Information Notice No. 84-55, Supplement 1: Seal Table Leaks at PWRs

                                                     SSINS No.: 6835       
                                                     IN 84-55, Supplement 1 

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

                                May 14, 1985

Information Notice No. 84-55, SUPPLEMENT 1:    SEAL TABLE LEAKS AT PWRs 

Addressees: 

All nuclear power reactor facilities holding an operating license (OL) or a 
construction permit (CP). 

Purpose: 

This supplement to Information Notice (IN) 84-55 provides additional 
information concerning seal table leaks and failures of compression type 
mechanical fittings. A correction also has been included in this supplement 
concerning the use of the term "SWAGELOK" fitting. It is expected that 
recipients will review the information for applicability to their facilities
and consider actions, if appropriate, to preclude a similar problem 
occurring at their facilities. However, suggestions contained in this 
supplement do not constitute requirements; therefore, no specific action or 
written response is required. 

Background: 

IN 84-55 described two events that led to primary reactor coolant leaks at 
the seal table. The leaks were caused by failure of the mechanical seals 
during maintenance under high-pressure conditions. These events occurred at 
Sequoyah 1 and Zion 1. INPO Significant Event Report (SER) 43-84 and 
Supplement 1 to SER 43-84 also discuss these two events as well as the 
Trojan event (LER 84-014) that occurred on September 13, 1984. 

Further review of the Sequoyah event by the Tennessee Valley Authority (TVA)
determined the exact cause of the high-pressure seal failure. Their review 
showed that the design of the as-supplied cleaning tool had been modified by
Sequoyah personnel and that these modifications produced a tool that would 
transfer force to the thimble tube high-pressure seal at a level sufficient 
to cause separation of the seal assembly. The original cleaning tool, 
supplied by Teleflex, Inc., consisted of a drivebox and flexible plastic 
tube that connected to the upper flare fitting at the end of the thimble 
tube. Use of the handcrank on the drivebox moved the cleaning tool into or 
out of the thimble tube. Excessive force could not be transmitted to the 
fitting of the high-pressure seal because the flexible plastic tube would be 
the weakest component subjected to stress. However, because of the 
flexibility of the plastic tube, the workers had difficulty feeding the 
cleaning device through the drivebox into the tube. During 1978 or 1979, 
plant personnel, to compensate for this difficulty, had modified the tool by 
a metal extension sleeve that threaded onto the thimble tube flare fitting. 
This modified tool was remodified on several occasions because of missing 
pieces or further attempts to improve tool stability. 
.

                                                     IN 85-55, Supplement 1 
                                                     May 14, 1985          
                                                     Page 2 of 4           

The tool that was used during the Cycle 3 refueling consisted of a base that
slipped over the thimble tube assembly at the seal table and a metal 
extension piece that threaded on to the thimble tube flare fitting to which 
the drivebox was attached. This tool was lost during containment cleanup 
activities before plant startup. When the decision was made to clean the 
tubes at power, another tool was fabricated with a slightly shorter base, 
which did not mate flush with the upper extension piece. Shims were used to 
try to correct this problem, but they appear to have still allowed some 
fulcrum effect causing the seal to fail. 

In addition to the events discussed in IE IN 84-55, several other similar 
events occurred during the past year. These are briefly described below. 

1.   On October 23, 1984 an unisolatable reactor coolant leak occurred at 
     Catawba Nuclear Station, Unit 1 (LER 84-18). The unit was in hot 
     standby with reactor coolant system (RCS) pressure at 1500 psig and 
     temperature at 440F. The leak occurred when the stainless steel 
     conduit containing an incore thermocouple separated from the mechanical 
     compression tube fitting. An approximate 5 gpm leak rate occurred and a 
     total of 12,000 gallons of coolant from the RCS leaked to the 
     containment floor and equipment sump. 

     Evaluation of the failure indicated that the conduit had not been fully
     inserted into the fitting and the fitting had been tightened only 13/20
     turns. The required number of turns is 11/4 turns. Another thermocouple
     also was found to be loose. The cause of the failure was evaluated as 
     a construction/installation deficiency. 

2.   Two events occurred at the Rancho Seco Nuclear Plant. The first 
     occurred on April 20, 1984, when a 3/8 stainless steel sensing line 
     blew out of a compression fitting under system pressure (2200 psig) 
     following routine recalibration of a pressurizer level transmitter. The 
     second event occurred on July 31, 1984. A leak was noticed on a steam 
     generator level transmitter sensing line fitting. An attempt to tighten 
     the fitting worsened the leak. Before attempting to further tighten the 
     fitting the technician tapped the sensing line with a wrench, whereupon 
     the stainless steel line blew out of the fitting. 

     Both of the fittings involved were installed during or after a 1983 
     refueling outage. A decision was made to inspect 1444 compression 
     fittings installed during that period. The inspection revealed that 
     approximately 3% of the tube fittings were improperly made up as a 
     result of poor workmanship. A few instances of improperly oriented or 
     missing ferrules were found as gross errors, but the majority of the 
     deficiencies found concerned the improper location of the ferrule with 
     respect to the end of the tube. 

     These problems were attributable to poor  workmanship in the cutting of
     the tubing and also to potential spring back of the tubing after firm 
     seating in the fitting during initial assembly. 
.

                                                    IN 85-55, Supplement 1 
                                                    May 14, 1985          
                                                    Page 3 of 4           


3.   On October 25, 1983, a dewater and air drying operation was being 
     conducted at 39 thimble locations on the seal table at D. C. Cook Unit 
     1. The reactor was at about 50% power throughout the 2-day effort. At 
     the conclusion of the maintenance effort, all guide tube connections 
     were checked and found satisfactory except for one. This fitting had 
     apparently not been properly assembled because the middle high-pressure
     seal came loose causing the thimble to rise approximately 5 inches 
     above the seal table before it stopped. Clamps were reconnected to 
     prevent any further rising of the thimble out of the core. 

Discussion: 

Compression fittings are used throughout nuclear plants in primary, 
secondary, and auxiliary systems. In some applications, failure of a fitting
may be nothing more than a nuisance. In other applications serious events 
could follow the failure of one of these fittings; however, past failures of
fittings cannot be traced to any one single source. The examples cited 
indicate that the failures are varied and can include the following: 
improper maintenance, improper installation, improperly designed tools, and 
improper materials. 

In the Sequoyah event, modifications made to the cleaning tool were not 
controlled adequately by existing formal programs. There were no records 
that any of the modifications to the tool had been technically evaluated or 
tested to determine their effect on the thimble tube seal. Lack of such 
controls appears to have been a significant precursor to the event that 
occurred. 

In the Trojan, Rancho Seco, D. C. Cook, and Catawba events, the immediate 
cause was improper installation of the fittings coupled with corrective 
maintenance on the fittings while the RCS was hot and pressurized. 
Compression fittings exhibit a mode of failure involving pullout of the 
tubing from the fitting when subjected to an axial load. Tensile tests have 
demonstrated this mode of failure can occur when the ferrule is improperly 
located on the tube during initial assembly. When this occurs, it can cause 
a substantial reduction in the-safety factor of the fitting. Thus, 
maintenance on the fitting under these conditions has led to a substantial 
number of fitting failures. 

Some events involving the failure of compression fittings have become known 
because of the circumstances that followed the fitting failure. However, 
because of the varied applications of the compression fitting, many failures
of fittings, or problems associated with the fittings, probably go 
unreported. Because of the number of severe events that have occurred during
the past year, it is appropriate that positive steps be taken to reduce the 
number of similar events from occurring. 

In this regard, some utilities have revised procedures on the assembly and 
maintenance of these fittings to address the problems mentioned above. They 
also have upgraded training efforts on the assembly and maintenance of the 
fittings and have enacted stricter controls on when maintenance can be 
conducted (e.g., restricting maintenance on fittings used in hot, 
pressurized systems). Additionally, Westinghouse is developing fitting 
inspection guidelines and detailed acceptance criteria. 
.

                                                  IN 85-55, Supplement 1 
                                                  May 14, 1985 
                                                  Page 4 of 4 

In the original IE information notice, reference was made to the failure of 
"SWAGELOK" fittings. It was found that at Sequoyah and Trojan the 
compression fittings used were hybrid fittings assembled using parts from 
other vendors, including Crawford Fitting Co. Brand names of other 
compression fittings include SNO-Trik, Gyrolok, Megalok, Ringlok, Unilok, 
Wadelok, etc. Apparently, many of the parts are used interchangeably. In the 
case of Sequoyah, "Gyrolok" and "Swagelok" parts were used to construct the 
fittings. Therefore, use of the name "SWAGELOK" fitting in the above 
applications was inaccurate. A "SWAGELOK" fitting has a nut, front and back 
ferrule, and body, all manufactured by Crawford. In fact, Crawford cautions 
against interchanging parts of tube fittings made by other manufacturers 
with "SWAGELOK" tube fitting parts. 

No specific action or written response is required by this information 
notice. If you have any questions regarding this matter, please contact the 
Regional Administrator of the appropriate NRC regional office or the 
technical contact listed below. 



                                   Edward L. Jordan Director 
                                   Division of Emergency Preparedness 
                                     and Engineering Response 
                                   Office of Inspection and Enforcement 

Technical Contact:  D. Powell, IE
                    (301) 492-7155

Attachment: List of Recently Issued IE Information Notices