ML20133A606
Text
I 4
i ~...i 'e O
Q.h
. M.T.T.!u.'M ';."<!P G e
/
-.- 1418 E. Linden Avenue
~
~ _ _ _ -.
a g
P.O. Box 152
.44 C@m2 p3 @Co EM"
~
mo. ment ane con:coi systems
/
3030 Red Hat Lane Whitt er. CA 90601 (213)692 9021 Teles 67 4758 April 8, 1981 Uldis Potapors - Chief Nuclear Regulatory Commission Inspection & Enforcement Region IV Vendor Inspection Branch 611 Ryan Plaza Drive, Suite 1000 Arling ton, Texas 76011
Dear Sir:
On 3 April 1980 and again on 31 July 1980 and 22 August 1980,
, letters were sent pursuant of a 10CFR21 report of a possible defect in the Comsip sample pump.
Since these dates further data has been gathered.
Pump'diaphram f ailures occurred at the Hatch nuclear station of Georgia Power Company.
These f ailures initiated the 10CFR21 report by Comsip.
Further investigation revealed that repeated and prolonged closure of the containment isolation valves to the Comsip System without shut down of the pump was the cause of the failures.
Tests at Comsip on a dead headed pump did not end ih the diaphram failure but did break the motorshaf t ex-tension.
in both cases the system pressure specifications were grossly exceeded.
No other failures have been reported from the field.
To provide additional margin, a kit has beeft made available.to install a bypass around the pump to limit the pressure under deadheaded operation and the motor shaft extension has been re-designed for greater strength.
Tests of this arrangement have yielded no excessive temperature rise which would be detrimental to the system.
This arrangement is also included in'all K III and K IV systems of Comsip's present production.
kn Q 5 850227 QQc(
DERNABE84-206 PDR fJyf
s h
Uldis Potapors - Chief Nuclear Regul'atory Co= mission April 8, 1981 Page.2 The high head pres ~sure does not load up the motor because the 1 hp motor used is larger than needed.
However, this motor is the smallest lE motor available.
Therefore, overload as a means of limiting dead heading damage cannot be used.
There are low flow alarms on the Comsip system both at the local and remote stations, thin alarms were ignored in the case of the Harch failure.
We know of no way, without reducing the relia-bility of the system, of the insuring against operator error.
hnC E [ " $ MU E M U G'1Aded~
rIt is nW
%at posb$#JP' egg'c '
ntial n W iazzar,JaEf~ifie# 55~ifETYIPfiYi"'
' ver reaction'in zealous effort to comply,with 10 CFR21.
o Therefore, Comsip rescinds the 10CFR21 reports of 3 April, 31 July and 22 August of 1980 in so far as disclosing a report-able defect.
Very truly yours, COMSIP, INC.
D phi Systems Division 1
kW h
$'Y Robinson echnical Director AR/cld Encl.
D e
b e
4 l
l l
1
a Q-)
G.ey
..\\
.y1 c.-,e ; er nc.<; 6
~
./ g 1418 E. Unden Avenue ~'~~
g a
P.O. Box 152
.(
Uncen. NJ 07036 "h-2WRh h,9 h E,'fi,
as a
A
' r 0 ~ -
A.. /
- nSifU:nF.! Gnci Control SySierns
- /
3030 Red Het Lane
./
Wruttier. CA 90601 (213)692 9021 Teler 67-4768 E.E 10CFR21 Reports 8 April 81 In reviewing the letters concerning the 10CFR reports of possible defect in the Comsip pump, it was noticed that the description of events in the letter of 3 April 80 can be interpreted in a way not strictly in accordance with the facts.
The pump in which the shaf t failure occurred was being used in the Comsip IEEE 323, 1974 Qualification Program.
The pump was at the test facility of E. A. &T. in Playa Del Rey.
The pump had been' dis-connected from the test bed for leak testing which included capping the outlet of the pump.
After testing,the pump was started with-out removing the outlet cap.
Within a short period of time, es timated by E. A. &T. personnel as 20 seconds, the motor shaf t extention failed.
Inspection of the pump by Comsip disclosed a broken motor shaf t but no. other damage.
The motor was returned to Reliance Electric while engineers at Reliance and Comsip technical people investigated new shaf t designs.
A motor with the new shaf t design was installed in the old pump for the qualification program.
The details of the new shaf t design are outlined in the letter of 31 July 1980.
No measure of the pressure at the outlet of the dead headed pump was made.
However, measurement's made on a pump with the new motor s.haf t gave pressures in excess of 250 psig.
Auditable records exist at E. A&T.
There was an in-house f ailure of a pump mmotor shaf t of the old design on Job 80-637.. The failure occurred after four hours of dead-headed service with a bypass alve set at 100 psig as de-scribed in the letter of 31 July 1, 30.
E W
s y' ALFRED ROBINSON Technical Director Comsip, Inc.
fo/& $V 306 GM1
-