ML20046D389
| ML20046D389 | |
| Person / Time | |
|---|---|
| Issue date: | 06/18/1993 |
| From: | Cwalina G, Wilson R Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML20046D385 | List: |
| References | |
| REF-QA-99901264 99901264-93-01, 99901264-93-1, NUDOCS 9308190088 | |
| Download: ML20046D389 (17) | |
Text
_ _ _ _ _ _ _ _
O O
ORGANIZATION:
FLUID COMPONENTS, INCORPORATED 1755 LA COSTA MEADOWS DRIVE SAN MARCOS, CALIFORNIA 92069 REPORT NO.:
99901264/93-01 CORRESPONDENCE Malcolm M. McQueen, President ADDRESS:
Fluid Components, Inc.
1755 La Costa Meadows Drive San Marcos, California 92069 ORGANIZATIONAL Stephen R. Mitchell, Quality Assurance CONTACT:
Manager (619) 744-6950 NUCLEAR INDUSTRY Flow and level instruments for all types of ACTIVITY:
commercial nuclear power plants INSPECTION April 13-15, 1993 CONDUCTED:
TEAM LEADER:
h 6
7[J77 Richard C.
Wilson, Senior Engineer Date Reactive Inspection Section 2 (RIS2)
Vendor Inspection Branch (VIB)
OTHER INSPECTOR:
Stephen D. Alexander, Equipment Qualification-and Test Engineer, RIS2, VIB l
l/A W O. /,,
f
./
APPROVED:
//i P 7A.'s
/~O
~ Gregory C.
Cwalina, Chief Date Reactive Inspection Section 2 Vendor Inspection Branch INSPECTION BASES:
10 CFR Part 21 and 10 CFR Part 50, Appendix B I
INSPECTION SCOPE:
To selectively review the implementation of FCI's quality assurance program for supplying nuclear safety-related. equipment, with emphasis on the calibration of flowmeters.
PLANT SITE Numerous APPLICABILITY:
9308190088 930628 N"f PDR GA999 EMVFLUID 99901264 PDR m
___ 2 _ - _ ______-_-____ - -
O U
w 1
INSPECTION
SUMMARY
1.1 Violation 99901264/93-01-01 (Open)
Contrary to the requirements of 10 CFR Part 21, Fluid Components, Incorporated (FCI) did not begin timely evaluations of discrepancy reports describing potential deviations involving flow switch accuracy (see Section 3.6 of this inspection report).
1.2 yiolation 99901264/93-01-02 (Open)
Contrary to the requirements of 10 CFR Part 21, FCI procedures reflected an obsolete revision of 10 CFR Part 21 and did not properly address reporting and evaluation requirements, and FCI had not posted a current copy of 10 CFR Part 21 (see Section 3.7 of this inspection report).
1.3 Nonconformance 99901264/93-01-03 (Open) i Contrary to several criteria of Appendix B to 10 CFR Part 50, which was invoked on FCI by licensee purchase orders, FCI certified the accuracy of flowmeters as 3% of full scale without an adequate basis because of numerous flaws in the calibration
, process (see Sections 3.4 and 3.5 of this inspection report).
2 STATUS OF PREVIOUS INSPECTION FINDINGS There was no previous NRC inspection of this facility.
3 INSPECTION FINDINGS AND OTHER COMMENTS 3.1 Entrance and Exit Meetinos In the entrance meeting on April 13, 1993, the NRC inspectors discussed the scope of the inspection, outlined the areas to be inspected, and established interfaces with FCI management and staff.
In the exit meeting on April 15, 1993, the inspectors discussed their findings and concerns with FCI management and staff.
3.2 Inspection Scope FCI designs and manufactures fluid flow and liquid level instruments.
The company was established in 1964, has about 140 employees, and occupies about 50,000 square feet.
Commercial nuclear power plant business peaked at about 40% of the total, and recently has been in the 10-15% range.
The inspection concentrated on the calibration of type LT81A and similar mass flowmeters, which are frequently supplied for air 2
m
O O
flow meas
'ments in commercial nuclear power plants.
Other areas addressed were the handling of discrepancy reports and the program for meeting the reporting requirements of 10 CFR Part 21.
The FCI facilities were observed, again with emphasis on calibration.
FCI personnel stated that materials were procured commercial grade, certified material test reports were obtained, and material from every vendor was sampled every six months or upon purchase.
The FCI Type LT81A mass flowmeter uses a patented thermal l
dispersion principle.
A fluid flows across two resistance temperature detectors (RTDs), one of which is preferentially heated by a heating element.
The temperature difference between the RTDs varies with fluid flow, and is greatest at zero flow.
l Electronic circuitry converts the difference in RTD resistances to an output signal that is essentially linear with flow.
The signal from the unheated RTD is also used for process fluid temperature compensation.
The NRC inspectors selected seven safety-related purchase orders (POs) for flowmeters from a list of about 150 nuclear plant POs provided by FCI.
One of the seven was in-process and had not reached the calibration stage, so the calibration was not reviewed.
Another PO covered steam flowmeters, which are calibrated differently.
Their calibration was not reviewed, but the inspector noted that the drawing specified " Accuracy 13%."
The customer requirements in each PO were reviewed.
The Pos covered original and replacement equipment.
They specifically invoked 10 CFR Part 21 and Appendix B to 10 CFR Part 50, as well as the technical requirements of applicable earlier POs.
In some cases calibration traceable to the National Institute of Standards and Technology (NIST) was required.
FCI prepared an assembly drawing for each original equipment PO, and the original equipment drawings were used for the replacement equipment Pos'.
In each case the drawing specified the "linearizable flow range" and stated " Accuracy:
3% of full scale."
The only exception reviewed by the inspectors was drawing no. 706146 sheet 3, Revision R, for Georgia Power Company (GPC), which specified repeatability as 1%.of range and accuracy as 5% of range.
This drawing applied to six flowmeters on Sales / Shop Orders (S.O.s) 17586 and 17596, GPC POs P-50658 and P-50659, all dated in February 1989 with calibrations performed on Stand CL on March 9 and 16, 1989..The GPC orders did not clearly invoke Appendix B to 10 CFR Part 50, but did invoke industry standards for environmentally qualified equipment, referenced the FCI qualification test report, and stated that 10 CFR Part 21 applies to nuclear safety items.
All of the flowmeters reviewed had turndown ratios (ranges) of 10:1 or less.
The catalog provided by FCI during the inspection 3
n-g 7
U w
t
_specified an accuracy of i1% of full. scale or 3% of reading,
' j whichever-is better, for turndown ratios of 10:1 or less in air,_
i and repeatability.of i1% of full scale.
The FCI Certificates of Conformance stated that the instruments were certified to have
?
been manufactured, tested, and inspected to the requirements of 1
the PO.
l 3.3 Descrintion of Flowmeter Calibration i
The calibration of an LT81A air mass flowmeter involves the following major steps (excluding the display and totalizer
[
calibrations, which the inepectors did not review), using
" Document # 008072, LT81 Calibration Procedure Board #0017
.i Rev.
B," Revision B, February 7, 1989, for step (1), with no documented procedure for the remainder.
(1)
After temperature compensation, the LT81A is installed in a calibration test stand containing three turbine flowmeters used as transfer standards.
The nonlinearized voltage from the sensor 1 head
(" Pin 6 voltage").is recorded at ten different. flow rates covering the range of the instrument.
The computer-processed l
reading of one of the turbine meters is taken as.the actual air flow through the LT81A at each flow rate.-
Potentiometers in the l
circuitry are adjusted to provide an output signal ~(usually 4-20 hA) that is linear'with flow.
A five-point final check'of the linearized output signal against the Pin 6 voltage is then made.
in the test stand.
i The calibration procedure states that the allowable tolerance on_
i the output is 1% of full scale, and specifies that-if any signal a
does not fall within tolerance "some slight adjustments to the i
calibration will be required," as directed by an experienced q
technician during the final check.
(As-noted above, FCI'
.i specifies and certifies 3% of full scale accuracy; the'1% value is an in process criterion.
The "4 to;20 mA calibration table" i
sheet, used to record the final five-point. check of linearized
[
output vs. flow that is specified in the calibration procedure, states at the bottom:
" max. deviation = 1 0.16 mA @ 1.0 %."
This is another in-process criterion, amounting to 1% of span in this case.)
+
'i The calibration procedure also. states "look for Pin 6 repeat-l ability from the original Pin 6 data," but does not specify any j
further action with regard to the Pin 6 data. _The-4-20.mA sheets observed by the inspectors showed that, on the-final check, if i
the Pin 6 voltages did not agree.with the values' corresponding to the desired flow for a given output within well under 1%, the flow was adjusted until close agreement was reached.
With some exceptions, manufacturer's records and data for the i
standard flowmeters were not provided to the inspectors.
At least three manufacturers were represented, one of the meters was 4
i l
\\
F O
W w
. described as custom, and the manufacturer of one was not evident by observation.
Two open loop test stands, A and B, have normally been used for safety-related flowmeters.
The inspectors noted that a 1989 safety-related PO used Stand CL.
It contained an anemometer as a transfer standard, as did Stand A for a 1988 safety-related calibration, in lieu of the presently-used turbine flowmeters.
(2)
The transfer standard turbine flowmeters are individually calibrated at multiple flow values against sonic nozzle traveling standards (low flows) or against traveling standard 4" turbine flowmeter FM-46 (insufficient air pressure is available to drive sonic nozzles in 4" pipes).
A computer th, fits a curve to the specific calibration data points for each transfer standard.
When calibrating delivered flowmeters per step (1) above, it is the values from the curve that are used as " actual" flow values.
FCI did not consider possible changes with time (drift) of the turbine flowmeters in the calibration process; this concern is addressed in Section 3.5(3) of this inspection report.
(3)
For low flow rates, four sonic nozzles (FM-81,
-82,
-83, and
-84) served as traveling standards used to calibrate the smaller turbine flowmeters installed in the test stands.
They were
' calibrated against other sonic nozzles at Flow Dynamics, Inc. in Scottsdale, Arizona, in October 1991, with an estimated error of 10.25 percent against the NIST standard.
The curves fit the data points within 0.17% of reading or better, whereas the inspectors observed errors of as much as 0.4% of reading in the turbine flowmeter curve fits.
The NRC inspectors did not observe any data reflecting drift of the sonic nozzles with time, and FCI did not include drift error in accuracy determinations.
The inspectors did not investigate calibratio-cf the transfer standards used for low flows prior to use of the sonic nozzles.
(4)
For higher flow rates, 4" turbine flowmeter FM-46 served as a traveling standard for the turbine flowmeters installed in the test stands.
It was calibrated annually, beginning in 1989, at the Colorado Engineering Experimental Station, Inc. in Nunn, Colorado, with an error estimated by the laboratory each time at 10.5% of raading vs. the NIST standard.
As for the other standard flowmeters, FCI did not include dr.ft errors in determining accuracy.
i 3.4 Calibration Error Sources The NRC inspectors had numerous concerns 1ith the calibration of FCI type LT81A air flowneters for nuclear safety-related service.
These concerns generally related to the fai.ure to issue, and follow, adequate calibration procedures.
Specific concerns iden-tifica by the inspectors during the inspection, in reviewing cer-tain calibration records after the inspection, and in telephone 5
O O
discussions with the FCI QA manager from May 10-26, 1993, are
' addressed in this section (calibration error sources) and'the next section (additional concerns) of this inspection report.
i FCI personnel stated that licensee inspectors generally did not l
look beyond the simple calibration of a delivered flowmeter against the transfer standard.
The NRC inspectors also observed that a recent audit of FCI by an industry group team did not i
raise the concerns addressed in this inspection report.
i The NRC inspectors found no evidence that FCI had identified.or I
combined all of the various calibration errors to determine the absolute acraracy of delivered flovmeters.
The failure of FCI to adequately consider all error sources in calibrating safety-related flowmeters constitutes a portion of Nonconformance i
99901264/93-01-01.
The NRC inspectors tabulated the identifiable quantified error sources involved in calibrating two flowmeters on Stand B on March 10, 1992, under FCI S.O.
32263.
These flowmeters were type LT81A, drawing 88-138561, Revision A.
The customer was Virginia Electric Power Company (VEPCO) under PO SSY-368340, dated December 3, 1991, and the FCI sales representative was United Control Co. of Richmond, Va.
The invoice stated that the shipping date was March 11, 1992.
For flowmeter serial number l
3680-1 (1-10 ft/sec range) the calibration-path to NIST with the.
largest identified errors was through turbine meter FM-78 and sonic nozzle FM-84.
The sum of the absolute values of the quantifiable errors in this path was about 1.4% of full scale.
For serial 3681-1 (8.61-51.66 ft/sec) the calibration path to NIST with the largest identified errors was through turbine meters FM-63 and FM-46.
The absolute sum of the errors for this path was about 2.0% of full scale.
The NRC inspectors also estimated the quantified errors in calibrating four type LT81A flowmeters on Stand A on 9/27/91 for VEPCO under FCI S.O.
29907, shipped on September 30, 1991.
Two flowmeters were 1-10 ft/sec units (serial numbers 3491-1 and 3492-1), and two werc 8.61-51.66 ft/sec units (serial numbers 3493-1 and 3494 '.).
The absolute sum of the errors estimated for each of the four totalled about 2.0 percent of full scale.
The inspectors considered the following error sources in these tabulations:
output voltage and Pin 6 voltage deviations during the final check calibration, curve fit, and drift errors for turbine meter standards calibration and curve fit errors for sonic nozzle standards 6
l J
O O
U V
.These are the only error sources for which the inspectors.could
[
obtain values.
The following additional error sources could not i
be quantified.
(1)
The practice of making circuit adjustments against Pin 6 i
voltages rather than standard turbine' meter readings, during the final check of the LT81A flowmeter, makes its calibration subject to the repeatability error of the sensing head, which is in effect used as a transfer standard for the final check.
The final check then becomes only a check of the circuit adjustments, l
rather than of the entire flovmeter, and that check is influenced by the repeatability of the sensing head.
The inspectors had no basis for estimating the drift error, although the FCI catalog and one of the assembly drawings stated a repeatability of 1% of full scale for the entire flowmeter.
i (2)
The inspectors found no data for estimating the drift of the sonic nozzle standards.
(3)
The inspectors observed range gaps in the calibration of flowmeters against standards for the specific cases discussed above.
[This matter is discussed from the procedural standpoint in Section 3.5(2) of this inspection report.)
Specifically, in Stand B, there was a gap in the calibration of standard turbine beter FM-78 against sonic nozzles FM-83 and FM-84, from 21.4 to 24 ft/sec, and a gap in the calibration of standard turbine meter FM-63 vs. sonic-nozzle FM-84 and turbine meter FM-46, from 114 to 144 ft/sec.
In Stand A, there was a gap in the calibration of standard turbine meter FM-47 from 4.35 to 5.65 ft/sec against conic nozzles FM-56 and FM-57, and a gap in calibrating the delivered flowmeters serial numbers 3493-1 and 3494-1 against turbine meters FM-47 and FM-59 from 17.2 to 26.7 ft/sec (more than 20% of their calibrated range).
For stand B, in each case one of the ten calibration points, but none of the final check points, fell in the gap.
For Stand A, in each case two calibration points and one final check point fell in the gap.
The inspectors did not attempt to estimate the errors resulting when calibration or final check points fell within uncalibrated gaps.
Where standards overlapped, the inspectors included the larger of the two errors in the estimates given above.
(4)
Data sheet entries discussed in Section 3.5(4) of this inspection report suggest that the air flow used in open loop test stands was not pressure-corrected, so that necessary density corrections were not made and additional error was introduced into the calibration.
(5)
For flowmeter serial 3680-1, neither the top of range or bottom of range point was included in the final check, as discussed in Section 3.5(9) of this inspection report.
The effect of this omission was not estimated, but it could be 7
r
(l V
.significant.
Slightly less serious was the omission of the bottom of range points for serial numbers 3491-1, 3492-1, and l
3493-1.
3.5 Additional Inspector Concerns Recardina Calibration In addition to the sources of experimental error discussed in Section 3.4 above, the inspectors noted the following additional anomalies and inconsistencies in the flowmeter calibration process.
The failure of FCI to prevent or correct these deficiencies in calibrating safety-related flovmeters constitutes a portion of Nonconformance 99901264/93-01-03.
(1)
The "In-House Certificate of Calibration" data sheets.used to document calibration of the transfer standard flowmeters did not provide a space for identifying the calibration test stand, although "A Stand" or "B Stand" was sometimes written on the sheet in the title block area.
The " Test Department Calibration Data:
Actual Test Conditions" form used to document calibration of delivered flowmeters did provide a space for identifying the stand, and provided for identifying the decade resistance boxes used for calibration, but did not identify the standard '.1rbine flowmeters.
The sheet titled "4 to 20 mA Calibration Ta,ie,"
which contains the results of the final five-point calibration bheck, identified neither stand nor standard.
In a few instances, in response to questioning, FCI produced a
" Final Acceptance Test Procedure" form that sometimes_gave additional information.
This form, although listed on the Shop Order front sheet, is not listed on the " Pre-Flight" sheet for inclusion in the flowmeter calibration files, and was, in fact, not included in any of the calibration files reviewed by the NRC inspectors.
The NRC inspectors noted that the transfer standards (but not the test stands) used for calibrating nonsafetv-related flowmeters were identified on " Certificates of Calibration,"
where traceability to NIST was specified; a commercial grade procurement by a licensee was so documented.
However, the
" Certificates of Calibration" observed by the inspectors for safety-related POs did not identify the transfer standards.
FCI personnel stated that the transfer standard flowmeters installed in the test stands were not changed.
However, the seven POs reviewed by the NRC inspectors contained four instances of transfer standard identification discrepancies that involved use of different transfer standards, including use of a different type of standard and an unidentifiable standard.
The 4" standard turbine meter apparently used in Stand A for the example discussed in Section 3.4 above was FM-59.
According to its rscords, FM-59 was previously calibrated on 8
w w
October 11, 1990; April 16, 1991; and September 11, 1991.
The April 16, 1991, "In-house Certificate of Calibration" sheet indicated that FM-59 was in Stand B, while the other two calibration sheets designate Stand A.
FCI stated that the discrepancy was due to a clerical error in entering information in the title block area of the calibration sheet, but could offer no evidence in support of this assertion.
The " Equipment Used" block of the "In-House Certificate of Calibration Sheet" for standard turbine meter FM-47 dated September 10, 1991, showed sonic nozzles FM-55 and FM-56 as the traveling standards used to calibrate it, while the data tables on the same sheet showed FM-56 and FM-57.
Stand CL was used to calibrate six flowmeters under S.O.s 17586 and 17596, Georgia Power Co. PCs P-50658 and P-50659, on March 9 and 16, 1989.
For each of these flowmeters the calibration data sheet states that the transfer standard was a Davis anemometer, FCI tag no. ELO10.
However, FCI's records show that ELO10 is a digital voltmeter.
FCI personnel were unable to identify the transfer standard during the inspection.
On August 29, 1988, FCI calibrated two type LT81A flowntcers on S.O.
13856, shipped on August 30, 1988, under VEPCO PO SSY-178975 dated March 28, 1988, which imposed the requirements of 10 CFR Part 50, Appendix B, and invoked 10 CFR Part 21.
However, the August 29, 1988, calibration date for these delivered instruments preceded the earliest calibration records for the turbine meter transfer standards examined by the inspectors.
Upon questioning, FCI personnel produced a " Final Acceptance Test Procedure" sheet for the S.O.
that listed the standard as "EL-74."
FCI's commercial grade calibration records also showed that flowmeter EL-74 was used in Stand A around August 1988.
The calibration records for EL-74 identified it as a 0-80 ft/sec Davis anemometer.
It was calibrated on August 9,
1988, using four "rotometers" [ sic), numbers FM-139,
-145, -146, and -147.
The required accuracy was listed as "1
.75 FPS (1% of range)."
However, EL-74 was only calibrated up to 25.04 ft/sec, whereas it was supposedly used to calibrate LT81A serial number 2791-1 up to 51.66 ft/sec.
Other discrepancies on the calibration data sheet were that it was not signed by the QA representative as required, nor were the as-received data, procedure number, and serial number recorded.
A previous calibration of EL-74 (February 8, 1988) only extended up to 36.51 ft/sec, and letters from the manufacturer indicated calibration only up to 34 ft/sec.
9 l
i I
y O
O l
'When questioned by the NRC inspector, FCI justified extension of EL-74's use beyond its calibrated range on the basis of " engineering R&D" testing performed on November 1, 1988 (two months after hardware shipment), when EL-74 was compared to sonic nozzles (of unspecified identity and I
accuracy) up to 70 ft/sec and its K-factor (analogous to that of a turbine meter) varied from 152.5 to 155 cycles per i
cubic foot.
These uncertainties as to which test stands and standard flowmeters were used for calibrating previously delivered flowmeters raise questions concerning the accuracy _of the.
calibrations.
Further, the NRC inspectors did not attempt to' i
identify or evaluate the accuracy of standard flowmeters other than for the specific samples selected.
(2)
As addressed in Section 3.4(3) of this inspection report, j
FCI's calibration of a flowmeter often requires the use of multiple standards, whose ranges may or may not overlap, each covering part of the range of the meter being calibrated.
The NRC inspectors found no evidence that FCI had evaluated whether the multiple transfer standards used to-cover the range of a delivered flowmeter covered the full range, with overlap and with
. acceptable accuracy.
i The inspectors also found no documentation addressing setup or use of a test stand that addressed the gaps or overlaps of.the standard flowmeter ranges.
Therefore, the origin of the data points for the so-called " actual flow" on the Test Department Calibration Data sheets in the gap regions was not clear.
The inspectors further noted that the gaps changed--and could appear and disappear--with the semi-annual or annual recalibrations of the standards, depending on the end points of the calibrations.
4 For calibration of transfer standard turbine flowmeters against traveling standards (sonic nozzles and turbines), the "In-house Certificate of Calibration" form shows which standard was used as the reference for each flow point.
For calibration of delivered flowmeters, neither the " Test Department Calibration Data" sheet used to record calibration data, nor the "4 to 20 mA Calibration Table" sheet used to record the final five-point check, nor any other sheet in the file, identifies which transfer standard meter was used for which flow point where the standards overlap.
The inspectors noted that the upper boundary of each gap was often formed by the low-flow end of a turbine flowmeter's calibration, where errors are usually largest, the calibration curves nonlinear, and extrapolation questionable.
The lower boundary was formed by the high-flow end of the calibration of either a turbine flowmeter--where overspeed is a concern--or a sonic nozzle--where an adequate supply of high-pressure air is necessary.
10
o O
O (3): The NRC inspectors found no evidence that FCI had-accounted
+
for the effect of transfer standard drift on the accuracy of delivered LT81A'flowmeters, between calibrations of the standards.
A transfer standard calibration was usually used for approximately six months (one year for the traveling standard),
or until a calibration shift was suspected.
If the next calibration differed significantly, bearing replacements and possibly other repairs were made, and new calibrations performed, without consideration for the LT81A flowmeters calibrated and shipped during that interval.
For the traveling transfer standard FM-46, the 1990 and 1991 calibration curves agreed very closely, but were consistently about
% higher than the 1989 curve and
% lower than the 1992 curve (except at very low flows, where the changes were larger).
All of the calibrations were performed at CEESI.
Upon review of the December 11, 1992, data FCI decided to replace the turbine bearings and have the meter recalibrated.
That effort was in progress during the inspection.
FCI personnel suggested that, since FM-46 was used less than the transfer standard turbine meters installed in the test stands, its bearings may have seated and worn more slowly.
Transfer standard flowmeter FM-47 was used in Stand A during calibration of the delivered flowmeters discussed in Section 3.4 above, based on its calibration on September 10, 1991.
The next recorded calibration of a 1.5" flowmeter in Stand A was on March 25, 1992, for FM-87.
On the next FM-47 calibration sheet dated October 26, 1992, the technician noted that it had been modified and out of service.
The inspectors could not determine at what point--before or after its presumed use for calibrating the LT81As--FM-47 developed a non-correctable error or~other condition that caused it to be replaced and modified.
- Also, there was no indication in the records that FCI evaluated the impact of FM-47's presumed failure on the accuracy of the delivered flowmeters that it was used to calibrate.
It is therefore possible that significant error could have been introduced into LT81A calibrations.
The failure of FCI to address discrepancy reports concerning transfer standard flowmeter drift in a timely manner is identified as a violation of 10 CFR Part 21 in Section 3.6 of this inspection report.
(4)
On the sheet titled " Virginia Power LT81 Equivalent Air Calibration Data 9 March 92" for flowmeter 3680-1, the loop test stand calibration pressure is stated as 0.00 psig.
For the " Test Department Calibration Data" sheets for flowmeters 3680-1 and 3681-1, "amb" is entered in the block titled " Pressure".
No pressure is recorded for the final check.
The purpose of the pressure measurement is to permit density-correcting the measured air flow rate to standard cubic feet per minute.
Gauge pressure 11
e
?
0 cannot'be used for this purpose, since it represents only the difference between the test loop pressure and the ambient atmospheric pressure at the time, neither of which was recorded.
This suggests that calibration corrections were not made for
- l pressure.
Even if corrections were made, the data sheets do not record the basis for the corrections.
(5)
The inspectors noted that the calibration process did not address response to variations in ambient conditions such as temperature, humidity, voltage, and frequency that should be included in accuracy determinations.
In this context the temperature concern is with error introduced by ambient air effects, most likely on the circuitry, rather than the process air temperature variation for which the LT81A flowmeters are temperature-compensated.
The inspectors did not investigate the treatment of harsh environment effects related to environmental qualification, which would not be related to normal environment calibration.
(6)
F'CI used Document # 008072, "LT81 Calibration Procedure, Board #0017 Rev.
B," Revision B, issued February 7, 1989, to calibrate delivered flowmeters.
This procedure was not signed or approved, it was not under Appendix B control, and there was no requirement in the Appendix B document hierarchy to use it or any other calibration procedure.
Furthermore, no procedures for calibration of the transfer standard flowmeters were found.
(7)
The calibration accuracy of delivered flowmeters was never recorded in the files reviewed by the inspectors, nor were accuracy calculations included.
(8)
The "4 to 20 mA Calibration Table" sheets used to record the final calibration check data for delivered flowmeters lack information such as technician identity (signature, stamp), date, identification of test stand and standard flowmeters.
(9)
The inspectors noted two significant discrepancies in the documentation for flowmeter serial number 3680-1 for VEPCO:
In the final calibration check on the "4 to 20 mA Calibration Table" sheet, the technician did not check the 4 mA output point (bottom of range) as typed on the sheet.
Near the title block of the " Test Department Calibration Data" sheet, "10.70" was entered as the range high, with the unsigned, undated notation " wrong, should be 10.07."
A similar notation was made near the 10.70 ft/sec entry in the calibration data table, and both the original calibration and final check Pin 6 voltages are crossed out.
A new and slightly lower Pin 6 voltage was entered next to the crossed out values (presumably the voltage measured for 10.07 ft/sec at the time of the notations), but there was no original 12
- ~ -
4 O
O W
' calibration voltage recorded to compare with the check value.
The original calibration evidently contained an error, which apparently was discovered during the final check, at which time the notations apparently were added.
The effect of these two errors is that two of the five points specified in the calibration procedure for final check, including both endpoints, were not checked for flowmeter 3680-1.
The 120%
over-range point at 12.09 ft/sec was also checked, although it was not covered by procedure or noted on the final check sheet and no deviation was recorded.
The only in-range points checked were at about 3.3, 5.4, and 7.8 ft/sec for the 1-10 ft/sec instrument.
(NOTE:
the difference between the 1-10 ft/sec range specified by the licensee and the 1.007-10.07 ft/sec calibration range reflects the difference between the licensee's specified 74*F normal operating temperature and FCI's 70*F standard temperature.)
(10)
The practice of making circuit adjustments against Pin 6 voltages rather than standard turbine meter readings during the final check of the LT81A flowmeter is not spelled out in the calibration procedure.
Further, the technician's entries appeared to deviate from the intended format of the "4-20 mA sheet" for the sheets observed by the inspectors.
The first
'c o l u m n, headed " Indicated Flow," was left blank.
The second and third columns, headed " Actual Flow" and " Signal Output,"
contained typed entries of the form "8.673 f/sec = 4.000 mA,"
with lines for entering " Indicated" and " Deviation" in the third column; it is in these spaces that output voltages were recorded.
The last two columns share the heading " Pin 6 Volts," but entries were made in only the first of these columns; those values were also entered in an unlabeled column on the " Test Department Calibration Data" sheet containing the original calibration data for the oroduct being calibrated.
Also on the "4-20 mA" sheets, the technician used " signal output" values of about \\ of those typed on the sheet (e.g.,
4.999 instead of 20 mA) without explanation.
The calibration procedure specifies " flow the unit at the 5 points indicated on the 4-20 sheet."
FCI personnel explained that the recorded values are voltages measured across a 250 ohm resistor, but the "4-20 sheet" continues to show mA.
The inspectors did not investigate the tolerance of the resistor or the error resulting from its use.
(11)
There was no documentation that either calibration laboratory used by FCI--Colorado Engineering Experimental Station, Inc. or Flow Dynamics, Inc.--was capable of performing safety-grade calibrations traceable to NIST; e.g.,
no evidence of a technical or QA audit by FCI or indication of an Appendix B program at either place.
13
~J
1 l
4 O
O
.(12}' The inspectors noted that three different units of flow were involved in calibrations, which complicates the overall error analysis.
Calibrations of the traveling standard turbine meter FM-46, and of the transfer standard turbine meters against any of the traveling standards, were expressed in units of K-factor (cycles per cubic foot) vs. output frequency?in Hertz; the readings from the turbine meters during calibration of the delivered flowmeters were expressed in standard cubic feet per minute (scfm); and the LT81A meters were calibrated in feet per second (ft/sec).
The calibration procedure (Document # 008072) did include formulas for computing flow rates and tables of conversion factors.
3.6 Mandlina of Discrepancy Reports The NRC inspectors reviewed FCI's Discrepancy Report (DR) logs for 1992 and 1993 and selected two DRs affecting the calibration
~
of turbine flowmeters installed in the A and B stands used to i
calibrate nuclear safety-related flowmeters:
DR 02726, dated October 4, 1992, and DR 02914,, dated February 15, 1993.
These DRs reported that two of the transfer standatd turbine flowmeters differed by as much as 7 to 16% of reading from the sonic nozzle traveling standards six months after their previous calibration.
FCI had not dispositioned either DR at the time of the
' inspection.
For three other DRs issued during the same time period that did not involve nuclear safety-related equipment, the inspectors found that one had been closed after notifying the customer, one had not been acted on, and for the third the customer had been notified and the DR not yet closed.
The NRC inspectors noted that Section 5 of FCI QA Procedure 704029, " Evaluation of Measuring and Test Equipment," Revision D, dated June 26, 1987, requires that if a piece of calibrating equipment is found to have been received by the calibrator in an out of tolerance condition, records shall be checked to determine if the equipment was used to perform any final acceptance tests; if so, a DR should be submitted to the Material Review Board for disposition.
QA Procedure 704004, " Discrepancy Report,"
Revision F, dated June 17, 1991, discusses the processing of DRs, but no required time frame is identified.
The NRC inspectors found no evidenc,e of review or evaluation of DRs 02726 and 02914.
On May 10, 1993, the NRC inspector asked FCI by telephone for the results of evaluations of the two nuclear-related DRs. The FCI QA manager stated that neither had been dispositioned, and initiated evaluations in response to the telephone call.
7 DR 02726 addressed turbine flowmeter FM-87, installed in calibra-tion Stand A.
In a May 13, 1993, telephone call FCI stated that Stand A was used for three nuclear safety-related Pos in the six-month period prior to issue of the DR (i.e., the interval when i
drift occurred after the previous semi-annual calibration was l
14 j
i i
e O
O performed).
Each PO involved type FR72-4 flow switches, rather than analog flowmeters.
The affected S.O.s are:
S.O.
35607, Tennessee Valley Authority, two units calibrated September 29, 1992 S.O.
34622, Tennessee Valley Authority, five units calibrated July 23, 1992 S.O.
35682, Northeast Utilities Corporation, five units calibrated October 5, 1992 In a May 26, 1993, telephone call the FCI QA manager stated that all of the subject flow switches were certified to 3% of full scale accuracy, and full scale was at least 50 ft/sec.
The maximum observed drift in FM-87 as reported in DR 02726 was 0.22 ft/sec, which is 7.6% of reading at about 3 ft/sec, but only about 0.44% of full scale or less for the delivered flow switches.
Thus, the DR focused on a bottom-of-range error that represented a large percentage of the actual reading, but only a small error in terms of the full scale accuracy certified for the delivered flow switches.
DR 02914 addressed turbine flowmeter FM-78, installed in Stand B.
FCI reported in the May 13, 1993, telephone call that no nuclear POs used Stand B in the six months prior to the date of the DR.
DR 02726 was written on October 4, 1992, and DR 02914 was written on February 15, 1993.
Each reported a potential deviation for basic components that FCI should have evaluated for reportability.
10 CFR 21.21 requires evaluating deviations within at most 60 days of discovery, or providing an interim report.
FCI did not begin evaluation until telephoned by the NRC inspector on May 10, 1993.
Even though subsequent evaluation by FCI reportedly showed that the concerns raised in these DRs did not necessarily violate the certified accuracy specification of delivered flow switches, the failure to evaluate DRs 02726 and 02914 in a timely manner constitutes Violation 99901264/93-01-01.
3.7 10 CFR Part 21 Program The NRC inspectors reviewed FCI QA Procedure 704011, "10CFR21 Reporting of Defects and Non-Conformances," Revision B, dated October 31, 1988.
This was FCI's current procedure for reporting defects and noncompliances pursuant to 10 CFR Part 21.
The procedure did not contain the time limits for notification or the requirements for an interim report that have been added to 10 CFR Part 21 since 1988.
The inspector pointed out to FCI that the term "Non-Conformances" used in the title of Procedure 704011 has no meaning in the context of the language of Part 21.
The title of 10 CFR Part 21 I
l 15 l
J
,'O j]
g V
W is " Reporting of Defects and Noncompliance," and the term
" noncompliance" is defined in the regulation.
The inspector also found that the reporting requirements of both the earlier and the current versions of 10 CFR 21.21 were improperly addressed in Procedure 704011.
Paragraph 3.1 stated the responsibility of an FCI employee to notify the QA manager if the employee learned that a basic component supplied by FCI "contains a defect or fails to comply with 10CFR21."
This requirement confused deviations from technical specifications, which an employee may discover, with defects, which are deviations that could create substantial safety hazards.
The procedure effectively established such a high threshold for employee reporting that it precluded any reporting by employees.
Conversely, employees could identify deviations and are required to report them by 10 CFR Part 21, but not by the'FCI procedure.
Paragraph 3.1 of QA Procedure 704011 also improperly restricted the ongoing notification and evaluation requirements of 10 CFR 21.21(a) (3) (1), which addresses f ailures to comply with the Atomic Energy Act of 1954, as amended, or any applicable rule, regulation, order, or license of the Commission relating to a substantial safety hazard, and not just with Part 21 as stated in the FCI procedure.
Paragraph 4.1.c of QA Procedure 704011 required employees to submit a written report to the QA manager to define the nature of the defect or failure to comply, and the safety hazard which was or could be created.
The QA manager agreed that FCI employees would seldom if ever be able to make such determinations.
This requirement indicated further lack of understanding of NRC requirements relating to defects and safety hazards.
Paragraph 4.2 of QA Procedure 704011 stated that the QA manager would determine if the defect or failure to comply is a
" reportable incident."
In fact, 10 CFR 21.21(c) requires notifying the NRC in case of a failure to comply or defect affecting a basic component.
The procedure gave no further definition of what constitutes a so-called reportable incident.
Finally, QA Procedure 704011 did not reflect the provisions required by 10 CFR 21.21 to ensure that all affected licensees or purchasers are informed of deviations that FCI determines it cannot evaluate.
The procedure did not address this situation, which the FCI QA manager conceded is the most likely case for a deviation.
Based on these deficiencies, the NRC inspectors concluded that QA Procedure 704011 did not ensure that deviations will be evaluated, that defects or failures to comply will be reported to the responsible officer, or that all affected purchasers or
)
licensees will be informed of deviations when FCI cannot perform 16 l
i
e O
o W
W
.the evaluation.
These deficiencies, together with the failure to incorporate new requirements of Part 21 and to post the current revision of Part 21--the issue effective October 29, 1991, was posted--constitute Violation 99901264/93-01-02.
i 4
PERSONNEL CONTACTED FCI:
+
M.M. McQueen, President
+
R.A.
Deane, Treasurer
+
R.E.
Ogle, Director of Administration
+
A.D.
Johnson, Director of Engineering
+
S.R.
Mitchell, QA Manager
+
M.
Bess, Test Engineering Manager
+
R.
Thorpe, Contracts Manager
+
W.
Franz, Consulting Engineer
+
Attended the entrance meeting on April 13, 1993 Attended the exit meeting on April 15, 1993 17
-