Bulletin 80-19: Failures of Mercury-Wetted Matrix Relays in Reactor Protective Systems of Operating Nuclear Power Plants Designed by Combustion Engineering

                                                       SSINS No.: 6820 
                                                       Accession No.: 
                                                       8006190022 

                               UNITED STATES 
                        NUCLEAR REGULATORY COMMISSION
                    OFFICE OF INSPECTION AND ENFORCEMENT
                           WASHINGTON, D.C. 20555
                                     
                                July 31, 1980

                                                      IE Bulletin No. 80-19 

FAILURES OF MERCURY-WETTED MATRIX RELAYS IN REACTOR PROTECTIVE SYSTEMS OF 
OPERATING NUCLEAR POWER PLANTS DESIGNED BY COMBUSTION ENGINEERING 

BACKGROUND: 

This bulletin addresses the failures of mercury-wetted relays used in the 
logic matrix of the reactor protective system (RPS) of nuclear power plants 
designed by Combustion Engineering (C-E). Except for Arkansas Nuclear One 
Unit 2 and Palisades, both of which use dry-contact matrix relays, the NRC 
understands that all other operating C-E plants use C.P. Clare Model 
HG2X-1011 mercury-wetted matrix relays in the RPS. 

Mercury-wetted matrix relays manufactured by the Adams and Westlake Company 
were initially used in the Palisades plant; however, because of repeated 
failures of these relays, they were subsequently replaced with relays having
dry-contacts. GTE, the manufacturer of these dry-contact relays, however, 
has since discontinued their production. Thus, although the dry-contact 
relays used at Palisades have performed without a failure since they were 
installed, they are not available for the other operating nuclear power 
plants designed by C-E. 

OPERATING EXPERIENCES AND EVALUATION: 

To date, operating nuclear power plants designed by C-E have reported 
thirty-one (31) failures of mercury-wetted relays used in the logic matrix 
of the RPS. 

Most of the reported failures were "failed-closed" type (i.e., the type that
could inhibit a reactor trip), and four of the reported events involved 
multiple failures (i.e., three relay failures were detected during two 
tests; two other failures were detected during two different tests). Because 
of the redundancy within the RPS, no reported event would have prevented a 
reactor trip; however, the build-up of coincident "failed-closed" failures 
of certain sets of relays could result in trip failures for off-normal 
events. 

The number of single and multiple relay failures reported gives rise to two 
concerns: (1) the total number of failures yields a much higher random 
failure rate than that used in other relay failure estimates*, and (2) the 
number of 


  *  Other relay failure estimates include (1) WASH-1400, "Reactor Safety 
     Study", NRC, October 1975; (2) IEEE Std 500-1977,"IEEE Guide to the 
     Collection and Presentation of Electrical, Electronic, and Sensing 
     Component Reliability Data for Nuclear Power Generating Stations", 
     IEEE, New York; and (3) NUREG/CR-0942, "Nuclear Plant Reliability Data 
     System, 1978 Annual Reports of Cumulative System and Component 
     Reliability", NRC. 
.

IE Bulletin No. 80-19                                      July 31, 1980 
                                                            Page 2 of 2 

multiple failures detected suggests the presence of a common-mode failure 
mechanism. Such a common-mode failure mechanism could result in the build-up
of specific "failed-closed" failures which, in turn, could result in 
anticipated transients without scram (ATWS). Thus, the relatively high 
random failure rate and the suggested common-mode failure mechanism, 
indicate that plants using mercury-wetted matrix relays in the RPS are more 
susceptible to scram failures than predicted in other studies. 

ACTIONS TO BE TAKEN BY HOLDERS OF CONSTRUCTION PERMITS OR OPERATING LICENSES
FOR NUCLEAR POWER FACILITIES: 

1.   Review your facility to determine whether or not mercury-wetted relays 
     are used in the RPS. If no such relays are used, you should submit a 
     negative declaration to this effect and you need not respond to the 
     remaining items in this bulletin. Your negative declaration shall be 
     submitted to the appropriate NRC regional office within thirty (30) 
     days of the date of this bulletin and a copy forwarded to the Director, 
     Division of Reactor Operations Inspection, Office of Inspection and 
     Enforcement, NRC, Washington, D. C. 20555. 

2.   Licensees of operating nuclear power plants using mercury-wetted relays
     in the RPS should increase the frequency of their surveillance tests. 
     Until further notice, or until the mercury-wetted relays have been 
     replaced with qualified relays of a different design, surveillance 
     testing of the relays shall be initiated within ten (10) days of the 
     date of this bulletin and repeated at intervals not exceeding ten (10) 
     days thereafter. Upon detecting a failed relay, the failed unit shall 
     be replaced with a qualified dry-contact relay or a new mercury-wetted 
     relay. (The removed relay shall not be reused in the RPS.) 

3.   Nuclear power facilities which are using or whose design includes the 
     use of mercury-wetted matrix relays in the RPS shall submit either 
     their plans and schedules for replacing the mercury-wetted relays with 
     qualified relays of a different design, or justification for using the 
     mercury-wetted relays. Responses to this item shall be submitted to the 
     offices listed in Item 1, above, within ninety (90) days of the date of 
     this bulletin. 

Approved by GAO, B180225 (ROO72); clearance expires July 31, 1980. Approval 
was given under a blanket clearance specifically for identified generic 
problems.