United States Nuclear Regulatory Commission - Protecting People and the Environment

Potter & Brumfield Model MDR Rotary Relay Failures

                                UNITED STATES
                           WASHINGTON, D.C.  20555

                               January 6, 1992



All holders of operating licenses or construction permits for nuclear power 


The U.S. Nuclear Regulatory Commission (NRC) is issuing this information 
notice to alert addressees of failures experienced with MDR series Potter & 
Brumfield (P&B) rotary relays installed in safety-related systems at certain 
nuclear power plants.  It is expected that recipients will review the 
information for applicability to their facilities and consider actions, as 
appropriate, to avoid similar problems.  However, suggestions contained in 
this information notice are not NRC requirements; therefore, no specific 
action or written response is required.

Description of Circumstances

On January 14, 1986, September 17, 1987, and December 8, 1987, an emergency 
diesel generator (EDG) failed an operability surveillance test at the 
LaSalle County Station, Units 1 and 2.  In each case, while the Commonwealth 
Edison Company (CECO) attempted to synchronize the EDG to its bus, the EDG 
output breaker would not close.  CECO replaced all P&B MDR relays in the 
output breaker closing circuits with General Electric HFA relays.  The NRC 
staff has received no reports of relay failures at LaSalle affecting EDGs 
since these were replaced.

On October 10, 1988, the Arizona Public Service Company (APSC), the licensee 
for the Palo Verde Nuclear Generating Station, submitted a report in 
accordance with Title 10 of the Code of Federal Regulations, Part 21 (10 CFR 
Part 21).  This report documented 18 instances over a 2-year period in which 
P&B MDR relays failed to change position. 

APSC detected these failures during either routine surveillance testing or 
actuation of the engineered safety features (ESF) actuation system or the 
reactor trip switchgear.  After replacing all P&B MDR series relays, APSC 
experienced only two failures; improperly sized coils or contamination in 
the insulating material of the switch caused these two failures.


                                                            IN 92-04
                                                            January 6, 1992
                                                            Page 2 of 3

On July 19, 1991, during a monthly surveillance test, the River Bend Station 
experienced ESF actuation of containment isolation valves, control room 
filter trains, the standby gas treatment system, and the fuel building 
filter trains because of an MDR relay failure.  

On July 23, 1991, during a monthly surveillance test at the River Bend 
Station, the failure of an MDR-5111-1 relay caused an ESF isolation of a 
reactor water sample valve.  The Gulf States Utilities Company (GSU), the 
licensee, committed to replace all P&B MDR relays.


P&B MDR relays are used in various safety-related applications in commer-cial 
nuclear power plants with reactors manufactured by the Babcock and Wilcox 
Company; Combustion Engineering, Incorporated; the General Electric Company; 
and the Westinghouse Electric Corporation.  Industry records identify over 
60 instances of P&B MDR rotary relays failing to operate properly since 

An MDR relay failure may cause the loss of a train of the ESF actuation 
system, the emergency core cooling system, or the reactor protection system.  
A common-mode failure may result in the loss of one or more of these 
systems.  GSU performed a probabilistic risk assessment (PRA) of the reactor 
protection system at River Bend, based on plant-specific, P&B MDR relay 
failure rates that were greater than the generic failure rates by a factor 
of 5.1.  This PRA showed that the reactor protection system failure rate 
increased by a factor of 25 to 3.3E-4.  

The principal failure mechanism of P&B MDR rotary relays appears to be 
mechanical binding of the rotor caused by deposits from coil varnish 
outgassing and chlorine corrosion products.  A secondary failure mechanism 
appears to be the intermittent continuity of electrical contacts.  A number 
of variables contribute to these failure mechanisms and cause the relays to 
fail at random mostly within 2 to 5 years of the in-service date.  Failures 
may occur regardless of current or wattage, the use of ac or dc power, or 
whether normally energized or de-energized.  It is also important to note 
that a relay rotor can bind immediately after a surveillance test.

Attachment 1 provides a detailed description of the failure mechanisms, con-
tributing causes, and failure investigations.  Attachment 1 also discusses 
modifications made to P&B MDR series relays by the manufacturer to reduce 
susceptibility to the failure mechanisms discussed above.


                                                            IN 92-04
                                                            January 6, 1992
                                                            Page 3 of 3

This information notice requires no specific action or written response.  If 
you have any questions about the information in this notice, please contact 
one of the technical contacts listed below or the appropriate Office of 
Nuclear Reactor Regulation (NRR) project manager.

                                   Charles E. Rossi, Director
                                   Division of Operational Events Assessment
                                   Office of Nuclear Reactor Regulation

Technical contacts:  K. R. Naidu, NRR
                     (301) 504-2980

                     R. A. Spence, AEOD
                     (301) 492-8609

1.  Potter & Brumfield Model MDR Rotary Relays
2.  Figure 1, Potter-Brumfield Model MDR Rotary Relay
3.  Figure 2, MDR Non-Latching Relay
4.  List of Recently Issued NRC Information Notices

                                                            Attachment 1
                                                            IN 92-04
                                                            January 6, 1992
                                                            Page 1 of 4

                 Potter & Brumfield Model MDR Rotary Relays

Description of the MDR Rotary Relay

Potter & Brumfield (P&B) manufactures two types of MDR rotary relays:  
latching and non-latching.  Various series of these relays are provided for 
service at 28 and 125 volts (V) dc and 115 and 440 Vac, with from 4 to 24 
pole double throw (PDT) contacts.  While each series has a different number 
of contact stacks and has a different coil, power, and current capacity, 
each of the series is similarly constructed and exhibits similar failure 

Non-Latching Relays

The non-latching MDR relay has two coils connected in series inside the 
relay which, when energized, rotate the relay rotor shaft, which operates 
the contacts through a shaft extension.  The stator faces and stop ring 
limit the rotor movement to a 30-degree arc.  Two springs return the rotor 
to the stop ring and the contacts to their normal positions when the coils 
are de-energized.  The non-latching MDR relays have two positions: 
"energized" and "de-energized."  (See figures 1 and 2).

Latching Relays

Each relay in the MDR latching series has two sets of series coils, which 
provide a latching two-position operation.  When one set of coils is 
energized, the rotor shaft rotates through a 30-degree arc, changing the 
state of the contacts.  The other set of coils must be energized to return 
the relay to its original position. 

Failure Investigations

The Commonwealth Edison Company (CECO) determined that the three events at 
the LaSalle County Station resulted from the failure of P&B MDR-137-8 or 
MDR-138-8, 125 Vdc normally energized relay contacts to close.  CECO 
performed diagnostic testing after the earlier events but could not repeat 
the failure.  This lack of repeatability is typical of MDR intermittent 

The Arizona Public Service Company (APSC) found that three of the P&B MDR 
relay rotors at Palo Verde Nuclear Generating Station (PVNGS) would not move 
more than 12 degrees of the complete 30-degree arc.  The failed relays, 
located in cabinets without forced ventilation, were in an ambient 
temperature of 95 to 104�F (the design limit is 149�F) and had an external 
surface temperature of 157�F.  

                                                            Attachment 1
                                                            IN 92-04
                                                            January 6, 1992
                                                            Page 2 of 4

APSC detected no relay failures in cabinets with forced ventilation which 
provided an ambient temperature of 81�F or less.  Such relays had a 
temperature of 112�F on their external surfaces.  APSC determined that it 
had applied up to 39.8 Vdc to the 28 Vdc coils.  APSC tested 7 of the 18 
failed relays on an 18-month frequency and 10 on a 62-day frequency.  APSC 
had the relay failures analyzed and determined that varnish on the relay 
coils outgassed, condensed, and accumulated between the rotor shaft and the 
end-bell bearings, binding the rotor and the bearings together.  The 
outgassing was due to excessive coil temperatures that occurred when the 
coils were continuously energized at voltages above their nominal ratings.  
The heat also may have caused the release of chlorine from (1) the PVC 
coating on the fiberglass tubing covering the solder joint between the 
magnet wire and the Teflon coated lead wire, and (2) the Neoprene rubber 
grommet through which the coil lead wires penetrate the base of the relay.  
The chlorine corroded brass parts inside the relay.  P&B and APSC concluded 
that long intervals between de-energizing of the relays may have also 
contributed to the failures.  

In May 1989, APSC installed replacement P&B relays at PVNGS that were 
manufactured with coils coated with epoxy instead of varnish.  APSC 
conducted tests and found that 5 of the 42 relays tested would not rotate to 
their de-energized position and that 5 other relays operated slowly.  Two 
independent laboratories observed that; (1) the relays' epoxy was not 
properly cured, (2) uncured epoxy contaminated the rotor and (3) P&B did not 
de-aerate the epoxy prior to use, contrary to the manufacturer's 
recommendations.  This caused the rotor and stator surfaces to bond 
together, preventing the rotor from rotating freely.  P&B informed the NRC 
that APSC returned the 42 relays and that P&B rebuilt them.

On September 10, 1990, the General Electric Nuclear Energy Division (GENE) 
issued Rapid Information Communication Services Information Letter 053 to 
address P&B MDR relay failures reported at two GE boiling water reactors.  
P&B believed that chlorine released from rubber grommets and polyvinyl 
chloride sleeves caused corrosion and that varnish on the coils outgassed 
while the relays were continuously energized.  Both chlorine-corrosion 
products and varnish accumulated in the bottom end-bell bearing and caused 
the rotor shaft to bond to the bearing.  P&B suspected that the failed 
relays were exposed to high ambient temperatures and could have been exposed 
to high coil voltages or could have been rarely cycled.

On November 2, 1990, GENE issued Potentially Reportable Condition 90-11 in 
which it stated that both 24 Vdc and 120 Vdc coils had lower coil powers 
than the 125 Vdc relays and were therefore not vulnerable to this failure 
mode.  GENE concluded that no substantial safety hazard existed.  However, 
upon investigating the failed MDR relays at River Bend as discussed below, 
the NRC obtained results that may contradict these conclusions.

On July 19, 1991, a high resistance on one set of contacts on a P&B 24 Vdc, 
MDR-5111-1 rotary relay, which should have been closed, caused a voltage 
drop to the downstream relays which opened their contacts and resulted in an 

                                                            Attachment 1
                                                            IN 92-04
                                                            January 6, 1992
                                                            Page 3 of 4

actuation at the River Bend Station.  The Gulf States Utilities Company 
(GSU), the licensee, later performed bench testing of this failed relay and 
verified that the relay actuated properly and all contacts changed state 
properly, and exhibited proper continuity.  The coil was meggered and found 
to be acceptable.  The contacts all appeared to be clean and shiny, with no 
evidence of pitting or residue.  GSU found no foreign material in the relay 
or on the rotor shaft and found nothing that may have contributed to the 
high resistance across the contacts.  

On July 23, 1991, GSU investigated another MDR relay failure at River Bend 
and found two MDR-5111-1 relay contacts open that should have been closed 
when the coil was energized.  GSU also found that the contacts operated 
intermittently with some contacts closing several minutes after the coil was 
energized or sometimes not at all.  

Both River Bend failed relays had been in service within tightly-regulated 
design voltage and temperature conditions and were mounted inside stainless 
steel isolation cans for divisional separation.  GSU measured the 
temperature inside the isolation can at 113�F, while the ambient cabinet 
temperature was 92�F.  In each case, the failed relay had been recently 
cycled because of a short loss of power to the coil that had occurred a few 
days before the relay failure was discovered, and it appears that not all 
contacts engaged properly when power was restored.  

Failure Mechanisms

The primary failure mechanism of the P&B model MDR rotary relay appears to 
be a mechanical binding of the rotor caused by organic outgassing and 
deposition of contaminants and corrosion particles on the relay rotor shaft.  
The contaminants are deposited in the end bell bearings and sleeves and 
cause the rotor shaft to bond or stick to the bearing, preventing the rotor 
shaft from fully rotating when the relay coils are energized or 
de-energized.  The principal contaminant is outgassed material emitted from 
the brown enamel varnish used to coat the relay coils.  This contamination 
may not be apparent to the naked eye.  The corrosion results from chlorine 
released from the rubber grommets and the polyvinyl chloride sleeves.  Gulf 
States and P&B disassembled six operable and two failed relays that had been 
in service since December 1984.  The thickness and color of the deposits on 
the rotor, sleeve, and end-bell bearings of the relays varied widely among 
the eight relays, indicating varnish outgassing. 

A secondary failure mechanism appears to be intermittent continuity of the 
electrical contacts.  High resistance and intermittent continuity may result 
from chemical reactions on the fixed and movable silver contacts.  P&B 
tested a MDR-5112-1, 125 Vdc relay that had been in service at River Bend 
and found intermittent continuity on a set of clean, unused contacts.

A number of variables contribute to these failure mechanisms and reduce the 
length of the operating life of the complex P&B MDR rotary relays.  These 
variables include coil wattage, applied ac or dc voltage, normally energized 
or de-energized coil, manufacturing tolerances, ambient and coil 
temperatures, varnish thickness, mounting configurations and enclosures, 
cabinet ventilation, 

                                                            Attachment 1
                                                            IN 92-04
                                                            January 6, 1992
                                                            Page 4 of 4

relay breathing, testing frequency, operational cycling, the number of 
contact decks, and the amperage and voltage of the contact load.  These 
contributory factors cause an apparent random failure history.  While each 
of the MDR relays failed between 1 month to 13 years after it was placed in 
service, most failed within 2 to 5 years.  

Modifications to MDR Relays

P&B has made the following design changes to MDR series relays:

     Changed the movable contacts from silver to silver-cadmium-oxide in 
     October 1985.  However, P&B recommends against using MDR relays with 
     either silver or silver-cadmium-oxide in low current circuits.

     Changed the coil coating from varnish to Dolphon CC-1090 epoxy resin in 
     February 1986.  This reduced the coil outgassing rate.  However, P&B 
     does not de-aerate Dolphon CC-1090 prior to use, contrary to Dolphon's 
     recommendations.  P&B informed the NRC that the epoxy manufacturer 
     plans to cease production of this currently used and tested epoxy.  The 
     NRC is unaware of when P&B will change to a new epoxy.  Licensees may 
     wish to determine if P&B has examined the replacement epoxy for 
     susceptibility to outgassing after aging.  Licensees may also wish to 
     determine if P&B applies the epoxy in accordance with the 
     manufacturer's recommendations.

     Replaced the brass switch studs in medium size MDR relays with 
     stainless steel studs in November 1986.

     Began lubricating end-bell bearings in July 1988.

     Changed chloride-containing materials to chloride-free materials in 
     June 1989.

     Changed the rotor spacers from brass to stainless steel in May 1990.

     Changed the brass spring retainer in small size MDR relays from brass 
     to stainless steel in May 1990.

     Changed shims from brass to phosphor bronze in May 1990.

P&B had implemented all these modifications to its MDR rotary relay design 
by May 1990.  

When APSC reported having problems with MDR relays at Palo Verde in 1988, 
P&B believed that only relays normally energized with excessive voltage and 
operated infrequently were susceptible to the corrosion and outgassing 
failure mode.  P&B did not notify other licensees about these problems since 
this condition appeared to occur only at plants with reactors manufactured 
by Combustion Engineering, Incorporated.  P&B informed the NRC that since 
1988 it has only supplied MDR relays as commercial grade components without 
accepting the reporting requirements of 10 CFR Part 21.
Page Last Reviewed/Updated Friday, May 22, 2015