Part 21 Report - 1995-188

ACCESSION #: 9508170275

                                             KEYSTONE
                                        
3950 Greenbriar, Stafford, Texas 77477       Anderson, Greenwood & Co.

August 10, 1995

VIA FAX TO NRC OPERATIONS CENTER (301) 816-5151

Document Control Desk
U.S. Nuclear Regulatory Commission 
Washington, D.C. 20555

Ref: Our July 14, 1995 10CFR21 Noncompliance Report

Gentlemen:

The purpose of this correspondence is to provide results of testing
performed to evaluate the problem reported in our July 14, 1995, letter.

Testing was performed to establish the amount of deviation from nameplate
set pressure that could result from testing AGCO's 81P using air as the
test media instead of water.  Valves were assembled and set pressure
adjustments were made following the factory procedure using water as the
test media.  The valves were then checked using air as the test media. 
If the air test method did not result in the same set pressure value,
further adjustments were made to satisfy the air method criteria.  For
valves where adjustments were necessary, the factory water method was
repeated and the resulting set pressure recorded.  The difference between
the first water test setting and the second water test reading represents
the potential deviation from nameplate set that could result from
adjusting the valve using the air method.

Two series of tests were performed.  One using KEL-F seat material and
another using PEEK seat material.  Set pressures ranging from 100 psi to
2000 psi were tested.  Results of the tests indicated the following:

                                                  Potential Set
                         Set Pressure Range     Pressure Deviation
Valve/Seat Material             (psig)          Above Nameplate (%)

     81P/KEL-F                100-750                  4-15
     81P/KEL-F                   >750                  None
     81P/PEEK                 100-750                  20-40
     81P/PEEK                    >750                  None

Since the air test method is based on the valve's first leakage on
increasing pressure, results are very dependent on the effectiveness of
the valve seat.  PEEK is a very hard plastic relative to

P.O. Box 944,                      Phone: (713) 274-4400,
Stafford, Texas 77497-0944         Telex: 0775219, Fax: (713) 240-1800



Document Control Desk
U.S. Nuclear Regulatory Commission 
August 10, 1995
Page 2

KEL-F and requires more seating load to effect a seal.  Therefore the
potential deviation for PEEK is greater than for KEL-F.  For both
materials, the higher set pressures (>750 psi) showed no deviation
between water and air test methods.  For the lower set pressures, the
range of deviations observed in our tests showed no significant trend.

The deviations shown in the above table are only applicable to AGCO's
Model 81P liquid relief valve that has had its set pressure adjusted
after shipment from the factory using air as the test media and first
crack (leak) at 95% of set pressure as the criteria.  Any valve that has
AGCO's factory lead seal unbroken or has been adjusted using the water
test method is unaffected.

For set pressures below 750 psi, the 81P should only be set pressure
adjusted using the water test method described in Procedure N05.9040.180,
Revision F.  For set pressures above 750 psi, either the water or air
test method may be used.

We are in the process of preparing letters to all customers who have
purchased this model explaining the above findings and recommended
procedures.

Please do not hesitate to contact me if you have any questions.

Very truly yours,

ANDERSON, GREENWOOD & CO.


J. Alan West, P.E.
Vice President, Engineering

JAW:mnp
Atta.

c:   Mr. Greg Hyland 
     Mr. Juan Gomez 
     Mr. Ed Ross 
     Mr. Doug Snyder 
     Mr. Steve Stansbury 
     Mr. Jim Baca

NRC08.095

*** END OF DOCUMENT ***