ML20134D782
| ML20134D782 | |
| Person / Time | |
|---|---|
| Site: | Davis Besse  |
| Issue date: | 05/08/1996 |
| From: | Donnellen R, Gale C, Hooten D CENTERIOR ENERGY |
| To: | |
| Shared Package | |
| ML20134D740 | List: |
| References | |
| NUDOCS 9702050367 | |
| Download: ML20134D782 (200) | |
Text
.
3
)
In~strument Drift Data Analysis Page Revision j
Methodologyand Assumptions 1 of 19 01 1
1
Title:
Instrument Drift Data Analysis Methodolonand Assumptions i
i Prepared by:
<ft 7
Y-26-76 Date Reviewed by:
CAW d
"/-2 6 - 76 Date l
l Approved by:
YlM/7 U
'~
Manager-Plant Engineering Date Approved by:
Ok -
5/P/74
[ rector-Engineering and Services bate l
4 i
i 9702050367 970130 PDR ADOCK 05000346 P
Instrument Drift Data Analysis Page Revision Methodology and Assumptions 2 of to 01 4
Table Of Contents e.
EASS f
Methodology and Assumptions 3
4 i
References 12 r
1 List of Affected Instrument Strings 14 Schedule 16 l
Standard Review Form 17 Instrument Drift Study Flowchart 19 I
i f
i 6
l a
i
.i N
g 4
d 4
4 3
4 1
Instrument Drift Data Analysis Page Revision Methodology and Assumptions a of is 01
- 1. Select the Technical Specifications (TS) section to be evaluated using the list of affected instrument strings and the schedule (copy of each for phase I attached).
- 2. Identify all redundant channels associated with that TS section and any identical instmment strings not associated with that TS section.
- 3. Obtain surveillance test procedures, data packages, drawings, etc. as 'needed to create a block diagram, showing all components in the instrument string used to perform the TS required -
function, for each redundant channel. Identify the surveillance test procedure used for each component in the block diagram
- 4. Verify that all individual components are identical (have the same make, model number, and range) to their counterparts in the redundant channels.
- a. If they are not all identical, then either the data from each channel has to be analyzed separately or a written justification must be prepared for combining the data from the non-identical channels.
- 5. Obtain historical as-found/as-left calibration data from Surveillance / Periodic Tests (STs/ pts), Maintenance Work Orders (MWOs), and I&C shop records. Enter applicable data into the spreadsheet.
Note: The data is entered into an Excel spreadsheet. The spreadsheet template contains the algorithms that determine the values used for this evaluation.
The general underlying algorithms and lookup tables were independently verified to be correct. This verification consisted of formula checks, checking basic statistics against an independent Lotus 123 version, and line by line verification of the lookup tables as well as checking against hand calculations.
If a field change was performed (e.g., a transmitter was changed to one of a different a.
make or model number), then only enter data obtained subsequent to the most recent field change for that group of redundant channels.
- b. If the channel performs an automatic protective action, then there will only be " trip" and/or " reset" data to record for each test. If the channel provides indication of a process variable (flow, pressure, temperature, etc.), then several (up to nine) different data points will be recorded for each test. Data taken at x% of span (increasing) and x% of span (decreasing) will be treated separately and not lumped together because they do not provide independent drift information when analyzed from one test to another.
Instrument Drift Data Analysis.
Pege Revision Methodology and Assumptions 4 of 1s 01
- c. String data (as opposed to component data) should be used in most cases.
This is because Davis-Besse's calibration procedures are generally stmetured such that if the string check satisfies all acceptance criteria, then the surveillance test is considered complete, an( no data is taken on individual components within the string.
.d.
Some channels aie functionally tested at more frequent intervals than once every 18 months. These tests provide greater amounts of data on all non-sensor components than do the calibrations performed on the entire channel each refueling outage. For this reason, these strings may be analyzed using calibration data for the sensor by itself and channel functional test data for the rest of the string. This requires a sufficient amount of data to be available for the sensor alone. Sensors of the same make, model number, and range that are used in other applications with similar operating environments may be used to provide additional data. This approach should be used with caution because data may only be available for some sensors for instances when they are out of tolerance, and this would skew the results (making them worse than the actual sensor performance).
- e. The percent (s) of span at which data was taken may have changed at some point (s) in time (e.g., data taken at 10, 30, 50, 70, and 90% of span previously is now taken at 5, 25, 50,75, and 95% of span). In these cases, a data column may be " shifted" to align with another data column that is within *10% of span in order to increase the sample size for that data point. If shifting of more than 10% of span is done, justification must be provided. When shifting data "up" use a ratio of percents of span (e.g., multiply drift values for 90% of span by (95/90) to obtain drift values for 95% of span). When shifting data "down" use a straight biasing approach (e.g., a drift value at 10% of span stays the same when shifted to 5% ofspan).
Note: After shifting data columns, as-found data from one test need not be compared with as-left data from another test when the two tests took data at different percents of span, however, it is acceptable to do so.
- 6. Calculate the drift that occurred between pairs of consecutive tests in either % span (typically) or process units (rarely). Segregate the results by data point for instrument strings providing process variable indication.
- 7. Calculate basic statistics for the drift values calculated in item 6 - sample mean (x), sample standard deviation (s), number of sample data points (n), 95/95% tolerance factor (k), and 95/95% toleranceinterval(x*ks).
- a. For instrument strings providing process variable indication there will be a set of basic statistics for each data point.
- b. For some instrument strings, measurement uncertainty in only one direction is of concern (e.g., reactor coolant flow measurement uncertainty for input to RPS is only of
instrument Drift Data Analysis Pepe Revision Methodology and Assumptions 5 of to ot concem if measured flow is greater than actual flow because ' the RPS power / imbalance / flow trip occurs on a decreasing flow signal, therefore a higher l
measured flow than actual flow delays the trip). For these single side ofinterest strings a i
one-sided tolerance factor may be used to determine the 95/95% tolerance interval.
Justification for use of a one-sided tolerance factor must be provided.
- 8. Identify any potential outliers among the data by performing the T-Test as described in j
ANSI / ASTM E178-1994, " Standard Practice for Dealing With Outlying Observations".
3 9, Send a copy of the spreadsheet, the block diagram, and a standard review form (copy attached) to the System Engineer,
- a. Resolve all comments resulting from the System Engineer's review.
- b. The standard review form must be signed by the System Engineer and kept with the file for that instrument string.
- 10. Utilizing the System Engineer's input, provide written justification for any outlying data '
point that is removed from the sample set. Unless clear evidence exists to demonstrate that the outlier is not representative of actual instmment drift (e.g., a failure or data entiy error occurred), the outlier should be retained in the sample set.
l
- 11. Recalculate the basic statistics if any spreadsheet data was changed as a result of either the j
System Engineer's review or the analysis of outliers.
- 12. Verify that the assumption of drift data being normally distributed is not unreasonable by performing the W test (for sample sizes less than or equal to 50) or the D' test (for sample i
sizes greater than 50), as described in ANSI N15.15-1974," Assessment of the Assumption of Normality (Employing Individual Observed Values)".
l
- a. When performing the W or D' test, use the 0.05 significance level. This means that if the sample data is randomly selected from a population that is normally distributed, there is less than a 5% probability that the test will reject the assumption of normality for that sample.
]
)
- b. To supplement the applicable normality test, create a histogram which plots number of drift data points versus the number of standard deviations from the mean. A group of l
" bins" will be defined, with each one including a range of values for number of standard
. deviations from the mean. Each bin will display two bars - one representing the actual
)
number of drift data points contained in that bin and the other representing the number of data points that bin would contain if the sample distribution were perfectly normal.
i i
Instrument Drift Data AnClysis Page Revision Methodology and Assumptions e of 1s of l
Comparison of the two bars in each bin provides additional evidence as to whether or not i
the driR data is normally distributed. The smaller the sample size the fewer the number of bins. For example, with only ten data points, the bins are:
j
- 1. More than 2 standard deviations below the mean
- 2. Between 2/3 and 2 standard deviations below the mean
- 3. Within.2/3 standard deviations of the mean
- 4. Between 2/3 and 2 standard deviations above the mean l
l S. More than 2 standard deviations above the mean l
- c. If the applicable test indicates that the assumption of normality should be rejected, then i
the histogram should be used to verify that the dria data is bounded by a normal distribution. This requires that ~95% or more of the driR data be contained within 2 standard deviations of the mean. To facilitate making this comparison, the bin sizes and l
locations should be chosen carefully to ensure that a bin " boundary" exists at exactly 2 standard deviations above and below the mean. (It is expected that all samples of driR data will be normal or bounded by the assumption of normality).
\\
\\
- d. For instmment strings which have data taken at multiple points, these normality l
assessment tools may be used on the sum of all data points and/or on individual data
)
points. At a minimum, the worst case individual point should be assessed.
L
- 13. Evaluate time dependency of the driR data. No single test or technique can be used to l
determine whether or not an instrument string's dria will increase with time. A variety of tools must be used to build a case for or against a given instrument string's dria being time dependent. If the results are inconclusive, then default to the assumption that some time
)
dependency exists. The various tools for performing this evaluation are described below.
)
l
- a. Plot driR vs. time since last test for all data points. Plots may also be made for individual l
data points, if applicable.
- b. Plot driR vs. time since last adjustment for all data points. Plots may also be made for individual points, if applicable.
DriR vs. time since last adjustment plots must be interpreted carefully. Although they can provide dria information for intervals longer than the normal calibration interval, that data could be misleading if the instrument string was regularly adjusted.
Adjustments " set the clock" back to " time =0", thereby preventing the creation of a possibly large driR value at a long time interval, i.e., only
" good data" avoids adjustment long enough to achieve a long time interval. Adjustments must be relatively rare to justify this approach to support a claim of time-independence.
)
l 1
l I
.-r= - - - -
.I y---
w
- ~. - - -. - _ -. ~.. -. -. -.. - -
. - -. ~.- -. --.;
i b
Instrument Drift Data Analysis Page Revision i
Methodology and Assumptions 7 of is 01 4
i
- c. The plots generated for items 13a and 13b can be redone using absolute value of drift j
instead of drift.
- d. The plots generated for items 13a and 13b can be used to evaluate sample mean and sample standard deviation at various calibration intervals. The data in a given plot is
, divided into several groups, each representing a range of calibration intervals. The mean and standard deviation are computed for each group and the results displayed in tabular I
form. 22.5 months is a helpful calibration interval to use as a boundary between groups since it is the maximum permitted by Tech Specs for instrument strings with a nominal i
}
calibration interval of18 months.
- e. Hypothesis testing can be used to help determine whether or not the variations in i
standard deviation observed among the different groups in item 13d are due to drift time-l 4
dependence. Sample standard deviations are used to evaluate the likelihood that two different samples were drawn from the same population. Results indicating that the same i
j' population produced all the samples constitute evidence that the drift is not time-j dependent. To implement this, use the F statistic for testing the equality of two 2
2 2
2 j
variances. The hypothesis to be tested is oi = o2, and the alternative is oi > e2, where o,2 is the variance for the range oflonger calibration intervals, and o22 is the variance for I
the range of shorter calibration intervals. The test statistic is si'/s2 and is an observed 2
i value of a random variable which has an F-distribution with (ni-1, n2-1) degrees of freedom, provided the hypothesis is true. The critical value is found from a table of l
l values of a random variable, z, which has an F-distribution with (m,n) degrees of
~
l freedom for which the distribution function, F(z), has the value 0.95. This corresponds j
to a significance level of 0.05, which means that there is a 5% probability that a true i
hypothesis will be rejected. If the test statistic is less than or equal to the critical value, l
then the hypothesis is not rejected, but if the test statistic is greater than the critical value, then the hypothesis is rejected. See references 4 and 23 for further discussion of the F
]
statistic.
- f. Regression analysis can be performed on the plots generated for items 13a and 13b to see-if a meaningful correlation exists. Based upon the results we've obtained on some actual drift data and those discussed in EPRI TR-103335, " Guidelines for Instrument l
t'alibration Extension / Reduction Programs", it is expected that regression analysis will j
rarely, if ever, show a significant correlation between drift and calibration interval. If the 2
2 i
R correlation is close to 1, then a meaningful drift rate can be calculated. If the R i
correlation is closer to 0, then the lack of correlation may be used as evidence that the drift is not time-dependent.
i l
- 14. Determine the projected 95/95% tolerance interval for the expected 30 month drift.
?
4 I
l
?
l s
k
instrument Drift Data Analysi$
Page Revision Metflodologyand Assumptions a of 19 01
- a. If the drift has been determined to be time-independent, th en the 95/95% tolerance interval previously calculated (see item 7 or item 11, as applicable) applies to a 30 month calibration interval.
- b. If the drift could not be demonstrated to be time-independent, then extrapolate each individual drift data point that was calculated for a calibration interval of less than 30 months to 30 months. If the distribution is clearly time dependent or there is insufficient evidence to assess the time dependency, use linear extrapolation and multiply each drift sincelast test data point by
(# months since last test 30 month:
- c. If the drift could not be demonstrated to be time-independent, but the degree of time dependence is less than linear, then extrapolate as in item 14b, but multiply the drift since 30 months 1/2 last test by (# months since last test Providejustification for using this method of extrapolation instead of the linear method.
- d. Some data sets may contain several as-found vs. as-left drift values with short test-to-test intervals. When extrapolated, these data points may cause the results to be overly conservative and unacceptable. In such cases, these drift data points with short test-to-test intervals may be deleted from the extrapolation process. The " months since last test" value selected as the threshold for drift data deletion must bejustified.
- e. For each new, extrapolated data set (which includes drift data points calculated for calibration intervals greater than or equal to 30 months in addition to the extrapolated drift data points), perform the following:
- 1. Calculate the basic statistics
- 2. Identify and analyze potential outliers
- 3. Recalculate the basic statistics if any outliers were deleted
- 4. Verify that normality is not an unreasonable assumption
- 15. Evaluate the results (i.e., the 30 month drift 95/95% tolerance interval) against the design basis for the instrument string.
For instrument strings that perform an automatic protective action this requires an a.
analysis of the calculation that establishes the setpoint. For instrument strings that provide process variable indication this requires verification that they can still be used to effect a safe plant shutdown. A calculation may not exist for these process variable indication instrument strings.
. _ _ ~.. _. _ _ _
Instrument Drift Data Analysis Page Revision Methodology and Assumptions e of 1s 01
- b.. Since as-found vs. as-left calibration data includes several sources of uncertainty m addition to "true drift", these uncertainty terms are contained in the 95/95% tolerance intervals obtained through the analysis of as-found vs. as-left calibration data. For this project only three terms will be used when comparing design basis information with these 95/95% tolerance intervals.
They are drift, reference accuracy, and M&TE uncertainty. Obtain values for these terms from the associated calculation, if one exists.
t If not, use either the equipment specification or vendor supplied information.
- c. Combine the uncertainties obtained in item 15b for all components in the string that as-found/as-left data was collected for. These terms must be combined in the same manner i
as in the calculation- (e.g., square-root-sum-of-the-squares, algebraic, etc. ). If no calculation applies, then combine them as would be done if a calculation were being created.
- d. Verify that the as-found minus as-left historical data has not exceeded the total uncertainty obtained in item 15e except on rare occasions. If applicable, also confirm that as-found data has not exceeded its Tech Spec Allowable Value on more than rare occasions. If this cannot be confirmed, then verify that corrective action has been taken to prevent future violations of the Allowable Value.
- e. If the total uncertainty obtained in item 15c bounds the 30 month drift 95/95% tolerance interval, then extension of the surveillance interval for calibration from 18 to 24 months is justified. Typically, for process variable indication instrument strings and automatic protective action instrument strings for which the sensor is analyzed separately (see item 5d), the worst case data point should be chosen to represent the drift characteristic (i.e...
to provide the 30 month drift 95/95% tolerance interval). In some cases, depending on j
the instrument string's function, a " critical" portion of the string's range may be identified that best represents the drift characteristic (e.g., those data points near a trip setpoint).
Written justification must be provided when utilizing this concept of the " critical" portion of a string's range. Also, caution must be used because a setpoint could be changed such that the " critical" portion no longer envelops the setpoint. Therefore, " critical" portions must be chosen so they cover reasonably expected changes in setpoints.
f.
If the 30 month drift 95/95% tolerance interval exceeds the total uncenainty obtained in item 15c, then further evaluation is required. When a calculation exists or needed to be created, send a Request For Assistance (RFA) to the supervisor of the unit responsible for the calculation. The RFA must provide the results of the instrument string's drift data analysis, including the comparison with design basis information. Ask the responsible organization to make whatever changes are necessary to the calculation so it adequately addresses the drift data analysis results. If they determine the calculation is acceptable as is, the basis for that conclusion must be documented thoroughly in the calculation. When l
J l
Instrument Drift Data Analysis Page Rzvisi:n Methodology and Assumptions to of 1s oi
)
i no calculation exists, contact Nuclear Engineering and Operations, providing the results of the drift data analysis and soliciting input regarding use of the subject instrument string (s) to effect a safe plant shutdown. Ask Nuclear Engineering to look at use of the instrument string (s) as described in the Updated Safety Analysis Report. Ask Operations to look at use of the instrument string (s) as directed by the Emergency Procedure (DB-OP-02000). The thrust of each organization's review should address the acceptability of i
the projected 30 month drift 95/95% tolerance interval for each decision or action that would be based on information provided by the instrument string (s). Use these responses and the expertise of the System Engineer to evaluate the acceptability of using instmment strings that provide control or indication only functions to safely shut down the plant.
- g. Review existing surveillance test procedures (channel checks, channel functional tests, and channel calibrations) for the instrument string (s) to verify that their acceptance criteria appropriately reflect all applicable conditions and assumptions of any associated setpoint and safety analyses. If an instrument string is not addressed in any existing setpoint or safety analysis, then this review is not applicable.
Note: Deficiencies found during this review should be brought to the drift study team's attention to be evaluated for possible initiation of a Potential Condition Adverse to Quality Report (PCAQR).
- 16. Prepare a written overall result summary that addresses the first six of the seven issues described in NRC Generic Letter 91-04, Enclosure 2.
Any calculation revisions resulting from item 15f do not have to be complete before a.
writing this result summary. Once the revisions are complete, those results will be incorporated into the appropriate License Amendment Request (LAR) submittal, and the overall result summary will be updated.
- b. All instrument string components not providing drift data (e.g., RTDs) must be qualitatively discussed as part of the justification for increasing the surveillance interval l
to 24 months.
- c. To provide consistency among the result summaries, the phrase "30 month projected drift" should be used to identify the drift study result for a particular instrument string, j
and the phrase " design basis / reference uncertainty" should be used to identify the appropriate combination of drift, reference accuracy, and M&TE uncerta'mty (see items 15b and 15c) to which the drift study result is compared (see items 15e and 15f).
l
- d. Each written overall result summary will be signed and dated by both the preparer and a reviewer.
.~,.
Instrument Drift Data Analysis Page Revision Methodology and Assumptions 11 of to oi i
- e. The seventh issue from NRC Generic Letter 91-04, Enclosure 2 requires us to " provide a summary description of the program for monitoring and assessing the effects ofincreased calibration surveillance intervals on instrument drift and its effect on safety". This is being addressed under a separate document.
.l l
,1 1
s 4
v j
4 t
4 4
1 9
4 l
. ~
a Instrument Drift Data Analysis Page Revision Methodology and Assumptions 12 of to ot Instrument Drift Study References 1.
Eisenhart, Ha' stay, and Wallis, Technigass ofStatistical Analysis, McGraw-Hill,1947.
2.
Lieberman, Gerald J., Tablesfor One-SidedStatistical Tolerance Limits, Industrial Quality Control, April 1958 5.
ANS1 N15.15-1974, Assessment of the Assumption ofNormality (EmployingIndividual Observed Values).
4.
Kreyszig, Erwin, AduancedEngineeringMathematics, Fourth Edition, Wiley,1979
- s. Odeh, R.E. and Owen, D.B., Tablesfor Normal Tolerance Limits, Sampling Plans, arul Screening, Marcel Dekker, Inc.,1980 6.
Beggs, William J., Statisticsfor Nuclear Engineers arulScientists, Part 1: Basic Statistical Inference, DOE Research and Development Report No. WAPD-TM-1292, February 1981.
7.
Regulatory Guide 1.97, instrumentationfor Light-Water-CooledNuclear Power Plants to Assess Plant andEnvirons Condition During andFollowing an Accident, Revision 3, May 1983.
- s. NUREGICR-5560, Aging ofNuclear Plant Resistance Temperature Detectors, June 1990.
9.
NRC Generic Letter 91-04, Changes in TechnicalSpecification Surveillance Intervals to Accommodate a 24-Month Fuel Cycle, dated April 2,1991.
- 10. Letter from R. P. Zimmerman, NRC, to H. B. Ray, Southern California Edison Company, dated April 12,1991.
11, Letter from D.F. Kirsch, NRC, to H.B. Ray, Southern California Edison Company, dated April 26,1991.
- 12. Letter from H. B. Ray, Southern California Edison Company, to NRC, dated May 21,1991.
- 13. Letter from R. P. Zimmerman, NRC, to H. B. Ray, Southern California Edison Company, dated June 14,1991.
- 14. Toledo Edison Memorandum NEN-91-10459 from R. C. Zyduck to J. H. Lash, dated October 23,1991.
Instrument Drift D:ta Analysis Page R:visiin Methodology and Assumptions 13 of ts 01 1s. Webb, R. C. and Beuchel, B. E., A Graded Approach to Setpoint Calculation Programs, Proceedings of the Thirty-fiRh Power Instrumentation Syraposium of the Instrument Society of America,1992.
- 16. Letter from T. G. Broughton, GPU Nuclear Corporation, to NRC, dated June 24,1992.
- 17. Letter from T. G. Broughton, GPU Nuclear Corporation, to NRC, dated May 28,1993.
- 18. Letter from R. W. Hernan, NRC, to T. G. Broughton, GPU Nuclear Corporation, dated June 23,1993.
I9. EPRI TR-103457, Non-ProcessInstrumentation Surveillance and Test Reduction, December 1993.
- 20. ANSIIASTM E 178-1994, StandardPracticefor Dealing with Outlying Obsen>ations.
- 21. IS A-567.04, Part I-l994, Setpointsfor Nuclear Safety-RelatedInstrumentation.
- 22. IS A-RP67.04, Part II-l994, Methodologiesfor the Determination ofSetpointsfor Nuclear Safety-RelatedInstrumentation.
- 23. NUREG-1475, Applying Statistics, February 1994.
l i
- 25. EPRI TR-103099, Effects ofResistance Temperature Detector Aging on Cross-Calibration Techniques, June 1994.
1
- 26. Toledo Edison Memorandum NED-95-40001 from J. H. Lash to R. C. Zyduck, dated January
)
18,1995.
- 27. Letter from G. L. Boldt, Florida Power Corporation, to NRC, dated May 31,1995.
i
' i
'p
$5 I
- 2 i
Tech Spec Instrument Drift Study /24-Month Fuel Cycle fg
- 5. 3 E
I List of Affected Instrument Stringst (Note Only first component in string is listed.)
<V eg Phase I (Must do to implement 24-month fuel cycle.)
Make/Model Cal Test /PM U
M*
Instru 0/Name Tech Spy 5
FT RCIA1 (RCS Flowl 4.3.1.1.1. Table 4.3-1, Item 4, Note 7 Rosemount 1153HD6 MI 3061 TT RCIA2 (MCS Flow) 4.3.1.1.1, Table 4.3-1, Item 4, Note 7 Rosemount 1153HD6
'MI 3062 3 )3 i
C a
FT RCIA3 (RCS Flow) 4.3.1.1.1, Table 4.3-1, Item 4, Note 7 Rosemount 1153HD6 MI 3063 FT RCIA4 (RCS Flow) 4.3.1.1.1, Table 4.3-1, Item 4, Note 7 Rosemount 1153HD6 MI 3064 FT RCIB1 (MCS Flow) 4.3.1.1.1. Table 4.3-1, Item 4, Note 7 Rosemount 1153HD6 MI 3065 3 E.
FT RC182 (PCS Flowl 4.3.1.1.1, Table 4.3-1. Item 4, Note 7 Rosemount 1153HD6 MI 3066 gj PT RC1B3 (RCS Flowl 4.3.1.1.1, Table 4.3-1, Item 4, Note 7 Rosemount 1153HD6 MI 3067 3 f,"
48 f7 BC184 (MCS Flow) 4.3.1.1.1, Table 4.3 1. Item 4, Note 7 Rosemount 1153HD6 MI 3069 r
Lit 4617 (CTNT sump level) 4.3.3.6, Table 4.3-10, Item 15s 4.4.6.1.b Transamerica Delaval TD/RE-36562 MI 3722 LIT 4619 (CTMT sump level) 4.3.3.6, Table 4.3-10, Item 15s 4.4.6.1.b Transamerica Delaval RM-54852-48-2700
.MI 3721 l
LT 4594 (CINT wtr level) 4.3.3.6, Table 4.3-10, Item 16 Rosemount 1153AD7 MI 3726 -
LT 4595 (CTMT wtr level) 4.3.3.6, Table 4.3-10, Item 16 Rosemount 1153AD1 MI 3725 LT 5448A (RCS hot leg level) 4.3.3.6, Table 4.3-10, Item 19 Rosemount 1153HD6 MI 3712
(
I.T 54488 (RCS hot leg level) 4.3.3.6, Table 4.3-10, Item 19 Rosemount 1153HD6 MI 3711 LT RC14-1 (pressurizer levell 4.3.3.5.1, Table 4.3-6, Item 4s 4.3.3.6, Rosemount 1153HD5 MI 3640 Table 4.3-10, Item 4 LT RC24-3 (pressurirer levell 4.3.3.5.1. Table 4.3-6, Item 4: 4.3.3.6, Rosemount 1153HD5 MI 3641 5
Table 4.3-10, Item 4 LTSP9A3 (SG 1evell 4.3.3.5.1. Table 4.3-6, Item 6: 4.3.3.6, Rosemount 1153005 MI 3653, MI 3658 Table 4.3-10 Item 5: 4.7.1.2.1.d I-LTSP9A4 (SG 1evel) 4.7.1.2.1.d Rosemount 1153DD5 MI 3655, MI 3660 l
i l
LTSPSA6 (SG 1evel) 4.3.2.2.1, Table e.3-11, Item Ib, 4.3.3.6, Rosemount 1152DP5 MI 3237 I
Table 4.3-10, Item 5 3
LTSP9A7 (SG 1evell 4.3.2.2.1 Table 4.3-11, Item Ib Rosemount 1152DP5 MI 3239 LTSP9A3 (SG 1evell 4.3.2.2.1 Table 4.3-11, Item 1b Rosemount 1152DP5 MI 3239 i
LTSP9A9 (SG 1evell 4.3.2.2.1, Table 4.3-11, Item ib Rosemount 1152DP5 MI 3240 LTSP933 (SG 1evell 4.3.3.5.1, Table 4.3-6, Item 6s 4.3.3.6, possmount 1153DD5 MI 3654, MI 3659 i
LTSP934 (SG 1evell 4.7.1.2.1.d Rosemount 1153DD5 MI 3651, M13661
{
"f[
f Table 4.3-10 Item 5: 4.7.1.2.1.d
, 3 1
LTSPSB6 (SG 1evel) 4.3.2.2.1, Table 4.3-11, Item ib Rosemount 1152DP5 MI 3241 i
4.3.3.6 Table 4.3-10, Item 5 LTSPSB7 (SG 1evel) 4.3.2.2.1, Table 4.3-11, Item Ib Rosemount 1152DP5 M1 3242
{
LTSP986 (SG 1evel) 4.3.2.2.1, Table 4.3-11. Item Ib Rosemount 1152DP5 MI 3243
- i LTSP989 (SG 1evell 4.3.2.2.1, Table 4.3-11, Item Ib Rosemount 1152DP5 MI 3244 I
PSH RC2J4 (RCS pressure) 4.5.2.d.1 SOR STA-84-NK-C1A-JJTTK6 PM 2655, SP 3130 PSL 106A (AFFT inlet St press) 4.7.1.2.2 SOR 6TA-84-NK-C1A-JJTTK12 MI 3903 PSL 1068 (AFFT inlet st press) 4.7.1.2.2 SOR 6TA-84-NK-C1A-JJTTK12 MI 3903 PSL 106C (AFPT inlet at press) 4.7.1.2.2 SOR 6TA-B4-NK-CIA-JJTTK12 MI 3903 PSL 106D (AFFT inlet at press) 4.7.1.2.2 SOR 6TA-84-NK-CIA-JJTTK12 MI 3903 PSL 107A (AFFT inlet st press) 4.7.1.2.2 SOR 6TA-84-WK-CIA-JJTTK12 M1 3906
(
PSL 1978 (AFPT inlet st press!
4.7.1.2.2 SOR STA-84-NK-C1A-JJTTK12 MI 3906 i
PSL 107C (AFFT inlet St press) 4.7.1.2.2 SOR 6TA-84-NK-CIA-JJTTK12 MI 3906 PSL 1070 (AFFT inlet at press) 4.7.1.2.2 SOR 6TA-94-NK-C1A-JJTTK12 MI 3906 3
e$.
?
" g-o s
L
m -. m._.
.u
.m._.
=- -
4_.m..
.m
~.
... - ~
- --.- ~ ___ -
m.
..m..
m _... _ _ -. _.. - -. _ _.
t 55 i
7 Instru 8/Name Tech Spec Mate /Model Cal Test /PM C
k3 PSL 4926A (AFP suction press) 4.7.1.2.1.e SOR 12V2-E4-M2-CIA-LLTTK3 MI 3901 PSL 4528B (AFP suction press) 4.7.1.2.1.e SOR 12V7-E4-N4-BIA-TTLLX3 MI 3901 Oe PSL 4f29A (AFP suction press) 4.7.1.2.1.e SOR 12V2-E4-M2-CIA-LLTTK3 MI 3904 O3
~
PSL 4925B (AFP suction press) 4.7.1.2.1.e SOR 12V2-E4-M2-CIA-LLTTK3 MI 3904
$g PSL 4930A (AFP suction press) 4.7.1.2.1.e SOR 12V2-E4-M2-CIA-LLTTK3 MI 3902 PSL 49303 (AFP suction press) 4.7.1.2.1.e SOR 12V2-E4-M2-CIA-LLTTK3 MI 3902 PSL 4931A (AFP suction press) 4.7.1.2.1.e SOR 12V2-E4-M2-CIA-LLTTK3 MI 3905 CLyh PSL 4931B (AFP suction press) 4.7.1.2 1.e SOR 12V2-E4-M2-C1A-LLTTK3 MI 3905 PT 4537 (CTMT pressure) 4.3.3.6, Table 4.3-10, Items 16 s 17 Rosemount 1153AD7 MI 3723 es g PT 4583 (CTMT pressure) 4.3.3.6 Table 4.3-10, Items 16 6 17 Rosemount 1153AD7 MI 3724 1
E >
5 PT 6365A (RCS pressure) 4.3.3.5.1. Table 4.3-6, Item 3: 4.3.3.6, Rosemount 1154GP9 MI 3733 Table 4.3-10, Item 3 l
PT 63553 (RCS pressure) 4.3.3.5.1 Table 4.3-6, Item 38 4.3.3.6, Rosemount 1154GP9 MI 3734 e, y Table 4.3-10, Item 3 O es PT RC2A1 (RCS pressure) 4.3.1.1.1 Table 4.3-1, Items 5,6,7,14 Rosemount 1152GP9 MI 3054 g et PT PC2A2 (prs pressure) 4.3.1.1.1 Table 4.3-1 Items 5,6,7,14 4.4.3 Rosemount 1152GP9 MI 3052, MI 3742 PT PC2A3 (RCS pressure) 4.3.2.1.1. Table 4-3.2. Items id, le, Foxboro NE11GH-IIM2 MI 3134 Sa a Sbs 4.5.2.d.1 PT PC2A4 (RCS pressure) 4.3.2.1.1, Table 4-3.2 Items Id, le Foxboro NE11GH-IIM2 MI 3132, MI 3102 4.3.3.6, Table 4.3-10. Items 3, 10, and 19 PT RC2B1 (RCS pressure) 4.3.1.1.1, Table 4.3-1, Items 5,6,7,14 Rosemount 1152GP9 MI 3053 PT RC232 (RCS pressure) 4.3.1.1.1, Table 4.3-1 Items 5,6,7,14:
Posemount 1152GP9
- MI 3051, MI 3742 4.4.3 PT RC2B3 (RCS pressure) 4.3.2.1.1, Table 4-3.2, items 1d, le, $b Foxboro NC11GH-IIM2 MI 3133 l
l PT RC254 (RCS pressure) 4.3.2.1.1, Table 4-3.2, Items 1d, le, 5b Foxboro NE11GH-IIM2 MI 3131 MI 3701 l
+
4.3.3.6, Table 4.3-10 Items 3, 10, and 19: 4.5.2.d.1 PT SP12A2 (SG outlet st press) 4.3.3.5.1, Table 4.3-6, Item 5:
Foxboro NE11GM-HIE 2 MI 3651
[
4.3.3.6, Table 4.3-10, Item 1 PT SP1281 (SG outlet st press) 4.3.3.5.1, Table 4.3-6, Item 5s 4.3.3.6, Foxboro NE11Gt-HIE 2 MI 3652 Table 4.3-10, Item 1 PI 4556A (CTMT radiation) 4.3.3.6 Table 4.3-10, Item 6a General Atomics RD-23 (0360-2062-01 Rev D)
MI 3407 KE 45953 (CTMT radiation) 4.3.3.6, Table 4.3-10, Item 6a General Atomics RD-23 (0360-2062-01 Rev D)
MI 3408 TE RC3A2 (RCS Temperature) 4.3.1.1.1, Table 4.3-1 Items 3 s ?
Rosemount 177HW MI 3054, SC 4111 TE RC3A4 (RCS Temperature) 4.3.1.1.1, T. ole 4. 3-1, Items 3 s 7 s 4.3. 3.5.1 Rosemount 177HW MI 3052, SC 4111 l
Table 4.3-6, Item 2 TE RC382 (RCS Temperature) 4.3.1.1.1, Table 4.3-1, Items 3 s 7s 4.3.3.5.1 Rosemount 177HW MI 3051, SC 4111 m
.g i
o a Table 4.3-6 Item 2 TE RC354 (RCS Temperature) 4.1.1.1.1, Table 4.3-1, Items 3 s 7 Rosemount 177W MI 3053, SC 4111 3S TTRC3A5 (RCS Temperature) 4.3.3.6 Table 4.3-10, Items 2, 10, s 19 Foxboro N-2AI-P2V MI 3701 e
TTRC3A6 (RCS Temperature) 4.3.3.6, Table 4.3-10, Items 2,10 Foxboro N-2AI-P2V MI 3702 TTRC3B5 (RCS Temperature) 4.3.3.6, Table 4.3-10, Items 2, 10, s 19 Foxboro N-2AI-P2V HI 3701 TTRC3B6 (RCS Temperature) 4.3.J.6, Table 4.3-10, Items 2, 10 Foxboro N-2Al-P2V MI 3702 TT5441B tacs hot leg lov temp ccep) 4.3.3.6, Table 4.3-10, Item 19 Alison Control A888-R103 MI 3712 TT54503 tacs het les lov temp ecup) 4.3.3.6. Table 4.3-10, Item 19 Alison Control A888-R103 MI 3711 ZT 4263 (PoRY position) 4.3.3.6, Table 4.3-10. Item 11 TEC 504A MI 3743 ZT 4264 (PORY position) 4.3.3.6, Table 4.3-10, Item 11 TEC 504A MI 3744 ZT 4265 (PRZR saf valve position) 4.3.3.6, Table 4.3-10, Item 13 TEC 504A MI 3743 ZT 4266 (PRIR saf valve position) 4.3.3.6, Table 4.3-10, Item 13 TEC 504A MI 3744 ZT 4267 (PRZR saf valve position) 4.3.3.6, Table 4.3-10, Item 13 TEC 504A MI 3743 IT 4268 (PRZR saf valve position) 4.3.3.6, Table 4.3-10. Item 13 TEC 504A MI 3744
?s.
h O-" :s F
.-__.-m, - - -,
e e.
v,.,
,.,.._,..e.....o m
!l, j
i1I
]!I l
,ll)ll l
1
'l l
j
,l
)
il 1
,lJ;{J
,Ii) 1 1li lll ll11 i
i e
taD e
uD IX n
1 0
is X
iv 2
R X
I il lV I
IV IV V
)
e V
9 s
a I
e1 h
gf P(
l a o l
e P
I 6
h 1
1 1
F y
d tu I
)
S e
i bl A
a r
c D
t g
s l) ilp
)
)
s e
s p
n e
e v
i a
r e
l P
s s
m to
)
n
)
)
a f
I r
r e
b i
)
n(
e o
e lbd ned u
e H
e r
a
)
ta n
ae gt e e
r r
)
r t
ts u
k u
l) S n
)
n i
t n
s c
s e
t R
r i
d l) ig icf d isi t
o c
i e e
e a
e s
i e
n r
o p
s I
s r
e t
n v
d ed lpe wu
)
l m
e v
t U
a i
I s n
)
e e pv tad d.
e w
r r
e l
)-
t d
e R
te u
a e
r l
)
P e
e P l
u y q c
)
c o
tn T
O S mt a u e
- t. r e i
y w
L, h l
I d
)
r a
S v
e nl t O
p e )e W dl
)
S R
el f a c r.
a r
)
e C
e pt pi o p
R r
)
o 4 is
)
ejwmn C C Z O O v e w
w e
o w R R P
S S
P m
l r
r msymd p o
mi
)
t (i
)
o e
l m
e r
ta u
r u
l k
t e
t r
e
)
l) o l
r F
ll a m la t
i s
v a
e u
)
s L
l m
a a a g
a T e
s o c s oe m.pe ui rd s
T e
v n
t i
ev S e
s e
u t
x c
u l.
c s
e s
v o
yct r
e a
m d
ta e
F a
l
)e e s
s v a u
. A ta t e
V n
m n n
n n
a r
e nnb s r
e l
R o
n la l
W P l
l o o c
p e
e i
r n n i
n m
r a
st r
t u
r u
s h
e F
ru u y
u u
oi r
l a
tm t
t t
t t
s e
h M
I V
m u
u e
a s
S g
te i idi l a.di lul s
r r
n o
o o
o a
l s
p A
s tn e
t t et a
- r u
s T w
tu u
s m S r
n n
)
c ci s v a s o
me x
e e
e s
s ta ta s r
o r
o o
O P l) n l
M O ty r R R s
s e
B s
s o
o lef y
v e o o
e e
i t
e a
i e
l r
k l l i i e I
s e
,h eri t
is s H P o ig e d
S S
R S D D D D D t
p p
v R C
P f
N W W M
T u
r t
r r a ia2 nsll l
,t r
n e
e l
H M M M M M l
o o
e T
S V S
T T T o
t
)
)
ta r
r o
e a
t e
c looet l
P P
u w
P P
l sl s r s
o c c v e
4 h
F h
s oh a u G
a m
a me mu s
w s
o ci k
l a
l r
U M S R R M M M H l
u l
a ta tail 2
ig S u
o o
l c
e r
L H P t
C C H Z S
S S
S S
S R
T C O Z T T T C l
ta c
tn DDDBONT3 ROC o
t a a a uoi0mes el e
s t r
r O
v a l
l S
/
/
/
/
/
u yn R
h R R D P v
O C C Z O C R P
P C C C R R
S e
I C
r o
C
(
C C C S
- e te e
e lai tn t
S R R P
S Z
l t
e t
R 7
R R R S
S S
S R
t I
u
+ s o
o M M M M M M M M P
o n
r s
o S A A A A F np m
m M M M M M A A A A A A A A d
e s
e S
e S
S S
P F
F F
F S
m e
e A A A A A P
P P
P P
P P
P n tr s
Am m P o P P
P R
S S
S S
(
e t
n r
e R R R
(
(
(
(
(
b R
R R
R R
R N,
P P
P P
P
(
(
(
(
(
(
(
(
a o
u r
u e
a 0 I
3 6
l)
C s
P
(
(
(
4 d
e a
b
(
(
(
(
(
a s r
(
1 2
3 4
5 6
1 I
1 51 1
71 91 v
v l
e t s i
3 4
5 6
7 1
i l
S I
2 3
4 5
6
(
(
(
(
(
e e
s u
m m
e r
aA qe m
m r
m m m m
m m n t
m m
m m
m m
m m
m m
m m
m m
m m
m m
s la v P a
I l I
l n
e D
s t
V e
n t
V VVVVI XX iX I I e
1 1
1 i,
te e
t t
t i
i t
te e
t t
t te e
t t
t te te e
t te t
t p
n o
I 1 i,
t te te e
e e
e e
e e
e e
d R
t I, I, 1
i, i, i, I, I, l, 1, i, i, I, i, i, i, i, i, i, i
i, i, i, i, i, i, i
i, m
i O
i S
I I 1 I t
t e
e e
e e
e e
t t
l t n o
t t
4 4 4 6 4 0 0
0 0
0, 0 0
0, 0 u
t S
u l
f 1
1 2
2 2
2 1
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
1 1
1 1
1 1
0 01 0 01 1
1 1
S la c
e i a 1
1 1
r n
T S
I Dy n
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
3 3
3 3
3 3
3 3
3 3
3 3 3 T
i F
F F
o F P
T
'v g
s e
e e
e l
l b
l l
b le e
e e
l l
l 4
4 4
4 4
4 4
4 4
4 4
4 4
M t
A A F e
e e
e e
e e
e e
t o
r l
b b
l l
b b
l b
b l
b l
l l
b b
b le le l
b b
b b
b le l
b b
b b
b b
b e
e e
e e
e
)
a a
a a
a a
a a
a a
a a
a a
a a
le le le l
l b
b l
le l
V T i
(
(
nl u
a a
a a
a a
a a
a a
a a
a R C D A
eo S
b b
b u
p 1
1 1
1 1
1 I
T, T, T, T, T, T. T, T, T, T, T, T. T, T, T, T, d.
e.
(
s
(
(
md 1
1 1
1 1
1 T, T. T, T. T. T, T, T. T, T, T, T, T, O b.
1 1
- l. 2 n
c e
e o
1 1
1 P
- l. d.
2 2
2 to rh S
1 1
1 1
1 1
1 I
1 1
2 5
5 5
5 5
6 6
6 6
6 6
6 6
6 6
6 6
6
(
s 1
l.
1 1
2 2
2 2
2 3
3 3
3 3
3 3
3 3
3 3
3 3 3
3 3
3 3 3
6 2
1 1
l.
l t t h
1 1
e se ce 3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 3
3 4
4 5
7 7
7 i
InM T
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 4
4 M
Instrument DriA Data Analysis Page Revision Methodology and Assumptions 17 or19 01 TO:
FROM: 24-Month Fuel Cycle Instrument Dria Study Team DATE:
SUBJECT:
Review and Verification ofData Collection Accuracy Attached is a copy of the data collected pertaining to the following:
Tech Spec Surveillance Requirement (s):
Functional Unit (s):
Applicable Sensor (s):
Calibration Docurnent(s):
l The data has been entered into a spreadsheet and formatted to calculate dria in percent span (usually) or process units (rarely) using the "as-found minus as-leR" approach. This data was obtained from STs/ pts, MWOs, and I&C shop records. A block diagram for each instrument string is also attached.
i Please review all the attached information, in accordance with the criteria sheet, making any comments on a document review form. Once all comments have been resolved, sign and date the statement below.
I have reviewed the attached block diagram (s) and spreadsheet information in accordance with the criteria sheet and have found them to be correct.
System Engineer Date DPH/ljk Attachment i
i
a hatrument DriA Data Analysis Page Rcvision Methodologyand Assumptions isat19 on Instrument Drift Study Data Review Ctiteria I. Verify that a block diagram for all redundant instrument strings is attached. Identify any identical strings not included. If any non-identical ones are included, verify that either the data is analyzed seperately or written justification for combining data from the non-identical strings is included.
- 2. Verify that each block diagram is correct, showing all components in the instrument string used to perform the Tech Spec required function, and showing the correct procedure (or PM) used to calibrate each component. Components not used to comply with the Tech Spec requirment need not be included in the block diagram
- 3. Verify completeness of the as-found/as-left calibration data. The drift study team reviewed STs/ pts, MWOs, and I&C shop record. At a minimum, please search the Nuclear Records Management database and the SPB to look for cases where calibration and maintenance activities were performed without the spreadsheet reflecting it. Consider the following during your review:
- a. If a field change (FCR, MOD, FPR, etc.) was performed on any component in the instrument string, then only data obtained subsequent to the most recent field change is relevant unless written justification for claiming the field change had no significant impact on drift characteristics of the afected component (s) is included.
- b. If an instrument string performs an automatic protective action, then only " trip" and/or " reset" data is recorded for each test. If an instrument string provides process variable indication, then several diferent data points are recorded for each test.
- c. If maintenance was performed on a component in the instrument string between complete string checks (e.g., a module in the Control Room was replaced, repaired, or adjusted with the plant on-line), then the spreadsheet will indicate "NA" for both as-found and as-left data and will indicate an adjustment was made. If a string check (including sensor) was performed as post maintenance testing, then the spreadsheet will indicate "NA" for as-found data, record as-left data, and indicate an adjustment was made. These practices prevent use of as-found data from the subsequent outage minus as-left data from the previousoutage as a meaningful measure of drift for that string (since tb difference reflects the efects of the maintenance performed in addition to the drift occurring over time).
- 4. Verify accuracy of the as-found/as-left calibration data by spot checking a small sample if spreadsheet data points. Enough points should be checked to enable signing the cover sheet statement with confidence. It is not necessary to check each and every piece of data. Each outlier (a data point significantly different in value from the rest of the sample) identified on the spreadsheet should be checked for data entry error. If the outlier is not a data entry error, then maintenance records should be reviewed to see if the outlier can be explained.
... ~
. - _ _.. ~. -
InstrumentDriAData Analysis Pay Revision Methodologyand Assumptions 19 or19 01 e
r sawTs Instrument Dnft Study 1
ai.m to be evaluated Flowchart i
1 idemify all muerument string to be 5
evaluated 1
Collect lusional AF/ALdatafreen shop records, j
MWOs,sTs, etc.
1 i
Canyute basic am.m m using g
Test toTest perosnt dria data (x, s 95S5%)
d' Compute basic t-owl.s,s statistia using l
Identify potentsal Rejected extrapolated outher' M data i
omtars J,
Rejected l
/
l c
~,
Have System Ensmeer review data Identify and A analyze outliers analyze potential
(
No embers
'"II'
i No ombers
~
Rejected VerifydriA datais normally absenbuted or Verifydria detais V
_a Normshty is en wwossansble nonnality is a n nnally distributed or To be deternuned answnpu (not IAety) conservative e normality is a later,only if cmative usinpon necessary i
/
Evai le
/r. Time>
Imensnant Dna may
)
be Tune Dependent I'
Insensna Dna a
t a ad Tune Extrapolate drin l
Depende data to 30mienths Evaluate protected l
30 month driR l
y against desip j
basis Revise desisn basis documents, as i
necessary i
Prepare Resuhs i
Summary Ashirussing GL9104 lasues
LAR 95-0024 ATTACHMENT 2 FOR LICENSE AMENDMENT REQUEST NUMBER 95-0024 (40 pages follow) i l
l l
(
i l
l
- .i j
1 RPS RC FLOW The following overall result summary of the drift study performed for the RPS differemial pressure j
transmitters and I/E converters is applicable to Technical Specification Surveillance Requirement l
4.3.1.1.1, Table 4.3-1, Item 4, Note 7 (RPS Flux - A Flux - Flow).
Block diagrams for these instrument strings are shown on Attachments I through 4. The differential pressure transmitters were analyzed separate from the rest of the string, as were the I/E converters.
These components are both subject to testing activities that will be performed less frequently after switching to a 24-month fuel cycle. The other components in the string - the square root extractor, j
buffer amplifier, function generator, and bistable modules - are tested quarterly with the plant on-l line. Since this test frequency is not affected by the increase in fuel cycle duration, these modules were not included in this drift study. They were, however, part of a previous drift study to support l
extending the on-line surveillance test interval from one to six-months. (See LAR 90-002, which resulted in License Amendment No.185.)
a Differential pressure transmitter data taken at approximately 25%, 50%,75%, and 95% of span, in both the increasing and decreasing directions was utilized in this study. Data at the low end of the range was not utilized, as it is not relevant because the RPS flux-A flux-flow trip is not needed to show protection for loss of both reactor coolant pumps in the same loop. (The RPS high j
flux / number of reactor coolant pumps on trip will automatically trip the reactor if both RCPs in the same loop are lost.) Since flow is proportional to the square root of differential pressure,25% of differential pressure span corresponds to 50% of flow span, which is adequate to cover the lower portion of the necessary range of the instrument string. Some minor data column " shifting," as described in item 5e of" Instrument Drift Data Analysis Methodology and Assumptions" (referred to as "the methodology document" henceforth), was performed because the static pressure effect on each transmitter was slightly different, resulting in calibration data being taken at percents of span that varied a little among transmitters.
All eight RPS differential pressure transmitters were replaced during 6RFO in response to the Rosemount transmitter loss of fill oil issue addressed by NRC Bulletin 90-01. The replacement 4
transmitters all have serial numbers above 500000, signifying that they were manufactured after July 11,1989. Since these transmitters were built using an improved manufacturing process, data from transmitters installed previously was not used in this drift study.
For all seven (after " shifting") test points, the sample mean (x) is less than 0.02% span, which can be considered to effectively be zero; The worst case 95/95% tolerance factor times sample standard deviation (k's)is 0.95% span (random error). A one-sided tolerance factor was used because the RPS flux-A flux -flow trip occurs on a decreasing flow signal, so its uncertainty is of concern only if measured flow exceeds actual flow. The supporting details for these results are shown on. These n esults were obtained after the removal of one outlier from the FTRCI A3 data.
The percentage drift between the as-left data on March 16,1993, and the as-found data on October 15,1994, was approximately -2% span for each test point and was affected by a sudden downward shift in transmitter output of about I mpph during the middle of 1993. Such a sudden shift in output
i 2
4 l
i is not representative of transmitter drift, but is reflective of some sort of abnormal transmitter l
behavior, therefore, this test's drift data is considered an outlier for each test point.
The assumption that the data is normally distributed was tested by performing the D' test'on the drift data for all test points and the W test on the drift data for test point #2, which was the worst case point. The assumption of normality was rejected by the D' test, with the D' value falling below the 4
i desired range of values. This indicates the distribution has a higher kurtosis than would be expected for a normal distribution (i.e., it's more sharply " peaked"). The assumption of normality was not rejected by the W test. Attachment 6 summarizes the D' and W test results. Histograms were created, plotting the number of drift data points versus drift (in number of standard deviations from the mean), for all test points and for test point #2. The histogram for all test points demonstrates that the data distribution is indeed slightly more sharply " peaked" than a normal distribution, with 94.8%
of the drift data points within two standard deviations of the mean. The histogram for test point #2 shows that the data distribution is approximately normal, with 95.5% of the drift datapoints within two standard deviations of the mean. The histograms and supporting detail are shown in Attachment
- 7. Taken together, these results support the assumption that the differential pressure transmitter drift data is normally distributed.
A drift versus time since last test plot for all points is shown in Attachment 8. Most of the drift data
^
points are for test intervals between 15 and 20 months, so no strong evidence is provided to either support or refute the assertion that drift is independent of time. A drift versus time since last i
adjustment plot for all points is shown in Attachment 9. The drift data points for intervals longer
)
than 30-months are all bounded by the worst case points for intervals between 15 and 20 months, i
however, there were many cases where the differential pressure transmitter was adjusted at an
]
interval less than 20-months since its previous adjustment. Therefore, it cannot be concluded that
)
drift is time independent for test intervals up to 30 months. On the other hand, it's also not clear that drift increases linearly with time. To gain additional insight into the time dependent characteristics of differential pressure transmitter drift, the data for all (drift versus time since last adjustment) points was divided into three groups representing various ranges of calibration intervals, and the I
mean and standard deviation were computed for each group. The results, shown in Attachment 10, indicate that the data for intervals greater than 30-months is more conservative, with respect to both mean and standard deviation, than the 15 to 20-month interval data. These results also indicate that the mean value decreases with time since last adjustment. For this reason, a regression line was fit to the drift versus time since last adjustment plot for all points, and it is shown in Attachment 9. Its 2
slope is small in magnitude (-0.007% span per month), and the correlation is low (R = 0.03). This provides further evidence that the transmitters exhibit very little time dependent drift.
Since there isn't strong evidence to support a conclusion of drift time dependency, but merely insufficient evidence to clearly demonstrate a lack thereof, the drift versus time since last test data was extrapolated to a 30-month interval using the square root method described in item 14c of the j
methodology document. The data for test intervals less than three months was excluded from the extrapolation because the drift experienced during those short intervals would likely not have been representative of the drift occurring over the next (at least) 9 intervals of equal length. This can be verified by referring again to Attachment 8 and observing that extrapolation of the short interval data to an 18-month interval would produce results worse than what was obtained for the actual data with
l.*
1 3
i f
intervals between 15 and 20 months. For the extrapolated data, shown on Attachment 11, the sample mean (x) is rtightly negative but close to zero (i.e., between 0 and -0.1% span). Since negative drift j
is conservative for these differential pressure transmitters, and because the magnitude is small, the mean will be considered zero. The worst case 95/95% tolerance factor times sample standard deviation (k's) is 1.?4% span (random error). No outliers were removed from the extrapolated data i
set.
}
The assumption that the extraoolated data is normally distributed was tested by performing the D'
]
test on the drift data for all test points and the W test on the drift data for test point #2, which was again the worst case point. The assumption ofnormality was rejected by the D' test, with the D' value falling below the desired range of values once again. The assumption ofnormality was not rejected by the W test. Attachment 12 summarizes the D' and W test results. Histograms were i
created, plotting the number of drift data points versus drift (in number of standard deviations from l
the mean), for all test points and for test point #2. The histogram for all test points demonstrates that the data distribution is again slightly more shar;,1y " peaked" than a nonnal distribution, with 93.2% '
i of the drift data points within two standard deviations of the mean. The histogram for test point #2 shows that the data distribution is approximately normal, with 94.7% of the drift data points within j
two standard deviations of the mean. The histograms and supporting detail are shown in Attachment
- 13. Taken together, these results support the assumption that the differential pressure transmitter extrapolated drift data is normally distributed.
I I/E converter dwa was taken at 0%,25%, 50%,75%, and 100% of span, in both the increasing and 2
j decreasing directions. For all nine test points, the sample mean (x) is less than 0.02% span, which can be considered to effectively lie zero. Therefore, the entire error associated with the I/E converters is considered random. The worst case 95/95% tolerance interval maxin.um is 0.15%
f span. A one-sided tolerance factor was used for the same reason as in the differential pressure transmitter case. The supporting details for these results are shown on Attachment 14. No outlier candidates were identified by the T-Test.
The assumption that the data is normally distributed was tested by perfonning the W test on the drift data for test point #5, which was the worst case point. The assumption of normality was not rejected. Attachment 15 summarizes the W test results. A histogram was created, plotting the I
number of drift data points versus drift (in number of standard deviations from the mean), for test point #5. It demonstrates that the data distribution is approximately normal and shows that all the data is within two standard deviations of the mean. The histogram and supporting detail are shown in Attachment 16.
A drift versus time since last test plot for all points is shown in Attachment 17. Most of the drift data j
points are for test intervals between 15 and 20 months, so no strong evidence is provided to either support or refute the assertion that drift is independent of time. A drift versus time since last adjustment plot for all points is shown in Attachment 18. This plot contains plenty of data for intervals beyond 30-months, with some data at intervals as high as 70 months. The drift data points for intervals longer than 30 months are all bounded by the worst case points for intervals between 15 and 20 months. This demonstrates that I/E converter drift is not time dependent. To quantitatively support this observation, the data for all (drift versus time since last adjustment) points was divided
into three groups representing various ranges of calibration intervals, and the mean and standard deviation were computed for each group. The results, shown in Attachment 19, indicate that the data for intervals between 30 and 40 months is equal to, with respect to mean, or better than, with respect i
to standard deviation, the data for intervals between 15 and 20 months. Based on the evidence discussed above, it is reasonable to conclude that IE converter drift is independent of time.
The design basis /referenct uncertainty is obtained from B&W document 32-1172392-00, Reactor Protection System String Error Calculations, dated 6/13/88. It lists the differential pressure transmitter's accuracy as 0.25% span (random error) and its drift as 0.25% span (random error). It also lists the I/E converter's accuracy as 0.25% span (random error). Combining the transmitter's two random terms using the square-root-sum-of-squares technique results in an overall random error of 0.35% span. Comparing these errors with the transmitter's 30-month projected drift of 1.34%
span reveals that the design basis / reference uncertainty does not bound the 30-month projected drift.
Therefore, even though the converter's design basis / reference uncertainty (0.25% span) bounds its j
30-month projected drift (0.15% span), B&W will be asked to revise the RPS string error calculation l
document to incorporate the 30-month projected drift determined for the differential pressure transmitter and for the IE converter. They will also be asked to calculate revised RPS Technical Specification Allowable Values, if necessary. The following uncertainty terms must be accounted for by B&W:
Differential Pressure Transmitter Accuracy, Drift, and M&TE Uncertainty
- l.34% span Differential Pressure Transmitter Calibration Tolerance
- 0.25% span
'(ref.: DB-MI-03061 through-03068)
IE Converter Accuracy, Drift, arid M&TE Uncertainty
- 0.15% span IE Converter Calibration Tolerance 0.25% span (ref.: DB-MI-03061 through -03068)
Also, the other uncertainties accounted for in B&W document 32-1172392-00 that are not covered by the above list must continue to be accounted for in the revised RPS string error calculation document.
3 Historical differential presme transmitter drift has exceeded its design basis / reference uncertainty in the non-conservative direction (greater than +0.35% span) during five of the 22 calibrations for which data l
is available. Historical drift exceeded the design basis / reference uncertainty in the conservative direction (less than -0.35% span) during five other calibrations. These results suggest that the design basis value for differential pressure transmitter drift,0.25% span (random error), is not large enough to adequately characterize transmitter performance. PCAQR 96-0278 was initiated to address this condition. Of the ten cases where historical drift exceeded the design basis / reference uncertainty, none would have exceeded the 30-month projected drift. Furthermore, only one of the extrapolated drift data points exceeds the 30-month projected drift - and that in the conservative direction. Therefore, after the 30-month projected drift is incorporated into the RPS string error calculation document, it can reasonably be expected that differential pressure transmitter drift will rarely exceed acceptable limits.
Historical I/E converter drift did not exceed its design basis / reference uncertainty ( 0.25% span) in any of the 27 calibrations for which data was reviewed.
5 Conditions and assumptions of the setpoint and safety analysis were identified by reviewing the following documents:
Updated Safety Analysis Report i
B&W document 32-1172392-00, Reactor Protection System String Error Calculations I
Technical Specifications Toledo Edison Calculation C-ICE-58.01-008, Revision 0 The following surveillance and periodic test procedures were reviewed and verified to appropriately l
. reflect all applicable conditions and assumptions of the setpoint and safety analyses:
j DB-MI-03061, Channel Calibration of FT-RC01 Al, RCS Loop 2 Flow Transmitter to RPS Channel 1 DB-MI-03062, Channel Calibration of FT-RC01 A2, RCS Loop 2 Flow Transmitter to RPS Channel 2 i
DB-MI-03063, Channel Calibration of FT-RC01 A3, RCS Loop 2 i
Flow Transmitter to RPS Channel 3 l
DB-MI-03064, Channel Calibration of FT-RC01 A4, RCS Loop 2 j
Flow Transmitter to RPS Channel 4 DB-MI-03065, Channel Calibration of FT-RC01B1, RCS Loop 1 Flow Transmitter to RPS Channel 1
{
DB-MI-03066, Channel Calibration of FT-RC01B2, RCS Loop 1 l
Flow Transmitter to RPS Channel 2 DB-MI-03%7, Channel Calibration of FT-RC01B3, RCS Loop 1 Flow Transmitter to RPS Channel 3 DB-MI-03068, Channel Calibration of FT-RC01B4, RCS Loop 1 Flow Transmitter to RPS Channel 4 DB-MI-03057, RPS Channel 1 Calibration of Overpower, Power / Imbalance / Flow, and Power / Pumps Trip Functions DB-MI-03058, RPS Channel 2 Calibration of Overpower, Power / Imbalance / Flow, and Power / Pumps Trip Functions DB-MI-03059, RPS Channel 3 Calibration of Overpower, Power / Imbalance / Flow, and Power / Pumps Trip Functions DB-MI-03%0, RPS Channel 4 Calibration of Overpower, Power / Imbalance / Flow, and Power / Pumps Trip Functions DB-SC-04117, RPS Channel 1 Flow Scaling Factor Determination DB-SC-04118, RPS Channel 2 Flow Scaling Factor Determination DB-SC-04119, RPS Channel 3 Flow Scaling Factor Determination DB-SC-04120, RPS Channel 4 Flow Scaling Factor Determination DB-OP-03006, Miscellaneous Instmment Shift Check
4 6
Some or all of these procedures may require alteration after the RPS string error calculation document is revised and any needed changes to RPS Technical Specification Allowable Values are determined.
i Preparer:
74%
C,F-/5'-76 Signature /Date Reviewer:
),
3 -/5 -76
/ Signature /Date 3
0 4
i
Reactor Coolant Hot Leg Flow j
Reactor Protection System Channel 1
Range = 0 to 80 mpph
,-------------T-------7 (for each loop) l I
l I
FTRC1A1 FYRC1A1 RPS1RC1404 I
l l
l l
(Differential Pressure (Current-to-Voltag l
Transmitter)
Converter) l (Square Root Extractor) l l
l L_____q l DB-MI-03061 Loop 2 I
I i
i
____1_____'
j e_________
FYRC1-1 l
l (Buffer Amplifier)
[
l l
l l
l l
T----
J l
l FTRC1B1 FYRC181 l
RPS1RC1407 l
ffer ial Pressure Current to-Voltag (Square Root Extractor)
Genera and Bistable Modules i
l l
l l
DB-MI-03057 I
DB-MI-03065 Loop 1 i
u______________1______________a l
Reactor Coolant Hot Leg Flow 1
Reactor Protection System Channel 2
r---------7 Range = 0 to 80 mpph (for each loop) l j
I FTRC1A2 FYRC1A2 RPS2RC1404 (Differential Pressure (Current-to-Voltage l
Transmitter)
Converter) l (Square Root Extractor) l l
l L_____q l DB-MI-03062 Loop 2 I
I u______________1_____,
i (Buffer Amplifier) l l
l I
I i
i
______________7_____a j
i FTRC1B2 FYRC182 l
RPS2RC1407 l
fferential Pressure Current to-Voltage Rd h%
r M
Bistable Modules t
l DB-MI-03066 Loop 1 l
l DB-MI-03058
______________x______________,
1
Reactor Coolant Hot Leg Flow Reactor Protection System Channel 3 r
Range = 0 to 80 mpph (for each loop)
I I
l FTRC1A3 FYRC1A3 RPS3RC1504 (Differential Pressure (Current-to-Voltage l
I Transmitter)
Converter) l (Square Root Extractor) l j
l L_____q l DB-MI-03063 Loop 2 I
I l
L I
FYRC1-3 (Buffer Amplifier) l l
i
__7_____J l
l I
FTRC183 FYRC1B3 l
RPS3RC1507 I
ffer ial Pressure Current to-Voltag (Square Root Extractor)
Genera and Bistable Modules i
I I
l DB-MI-03067 Loop 1 I
DB-MI-03059 I
u______________a_______________a t
I
Reactor Coolant Hot Leg Flow 1
Reactor Protection System Channel 4 Range = 0 to 80 mpph r -
T - - - - - -- -
7 (for each loop)
I l
I FTRC1A4 FYRC1A4 RPS4RC1504 I
I (Differential Pressure (Current-to-Voltag l
Transmitter)
Converter) l (Square Root Extractor) l l
l L_____q j
i DB-MI-03064 Loop 2 I
I u______________1_____'
FYRC1-4 I
l (Buffer Amplifier)
I l
l
__,_____a i
j l
FTRC184 FYRC184 I
RPS4RC1507 I
ns er)
Convert r (Square Root Extractor)
Genera and Bistable Modules l
l l
l DB-MI-03068 Loop 1 l
DB-MI-03060 l
u___________
___u______________a l
t 1
.i fr U
V I
W X
Y Z
AA E
A 8
C l
P Q
R l
S T
3 Instrument Span:
1.6 Pomt#2 Pomt #3 Point #4 Point #5 4
Sh$edData Sh$ed Data ShpedData Sh$rd Data 6
0.8 volts 1.2 volts 1.6 volts 2.0 volts 6
AF Date AL Date Instrument As Fcund As Left Adrusted 7 As Found As left Adjusted 7 As Found As Left Adiusted 7 As Found As Left Adiosted 7 7
IT-RC01A1
(
8 na 3/2490 nn 0.7980 y
na 1.1990
.y na 1.6000 y
na 1.9210 y
9 Gr25/90 4/25/90 0.8030 0.7990 y
1.2040 1.2010 y
1.6060 1.6020 y
1.9270 1.9230 y
10 9/11/91 9/11/91 0.8110 0.7990 y
1.2110 1.1990 y
1.6110 1.6000 y
1.9300 1.9200 y
11 3/16/93 3/16S3 0.8004 0.8004 a
1.2016 1.2016 n
1.6023 1.6023 n
1.9230 1.9230 n
12 10/13/94 10/13194 0.8010 0.8000 y
1.2020 1.2000 y
1.6020 1.6000 y
1.9230 1.9200 y
13 IT-RColBI 14 na 303/90 nn 0.7990 y
na 1.1990 y
na 1.5990 y
na 1.9190 y
16 5/2490 5/24S 0 0.8033 0.8007 y
1.2048 1.2024 y
1.6059 1.6024 y
1.9259 1.9226 y
16 9/12/91 9/12/91 0.7960 0.8020 y
1.1970 1.2030 y
1.5960 1.6030 y
1.9160 1.9230 y
17 3/18S 3 3/18/93 0.8070 0.8000 y
1.2060 1.2000 y
1.6060 1.6010 y
1.9260 1.9200 y
f 18 IT-RC01 A2 19 na 3/15/90 na 0.7990 y
na IJ990 y
na 1.5990 y
na 1.9190 y
20 9/17SI 9/I7S1 0.8030 0.8000 y
1.2020 1.r990 y
1.6020 1.5990 y
1.9220 1.9190 y
21 3/15/93 3/15/93 0.8004 0.8004 n
1.2009 1.2009 n
1.6006 1.6006 n
1.9211 1.92ll n
22 10/13/94 10/13/94 0.7950 0.8000 y
1.1950 1.2000 y
1.5960 I.6010 y
1.9160 1.9210 y
23 FT-RC01B2 24 na
$/23/90 nn 0.7980 y
na 1.1980 y
na 1.5980 y
na 1.9190 y
25 5/27/90 5/27/90 0.7940 0.7990 y
1.1940 1.1990 y
1.5950 1.6000 y
I.9150 1.9210 y
26 9/13/91 9/13/91 0.7990 0.7990 n
L.2000 1.2000 n
1.6000 1.6000 n
I.9200 1.9200 n
27 3/18/93 3/18S 3 0.7920 0.8000 y
1.1940 1.2010 y
1.5980 1.6000 y
1.9180 1.9210 y
28 10/11/94 10/11/94 0.8030 0.8010 y
1.2030 1.2010 y
1.6040 1.6020 y
1.9250 1.9230 y
29 FT-RC01 A3 30 na 3/29!90 nn 0.7990 y
na 1.1990 y
na 1.5980 y
na 1.9200 y
31 9/18/91 9/18:91 68020 68000 y
1.2020 1.1990 y
1.6030 1.5990 y
1.9240 1.9190 y
32 3/16/93 3/16/93 0.7977 0.7977 a
1.1961 1.1961 m
1.5965 1.5965 a
1.9169 1.9169 m
33 10/15!94 loil5/94 0.7660 A7980 y
1.1650 1.1980 y
1.5640 1.5990 y
1.8840 1.9200 y
34 IT-RC01B3 l
36 na 3/30/90 na R8000 y
na 1.1990 y
na 1.6000 y
na 1.9200 y
l 36 9/16/91 9tl6/91 0.7840 0.8010 y
1.1840 1.2000 y
1.5840 1.6010 y
1.9030 1.9200 y
37 3/l1!93 3:17/93 0.7978
& 7978 n
1.1974 1.1974 n
1.5984 1.5988 n
1.9171 1.9171 n
l 38 IT-RC01 A4 I
39 na 3/31/90 na A7990 y
na 1.1990 y
na 1.5990 y
na 1.9100 y
l 40 9/l7/91 9/11/9 L 0.8090
- 8010 y
1.2070 1.2000 y
1.6060 1.6000 y
1.9260 1.9200 y
41 3/l5!93 3/15/93 0.8001
&8001 n
1.2008 1.2008 n
1.6018 1.6018 n
1.9234 1.9234 n
42 FT-RC01B4 43 na 3/31/90 nn 0.7990 y
na 1.1990 y
na 1.5990 y
na 1.9190 y
44 9/14/91 9/l491 0.78 70 0 8000 y
1.1880 1.2010 y
1.5890 1.6020 y
1.9090 1.9220 y
46 3/17l93 3/l733 0.8015 0.7992 y
1.2041 1.1995 y
1.6060 1.6006 y
1.9278 I.9199 y
1
Attcchment 5 (cont)
A l
B j
C l
AD i
AE AF l
AG l
AH l
Al l
AJ 1
2 3
Instrument Span:
1.6 Potnt #6 Point #7 Point #8 4
ShiftdDaa Shiftd Data Shiftd Data 5
l.6 volts 1.2 volts 0.8 volts 6
AF Date AL Date Instrument As Found As Left Adiusted 7 As Found As Left Adiusted ?
As Found As Left Adiusted 7 7
FT-RC01A1 8
na F24/90 na 1.6000 y
na 1.1990 y
nn 0.7990 y
9 4/25/90 4/25/90 1.6060 1.6020 y
1.2050 12010 y
0.8030 0.7990 y
10 9/I1/9I 9/11/91 1.6120 1.6000 y
1.2120 1.1990 y
0.8120 0.7990 y
11 3/16/93 3/16 S 3 1.6031 1.6031 n
1.2016 1.2016 n
0.8004 0.8004 n
12 10/13/94 10/13/94 1.6020 1.6000 y
1.2020 1.2000 y
0.8010 0.8000 y
13 FT-RC01BI 14 na 3/2360 na I.5990 y.
na 1.1990 y
na 0.7990 y
15 5/24 S0 5/24/90 1.6058 1.6027 y
1.2048 1.2020 y
0.8037 0.8009 y
16 9/12/91 9/12/91 1.5960 1.6030 y
1.1960 1.2030 y
0.7960 0.8020 y
17 3/18/93 3/18/93 1.6070 1.6010 y
1.2070 1.2000 y
0.8070 0.8000 y
18 FT-RC01A2 19 na 3/15/90 na 1.6000 y
na 1.1990 y
na 0.7990 y
20 9/17/91 9/17/91 1.6020 1.5990 y
1.2020 1.1990 y
0.8030 0.8000 y
21 3/15/93 3/15/93 1.6010 1.6010 n
1.2002 1.2002 n
0.7999 0.7999 n
22 10/13/94 10/13 S4 1.5960 1.6000 y
1.1950 1.2000 y
0.7950 0.8000 y
23 FT-RC01B2 24 na 5/23/90 na 1.5980 y
na 1.1980 y
na 0.7980 y
25 5/27/90 5/27/90 1.5950 1.6000 y
1.1950 1.2000 y
0.7940 0.7995 y
26 9/13/91 9/1331 1.5990 1.5990 n
1.1990 1.1990 n
0.7990 0.7990 n
27 3/18/93 3/18/93 1.5960 1.6020 y
1.1950 1.2010 y
0.7930 0.8000 y
28 10/11/94 10/1164 1.6040 1.6020 y
1.2040 1.2020 y
0.8030 0.8010 y
29 FT-RC01 A3 30 na 3/2960 na 15990 y
na L1990 y
na R8000 y
31 9/18/91 9/18/91 L6030 15990 y
L2020 L1990 y
A8020 A3000 y
32 3/1663 3/16/93
L5?68 15968 n
L1965 L1965 m
A 7976 A 7976 n
33 10/15/94 10/15S4 L5640 L5990 y
L1650 L1990 y
0 7660 6 7980 y
34 FT-RCOIB3 35 na 3/30/90 na 16010 y
na L2000 y
na A8000 y
36 9/16/91 9/l6/91 L5840 L6010 y
L IR40, L2000 y
G7840 68010 y
37 3/17/93 3/11/93 L5987 L5987 n
L1981 1.1981 l
n 0.79 79
& 79 79 n
38 FT-RC01 A4 39 na 3/31/90 na L5980 y
na L1980 y
na A 7990 y
40 9/17i91 9/11/91 L6060 L6000 y
L2080 L2000 y
08090 68010 y
41 3/15/93 3!15/93 L6020 L 6020 n
L2005 L2005 a
68002 0.8002 n
42 FT-RC01B4 43 na 3/31/90 na L5990 y
na L1990 y
na R 7980 y
44 9/14S1 9ll4/91 L5890 L6020 y
L1880 L2010 y
& 78 70 68000 y
45 3/17/93 3/17/93 L6066 1 6006 y
12049 L2005 y
a#020 a7995 y
Attachm:nt 5 (cont.)
A B
C AO l
Months Since 1,ast Test 2
j 3
Instrument Span:
1.6 4
5 6
AF Date AL Date Instrument Point #2 Point #3 Point #4 Point #5 Point #6 Point #7 Point #8 7
FT-RC01A1 8
na 3/24/90 9
4/25/90 4/25/90 1.051 1.051 1.051 1.051 1.051 1.051 1.051 10 9/11/91 9/11/91 16.559 16.559 16.559 16.559 16.559 16.559 16.559 11 3/16/93 3/16/93 18.136 18.136 18.136 18.136 18.136- - 18.136 18.136 12 10/13/94 10/13/94 18.924 18.924 18.924 18.924 18.924 18.924 18.924 13 FT-RC0181 14 na 3/23/90 15 5/24/90 5/24/90 2.037 2.037 2.037 2.037 2.037 2.037 2.037 16 9/12/91 9/12/91 15.639 15.639 15.639 15.639 15.639 15.639 15.639 17 3/18/93 3/18/93 18.168 18.168 18.168 18.168 18.168 18.168 18.168 j
18 FT-RC01A2 19 na 3/15/90 20 9/17/91 9/17/91 18.103 18.103 18.103 18.103 18.103 18.103 18.103
^
21 3/15/93 3/15/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 22 10/13/94 10/13/94 18.957 18.957 18.957 18.957 18.957 18.957 18.957 23 FT RColB2 24 na 5/23/90 25 5/27/90 5/27/90 0.131 0.131 0.131 0.131 0.131 0.131 0.131 26 9/13/91 9/13/91 15.573 15.573 15.573 15.573 15.573 15.573 15.573 27 3/18/93 3/18/93 18.136 18.136 18.136 18.136 18.136 18.136 18.136 28 10/11/94 10/11/94 18.793 18.793 18.793 18.793 18.793 18.793 18.793 29 FT-RC01A3 30 na 3/29/90 31 9/18/91 9/18/91 17.676 17.676 17.676 17.676 17.676 17.676 17.676 32 3/16/93 3/16/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 33 10/15/94 10/15/94 18.990 18.990 18.990 18.990 18.990 18.990 18.990 34 FT-RC01B3 35 na 3/30/90 36 9/16/91 9/16/91 17.577 17.577 17.577 17.577 17.577 17.577 17.577 37 3/17/93 3/17/93 18.004 18.004 18.004 18.004 18.004 18.004 18.004 38 FT-RC01 A4 39 na 3/31/90 40 9/17/91 9/17/91 17.577 17.577 17.577 17.577 17.577 17.577 17.577 41 3/15/93 3/15/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 42 FT-RC01B4 4
43 na 3/31/90 44 9/14/91 9/14/91 17.478 17.478 17.478 17.478 17.478 17.478 17.478 45 3/17/93 3/17/93 18.070 18.070 18.070 18.070 18.070 18.070 18.070
A Attachment G (cont.)
BX BY BZ CA l CB l CC l CD CE I CF i CG 1
Percent Dnft Since Las Test - one sided interval 2
3 Instrument Span:
1.6 4
6 6
AF Date AL Date Instrument Point #2 Point #3 Pomt#4 Point #5 Point #6 Point #7 Point #8 7
FT-RCotAl j
8 na 3/24/90 j
9 4/25/90 4/25/90 03125 03125 0375 0.375 0375 0.375 0.25 10 9/11/91 9/11/91 0.75 0.625 0.5625 0.4375 0.625 0.6875 0.8125 11 3/16/93 3/16/93 0.0875 0.1625 0.14375 0.1875 0.19375 0.1625 0.0875 12 10/13/94 10/13/94 0.0375 0.025
-0.01875 0
-0.06875 0.025 0.0375 13 FT RC01BI 14 na 3/23/90 16 5/24/90 5/24/90 0.26875 0.3625 0.43125 0.43125 0.425 03625 0.29375 16 9/12/91 9/12/91
-0.29375
-03375
-0.4
-0.4125
-0.41875
-0375
-030625 iT 3/18/93 3/18/93 0.3125 0.1875 0.1875 0.1875 0.25 0.25 0.3125 18 FT-RC01A2 19 na 3/15/90 20 9/17/91 9/17/91 0.25 0.1875 0.1875 0.1875 0.125 0.1875 0.25 21 3/15/93 3/15S3 0.025 0.11875 0.1 0.13125 0.125 0.075
-0.00625 22 10/13/94 10/13/94
-03375
-0.36875
-0.2875
-0.31875
-03125
-0325
-030625 23 FT-RC01B2 24 na 5/23/90 26 5/27/90
$/27/90
-0.25
-0.25
-0.1875
-0.25
-0.1875
-0.1875
-0.25 26 9/13/91 9/13/91 0
0.0625 0
-0.0625
-0.0625
-0.0625
-0.03125 27 3/18/93 3/18/93
-0.4375
-0375
-0.125
-0.125
-0.1875
-0.25 0.375 l
28 10/11/94 10/11/94 0.1875 0.125 0.25 0.25 0.125 0.1875 0.1875 29 FT-RC01A3 30 na 3/29/90 j
31 9/18/91 9/18/91 0.1875 0.1875 0.3125 0.25 0.25 0.3 /5 0.125 1
32 3/16/93 3/16/93
-0.14375
-0.18125
-0.15625
-0.13125
-0.1375
-0.16875
-0.15 33 10/15/94 10/15/94
-1.98125
-l.94375
-2.03125
-2.05625
-2.05
-1.95625
-1.975 34 FT-RCO1B3 36 na 3/30/90 36 9/16/91 9/16/91 1
-0.9375 1
-1.0625
-1.0625
-l
-1 37 3/17/93 3/17 S 3
-0.2
-0.1625
-0.1375
-0.18125
-0.14375 0.11875
-0.19375 38 FT-RCot A4 39 na 3/31/90 40 9/17/91 9/17/91 0.625 0.5 0.4373 0375 0.5 0.625 0.625 i
di 3/15/93 3/15/93
-0.05625 0.05 0.1125 0.2125 0.125 0.03125
-0.05 42 FT-RC01B4 43 na 3/31S0 n
44 9/14/91 9/14/91
-0.75
-0.6875
-0.625
-0.625
-0.625
-0.6875
-0.6875 46 3/17/93 3/17/93 0.09375 0.19375 0.25 03625 0.2875 0.24375 0.125 j
46 47 48 49 60 Mean
-0.101
-0.093
-0.070
-0.080
-0.080
-0.075
-0.097 4
61 Sato 0.570 0.544 0.560 0.568 0.573 0.562 0.566 62 Count 23 23 23 23 23 23 23 63 k (one sided) 2328 2.328 2328 2328 2.328 2328 2328 64 k*s 1327 1.267 1.305 1321 1.335 1309 1.318 l
Y~
95/95 Max 1.226 1.174 1.234 1.241 1.255 1.234 1.221 66 2
67 68 Outlier Analysis 69 T
2.62 2.62 2.62 2.62 2.62 2.62 2.62 3
x-Ts
-1.594 1.519
-1.539
-1.567
-1.583
-1.548 1.580 61 x+Ts 1393 1.333 1.398 1.407 1.422 1398 1386 J
Outliers 1
1 1
I I
I I
3 Mean
-0.015
-0.009 0.019 0.010 0.009 0.010
-0.011 64 Sstd 0.405 0.374 0.371 0378 0.389 0.394 0.400 g
Count 22 22 22 22 22 22 22 3
k (one sided) 2349 2.349 2349 2.349 2.349 2.349 2.349 67 k*s 0.952 0.879 0.872 0.888 0.914 0.925 0.939 Y
95/95 Max 0.937 0 870 0.890 0.898 0.923 0 935 0.928 i
___ _. ~.._ _.. _,. _ _ _
Attachm:nt 5 (cont.)
A B
C BG l
BH l
BI BJ BK BL BM 1
Months Since Last Adjustment 2
3 Instmment Span:
1.6 4
5 6
AF Date AL Date Instrument Point #2 Point #3 Point #4 Point #5 Point #6 Point #7 Point #8 7
FT-RC01A1 8
na 3/24/90 9
4/25/90 4/25/90 1.051 1.051 1.051 1.051 1.051 1.051 1.051 10 9/11/91 9/11/91 16.559 16.559 16.559 16.559 16.559 16.559 16.559 11 3/16/93 3/16/93 18.136 18.136 18.136 18.136 18.136 - 18.136 18.136 12 10/13/94 10/13/94 37.060 37.060 37.060 37.060 37.060 37.060 37.060 13 FT-RC01B1 14 na 3/23/90 15 5/24/90 5/24/90 2.037 2.037 2.037 2.037 2.037 2.037 2.037 16 9/12/91 9/12/91 15.639 15.639 15.639 15.639 15.639 15.639 15.639 17 3/18/93 3/18/93 18.168 18.168 18.168 18.168 18.168 18.168 18.168 18 FT-RC01A2 19 na 3/15/90 20 9/17/91 9/17/91 18.103 18.103 18.103 18.103 18.103 18.103 18.103 21 3/15/93 3/15/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 22 10/13/94 10/13/94 36.862 36.862 36.862 36.862 36.862 36.862 36.862 23 FT-RC01B2 24 na 5/13/90 25 5/27/90 5/27/90 0.131 0.131 0.131 0.131 0.131 0.131 0.131 26 9/13/91 9/13/91 15.573 15.573 15.573 15.573 15.573 15.573 15.573 1
27 3/18/93 3/18/93 33.708 33.708 33.708 33.708 33.708 33.708 33.708 28 10/11/94 10/11/94 18.793 18.793 18.793 18.793 18.793 18.793 18.793 29 FT-RColA3 30 na 3/29/90 31 9/18/91 9/18/91 17.676 17.676 17.676 17.676 17.676 17.676 17.676 32 3/16/93 3/16/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 33 10/15/94 10/15/94 36.895 36.895 36.895 36.895 36.895 36.895 36.895 34 FT-RC01B3 35 na 3/30/90 36 9/16/91 9/16/91 17.577 17.577 17.577 17.577 17.577 17.577 17.577 37 3/17/93 3/17/93 18.004 18.004 18.004 18.004 18.004 18.004 18.004 38 FT-RC01A4 39 na 3/31/90 40 9/17/91 9/17/91 17.517 17.577 17.577 17.577 17.577 17.577 17.577 41 3/15/93 3/15/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 42 FT-RC01B4 43 na 3/31/90 44 9/14/91 9/14/91 17.478 17.478 17.478 17.478 17.478 17.478 17.478 45 3/17/93 3/17/93 18.070 18.070 18.070 18.070 18.070 18.070 18.070 (cont.)
A B
C BP l
BQ l
BR BS BT BU BV 1
Percent Drift Since Last Adjustment 2
3 Instrument Span:
1.6 4
5 6
AF Date AL Date Instrument Point #2 Point #3 Point #4 Point #5 Point #6 Point #7 Point #8 7
FT-RC01 A1 8
na 3/24/90 9
4/25/90 4/25/90 0.3125 0.3125 0.3750 0.3750 0.3750 0.3750 0.2500 10 9/11/91 9/11/91 0.7500 0.6250 0.5625 0.4375 0.6250 0.6875 0.8125 11 3/16/93 3/16/93 0.0875 0.1625 0.1437 0.1875 0.1937 0.1625 0.0875 12 10/13/94 10/13/94 0.1250 0.1875 0.1250 0.1875 0.1250 0.1875 0.1250
__13 FT-RC01B1 14 na 3/23/90 15 5/24/90
$/24/90 0.2687 0.3625 0.4313 0.4312 0.4250 0.3625 0.2937 i
16 9/12/91 9/12/91
-0.2937
-0.3375
-0.4000
-0.4125
-0.4187
-0.3750
-0.3062 17 3/18/93 3/18/93 0.3125 0.I875 0.I875 0.I875 0.2500 0.2500 0.3125 18 FT-RCO1A2 19 na 3/15/90 20 9/17/91 9/17/91 0.2500 0.1875 0.1875 0.1875 0.1250 0.1875 0.2500 21 3/15/93 3/15/93 0.0250 0.1188 0.1000 0.1312 0.1250 0.0750
-0.0062 22 10/13/94 10/13/94
-0.3125
-0.2500
-0.1875
-0.1875
-0.1875
-0.2500
-0.3125 23 FT-RC01B2 24 na 5/23/90 25 5/27/90 5/27/90
-0.2500
-0.2500
-0.1875
-0.2500
-0.1875
-0.1875
-0.2500 26 9/13/91 9/13/91 0.0000 0.0625 0.0000
-0.0625
-0.0625
-0.0625
-0.0312 27 3/18/93 3/18/93
-0.4375
-0.3125
-0.1250
-0.1875
-0.2500
-0.3125
-0.4062 j
28 10/11/94 10/11/94 0.1875 0.1250 0.2500 0.2500 0.1250 0.1875 0.1875 29 FT-RColA3 j
30 na 3/29/90 31 9/18/91 9/18/91 0.1875 0.I875 0.3125 0.2500 0.2500 0.I875 0.1250 32 3/16/93 3/16/93
-0.1437
-0.1812
-0.1562
-0.1313
-0.1375
-0.1687
-0.1500 33 10/15/94 10/15/94
-2.1250
-2.1250
-2.1875
-2.1875
-2.1875
-2.1250
-2.1250 34 FT-RColB3 7
35 na 3/30/90 36 9/16/91 9/16/91
-1.0000
-0.9375
-1.0000
-1.0625
-1.0625
-1.0000
-1.0000 37 3/17/93 3/17/93
-0.2000
-0.1625
-0.1375
-0.1812
-0.1437
-0.1188
-0.1937 38 FT RC01 A4 1
39 na 3/31/90 40 9/17/91 9/17/91 0.6250 0.5000 0.4375 0.3750 0.5000 0.6250 0.6250 41 3/15/93 3/15/93
-0.0563 0.0500 0.1125 0.2125 0.1250 0.0312
-0.0500 42 FT-RC01B4 45 na 3/31/90 4
44 9/14/91 9/14/91
-0.7500
-0.6875
-0.6250
-0.6250
-0.6250
-0.6875
-0.6875 45 3/17/93 3/17/93 0.0938 0.1937 0.2500 0.3625 0.2875 0.2437 0.1250 4
De D'=T/S value of $23.02 for a 0.05 significance level of normality should fall between approximately $29 and $46.
He D' value for this test is less than the values it should fall in, which shows it has a high kurtosis.
l A high kurtosis says that the data came fmm a distribution that is more sharply " peaked" than the normal distribution.
The assumption of normality is rejected at the 0.05 significance level.
I ETRCI AI - B4 allpoints ne D' Test i
Ordered Xi stdev's count:
154 T - term 1
-1.0625
-2.8004 S^2:
22.09842 81.2813 2
-1.0625
-2.8004 89.2187 3
-1
-2.6360 T:
2458.66250 74.5 g
A
-1
-2.6360 D'=T/S:
523.02007 73.5 5
-1
-2.6360 72.5
[
6
-1
-2.6360 average:
0.0018 71.5 7
-0.9375
-2.4715 stdev:
0.3800 66.0937 8
-0.75
-1.9781
$2.125 9
-0.6875
-1.8137 47.0938 i
r F
l t
i L
I
' (cont.)
i FTRCI AI - B4 l
l W test for normality - FTRCI Al - B4 point #2 step #
4.2.2 S^2=
3.447 4.2.3 0.8033 l
1 1.7500 0.4590
=
2 1.3750 0.3156 0.434 3
0.7500 0.2571 0.1928
=
4 0.6500 0.2131 0.1385
=
5 0.5625 0.1764 0.0992
=
6 0.5000 0.1443
=__
0.0722 7
0.3875 0.1150 0.0446
=
8 0.3312 0.0878 0.0291
=
9 0.1500 0.0618 0.0093
=
10 0.0875 0.0368 00032
=
11 0.0125 0.0122 0.0002 r
12 0.0000 0
=
l 13 0.0000 0
=
14 0.0000 0
=
15 0.0000 0
=
16 0.0000 0
=
17 0.0000 0
=
18 0.0000
=
0 0
19 0.0000
=
20 0.0000 0
=
21 0 % 00 0
=
22 0 8000 0
=
23 0.0000 0
=
0 24 0.0000
=
25 0.0000 0
=
b=
1.82620 4.2.4 b^2 =
3.33502 W=b^2/S^2=
0.96740 (test statistic) l W critical value 0.9110 The assumption of normaility is NOT rejected at the.05 level.
l I
l l
i i
)
4 i;
i I
i i
n 154 j
average 0.002
{
stdev 0.380 Din Description Upocr ilound Expected Obsened j
1
(-e,X-20)
-0.76 3.50 7
j 2
(X-20,X-(4/3)o)
-0.50 10.54 7
j 3
(X-(4/3)o,X-(2/3)o)
-0.25 24.84 17 j
4 (X-(2/3)o,X) 0.00 38.12 37 l
5 (X,X+(2/3)o) 0.26 38.12 54
{
6 (X+(2/3)o,X+(4/3)o) 0.51 24.84 23 l
7 (X+(4/3)o,X+20) 0.76 10.54 8
l 8
(X+20,X+m) 9999.00 3.50 1
I U D*d j
Verification of Normality-Binning Method 3
FTRCI Al-B4 - All Points sobed
{
g 50 I
~
f 40 3
s
>T' i
a
'I tj S
g 30 y
f;g k J.
l 4
il Q2
-~
l
}
p e
2 20 p
c.
b h
2 p
g-s si. n.se i
i i
i i
i i
i i
.a 4
l l
I
1 (cont.)
T n
22 average
-0.015 stdev 0.405 Djn Descrintion Upper Bound Expected Obsmed I
(-e,X-20)
-0.83 0.50 1
2 (X-2o,X.6670)
-0.29 5.05 4
4 3
(X.667o,X+.6670) 0.26 10.90 12 4
(X+.667o,X+2a) 0.80 5.05 5
l 5
(X+20,+e) 9999.00 0.50 0
1 i
\\
l i
Verification of Normality - Binning Method y
d FTRCI A1 B4 - Point #2 1.'
i J
lIi!.
9 Ol?:#}
O
'h $1 ^#
.?'%E
.l 5NM I'
j.g(f:g; is a)6 Q
y 8f
.y;5-)
<1%f y
pf g[:
i, a 4
e i
m
- b; f.V W;;.1 3
e ug?
?
,4 N.h??
(4.X-2e)
(X-2o.X.M7o)
(X.uk.X+.ulo)
(X, m%,X,20)
(X.2e,
)
i ihn Range i
a i
J e
02 O:
l y.
8 1
+ * *
+ ; 4 a 44*
4 6
1 e-t 4
1 l
2 1
t
)
)
s s
s et h
t Tn no t io t
aP O
s 8
Ll ts l
e eA T
tn c(
1 t
0 e
s in4 m
a SB L
hc e -
e c
a m1 n
t i
A S
t i
A T1 e
0 m
.vC T
i R
tf 8
iT rDF 6
4 I +*
2 0
0 0
0 0
0 0
0 0
0 0
0 0
0 8
6 4
2 0
2 4
6 8
0 2
1 0
0 0
0 0
0 0
0 0
1 1
y~Dje2s5
0 4
7
,4 e++
7 3
1 0 69 5
+ 2 3
x 0 1
0
.4++
g 70 =
02 R 0-
=
y 0
3 5
2 t
)
n s
h e
m) t n
o t
M s s t
u n
(
t ji n
d o e
AP m
9 t
l t
l s
s u
tn aA 0
j d
e L(
2 m
4 A
h eB t
c s
t i -
-.,+
a c
n a
L
- 5.g t
SI E,e gy e
A e A c
, 9+
+ ^
4 n
mI i
iC S
R
+0+*
e T. T m
i vF
- -i t$
T tf 5
i 1
rD 0
1 5
- - g
-4+.
h >
0 0
0 0
0 0
0 0
0 0
0 0
0 0
8 6
4 2
0 2
4 6
8 0
2 1
0 0
O 0
0 0
0 0
0 1
1 7
g asE cE
. 0 Binning Method - Percent Drift i
i B.in Months Months Count hican Sid
[0,10]
>0
<=3 21 0.16 0.28 l
Months Months
[10,30]
>15
<=20 112 0.01 0.41 Months Months
[30,50]
>30
<=40 21
-0.13 0.21 Drift v. Time Since Last Adjustment FrRCI AI - B4 (All Poir.ls) a g
~
0.80 -
0.60
-- g 4 j'
O.40 -
-h m
0.20 -
g-
]
0.00 -
t:
-0.20 3 E
-0.40 -
l I
g e:
I
-0.60
-0.80
.l.00 -
1.20 0
5 10 15 20 25 30 35 40 Time Since Last Adjustment (Months) l l
i
I 4 1 l
BX BY l BZ CH l Cl l CJ l CK CL CM I CN i
1 Percent Drin Since Lad 7 est l
2 extrapolated to 30 months via square root method 3
Instrument Span:
1.6 (all ted-ted intervals <3 months neglected) 4 5
6 AF Date AL Date Instrument Point #2 Point #3 Point #4 Point #5 Point #6 Point #7 Point #8 7
FT-RC01A1 8
na 3/24/90 9
4/25/90 4/25/90 10 9/11/91 9/11/91 1.010 0.841 0.757 0.589 0.841 0.925 1.094 I
11 3/16/93 3/16/93 0.113 0.209 0.185 0.241 0.249 0.209 0.113 12 10/13/94 10/13N4 0.047 0.031
-0.024 0.000
-0.087 0.031 0.047 13 FT RColBI 14 na 3/23/90 15 5/24/90 5/24/90 16 9/12/91 9/12/91
-0.407
-0.467
-0.554
-0.571
-0.580 0.519
-0.424 17 3/18s3 3/18/93 0.402 0.241 0.241 0321 0.321 0.402 18 FT-RC01 A2 19 na 3/15/90 20 9/17/91 9/17/91 0.322 0.241 0.241 0.241 0.161 0.241 0.322 21 3/15/93 3/15/93 0.032 0.154 0.129 0.170 0.162 0.097
-0.008 22 10/13/94 10/13/94
-0.425
-0.464
-0.362
-0.401
-0.393
-0.409
-0385 23 FT-RColB2 24 na 5/23/90 25 5/27/90 5/27S 0 26 9/13/91 9/13/91 0.000 0 '87 0.000
-0.087
-0.087
-0.087
-0.043 27 3/18/93 3/18/93
-0.563
' ! 12
-0.161
-0.161
-0.241
-0.322
-0.482 28 10/11/94 10/11/94 0.237
,[
0.316 0.316 0.158 0.237 0.237 29 FT-RC01 A3 30 na 3/29/90 31 9/18/91 9/18/91 0.244 0.244 0.407 0.326 0326 0.244 0.163 32 3/16/93 3/16/93
-0.186
-0.235
-0.202
-0.170
-0.178
-0.218
-0.194 1
33 10/15/94 10/15/94 i
34 IT-RC01B3 35 na 3/30/90 36 9/16/91 9/16/91
-1306
-1.225 1.306
-1.388 1388
-1306
-1.306 37 3/17/93 3/17/93
-0.258
-0.210
-0.177
-0.234
-0.186
-0.153
-0.250 38 IT-RCot A4 39 na 3/31/90 40 9/17/91 9/17191 0.817 0.653 0.572 0.490 0.653 0.817 0.817 41 3/15/93 3/15/93
-0.073 0.065 0.146 0.275 0.162 0.040
-0.065 42 FT-RColB4 43 na 3/31/90 44 9/14/91 9/14/91
-0.983
-0.901
-0.819
-0.819
-0.819
-0.901
-0.901 45 3/17/93 3/17/93 0.121 0.250 0.322 0.467 0370 0.314 0.161 46 47 48 49 50 Mean
-0.045
-0.043
-0.015
-0.025
-0.029
-0.023
-0.037 51 Sdd 0.553 0.501 0.493 0.499 0.519 0.531 0.548 52 Count 19 19 19 19 19 19 19 53 k (one sided) 2.423 2.423 2.423 2.423 2.423 2.423 2.423 54 L*s 1339 1.214 1.193 1.209 1.259 1.287 1.328 55 95/95 Max 1.294 1.172 1.178 1.184 1.230 1.264 1.291
~56 57 58 Outlier Analysis SS T
2.53 2.53 2.53 2.53 2.53 2.53 2.53 60 x Te 1.444
-1310
-1.261 1.287 1.343
-1.367
-1.424 61 x+Ts 1.353 1.225 1.211 1.237 1.285 1.321 1350 g
Outliers 0
0 1
1 1
0 0
63 Mean
-0.045
-0.043 0.057 0.051 0.046
-0.023
-0.037 64 Sdd 0.553 0.501 0392 0 385 0.414 0.531 0.548 65 Count 19 19 18 18 18 19 19 66 k (one sided) 2.423 2.423 2.453 2.453 2.453 2.423 2.423 67 L'a 1340 1.214 0.961 0.944 1.015 1.287 1.329 68 95/95 Max 1.295 1.172 1.017 0.995 1.061 1.264 1.292 69 i
l 4
l l
i 2
[
i The D'=T/S value of 417.04 for a 0.05 significance level of normality should fall between approximately 423.6 and 439.1.
l The D* value for this test is less than the values it should fall in, which shows it has a high kurtosis.
l A high kirtosis says that the data came from a distribution that is more sharply " peaked" than the normal distribution.
The assumption of normality is rejected at the 0.05 significance level.
l l
l l
l FTRCI AI - B4 all points extrapolated to 30 months via square root method (all intervals < 3 months neglected)
The D* Test i
Ordered Xi stdev's count:
133 T-term 1 -13 880885
-2.6649 S^2:
34.22917 91.6138427 2 -13 880885
-2.6649 90.2257542 3 -1.3064363
-2.5046 T:
2439.95627 83.6119206 4 -1.3064363
-2.5046 D'=T/S:
417.04593 82.3054844 5 -13 064363
-2.5046 80.9990481 6 -1 3 064363
-2.5046 average:
-0.0310 79.6926118 7
-1.224784
-23442 stdev:
0.5092 73.48703 %
8
-0.982586
-1.8686 57.9725729 i
i 1
i i
l
AttaM1 ment 12 (cont.)
FIRCI Al - B4 l
l l
l l
l l
W test for normality - FTRCI AI - Bi point #7 extrapolated to 30 months via square root method (intervals <3 months neglected) step #
4.2.2 S^2=-
5.500 4.23 1.1135075 1
23159 0.4808
=
0.5814719 2
1.7991 03232
=
0.2469464 3
0.9643 0.2561
=
0.1536845 4
0.7464 0.2059
=
0.1M8501 5
0.65II 0.1641
=
6 0.4951 0.1271 0.0629236
=
0.0285999 7
0.3 % 9 0.0932
=
0.0113434 l
8 0.1853 0.0612
=
0.0014306 9
0.0472 0.0303
=
0 10 0.0000
=
0 11 0.0000
==
0 12 0.0000
=
0 13 0.0000
=
14 0.0000
=
0 0
15 0.0000
=
0 16 0.0000
=
i 0
17 0.0000
=
I 18 0.0000
=
0 19 0.0000 0
=
0 20 0.0000
=
21 0.0000
=
0 22 0.0000
=
0 23 0.0000
=
0 24 0.0000
=
n 25 0.0000
=
0 b=
2J0676 4.2.4 b^2=
53 2113 i'
W=b^2/S^2=
- 0. % 755 (test statistic)
I W critical value 0.9010 The assumption of normaility is NOT rejected at the.05 level.
I I
I I
i
1 3 n
133 average
-0.031 stdev 0.509 Bin Descrintion (JpperBound Expected Ohsened 1
(-e,X-2a)
-1.05 3.03 7
2 (X-20,X-(4/3)o)
-0.71 9.11 7
3 (X-(4/3)o,X-(2/3)o)
-0.37 21.45 16 4
(X-(2/3)o,X)
-0.03 32.92 26 5
(X,X+(2/3)o) 0.31 32.92 48 6
(X+(2/3)o,X+(4/3)o) 0.65 21.45 18 7
(X+(4/3)o,X+20) 0.99 9.11 9
8 (X+20,X+e) 9999.00 3.03 2
Verification of Normality - Binning Method O ExW
" h'd
~
FTRCI Al-B4 - All Points
- Etr8Polsted to 30 months via square root method (intervals <3 months neglected) 50 45 40
~
j 35 j
e y
{
30
~
(~f" o
e Y
[?
2$
E,'
T
~
E 20 n.
[?
[^[
?i Q
~
i}.;j 15 Uf h
~
t?
~
1-10
.?
(*J-3e)
(K-3e.X145pt 0141p.X411pt
@-(ISp2)
(X.K.GSpl G.00p3M45p) merespJ.3s)
(X.3eJeep Bin Range
~
T I
I i
I i
Att chm:nt 13 (cont.)
i n
19 average
-0.045 stdev 0.553 Hin Description Upper Bound Expected Observed I
(-*,X-20)
-1.15 0.43 1
2 (X-20,X.6670)
-0.41 4.36 3
3 (X.6670,X+.667o) 0.32 9.41 12 4
(X+.667o,X+2a) 1.06 4.36 3
5 (X+20,+=)
9999.00 0.43 0
Verification of Normality - Binning Method DExpected FTRCI Al-B4 - Point #2 e observed extrapolated to 30 months via square root method (latervals<3 months excluded) 10.00
- ll;N.
S 8.00 a
%h h
bl$5 6 00 -
{jlyl
t s.
3 4 00 -
n.4.s
~~
t 2 00
}.,[;p frf.,
)
3 0#F B%1-j
- h
[ N'k.
d
(-e,X-2o)
(X-2a,X.M7o)
(X.M7o,X+.M7o)
(X+.M7e,X42a)
(X+2o,+e)
Bin Range i
1 I
I I
1 I
I I
I I
L i
E I
f f
I
1 4 6
{
A i
li i
C M
N O
P Q l R
I S
T U
1 2
3 Instrument Span:
10 4
l 5
DATE Point #1 0.00 %
Pomt #2 25.00 %
Point #3 50.00 %
I 6
As Found As i.ett instrument As Found As Left Adiusted As Found As 1.cft Adiusted As Found As Left Adiusted Y
FY-RC01A1 8
3/22/90 3/22/90
-0.001 0
y 2.496 2.492 y
5 4.988 y
9 5/25/90 5/25/90 0
0 n
2.492 2.492 n
4.988 4.988 n
10 9/11/91 9/1I/91 0.003 0.003 n
2.497 2.497 n
4.997 4.997 n
11 3/16/93 3/16/93 0
0 n
2.492 2.492 n
4.994 4.994 n
12 FY-RC01B1 13 3/22/90 3/22/90 0.003 0.003 n*
2.504 2.504 n'
5.007 5.007 n*
14 5/24/90 5/24/90 0.003 0.002 y
2.506 2.498 y
5Ur 4.997 y
13 5/26/90 5/26/90
-0.001
-0.001 n
2.497 2.497 n
4.996 4.996 n
16 9/12/91 9/12/91 0.001 0.001 n
2.501 2.501 n
5.001 5.001 n
17 3/18/93 3/18/93
-0.002
-0.002 n
2.498 2.498 n
4.998 4.998 n
18 FY-RC01A2 19 3/9/90 3/9/90 0
0 n*
2.496 2.496 n*
4.996 4.996 n'
20 6/1/90 6/1/90 0
0 n
2.496 2.496 n
4.996 4.996 n
21 9/17/91 9/17/91
-0.005
-0.005 n
2.497 2.497 n
4.997 4.997 n
22 3/15/93 3/15/93 0.0008 0.0008 n
2.4962 2.4962 n
4.9968 4.9968 n
23 FY-RColB2 24 10/15/88 10/15/88 0.008 0
y 2.51 2.495 y
5.015 4.993 y
25 3/10/90 3/10/90 0.001 0.001 n
2.498 2.498 n
4.998 4.998 n
26 6/1/90 6/1/90 0.001-0.001 n
2.499 2.499 n
4.998 4.998 n
27 9/13/91 9/13/91 0.002 0.002 n
2.499 2.499 n
5 5
n 28 3/18/93 3/18/93 0.002 0.002 n
2.499 2.499 n
5 5
n 29 10/11/94 10/11/94 0.002 0
y 2.5 2.4%
y 5.001 4.994 y
30 IT-RC01A3 31 3/28/90 3/28/90 0.001 0.001 n*
2.499 2.499 n*
4.997 4.997 n*
32 5/14/90 5/24/90 0
0 n
2.496 2.496 n
4.995 4.995 a
33
- 9..d/91 9/18/91 0
0 n
2.497 2.497 n
4.999 4.999 n
34 3/16/93 3/16/93 0
0 n
2.501 2.501 n
5 5
n 35 FY-RColB3 36 3/28/90 3/28/90
-0.002
-0.002 n*
2.496 2.496 n'
4.994 4.994 n'
37 5/24/90 5/24/90 0
0 n
2.496 2.496 n
4.995 4.995 n
30 9/16/91 9/16/91
-0.001
-0.001 n
2.497 2.497 n
4.998 4.998 n
39 3/17/93 3/17/93
-0.001
-0.001 n
2.496 2.496 n
4.996 4.996 n
40 FY-RCol A4 41 3/30/90 3/30/90
-0.001
-0.001 n'
2.494 2.494 n'
4.992 4.992 n*
42 5/28/90 5/28/90
-0002
-0.002 n
2.493 2.493 n
4.991 4.991 n
43 9/17/91 9/17/91
-0.005 0
y 2.489 2.497 y
4.988 4.997 y
44 3/15/93 3/15/93 0.003 0.003 n
2.503 2.503 n
5.004 5.004 n
45 FY-RColB4 46 3/30/90 3/30/90 0
0 n'
2.495 2.495 n*
4.994 4.994 n'
47 5/28/90 5/28/90 0
0 n
2.496 2.496 n
4.994 4.994 n
48 9/14/91 9/14/91
-0.001
-0.001 n
2.492 2.492 n
4.992 4.992 n
49 3/17/93 3/17/93 0
0 n
2.495 2.495 n
4.993 4.993 n
50 51 n* indicates no adjustment was made, 52 but the test is used as a 'last adjustment' 53 point for calculation purposes i
i 1
)
+
, 4 (cont.)
A B
C V
W X
Y l
Z, i
2 3
Instrument Span:
10 4
5 DATE Point #4 75.00 %
Point #5 100.00 %
Point #6 75.00 %
6 As Found As Left Instrument As Found As LeB Adiusted As Found As Left Adjusted As Found As Left Adiusted 7
FY-RC01A1 8
3/22/90 3/22/90 7.51 7.492 y
10.026 10 y
7.511 7.492 y
9 5/25/90 5/25/90 7.491 7.491 n
9.999 9.999 n
7.492 7.492 n
10 9/11/9I 9/I1/91 7.503 7.503 n
10.009 10.009 n
7.501 7.501 n
11 3/16/93 3/16/93 7.498 7.498 n
10.008 10.008 n
7.498 7.498 n
12 FY-RC0181 13 3/22/90 3/22/90 7.511 7.511 n*
10.013 10.013 n'
7.512 7.512 n*
14 5/24/90 5/24/90 7.514 7.5 y
10.025 10 y
7.5 74 7.498 y
1W 5/26/90 5/26/90 7.498 7.498 n
9.996 9.996 n
7.498 7.498 n
16 9/12/91 9/12/91 7.503 7.503 n
10.007 10.007 n
7.503 7.503 n
17 3/18/93 3/18/93 7.499 7.499 n
10 10 n
7.5 7.5 n
18 FY-RC01A2 19 3/9/90 3/9/90 7.498 7.498 n'
10.004 10.004 n'
7.499 7.499 n'
20 6/1/90 6/1/90 7.5 7.5 a
10.005 10.005 n
7.5 7.5 n
31 9/17/91 9/17/91 7.503 7.503 n
10.008 10.008 n
7.503 7.503 n
22 3/15/93 3/15/93 7.4994 7.4994 n
10.0094 10.0094 n
7.5008 7.5008 n
33 FY-RC01B2 24 10/15/88 10/15/88 7.522 7.496 y
10.03 10 y
7.524 7.508 y
25 3/10/90 3/10/90 7.502 7.502 n
10.01 10.0 t n
7.502 7.502 n
26 6/1/90 6/1/90 7.502-7.502 n
10.007 10.007 n
7.502 7.502 n
27 9/13/91 9/13/91 7.504 7.504 n
10.01 10.01 n
7.504 7.504 n
18 3/18/93 3/18/93 7.503 7.503 n
10.01 10.01 n
7.502 7.502 n
29 10/11/94 10/11/94 7.504 7.496 y
10.011 10 y
7.505 7.496 y
30 FY-RCol A3 31 3/28/90 3/28/90 7.503 7.503 n*
10.009 10.009 n*
7.504 7.504 n*
32 5/24/90 5/24/90 7.498 7.498 n
10.002 10.002 n
7.497 7.497 n
33 9/18/91 9/18/91 7.503 7.503 n
10.009 10.009 n
7.504 7.504 n
34 3/16/93 3/16/93 7.504 7.504 n
10.012 10.012 n
7.505 7.505 n
35 FY-RCJ1B3 36 3/28/90 3/28/90 7.499 7.499 n*
10.006 10.006 n'
7.499 7.499 n'
37 5/24/90 5/24/90 7.499 7.499 n
10.003 10.003 n
7.497 7.497 n
38 9/16/91 9/16/91 7.503 7.503 n
10.012 10.012 n
7.504 7.504 n
39 3/17/93 3/17/93 7.503 7.503 n
10.007 10.007 n
7.503 7.503 n
d FY RC01 A4 41 3/30/90 3/30/90 7.491 7.491 n'
9.992 9.992 n'
7.492 7.492 n'
43 5/28/90
$/28/90 7.491 7.491 n
9.991 9.991 n
7.491 7.491 n
43 9/17/91 9/17/91 7.485 7.498 y
9.987 9.999 y
7.487 7.498 y
44 3/I5/93 3/15/93 7.507 7.507 n
10.009 10.009 n
7.507 7.507 n
45 FY-RCO1B4 46 3/30/90 3/30/90 7.495 7.495 n'
9.999 9.999 n*
7.496 7.4%
n' 47 5/28/90 5/28/90 7.495 7.495 n
9.998 9.998 n
7495 7.495 n
48 9/14/91 9/14/91 7.492 7.492 n
9.994 9.994 n
7.492 7.492 n
49 3/17/93 3/17/93 7.494 7.494 n
9.997 9.997 n
7.494 7.494 n
50 n indicates no adjustment was made.
o 51 Q3 but the test is used as a 'last adjustment' 53 point for calculation pmposes i
I f 4 (cont.)
i A
B C
Ali Al/
AG l
2 3
Instrument Span:
10 4
l 5
DATE Point #7 50.00 %
Point #8 25.00 %
Point #9 0.00 %
6 As Found As 1.cft instrument As Found As 1,ef t Ad usted As Found As 1. cit Ad usted As Found As I,cft Adiusted l
1 7
FY-RC01 A1 8
3/22/90 3/22/90 5.001 4.989 y
2.497 2.492 y
-0.001 0
y 9
5/25/90
$/25/90 4.989 4.989 n
2.493 2.493 n
0 0
n 10 9/11/91 9/11/91 4.997 4.997 n
2.497 2.497 n
0.003 0.003 n
11 3/16/93 3/16/93 4.994 4.994 n
2.494 2.494 n
0 0
n 12 FY RColBI
)
13 3/22/90 3/22/90 5.008 5.008 n'
2.505 2.505 n'
O.003 0.003 n'
14 5/24/90 5/24/90 5.01 4.999 y
2.506 2.498 y
0.003'
-0.001 y
15 5/26/90 5/26/90 4.997 4.997 n
2.497 2.497 n
-0.001
-0.001 n
16 9/12/91 9/12/91 5.001 5001 n
2.501 2.501 n
0.001 0.001 n
17 3/18/93 3/18/93 5
5 n
2.498 2.498 n
-0.001
-0.001 n
18 lY RC01A2 19 3/9/90 3/9/90 4.995 4.995 n'
2.496 2.496 n'
0 0
n' 20 6/1/90 6/1/90 4.996 4.996 n
2.496 2.496 n
0 0
n 21 9/17/91 9/17/91 4.999 4.999 n
2,498 2.498 n
0 0
n 22 3/15/93 3/15/93 4.9981 4.9981 n
2.4983 2.4983 n
0.0005 0.0005 n
23 FY-RColB2 24 10/15/88 10/15/88 5.013 4.994 y
2.508 2.495 y
"~ 0.002 0.002 n
0.007 0.001 y
25 3/10/90 3/10/90 5.003 5.003 n
2.502 2.502 n
26 6/1/90 6/1/90 4.999 4.999 n
2.499 2.499 n
0.002 0.002 n
27 9/13/91 9/13/91 5.002 5.002 n
2.502 2.502 n
0.003 0.003 n
28 3/I8/93 3/18/93 5
5 n
2.5 2.5 n
0.001 0.001 n
29 10/ll/94 10/11/94 5 001 4.994 y
2.501 2.496 y
0.002 0
y 30 FY-RC01A3 31 3/28/90 3/28/90 4.998 4.998 n'
2.499 2.499 n'
O O
n*
32 5/24/90 5/24/90 4.994 4.994 n
2.496 2.496 n
0 0
n 33 9/18/91 9/18/91 4.999 4.999 n
2.498 2.498 n
-0.002
-0.002 n
34 3/16/93 3/16/93 4.999 4.999 n
2.499 2.499 n
0 0
n 35 I'Y-RColB3 36 3/28/90 3/28/90 4.994 4.994 n'
2A96 2.496 n'
-0.001
-0.001 n'
37 5/24/90 5/24/90 4.994 4.994 n
2.496 2.496 n
0 0
n 38 9/16/91 9/16/91 4.999 4.999 n
2.498 2.498 n
0 0
n 39 3/17/93 3/17'93 4.997 4.997 n
2.496 2.496 n
-0.001
-0.001 n
40 FY-RC01 A4 41 3/30/90 3/30/90 4.993 4.993 n'
2.494 2.494 n'
-0.001
-0.001 n'
42 5/28/90 5/28/90 4.992 4.992 n
2.494 2.494 n
-0.002
-0.002 n
43 9/17/91 9/17/91 4.989 4.997 y
2A91 2.498 y
-0.004 0
y 44 3/15/93 3/15/93 5.004 5.004 n
2.503 2.503 n
0.003 0.003 n
4J IT-kC01B4 46 3/30/90 3/30/90 4.995 4.995 n*
2.496 2A96 n'
O O
n' 47 5/2090 5/2 &90 4.994 4.994 n
2A96 2.496 n
0 0
n 48 9/14/91 9/1 d Y 4.992 4 992 n
2 493 2.493 n
-0.002
-0.002 n
49 3/17/93 3/1 f/II 4.993 4.9 13 n
2.495 2A95 n
0 0
n 50 51 n* indicates no adjustmeK was made, 52 but the test is used as a 'last ad.iustment' 53 point for calculation purposes
~.-
. -. - ~. - - - _ ~ ~.
~
Attaciunent 14 (cont.)
A B
C i
AN l AO l AP AQ AR AS AT AU AV l
1 Months Since Last Test i
2 3
instrument Span:
10 4
l 5
DAE 6
&!!fgugd As 1.c.!J instrument Point #1 Point #2
['oint #3 Point #4 Point #5 l'oint #6 Point #7 Point #8 Point #9 7
lY-RColAl 8
3/22/90 3/22/90 9
5/25/90 5/25/90 2.103 2.103 2.103 2.103 2.103 2.103 2.103 2.103 2.103 I
le 9/11/91 9/11/91 15.573 15.573 15.573 15.573 ' 15.573 15.573 15.573 15.573 15.573 11 3/16/93 3/16/93 18.136 18.136 18.136 18.136 18.136 18.136 18.136 18.136 18.136 12 lY RC0181 13 3/22/90 3/22/90 14 5/24/90 5/24/90 2.070 2.070 2.070 2.070 2.070 2.070 2.070 2.070 2.070 i
15 5/26/90 5/26/90 0.066 0.066 0.066 0.066 0.066 0.066 0.066 0.066 0.066 16 9/12/91 9/12/91 15.573 15.573 15.573 15.573 15.573 15.573 15.573 15.573 15.573 17 3/18/93 3/18/93 18.168 18.168 18.168 18.168 18.168 18.168 18.168 18.168 18.168 18 lY-RColA2 19 3/9/90 3/9/90 20 6/1/90 6/1/90 2.760 2.760 2.760 2.760 2.760 2.760 2.760 2.760 2.760 21 9/17/91 9/17/91 I5.540 15.540 15.540 15.540 15.540 15.540 15.540 15.540 15.540 22 3/15/93 3/I5/93 17.906 17.906 17.906 17.906 17.9 %
17.906 17.906 17.906 17.906 23 lY RColil2 24 10/15/88 10/15/88 25 3/10/90 3/10/90 16.789 16.789 16.789 16.789 16.789 16.789 16.789 16.789 16.789 26 6/1/90 6/1/90 2.727 2.727 2.727 2.727 2.727 2.727 2.727 2.727 2.727 27 9/13/91 9/13/91 15.409 15.409 15.409 15.409 15.409 15.409 15.409 15.409 15.409 28 3/18/93 3/18/93 18.136 18.136 18.136 18.136 18.136 18.136 18.136 18.136 18.136 29 10/11/94 10/11/94 18.793 18.793 18.793 18.793 18.793 18.793 18.793 18.793 18.793 30 lY-RC01A3 31 3/28/90 3/28/90 32
$/24/90 5/24/90 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 33 9/18/91 9/18/91 15.836 15.836 15.836 15.836 15.836 15.836 15.836 15.836 15.836 34 3/16/93 3/16/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 17.906 17.906 35 lY-RC01B3 36 3/28/90 3/28/90 37 5/24/90 5/24/90 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 38 9/16/91 9/16/91 15.770 15.770 15.770 15.770 15.770 15.770 15.770 15.770 15.770 39 3/17/93 3/17/93 18.004 18.004 18.004 18.004 18.004 18.004 18.004 18.004 18.004 de IY-RC01A4 41 3/30/90 3/30/90 42 5/28/90 5/28/90 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 43 9/17/91 9/17/91 15.671 15.671 15.671 15.671 15.671 15.671 15.671 15.671 15.671 44 3/I5/93 3/15/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 17.906 17.906 45 lY-RColB4 46 3/30/90 3/30/90 47 5/28/90 5/28/90 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 j
48 9/14/91 9/14/91 15.573 15.573 15.573 15.573 15.573 15.573 15.573 15.573 15.573 49 3/17/93 3/17/93 18.070 18.070 18.070 18.070 18.070 18.070 18.070 18.070 18.070 50 i
Anacianent 14 (ConL) 11 %
HY HZ CA l Cli l CC 1 CD i CE i. CF {
CG Cll Cl Percent Dnft $mm last lost 2
K)ne Euled Interval) 3 instnanent Span:
to e
S DATli 6
As f ound As l.c.h tratrurrent Pant #1 Pmnt_#2 Pmnt #3 Pomt #4 Pomt #5 Pmnt #6 Pomt #7 Pomt#8 Point #9 1
fY.RCol Al 2/22/90 v22NO 9
$/25/90 5/25NO 0.(XOO 0.Ow0 0.0000 0.0100 0.0100 0.0000 0.0000 0.0100 0.0000 10 9/ll/91 9/l lN) 0.0300 0.0$00 0 09(x) 0.1200 0 l(x10 0.0900 0.0800 0.0400 0.03(X) 11 3/lu93 3/1993 0.0300
-0.0500
-0.0300 0.0500 0.0t00
-0.0300 0.0300 0.0300 0.0300 12 IY.RCO1111 13 V22/90 3/22/90 14
$/2440
$/24/90 ORIGO 0.0200 0.0300 0.0300 0.1200 0.0200 0.0200 0.0100 0.0000 45 S/2u90 5/2 & 90
-0.03Jo
-0.0100 4.0100 0.0200
-0 0400 0.0000
-0.0200 0.0100 0.00(x) 16 9/12NI 9/12/91 0.020')
0.0400 0.0500 0.0500 0.1100 0.0500 0.0400 0.0400 0.0200 17 3/isN3 3/1s/93 0.0300 0.0300 0.0300
-0.0400 0.0700 0.0300 0.0100
-0.0300
-0.0200 is iY.RCut A2 19 3N/90 3NNO 20 WINO
&lNO 0.0000 0.0000 0.0000 0.0200 0.0100 0.0100 0.0100 0.0000 0.0000 Il 9/17/91 9/17NI 0.0500 0.0100 0.0100 0.0300 0.0300 0.0300 0.0300 0.0200 0.0t 10 3
22 3/15/93 3/15/93 0.0580 0.0080 0.0020 0.G360 0.0140
-0.0220 0.0090 0.0030 0.0050 Y
FY.RColil2 24 10/15/8s 10/15/su 25 3/10/90 3/10$0 0.0100 0.0300 0.0500 0.0000 0.1000 4).0600 0.0900 0.0700 0.0100 26 6/1/90 6/1/90 0.0000 0.0100 0.00(x) 0.0000 0.0300 0.(xxio 0.04tx) 0.0300 0.lo)0 27 9/13/91 9/13/91 0.0400 0.0txx) 0.0200 0.0200 0.0300 0.0200 0.0300 0.0300 0.0100 2a 3/18$3 3/18/93 0.0000 OAx)O OAxio
-0.0100 0.0000
-0.0200
-0.0200 0.0200 0.0200 29 10/11/94 10/llN4 0.0tx)0 0.0100 0.0100 0.0100 0.0100 0.0300 0.0100 0.0100 0.0100 30 FY.kC01A3 31 3/28/90 3'28Nu 32 5/2490 224No
-0.0100 0.0300 0.0200 0.0500
-0.0700
-0.0700
-0.0400
-0.0300 0.00uo 33 9/18/91 9/18NI 0.00tX) 0.0100 0.0400 0.0500 0.0700 0.0700 0.0500 0.0200 0.0200 34 3/16N3 3/16/93 0.0fx10 0.0400 0.0100 0.0100 0.0itxi 0.0100 0.0(x'io 0.0100 0.0200 as FY.RColH3 36 3'2tNO 3/28/90 37 5/24$0 S/24N0 0.020u 0.0000 0.0100 0.0000
-0.0300 0.0200 0.0000 0.0(XX) 0.0100 3a 9/16/91 9/16&l 0.0100 0.0100 0.0300 0.0400 0.0900 0.0700 0.0500 0.0200 0.00f ai 39 3/l?N3 3/17/93 0.0000 0.0100
-0.0200 0.0tXso
-0.0500
-0.0100
-0.0200
-0.0200 0.0100 40 FY-RC01 A4 41 3/30/90 3/30NO 42 5/28/90 5/28/90
-0.0100
-0.0100 0.0100 0.0000 0.0100
-0.0100 0.0100 0.0000
-0.0100 43 9/l?N1 9/17/91 0.0300 0.0400
-0.0300
-0.0600
-0.0400
-0.0400 0.0300 0.0300
-0.0200 44 3/15/93 3/15N3 0.0300 0.0n00 0.0700 0.090u 0.1000 0.0900 0.0700 0.05uo 0.0300 45 FY.kColB4 46 3/30NO 3/3GNO 47 5/2kNO 5/28t90 0.0000 0.0100 0.0000 0.0000
-0.0100 0.0100
-0.0100 0.0000 0.000t>
an 9/14/91 9/1491
-0.0100 0.0400 0.0200 0.0300
-0.0400 0.0300
-0.0200 0.0300 0.0200 49 3/17$3 3/17/93 0.0100 0.0300 0.0100 0.0200 0.0300 0.0200 0.0100 0.0200 OE 50 El 52
!Li Memn
-0.001 0.004 0.010 0.uts 0.016 0.006 0.0tn
- 0. Oui 0.001 54 Satd 0.022 0.027 0.030 0.042 0.057 0.042 0.036 0.027 0.016 65 Count 27 27 27 27 27 27 27 27 27 So k (one sided) 2.26 2.26 2.26 2.26 2.26 2.26 2.26 2.26 2.26 67 k*s 0.050 0.062 0.069 0.096 0.130 0.094 0.081 0.061 0.036 58 95/95 Max 0.049 0.066 0.078 0.105 0.146 0.100 0.090 0.066 0.036 69 60 el Outher Analynas 62 T
2.66 2.66 2.66 2.66 2.66 2.66 2.66 2.66 2.66 Y
u.Ts
-0.000 0.069 0.071 0.104 0.137
-0 105
-0.087 0.068
-0.042 Y
x+1s 0.058 0.077 0.090 0.122 0.169 0.117 0.104 0.077 0.043 7
Outhers 0
0 0
0 0
0 0
0 0
7 Mean
-0.001 0.004 0.010 0.009 0.016 0#)6 0.009 0.005 0.001 7
Satd 0.022 0.027 0.030 0.042 0.057 0.042 0.036 0.027 0.016 7
Count 27 27 27 27 27 27 27 27 27 g
k (one sided) 2.26 2.26 2.20 2.26 2.26 2.26 2.26 2.26 2.26 7p h *s 0 050 0.062 0 069 0.096 0.130 0.094 0.ON I O.061 0.036 Y
95/95 Max 0.049 0.066 0.07E 0.105 0.146 0.100 0.090 0.066 0.03t.-
[1T I
{
i i
I j
i
Attachtnent 14 (cont.)
A B
C l
BF l BG l Bil Bt IU BK l BL BM j BN 1
Months Since Last Adjustment 2
3 Instrurnent bpan:
10 4
As 1'ound As Lell Instrument Point #1 Point #2 Point #3 Point #4 Point #5 Point #6 Point #7 Point #8 Point #9 7
IY-RC01A1 8
3/22/90 3/22/90 9
5/35/90
$/25/90 2.103 2.103 2.103 2.103 2.103 2.103 2.103 2.103 2.103 10 9/11/91 9/11/91 17.676 17.676 17.676 17.676 17.676 17.676 17.676 17.676 17.676 11 3/16/93 3/16/93 35.811 35.811 35.811 35.811 35.811 35.811 35.811 35.811 35.811 12 lY-RColBI 13 3/22/90 3/22/90 14
$/24/90 5/24/90 2.070 2.070 2.070 2.070 2.070 2,070 2.070 2.070 2.070 15 5/26/90 5/26/90 0.066 0.066 0.066 0.066 0.066 0.066 0.066 0.066 0.066 16 9/12/91 9/12/91 15.639 15.639 15.639 15.639 15.639 15.639 15.639 15.639 15.639 17 3/18/93 3/18/93 33.807 33.807 33.807 33.807 33.807 33.807 33.807 33.807 33.807 18 lY-RC01A2 19 3/9/90 3/9/90 20 6/1/90 6/1/90 2.760 2.760 2.760 2.760 2.760 2.760 2.760 2.760 2.760 21 9/17/91 9/17/91 18.300 18.300 18.300 18.300 18.300 18.300 18.300 18.300 18.300 22 3/15/93 3/15/93 36.205 36.205 36.205 36.205 36.205 36.205 36.205 36.205 36.205 23 IY-RColB2 34 10/15/88 10/15/88 25 3/10/90 3/10/90 16.789 16.789 16.789 16.789 16.789 16.789 16.789 16.789 16.789 26 6/1/90 6/1/90 19.515 19.515 19.515 19.515 19.515 19.515 19.515 19.515 19.515 27 9/13/91 9/13/91 34.924 34.924 34.924 34.924 34.924 34.924 34.924 34.924 34.924 28 3/18/93 3/18/93 53.060 53.060 53.060 53.060
$3.060 53.060 53.060 53.060 53.060 29 10/11/94 10/11/94 71.852 71.852 71.852 71.852 71.852 71.852 71.852 71.852 71.852 30 IY-RColA3 31 3/28/90 3/28/90
)
33 5/24/90 5/24/90 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 J
33 9/18/91 9/18/91 17.708 17.708 17.708 17.708 17.708 17.708 17.708 17.708 17.708 j
34 3/16/93 3/16/93 35.614 35.614 35.614 35.614 35.614 35.614 35.614 35.614 35.614 35 IY-RColB3 36 3/28/90 3/28/90 37 5/24/90 5/24/90 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 1.873 38 9/16/91 9/16/91 17.643 17.643 17.643 17.643 17.643 17.643 17.643 17.643 17.643 39 3/17/93 3/17/93 35.647 35.647 35.647 35.647 35.647 35.647 35.647 35.647 35.647 40 IY-RColA4 41 3/30/90 3/30/90 43 5/28/90 5/28/90 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 43 9/17/91 9/17/91 17.610 17.610 17.610 17.610 17.610 17.610 17.610 17.40 17.610 44 3/15/93 3/15/93 17.906 17.906 17.906 17.906 17.906 17.906 17.906 17.906 17.906 45 IY-RColB4 46 3/30/90 3/30/90 47 5/28/90 5/28/90 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 1.938 j
48 9/14/91 9/14/91 17.511 17.511 17.511 17.511 17.511 17.511 17.511 17.511 17.511 49 3/17/93 3/17/93 35.581 35.581 35.581 35.581 35.581 35.581 35.581 35.581 35.581 50
..s_
.. -. -.. ~. -.. -.~ ~ _.~. -, _
-.~.
~-.- -
a l
Attcchment 14 (cont.)
i l
A L
B C
BO l BP l
IM)
BR I BS BT BU BV BW 1
Percent Drift Since Last Adjustment j
Instrument Span:
10 4
5 DATE I
As.Eognd As Left Instrument Pointfl Point #2 I oint #3 Point #4 Point #5 Point #6 Point._#]
Point #8 Point #9 s
1 7
FY-RC01A1 8
3/23/90 3/22/90 9
5/25/90 5/25/90 0.0000 0.0000 0.0000
-0.0100
-0.0100 0.0000 0.0000 0.0100 0.0000 Y
9/11/91 9/11/91 0.0300 0.0500 0.0900 0.1100 0.0900 0.0900 0.0800 0.0500 0.0300 I
11 3/16/93 3/16/93 0.0000 0.0000 0.0600 0.0600 0.0800 0.0600 0.0500 0.0200 0.0000 j
12 FY-RColB1 13 3/22/90 3/22/90 14 5/24/90 5/24/90 0.0000 0.0200 0.0300 0.0300 0.1200 0.0200 0.0200 0.0100 0.0000 4
15 5/26/90 5/26/90
-0.0300
-0.0100
-0.0100
-0.0200
-0.0400 0.0000
-0.0200
-0.0100 0.0000 l
Y 9/12/91 9/12/91
-0.0100 0.0300 0.0400 0.0300 0.0700 0.0500 0.0200 0.0300 0.0200 Y
3/18/93 3/18/93
-0.0400 0.0000 0.0100
-0.0100 0.0000 0.0200 0.0100 0.0000 0.0000 3
- 3 FY-RCol A2 l
19 3/9/90 3/9/90 20 6/1/90 6/1/90 0.0000 0.0000 0.0000 0.0200 0.0100 0.0100 0.0100 0.0000 0.0000 l
21 9/17/91 9/17/91
-0.0500 0.0100 0.0100 0.0500 0.0400 0.0400 0.0400 0.0200 0.0000 l
22 3/15/93 3/15/93 0.0080 0.0020 0.0080 0.0140 0.0540 0.0180 0.0310 0.0230 0.0050 23 FY RColB2 J
24 10/15/88 10/15/88 25 3/10/90 3/10/90 0.0100 0.0300 0.0500 0.0600 0.1000
-0.0600 0.0900 0.0700 0.0100 l
r l
26 6/1/90 6/1/90 0.0100 0.0400 0.0500 0.0600 0.0700
-0.0600 0.0500 0.0400 0.0100 27 9/13/91 9/13/91 0.0200 0.0400 0.0700 0.0800 0.1000
-0.0400 0.0800 0.0700 0.0200 28 3/18/93 3/18/93 0.0200 0.0400 0.0700 0.0700 0.1000
-0.0600 0.0600 0.0500 0.0000 23 10/11/94 10/11/94 0.0200 0.0500 0.0800 0.0800 0.1100
-0.0300 0.0700 0.0600 0.0100 l
30 lY-RC0lA3 j
31 3/28/90 3/28/90 j
32 5/24/90 5/24/90
-0.0100
-0.0300
-0.0200
-0.0500
-0.0700
-0.0700
-0.0400
-0.0300 0.0000 1
33 9/18/91 9/18/91
-0.0100
-0.0200 0.0200 0.0000 0.0000 0.0000 0.0100
-0.0100
-0.0200 l
34 3/16/93 3/16/93
-0.0100 0.0200 0.0300 0.0100 0.0300 0.0100 0.0100 0.0000 0.0000 35 FY-RColB3 36 3/28/90 3/28/90
+
i 37 5/24/90 5/24/90 0.0200 0.0000 0.0100 0.0000
-0.0300
-0.0200 0.0000 0.0000 0.0100 38 9/16/91 9/16/91 0.0100 0.0100 0.0400 0.0400 0.0600 0.0500 0.0500 0.0200 0.0100 39 3/17/93 3/17/93 0.0100 0.0000 0.0200 0.0400 0.0100 0.0400 0.0300 0.0000 0.0000 40 IT-RC01A4 j
41 3/30/90 3/30/90 42 5/28/90 5/28/90
-0.0100
-0.0100
-0.0100 0.0000
-0.0100
-0.0100
-0.0100 0.0000
-0.0100 i
43 9/17/91 9/17/91
-0.0400
-0.0500
-0.0400
-0.0600
-0.0500
-0.0500
-0.0400
-0.0300
-0.0300 44 3/15/93 3/15/93 0.0300 0.0600 0.0700 0.0900 0.1000 0.0900 0.0700 0.0500 0.0300 l
4J FY-RC01B4 1
46 3/30/90 3/30/90 47 9/28/90 5/28/90 0 0000 0.0100 0.0000 0.0000
-0.0100
-0.0100
-0.0100 0.0000 0.0000 48 9/14/91 9/14/91
-0.0100
-0.0300
-0.0200
-0.0300
-0.0500
-0.0400
-0.0300
-0.0300
-0.0200 1
=
5 Y 3/17/93 3/17/93 0.0000 0 0000
-0.0100
-0.0100
-0.0200
-0.0200
-0.0200
-0.0100 0.0000 l
i 58 l
C --
i A.
4 4
3 8
2 2
6 8
2 2
6 9
8 0
0 0
0 0
0 0
0 0
0 0
0 3
5 2
8 2
7 9
4 6
9 1
1 5
7 7
9 3
7 1
8 5
3 7
6 9
2 8
1 1
2 4
3 0
5 0
1 6
3 2
0 0
0 8
8 5
0 3
2 2
0 0
0 0
0 0
0 2
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
- I
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=b 6
8 2
2 8
4 6
8 3
8 0
8 8
6 1
2 5
4 8
4 2
2 2
4 5
7 3
0 5
1 8
5 3
1 9
7 5
3 1
4 3
2 2
1 1
1 1
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
le
)c v
i e
t l
s 5
i 5
1 ta 0
t t
n s
e e
t h
s m
e t
t t
h
(
a i
e d
t tA je r
T l
5 O
I 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
6 0
N t
0 0
0 0
0 0
0 0
0 0
0 4
0 0
0 0
0 0
0 0
0 0
0 0
0 8
3 s
n 9
8 5
4 4
3 0
6 4
4 4
2 1
0 0
0 0
0 0
0 0
0 0
0 0
4 2
i io 1
1 1
1 1
1 1
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
2 9
y p
0 0
0
. 0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 9
0 t
i 4
0 l
l B
ian n
I 6
7 o
I 8
3 n
A 0
9 0
7 f
I o
C 0
e n
R 0
u o
Y
=
l v
p I
l i
a t
l F
2^
1 2
3 4
5 6
7 8
9 0
1 2
3 4
5 6
7 8
9 0
1 2
3 4
5 m
l 1
1 1
1 1
1 1
1 1
1 2
2 2
2 2
2 S
a i
y
/
c e
t 2
i l
^
t i
a
=
=
b n
=
4 m
2 2^
=
c e
B r
^
W W h o
S b
T n
I r
A o
f I
C s
t R
e 2
3 4
t p
Y e
2 2
2 F W ts 4
4 4
. 6 4
n 27 average 0.016 stdev 0.057 3
I)Jn Descrintion Unner Bound fixnected Observed I
(-e,X-2n)
-0.10 0.61 0
2 (X-2c,X.6670)
-0.02 6.20 8
3 (X.667o,X+.6670) 0.05 13.37 12 4
(X+.667o,X+2a) 0.13 6.20 7
5 (X+20,+m) 9999.00 0.61 0
Verification of Normality - Binning Method DErpected FYRCI Al-B4 - Point #5 mobserved j
14.00 -
~
12.00 e
9
~
2 10.00
~
g
.rg
.lu$.
A 8.00 F
X
~ f 6.00 m
y.
.,l i ll'
~
ll, f
~
,llf
~
~
2.00 lkV
,l
'2-
,j
, ~;
- f'
~
~
0.00 -
1 1
I I
(*,X-2o)
(X-2a,X-667c)
(X.667a,X+.667o)
(X+.667c,X+20)
(X+2o,w)
Bin Range l
I I
I I
I I
I I
1
lt i!
i,I 1'
I='
- iI l!itt!b[!
- t
.f[ $r I*1
+I bt 0
~
2 4
s -
e 6
e
_d e !-
e 4
1
^.
4
-, e ^.Im1 8
t
^.
4 -. 9 6
^.
6 9
S - _ 9 :.-_
1
=e e. ^.
^.
O
^.
. 9 r.-_
9 4
-l 4
1 t
s e s Tt 2
n 1
t
)
i s o sh aP t
7 Ll n
1 o
l cA M
e t
n
(
e n -
t s
m S
i 4
e h
B T
0 t
c e -
1 s
a m 1 a
mA L
t t
i A
TI e
c
.C n
vR iS tf Y ee irF h
D T
8 6
4 e.
o
,1 D
2 0
e O
e 0
0 0
0 0
0 0
0 0
0 0
0 0
4 2
0 8
6 4
2 0
2 4
6 8
1 1
1 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
0 0
^cL$,.f0 gag
- ii
,.^
t:
j ^
1
08 s
0 0
e.
0 07 06 0
e 0
+
0 0
O 0
0 5
tn
)
e s
h m
t s
n t
t s
o n
ui M
jo
(
d P n
t a
8 e
l m
t l 1
saA t
t s
n L -
u 0 j e
e4 4 d m
cB A
h t
n s
c a
i a
t
.+
- e*g +.,.+0 A
eA 0
0 t
e 0
c mI 0
0
+
0 O
0 n
iC iS TR 0
e 0
e
- v. Y h
F tf ir 0
D 3
0
+
0
+
0 0
2 e
0 e
0 e 0 m0 o M M o 0 4 0 4 0
0 9
0 e e
0
+
e 0
+
+
0 e
0 e
0 1
0 0
0 e 0 0 e 0
e 0
4 9
o o
0 4
2 1
8 6
4 2
0 2
4 6
8 1
1 0
0 0
0 0
0 0
0 0
0 0
0 0
y
- 0. I t 10 0
0 0
0 0
5 i,
. 9 i
l i
Binninr> Method -Iweni Drift Months Months fQunt Mean Sid
[0,3)
>0
<=3 72 0.00 0.02 Months Mondn
[t 5,20J
>I5
<=20 90 0.02 0.04 Mo.nths Months
[30,40J
>30
<=40 63 0.02 0.03 Drift v. Time Since Last adjustment FYRCI AI. B4 - All Points 0.14 0.12
-+
a e
4 0.1
+ 6-
+
0.08 0
00 6te
- e e
e
~
{0,06 -
=&
t eoesl 4
e o
- 0.04
- e4 ep
+
l 4
ee.
l t0.02 -ee em
+
0
]
1 6
e
=*
4 4
0<> -g6 1
<>9 99 44 l
0.02 <>-+
e
+
<>e e
4
-0.04 <> - +
+
+c l
4 46 l
0.06 0; c l
6
-0.08 0
10 20 30 40 50 60 70 80 -
lime Since Last Adjustment (Months)
E E
u E
E t
i 1
E i
1 LAR 95-0024 ATTACHMENT 3 FOR LICENSE AMENDMENT REQUEST NUMBER 95-0024 (69 pages follow)
LAR 95-0024 Attcch::nt 3 i
Page 1 Summary of Instrument Drift Study for j
Reactor Protection System Instrumentation i
Reactor Coolant Temperature (TERC3A2, 1ERC3A4, TERC3B2, TERC3B4) i l
The following result summary of the instrument drif t study per-formed for the RPS RTDs and linear bridge modules is applicable to Technical Specification Surveillance Requirement 4.3.1.1.1, Table 4.3-1, Functional Units 3 and 7,
Reactor Coolant High j
Temperature and Reactor Coolant Pressure-Temperature).
Each instrument string consists of an RTD, a signal converter, a linear bridge and a bistable module. The RTDs and linear bridge modules were analyzed separate from the rest of the string. The signal converter and bistable modules are tested every six-months i
with the plant on-line. Since this test frequency is not affected l
by the increase in fuel cycle duration, these modules were not included in this drift study. They were, however, part of a previous drift study to support extending the on-line surveillance test interval from one to six-months. These instrument strings also feed indicators at the Auxiliary Shutdown Panel or in the Control Room. They are evaluated in the overall result summary for Technical Specification Surveillance Require-ment 4.3.3.5.1, Table 4.3-6, Item 2 (Remote S/D Monitoring Reactor Coolant Temperature-Hot Legs, Enclosure 3).
RTD Analysis:
The RTDs were not evaluated according to the DBNPS " Instrument Drift Data Analysis Methodology and Assumptions" (DBNPS Methodology) because they are tested using the cross calibration method (Ref. NUREG/CR-5560). Data was taken during the seventh, eighth and ninth refueling outages (between 8-31-91 and 10-1-94, inclusive).
Testing was performed durirg all three plant cooldowns and once during plant heatup. The temperature range covered was 314'F to 533'F. This test's acceptance criteria for RPS RTDs is 10,5'F. The maximum deviation determined for an RPS RTD during any data run was 0.427'F, which is within acceptable limits. For data taken above 520*F (i.e., within the range of the signal processing modules and bistables),
the worst case deviation was only 0.266*F - barely over half the acceptance limit. Since RTDs are not adjusted between tests, these results provide sufficient evidence to reasonably conclude that the RPS RTDs vill continue to meet the 10.5'F acceptance criteria if cross calibration testing is performed at 24 month intervals.
LAR 95-0024 Attschunt 3 Page 2 Linear Bridge Analysis:
The linear bridge modules are calibrated every 18-months using a decade box to simulate RTD inputs. In addition, they are checked i
during functional tests performed at six-month intervals (with the plant on-line) using fixed resistance inputs from the temp-erature test modules, which are adjacent to the linear bridge modules in the RPS cabinets. Although these checks don't satisfy the formal calibration surveillance requirement because the test resistors are not H&TE with accuracy traceable to NIST, they nevertheless provide useful data for determining historical drift. Furthermore, the data from these checks is ideally suited I
for evaluating time dependency because it was collecte fre-quently (monthly through September 1994, and every six-months since then) and because only two adjustments have been made since the end of 1988. Since the data from the checks using test re-sistors was used, the data from the tests using a decade box was not used, as these two sets of data are not independent of each other. Also, conditions addressed by Potential Conditions Adverse to Quality Reports 87-0609 (vorn module frame and cabinet connec-tors) and 88-0702 (various deficiencies in the temperature test module that provided the resistance inputs) rendered invalid all channel 4 linear bridge module data taken in 1988 prior to October 31.
Linear bridge module data was taken at 0%, 50%, and 100% of span.
For all three points, the mean was 0, and the 95/95% tolerance interval was 10.04% of span. A few outlier candidates were ident-ified by the T-Test, but they were not evaluated for removal from the data set because doing so would not have impacted the results significantly (i.e., no outliers were removed from the data set).
The assumption that the data is normally distributed was tested 3
using the D' test, and the assumption was rejected. Data taken at 0%, 50%, and 100% of span was combined for this assessment. A j
histogram was created, plotting number of drift data points versus drift in percent of span. The graph had a large peak in the center, and greater than 95% of the data is within two stand-ard deviations of the mean, therefore, the normal distribution bounds the drift data and is acceptable.
Drift versus time since last test plots for all points and for each of the three individual points were created. In each plot, the vorst case drift data points observed within the first month either equal or exceed the vorst case points observed af ter the first month. However, no individual test intervals exceeded j
nine-months, so it could not be concluded from this alone that drift is time independent for test intervals up to 30 months.
Drift versus time since last adjustment plots for all points and for the individual points were also created. These plots contain considerable data for intervals beyond 30 months, with some data at intervals as high as 90 months. Since there vere only two cases of an adjustment being made after the initial data was taken (at time = 0), this data demonstrated that linear bridge
LAR 95-0024 Attcch=nt 3 Page 3 module drift is not time dependent. To quantitatively support this observation, the data for all points was divided into groups representing various ranges of calibration intervals, and the mean and standard deviation vere computed for each group. The consistency of these statistics supports the assertion that the drift is independent of time. However, since there were numerous instances on the drift versus time since last adjustment plots of data falling outside the 95/95% toleran :e interval of 10.04% of span, the 30 month projected drif t was chosen to be 10.10% of span. This is conservative, as it encompasses all the data on the 1
drift versus time since last adjustment plots.
i The results of the drif t study were incorporated by Framatome into the " Reactor Protection System String Error Calculations" l
(docummt 32-1172392-02, dated July 26, 1996) and the "DBNPS Unit 1
RPS Setpoint Allovable Values Calculation" (document 32-1257719-02, dated September 25, 1996). This resulted in less restrictive Technical Specification Allovable Values for the l
Reactos Coolant High Temperature trip and Pressure-Tempereture trip. However, no revision to the current Technical Specification j
Allovable Values for these trips is being proposed.
Historical drift values for the linear bridge module could not be directly compared with a corresponding design basis / reference uncertainty due to Framatome methods for combining uncertainties.
However, reasonable assurance that historical drift has not, except on rare occasions, exceeded acceptable limits for a cali-bration interval can be obtained by considering that the worst case drift recorded (0.10% of span) is as small as the tightest of the three calibration tolerances used (the one for input @
100% of span) and is less than half the value used in Framatome's RPS string error calculation document for the drift of the RTD/ linear bridge combination (0.23% of span). Also, during the sixth through ninth refueling outages, the channel functional tests performed for the RPS RC high temperature trip as part of the refueling channel calibrations measured the temperature signal at the linear bridge module input, thereby including the effect of its drift in the test results. There were no instances of these test results exceeding the Technical Specification Allovable Value.
The applicable surveillance and periodic test procedures were reviewed and verified to appropriately reflect all applicable conditions and assumptions of the setpoint and safety analysis.
Reactor Coolant Flov (FTRCIA1, FTRCIA2, FTRCIA3, FTRCIA4, FTRC1B1, FTRC1B2, FTRC1B3, FTRC1B4)
The following overall result summary of the drift study performed for the RPS differential pressure transmitters and I/E converters is applicable to Technical Specification Surveillance Requirement 4.3.1.1.1, Table 4.3-1, Item 4, Note 7 (RPS Flux - AFlux - Flow).
LAR 95-0024 Attachnnt 3 Enclosure'l Page 4 These instrument strings consist of differential pressure transmitters, I/E converters, square root extractors, a buffer amplifier, a function generator, and a bistable module. The differential pressure transmitters were analyzed separate from j
2 the rest of the string, as were the I/E converters. The square root extractor, buffer amplifier, function generator, and bistable modules are tested quarterly with the plant on-line.
Since this test frequency is not affected by the increase in fuel j
cycle duration, these modules were not included in this drift study.
Differential Pressure Transmitter Analysis:
Differential pressure transmitter data taken at approximately 25%, 50%, 75%, and 95% of span, in both the increasing and-de-creasing directions was utilized in this study. Data at the lov end of the range was not utilized, as it is not relevant because the RPS flux-oflux-flow trip is not needed for a loss of both i
reactor coolant pumps in the same loop as the RPS High Flux / Number of Reactor Coolant Pumps On trip will automatically trip the reactor if both RCPs in the same loop are lost. Since flow is proportional to the square root of differential pressure, 25% of differential pressure span corresponds to 50% of flow span, which is adequate to cover the lower portion of the necessary range of the instrument string. Some minor data column shif ting as described in item 5e of the DBNPS Methodology was performed because the static pressure effect on each transmitter was slightly different, resulting in calibration data being taken at percents of span that varied slightly among transmitters.
All eight RPS differential pressure transmitters were replaced during the sixth refueling outage in response to the Rosemount transmitter loss of fill oil issue addressed by NRC Bulletin 90-01. The replacement transmitters all have serial numbers above 500000, signifying that they were manufactured af ter July 11, 1989. Since these transmitters were built using an improved manufacturing
- process, data from transmitters previously installed was not used in this drift study.
For all seven test points, the sample mean (x) is less than 0.02%
of span, which can be considered to effectively be zero. The worst case 95/95%
tolerance factor times sample standard deviation (k*s) is 0.95% of span (random error). A one-sided tolerance factor was used because the RPS flux-oflux-flow trip occurs on a decreasing flow signal, so its uncertainty is of concern only if measured flow exceeds actual flow. These results were obtained after the removal of one outlier from the FTRCIA3 data. The percentage drift between the as-left data on March 16, 1993, and the as-found data on October 15, 1994, was approxi-mately -2% of span for each test point and was affected by a 3
sudden downward shift in transmitter output of about 1 mpph during the middle of 1993. Such a sudden shift in output is not representative of transmitter drift, but is reflective of abnormal transmitter behavior, therefore, this test's drift data
LAR 95-0024 Attcchnnt 3 Page 5 is considered an outlier for each test point.
The assumption that the data is normally distributed was tested by performing the De test on the drif t data for all test points and the V test on the drift data for test point #2, which was the vorst case point. The assumption of normality was rejected by the D'
test, with the D' value falling below the desired range of values. This indicates the distribution has a higher kurtosis than would be expected for a normal distribution (i.e., it's more sharply peaked). The assumption of normality was not rejected by the W test. Histograms were created, plotting the number of drift data points versus drif t in number of standard deviations from the mean, for all test points and for test point #2. The histo-gram for all test points demonstrated that the data distribution is indeed slightly more sharply peaked than a normal distribu-tion, with 94.8% of the drif t data points within two standard deviations of the mean. The histogram for test point #2 showed that the data distribution is approximately normal, with 95.5% of the drift data points within two standard deviations of the mean.
Taken together, these results support the assumption that the differential pressure transmitter drift data is normally dis-tributed.
A drift versus time since last test plot for all points was created. Most of the drif t data points were for test intervals between 15 and 20 months, so no strong evidence was provided to either support or refute the assertion that drift is independent of time. A drif t versus time since last adjustment plot for all points was also created. The drift data points for intervals longer than 30 months were all bounded by the vorst case points for intervals between 15 and 20 months, however, there vere numerous cases where the differential pressure transmitter was adjusted at an interval less than 20-months since its previous adjustment. Therefore, it could not be concluded that drift is j
time independent for test intervals up to 30 months. Also it was not clear that drift increased linearly with time. To gain additional insight into the time dependent characteristics of differential pressure transmitter drif t, the data for all drif t versus time since last adjustment points was divided into three groups representing various ranges of calibration intervals, and the mean and standard deviation were computed for each group. The results indicated that the data for intervals greater than 30 months is more conservative, with respect to both mean and standard deviation, than the 15 to 20 month interval data. These results also indicated that the mean value decreased with time since last adjustment. For this reason, a regression line was fit to the drift versus time since last adjustment plot for all points. The slope of this line was small in maggitude (-0.007%
of span per month), and the correlation was lov (R - 0.03). This provides further evidence that the transmitters exhibit very little time dependent drift.
Since there isn't strong evidence to support a conclusion of drift time dependency, but merely insufficient evidence to
LAR 95-0024 Attcch=nt 3 Page 6 clearly demonstrate a lack thereof, the drift versus time since last test data was extrapolated to a 30 month interval using the square root method. The data for test intervals less than three months was excluded from the extrapolation because the drift experienced during those short intervals would likely not have been representative of the drift occurring over at least the next 9 intervals of equal length. This was verified by observing that extrapolation of the short interval data to an 18 month interval vould produce results more unfavorable than what was obtained for the actual data with intervals between 15 and 20 months. For the extrapolated data the sample mean (x) is slightly negative but close to zero (i.e., between 0 and -0.1% of span). Since negative drift is conservative for these differential pressure trans-mitters, and because the magnitude is small, the mean vill be considered zero. The vorst case 95/95% tolerance factor times sample standard deviation (k*s) is 1.34% of span (random error).
No outliers were removed from the extrapolated data set.
The assumption that the extrapolated data is normally distributed was tested by performing the D' test on the drift data for all test points and the V test on the drift data for test point #2, which was again the vorst case point. The assumption of normality was rejected by the D' test, with the D' value falling below the desired range of values. The assumption of normality was not rejected by the V test. Histograms were created, plotting the number of drif t data points versus drif t in number of standard deviations from the mean, for all test points and for test point
- 2. The histogram for all test points demonstrates that the data distribution was again slightly more sharply peaked than a normal distribution, with 93.2% of the drift data points within two standard deviations of the mean. The histogram for test point #2 1
shows that the data distribution is approximately normal, with 94.7% of the drift data points within two standard deviations of the mean. Taken together, these results support the assumption that the differential pressure transmitter extrapolated drift data is normally distributed.
Historical differential pressure transmitter drift has exceeded its design basis / reference uncertainty in the non-conservative direction (greater than +0.35% of span) during five of the 22 calibrations for which data is available.
Historical drift exceeded the design basis / reference uncertainty in the conser-vative direction (less than -0.35% of span) during five other calibrations. These results suggest that the design basis value for differential pressure transmitter drift, 0.25% of span (random error), is not large enough to adequately characterize transmitter performance.
Potential Condition Adverse Quality Report 96-0278 was initiated to address this condition and resulted in corrections to the design basis / reference uncertainty. Of the ten cases where historical drift exceeded the design basis / reference uncertainty, none vould have exceeded the 30 month projected drift.
Furthermore, only one of the extrapolated drif t data points exceeds the 30 month projected drift and that in the conservative direction. Therefore, now that
Attcchr:nt 3 l
i Page 7 the 30 month projected drif t has been incorporated into the RPS string error calculation document, it can reasonably be expected l
that differential pressure transmitter drif t will rarely exceed l
acceptable limits.
l I/E Converter Analysis:
I/E converter data was taken at 0%, 25%, 50%, 75%, and 100% of span, in both the increasing and decreasing directions. For all nine test points, the sample mean (x) is less than 0.02% of span, which can be considered to effectively be zero. Therefore, the entire error associated with the I/E converters is considered random. The vorst case 95/95% tolerance interval maximum is 0.15%
of span. A one-sided tolerance factor was used for the same reason as in the differential pressure transmitter case. No outlier candidates were identified by the T-Test.
The assumption that the data is normally distributed was tested by performing the V test on the drif t data for test point #5, which was the worst case point. The assumption of normality was not rejected.
A histogram was created, plotting the number of drift data points versus drift in number of standard deviations from the mean, for test point #5. This demonstrated that the data distribution is approximately normal and shows that all the data is within two standard deviations of the mean.
A drift versus time since last test plot for all points was created. Most of the drif t data points are for test intervals between 15 and 20 months, so no strong evidence was provided to either support or refute the assertion that drift is independent of time. A drif t versus time since last adjustment plot for all points was also created. This plot contained considerable data for intervals beyond 30 months, with some data at intervals as high as 70 months. The drif t data points for intervals longer j
than 30 months are all bounded by the vorst case points for intervals between 15 and 20 months. This demonstrated that I/E converter drift is not time dependent. To quantitatively support this observation, the data for all drif t versus time since last adjustment points was divided into three groups representing various ranges of calibration intervals, and the mean and stand-ard deviation vere computed for each group. The results indicated that the data for intervals between 30 and 40 months is equal to, with - respect to the mean, or better than, with respect to the standard deviation, the data for intervals between 15 and 20 months. Based on the evidence discussed above, it is reasonable to conclude that I/E converter drift is independent of time.
Historical I/E converter drift did not exceed its design basis / reference uncertainty ( 0.25% of span) in any of the 27 calibrations for which data was reviewed.
The results of the drif t study were incorporated by Framatome into the " Reactor Protection System String Error Calculations" (document 32-1172392-02, dated July 26, 1996) and the "DBNPS Unit
LAR 95-0024 Attcchrnt 3 Page 8 1
RPS Setpoint Allowable Values Calculation" (document 32-1257719-00, dated August 21, 1996). This resulted in limits that are bounded by the previous Cycle 11 calculation for Flux -
AFlux/ Flow Allovable Value. These new limits will be reflected in a revision to the Core Operating Limits Report which currently reflects Cycle 9 limits which are bounded by the Cycle 11 limits.
The applicable surveillance and periodic test procedures were reviewed and verified to appropriately reflect all applicable conditions and assumptions of the setpoint and safety analysis.
Reactor Coolant Pressure i
(PTRC2A1, PTRC2A2, PTRC2B1, PTRC2B2)
The following overall result summary of the drift study performed for the RPS pressure transmitters is applicable to Technical Specification Surveillance Requirement 4.3.1.1.1, Table 4.3-1, Items 5, 6,
7, and 14 (Reactor Coolant Low Pressure, Reactor Coolant High Pressure, Reactor Coolant Pressure - Temperature, and Reactor Protection System Shutdown Bypass High Pressure).
Each instrument string consists of a transmitter, a buffer amp-lifier and a bistable module. The pressure transmitters were analyzed separate from the rest of the string. The buffer ampli-fier and bistable modules are tested every six months with the plant on-line.
Since this test frequency is not affected by the increase in fuel cycle duration, these modules were not included in this drift study. The Reactor Protection System Channel 1 and 2 pressure transmitters also feed pressurizer pilot operated relief valve (PORV) control circuitry. These strings are evalu-ated in the overall result summary for Technical Specification Surveillance Requirement 4.4.3.
Pressure transmitter data has been taken at 23 different test points throughout the 1700 to 2500 psig range. In order to make the data manageable, nine points at equal intervals were selected to characterize pressure transmitter performance. This neces-sitated a data column shif ting, as described in item Se of the DBNPS Methodology. Two exceptions were taken, however, to item Se. First some data columns were shif ted by more than 10% of span. The maximum shif ts were made downward by 12.5% of span.
Exceeding the 10% limit was unavoidable given the 23 test points involved. Secondly, af ter shif ting data columns, as-found data from one test was compared with as-lef t data from another test even when the two tests took data at different percent of span.
This was done to avoid a negative impact on the results from a large increase in tolerance factor due to a large data loss due to the frequent changes in test points.
LAR 95-0024 Attachm nt 3 Page 9 Data from several surveillance tests was adjusted to address M&TE found to be out of calibration after the test was completed.
Among the nine test points, the worst case sample mean (x) is 0.14% of span (correlated error), and the worst case 95/95% tol-erance factor times sample standard deviation (k*s) is 0.94% of span (random error). These results were obtained after the removal of two outliers from the channel 3 data for test point
- 3. These outliers (-11% of span and 11.24% of span) were at least an order of magnitude larger than all the other drift data points and resulted from what was probably a data entry error during the calibration performed on 4/15/82.
The desired transmitter output voltage was 6.548 volts, but the value recorded was 5.643 volts. It appears that the actual voltage was probably 6.543 volts, with the "6" and the "5" being transposed.
Had this value been recorded, the drif t data points would have been 0.25% of span and -0.01% of span instead of -11% of span and 11.24% of span, respectively. These values are more in line with the other drift data points both for test point #3 in general and for the channel 3 tests on 4/15/82 and 8/6/83 in particular.
The assumption that the data is normally distributed was tested by performing the W test on the drift data for both test point #1 and test point #6. These test points were selected because they produced the vorst case random and correlated errors, respec-tively. The assumption of normality was not rejected in either case. Histograms were created, plotting the number of drift data points versus drif t in number of standard deviations from the mean. Both demonstrate that the data distribution is approxi-mately normal.
A drift versus time since last test plot for all points was created. The drif t data points for test intervals longer than 22.5 months were all within the range of points for test in-tervals between 15 and 20 months, however, only three tests with intervals longer than 22.5 months were obtained. A drif t versus time since last adjustment plot for all points was also created.
Again, the drift data points for intervals longer than 22.5 months are all bounded by the vorst case points for intervals between 15 and 20 months. Although more drift data for intervals beyond 22.5 months was available than was the case for the drift versus time since last test plot, there were still many cases where the pressure transmitter was adjusted at an interval less than 20 months since its previous adjustment. Given these obser-vations, it cannot be concluded that drift is time independent for test intervals up to 30 months. Also, it was not clear that drift increases linearly with time. To gain additional insight into the time dependent characteristics of pressure transmitter drift, the data for all drif t versus time since last adjustment points was divided into three groups representing various ranges of calibration intervals, and the mean and standard deviation were computed for each group. The results ?ndicated that while there was a slight tendency for the mean to increase with time when comparing the 15 to 20 month interval data with the data for
LAR 95-0024 Attachunt 3 Page 10 intervals greater than 22.5 months (including intervals up to approximately 50 months),
the standard deviations for these groups are almost equal.
Since there was not strong evidence to support a condusion of drift time dependency, but merely insufficient evidence to clearly demonstrate a lack thereof, the drift versus time since last test data was extrapolated to a 30 month interval using the square root method described in item 14c the DBNPS Methodology.
The data for test intervals less than two months was excluded from the extrapolation because the drift experienced during those short intervals vould likely not have been representative of the
)
drift occurring over the next (at least) 15 intervals of equal length. This was verified by observing that extrapolation of the short interval data to an 18-month interval vould produce results more unfavorable than what was obtained for the actual data with 1
intervals between 15 and 20 months. For the extrapolated data the vorst case sample mean (x) is 0.25% of span (correlated error),
and the vorst case 95/95% tolerance factor times sample standard deviation (k*s) is 1.24% of span (random error). No outliers were removed from the extrapolated data set.
The assumption that the extrapolated data is normally distributed was tested by performing the V test on the extrapolated drif t data for both test point #1 and test point #6. These test points were selected because they produced the vorst case random and correlated errors, respectively. The assumption of normality was
]
not rejected in either case. Histograms were created, plotting I
the number of drift data points versus drift in number of stand-ard deviations from the mean. Both demonstrate that the extrap-j olated data distribution is approximately normal.
The results of the drif t study were incorporated by Framatome into the " Reactor Protection System String Error Calculations" (document 32-1172392-02, dated July 26, 1996) and the "DBNPS Unit 1
RPS Setpoint Allowable Values Calculation" (document 32-1257719-02, dated September 25, 1996). This resulted in less restrictive Technical Specification Allowable Values for the Reactor Coolant Lov Pressure, Reactor Coolant High Pressure and Reactor Coolant Pressure-Temperature trips. However, no revision j
to the current Technical Specification Allovable Values is being i
proposed for these trips. The Allovable Value calculated by Framatome for the Shutdown Bypass High Pressure trip remains at the current Technical Specification value.
Historical drift has exceeded the design basis / reference uncertainty (greater than +0.65% of span or less than -0.725% of span) during six of the 37 calibrations for which data is available. Also, during the eighth refueling outage, the channel i
functional test performed for the RPS channel 1 RC high pressure trip as part of the refueling channel calibration resulted in a i
trip slightly above the Technical Specification Allowable Value (2355 psig). The field setpoints for the RPS RC High Pressure, Lov Pressure, Pressure-Temperature, and Shutdown Bypass High
4 i
j
-LAR 95-0024 l
Attschunt 3 l
Enclosure'l l
Page 11 i
P Pressure trip functions have all been made more conservative as corrective action for that event, and no further violations of j
Allowable Values have occurred. In addition, of the six cases where historical drift -exceeded the design basis / reference i
6 uncertainty, none would have exceeded the 30 month projected j
drift. Furthermore, none of the extrapolated drif t data points j
exceed the 30 month projected drif t. Therefore, now that the 30 month projected drif t has been incorporated into the RPS string error calculation document, it can reasonably be expected that j
drift vill rarely, if ever, exceed acceptable limits.
l The applicable surveillance and periodic test procedures were -
l reviewed and verified to appropriately reflect all applicable conditions and assumptions of the setpoint and safety analysis.
i 1
1 i
i i
i i
i r
4
LAR 95-0024 Attechu nt 3 Page 1 Summary of Instrument Drift Study for Steam and Feedvater Rupture Control System Steam Generator Level Low (LTSP9A6, LTSP9A7, LTSP9A8, LTSP9A9, LTSP9B6, LTSP9B7, LTSP9B8, LTSP9B9)
There are no other identical instrument strings in the plant.
Each instrument string consists of a level transmitter, an associated signal converter, and a bistable. Other instruments fed by the transmitter or by the signal converter are not anal-yzed here. The level transmitters are located in containment with the remainder of the instrument string located in the Control Room Cabinet Room.
Because of the availability of vast amounts of monthly functional test data available for the bistable trip point and a reasonable amount of transmitter data from each refueling outage, it was decided to analyze the transmitter data separate from the signal converter through bistable data. Although transmitter through bistable data was not analyzed, it was hoped that the large number of functional test data points would provide a small tolerance factor and when combined with the transmitter alone data, would result in a lover overall 30 month projected drif t.
It should be noted that because the functional test surveillance interval is not being extended, time dependency was not
)
evaluated for the bistable functional test data.
Transmitter Analysis:
The range of the transmitters was changed from 388 inches to 250 inches via Mod 85-0157 in October 1988. Therefore, the previous data taken on these strings is not included in the study.
Additionally, since 1988, there have been a few minor changas to the calibrated ranges (much less than 10% of span) of the trans-mitters, and the drif t data conservatively contains appropriate discontinuities.
1 During each calibration of a transmitter, output current data is i
recorded as the voltage dropped across a test resistor in the SFRCS cabinet. Additionally, if a transmitter requires adjust-I ment, transmitter output data is taken across a test resistor installed in the current loop at the transmitter. After the adjustment is made, as-left data is typically taken in both locations. Since there are more data points available in histor-
)
ical data for the location at the cabinet (taken every calibra-tion), these points were used in the drif t study as opposed to the transmitter data at the transmitter.
l
LAR 95-0024 Attech=nt 3 Page 2 Since the lov level bistable trips at 9.4% of span, decreasing percent indication or increasing differential pressure between data points 8 (24% of span) and 9 (4% of span), the largest 95/95 Hax. drift since last test value of the two points vill be conservatively utilized as the 30 month projected drif t if no time dependency is found. Additionally, since the k*s values were the same for points 8 and 9, and the k*s values for points 8 and 9 vere larger than the 95/95 Max. value, and point 9 is closer to the bistable trip point than point 8,
point 9 data vill be individually examined for normality and time dependency as well as all data points together. No outliers vere identified in the Test to Test drift data.
For point 9 (4% of span) plotting the number of data points versus the number of standard deviations from the mean was performed. The comparison to that of the normal distribution indicates that the data is approximately normal. Although only 91.3% of the points fell within two standard deviations, the data passed the V Test. Also, if only one more point had fallen within two standard deviations, over 95% vould have been within the two i
standard deviation bounds. For the purposes of this study, the assumption of normality is justified.
j For all data points plotting the number of data points versus the number of standard deviations from the mean was performed. The comparison to that of the normal distribution indicates that the i
data is approximately normal. Nearly 95% (94.2%) of the points fell within two standard deviations and the data passed the D' Test. For the purposes of this study, the assumption of normality is justified. The assumption of normality is therefore confirmed 4
for the transmitter data.
The Drift vs. Time Since Last Test plot for point 9 (4% of span) indicated a slight non-conservative increase over the short in-terval of approximately 16.5 to 19 months. However, the overall mean of point 9 was negative, as is the slope of the Drif t vs.
2 Time Since Last Adjustment plot. Additionally, a low R value in both plots supports the conclusion that the drift with respect to time is not time dependent, but random.
Additional support for time independence was provided by the slight decrease in standard deviation for the greater than 30 months Since Last Adjustment group of data when compared to the less than 30 months group. However, since only eight of the 23 data points were lef t unadjusted past the approximate 18 month interval, many potentially poor drift data points did not appear in the longer time intervals. Thus, it may not be concluded that the drift is independent of time.
For all data points the Drift vs. Time Since Last Test plot indicated a slight non-conservative increase of the short inter-val of approximately 16.5 to 19 months. However, the overall means of all points were negative as was the slope of the Drift 2
vs. Time Since Last Adjustment plot. Additionally, a low R value s
-_ _~ __ _ _ _ _ _ _ _ _
LAR 95-0024 Attechmnt 3 Page 3 in both plots supports the conclusion that the drift with respect to time is not time dependent, but random.
As opposed to the slight decrease in standard deviation for Point 9 data, support for time dependence was provided by the slight increase in standard deviation for the greater than 30 months Since Last Adjustment group of data when compared to the less than 30 months group. Again, since only 72 of the 135 data points vere left unadjusted past the approximate 18 month interval, many potentially poor drift data points did not appear in the longer l
time intervals. Thus, it may not be concluded that the drift is i
independent of time.
Since it could not be concluded that the drift is independent of time, the Since Last test data was extrapolated to 30 months.
Linear extrapolation was considered overly conservative and Square root extrapolation was used. This is based on the reasons, given above which indicate the drift may be conservative or time independent.
The largest k*s or 95/95 Max. drift value from the Square Root l
Extrapolated all Test-Test data to 30 months statistics for points 8 and 9 was 0.446 percent of span. This is the 30 month projected drif t value for the transmitter. This is equivalent to 1.115 inches of the evaluated 0-250 inch range.
The point 9 square root extrapolated data was tested for nor-mality as was done for the original test-test data. The results, similar to those discussed for the non-extrapolated data above and indicated the assumption of normality is acceptable for the purposes of this study.
1 Bistable Analysis:
From functional test data for the eight bistables, as measured across an installed test resietor at the current loop input to the SFRCS cabinet, basic statistics were calculated. As was done with the transmitters, a one-sided tolerance factor was used. The desired functional test trip value was consistent throughout all the tests since October 1988. Prior to that time, the SFRCS cabinet and string components were different model numbers (replaced via Mod 87-1107). Therefore, only the functional test trip data since October 1988, was utilized.
Although 11 outliers were identified in the functional test data, the high outliers were conservatively removed and the lov out-liers were conservatively lef t in the data set for calculating basic statistics. The resulting 95/95% Min. value of -0.000589 VDC (0.015% of the 1 - 5 VDC range) conservatively represents the vorst case expected drift value over a (typically less than) one month time interval. Since time dependency was not evaluated, it does not necessarily represent the vorst case expected drift value for a 125% times one month interval.
LAR 95-0024 Attechmrnt 3 Page 4 j
The bistable data, including all outliers, failed the D' Test.
l Since the calculated D'
value was less than either critical value, it can be concluded that the data had high kurtosis, or more sharply peaked than a perfectly normal distribution, A histogram of the expected and observed data points with renpect to the number of standard deviations from the mean confirms this conclusion. Similarly, the data, with only the high outliers removed failed the D' Test and similarly has high kurtosis.
Since the high kurtosis conservatively reflects a more than normal distribution of good points and nearly 95% of the points were covered by two standard deviations, the assumption of nor-mality is assumed for purposes of this study.
Although not evaluated for time dependency, these bistable basic statistics were considered in the calculation of a new Allowable Value.
Design Basis Evaluation:
Calculation C-ICE-083.03-001 Rev. 8 changed the setpoint from >
15.6" to 1 16.9".
Modification 95-0062 rescaled/recalibrated the instrument strings associated with the SFRCS Steam Generator Low Level trip from a 0 to 250 inch range to a 0 to 300 inch range during the tenth refueling outage. The range change was required due to the in-creased fouling of the steam Generators. As a Steam Generator fouls, the indicated startup, operate, and full range levels increase. The modification required no change in the actual physical components utilized in the instrument strings.
The results of this drift study, based on the 0 to 250 inch range, were reflected into the new design basis calculations for the increased range of the SFRCS Steam Generator Low Level trip.
This resulted in a change to the SFRCS Steam Generator Low level trip allowable value. The new allowable value is 216.9 inches.
LAR 95-0024 Attach:nnt 3 Page 1 SummaryofInstrumentDriftStug for Remote Shutdown System Instrumentation Reactor Coolant Temperature - Hot Legs (TERC3A4 and TERC3B2 as read on TIRC3A4 and TIRC3B2)
Each instrument string consists of four elements: an RTD, a linear bridge module, a signal converter, and a temperature indicator.
The linear bridge modules are analyzed separately in the Reactor Protection System enclosure and found to have a 30 month pro-jected drift of 20.1% of span. The RTDs are also analyzed in the Reactor Protection System Instrument Drift Study (Enclosure 1) and vere evaluated in a manner different than that prescribed in the DBNPS " Instrument Drift Data Analysis Methodology and Assumptions," document (DBNPS Methodology) and found to have a drift of 10.5% of span.
The auxiliary output isolation amplifier of the signal converter was not evaluated using historical drif t, but on a qualitative basis, using engineering judgment. This is discussed further below.
Temperature Indicator Analysis:
For regular tests, the temperature indicators are tested by injecting a voltage signal into the indicator and reading the temperature value displayed. This value is compared with the desired temperature reading to determine if the instrument needs adjustment. If an adjustment is necessary, the zero screw on the indicator is manipulated to provide the desired output. In addition, in some tests a reverse calibration is performed. In these instances, the voltage input is adjusted until the desired output is reached. The technician then records the input voltage required to produce the desired output. This value is compared with the desired input voltage to determine if the instrument needs adjustment. For calculation purposes, data taken during reverse calibration tests was converted to data that could be trended with the values from tests where a regular calibration was performed. The conversion process is as follows. The voltage found is converted to a percent of span and then multiplied by i
the span of the instrument in degrees. This determines the offset (in degrees) that the voltage represents. Second, the calculated
)
offset value is added to the bottom of scale value. Third, the degree value calculated is reflected about the desired temper-ature to determine what the temperature value vould have been had a regular calibration been done. This reflection is necessary because of the inherent differences between a reverse calibration and a regular calibration. If the voltage necessary in a reverse calibration is lover than desired, for example, the temperature
LAR 95-0024 Attschm2nt 3 Page 2 reading would have been higher than expected had a regular cali-bration been done.
Because there were so few data points on the two temperature indicators alone, data from other instruments with identical makes and model numbers was added. First, two control room in-dicators (TIRC3A2 and TIRC3B4) were added. These indicators are members of identical strings and are scaled to the same range as the original two indicators. Also, both the auxiliary shutdown panel (ASP) room (where TIBC3B2 and TIRC3A4 are located) and the control room are environmentally controlled, so variations in drift data due to different environmental conditions would be negligible. Second, several indicators located on the ASP vere added. The additional instruments used in the drift study are two steam generator start up level indicators (LISP 9A3 and LISP 9B3),
two steam generator outlet steam pressure indicators (PISP12A1 and PISP12B1), and two reactor coolant pressurizer level indi-cators (LIRC14-1 and LIRC14-2). The additional indicators are all Edgewise 180 indicators, like the temperature indicators. The indicators located on the ASP vere chosen because they are oper-ating under the same environmental conditions as temperature 4
indicators TIRC3A4 and TIRC3B2. Some of the additional indicators have a different input range (-10 to 10 volts versus 0 to 10 volts). The. manufacturer was consulted and confirmed that the indicators have identical physical mechanisms and therefore, should have identical drift characteristics.
i Fourteen different percent span points have been used during the time period of interest to satisfy the nine point calibration check of the temperature indicators. To provide a manageable data 4
set, 9 of the 14 points were selected to characterize indicator performance. This necessitated data column shifting, as described in item Se of the DBNPS Methodology. All of the indicator data (including the supplementary indicators on the ASP) was shif ted so as to line up in Lino columns. However, one exception to the DBNPS Methodology item 5e was made. Originally, only the four temperature indicators were considered in the study. It was found that the results were not representative of the actual drif t characteristics (due to the large increase in tolerance factor) due to data lost as a result of the frequent changes in test points. Therefore, the as-found data from one test was compared with the as-lef t data from the previous test even when the two tests took data at different percent of span. When the additional indicators were added, these cross-column comparisons were not deleted. A check of the complete data set shows that there are no significant differences between the set where the cross-column comparisons are included and the set where they are not.
Therefore, cross-column comparisons vere utilized in this drif t study. Among the nine, after shifting, data points, the vorst case 95/95% interval is 2.507% of span.
The assumption of normality was tested by performing the V test on the worst case point. A histogram plotting the number of drift data points versus drift (in number of standard deviations from
i LAR 95-0024 i
Attachm:nt 3 l
Page 3 l
the mean) was created. Since the V test passed and the histogram shows that the data is approximately normal, the assumption of normality is verified for this data set.
i i
Previously testing of the indicators alone was not required when i
the string check passed. Since the string check rarely failed, j
there are several occasions where indicator test to test in-i tervals exceed one hundred months. These long points show that the instrument has some time dependent drift, as they are not bounded by the peints of lesser intervals. The degree of time dependence is mitigated by the fact that on four of the seven occasions where the indicator test to test interval exceeded 100 months, the indicator was not adjusted when it was tested. In order to further assess the amount of time dependency, the i
percent drift since last adjustment data was grouped into bins and the mean and standard deviation of each of these bins was l
calculated. The results show that the time dependent drift is less than linear. The mean appears to fluctuate randomly, and the f
only significant change in standard deviations is between the lesser intervals (those up to 50 months) and the interval greater than 100 months. Ther2 fore, the data was extrapolated to 30 i
months using the square root method described in item 14c of the i
DBNPS Methodology. One exception to standard drift projection practice vas made.. Those points exceeding 100 months were not-included in the extrapolated set because they would not be re-presentative of expected drif t over a 30 month calibration in-terval. Therefore, only points with intervals between 2 and 50 months were included in the data. Points with intervals less than two month:. vere excluded in accordance with item 14d of the DBNPS Methodology. The extrapolated data has a worst case 95/95% in-terval of 2.710% of span.
The assumption of normality for the extrapolated data set was tested by performing the W test on the vorst case point. Also, a histogram plotting the number of drif t data points versus drif t t
(in number of standard deviations from the mean) was created.
i Because the W test passes and the histogram demonstrates that 95%
l of the data is within 2 standard deviations, the assumption of normality is verified for this data set.
i Signal Converter Analysis:
1 i
The signal converters are currently tested every 18 months.
Before 1990, the modules were tested on a monthly basis via a i
three point check. In order to most efficiently utilize all of the available data, it was decided to analyze the three points for which the greatest amount of data is available (100% of span, 50% of span, and 0% of span). Among the three data points, the vorst case 95/95% interval is t0.182% of span.
6 1
- - _ _ ~. -
-. ~ ~ - _. - -. -. - - _ -
.~.
1 i
- LAR 95-0024 j
Attachmznt 3 i
l l
Page 4 The assumption of normality was tested by running a D' test on i
all percent drift since last test points. The D' test failed low, indicating high kurtosis in. the data. The D' test was supple-4 j
mented by creating a histogram and plotting the number of drift data points versus drif t, in number of standard deviations from the mean. The histogram shoved that the distribution is bounded by normality, as 271' of 280 (96.8%) of the. terms are. within 2 j
standard deviations. Therefore, the assumption of normality is i
verified for this data set.
i 1
j A percent drift since last test graph was developed. Several data points, all found in May 1990, indicated a large positive drift at approximately 7,10 and 17 months since last adjustment. Since the deletion'of these points as outliers could not be justified, l
i the data is assumed to be time dependent. However, since there j
are only three tests (out of n; ore than 100) that indicate some j
time dependency, the data was extrapolated to 30 months via the j
square root method. The worst case 95/95% tolerance interval is l
10.69% of span.
i i
The assumption of normality for the extrapolated data set was tested by running a D' test and plotting a histogram. The D' test j
and histogram both indicated that the distribution of these points appears to be non-normal because the data set is not bounded by two standard deviations and has high kurtosis.
j i
However, 93.6% of the terms are within two standard deviations so i
the non-conservative error introduced by treating this sample as I
normal is very small. Therefore, engineering judgment concludes that the conservative assumptions made elsewhere, the time de-i pendence assumption above and the assumed drif t of the auxiliary i
output isolation amplifier that follows, more than account for j
the potential error introduced by the distribution of the sample.
1-One deviation from normal procedure was involved in analyzing the I
signal converters. The path of interest runs from the module j
input through the auxiliary output-isolation amplifier which is a parallel path with the primary output isolation amplifier af ter L
the module input. However, the only instrument data available was j
taken from the module input through the primary output isolation amplifier. Since the gain takes place in a common amplifier, the j
two paths will have nearly identical drifts. Also, the only means of adjusting this instrument is via an operational amplifier common to both paths. However, to be conservative, the drift of l
the auxiliary output isolation amplifier was assumed to be 1%
i of span, random and independent of the 10.69% of span drif t for 2
the rest of the module. The value of 1% of span, obtained via engineering judgment, was chosen because this value is reasonably conservative with respect to the amount of drift that the aux-iliary output isolation amplifier vould experience. It is ex-pected.that this vill account for any performance differences between the two isolation amplifiers.
/'
-s u
.i
~
---e.,.y--
%--y..s w
-y
---y--,,,-
LAR 95-0024 Attachm:nt 3 Page 5 Total String Analysis:
The 30 month projected drift of 3.01% of span for these instrument strings was determined by combf.ning the individual drift values for the various components utilizing the square-root-sum-of-the-squares technique.
The design basis / reference uncertainty of 2.32% of span was obtained by combining individual terms for the various components.
Historical signal converter drift exceeded acceptable limits only three times out of 101 calibrations for which data was reviewed.
Historical indicator drift exceeded acceptable limits only twice out of 36 calibrations for which data was reviewed.
Although the 30 month projected drif t for the string (3.01% of span) exceeds the design basis / reference uncertainty (2.32% of span), the Auxiliary Shutdown Panel Reactor Coolant hot leg temperature strings are still considered acceptable for use in effecting a safe shutdown. The purpose of these instruments being located at the Auxiliary Shutdown Panel is to support decay heat i
removal via the auxiliary feedvater system and the main steam safety valves and/or atmospheric vent valves. This is required to maintain the plant in Mode 3,
hot
- standby, under normal non-accident conditions.
Davis-Besse Nuclear Power Station procedure, DB-OP-06903, " Plant Shutdown and Cooldown," section 7.0, "Cooldown on Natural Circulation," provides the required guidance for how the Reactor Coolant hot leg temperature instru-ments are used to support this function. The establishment of natural circulation heat removal is verified by observing the following indications:
The difference between RC hot and cold leg temperatures is stable at less than 50
'F.
Incore thermocouple and Reactor Coolant hot leg temperature indications are coupled and tracking.
The 30 month projected drif t of 3.01% of span, which is approx-imately 3
'F, is judged to be acceptable for using the Auxiliary Shutdown Panel Reactor Coolant hot leg temperature indications in accomplishing these two verifications. The typical difference between Reactor Coolant hot and cold leg temperatures during natural circulation is approximately 30
'F, so 3 'F of RC hot leg temperature drift would not impair the operator's ability to assess whether or not natural circulation heat removal had been established.
These instrument strings are not included in any trip setpoint calculations and there are no conditions or assumptions of set-point and/or safety analyses that must be reflected in plant surveillance test procedure acceptance criteria.
i LAR 95-0024 j
Attachmtnt 3 Page 6 Reactor Cool; int au. tem Pressure (PT6365A and PT6365B as read on PI6365A1 and PI6365B1)
The two instrument strings consist of a transmitter two con-wrters, (PY6365A1/B1 and PY6365A3/B3), and a pressure indicator.
No safety related setpoints are associated with these instrument I
strings. The 30 month projected drift was determined to be 20.997% of span. The design basis / reference uncertainty was determined to be il.3688% of span. Historical drift data was i
acceptable except on rare occasions.
The assumption of normality was tested using the V test. The assumption of normality was not rejected by the V test for the vorst case point. The data for the vorst case point and all points appeared normal when plotted on a histogram. For the vorst case data point set, no points were greater than two standard deviations from the mean. The histogram for all points appeared normal. but only 93 of 99 (93.94%) of all data points are within 2 i
standard deviations. This is very close to the required 95% thus it is considered normal. This supports the assumption of a normal i
distribution.
The assumption of normality was also tested for the data extrap-olated to 30 months. The assumption of normality was not rejected by the V test for the vorst case point. The data for all points and for the worst case point appeared normal when plotted on a histogram. For the worst case data point set, no data points were greater than 2 standard deviittions from the mean. For all points, 61 out of 63 (96.83%) were within 2 standard deviations of the mean. Only 95% of all points are required to be within 95% of the mean. This supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment. The long term plots of drift versus time since last adjustment did not show any time dependent characteristics, however, only 1 data point set is beyond 30 months. Since only one data point was longer than 30 months, all drift versus time since last test data points greater than 5 months were linearly extrapolated to 30 months. Only points that vere originally longer than 5 months were used because points less than 5 months vould add error which would be large and not representative of drift. This was demon-strated by an analysis of the two groups of data (points less than 5 months and points greater than 5 months), which determined that both the mean and the standard deviation are smaller for the points greater than 5 months. This indicates that drift does not increase when going from the points less than 5 months to the points greater than 5 months, therefore, it is acceptable to use only the points greater than 5 months for determining the 30 month projected drift.
Since the 30 month proj ected drift is less than the design basis / reference uncertainty, all current assumptions based on the
LAR 95-0024 Attachmant 3 Page 7 design basis / reference uncertainty are still applicable and conservative because the 30 month projected drif t is bounded by the design basis / reference uncertainty.
Pressurizer Level 1
(LTRC14-1 and LTRC14-3 as read on LIRC14-1 and LIRC14-2)
The 30 month projected drif t of 25.386% of span is greater than the design basis / reference Uncertainty of il.6038% of span. The 30 month projected drif t is a worst case value with a 95/95%
confidence factor and is conservative. This tolerance interval was based on a sample set containing only 5 points. This gives a tolerance factor of 5.079.
The historical drift data was acceptable except on rare occasions.
l The assumption of normality was tested using the W test. The assumption of normality was not rejected by the V test for points 1 and 9. The data for points 1 and 9 appears normal when plotted on a histogram. For these data point sets, no data points were
)
greater than 2 standard deviations from the mean. This supports 1
i i
the assumption of a normal distribution.
The drift data that was extrapolated to 30 months was also tested using the V test for the worst case point. The assumption of normality was not rejected by the V test for the vorst case point. The data for the vorst case point appeared normal when plotted on a histogram. For the worst case data point set, no data points were greater than 2 standard deviations from the mean. This supports the assumption of a normal distribution.
The drift data was plotted to show the relationship between the Drift (percent of span) versus the time since last adjustment and drift versus time since last test. These instrument strings have gone over 35 months without requiring an adjustment, however, only one test is longer than 30 months. Since only one data point was longer than 30 months, all drift versus time since last test data points greater than 5 months were linearly extrapolated to 30 months. Only points that vere originally longer than 5 months
)
were used because points less than 5 months vould add error which would be large and not representative of drif t.
Therefore, the drift was determined to be time dependent.
I Although the 30 month projected drif t is greater than the design basis / reference uncertainty, these instruments are still con-sidered acceptable. The operator actions associated with these instruments involve controlling pressurizer level during a plant shutdown from the Auxiliary Shutdown Panel following a control room fire or an evacuation of the control room. During these events, the pressurizer level vould be controlled at approxi-mately 100 inches. This number was chosen because it is well above the level of the pressurizer heaters and well below the top of the pressurizer thus it minimizes the possibility of the
LAR 95-0024~
Attachmsnt 3 Page 8 I
pressurizer going solid (overfilling the pressurizer) or un-covering the pressurizer heaters. The error associated with LIRC14-1 and LIRC14-2 is not large enough to cause the pres-surizer level to increase or decrease significantly enough to cause the pressurizer to overfill or to uncover the pressurizer heaters.
No calculations or safety related setpoints are associated with LIRC14-1 or LIRC14-2 nor is this instrumentation i
related to any safety analysis assumptions. Based on this, these instruments are considered acceptable to provide Pressurizer Level indication.
Steam Generator Outlet Steam Press (PTSP12A2 and PTSF12B1 as read on PISP12A1 and PISP12B1)
The 30 month projected drif t of i 3.632 psig is greater than the 4
design basis / reference Uncertainty of i26.35 psig. The 30 month projected drif t from the drift--study data is a worst case value with a 95/95% confidence factor and is conservative. This tol-erance interval was based on a sample set containing only 9 points. This gives a tolerance factor of 3.532. Historical drift was acceptable except on rare occasions.
The assumption of normality was tested for the drif t data using j
the V test. The assumption of normality was not rejected by the V test. The data appeared normal when plotted on a histogram. For the worst case data point set, no points were greater than 2 standard deviations from the mean. This supports the assumption of a normal distribution.
The assumption of normality was also tested for the data extrap-olated to 30 months. The assumption of normality was not rejected by the V test for the vorst case point. The data for all points and for the worst case point appeared normal when plotted on a histogram. For the worst case data point set, no data points were greater than 2 standard deviations from the mean. For all points, 71 out of 75 (94.67%) were within 2 standard deviations of the mean. This is considered acceptable for a normal distribution as 95% of the points are _ required to be within two standard devia-tions for a normal distribution. This supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment and drif t versus time since last test. These instrument strings have gone 58 months without requiring an adjustment. Regression analysis was also performed on all points and on the worst case point for drift versus time since last test and indicated that the drift is not time dependent. The standard deviation increased slightly as the. time since last adjustment increased. Although the increase is small, it does seem to indicate that there is some time de-pendent drift. To be conservative, the data was extrapolated to 30 months using all points greater than 2 months. All drift versus time since last test data points greater than 2 months
LAR 95-0024 Attcchm:nt 3 Page 9 vere extrapolated to 30 months using the square root method. The data was extrapolated to 30 months using the square root sum of squares method because the time dependent drift associated with PISP12A and PISP12B is clearly not linear based on the analysis of the data (Regression Analysis and plots of Drift Versus Time Since Last Adjustment). Since the drift is not linear, the square root sum of squares method of data extrapolation is considered a conservative method to extrapolate the data. Only points that were originally longer than 2 months were used because points less than 2 months would add error which would be large and not representative of drift.
Pressure indicators PISP12Al and PISP12B1 are located in the Auxiliary Shutdown Panel. These indicators are used to place the plant in a safe shutdown condition from the Auxiliary Shutdown Panel if the control room becomes inaccessible. Steam Generator pressure indication supports decay heat removal via the Auxiliary Feedvater system and the Main Steam Safety Valves or Atmospheric Vent Valves. Based on engineering judgment, the 30 month pro-jected drif t of 43.63 psig is acceptable because it vould not prevent the operator from removing decay heat via the Steam Generators (SG), nor would it prevent them from adequately con-2 trolling SG pressure. If SG pressure indication experienced the vorst case drift in either direction, the operator vould observe the discrepancy between indicated pressure and the expected Main Steam Safety Valve and/or Atmospheric Vent Valve setpoints. The operator would then control SG pressure at a different indicated value to compensate for the instrument error.
Steam Generator Startup Range Level (LTSP9A3 and LTSP9B3 as read on LISP 9A3 and LISP 9B3)
The 30 month projected drif t of 14.143% of span is greater than the design basis / reference Uncertainty of i.65% of span. The 30 l
month projected drif t from the 24 month drif t study data is a worst case value with a 95/95% confidence factor. This tolerance interval was based on a sample set containing only 8 points. This gives a tolerance factor of 3.732. Historical drif t was accep-table except on rare occasions. Also, any error introduced by M&TE was not included as it was considered negligible when com-pared to the other uncertainty terms.
The assumption of normality was tested using the V test. The assumption of normality was not rejected by the V test for the vorst case point. The data for all points appeared normal when plotted on a histogram. For the worst case data point set and for all points, no points were greater than 2 standard deviations from the mean.
This supports the assumption of a normal distribution.
The assumption of ne mality was tested using the V test for the data extrapolated to a0 months. The assumption of normality was not rejected by the V test for the vorst case point. The data for
Attrechm:nt 3 Page 10 all points appeared normal when plotted on a histogram. For the vorst case data point set and for all points, no points were greater than 2 standard deviations from the mean. The data for all points also passed the D prime test. This supports the assumption of a normal distribution, j
The data was plotted to show the relationship between the Drif t I
(percent of span) versus the time since last adjustment and drift versus time since last test for all points. There was a decrease in the standard deviation as the time since last adjustment increased. Regression analysis was performed on all points and on the vorst case point for drift versus time since last test and for drift versus time since last adjustment. For drif t versus time since last test the regression analysis indicated that the data is not time dependent. These instrument strings have gone over 30 months without requiring an adjustment. The long term plots of drif t versus time since last adjustment do not show any time dependent characteristics. However, only one data,acint is longer than 30 months for the vorst case point. Sir:ee only one point was longer than 30 months, all data was extrapolated to 30 months using the square root method. The square roat method was chosen because the time dependent drift associated with LISP 9A3 and LISP 9B3 is clearly not linear. This is based on regression analysis and plots of drift versus time since last test and drift versus time since last adjustment.
Although the 30 month projected drif t is greater than the design basis / reference uncertainty, these instruments are still capable of being used to indicate steam generator level. No operator actions are taken based on these instruments during normal opera-tions. These indicators are used when the control room is evacu-ated to maintain steam generator level at 49 inches (post reactor trip). The 30 month projected drif t is 24.143% vhich is 210.36 inches. The 49 inch level is a guide based on convenience and maintaining some inventory in the steam generators. There are no precise level requirements that need to be maintained. The pur-pose of these instruments is to maintain the plant in a safe shutdown condition. A safe shutdown condition can be maintained with the steam generator level at 49 110.36 inches. No calcula-tions or safety assumptions use these instruments, therefore, no calculations or safety analysis require changes.
_ _ - _ ~
LAR 95-0024 Attach: Int 3
~Page 1 Summary of Instrument Drift Study for Post-Accident Monitoring System Instrumentation Steam Generator Outlet Steam Pressure (PTSP12A2 AND PTSP12B1 as read on PISP12A and PISP12B)
The 30 month projected drif t of -22.548 psig (a one side tol-erance factor was utilized) exceeds the design basis / reference uncertainty of 111.76. The 30 month proj ected drift from the drift study data is a vorst case value with a 95/95% confidence factor and may be too conservative. This tolerance interval was based on a sample set containing only 5 points. This gives a tolerance factor of 4.203.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Steam Generator Outlet Steam Pressure indicators.
The assumption of normality was not rejected by the V test. The data appeared normal when plotted on a histogram. For the vorst case data point set, one data point was greater than 2 standard deviations from the mean, however, a large peak of data points at the center of the distribution causes the standard deviation to be very small. This may be the cause of the single point being more than two standard deviations from the men A large peak of points at the center of the distribution is cc-
- vative, thus it is bounded by the normal distribution.
This supports the assumption of a normal distribution.
The assumption of normality was also tested for the data extrap-olated to 30 months. The assumption of normality was rejected by the V test for the vorst case point. This failure may have been caused by the large peak of data at the center of the distri-bution. Four of the five points are at the center of the distri-bution. A large peak of data at the center of the distribution is conservative, thus it is bounded by the normal distribution. The data for all points and for the vorst case point appeared normal when plotted on a histogram. For the vorst case data point set, no data points were greater than 2 standard deviations from the mean. For all points, 42 out of 45 (93.33%) were within 2 standard deviations of the mean. For a normal distribution 95% of the points are required to be within two standard deviations, however, this distribution also has a large peak of data at the center of the distribution (29 of 45 points). A large peak of data at the center of the distribution is conservative, thus it is bounded by the normal distribution.
This supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment. The long term plots of drift versus time since last adjustment did
LAR 95-0024 AttatchmInt 3 Page 2 not show any time dependent characteristics, however, only 2 data point' sets are beyond 30 months. Since only two data points were longer than 30 months, all drift versus time since last test data points greater than 2 months were extrapolated to 30 months using the Square Root Sum Of Squares method. The data was extrapolated to 30 months using the Square Root Sum Of Squares method because the time dependent drift associated with PISP12A,B vas clearly not linear based on the analysis of the data (Regression Analysis and plots of Drif t Versus Time Since Last Adjustment). Since the drift is not linear, the Scuare Root Sum Of Squares method of data extrapolation is considered a conservative method to extrap-olate the data. Only poin'.s that were originally longer than 2 months were used becaus2 points less than 2 months would add error which would be h rge and not representative of drift.
These indicators provide Regulatory Guide 1.97, type A variable information to the operators during a Steam Generator Tube Rupture design basis accident. They are used for controlling cooldown rate during RCS cooldown and depressurization to 500 'F and 1000 psig. They are also used to control SG pressure between 980 and 1000 psig, in accordance with Emergency Operating Procedures.
While the drift is larger than the design basis / reference uncertainty, it is span independent meaning that the relationship between the points does not change as the instruments drift.
There does not appear to be any drift component that affects the relationship between the individual points over the range of the i
instruments. The basic statistics calculated for these instrument strings support the assumption that the drift is span independ-ent. No individual point is significantly different than adjacent points. Based on this, it is acceptable to use the Steam Gener-
)
ator Pressure Indicators to show a change in pressure (an increasing or decreasing trend) but not the actual pressure for 1
controlling RCS cooldown rate.
The SG pressure control band allows 50 psig of margin to the Main Steam Safety Valve (MSSV) setpoint of 1050 psig. The accident analysis assumes that after the affected SG is isolated, no additional release of radioactivity occurs, therefore, it is necessary to verify that the 30 month projected drif t of 22.55 psig vill not cause the total instrument string uncertainty to become large enough to challenge the MSSV setpoint. A calculation vill be created for this purpose. If 50 psig of margin is in-sufficient, the control band vill be lowered in the applicable procedure.
Reactor Coolant Loop Outlet Temperature (TERC3A5, TERC3A6, TERC3B5 and TERC3B6 as read on TIRC3A5, TIRC3A6, TIRC3B5 and TIRC3B6)
The temperature elements (TERC3A5,A6,B5,B6) in these four instru-ment strings (two per loop) is a passive device (RTD) and is not
e f
l LAR 95-0024 Attachnent 3-l j
-Page 3 f'
calibrated. Since the RTD is not a part of the calibration there J
is no as-found or as-lef t data associated with it. Also,
.the RTDs are periodically replaced under the station Environmental Qualification program. Any error associated with the. temperature 4
element is not included in this comparison.
}
The 30 month projected drif t of
.3285% is less than the design
_1 i
basis / reference uncertainty of t.6161%. The data does not exceed i
1 acceptable limits for the calibration interval. No outliers were i
i removed from the data. Historical data has been acceptable except l
on rare occasions.
l l
Since the 30 month projected drift is less than the design i
basis / reference uncertainty, all current assumptions based on the j
i design basis / reference uncertainty are still applicable and conservative because the 30 month projected drif t is bounded' by t
the - design basis / reference uncertainty. No calculations, cali-brations, functional tests, or channel checks require alteration.
~
I The assumption of normality was tested using the W test. The data for point seven (the worst case data point) failed the V test.
I This failure was caused by a large peak of data points around the I'
center of 'the distribution. This indicates that the instruments perform better than expected. Also, the distribution of all l'
points and point seven with respect to the standard deviation was examined. These distributions of the points are affected by the digital resolution of the readout. The indicator only reads to
^
the nearest degree which means any error vill cause the indicator h
to be off by one degree.
This causes the drift values to occur in discrete multiples of 0.125% of span when there is any error.
j~
When the distribution of all points with respect to the standard.-
deviation was examined, they fell exactly on 0.00, 1.25 or 2.50 standard deviations from the mean.-.This plot also showed a large peak at the center of the distribution indicating that the in-j
.ar9ments perform better than expected. The distribution'is not j
graal bec since the peak occurs at ' the center of' the distri-bution it is conservative thus it is bounded by the normal dis-tribution. All points for the worst case point vers within 2 i
standard deviations of the mean. For all points, 120 of 134
}
points (94.02%) were within 2 standard deviations. _ This is very close to the required 95% and is considered acceptable due to the i
small amount of drift associated with these instrument strings.
l Since the performance of the instruments is bounded by the normal distribution, the assumption of normality is valid.
-]
a The data was plotted to show the relationship between the Drif t j
(percent of span) versus the time since last adjustment and Drift i
(percent of span) versus time since last test. At no point in j
j time did the data exceed design basis / reference uncertainty.
These instruments are rarely adjusted and do not show any time i
dependent characteristics and very little drift. These terp-
)
erature strings have gone 75 months without requiring an
}
adjustment. The absolute value of the Drif t (percent of span)
(
t
LAR 95-0024 Attschnnt 3 Page 4 versus the time since last adjustment was plotted. No increase vas seen in the absolute value of th+ drif t as time increased indicating that there is no time dependent component of drif t associated with this instrument string. There was a decrease in the standard deviation as the time since last adjustment increased. The standard deviation does not increase as a function of time since last adjustment. The data indicates that the temperature strings vill remain relatively constant as a function of time.
The Reactor Coolant Loop Outlet Temperature indicators may be used for manual calculation of subcooling margin if the sub-cooling margin monitors are unavailable. Currently, no calcula-tion exists supporting the use of these indicators for this purpose. This use is being evaluated and applicable procedures vill be revised as necessary.
l Reactor Coolant Loop Pressure Monitors dxtended Range Instrument Strings:
(PT6365A and PT6365B as read on PI6365A and PI6365B):
Since the 30 month projected drif t of il.333% of span is less than the design basis / reference uncertainty of 11.7632% of span, all current assumptions based on the design basis / reference uncertainty are still applicable and conservative.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the RCS Loop Pressure Indicators.
l The assumption of normality was not rejected by the V test. The data for all points and for the worst case point appeared normal when plotted on a histogram. For the vorst case data point set, j
one data point was greater than 2 standard deviations from the mean, however, this is not unusual since 0.64 points were expec-ted to be greater than two standard deviations from the mean. For all points, 120 out of 126 (95.24%) were within 2 standard devia-tions of the mean. Only 95% of all points are required to be within 2 standard deviations of the mean. This supports the assumption of a normal distribution.
The assumption of normality was also tested for the data extra-polated to 30 months. The assumption of normality was not re-jected by the V test for the worst case point. The data for all points and for the vorst case point appeared normal when plotted on a histogram. For the vorst case data point set, no data points were greater than 2 standard deviations from the mean. For all points, 66 out of 72 (91.67%) were within 2 standard deviations of the mean. This is close to the 95% required for a normal distribution and is considered acceptable. This supports the j
assumption of a normal distribution.
)
LAR 95-0024 Attechm:nt 3 Page 5 The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment. The long term plots of drift versus time since last adjustment do not show any time dependent characteristics, however, only 1 data point j
set was beyond 30 months. Since only one data point was longer 1
than 30 months, all drift versus time since last test data points greater than 5 months were linearly extrapolated to 30 months.
Only points that were originally longer than 5 months were used because points less than 5 months would add error which would be large and not representative of drift. This was demonstrated by an analysis of the two groups of data (points less than 5 months and points greater than 5 months), which determined that both the mean and the standard deviation are smaller for the points greater than 5 months. This indicates that drift does not in-crease when going from the points less than 5 months to the
)
points greater than 5 months, therefore, it is acceptable to use only the points greater than 5 months for determining the 30 month projected drif t.
The Reactor Coolant System Loop Pressure indicators may be used for manual calculation of subcooling margin. Currently, no calcu-lation exists supporting the use of these indicators for this purpose. This use is being evaluated and applicable procedures vill be revised as necessary.
)
l Vide Range Instrument Strings:
(PTRC2A4 and PTRC283 as read on PIRC2A4 and PIRC2B3)
These strings are associated with Safety Features Actuation System (SFAS) Channels 1 and 2.
Channels 3 and 4 of SFAS have virtually identical instrument strings providing the same in-dications, however, they are not utilized for post accident monitoring. Two extended range (0 - 3000 PSIG) pressure strings also support Technical Specification Requirement Table 4.3-10, Item 3.
However, along with range differences, the components within the extended range strings are unlike the SFAS strings.
Therefore, the extended range strings were not used within this specific drift study for the SFAS RCS Pressure Vide Range Indications. There are no other instrument strings in the plant identical to the SFAS strings.
Each instrument string consists of a pressure transmitter, an associated current - to - current isolator, and a pressure in-dicator. Except for the current - to - current isolator the model numbers of all components in the four SFAS strings are the same.
Not shown in the diagrams are other instruments fed by the trans-mitter or by the signal converter.
Channels 1 and 4 of SFAS contain replacement current - to -
current isolators, Consolidated Controls model 6N548, installed in October 1991 and March 1993, respectively. The other channels contain Fisher and Porter model 50EK1000 (the original isolators
=
LAR 95-0024 Attachunt 3 Page 6 supplied with the system). Although a drif t value is not speci-fled within either the original or the replacement isolator vendor literature, the replacement isolator was chosen by the 4
original supplier of the SFAS system (Consolidated Controls) as an appropriate replacement. Since the replacement is built of electrical subcomponents at least one decade never than those of the original, it is expected that the drift of the replacement is better than that of the original. The specified accuracy value for both the original and the replacement isolators is the same (0.25% of full scale). Additionally, with the assumption that drift of a higher accurate device (such as 0.25% for the i
isolator) is less than that of a less accurate device (1%
accuracy for the indicator and 0.65% accuracy for the trans-mitter), it is expected that the drift of the isolator vill be much less than that of the remaining components in the string.
For these reasons, the drift of the replacement was assumed to be comparable to or better than the original and data for the four strings was considered to be from the same statistical population for the purposes of this study.
The history of the strings showed the original Foxboro E11GH transmitters were replaced with Environmentally Qualified Foxboro NE11GH transmitters in May and June of 1986. Since the new trans-mitter's internal parts are different than the old transmitter's and the new transmitter has a performance uncertainty larger than the replaced transmitter (0.65% of calibrated span vice 0.5%), it was conservatively decided to use only string test data obtained cince the new transmitters were installed.
Since there were only two outliers identified in the 108 drif t data values calculated, and neither was significantly outside the outlier criteria, no outliers were removed. Also, there were no outliers identified for the vorst case Point 4.
The data for the worst case point, Point 4, passed the V Test.
The data for All Points did not pass the D' Test, having a lov D' value indicating high kurtosis (too much data near the mean).
Plotting the number of data points versus the number of standard deviations from the mean was performed for both Point 4 and All Points. The comparison to that of the normal distribution in-dicated that both sets of data were approximately normal. Both sets had over 95% of their points within 2 standard deviations.
Based on these results, for the purposes of this study, the assumption of normality is justified.
For the short interval of 16 to 20 months, the Drif t vs. Time Since Last Test plot for Point 4 indicated a slight decreasing trend and a possibly decraasing spread of data for increasing time. The similar plot for All Points indicated much the same.
However, the Drif t vs. Time Since Last Adjustment plot for Point 4 indicated a slight increasing trend. The similar plot for All Pgints indicated a slight decreasing trend. There is a vgry lov R value in three of the four plots and a relatively low R value in the fourth plot. T' e standard deviations of bins of data below
LAR 95-0024 Attcchm:nt 3 Page 7 and above 22.5 months indicated the spread improves with in-creasing intervals. Due to there being only three of the possible twelve Since Last Adjustment data points having intervals past approximately 20 months, potentially bad drift data may have been masked by making string adjustments. For this reason, time independence is not justified and the test to test data vill be extrapolated to 30 months by the square root method.
The largest 95/95 Max. drift value from the extrapolated 30 month Test-Test data for all points (one sided tolerance intervals) is 2.818 percent of span or 70.45 PSIG. This is the 30 month projected drif t value for the indicators when used for monitoring decreasing pressure scenarios.
The largest 95/95 Min. or 95/95 Max. drift value from the extrapolated 30 month Test-Test data for all points (two sided tolerance intervals) is 3.449 percent of span or 86.225 PSIG.
This is the 30 month projected drif t value for the indicators when used for monitoring both increasing and decreasing pressure scenarios.
The Point 4 square root extrapolated data was tested for nor-mality as was done for the original test-test data. The results, similar to those discussed for the non-extrapolated data above indicated the assumption of normality is acceptable for the purposes of this study.
The All Points extrapolated data was similarly tested for no-rmality. The results were again similar to those discussed for the non-extrapolated data above and added further justification for the assumption of normality.
The the Design Basis Reference Uncertainty for normal operating conditions was determined to be 234.168 PSIG (or 1.367% of span).
For LOCA/HELB Conditions the Design Basis Reference Uncertainty was determined to be 30.056 PSIG (or 1.202% of span).
These instrument strings are used to verify compliance with Technical Specification 3.4.9.1, Reactor Coolant System Pressure / Temperature Limits.
The Technical Specification pressure / temperature curves are error corrected for the instru-ment string uncertainties to generate the curves utilized in DBNPS operating procedures. These procedures vill be revised as necessary. No change to Technical Specification 3.4.9.1 is re-quired.
The Reactor Coolant System Loop Vide Range Pressure indicators may be used for manual calculation of subcooling margin.
Currently, no calculation exists supporting the use of these indicators for this purpose. This use is being evaluated and applicable procedures will be revised as necessary.
l l
LAR 95-0024 Attachmsnt 3 Page 8 A review of historical drift data found that there vere few exceptions to the drift data exceeding that of the smaller design i
basis / reference uncertainty value determined, above (1.202% of span).
Surveillance test procedures were reviewed and found to be con-sistent with the design assumptions.
Pressurizer Level 1
(LTRC14-1 and LTRC14-3 as read on LIRC14-3 and LIRC14-4)
The 30 month projected drift of 3.483% (111.15 inches) is greater than the design basis / reference uncertainty of ii.607%.
The 30 month projected drift is a worst case value with a 95/95%
confidence factor and is conservative. The tolerance interval was based on a sample set containing only 7 points. This gives a drift correction factor of 4.007. Also, any error introduced by M&TE was not included as this uncertainty was considered to be negligible when compared to the other uncertainties.
j Historical drift has been acceptable except on rare occasions.
No safety related setpoints are associated with the Pressurizer Level Indicators.
The assumption of normality was not rejected by the V test for the worst case point. The data for the vorst case point and all points appeared narmal when plotted on a histogram. For the worst case data point set, no points were greater than 2 standard deviations from the mean. For all points, 90 of 97 points (92.78%) of all points were within 2 standard deviations. This is very close to the required 95% and is considered acceptable. This supports the assumption of a normal distribution.
The assumption of normality was tested for the data extrapolated to 30 months. The assumption of normality was not rejected by the V test for point 1 or for point 5. The. data for all points, point 5 and point 1 appeared normal when plotted on a histogram. For the data sets of point 1 and point 5, no points were greater than 2 standard deviations from the mean. For all points, 74 of 79 points (93.67%) of all points were within 2 standard deviations.
This is very close to the required 95% and is considered accep-table. This supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment and drift versus time since last test. These instrument strings have gone over 35 months without requiring an adjustment, however, only 2 tests are longer than 35 months. Since only two data point sets were longer than 30 months, all drift versus time since last test data points greater than 5 months were extrapolated to 30 months using the square root method. The square root method was chosen based on an analysis of the standard deviation as time increased.
LAR 95-0024 Attechm:nt 3 Page 9 The standard deviation decreased as the time since last adjust-ment increased. Also, the results of regression analysis did not indicate that the drif t was time dependent. Based on this, the square root method of data extrapolation is a conservative method of extrapolating the data. Only points that were originally longer than 5 months were used because points less than 5 months vould add error which would be large and not representative of drift. It is considered acceptable to use only the points greater than 5 months for determining the 30 month projected drift.
One outlier was removed from the data extrapolated to 30 months for point #2. The percent drift for the data taken in October 1994 did not appear to be valid for point #2. All points for that test were symmetric in percent drift based on the value where the reading was taken (i.e. values at points 1 and 9, points 3 and 7 and points 4 and 6 did not show any hysteresis ef fects) except for points 2 and 8 vhich showed a large difference between the two. The value of point 2 had the largest percent change of any point taken for all of the data. Based on the large discrepancy between points 2 and 8 and the large amount of change in point 2, this point is considered an outlier and was not included in the analysis.
The pressurizer level instruments are used to determine pres-surizer level if Non-Nuclear Instrument (NNI) power is lost to prevent the pressurizer heaters from being uncovered or the pressurizer from going solid (full) during normal operations or severe transients (accidents).
Most actions do not require precise level indication only that level be maintained in the operating region of the pressurizer (100 inches) which is well above the level of the pressurizer heaters and well below the top of the pressurizer. Thus minimizing the possibility of over-
)
filling the pressurizer or uncovering the pressurizer heaters.
The drift associated with LIRC14-3 and LIRC14-4 (3.483% or 211.15 inches) is not large enough to cause the pressurizer level to increase or decrease significantly enough to overfill the pres-surizer or to uncover the pressurizer heaters when the level is l
controlled at 100 inches. The only actions taken based on the exact readings of the instruments are based on readings taken at the lov end of the instrument range (50 inches decreasing). The points at the lov end of the instrument range performed signifi-cantly better than the vorst case point which was at 100% of scale (320 inches). Using data from the points at 5% and 25% of scale (16 and 80 inches) provides results which are more reason-able and reflect the actual instrument range used for manually l
controlling pressurizer level if NNI power is lost. The worst case 95/95% max./ min. for the points at 5% and 25% of scale is 2.154% (data extrapolated to 30 months) which is 16.893 inches.
This value is greater than the design basis / reference uncertainty (1.607%). The top of the pressurizer heaters is located at 28 l
inches (compensated) which is approximately 43.8 inches uncomp-ensated. The current setpoint for turning off the heaters is when
LAR 95-0024 Attcchm:nt 3 Page 10 indicated level (uncompensated) reaches 50 inches which cor-responds to 40 inches compensated level. The value of the uncompensated setpoint (50 inches) minus the error (6.893 inches) is 43.107 inches which is less than 43.8 inches. Based on this analysis, the pressurizer heaters may become uncovered if the 50 inch (uncompensated) setpoint is maintained. The current setpoint of 50 inches vill be revised to assure the pressurizer heaters remain covered.
Steam Generator Startup Range Level Steam and Feedvater Rupture Control System Associated Instrument Strings:
(LTSP9A6 and LTSP9B6 as read on LISP 9A6A and LISP 9B6A)
Two of the eight instrument strings (LISP 9A6A and LISP 9B6A) associated with the Steam and Feedvater Rupture Control System (SFRCS) steam generator level are required Post-Accident Monitoring Instrumentation. Data for theses strings was combined with two identical strings (LISP 9A8A and LISP 9B8A) for analysis.
The tenths digit is not utilized on LISP 9A6A and LISP 9B6A but is utilized on LISP 9A8A and LISP 9B8A. The tenths digit is the least significant digit of the instruments. This created two sets of data, one set with the tenths digit and one set without the tenths digit. The two sets of data were analyzed using hypothesis testing to determine if the two sets of data could be combined (i.e. vere from the same distribution). The F statistic was used for testing the equality of the two variances. The test statistic is equal to the variance of the smaller group divided by the variance of the larger group. This value was less than the critical statistic determined based on the F statistic vith (17,
- 62) degrees of freedom. Since the test statistic was less than the critical statistic, the hypothesis was not
- rejected, therefore, the data could be combined.
Prior to 1994, the test inputs specified in the data packages for the four instrument strings utilized in this analysis were changed less than 0.3% of span. The effects of these changes were analyzed to determine if any changes or adjustments in the data vere required. All data from the 1994 tests was adjusted by the correction factors obtained using the original input specified in the data packages and the new input from the data packages (see attached tables). The data for all of the indicators changed when the correction factor was included and the data was rounded to the resolution of the indictors. Based on this, the data cannot be used in this instrument drift study because the changes that resulted were not insignificant when compared to the accuracy of the string. The changes that resulted from the test input changes may affect the results of the drift analysis. These points were counted as an adjustment to the instrument string so the as-left data from the previous test was not compared to the as-found data of the first test with the new input values.
LAR 95-0024 Attachmant 3 Page 11 The 30 month projected drif t of 21.744% of span is greater than the Design Basis Reference Uncertainty of i.522% of span. The 30 month projected drif t from the 24 month drif t study data is a vorst case value with a 95/95% confidence factor. The 30 month projected drift is based on 8 data points which gives a tolerance factor of 3.732.
No outliers were removed from the data.
Historical drift has been acceptable except on rare occasions. No l
safety related setpoints are associated with the St7am Generator Startup Range Level Indicators.
The assumption of normality was not rejected by the W test for the worst case point. The data for all points appeared normal when plotted on a histogram. For the worst case data point set and for all points, no data points were greater than 2 standard deviations from the mean. The data for all points also passed-the D' test. This supports the assumption of a normal distribution.
The data extrapolated to 30 months was also tested by the V test.
The assumption of normality was not rejected by the W test for the vorst case point. The data for all points appeared normal when plotted on a histogram. For the vorst case data point set and for all points, no data points were greater than 2 standard deviations from the mean. The data for all points also passed the D' test. This supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drift (percent of span) versus the time since last adjustment and drift versus time since last test. Regression analysis was also per-formed on all points and on the worst case point for drift versus time since last test. Regression analysis performed indicated that the data is not time dependent. These instrument strings have gone over 35 months without requiring an adjustment, however, only two points of the vorst case data are longer than 30 months. Since only two points were longer than 30 months, all data except for points less than 1 month were extrapolated to 30 months using square root extrapolation. The square root method was chosen based on regression analysis and the fact that the i
standard deviation of all points decreased with time for a plot of drif t vs. time since last adjustment. Points less than one month were not included since they would add error that would be large making the 30 month projected drift overly conservative.
Although the 30 month projected drif t is greater than the design Basis Reference Uncertainty, no operator actions are taken based on these two instrument strings during normal operations. No calculations, safety related setpoints or safety analysis are associated with these strings. These indicators vill still dis-play steam generator level. Based on this, it is acceptable for the 30 month projected drif t of these indicators to be greater than the design basis / reference uncertainty.
Level transmitters LTSP9A6 and LTSP9B6 and indicators LISP 9A3 and LISP 9B3 vere recalibrated/rescaled from a 0 to 250 inch range to a 0 to 300 inch range by DBNPS Modification 95-062 during the
1 LAR 95-0024 Attachm nt 3 Page 12 i
tenth refueling outage. The range change was required due to the increased fouling of the Steam Generators. As the Steam Gener-i ators foul the indicated startup, operate and full range levels increase. The resulting increased error in indicated level is negligible and therefore, the indicators remain acceptable.
Post Accident Monitoring Instrument Strings:
j l
(LTSP9A3 and LTSP9B3 as read on LISP 9Al and LISP 9B1)
The 30 month proj ected drift of 16.123% is greater than the design basis / reference uncertainty of 11.65%. The 30 month projected drif t from the 24 month drif t study data is a worst case value with a 95/95% confidence factor. The 30 month pro-jected drift is based on 8 data points which gives a tolerance factor of 3.732. Also, any error introduced by M&TE was not included as it was considered negligible when compared to the other uncertainty terms.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Steam Generator b rtup Range Level Indicators.
The assumption of normality was not rejected by the V test for the vorst case point. The data for all points and the worst case point appeared normal when plotted on a histogram. For the vorst case data point set, no points were greater than 2 standard deviations from the mean. For all points, 78 of 81 points (96.30%) were within two standard deviations. Only 95% of the points are required to be within 2 standard deviations for a j
i normal distribution. This supports the assumption of a normal distribution.
The assumption of normality was tested for the data extrapolated to 30 months. The assumption of normality was not rejected by the V test for the vorst case point. The data for all points and the worst case point appeared normal when plotted on a histogram. For the worst case data point set, no points were greater than 2 standard deviations from the mean. For all points, 68 of 72 points (94.44%) were within two standard deviations. This is very close to 95% vhich is the number of points required to be within 2 standard deviations for a normal distribution and is considered acceptable. This supports the assumption of a normal distri-bution.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment and drift versus time since last test. Regression analysis was performed on all points and on the vorst case point for drift versus time since last test. The regression analysis performed indicates that the data is not time dependent. The long term plots of drif t versus time since last adjustment did not show any time dependent characteristics. However, only one data point is longer than 30
LAR 95-0024 Attcchm:nt 3 Page 13 months. Since only one point was longer than 30 months, all da'a was extrapolated to 30 months using the square root method. The square root method was chosen based on regression analysis and the plots of drift versus time since last adjustment which do not indicate that the data is linearly time dependent.
Although the 30 month projected drift is greater than the design basis / reference uncertainty, these instruments are still capable of being used to control steam generator level. There are various actions taken based on the indicated level, however, these actions do not require exact indication. The setpoints or action levels associated with indicators LISP 9Al and LISP 9B1 vere chosen based on convenience not precise levels. Only the dry steam generator level (16 inches per DBNPS Procedure) could be affected by the error associated with LISP 9Al and LISP 9B1, however, other indications are also used (pressure, temperature) to determine if the steam generator is actually dry. This is also not affected as the 16 inch level is a guide. No calculations or safety essump-tions use these instruments, therefore, no calculations or safety analysis require changes. Based on this, it is acceptable for the 30 month projected drift of these instrument strings to be j
greater than the design basis / reference uncertainty.
i Containment High Range Radiation Monitors (RE4596A and RE4596B as read on RI4596AB and RI4596BB)
The radiation elements (RE4596A and RE4596B) are passive devices and are not calibrated. Operability of these instruments is verified using a radioactive source and verifying that the re-sponse is within 20% of a desired value. Twenty percent is the total error of the instrument string as determined by the vendor.
The response values have always been vell within 20% of their desired value. This check verifies the response of the radiation element and the entire instrument string.
The 30 month projected drif t of il.973% is greater than the design basis / reference uncertainty of 1.5%.
the 30 month projected drif t from the 24 month drif t study data is a worst case value with a 95/95% confidence factor. This tolerance in-terval was based on a sample set containing only 9 points. This gives a drift correction factor of 3.532. also, any trror intro-duced by M&TE was not included as this infornation was not avail-able.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Containment High Range Radiation Indica-tors.
The assumption of normality was rejected by the V Test for the vorst case point. For the vorst case data point set, one data point was greater than 2 standard deviations from the mean,
LAR 95-0024-Attachm:nt 3 Page 14 however, a large peak of data points at the center of the distri-bution caused the standard deviation to be very small.
This may be the cause of the single point being more than two standard deviations from the mean. for all data points, 50 out of 54 (92.59%) points were within 2 standard deviations of the mean.
However, 49 of 54 points were at the mean or zero standard deviations from the mean. This creates a large peak of data at --
the center of the distribution which causes the standard devia-tion to be small. A large peak of data at the center of the distribution is conservative thus it is bounded by the normal distribution. this supports the assumption of a normal distri-bution.
The data was plotted to show the relationship between the drif t (percent of span) versus the time since last adjustment and drift versus time since last test. There was a decrease in the standard deviation as the time since last adjustment increased. Regression analysis was also performed on all points and on the worst case point for drift versus time since last test. The regression analysis indicates that the data is not time dependent. These instrument strings have gone over 70 months without requiring an adjustment. In fact, these instruments have not required any adjustments since they were installed. The long term plots of drift versus time since last adj ustment do not show any time dependent characteristics. This justifies the assumption that the instruments are not time deper. dent.
Although the 30 month projected drift is greater than_the design basis reference uncertainty, these instrument strings have never-required an adjustment. The instrument drift has always been small and has never taken the instruments out of their desired tolerance. The performance of these instruments has been ex-ceptional. The 30 month projected drif t is not considered. reason-able based on the actual performance of the instrument strings.
These radiation indicators are required for indication of radia-tion levels in containment following an accident. The specified 20% accuracy is consistent with the requirements of Regulatory Guide 1.97, " Instrumentation for Light-Water Cooled Nuclear Power Plants to Assess Plant and Environs Conditions During and fol-loving an Accident" (RG 1.97),
for radiation indication.
Currently, 2.2% of the 20% total accuracy is attributed to the electronics in the instrument string when estimating the accuracy of the monitors. The 30 month projected drif t (1.973%) is less than 2.2% therefore it is still acceptable under RG 1.97.
Reactor Coolant System Subcooling Margin Monitor The Reactor Coolant System (RCS) Subcooling Margin Monitors are utilized by plant operators to determine the need to trip
LAR 95-0024 Attachm:nt 3 Page 15 the reactor coolant pumps (RCPs). The Subcooling Margin Monitors have multiple sources for temperature inputs. The inputs are T RCS Loop 1, T RC Loop 2 and Incore Thermocouples (16 totabT HOT Either T manually selected by the operator, is the normal input tone,SubcoolingMarginMonitors.Theincorethermocouples were not evaluated in this drift study since they would only be used after adequate subcooling was lost and the RCPs were tripped. The incore thermocouples were not evaluated in this drift stufy because they vill still be calibrated at their current calibration interval (every 18 months). The only part of the instrument strings which vill not be calibrated every 18 months is the TSAT meter (TDY4950 and TDY4951) which does not have any time dependent drift associated with it, only meter accuracy
(+2
- F).
Redundant indicators for the incore thermocouples (TI4627 and TI4628) are available in the Post Accident Monitoring panels directly below the Subcooling Margin Monitors. These indicators vill be calibrated every 18 months and may be used to determine if the value displayed on the SCM is not accurate. Also, the incore thermocouples arenot used to determin subcooling margin. These instruments would only be used af ter adequate subcooling was lost and the RCPs were tripped.
Reactor Coolani Loop Outlet Temperature:
(TERC3AS,B5,A6,B6 as' read on TDY4950 and TDY4951)
The 30 month projected drift of 2.736 'F is less than the design basis / reference uncertainty of 14.472 'F therefore, all current assumptions based on the design basis / reference uncertainty are still applicable and conservative because the 30 month projected drift is bounded by the design basis / reference uncertainty. No calibrations, functional tests, or channel checks require alter-ation because the 30 month projected drift is less than the design basis / reference uncertainty.
No outliers were removed from the data. Historical data has been acceptable except on rare occasions. No safety related setpoints are associated with these instrument strings.
The temperature elements in these instrument strings are passive devices (RTDs) and are not adjusted but are verified against a standard which is impractical to do. These RTDs do not exhibit time dependent uncertainties. Since the RTDs are not a part of the calibration, there is not any as-found or as-left data associated with them. Also, - the RTDs are periodically replaced under the DBNPS Environmental Qualification (EO) program. Based on the above, it is acceptable not to include the RTDs in this the determination of the design basis / reference uncertainty. Any error associated with the RTDs has not been included in this comparison.
LAR 95-0024 Attcch=nt 3 Page 16 The assumption of normality for the vorst case point was tested using the V test. The data failed the V test. The distribution of the worst case point with respect to the standard deviation using a histogram was alse examined. This distribution of the points is affected by the digital accuracy of the readout. The indicator only reads to the nearest degree which means any error vill cause the indicator to be off by one degree. This causes the standard deviations to occur in discrete intervals of 10.9835 percent of span centered around -0.0757 when there is any error. All points fell exactly on -2.0426,
-1.0591,
-0.0757, 0.9078 or 1.8913 standard deviations from the mean. Only one point out of 26 (96.15%) was greater than two standard deviations from the mean.
Only 95% of all points are required to be within 2 standard deviations for a
normal distribution.
This supports the
^
assumption of a normal distribution, thus the assumption of normality is reasonable.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment. At no point in time did the data exceed the design basis / reference uncertainty.
These instruments are rarely adjusted.
These temperature strings have not required adjustment for up to 75 months. Regression analysis was performed on all points and on the vorst case point.
In both cases, regression analysis indicated that the data is not time dependent. The standard deviation does not increase as a function of time since last adjustment.
In fact, analysis demonstrated that the sample standard deviation decreased as the time since last adjustment increased. Also, the mean was always close to zero. The data indicates that the temperature strings vill remain relatively constant as a function of time. Based on the above evidence, these instrument strings are not considered time dependent.
Reactor Coolant Hot Leg Vide Range Pressure:
(PTRC2A4 and PTRC2B4 as read on TDY4950 and TDY4951 (TDY))
The 30 month projected drift 113.825 psig is less than the design basis / reference uncertainty of 121.65 psig even though the accuracy of the indicator was ignored when calculating the design i
basis / reference uncertainty. The analysis only considered part of
)
this instrument string. The remainder of the string is calibrated under the Safety Features Actuation System (SFAS) calibration.
(PTRC2A4 and PTRC2B4 Through I/I Converter (PTRC))
The 30 month projected drift of 1.762% is greater than the design basis / reference uncertainty of 10.9316% for the SFAS portion of the pressure string. This is probably affected by the small sample size which has a large correction factor (3.708).
Also, M&TE effects vere not included and it was assumed that all instrument drift would occur in the first 22.5 months with no drift occurring after that. This corresponds with the value used in calculation C-ICE-064.02-007.
LAR 95-0024 Attechunt 3 Page 17 Only the SPAS portion of the instrument string was analyzed here.
The remainder of the instrument string is calibrated under the Saturation Margin Monitor Calibration.
Combined Tolerance of Pressure Instrument String The 30 month projected drift value of 1 6.17 psig is greater than 4
the design basis / reference uncertainty of 131.80 psig (determined by the square-root-sum-of-the squares method) for the entire string. This is probably affected by the small sample size for the SFAS portion of the instrument string which has a large correction factor (3.708). Also, M&TE effects and drif t of no instruments other than the pressure transmitter vere not included in the design basis / reference uncertainty and it was assumed that all drift for the pressure transmitter vould occur in the first 22.5 months with no drift occurring after that.
No outliers vere removed from the data. Historical data has been acceptable except on rare occasions. No safety related setpoints are associated with these instrument strings.
The assumption of normality for the vorst case point for both sections of the pressure string was tested using the V' test. The assumption of normality was not rejected by the V test for the vorst case point of either section of the pressure string. The distribution of the points with respect to the standard deviation using a histogram was also examined. For the vorst case point of both portions of the pressure instrument string no points were greater than two standard deviations from the mean. For all points, 111 of 117 (94.87%) were within 2 standard deviations for the TDY portion f the string which is very close to the required 95% and is considered acceptable. No points were greater than 2 standard deviations from the mean for the PTRC portion of the string when all points were plotted on a histogram. This supports the assumption of a normal distribution. The assumption of normality is considered valid.
The assumption of normality for the PTRC portion of the string with the data extrapolated to 30 months was tested using the J test. The assumption of normality was not rejected by the V test for the vorst case point. The distribution of the points with respect to the standard deviation using a histogram was also examined. For the vorst case point and for all points no points were greater than two standard deviations from the mean. This supports the assumption of a normal distribution. The assumption of normality is considered valid.
For the TDY portion of the pressure string the Drif t vs. Time Since Last Test and Drif t vs. Time Since Last Adjustment plots for both Point 8
and All Points indicate more positive (non-conservative for monitoring for loss of subcooling) drift values for the increasing time intervals. This trend is supported by the regression analysis performed on the four plots. When mean and standard deviation values were calculated for (all points) i
I LAR 95-0024 Enc)nsure 4 Page 18 Drift vs. Time Since Last Adjustment bins less than 30 months, 30 to 50 months, and greater than 50 months, the deviation remained approximately unchanged or decreased for the longer interval bins. Since it appears that the standard deviation is not time dependent and the mean is time dependent, linear or square root 1
extrapolation of all points to 30 months was deemed inappropriate and was not performed. Rather, the vorst case k*s was added to the value of the Drif t vs. Time Since Last Adjustment for All Points regression line at 30 months with the zero intercept conservatively removed. The resulting 30 month projected drif t is 0.553% of span.
The Subcooling Margin Monitor is used to determine when to turn off the RCPs due to a loss of subcooling margin. The worst case i
point was not used since it is at the highest point of the in-strument range used in the calibration (2375 psig). The decision to turn off the RCPs on a loss of subcooling margin utilizes the i
remainder of the instruments' range (0 to 1875 psig). Due to the fact that the worst point is near the high point of the instru-ments' range, the drift of this point would not reflect the performance of the instruments at the range used to determine if subcooling margin was lost. Based on this, the worst case point i
was not used to determine the 30 Month Projected Drift for either portion of the pressure string.
1 The data for PTRC portion of the pressure string was plotted to show the relationship between the Drift (percent of span) versus the time since last adjustment and time since last test. These instruments were frequently adjusted. Since only one point is longer than 30 months, all of the data for PTRC portion of the pressure string was extrapolated to 30 months. The data was extrapolated to 30 months using the square root method of extrap-olation. The data was extrapolated to 30 months using the Square Root Sum Of Squares method because the time dependent drift associated with the pressure transmitters was clearly not linear based on the analysis of the data (Regression Analysis and plots of Drift Versus Time Since Last Adjustment). Since the drift is not linear, the Square Root method of data extrapolation is considered a conservative method to extrapolate the data.
The Subcooling Margin Monitor is used to determine when to turn j
off the RCPs due to a loss of subcooling margin. Since pressure signal drift in the decreasing direction vould provide a conser-vative subcooling margin indication, it is acceptable to use a 4
one-sided tolerance factor to determine 30 month projected drift.
The current conditions and assumptions of the safety analyses have been checked and are appropriately reflected in the accept-ance criteria of plant surveillance procedures associated with the Subcooling Margin Monitor for which surveillance intervals are being increased. The existing calculations related to the Subcooling Margin Monitor vill be reviewed and updated with the drift study data as required and the minimum subcooling margin l
LAR 95 0024 Attechm:nt 3 Page 19 allowed by DBNPS procedures vill be revised as necessary.
PORV and Pressurizer Safety Valves Position Indicators (ZE4263, ZE4264, ZE4265, ZE4266, ZE4267 and ZE4268 as read on ZI4263A, ZI4264A, ZI4265A, ZI4266A, ZI4267A and ZI4268A)
The vibration sensors (ZE4263 throguh ZE4268), charge converters (ZT4263 through ZT4268),
alarm initiation devices (ZIS4263A through ZIS4268A) and flow monitoring modules (ZY4263 through ZY4268) do not have any adjustments and are not calibrated.
These devices are checked during each calibration to verify that they still function properly. The vibration sensors and the charge converters are periodically replaced under the station Environmental Qualification (EQ) program. Since 1987, only three vibration sensors and two flow monitoring modules have been replaced in addition to the replacements required by the EQ
)
program. One of 'the vibration sensor failures was attributed to I
the sensor being dropped during maintenance on the valve which required the removal of the sensor. The alarm initiation devices have never failed to alarm properly.
The 30 month projected drift of i2.697% is greater than the design basis / reference uncertainty of 21.118%. The 30 month projceted drif t from the 24 month drif t study data is a worst case value with a 95/95% confidence factor. This tolerance in-terval was based on a sample set containing only 12 points. This gives a drif t correction factor of 3.162. Also, any error in-troduced by M&TE was not included as this error was considered negligible. No time dependent drift was included in the analysis since no vendor data was available on time dependent drift.
No outliers were removed from the data. Historical data has been acceptable except on rare occasions. No safety related setpoints are associated with the PORV or Pressurizer Safety Valve position indicators.
The assumption of normality was rejected by the V test. For the vorst case data point set, no points were greater than 2 standard deviations from the mean. The accuracy of the indicators appears to have affected the distribution of the points. Any error vill cause the indicator to be off by 1%. This caused the standard deviations to occur in discrete units of 0.0 of 21.1726 centered around the mean of 0.00 for the worst case data point set. The distribution appears flat with no points greater than 2 standard deviations from the mean. When all data points were plotted on a histogram, 187 of 192 points (97.4%) were within 2 standard deviations of the mean. Only 95% are required to be within 2 standard deviations. This supports the assumption of a normal distribution.
.__-._.__.__._._..-__._._-_~_.m J
Attachssnt 3 I
1 Page 20 The data was plotted to show the relationship bet' ween.the Drif t j
l
.(percent of span);versus the~ time since last adjustment and drift versus time since last test. These instrument strings have not i
required an adjustment for over 30 months. There was a decrease in the standard deviation as the time since last adjustment i
increased. The standard deviation does not increase as a function-l of time since last adjustment..The data indicates that the in-strument strings vill remain relatively constant as a function of j
time. Regression analysis vas.also performed on all points and on the worst case point for drift versus time since last test. The regression analysis indicates that the drift is not time l
dependent. This supports the assumption that the drif t is not-t l_
time dependent.
l The safety function of the valve flow monitoring system is de-l tection of flow not measurement of flow. The indicators do not show the actual position of the valves or the actual flow through l
the valve only.that the valve is open or closed. Gross indication of valve position (open or closed) is all that is required by j
NUREG-0578, "THI 2 Lessons Inarned Task Force Status Report and 1
Short-Term Recommendations,*
UREG-0737, " Clarification of THI Action Plan Requirements,"
btem II.D.3) and Table 2
of Regulatory Guide l'. 97, " Instrumentation For Light-Water Cooled 5
Nuclear Power Plants To Assess Plant and Environs Conditions l
During and Following An Accident." Since the indicators give only gross indication of valve position, that the valve is open or i
closed, the 30 month projected drif t is acceptable even though it
[
is greater than the design basis / reference uncertainty. No operator actions are taken based on incremental indication of i
valve position. Based on this information, no further evaluation of the PORV or. Pressurizer Safety Valve Position indicators is required.
Containment Normal Sump Level (LE4617 and LE4618 as read on LI4617 and LI4618) i These instrument strings are also used to satisfy Technical.
Specification 3.4.6, Reactor Coolant System Leakage.
The 30 month projected drif t of $0.223 feet is greater than the design basis / reference uncertainty of t0.2194. The 30 month projected drif t from the 24 month drif t study data is a worst -
case value with a 95/95% confidence factor. This tolerance in-terval was based on a sample set containing only 8 points. This gives a drif t correction factor of 3.732. Also, any error in-troduced by M&TE or time dependent drift was not included as this information was not available.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Containment Vessel Normal Sump ' Level Indicators.
4 LAR 95-0024 AttschmInt 3 Page 21 The assumption of normality was not rejected by the V test for the vorst case point. The data for all points and the worst case point appeared normal when plotted on a histogram. For the worst case data point set, no data points were greater than 2 standard deviations from the mean. For all data points, 52 out of points (92.86%) were within 2 standard deviations of the searn Ninety-five percent of the points are required to be within 2 standard deviations for a normal distribution, however, 51 of 56 points were at the center of the distribution which makes the standard deviation small. A large peak of data at the center of the distribution is conservative and is bounded by the normal distribution. This supports the assumption of a normal distri-bution.
The data was plotted to show the relationship between t'. e Drif t (percent of span) versus the time since last adjustment and drift versus time since last test. These instrument strings have gone over 70 months without requiring an adjustment. In fact, these instruments have not required an adjustment since they were installed. The long term plots of drif t versus time since last adjustment do not show any time dependent characteristics. This justifies the assumption that the instruments are not time dependent.
The 30 month projected drift is greater than the design / reference uncertainty. However, no safety related trip setpoints are associated with the containment normal sump level indication nor i
are these instruments used to effect a safe shutdown. A review of
'i the data indicates that these instruments perform better than the worst case point implies. The large tolerance associated with these instruments is caused in part by the digital accuracy of the instruments which read only to the nearest tenth of a foot (20.1 feet). This is large when compared to the 4 foot span of t2.5% of span). A total of 70 the instruments (0.1 feet
=
readings were taken for the instrument strings. Only 3 of the readings were not at the desired values. Five of the seven points
'Aere data is taken had no points that were not at the desired values (the mean and standard deviation of the percent drift since last test data vere zero). Based on this, these instrument strings are considered acceptable for Containment Normal Sump Level indication.
Containment Vide Range Vater Level (LT4594 and LT4595 as read on LI4594 and LI4595)
The Containment Vide Range Level instruments receive pressure compensation from the Containment Pressure instruments. The effects of this compensation on containment water level indi-cation are discussed in the section on Containment Pressure instrumentation.
The 30 month projected drift of $14.354% of span is greater than the design basis / reference uncertainty of 12.854% of span. The 30
LAR 95-0024 Attacharnt 3 Page 22 month projected drif t from the 24 month drif t study data is a worst case value with a 95/95% confidence factor. This tolerance interval was based on a sample set containing only 8 points. This gives a drift correction factor of 3.732.
No outliers were removed from the data. No safety related setpoints are associated with the containment Vessel Water Level (VR) Indicators.
The assumption of normality was not rejected by the W test for the worst case point. The data for all points and the worst case point appeared normal when plotted on a histogram. For the worst case data point set, no data points were greater than 2 standard deviations from the mean. For all data points, 67 out of 68 points (98.5%) were within 2 standard deviations from the mean.
Only 95% of the points are required to be within 2 standard deviations therefore this supports the assumption of a normal distribution.
The assumption of normality was also tested for the data extrap-olated to 30 months using the W test. The assumption of normality was not rejected by the V test for the worst case point. The data for all points and the vorst case point appeared normal when plotted on a histogram. For the vorst case data point set, no data points were greater than 2 standard deviations from the i
mean. For all data points, 66 out of 68 points (97.1%) were within 2 standard deviations from the mean. Only 95% of the points are required to be within 2 standard deviations therefore this supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment and drift versus time since last test. The long term plots of drift versus time since last adjustment do not show any time dependent char-acteristics, however, only one test is longer than 30 months.
Since these instruments were adjusted very frequently, the data for all tests was extrapolated to 30 months.
Point 9 was the vorst case point for all data. The standard deviation of point 9 was approximately twice that of the eight points. This point is at the bottom of the instrument range (2%
of span). Point I which is also at 2% of span point 9. It appears that point 9 is showing the effects of hysteresis. Since this point is at the bottom of the scale (2% of span) and is taken when the indicator is decreasing, it can be removed from this evaluation. The indicators are used for increasing water level.
Further, the range of the indicators is used to determine manual actions based on level indication of 570 feet which is approx-imately 60% of span as required by DBNPS procedure DB-0P-02000, "RPS, SFAS, SFRCS Trip, or SG Tube Rupture," (DB-0P-02000). For these reasons, point 9 was not used as the vorst case point for these instrument strings.
LAR 95-0024 Attechm:nt 3 Page 23 The 30 month projected drif t of the indicators is too large to use for precise level indication. While the drift is large, it is span independent meaning that all points drift equally high or equally lov. There does not appear to be any drift component that affects the relationship between the individual points over the range of the instruments, except for the hysteresis effects on point 9 vhich vere discussed previously. The basic statistics calculated for these instrument strings support the assumption that the drif t is span independent as all of the data points shif ted by approximately the same amount and in the same dir-ection. The direction in which all of the points shifted occurred in a randcm pattern (the drift was not always increasing or decreasing). Also, the magnitude of the drift appears to be random ranging from less than 0.1% drif t to greater than 3.0%
drift in an 18 month period.
All of the points in a given test for both the 18 month data and the data extrapolated to 30 months were analyzed to determine if their percent drift values could be combined. The results of the analysis showed that the standard deviation of the points in one test was much less than the standard deviation for a group of points from several tests. The average drift for all points in a test versus last test was plotted. These plots showed that the magnitude and direction (positive or negative) of the drif t is random and independent of the previous test or a previous value.
All points in a given test had approximately the same magnitude and direction of drif t regardless of their values (percent of span).
Data from DBNPS procedure DB-SC-03180, " Remote Shutdown, Post Accident Monitoring Instrumentation Monthly Channel Check," for Level Indicators LI4594 and LI4595 from the last five years were analyzed to determine if the actual drift experienced during an 18 month cycle was unacceptable. If the data corresponds to the 24 month Instrument Drift Study indications, then the values in the Monthly Channel Check should also vary greatly since it is unreasonable to assume that the instruments would drif t in the same manner. The difference in the instrument readings never exceed 1.2 feet (the tslerance is 6 feet). Only two tests had differences greater than 1 foot. The total range of values for both indicatore over the last five years was from 535.9 to 540.3 which is a difference of 4.4 feet.
i i
The effect from the change in ambient temperature was determined to be 21.641 feet. Based on this information, the temperature
{
effects could account for a total change of 3.282 feet which is i
the majority of the total range of values seem over the last five f
years (4.4 feet). This indicates that the instruments are not as j
erratic as the calibration data seems to indicate. These indica-tors tend to read in their desired range without drifting greatly l
in any direction. This supports the supposition that the instru-ments are in fact more accurate than the calibration data in-dicates.
1 l
LAR 95-0024 Attachm:nt 3 Page 24 Since the magnitude and direction of drif t is random and inde-pendent of span, the level indication relative to the span of the instrument will remain unchanged even though the overall drift of the instruments may be large. Based on this, it is acceptable to use the Containment Vater Level Indicators to show a change in water level but not the actual level.
Since the 30 month projected drif t of LI4594 and LI4595 is so large, DB-0P-02000 vill be revised to use the CTMT Vessel Vater i
l Level (VR) instruments for trending level inside CTHT but not for exact indication of CTMT vater level.
l Although the 30 month projected drif t of these instruments is large, no control functions are associated with these level indicators. Also, any decisions made based on these instruments vill not require precise level indication af ter DBNPS procedure DB-0P-02000 is revised. No calculations or safety related set-points are associated with the Containment Vessel Vide Range Level instruments 'or is this instrumentation related to any f
safety analysis assumptions.
Based on the above discussion these instruments are considered acceptable for providing qualitative data for operator actions.
Containment Vide Range Pressure (PT4587 and PT4588 as read on PI4587 and PI4588)
The 30 month proj ected drift is equal to the design basis / reference uncertainty of 11.07 psia therefore, all current assumptions based on the design basis / reference uncertainty are still applicable and conservative because the 30 month projected drift is bounded by the design basis / reference uncertainty. No calculations, calibrations, functional tests, or channel checks require alteration because the 30 month projected drif t is equal to the design basis / reference uncertainty.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Containment Pressure Indicators.
j The assumption of normality was not rejected by the V test for the vorst case point. The data for all points and the vorst case point appeared normal when plotted on a histogram. For the worst case data point set, only one data point was greater than 2 standard deviations from the mean. For all data points, 7 out of 99 points were greater than 2 standard deviations from the mean.
There was a large peak of data points at the center of this distribution which caused the standard deviation to be small. A large peak of points at the center of the distribution is con-r l
servative, thus it is bounded by the normal distribution. This J
l supports the assumption of a normal distribution.
LAR 95-0024 Attcchm:nt 3 Page 25 The data was plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment and drift i
versus time since last test. There was a decrease in the standard deviation as the time since last adjustment increased. Regression analysis was also performed on all points and on the vorst case point for drift versus time since last test. The regression analysis performed was inconclusive for determining time dependency.
Regression analysis provided no conclusive evidence that the i
drift is time dependent. However, these instrument strings have gone over 70 months without requiring an adjustment. The long term plots of drif t versus time since last adjustment did not show any tim dependent characteristics.
These plots also indicated that all data beyond 20 months is bounded by the vorst 4
data less than 20 months. In addition, the drift study tolet nce interval of 51.07 psia is roughly 3 times the vorst case point for data that goes out to the approximately 70 month interval.
Based on this, 51.07 psia is an acceptable 30 month projected drift. This justifies the assumption that the instruments are not time dependent.
l Pressure Indicators PI4587 and PI4588 are two of six containment i
pressure indicators that can be used to determine when to shut down the Hydrogen Dilution Blovers during post accident con-l ditions in accordance with DBNPS procedure DB-0P-02000. The Hydrogen Dilution Blowers are shut of f when the indicated con-y tainment pressure reaches 32 psia. The containment design pres-sure is 36 psig. One-half of containment design pressure equals
)
18 psig or 32.7 psia. The 30 month projected drift of PI4587 and i
PI4588 is 11.07 psi., which equals the design basis / reference uncertainty, and does not adversely effect the margin to contain-i
, ment design pressure assumed for securing the Hydrogen Dilution j
Blovers.
- Further, containment pressure vill increase slowly i
during the accident scenario providing sufficient time for oper-ators to take manual action to shut down the Hydrogen Blovers.
For these reasons, the 30 month projected drif t of pi 4587 and pi 4588 is considered acceptable. Since the 30 month projected drift is equal to the design basis / reference uncertainty, it is still acceptable to use PI4587 and PI4588 for this action as all current assumptions are still applicable.
Pressure Compensation for Containment Vide Range Water Level:
These instrument strings also provide pressure compensation for the Containment Vide Range Level instruments.
The 30 month projected drif t of 10.736% of span is greater than the design basis / reference uncertainty of 20.684% of span. The 30 month projected drif t from the 24 month drif t study data is a vorst case value with a 95/95% confidence factor. This tolerance interval was based on a sample set containing only 7 points. This gives a drift correction factor of 4.007.
i
LAR 95-0024 Attach =nt 3 Page 26 No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Containment Pressure instrument strings.
The assumption of normality was not rejected by the V test for the worst case point. The data for all points and the vorst case point appeared normal when plotted on a histogram. For the vorst case data point set, no points were greater than 2 standard deviations from the mean. For all points, 60 out of 63 vere within 2 standard deviations (95.24%). Only 95% of the points are required to be within 2 standard deviations therefore this supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drift (percent of span) versus the time since last adjustment and drift versus time since last test. There was a decrease in the standard deviation as the time since last adjustment increased. These instrument strings have not required an adjustment for over 70 months. The long term plots of drift versus time since last adjustment did not show any time dependent characteristics. These plots also show that all data beyond 20 months is bounded by the vorst data less than 20 months. In addition, the drift study tolerance interval of i0.736% is roughly 3 times the vorst case point for data that goes out to the ~70 month interval. Based on this, 20.736% is an acceptable 30 month projected drif t.
This justifies the assumption that the instruments are not time dependent.
The calibration of the CTMT Vessel Water Level (VR) instruments use a constant 50 psig input to instead of a variable input.
Because of this, the effects of the pressure inputs are not included in the calibration data. The Pressure instrument string inputs to the Containment Vide Range Level were analyzed separ-ately and it was determined that their 30 month projected drif t is 20.736% (10.0658Vdc based on a span of 0-8.94Vde). This has a maximum effect of 10.658% on the output. This shift vill be irrelevant because it is much less than the 114.354% 30 month projected drif t previously determined for the Containment Vide Range Level instrument strings. Since the applicable DBNPS pro-cedure vill be revised to use these instruments only for indic-ation of a general trend (reference previous discussion for Containment Vide Range Level) the effects of the pressure compen-sating input will not affect the use of the CTMT Vessel Water Level (VR) instruments.
Reactor Coolant Hot Leg Level Monitoring The Hot Leg Level Monitoring System is used to determine the reactor coolant inventory when all four reactor coolant pumps are off allowing operators to detect conditions that could lead to inadequate core cooling. This system is an aid utilized to pre-vent the core from being uncovered. The level is determined by measuring the differential pressure between the top and bottom of the hot legs using Level Transmitters LT5448A and LT5448B vhich
LAR 95-0024 Attachm:nt 3 Page 27 have density compensation in the reference legs provided by Temperature Element TE5449A and B, and TE5450A and E. All TE's are identical however, one sensor is in service and the other is an installed spare for each of the two sets of sensors. Tempera-ture Element TERC3A5, TERC3B5, Pressure Transmitter PTRC2B4, and PTRC2A4 provide density compensation for the actual water in the hot legs. The plant computer and TSC computer perform the necessary calculations to produce the hot leg level indication.
Reactor Coolant Hot Leg Level Temperature Compensations (TY5449, TY5450 as read on computer points T767, T768)
The temperature elements in these instrument strings are passive devices (thermistors) and cannot be adjusted. The temperature to current (T-I) board is calibrated to match the thermistor to provide maximum accuracy. The sensors cannot be removed for calibration due to their location and the difficulty involved with removing them and the potential for damaging the ther-mistors. These sensors do not require periodic maintenance (reference Davis-Besse Nuclear Power Station Electrical Equipment Environmental Qualification Package DB1-083, Alison Controls Temperature Sensor Assembly). Calibration data is not taken periodically for the thermistors. Any gross drif t of the ther-mistors would be identified through the monthly channel check.
Based on the above, it is acceptable not to include the sensors in this the determination of the design basis / reference uncertainty. Any error associated with the sensors has not been included in this comparison. The inaccuracies of the indicator (computer point) were assumed to be zero for calculating the design basis / reference uncertainty.
The 30 month projected drif t of il.308% of span is less than design basis / reference uncertainty of i3.0516% of span even though the accuracy of the computer point and time dependent drift were ignored for calculating the design basis / reference uncertainty. These instruments rarely required any adjustment.
The data was usually well within tolerance. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Hot Leg Level Monitoring System.
One outlier was eliminated frcm the data. This outlier on test date 10-12-88 was caused by the computer being scaled wrong. The output from the V/I converter to the computer was good, however, the computer point was scaled from 50-300 'F instead of 100-300
- F vhich caused the errors in the as-found data.
The assumption of normality was tested by analyzing the vorst case data point using the V test. The data failed the V tost.
This failure was caused by a large peak of data points at the center of the distribution. The data for the vorst case point was plotted on a histogram. Two points were greater than 2 standard deviations from the mean, however, 10 of 13 points (76.92%) are
LAR 95-0024 AttachmInt 3 Page 28 at the center of the distribution. This causes the standard deviation to be very small. The data for all points looks similar when plotted on a histogram. Five of 56 points (91.07%) are greater than 2 standard deviations from the mean, however, 44 of 56 points (78.57%) are at the center of the distribution which causes the standard deviation to be small. A large peak of data at the center of the distribution is conservative thus this distribution is effectively bounded by the normal distribution.
Sinse the performance of the ins.trument is effectively bounded by the normal distribution, the assumption of normality is valid.
The assumption of normality for the data extrapolated to 30 months was tested by analyzing the worst case data point using the W test. The data failed the V test. This failure was caused by a large peak of data points at the center of the distribution.
The data for the worst case point was plotted on a histogram. One of 10 points was greater than 2 standard deviations from the mean, however, 7 of 10 points (70%) are at the ' center of the distribution. This causes the standard deviation to be very small. The data for all points looks similar when plotted on a histogram. 3 of 41(92.7%) points are greate than 2 standard deviations from the mean, however, 29 of 41 points (70.7%) are at I
the center of the distribution which causes the standard devia-l tion to be small. A large peak of data at the center of the distribution is conservative thus this distribution is effec-tively bounded by the normal distribution. Since the performance l
of the instrument is effectively bounded by the normal distribu-tion, the assumption of normality is valid.
The data vas plotted to show the relationship between the Drif t (percent of span) versus the time since last adjustment and drift versus time since last test. At no point in time did the data exceed design basis / reference uncertainty. These instruments are rarely adjusted and do not show any time dependent character-istics and little drift. These temperature strings have not i
required adjustment for 58 months. The standard deviation does not increase as a function of time since last adjustment. In l
fact, analysis demonstrated that the sample standard deviation decreased as the time since last adjustment increased. The data l
indicates that the temperature strings will remain relatively l
constant as a function of time. Regression analysis was also l
performed on all points and on the vorst case point for drift l
versus time since last test. Regression analysis indicated that the drift is not time dependent. Since only 4 points were greater than 30 months, all points were extrapolated to 30 months via the square root method. The square root method was chosen based on regression analysis and the plots of drif t vs. time since last adjustment which indicate that the drift is not linear. Only
[
points with intervals greater than 1 month vere extrapolated to 30 months because points with intervals less than 1 month would add error which would be large and would not represent actual drift.
1
LAR 95-0024 Attachm2nt 3 Page 29 Reactor Coolant Hot Leg Temperature (TERC3A5 and TERC3B5 as read on computer points T782, T753)
The temperature elements in these instrument strings are passive devices (RTDs) and cannot be adjusted only verified against a standard. These RTDs do not exhibit time dependent uncertainties (reference Davis-Besse Nuclear Power Station Electrical Equipment Environmental Qualification Package DB1-30A, Rosemount Resistance Temperature Detector). Since the RTDs are not a part of the calibration, there is no as-found or as-left data associated with them. Also, the RTDs are periodically replaced under the station Environmental Qualification (EO) program. Based on the above, it is acceptable not to include the RTDs in this the determination of the design basis / reference uncertainty. Any error associated with the RTDs has not been included in this comparison.
The 30 month projected drift of i0.405% of span is less than the design basis / reference uncertainty of 10.7071% of span even though the accuracy of the computer point and time dependent drift were ignored for calculating the design basis / reference uncertainty. The data does not exceed acceptable limits for the calibration interval. Historical drift has been acceptable except on rare occasions. No outliers were removed from the data. No safety related setpoints are associated with the Hot Leg Level Monitoring System.
The assumption of normality was not rejected by the V test for the vorst case point. The data appeared normal when plotted on a histogram. For the vorst case data point set, no points were greater than 2 standard deviations from the mean. This supports the assumption of a normal distribution.
The data was plotted to show the relationship between the Drif t (percent of span) vs. Time Since Last Adjustment and the Drif t vs. Time Since Last Test. The data never exceeded the design basis / reference uncertainty.
These instruments are rarely adjusted and do not show any time dependent characteristics and very little drif t. These temperature strings have not required an adjustment for 134 months. The standard deviation does not increase as a function of time since last adjustment indicating that the instrument drift is not time dependent. In fact, analysis demonstrated that the sample standard deviation and the sample mean decreased as the time since last adjustment increased. Regression analysis was performed on all data points and on the worst case data point for plots of Drif t vs. Time Since Last Test. Regression analysis indicated that the drift is not time dependent.
LAR 95-0024 Attachm:nt 3 Page 30 Reactor Coolant Hot Leg Level:
(LT5448A and LT5448B As read on computer points L720, L721)
The 30 month projected drif t of 22.6176% is greater than the design basis / reference uncertainty of 11.2460% of span. Trans-mitter drift due to radiation effects was not included in the calculation of the design basis / reference uncertainty since this information was not available. For the design basis / reference uncertainty, all transmitter drif t was assumed to occur in the first six months with no drift occurring after the first six months. The 30 month projected drif t from the 24 month drif t study data is a worst case valte with a 95/95% confidence factor.
This tolerance interval was based on a sample set containing only 12 points. This gives a tolerance correction factor of 3.162.
Historical drift has been acceptable except on rare occasions. No outliers were removed from the data. No safety related setpoints are associated with the Hot Leg Level Monitoring System.
The assumption of normality was not rejected by the V test for the worst case point. The data for the vorst case point and for all points was plotted on a histogram. For the vorst case point, 9 of 12 points (75%) are at the center of the distribution. For all points, 53 of 71 points (74.65%) are at the center of the distribution. A large peak of data at the center of the distri-bution is conservative, thus it is effectively bounded by the normal distribution. This supports the assumption of a normal distribution.
There does not appear to be any significant time dependent drift j
associated with these instrument strings. These instruments are capable of remaining in service for over 70 months without an adj us tment. The daca was plotted as a function of drift (percent of span) versus time since last adjustment and drift versus time since last test. The plots of Drif t vs. Time Since Last Adjust-ment show a decreasing trend of the magnitude of the drift in the data as time increases. Regression analysis of the data was also performed on the plots of Drift Vs. Time Since Last Adjustment.
Regression analysis indicated that the drift is not time dependent.
Reactor Coolant Hot Leg Vide Range Pressure:
(PTRC2A4 PTRC2B4 as read on computer points P732 and P724, and PTRC2A3 and PTRC2B3 as read on computer points P733 and P725) l These pressure transmitters are associated with the Safety Features Actuation System (SFAS). Channels 1 and 4 of SFAS contain replacement current - to - current isolators, Consol-idated Controls model 6N548, installed in October 1991 and March 1993, respectively. Channels 2 and 3 contain Fisher and Porter model 50EK1000 (the original isolators supplied with the system).
Although a drift value is not specified withir either the
LAR 95-0024 Attachm:nt 3 4
Page 31 original or the replacement isolator vendor literature, the replacement isolator was chosen by the original supplier of the SFAS system (Consolidated Controls) as an appropriate re-placement.
Since the replacement is built of electrical subcomponents at least one decade never than _those of the original, it is expected that the drif t of tha replacement is better than that of the original. The specified accuracy value for both the original and the replacement isolators is the same (0.25% of full scale). Additionally, with the assumption that a
drift of a higher accurate device (such as 0.25% for the isolator) is less than that of a less accurate device (0.5%
accuracy for the computer point and 0.65% accuracy for the transmitter), it is expected that the drift of the isolator vill be much less than that of the remaining components in the string.
For these reasons, the drift of the replacement will be assumed to be comparable to or better than the original's and data for the four strings will be considered to be from the same statistical population for the purposes of this study.
For additional support of the isolators' drift being small, it may be noted that DBNPS calculation C-ICE-64.02.004 Revision 2, RCS Hot Leg Low & Vide Range Pressure Inst. String Uncertainties, assumed no drift for the isolator.
The 30 month projected drif t of $2.003% of span is greater than the design basis / reference uncertainty of il.0573% of span for the pressure string. This is probably affected by the small sample size of 9 data points which has a correction factor of 3.532. Also, M&TE effects were not included in the calculation of the design basis / reference uncertainty since this information was not included in DBNPS calculation C-ICE-064.02-004.
Historical drift has been acceptable except on rare occasions. No outliers were removed from the data. No safety related setpoints are associated with the Hot Leg Level Monitoring System.
The assumption of normality was not rejected by the V test for the vorst case point. The data appeared normal when plotted on a histogram. For the vorst case data point set, no points were greater than 2 standard deviations from the mean. For all points, 77 of 81 points (95.06%) were within 2 standard deviations of the mean. Only 95% of the points are required to be within 2 standard deviations from the mean for a normal distribution. This supports the assumption of a normal distribution.
The assumption of normality was tested using the V test for the data extrapolated to 30 months. The assumption of normality was not rejected by the V test for the vorst case point. The data for the worst case point and for all points appeared normal when plotted on a histogram. For the vorst case data point set, no points were greater than 2 standard deviations from the mean. For all points, 77 of 81 points (95.06%) vere within 2 standard deviations. Only 95% of the points are required to be within 2 standard deviations for a normal distribution. This supports the
LAR 95-0024 Attechmtnt 3 Page 32 assumption of a normal distribution.
The data was plotted as a function of drif t (percent of span) versus time since last adjustment and drif t versus time since last test. These pressure transmitters appeared to drift low initially after installation or adjustment and then remain constant as a function of time. Regression analysis of the data was also performed. Regression analysis indicated that the drift is not time dependent. Since only one data point was greater than 30 months, all of the data was extrapolated to 30 months using the square root method of data extrapolation. The square root method was chosen because the drift was clearly not linear based on regression analysis and the analysis of plots of drift versus time since last adjustment.
Although no specific evidence exists, there is the potential that adjustments could have been made to computer points without this information being recorded in a fashion that is retrievable. It is possible that an adjustment of a computer point may not be included in the data associated with the analysis of the HLLMS.
Considering the amount of data available, any error due to unavailable information vould be insignificant.
The Technical Support Center (TSC) computer is typically assigned a smaller tolerance than the MODCOMP computer. Therefore, it is reasonable to assume that the TSC computer points vill have less drift than the MODCOMP computer points. Only MODCOMP computer points were used in this study. This is conservative since the MODCOMP computer vill have more drift than the TSC computer.
No procedure revisions are required since tt:e existing procedures reflect current conditions and assumptions ior the Hot Leg Level Monitoring System. Existing engineering calculations vill be updated with the 30 month drif t projections.
!~
LAR 95-0024 j_
Page 1 Summary of Instrument Drift Study i
for i
i Pilot Operated Relief Valve i
I (PTRC2A2, PTRC2B2, RPS1RC2313, RPS2RC2313, PICRC2, and PSHLRC2-5) j 1
t I
Two Reactor Protection Sy3 tem Pressure Transmitters provide input j
to two separate tuffer amplifiers which provide input through a l
selector panel to the PORV control circuit located in the NNI j
cabinets. The PORV control instrument string was analyzed in i
three parts. The three separate portions are the pressure trans-mitters (PTRC2B2 and PTRC2A2), the buffer amplifiers (RPS1RC2313 1
and RPS2RC2313), and the summer through the pressure switch
)
(PICRC2 through PSHLRC2-5) located in the NNI cabinet.
l The presut? transmitters have been analyzed separately in the RPS Reacier Coolant System Pressure drif t study (Enclosure 1).
j=
Two additional identical pressure transmitters (PTRC2A1 and PTRC2B1) were used when calculating the projected 30 month drif t.
The data for the buffer amplifiers (RPS1RC2313 and RPS2RC2313) was supplemented with data from four essentially identical buffer i
i j
amplifiers from the RPS pressure strings (RPS1RC2304, RPS2RC2304, 1
RPS3RC2304, RPS4RC2304). ihe only difference between the two f
i buffer amplifier versions is that they have dif ferent response time ratings, therefore, their drift characteristics 'should be the same.
i The buffer amplifiers were analyzed for drift in the categories of balance, scaling and bias. Data is taken every 18 months on l
all six of the buffer amplifiers. The 95/95% intervals for bal-i ance, scaling and bias were calculated to be t0.060% of span, j
t0.087% of span, and 20.046% of span, respectively. Several points must be made regarding the manner in which the raw data i
vas manipulated. Voltage measurements for the balance and bias categories were made at the buffer amplifier module SCALED OUTPUT test jack until the procedures were changed (as part of the i
resolution to Potential Condition Adverse to Quality Report i
91-0441) to have the measurements made at the X1 OUTPUT test jack. Between these test jacks is a unity gain inverting opamp, so all as-found/as-lef t data for these - two categories that was taken at the SCALED OUTPUT test jack was multiplied by (-1) to i
make it comparable to data taken at the X1 OUTPUT test jack.
Also, the desired voltage for the scaling category varied from test to test but was always between 10.000 and 10.010 VDC, therefore, all as-found/as-left data for this category was decreased by the quantity, desired voltage minus 10.000 VDC. This normalized the data to a 10.000 VDC desired standard, making it comparable from one test to another.
The assumption of normality was tested for balance, scaling and bias. The histogram showed that the distribution was approxi-mately normal, ~and 100% of the points were within two standard deviations of the mean. The W test for balance did not reject the
.a
LAR 95-0024 Attcchmtnt 3 Page 2 assumption of normality. Therefore, the assumption of normality is confirmed for balance.
)
The histogram for the V test for scaling showed that only 93.33%
of the points are within two standard deviations of the mean and the V test rejects the assumption of normality. This distribution is similar to several others analyzed for normality in this drift study, where there is one data point that causes the distribution J
to appear non-normal. With such a small sample set, one point can cause less than 95% of the points to be within two standard deviations and the W test to fail. If this one point were re-J moved, both the histogram and the V test would indicate that the data is normally distributed. Even vite the point included, the shape of the histogram is evidence that the distribution is approximately normal. With these facts in mind, the distribution was treated as normal for the purposes of the drift study.
The histogram for the V test for bias showed that only 92.31% of the points are bounded by two standard deviations from the mean and the W test rejected the assumption of normality by a small margin. Further investigation showed that there is only one point outside two standard deviations and this caused the distribution not to be bounded and the V test to reject the assumption of normality. The histogram provided evidence that the distribution is approximately normal. Also, in this case, the V test is very close to passing (0.8658 calculated versus 0.8660 desired).
Therefore, the distribution vill be treated as normal for the purposes of the drift study.
Drift versus time since last test and drif t versus time since last adjustment graphs were generated to assess time dependency.
Because there were so few points and those points of longer i
intervals are not, in all cases, bounded by those of lesser intervals, the data was extrapolated linearly to 30 months. This resulted in nev 95/95% tolerance intervals of 10.111% of span, t0.183% of span, and 0.096% of span for balance, scaling, and bias, respectively. There were four points with test to test intervals of 0.559 months. These points vere not extrapolated to 30 months because the associated drif ts vould have been multi-plied by approximately 50 and the extrapolated points vould not be representative of actual expected 30 month drift. For example, the scaling point generated for the Channel i buffer amplifier to i
RPS during the 0.559 month interval between 9/26/91 and 10/13/91 has a drif t of 0.03% of span, which, when extrapolated to 30 months, becomes 1.61% of span. The actual 30 month drif t points taken from the drift versus time since last adjustment graphs j
showed that the vorst point is actually 0.08% of span.
The assumption of normality was tested for the extrapolated data sets. Since the V test for balance passed and the histogram demonstrated that 100% of the points are within two standard deviations of the mean, the assumption of normality is confirmed for balance.
i LAR 95-0024 Attcchm:nt 3 Page 3 The assumption of normality was tested for scaling. As with the unextrapolated data set, the distribution is not bounded by two standard deviations from the mean and the W test rejects the i
assumption of normality. There is one point outside two standard deviations that causes the distribution to appear slightly non-normal. This point was also present in the unextrapolated set and it caused the normality tests on that set to fail as well.
However, when plotted on a histogram, the distribution appears i
approximately normal and vill be treated as normal for the purposes of the drift study.
The histogram of the normality tests for bias showed that only 91.67% of the points are within two standard deviations from the mean. In this case, the W test did not reject the assumption of normality. Since the histogram appeared to be approximately normal, the assumption of normality is considered verified for the purposes of the drift study.
The PORV control circuitry located in the NNI cabinet is tested by injecting a voltage into the summer module (PICRC2). First, the input voltage is increased until the high level trip light comes on.
This voltage is recorded and used for trending purposes. Second, th-input voltage is decreased until the lov level trip light comes on. The voltage where the light comes on is recorded and used for trending purposes. The drif ts for the high and lov level trips were calculated to be 10.827% of span and 10.715% of span, respectively.
The High and Lov trips were tested for normality. Since the W for i
the high level trip data test did not reject the assumption of normality and the histogram showed that 100% of the points are bounded by two standard deviations from the mean, the assumption of normality is verified for the high level trip data. The V test i
for the lov level trip data did not reject the assumption of normality and the histogram showed that 100% of the points 'are i
bounded by two standard deviations from the mean, the assumption of normality is, therefore, verified for the lov level trip data.
Plots of drift versus time since last test and drift versus time since last adjustment vere generated for both the high level and lov level trip data in order to assess time dependency. Since the points greater than 30 months were not bounded by those less than 30 months the data set was extrapolated to 30 months. For addi-tional insight on the degree of time dependency, the data for drift since last adjustment was separated into two bins and the mean and standard deviation was calculated for each bin. The mean for the longer intervals (greater than 22.5 months) was larger than the mean for the shorter intervals (between.O and 22.5 months). However, both means are nearly negligible, as they are close to 10.10% of span. The standard deviations of the points in the longer intervals was less than the standard deviations of the points in the short intervals, which gives evidence that the random part of the drift is not increasing with time. These two facts taken together provide justification for using the square
LAR 95-0024 Attachm:nt 3 Page 4 root method to extrapolate the points to 30 months.
The 95/95% intervals calculated using the extrapolated set for high and low level trips are 11.098% of span and 10.959% of span, respectively. The extrapolated data sets for the high level trip were tested for normality via a histogram and a V test. The histogram shewed that 100% of the points are within two standard deviations from the mean and the V test did not reject the assumption of normality. The histogram for the low level trip data showed that 100% of the points are within two standard deviations from the mean and the V test does not rej ect the assumption of normality. Therefore, the extrapolated data sets are confirmed to be normal for the purposes of this drift study.
The basis for the pressurizer pilot operated relief valve (PORV) trip setpoint and allowable value of > 2435 psig listed in Technical Specification 3.4.3 is that the minimum PORV setpoint must accommodate the largest RCS pressure overshoot (50 psi) and the RPS high pressure instrument string error (30 psi), refer-enced to the RPS high pressure trip setpoint (2355 psig). This is so the PORV vill not open on any anticipated transients. Further, the PORV should be set so as to open prior to the code safety valves.
To verify that the original 30 psi RPS pressure string error remained bounding for a 24 month fuel cycle, it was necessary to substitute the results of the drif t study performed for the RPS s
pressure transmitters (pts) into the RC pressure string error equation. The new RPS high pressure instrument string error was determined to be 25.12 psi. Since this new error is less than the 30 psi that the PORV Technical Specification value (2435 psig) is based upon, that value remains acceptable as the minimum PORV setpoint.
The present PORV field setpoint is 2450 psig and was established to satisfy the criteria of meeting the Technical Specification minimum value e.nd allowing PORV actuation prior to lifting the pressurizer code safety valves. Since the code safety valve field i
setpoint is 2500 psig, 50 psi of margin is needed.
The total uncertainty in PORV versus code safety valve actuation was determined by combining the PORV string error with the code safety valve setting tolerance and setpoint uncertainty. This yields a total uncertainty of 43.3 psi. Since this is less than the required 50 psi of margin, PORV actuation should still occur prior to lifting the code safety valves.
j l
Finally, the PORV field setpoint of 2450 psig vill continue to provide adequate margin to the Technical Specification minimum value of 2435 psig. The PORV control circuitry located in the NNI cabinet and tested each refueling outage, has a setting tolerance of 4 psi and a 30 month projected drif t of 8.8 psi. Added to-i l
gether, they total 12.8 psi, which, when subtracted from the 2450 l
psig field setpoint, yields 2437.2 psig, which is still above the l
LAR 95-0024 Attachm:nt 3 Page 5 Technical Specification minimum value.
Historical drift values for buffer amplifier module balance, scaling, and bias have not exceeded the 30 month projected drif ts. The same holds true for the NNI PORV control circuitry.
Since the 30 month projected drif t values have all been shown to be acceptable, the historical drif t values must also have been acceptable. In addition, the as-found trips of the PORV control circuitry have not ever been below the Technical Specification minimum value. This test measures the trip voltage at the input to the NNI cabinet, so it does not reflect errors associated with either the pressure transmitter or the buffer amplifier. If the trip voltage vere to be measured at the buffer amplifier input, then the buffer amplifier balance, scaling, and bias setting tolerances and 30-month projected drift values would also be included in the evaluation.
Adding the setting tolerances algebraically and combining the 30-month projected drif t values using the square-root-sum-of-squares technique results in a total error of 13.4 psi, which, when subtracted from 2450 psig, yields 2436.6 psig, which is still above the Technical Specification minimum value. The pressure transmitter errors do not need to be included in this evaluation because they are the same ones usedi n the RPS high RC pressure instrument strings, therefore, their i
errors are already accounted for in the determination of the 2435 psig Technical Specification minimum value.
The PORV control instrument strings are not included in any setpoint calculations, and there are nc ~nditions or assumptions of setpoint and/or safety analyses n must be reflected in plant surveillance test procedure accep w.Se criteria.
LAR 95-0024 Attachmant 3 Encloture 6 Page 1 Summary of Instrument Drift Study for Reactor Coolant System Leakage - Leak Detection Systems Containment Normal Sump Level (LIT 4617 and LIT 4618)
These instrument strings also satisfy Technical Specification 3.3.3.6, Post-Accident Monitoring Instrumentation, Table 3.3-10, Item 15, containment Sump Normal Level.
The 30 month projected drif t of 10.223 feet is greater than the design basis / reference uncertainty of 10.2194. The 30 month projected drif t from the 24 month drif t study data is a worst case value with a 95/95% confidence factor. This tolerance inter-val was based on a sample set containing only 8 points. This gives a drift correction factor of 3.732. Also, any error intro-duced by M&TE or time dependent drif t was not included in the design basis / reference uncertainty as this information was not available.
No outliers were removed from the data. Historical drift has been acceptable except on rare occasions. No safety related setpoints are associated with the Containment Vessel Normal Sump Level Indicators.
l The assumption of normality was not rejected by the V test for the worst case point. The data for all points and the worst case j
point appeared normal when plotted on a histogram. For the worst case data point set, no data points were greater than 2 standard deviations from the mean. For all data points, 52 out of 56 points (92.86%) were within 2 standard deviations of the mean.
Ninety-five percent of the points are required to be within 2
]
standard deviations for a normal distribution, however, 51 of 56 i
points were at the center of the distribution which makes the standard deviation small. A large peak of data at the center of the distribution is conservative and is bounded by the normal distribution. This supports the assumption of a normal distri-bution.
The data was plotted to show the relationship between the Drift (percent of span) versus the time since last adjustment and drift versus time since last test. These instrument strings have gone over 70 months without requiring an adjustment. In fact, these instruments have not required an adjustment since they were i
installed. The long term plots of drif t versus time since last adjustment do not show any time dependent characteristics. This justifies the assumption that the instruments are not time dependent.
The 30 month projected drift is greater than the Design / Reference Uncertainty.
However, no safety related trip setpoints are associated with the Containment Normal Sump Level indication nor
' LAR 95-0024 Attcchm:nt 3 Page 2 are these instruments used to effect a safe shutdown. A review of the data indicates that these instruments perform better than the worst case point implies. The large tolerance associated with these instruments is caused in part by the digital accuracy of the instruments which read only to the nearest tenth of a foot (20.1 feet). This is large when compared to the 4 foot span of 12.5% of span). A total of 70 the instruments (0.1 feet readings were taken for the instrument strings. Only 3 of the i
readings were not at the desired values. Five of the seven points where data is taken had no points that were not at the desired values (the mean and standard deviation of the percent drift since last test data vere zero). Based on this, these instrument strings are considered acceptable for Containment Normal Sump Level indication.
i
[
l i
l l
l
}
1 4
4 r
LAR 95-0024 Attach:znt 3 Page 1 Summary of Instrument Drift Study for Auxiliary Feedvater System Auxiliary Feed Pump Turbine Steam Generator Level Control System (LTSP9A3, LTSP9A4, LTSP9B3, LTSP9B4, LY6451J, LY6452J, LY6459J, LY6460J, ZT6451, ZT6452, ZT6459, ZT6460)
The instrument strings are calibrated in several sections. The transmitters (LTSP9A3,
- LTSP9B3, LTSP9A4, LTSP9B4),
position transmitters (ZT6451,
- ZT6452, ZT6459, ZT6460) and signal converters (LY6451J, LY6452J, LY6459J, LY6460J) are calibrated separately and the remainder of the string is calibrated as a group.
The instrument drift study determined that the projected 30 month drift for the transmitters and the remainder of the instrument string exceeded the design basis / reference uncertainty. The 30 month projected drif t for the position transmitters and signal converters was less.than the design basis / reference uncertainty.
All drift data for the instrument strings was determined to be normally distributed.
The position transmitters and signal converters determined to be time independent. Historical data meet acceptance criteria except on rare occasions.
LIC 6451 AND LIC 6452 are used to control Auxiliary Feedvater (AFV) flow to the Steam Generators. During an overcooling event, these LIC's are used to diagnose the nature of the event. If the event is caused by a Steam Generator Overfill, the LIC's are used to verify that an overfill event is the cause of the overcooling.
During a minor Steam Line break, the LIC's may be used to deter-mine the af fected Steam Generator. Neither of these functions require an exact indication of Steam Generator level. Based on engineering judgment, the 30 month projected drift of 1.716% for the transmitter and 0.364% for the rest of the instrument string is acceptable because it would not prevent the operator from controlling AFV flow as required in an overcooling event.
Although the level transmitters 30 month projected drif t is much vorse than the design basis / reference uncertainty, these instru-ments are still acceptable for providing a trend of steam generator level.
Per the AFV System Description (SD-015 section 1.2.1.1),
the
" Automatic feedvater control is required for at least 10 minutes after Auxiliary Feedvater System initiation before credit for manual operation can be taken." Therefore, although early accident scenario manual override of the auto level control is possible, it vill not be assumed.
In the auto mode, the AFV Level Control system vill continuously change the AFP level control valve position in an attempt to maintain a start up range level of 49, 55, 124, or 130 inches (of the 250 inch span, setpoint depending on the Steam and Feedvater
LAR 95-0024 Attschm:nt 3 Page 2 Rupture Control System and Safety Features Actuation System status).- These settings were selected to provide assurance that the 36 inch and 120 inch analytical values are protected and to avoid interaction between two pumps feeding the same steam gener-ator. Interactions between competing pumps was determined to be not detrimental to providing sufficient flow to the steam genera-tors. A review was performed to determine if the 30 month pro-jected drift values impacted the margins between the 49 inch and 124 inch indicated field settings and the 36 inch and 120 inch actual analytical values (36 inch and 120 inch actual level in the steam generator is equivalent to 35.8 inch and 94.4 inch respectively in indicated level at low power levels). Using a graded approach to determining the instrument uncertainties, it was found that only the transmitter normal temperature effect needed to be combined with the 30 month projected drift to deter-mine an overall instrument uncertainty.
Square-root-sum-of-
-squares combining of these terms gives 11.45 inch. Thus, when f
this value is subtracted from the 49 inch and 124 inch indicated i
level settings, both associated analytical values (35.8 inch and j
94.4 inch) are found to be protected. Based on this uncertainty review and historical good performance of this level control loop, the projected instrument errors caused by drift were found acceptable for Auxiliary Feedvater control of the steam generator level.
)
Auxiliary Feed Pump Suction Pressure Interlocks (PSL4928A, PSL4928B, PSL4929A,
- PSL4929B, PSL4930A,
- PSL4930B, PSL4931A, PSL4931B)
Four Auxiliary Feed Water (AFV) Low Pressure Switches, PSL4928A, PSL4928B, PSL4929A, and PSL4929B are located on the suction line from the Condensate Storage Tanks (CSTs). These switches auto-matically shift the Auxiliary Feed Pump (AFP) suction to the Service Water System (SVS) in the event suction pressure falls below 2 psig for longer than 10 seconds (the setpoint at the time of the Drift Study). These switches open valves SW1382 and SV1383. When suction pressure from the CSTs increases above 10 psig, SVS valves SW1382 and SW1383 can be manually closed, realigning AFP suction back to the CSTs.
Four additional low pressure switches, two for each Auxiliary Feedvater Pump Turbine (AFPT), PSL4930A and PSL4930B for AFPT 1-1, and PSL4931A and PSL4931B for AFPT 1-2, are located on the AFP suction piping. These switches automatically close or block the opening of the AFPT steam valves if the suction pressure drops below 1 psig for more than 60 seconds. These switches prevent cavitation of and possible damage to the Auxiliary Feedvater Pumps due to insufficient Net Positive Suction Head (NPSH).
The Auxiliary Feed Pump Suction Transfer on Low Pressure Inte-lock setpoint is not based on a USAR Chapter 15 accident analysis nor is a setpoint specified in the Technical Specifications. The
LAR 95-0024 Attechm:nt 3 Page 3 safety function of these pressure switches is to ensure that the Auxiliary Feed Pumps are provided with a water supply to remove decay heat from the Reactor Coolant System. Therefore, these switches must provide transfer prior to a loss of water while not transferring unnecessarily. The current 2 psig setpoint with 10 seconds delay is chosen to maximize usage of the Condensate Storage Tank vater level while providing sufficient time to transfer AFP. vater supply to Service Vater prior to the AFP losing suction. There is no setpoint calculation since there is no analysis being protected and there is not an associated Tech-nical Specification Allovable Value. M&TE uncertainty is.05%.
The reference, accuracy for these switches is 2% or.116 psi. This value is rounded down to yield the calibration tolerance of 1 11 psi as referenced in the DBNPS calibration procedures (DB-MI-3901, DB-MI-3904). Since the M&TE uncertainty is so small it will not be combined with the accuracy for comparison purposes. No specific mention is made of any drif t allowance.
Comparing the procedure's acceptance tolerance of
.11 psi to a calculated 95/95% minimum value of.132 psi indicates that the tolerance may need to be increased to accommodate these new drift values. The difference between the allowed tolerance and the drift study value is less than 1 inch v.c. (where 1 inch v.c.
equals.0361 psi). This difference is negligible with respect to the required delay and the CST level.
The function of the Auxiliary Feed Pump Turbine Steam Isolation on Lov Pressure Interlock is to prevent damage to the Auxiliary Feedvater Pumps from extended pump cavitation. This setpoint is not based on a USAR accident analysis nor is a setpoint specified in the Technical Specifications. Therefore, these switches must isolate after a loss of NPSH vhile not isolating unnecessarily.
The current 1 PSIG setpoint with 60 seconds delay is chosen to balance these two requirements. There is no setpoint calculation since there is no analysis being protected and there is not an associated Technical Specification Allowable Value. M&TE uncer-i tainty is.05%. The reference accuracy fc-these switches is 2%
i or.116 psi. This value is rounded down to yield the calibration tolerance of 1 11 psi as referenced in the DBNPS calibration procedures (DB-MI-3902, DB-MI-3905). Since the M&TE uncertainty is so small it was not combined with the accuracy for comparison l
purposes. No specific mention was made of any drift allowance.
Comparing the procedure's acceptance tolerance of i.11 psi to the calculated 95/95% minimum value of.132 psi indicated that the tolerance may need to be increased to accommodate these new drift values. However, the difference between the allowed tolerance and the drift study value is less than 1 inch v.c. (where 1 inch v.c.
equals.0361 psi). This difference is negligible with respect to the 60 second delay and the fact that the manufacturer of the AFP states that it can cavitate for several minutes before pump damage occurs.
i Review of the as-found data indicates that there have been a few occasions that these devices required re-calibration when found i
out of tolerance. Since this trip is not based on an analytical
LAR 95-0024 Attachm:nt 3 Page 4 limit, videning the tolerance is acceptable. The instrument drift was determined to be time independent.
A review of design assumptions against channel functional, channel calibration, and channel check Surveillance Test procedures was conducted during the Drif t Study Review. It was found that for the Condensate Storage Tank / Auxiliary Feedvater Pump Suction Path Interlock switches, design documentation specified acceptable as-found setpoint tolerantes as one-sided and acceptable as-left setpoint tolerances as dual-sided. The surveillance procedures utilize the (potetially) non-conservative dual sided tolerances. Since the setpoint does not support a safety limit and no formal calculation exists for the setpoint, the existing procedure tolerances are adequate for the existing setpoint. Af ter a review is performed and possible calculation generated by Design, any necessary tolerance and setpoint-changes as well as other design assumptions vill be reflected in the surveillance procedures. C 3 remaining requirements were being i
met in the existing surveillance test procedures.
Auxiliary Feed Pump Turbine Inlet Steam Pressure Interlocks (PSL106A, PSL106B, PSL106C, PSL106D, PSL107A, PSL107B, PSL107C, PSL107D)
Pressure switches PSL106A through D and PSL107A through D close the Auxiliary Feed Pump Turbine (AFPT) Main Steam Inlet Isolation Valves if steam pressure decreases to less than 18 psig. Upon a low pressure signal from PSL106A and PSL106C (PSL107A and PSL107C) and/or PSL106B and PSL106D (PSL107B and PSL107D), a signal is sent to close the main steam isolation valves for the appropriate AFPT. Data for this drift study was taken beginning in October 1989, af ter new pressure switches were installed and the trip setpoint revised.
Review of the as-found data indicates that historical drift has been generally acceptable.
The percent drift since last test was plotted versus the months since last test. This plot does not have any points substantially past an 18 month interval that would support a conclusion regard-ing time dependency for a 30 month interval. This plot does support the conclusion that drift is time dependent within an 18 month interval.
The percent drift since last adjustment was plotted versus the months since last adjustment. This plot provides many points beyond an 18 month interval. The data points were then divided into two bins, for Time Since Last Test above and below 10 months. Statistics were calculated on each of the bins. Compar-ison of the two means reflects a non-conservative drift with respect to time. Comparison of the two standard deviations in-dicates that drift variation increases slightly with respect to time. This evidence supports the conclusion that the drift for
LAR 95-0024 Attechm:nt 3 Page 5 these pressure switches is time dependent.
Vithout clear indication that the drift is independent of time, each data point (percent drift since last test) was linearly extrapolated to 30 months. Basic statistics were calculated and the 95/95% Hin.,
(Max. vas not calculated since a one-sided tolerance factor was utilized) value generated based on the mean, standard deviation, and tolerance interval. The projected drif t is -4.020 percent of span per 30 months which is equal to -3.015 psi per 30 months.
The 30 month extrapolated data (percent drift since last test) was checked for normality by the V Test. Since the test passed, there was no reason to reject the assumption of normality.
A supplemental check of plotting the number of data points versus the number of standard deviations from the mean was performed.
The comparison to that of a normal distribution indicates that the data is approximately normal and 30 of 32 points (or 93.8%)
were within two standard deviations. The assumption of normality is therefore confirmed.
The Auxiliary Feed Pump Turbine Inlet Steam Pressure Interlocks setpoints are not based on Updated Safety Analysis Report Chapter I
15 analysis but are designed to function to allow decay heat removal via steam to the AFPT until the Steam Generator temp-erature is low enough that cooldown can be transferred to the Decay Heat Removal system. The existing calculation is not required to comply with ISA 67.04 since there is no analysis being protected and there is not an associated Technical Speci-fication allowable value. H&TE uncertainty, drift and reference accuracy are all specified in the calculation. M&TE accuracy is treated as negligible. The allowed drift is 3% and the reference accuracy is 1%. An allowance of 3.16% vas determined by applying the square-root-sum-of-the-squares method to these values.
Algebraic combination of these values results in an allowance of i
4%. Since.both values are less than the projected 30 month value a change to the calculation was required. The applicable calcu-1 lation was revised to incorporate the 30 month drift value resulting in a change to the field setpoint.
A review of DBNPS procedures DB-MI-3903 and DB-MI-3906 indicated that these procedures did not reflect the assumptions of the l
existing setpoint calculation with respect to the specified test j
i gauge accuracy. Potential Condition Adverse to Quality Report 95-0884 was initiated to address this issue. It was resolved by determining that sufficient conservative margin had been assumed in the calculation which could accommodate the larger test gauge accuracy. The revised setpoint calculation, to accon.odate the 24 Month drift, stated assumptions, including the revised field setpoint, vill be reflected in the surveillance procedures.
l b
_