

volume shall

be determined

operable

by verifying that

the drain

and vent valves close within 30 seconds after receipt of a signal

for control

rods to

and

open

when

the

signal

is reset,

at

least

once .per

18 months

from

a

normal

control

rod configuration of less

than or equal

to

50%

rod density.

With the

SDV

system

inoperable,

the

Action statement

requires

the plant to be in at least Hot Shutdown within

12 hours1.388889e-4 days 0.00333 hours 1.984127e-5 weeks 4.566e-6 months .

The licensee

immediately declared

the

SDV system inoperable

when the over-

due surveillance

was identified

and

made the appropriate

10 CFR 50.72

ENS

notification.

The licensee

conducted

a data

search

to determine if docu-

mentation

was available

from previous reactor .scrams

to show that the vent

and drain val,ves

had operated

properly,

but

none

was

found.

At 9:21

p.m.

October

18,

1984,

the plant

was manually

scrammed

from 55% power in order

to conduct the surveillance test.

During the

the

SDV vent valve closed in 32.4 seconds

and the drain

valve closed in 26:9 seconds.

Since the vent valve did not meet the accep-

tance criteria of 30 seconds,

the

SDV remained

inoperable

and investigation

commenced

on

the

cause

for the

slow stroke

time of the vent valve.

NRC

Region I requested

the licensee

to inform them of the corrective actions

yA

P

a'

4

k

~

13

completed to resolved

the inoperability,

and to demonstrate

that the valves

operated

in less

than

30 seconds prior to the Unit 1 startup.

Two redundant

vent valves

(XV-1F010A and XV-1F010B) are installed

on the

1

inch

SDV vent line and each is operated

by

a

separate

dual

solenoid pilot

valve

(SV-1F009A and

SV-1F009B).

The pilot valves also operate their as-

sociated

drain valve.

Currently

one vent

and

one drain valve

(XV-1F010B

and

(XV-1F011B) are mechanically

locked

open,

based

on

a commitment

made

in licensee letter PLA-1038, dated April 7,

1981,

which discussed

IE Bul-

letin 80-17.

The valves are blocked open pending completion of the instal-

lation of the

redundant

vent

and drain valve

systems

required

by the end

of the first refueling outage

by license

condition 2.C(17).

The redundant

valves

are

to

ensure

that

an uncontrolled

loss of reactor

coolant

would

not result

in the

event of

a single active fai lure.

The

SDV vent

and

drain pilot

solenoid

valves

are

ASCO,

dual-solenoid

valves

(Part

No.

HT8323A22),

and the vent valves are

Hammeldahl

Conoflow 500 Series

1 inch

globe valves.

On October

18 the licensee

issued

Work Authorization S44954 to investigate

the

cause

for

vent

valve

XV-1F010A

slow

stroking.

The

maintenance

technicians

initially lubricated

the

stem

and

checked

the

packing,

and

the

valve

then

operated

in

31.5

seconds.

air

actuator

was

disassembled,

with

no deficiencies

identified.

To retest

the vent valve

a temporary rig consisting of eight feet of 1/4 inch plastic

tubing

was

used

to operate

the valve.

The vent valve operated

in eight seconds.

The

normal

supply line was reinstalled,

and the valve operated

in

32

seconds.

Based

on this information the

problem was

assumed

to be with the time re-

quired for the pilot valve to vent the approximately

45 feet of one

inch

air supply header.

A different

model

of pilot solenoid

valve is installed

in Unit 2,

and

preoperational

test data indicated that these valves,

which are

much larger

than the

ASCO valve, vented the air header significantly faster.

The Unit

2

preoperational

test,

P255. 1A,

performed

on

September

24,

1983,

found

that the vent valve closed in 7.46

seconds

from when the

signal

was

initiated.

The associated

drain valve operated

in 14.7 seconds.

Initially, it was

conceived

that this

problem

could

be

scram pilot solenoid valve disc material

phenomenon,

since the pilot valve

was also

an

ASCO pilot solenoid of similar design to the

SPSVs.

Based

on

the test

data

showing

the valve consistently

operated

in

32 seconds,

and

the fact that the polyurethane

disc subassemblies

originally installed in

the valve were replaced with Viton-A, the reason for the failures does not

appear to be related to the

SPSV failures.

In order to correct

the venting problem,

the licensee

replaced

the

ASCO

pilot solenoid valve with a Valcor 3-way solenoid valve (Part

No. V70900-45)

on October 20,

1984.

Shutdown testing of the valve following the instal-

lation demonstrated

that the vent valve would close

in approximately

six

seconds after pilot de-energization.

In addition to replacing

the pilot solenoid valve, the licensee

conducted

a

100 percent

documentation

review to ensure that

no other similar admini-

strative errors

were

present

in their surveillance

tracking

system.

No

other deficiencies

were found.

The licensee

also modified the surveillance

documentation

forms to

emphasize

the

date

on which the surveillance

was

performed rather than the date of the form and created

a full time survei 1-

lance documentation

auditor position.

These actions

are intended to reduce

the potential for incorrect data entries

in the surveillance tracking com-

puter system.

The licensee

required

an emergency

Technical Specification

change prior to

startup,

which

was

issued

on

October

19,

because

the

Unit

1

Technical

Specifications

did not allow the unit to

be started

up with the

SDV vent

and drain valves inoperable,

but the surveillance

test required

the plant

to

be at

less

than

50 percent

rod density.

The Technical

Specification

change

allowed the unit to enter Operational

Condition

2 provided the sur-

veillance

was

performed within

12

hours

after

achieving

50

percent

rod

density.

Based

on

a review of the test results

and discussions

with Region I, the

licensee

restarted

Unit

1 and reached criticality at 3:52 a.m.

October 21,

1984.

At 1:29

p.m.

on

October

21,

the unit

was

manually

scrammed

from

approximately

7 percent

power to perform the Technical Specification sur-

veillancee

test.

The vent valve closure time for the test

was 5.2

seconds.

Unit

1 returned to criticality at 12:35 a.m.

on October

22.

7.0

Technical

S ecification Adherence

As noted in section 5.3. 1, the Technical Specifications

require that while

in Operational

Condition

1 or 2, the

average

insertion

time,

from

the fully withdrawn position for the three fastest

control

rods

in each

group of four control

rods

arranged

in

a two-by-two array,

based

on

de-

energization

of the

scram pilot valve solenoids

as

time zero,

shall

not

exceed

0.45

seconds

for notch position 45.

The applicable action statement

requires

that with the

average

insertion

times

of control

rods

exceeding

the insertion time limits, the control

rods with the slower

than

average

insertion

times

are

to

be

declared

inoperable

until

an

analysis is performed to determine

that required

scram reactivity remains

for

the

slow

four

control

rod

group.

Additionally,

the

surveillance

frequency

for

insertion

times

is

to

be

increased

to

60

days.

Otherwise

the plant is to

be in at least

Hot Shutdown within the next

12

hours.

Contrary to the above,

on June

25,

1984, Surveillance

Procedure

SR-155-003

was

performed

using data

from

a June

13

scram to meet the requirement of

Technical

Specification

3. 1.3.4.

The

data

indicated

that

the

four

rod

array containing control rods 38-39,

38-43,

42-39

and

42-43

exceeded

the

allowable average

scram insertion time to notch

45,

in that the insertion

time of the three fastest

rods

was 0.462

seconds.

Since the data

was not

properly reviewed,

and

the inoperability not identified,

the

applicable

Action requirements

were not completed.

This condition

remained until

a

4

4-

15

subsequent

reactor

on July 3,

1984.

The surveillance

was

not re-

performed until July 15,

1984.

As noted

in section

6.0,

Technical Specification Surveillance

Requirement

4. 1.3 . 1.4.a

states

that

the

discharge

volume

shall

be

determined

operable

by verifying that the drain and vent valves close within 30 seconds

after receipt of

a signal for control

rods to

scram at least

once per

18

months

from a normal control rod configuration of less

than or equal

to 50

percent

rod density.

Contrary

to

the

above,

at

2: 15

p.m.

on

October

18,

1984,

the

licensee

discovered that Surveillance

Procedure

S0-155-003,

SDV Vent and Drain Valve

Eighteen

Month Operability,

was last

performed

on

January

23,

1982,

and

was therefore

overdue

by approximately

15 months.

This violation was iden-

tified by the licensee

and promptly reported to the

NRC by an

ENS notifi-

cation.

Inspector review determined

that the corrective action

completed

and

planned

should prevent

recurrence.

Additionally, it was not a viola-

tion that

could

reasonably

be

expected

to

have

been

prevented

by

the

licensee's

corrective action for a previous violation.

Since

the criteria for licensee

identified violations

stated

in

10 CFR Part 2 Appendix

C

have

been satisfied,

a notice of violation will not be

issued,

but

the corrective

actions

implemented

by the 'licensee

will be

reviewed in a subsequent

inspection.

(387/84-35-03)

Safet

Si nificance

The

event discussed

in section

5.2

and 7.0 involved

one

2

x- 2 rod array

exceeding

the

average

insertion

time

from

the

fully withdrawn

position to notch 45.

The rods exceeded

this time by

10 milliseconds

and

they

met

the

insertion

times for the

remainder

of the

rod travel.

The inspector

discussed

the significance of this event with

NRR (members

of the

Reactor

Systems

and

Core

Performance

Branches)

and

reviewed T.S.

bases

and the

FSAR.

The purpose of the T.S.

scram insertion

times

are to

ensure that the control rods insert reactivity at

a rate within the bounds

of that assumed

in transient

and accident analysis.

The reactivity inser-

tion rates

must bring the

reactor

subcritical

at

a rate fast

enough

to

prevent

the

Minimum Control

Power

Ratio

(MCPR)

from becoming

less

than

1.06 during the limiting power transient

analyzed

in

FSAR chapter

15.

The

NRR reviewers

indicated that

exceeding

the

average

insertion

time

during the first few notches

of rod insertion

has

no effect on transient

and

accident

analysis

since

very little reactivity

is

inserted.

The

reactivity insertion rates

assumed

in analyses

are not affected unless

the

rods

are

slow in reaching

approximately

50 percent

rod insertion

to the

core.

t

1$

'1

P

1

p 7

16

The

rod

time data for the July

3

and July

15

indicate that

this four rod array met T.S. insertion times.

Based

on the time dependency

of the failure

mechanism

of the

SPSVs

(section

5.3. 1), it can

also

be

assumed

that

these

rods

would

have

met

the

insertion

time

requirements

immediately after the

June

13

Hence,

the

safety

significance

of

these

rods not inserting within the required time is minimal.

Nevertheless,

the

event

is significant

from the point of view that it

involved the improper review of surveillance

data

and the fact that two of

the affected

rods identified in this

occurrence

later failed to insert

during testing

on October 6,

1984.

Since

each control

rod drive mechanism

has its

own scram

and pilot valves,

only one drive

can

be affected if a

valve fails to open.

In other

words,

a single failure in

a hydraulic control unit would result

i n the

failure of only

one control

rod.

Due to the

presence

of the

inadequate

polyurethane

material in the

SPSV,

the potential

did exist,

however,

that

a

common

mode failure could have

caused

a significant number of rods to be

inoperable.

The mechanism that could have possibly identified the

problem

earlier,

the surveillance

procedure,

was not properly reviewed,

and there-

fore the precursor

event

on June

13

was

not investigated.

It should also

be

noted

that

due

to the

superior

capabilities

of the

plants

computer

systems,

significantly

more

data

is available,

and

much

more

than

the

required testing is actually completed.

In the event that

a large

number of individual SPSV's

were to fail to open

on

a scram,

the entire air

header

that supplies

the

SPSV

can

be depress-

urized

by either

Backup

valve

(SV-1F110A and

SV-1F110B).

Although

the

rods would

scram at

a

much

slower rate,

due to the

venting

process,

the

control

rods

would insert.

The

four rods did not

during

the

October

6 surveillance

testing,

because

only the local test

switches

were

utilized to

the rod.

During the preoperational

testing of Unit 1,

the time to depressurize

the air header

for each

backup

valve

was

43.3

seconds

for

F110A

and

28.21

seconds

for

F100B.

The

backup