ML20117C643

FPC Crngp Unit 3 Graded Approach Methodology for Instrument Uncertainty

ML20117C643
Person / Time
Site:	Crystal River
Issue date:	07/25/1996
From:	Koleff W, Frederick Sullivan FLORIDA POWER CORP.
To:
Shared Package
ML20117C640	List: 3F0896-10, Forwards FPC Crngp Unit 3 Graded Approach Methodology for Instrument Uncertainty ML20117C643
References
NUDOCS 9608270461
Download: ML20117C643 (13)

Text

-

j

~

Rev.O i

l Effective Date 7/drM i

i 1

i Florida Power Corporation Crystal River Unit 3 Graded Approach Methodology for Instrument Uncertainty Approved By: W.S. Koleff, Supervisor Nuclear Engineering Design ld $O0 l_WN i/

F.X. Sullivan, Manager Nuc ear Engineering Design h

/-

17 050ObO2 PDR DO P

PDR

)

I

1 Background

Instrumentation effectiveness is critical to the safe control of any nuclear powe Instrumentation effectiveness is as its compliance with applicable regulatory requirements.

adversely impacted by a large number of variables j

He methodology for and accident), drift, readability and installation considerations.

assessing the uncertainty of such a multi variant system is statistical in are assumed to simultaneously occur at the extreme of their hypothetical ra would be conservative but would not be an accurate representation of the a Rus, valid statistical methods are utilized to more precisely pred under most conditions.

uncertainty.

He key to an appropriate methodology is provid purpose (s). His involves at least two key elements:

(1) ne design (location, range, qualification, etc.) must be appropriate; l

(2) ne setpoints (for actuation, control, alarm, dis i

Some values must meet very rigorous standards (95% certainty w confidence), while others may rely on lower levels of rigor and some significance.

l values.

Here are a number of standard techniques used to establish such value This nese have been used over the years as well as in current applications.

document addresses application of these methods at Crystal River Unit

manner, j

relate to instrumentation. De methods also apply to sirnilar values (ele can be applied to all numerical values within the Technical Specifications.

Some techniques can be appropriately employed to only include unc accounted for in the various values. nese generally involve focusing the ana specific or appropriately limited conditions, statistically combining un l

D ese root of the sum of the squares), reliance on actual historical data, and other techniques are addressed under ' BASIC METHODS' in this docum l

any value.

Other techniques can be applied to a limited extent depending on

?

Dese involve varying the degree of conservatism by eliminating o Dese techniques are addressed under value.

factors, achieving lower confidence levels, and others.

" GRADED APPROACH" in this document.

1 Page 1 Rev.O

i Such In order to apply such an approach, the values must be categorized in some manner.

values may be categorized on a specific parameter, instrument string or even actuation where appropriate. The values are used in a number of places or documents:

l setpoints established by appropriate calibration procedures or other work cont operational procedures (e.g. EOPs, APs, ARs, etc.); various test procedures etc.) or routine operational procedures (ops, etc.). Derefore, for such categor' effective, it must involve integrated input from Operations, Nuclear Engineering Licensing, Maintenance and others at appropriate Nuclear Plant Technical Support, management levels. The basis should be technically sound, adequately cle i

readily retrievable.

Calculational methods that were followed prior to the recent development of detaile i

j standards are and will be considered acceptable. If actual nonconformances (cl inputs, significant errors in implied configuration or use of the values calc l

while revising calenlations to recent standards, they will be handled via standa j

Changes to field values or procedures, which are appropriate based and p m m.

revised calculations, are to be made in an orderly and controlled manner under F i

j prioritization and work control practices.

He following methods and techniques to achieve a Graded Approach are gu should be followed as closely as possible. Circumstances or situations particular The parameter or instrument string may dictate some deviation from this document.

development of instrument uncertainty calculations may deviate, with ap of developing the 1&C Design Supervisor, from this document to satisfy the intent uncertaintie; commensurate with their use and safety significance. This should on a case-by-case basis and justification for deviations should be provided document.

1 Page 2 Rev.O

BASIC METilODS l

in addition to the " Graded Approach" methodology, the following methods may be utilized a appropriate.

i

1. The readability error will be applied in the "Results" section of the calculation ins being treated as an additional variable (Reference 4). Actions will be specified to f scale division (major or minor). A half minor division could be used if defining the a at a full scale division unneceturity restricts the operating band.

An indicator on the Main Control Board has a range of 0 to 3000 psig, with Example: minor divisions every 50 psig. De total error for the loop to the indicator (without considering readability in the component error for the indicato psig. If the operator must maintain the process at a pressure greater than equal to 2200 psig, the indicator could read as high as 2217.6 psig wh actual pressure is 2200 psig.

Since the operator can only read the indicator to one-half minor division (

psig), the operator must be instructed to maintain the pressure to t minor scale increment in the conservative direction, which would be 2225 psig

2. Historical calibration information can be used. If records indicate t can be calibrated to a tighter tolerance or performs better between calibrations than calculated or provided by the vendor, the historical value can be used (As-left or Found tolerances).

3. A reduction in a variable's contribution to calculated uncertainty can be achieved by focusing on predetermined points of interest or limited ranges. Many instrume wide ranges for a variety of reasons. De uncertainty over that wide range ove likely uncertainty over narrower ranges of actuation or monitoring interest.

4. If a value is approached from a single side, appropriate statistical methods sh employed to appropriately consider uncertainty.

5. Only the environment for which the components need to operate will be con If a transmitter is located in the IB, but only needs to operate for a LOCA in Example:

RB, then the effects of an IB HELB will not be considered.

Page 3 Rev.O

i 1

i l

GRADED APPROACH I

A " Graded Approach" is a methodology for classifying values into different categories 1

I Section 4 of varying degrees of rigor being applied to develop instrument uncertainties.

Reference 2 notes that the full application of the methodology developed for automatic i

protection system actuation setpoints, to other values, may not always be appropriate its conservative nature. The classification of the value within this hierarchy of categories j

establishes the calculational methods which would be applied. The Instrumentation Society America (ISA) recommends the use of a " Graded Approach" for determining which and ho l

variables should be accounted for in uncertainty calculations. The approach used must be

{

commensurate with the importana to safety of the value. Reference 4 provides some guidance l

in the development of a Graded Approach but specifies the utility should provide the i

and technical basis for the classification scheme.

l i

Category Descriptions i

• Ihe four Graded Approach categories (A, B, C, and D) are listed below along with the i

l that defines the values included in each category. 'Ihese general criteria describe the content between the categories.

The criteria are of each category and the relative importance guidelines and should not be viewed or utilized as absolute definitions. Further, th l

may overlap somewhat within a category and between categories. That is part of th why it is important for personnel with a wide range of expertise, experience and r j

l to be involved in establishing categories and in the selection of calculational options avail f

within each.

i

\\

'Ihe categorization scheme is based on the specific function of a value, parameter, or l

indication. In some cases a single instrument or instrument string provides functions that j

reside in more than one category. Given this, it may be necessary to provide more than one l

In addition, some limits, parameters or values are set of channel uncertainty values.

considered to be " Nominal Values" and will require no offset or margin. 'Ihe justification l

considering a limit, parameter or value a " Nominal Value" should be documented.

i j

Category A t

1) Reactor Protection System (RPS) actuation setpoint i

2) Emergency Safeguards Actuation System (ESAS) actuation setpoint i

3) Emergency Feedwater Initiation & Control (EFIC) actuation setpoint J

?

l-

)

i i

Page 4 i

Rev.O i

i i

i Category B

1) A value which is directly used in applicable safety analysis. (The value has an impact on the performance of the safety function and no margin was allotted in the analysis.)

2) A value/ indication relied upon for the manual performance of an accident mitigation function (i.e. Reg. Guide 1.97 Type A variables)

3) A value relied upon to maintain the reactor in a safe shutdown condition, maintain the integrity of the safety related pressure boundary, or prevent exceeding 10CFR100 limit

4) A value relied upon to protect or support operation of equipment required for design accident mitigation Category C

1) Other actuation's, alarms, indications or control settings relied upon for continued operation and equipment protection, excluding items classified as Category D.1 (se below).

2) Vrhes derived from commitments in the Licensing Basis (including Technical Specification) not classified in Category A or B and which cannot be appropriatel as a Category D or nominal values.

3) Reg. Guide 1.97 Category B, C, and D variables Category D

1) Vendor supplied equipment (i.e.: Cardox system, turbine protective features, etc.)

2) Reg. Guide 1.97 Category E

3) Other values not meeting the criteria of Category A, B, or C

4) Values derived from commitments in the Licensing Basis (including Technical Specification) not classified in Category A, B or C and which can be appropria as nominal values with engineering judgment.

Page 5 Rev.O

TREATMENT OF ERRORS FOR GRADED APPROACH CATEGORIES Category A His is the highest category in the " Graded Approach". References 1 and 2 shall be used for development of the Analyses / Calculations.

All applicable errors shall be statistically accounted for in the Analyses / Calculations to provide a high confidence level. The following formula is used.

CEux, = fi(Eux,)* + (AFux,)']" i Em i Enmas Where:

CEm = Calibrated Loop Error - De overall instrument channel error, which is used to determine setpoints and action values from the design limit / analytical limit.

j Es,oo, = Calculated Loop Error - The instrument channel error, not taking into account calibration, drift, process errors and known biases.

AFm = As-Found Tolerance - De tolerance in which a channel / loop can be found after a period of operation, prior to calibration. His term includes the errors due to 3

M&TE and Drift / Stability.

Esas

= Bias Errors - Know biases that affect the ciperation of an instrument loop, such as static pressure shifts, IR effects, etc.

l Ermocsss= Process Errors - The error that results from the range of process operation limits, based on the scaling of the sensing instruments.

i Category B his is the second category in the " Graded Approach".

The methodology for this category will be the same as for Category A instrument strings, except that the Normal Process Errors will be combined via the SRSS method with the other random loop errors. In addition 2/3 of the M&TE error will be used.

i This method still ensures appropriate actuation's/ actions are taken. Nevertheless it does not compromise the ability to use the instrumentation by more accurately reflecting actual uncertainty.

CEux, = 1[(Eux,)* + (Em)' + (Alux, + [(20 MTEux,)* + (SBux,)']')')" i Em i Em Rev.O Page 6

2 Where:

= Normal Process Errors - De error that results from the range of Normal Eurg process operation limits, based on the scaling of the sensing instruments.

Alm, = As-left Tolerance or Calibration Tolerance - De tolerance to which a channel / loop is left after calibration. This term is determined from the Reference Accuracy of the components.

M&TE (Maintenance & Test Equipment) error - De errors due to the MTE oo,=

t M&TE used in the calibration of the loop.

i SBwo, = Stability / Drift - De error due to the stability / drift of the components 1

loop.

= Accident Process Errors - De error that results from the range of Accident Em process operation limits, based on the scaling of the sensing instruments.

In addition, if the AL oor (setting tolerance)is less than or equal to the SRSS of the Re 4

Accuracy of the components in the loop, ALwo, will not be accounted for. Dere t

above formula will be reduced as follows:

s CE oor = ii(E oor)' + (Em)* + (2/3 MTE oor)' + (SB on,)'j' i Emu i Em t

i t

t j

Category C

'Ihis is the third category in the " Graded Approach". De methodology for this catego l

be the same as for Category B values or variables, except the overall error is reduced De final values provided will be more in line with what can be expected from the below.

variable on a normal daily basis.

CE oo, = i 2/3 * (l(E oo,)* + (Em)* + (MTE oor)' + (SBw)']" i Emu i Em) t t

t If any biases or uncertainty data should not be reduced by 1/3 (i.e.: flow Note:

errors), they would be added onto the end of CE oop.

t Category D This is the fourth category in the " Graded Approach". The values or variables that into this category have no impact on plant safety. Errors / uncertainties for this category based solely on engineering judgment or can be treated as Nominal Values.

Page 7 Rev.O

l l

1 Additional instrument strings that would fall into this category are vendor supplied systems (primarily skid mounted). If the vendor provides initial or final setting values f equipment, it would not be appropriate for engineering to change these values correction. The values provided are normally from operational experience with their s!

and may be changed by the vendor upon installation to optimize the performance.

(

i l

l l

l r

l l

l

~

l l

Page 8 Rev.O

t I

TECilNICAL BMij i

1

1. SRSS of Process Errors:

l A method commonly used in the industry is to combine the normal process errors with other random errors using a SRSS methodology. Normal process errors vary randomly 1

between design limitations based on several factors (such as plant conditions, chemi compositions of the process and environmental conditions) and may have an addit j

subtractive impact on the total channel uncertainty. Given this, it is reasonable to assume j

that the normal process errors can be treated as a random event thus justifying their inclusion under the radical with other random errors. Treating these enors as random l

yielding a reduction in the overall channel uncertainty will more reasonably rep In addition, actual combination of errors seen at any given time during normal conditions.

this methodology is consistent with the methodology used by much of the industry.

I 4

d This method of combining normal process errors is not intended to be used on those errors l

which are considered bias errors such as IR drop errors, transmitter static pressure span l

compensation errors or reference leg temperature errors during accident conditio Reference 5 specifically cautions about the treatment of these type errors in the l

l development of channel uncertainties. Due to the potential impact on channel u l

as a result of biases and process errors during accident conditions, these terms will not b I,

reduced.

i

?

1

2. Omission of Alm Tolerance:

ALux, (calibration tolerance) is the acceptable parameter variation limits above i

i l

the desired output for a given input standard associated with the calibration of th instrument channel. Typically, this is referred to as the setting tolerance of the wid l

l the "As-Left" band adjacent to the desired response.

Depending on the method of calibration or performance verification, an allowance for the calibration tolerance j

If the method of calibration or performance be included in the uncertainty calculation.

verification verifies all attributes of reference accuracy (reference accuracy is typically assumed to have four attributes: linearity, hysteresis, dead band and repeatability) a method of calibration verifies all attributes of reference accuracy, then the calibratio l

tolerance does not need to be included in the total instrument channel uncertainty case, the calibration or performance verification has explicitly verified the instrum l

channel performance to be within the allowance for the instrument channel's r accuracy in the uncertainty calculation.

j i

i i

Page 9 Rev.O 1

l

1 o.

If the method of calibration does not verify all attributes of the reference accuracy, this method should not be used in the analyses / calculation. If specific values for linearity, hysteresis, deadband and repeatability are provided by the vendor, they are address the analyses / calculation under the calculated loop error.

His methodology is discussed in Section 6.2.6.2 of reference 2.

3. Reduction of MTEm by 1/3:

MTEu,,, (Measurement & Test Equipment uncenainty of the loop) errors are con production lot specifications and are certified for use in the field by our calibration Dese errors are considered to have a 3 sigma confidence level due to the tight controls nis error term will be reduced to the same confidena level as the applied to them.

overall analyses / calculation. Hence, MTE errors will be reduced to two-sigma (two-t l

of the 3-sigma values) to more reasonably represent the actual combination of errors seen i

l

4. Reduction of CE oo, by 1/3 (Category C only):

t CEu,,, (Calibrated Loop Errors) for instrumentation typically apply the worst case (maximum and minimum) design limits for each parameter, (reference accuracy, drift process errors, etc.), to assure the bounding or worst case errors are obtained. De

,l approach of assuming all parameters at the worst case design limits assures maxim j

conservatism, but can unnecessarily increase instrument total loop uncertainty and result' reduced operating margins or cause operations personnel to initiate manual actions at J

inappropriate time.

Typically systems will not be operated at worst case limits nor will accident transien l

force all process parameters to the worst case limits simultaneously. Hence a reduc total channel uncertainty is warranted for Category "C" parameters, nc comb'mation of 4

worst case errors encompass or represent 100% (3-sigma) of all errors. It is reasonable to l:

assume that the errors represented in the Category "C" calculations should provide a 95%

confidence level that the actual worst case errors are enveloped. Hence, the worst case error limits will be reduced to 2-sigma (two-thirds of the 3-sigma values) to more reasonably represent the actual combination of errors seen at any given time during l

normal or accident conditions.

His reduction in normal and accident errors is not intended to be used to reduce process errors which are known to be 2-sigma.

j

)

I i

Rev.O Page 10

1 TERMS USED IN THE ABOVE GRADED APPROACH DOCUMENT

,t CEwo, = Calibrated Loop Error - The overall instmment channel error, which is used to determine setpoints and action values from the design limit / analytical limit.

E

Ecow,

= Component Error - The SRSS of the errors associated with an individual j

component (i.e.: Reference Accuracy, Temperature Effect, etc.), with the exception of Drift.

E oo,

= Calculated Imp Error - The instrument channel error, not taking into account t

calibration, driR, process errors and known biases.

i[M + (Ecam)* +... (Ecm, )']"

8 En

=

i MTEwor= M&TE (Maintenance & Test Equipment) error - The errors due to the M&TE used j

in the calibration of the loop.

SBwo, = Stability / Drift - De error due to the stability / drift of the components in the loop.

ALwor = As-Ieft Tolerance or Calibration Tolerance - The tolerance to which a channel / loop is left after calibration. This term is determined from the Reference Accuracy of the components.

Alw = i[(COMPI-Eas,)* + (COMP 2-Eas,)' +... (COMPN Emar)'l" AFwo, = As-Found Tolerance - De tolerance in which a channel / loop can be found after a period of operation, prior to calibration. This term includes the errors due to M&TE and Drift / Stability.

h - *(h + I(MTEuw)* + (SBuw)'l")

Esas

= Bias Errors - Know biases that affect the operation of an instrument loop, such as static pressure shifts, IR effects, etc.

Ermoess: = Process Errors - De error that results from the range of process operation limits, based on the scaling of the sensing instruments.

Eups

= Normal Process Errors - The error that results from the range of Normal process operation limits, based on the scaling of the sensing instruments.

Em

= Accident Process Errors - The error that results from the range of Accident process operation limits, based on the scaling of the sensing instruments.

Nominal Value

= A value determined to require no error correction.

Rev.O Page 11

REFERENCES

1. IS A 567.04, Pan I, "Setpoints for Nuclear Safety Related Instrumentation", dated 1994

2. ISA RP67.04, Pan 11, " Methodologies for the Determination of Setpoints for Nuclear Safety-Related Instrumentation", dated 1994.

3. Draft Technical Report ISA-dTR67.04.03,

" Indication Uncertainties and Their Relationship with Indicated Values", Draft 4, April 1996

4. Draft Technical Repon ISA-dTR67.04.09, " Graded Approach to Setpoint Determination",

Draft 1,1994

5. IE Bulletin No. 79-21, " Temperature Effects on 12 vel Measurements", August 13,1979 i

Rev.O Page.12