Protecting People and the EnvironmentUNITED STATES NUCLEAR REGULATORY COMMISSION
ACCESSION #: 9711240143
Nuclear Reactor Laboratory The University of John G. Williams,
Engineering Building (20) Arizona Director
P.O. Box 210020 Tucson Arizona e-mail:
Tucson, Arizona 85721-0020 jgw@bigdog.
engr.arizona.edu
voice (520) 621-9729
FAX (520) 621-8096
November 13, 1997
U.S. Nuclear Regulatory Commission
Document Control Desk
Mail Stop P1-37 10CFR50.71
One White Flint North 10CFR21.21
11555 Rockville Pike
Rockville, MD 20852
Subject: Facility License No. R-52, Docket 50-113
Information report, concerning malfunction of Mode
Selector Switch
The occurrence documented on the following pages was reported to the NRC
by telephone and fax on 5 November 1997. This report supplement that
information and includes analysis, conclusions and remedial actions that
will be reviewed by the Reactor Committee for this facility.
John G. Williams, Director
Director, Nuclear Reactor Laboratory
JGW/dg
enclosure
copies:
Non-Power Reactors and Decommissioning Project Directorate, USNRC
Mr. Marvin Mendonca, USNRC
Dr. M. Cusanovich, Vice Pres. for Research, University of Arizona
Reactor Committee Members
[Illegible print]
FACILITY LICENSE R-52
Malfunction of Mode Switch in Pulse Mode
November 13, 1997:
Description of the Occurrence
On November 4, 1997, a pulse was performed with the mode switch in its
Pulse High setting and with the transient rod anvil set for a $1.75
pulse. The transient rod operated normally, fuel temperature rise was
recorded showing energy deposition in the range expected for a normal
pulse and the timer scram operated normally. No valid indication
appeared, however, for the peak reactor power or the integrated pulse
energy. The right safety channel indicated a trip on reactor power.
A digital data acquisition system connected to the right safety channel
showed that the signal on that channel displayed a normal reactor period
of approximately 11 ms, but then saturated This showed that the channel
was connected to a valid indication of reactor power level from the ion
chamber that is normally connected in steady state mode, not the high
range ion chamber which should be selected in pulse mode. ne trip on the
right safety channel was attributed to power in excess of 110 kW, which
is the steady state limit.
Testing of the mode switch revealed that changing the switch from its
manual to pulse positions did not reliably switch between the normal and
high range signals. The problem was corrected after removal of the rear
casing of the mode switch and application of switch cleaning fluid.
After reassembly, the system appeared to behave normally. The reactor
was then operated at 95 kW steady state while the high range channel was
calibrated and shown to behave normally. A pulse was then performed and
normal indications were found on all channels.
Analysis and diagnosis of the cause
A malfunction of the mode switch was the cause. It resulted in failure
to pick a relay which switches from the normal to the high range channel
when pulse mode is selected, and which also switches in the nvt circuitry
that Indicates Integrated pulse energy.
The following symptoms demonstrate that the high range channel was not
engaged, but Instead the normal range uncompensated ion chamber was
operating: (a) the reactor tripped on a high power scram from the right
safety channel, even though the calculated peak power was less than 100
MW, and the trip was set at 1000 MW for pulse mode; (b) fuel temperature
Indications showed a peak temperature of 186 C, in the normal range for a
$1.75 pulse, corresponding to energy deposition of 9.3 MW and peak power
of 92.4 MW; (c) a digital data collection system connected to the channel
(after the switching
relay) saturated, though it should not have done so below 1100 MW; (d)
prior to saturation, the same data showed an exponential increase with a
period of 11 ms, normal for a $1.75 pulse, an indication that the channel
was connected to a true power indication, though not the high range one.
On November 10, 1997, the mode switch was removed from the console for
inspection and cleaning (see attached memo and part drawing). As a
result of this inspection it was determined that the problem had actually
been caused by a mechanical misalignment of the rear wafer of the switch,
not by dirt. This was attributed to wear and aging that had loosened the
nuts holding the rear case. This also loosened the wafer assembly which
is held together by the same nuts. The problem was corrected not by the
cleaning fluid, as Initially supposed but by the retightening of these
nuts that occurred on reassembly.
Corrective and remedial action
The immediate problem was corrected before the reactor was placed back in
service on November 4. Tightening of the nuts holding the casing of the
mode switch also secured the switch wafers in their correct operating
alignment.
To prevent a recurrence, the Electronics Technician and the Reactor
Supervisor have proposed a small modification in the switch assembly, so
that the wafer assembly is held together independently of the nuts which
secure the rear casing. Details are given in the attached memo from Mr.
Lohmeier. These proposed changes will be submitted for approval to the
Reactor Committee before implementation.
Conclusions
A malfunction that was attributed in part to aging was found in the Mode
Selector Switch (General Atomics part number ELD 239-4300C). It was
corrected by tightening two nuts. To prevent a recurrence, a minor
change in the assembly has been proposed. Licensees of other facilities
using this GA part may wish to consider the applicability of this
experience to their systems.
The system failed safe causing actuation of reactor scram at 110 kW. The
reactor did not operate outside its normal envelope, and there was no
threat to health or safety of the public.
J. G. Williams, Reactor Laboratory Director
Date: 11/13/97
To: Dr. John Williams
From: Wayne Lohmeier
Subject: Mode Switch Problem
Dr. Williams,
On November 4, the day of the pulse problem, the Mode Selector Switch
(P/N ELD 239-4300C) rear case was removed and contact cleaner was sprayed
on the switch contacts. The case was then re-installed and the reactor
functioned normally.
On November 10, the switch was removed from the reactor console and
cleaned. Cleaning involved removing a foam like residue from the spray
contact cleaner. There was no obvious "dirt". The switch wafers are
constructed with clear plastic in front of and behind the switch
contacts. Although not hermetically sealed, the contacts are protected
from the outside elements.
The following observations were made in the switch construction. There
are two screws approximately 3 inches long that hold the switch frame,
the front case. the wafers, and the rear case. The assembly is secured
by the screws and a washer, a lock washer. and a nut on the outside of
the rear case.
My recollection of removing the rear case on November 4 is, the rear case
was not aligned with the front case (off by approximately 10 degrees) and
the nuts holding the rear case were not very tight.
The wafer that picks the "Pulse High/Lo" relay is the "D wafer" located
at the rear of the assembly. I suspect the problem was, the rear wafer
was lagging the mode switch knob because of the mis-alignment and nuts
not being tight.
Harry could feel a binding in the switch when going to the Pulse High/Lo
positions prior to removing the rear case and cleaning. When the
reassembly was completed, it felt smooth, and an audible sound of the
switch engaging at each position is now present.
Corrective Action:
1. Shorten the rear spacers between the rear wafer and rear case by the
thickness of a lockwasher and nut.
2. Assemble the switch wafers to the switch frame and front case using
a lockwasher and nut. Wafer assembly will now not rely on rear
cover for mechanical support.
3. Attach rear case using a lock washer and nut.
Figure "-1 ASSEMBLY and, DETAIL ITEM 9" omitted.
Figure "ELD 239-4300C Mode switch assembly" omitted.
*** END OF DOCUMENT ***
