ML19291B782
| ML19291B782 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 04/14/1975 |
| From: | Milman P GRINNELL CORP. (FORMERLY ITT GRINNELL CORP.) |
| To: | Knuth D NRC OFFICE OF INSPECTION & ENFORCEMENT (IE) |
| Shared Package | |
| ML19291B783 | List: |
| References | |
| NUDOCS 7911120523 | |
| Download: ML19291B782 (5) | |
Text
af v.9;;
g..._
r r n, r a
i L. ! u.
liTGnnneH Camontion Executive Offices 260 WestExchange Street Providence, RhodeIsland02901 Telephone (401)8317000 April 14, 1975
.,3' Dr. Donald Knuth, Director Office of Inspection and Enforcement U. S. Nuclear Regulatory Commission Washington, D. C.
20555
Dear Dr. Knuth:
On March 27, 1975, we were advised of an apparent problem with Grinnell Pipe Support Snubbers at the Three Mile Island Nuclear Station.
The difficulty was that Three Mile Island personnel could not get the snubbers to lock at the specified 10"/ minute.
After an extensive investigation, including tests at Three Mile Island and an independent testing laboratory, we have determined that,after cleaning, all valves tested performed according to predicted snubbing values and provide adequate seismic restraint.
Other conclusions of this investi-gation are as follows:
1)
The Type BH-1226 control valves furnished with the snubbers when used with General Electric SF-ll54 silicone fluid have a nominal locking velocity of 30"/ minute.
The snubbers in question were initially shipped with General Electric SF96-1000 fluid.
This fluid had a much higher viscosity than the SF-1154 fluid, giving the snubbers a locking velocity of 10"/ minute when they were initially shipped.
Vibration tests showed that with the SF-ll54 fluid and the locking velocity of 30"/ minute, these valves provide adequate seismic restraint.
They are also adequate for the restraint of relief valve lifts and similar system transients.
User acceptance limits could be modified accordingly.
5 02G5 1583 109 79111c0 N 3
[
r...,,,... e u PAGE 2
TO Dr. Donald Knuth, Director April 14, 1975 2)
The control valve malfunctions, resulting in a failure of the snubber to lock up, were caused by contamination contained in the user's hydraulic system.
3) vibration tests indicated that snubbers filled with General Electric SF96-1000 fluid may be marginal in their seismic response.
The following is a detailed account of our investi-gation into this matter:
This report summarizes an investigation into the performance of ITT Grinnell snubber valves of a type which was produced between 1966 and 1969.
This type of valve is designated BH-1226.
This is the number of the assembly drawing in the overhaul instructions.
The investigation was triggered by reports from the Three Mile Island Nuclear Station that EH-1226 valves were failing a functional test instituted at that site.
The acceptance limits for this test were based on data in the 1969 Grinnell Catalog, which indicated 10 inches per minute as the snubbing velocity of these valves.
Initial reports from the field suggested that none of the BH-1226 valves would snub - even when input velocities were over 50 inches per minute.
A subsequent visit to the site, however, revealed that about half of the valves in question locked at velocities between 20 and 30 inches per minute.
Other BH-1226 valves at the site had a conspicuous amount of contamination in the fluid reservoirs installed with them.
I.
Locking Velocity of a New Valve The first part of this investigation consisted of locking velocity measurements performed on a BH-1226 valve from our own supply.
These measurements were made on a closed-loop, servo-hydraulic machine at Teledyne Materials Research in Waltham, Massachusetts on 31 March and 1 April, 1975.
1583 110
ri...
m w. c u-PAGc 3
To' Dr. Donald Knuth, Director April 14, 1975 Tests were conducted with two different hydraulic fluids.
One fluid was SF96-1000 (General Electric Co.,
Silicone Products Dept., Waterford, New York).
This is the fluid used in the first snubbers made by Grinnell.
The other fluid was SF-ll54, which was supplied in the Three Mile Island snubbers.
Shock suppressors shipped for applications subject to nuclear radiation were filled with this latter fluid because of its greater tolerance for nuclear radiation.
In January, 1970, use of SF96-1000 was discontinued entirely.
All tests were performed on new, 2 1/2 inch bore cylinders.
Results of the tests were as follows:
Valve Serial No. S1318 Locking velocity Fluid Tension Compression SF-ll54 30 in/ min.
30 in/ min.
SF96-1000 10 in/ min.
10 in/ min.
II.
Tests of Field Valves The second part of this investigation consisted of locking velocity and dynamic tests performed on four valves returned from the Three Mile Island site.
Two of these valves had been installed in the plant prior to undergoing the field test.
The other two had been stocked as spares.
In each of these pairs of valves, one valve had had improved seals installed immediately prior to the field test; the other had not been disassembled since it was originally shipped.
All of these valves were tested with SF-1154 on a new, 2 1/2 inch bore cylinder.
Results are as follows:
1583 111
,i...
u,...... e a PAGE 4
To' Dr. Donald Knuth, Director April 14, 1975 Valve Lockinc Velocitv_
S/N History Tension Comoression Comments S1631 From Service No Lock Disassembled; Never dis-spool stuck; assembled build-up of contaminant.
Sil73 From Service 30 in/ min 35 in/ min Performed vi-Seals re-bration test placed (see below).
S1330 From Storage 35 in/ min 15 in/ min Build-up of Never dis-contaminant on assembled spool; particles deposited on springs.
S1422 From Storage 40 in/ min 40 in/ min Light build-up Seals re-of contaminant placed on spool; one spring bent.
Having determined that BH-1226 valves will snub (if not contaminated) at velocities around 30 inches / minute, we wanted next to assess the adequacy of these valves for a seismic event.
For seismic protection, specifications generally require a restraint to function at excitation frequencies from 3 Hz to 20 or 30 Hz.
Since the effect of increased locking velocity would be most pronounced at low frequencies, our seismic performance check consisted of a 3 Hz displacement sine wave; and a comparison run at 10 Hz.
Valve Serial No. Sil73 was used.
Results:
The peak-to-peak displacement required to develop the desired restraining force was the same at both frequencies.
Satisfactory seismic restraint was demonstrated in these tests.
Valves S1631 and S1330 were turned over to the ITT Grinnell Research and Development Division for analysis of the contaminant.
I583 112
r,ese r-.e v a PAGE 5
To '
Dr. Donald Knuth, Director April 14, 1975 III. Investication at Plant Site Since we were not yet aware that the Three Mile Island test rig had demonstrated the ability of any BH-1226 valves to lock, an ITT Grinnell engineer went to the site on 4 April, 1975.
He had with him valves serial numbers S1173 and S1422.
His findings were as follows:
A.
Approximately 50% of the BH-1226 valves at the site were locking at 20-30 in/ min.
B.
The remaining 50% locked in ranges from 30 in/ min.
to in excess of 50 in/ min.; however, the fluid contained in these units was heavily contaminated.
A sample of this fluid was turned over to the ITT Grinnell Research and Development Division for analysis.
We trust that the above information is useful to you and your consideration of this problem.
If we can be of any further assistance or furnish any further data concerning these products, please advise me and we will respond at once.
Very truly yours, ITT GRINNELL CORPORATION
<?
h.,
ca.w l /~ -e c, a ;
.y.
PAUL MILMAN, General Manager Pipe Hanger Division PM/tp cc:
R. Shewmaker - U.S.Nuc. Reg.Comm. (3)
H. Thielsch
- Prov.
E.
R.
Dobyns - Prov.
H.
E.
Eastwood - Prov.
1583 113