Text

Request Commission Approval to Implement Enhanced Performance Indicator Rept Which Incorporates Peer Group Comparisons & Operating Cycle Phase Adjustments
	ML20034D515
Person / Time
Issue date:	12/23/1992
From:	Taylor J; NRC OFFICE OF THE EXECUTIVE DIRECTOR FOR OPERATIONS (EDO)
To:	;
References
	SECY-92-425, NUDOCS 9301040050
	Download: ML20034D515 (32)
	v • d • e

.

...................ooooo-RELEASED TO THE PDR f

4h/95 4

f-A eas a

.................p......:-.

g

.g

'E o\\..e/

+

POLICY ISSUE December 23, 1992 SECY-92-425 For:

The Comissioners From:

James M. Taylor Executive Director for Operations Sub.iect:

PERFORMANCE INDICATOR PROGRAM - PEER GROUP AND OPERATING CYCLE PHASE ENHANCEMENTS

Purpose:

To obtain Comission approval to implement the enhanced Performance Indicator. Report which incorporates peer group comparisons and operating cycle phase adjustments.

Backoround:

In SECY-89-211, Performance Indicator Program - Cause Codes, the staff recomended that cause codes be displayed in a manner similar to that used for other performance indicators. However, the appropriate' Nuclear Steam Supply-System (NSSS) average would be used in place of the industry average in the " Deviations from Older Plant Means" chart.

In the Secrezary's memorandum of August 10, 1989, on SECY-89-211, the Comission approved the inclusion of cause code trends in 1de NRC quarterly. Performance Indicator Report.

However, the Comission did not approve the display of cause code deviations. Rather, the Comission requested that the staff assess the validity of comparing individual plants to their NSSS average in light of the plant-to-plant variations

Contact:

Donald E. Hickman, AEOD 492-4431 SECY NOTE: TO BE MADE PUBLICLY AVAILABLE WHEN THE FINAL SRM IS MADE AVAILABLE.

0 050066 ig )

y f) n i

CEoi04cosox0

(

i The Commissioners 2

3 within NSSS groups which could cause the sets of possible reportable events to differ among licensees.

In SECY-89-280, Performance Indica 1ior Program - Cause Codes, the staff discussed plans for developmental activities to address the differences in the sets of possible reportable events among licensees, and to make recommendations for plant-to-industry comparison methods.

The staff initiated a study at the Oak Ridge National Laboratory (ORNL) to determine appropriate " peer groups" for comparing reportable event data among licensees. Early in the peer group development effort it was recognized that cause code data were cyclic with a period approximating the refueling interval.

Further investigation of this behavior led to the conclusion that differences in operating phase (startup, power operations, refueling, etc.) could have as much or more effect upon the frequency of occurrence of reportable events as differences due to peer groups. To address this issue, the staff initiated a study at the Idaho National Engineering Laboratory (INEL) to identify those phases of operation in which the frequency of occurrence of reportable events varies significantly.

In SECY-90-3IO, Performance Indicator Program - The Status of Cause Codes Comparisons, the staff discussed the status of the peer group and operating cycle studies, as well as the development and evaluation of enhanced methods to incorporate the results of those studies into the PI report.

During the evaluation phase of the program, the staff discovered several issues concerning the enhanced methodology which required further analysis.

In SECY-91-236, Performance Indicator Program - The Status of Cause Codes Comparisons, the staff presented the results of the Peer Group and Operating Cycle studies, described the evaluation phase and the issues of concern that were identified, and discussed future plans for incorporating these proposed changes into the PI program.

In SECY-92-083, Performance Indicator Program - Progress on Incorporating Peer Groups and Operating Cycle Phases, the staff presented the peer groups, the operating cycle phases, and brief descriptions of the proposed calculational and display enhancements that had been aeveloped. That paper I

i also described a trial period for evaluation and refinement of the new techniques prior to making a recommendation to l

the Commission.

I

I V

The Commissioners 3

t Discussion:

This paper discusses the results of the trial period. A typictl plant presentation is shown in Attachment 1.

The final peer groups, operating cycle phases, and enhanced calculational and display methodologies are discussed in more detail in Attachment 2.

This paper also transmits to the Commission the enhanced Performance Indicator Report for the third quarter of 1992.

Trial Period Studies conducted by ORNL and INEL identified appropriate peer groups and operating cycle phases for calculating plant self-trends as well as deviations from the performance of similar plants.

In further work at both labs, enhanced methods for calculating and displaying plant trends and deviations were developed and refined. These methods were evaluated in three draft quarterly reports during the trial period between March and September 1992.

The Performance Indicator Interoffice Task Group (ITG) was involved in the evaluation and refinement of the techniques.

The members of the ITG found the enhanced format to be more informative than the current format.

In addition, the staff compared the enhanced report with the existing report to assure that the new methods improve the staff's understanding of plant performance. Among the issues evaluated were the following:

the type of information and level of detail provided in the displays, the choice of peer groups and operating cycle phases, and e

the appropriateness of the new calculational methods.

The enhanced report was distributed to senior NRC management at the Senior Management Screening Meetings. Their responses ranged from "no objection" to "quite good."

Displav Modifications Performance Indicator data are displayed for each plant in Part I of the PI report. The changes to these displays are designed to (1) show performance data from operational and i

shutdown periods separately, (2) show performance data relative to a group of similar plants, and (3) indicate the statistical significance of the calculated trends and deviations. These modifications allow substantially more information to be conveyed clearly and concisely (see

, ). These display changes include the following:

)

l

The romissioners 4

twelve quarters of data rather than eight, a time line showing plant status (operating or shutdown) e rather than critical hours, cause code quarterly values rather than moving averages, e

e event bars shaded to differentiate between events occurring in operation, shutdown, or startup, trends and deviations shown separately for operating and e

shutdown periods, cause code deviations added, and e

trends and deviations bars shaded to indicate their e

statistical significance.

Methodoloav ImDrovements Peer Groups and Operational Cycles The ORNL Peer Group study examined factors affecting event reperting, excluding operating cycle effects. The INEL Operating Cycle study identified phases of an operating cycle in which event reporting varies significantly. The results of these two studies are the nine peer groups and five operating cycle phases described in' SECY-90-310. A listing of the plants assigned to each peer group is provided in Attachment 2.

For display purposes, the five phases were combined into the two general categories of operation (startup, power operation, and pre-refueling) and shutdown (refueling and non-refueling outages).

Calculational Modifications The changes to the calculational methods are designed primarily to determine (1) operating and shutdown i

performance separately, (2) performance relative to a group of similar plants rather than the entire industry, and (3) the statistical significance of the calculated trends and deviations. These modifications provide a substantial increase in the amount of information to be presented in the PI report.

The methods of computing trends and deviations from event counts are based on simple and commonly accepted techniques.

Plant trends are based on the normalized rates of change of event counts, while deviations are based on the normalized differences between the plant and its peer group medians.

Both short term and long term measures are calculated to provide additional perspective on a plant's performance.

1 i

l l

i The Comissioners 5

The statistical significance of these measures is also computed to determine the probability that a calculated trend or deviation is due to a real change in plant performance rather than just the random variation that occurs in event counts.

==

Conclusion:==

In comparison to the current PI report, the enhanced report provides more information about plant performance trends.

It also provides visual indications of the significant contributors to those trends.

Recommendation:

That the Comission approve the use of the 1.ihanced Performance Indicator Report, as shown in the attached draft version. The staff recomends that the fourth quarter 1992 report be issued in both versions to introduce the new methodology and to complete the calendar year with the current version. The staff further recomends that the enhanced report replace the current report beginning with the first Quarter 1993 report, to be issued on or about June 15, 1993.

/

ames M.

or x

,/ Executive irector for Operations Attachments:

1.

Plant Presentation 2.

Peer Group and Operational Cycle Enhancements to the Performance Indicator Report (Copies for Commissioners, SECY, 0GC, and ED0 only) l

)

i

-i f

l 6

l s

t Commissioners' comments or consent should be provided directly to the Office of the Secretary by COB Monday, January 11, 1993.

Commission staff office comments, if any, should be submitted to

+

the Commissioners NLT Monday, January 4, 1993, with an information copy to the Office of the Secretary.

If the paper is of such a nature that it requires additional review and comment, the Commissioners and the Secretariat should be apprised of when r

comments may be expected.

DISTRIBUTION:

t' Commissioners f

OGC CAA IG i

OPP EDO ACRS SECY 9

1 f

5 t

I i

a 5

4 l

b

y e

4 i

I 1

t 9

1 ATTACHMENT 1 4

PLANT PRESENTATION L

i i

,o ATTACHMENT 1 1

QUARTERLY DATA l

Legend:

5*'

i PLANT NAME Feer Avg Trend Ceeretion s -. 3 i

inaastry Arg trend Shutoo.n m Fe'vehng R

tet Shown t>smo Co Cycie E,Ym r

a n

~

c f

R R

5

1 w-2 aw, r.->

w.i u-J w...

m.a ir, w sw>

w-u-J voar - Oater ver - Oster Autorrete Scrcrm wrine Criticcs safety System Actmtions 4

4 b

l"

$3.

Is.

x 4

E2.

2 t

i

{ '-

$1 9

m_;

= ' -

mo m

e t

337- ; j

.g an

',2 - s W -J a

w-i w-J in w u

. > si s C -J 4

%- 5 W-J in k l'

ew>

Yeor - Oater Year - Oater 1

Significor1t Events Safety Systern rcities I

f15 gs.

3

.2 a,

-0 t

o*

N' Q

2 2

2 2 L

e Jg t

e to q

r 2 c3 -

p.

g i.

m

._..4, I""

i C

~~w-J ' 'das w>

w-i li.J -

w-,'-~

m-J

.w.

swy w.i tm i

v.

i Year - Osorter Year - Oster Ecpoment Forced Outo; pes /

000 CorTyriercics Hxrs fcreed Outoge Rote (s) s

_ mo 5

7 E oa _

b M

e5'

-E '* -

w

'O ep l

I b, *,_

e er

^' E' '

E 55 =b E

bbbIN'7NI'I*

O E'N~ -

C

=

sws

%- s M -J

s e-}

M-5 W-J 6

by W.J ag e. W e

WB K-J Yea'b e b-Oster Year - Quar-ter r

Care Codes C. Admin

b. LiC Oper

c. Other Per Myic,n ge x

y

no.

h m.

c.

e.

i gE s.

s.

s,

..j e[fkbla.fE,e tv dPMm

".I ro.,' ":!N$. b.

'N

d. M:pnt

e. Desrqrn

f. MSc.

~

e f

l R

$2

=

g, 4e g

m.,i _8 n>w w+ _- 0,~te,

.7E s.

m.,

o v e,. -.

$,te Aver 0ge RCdotoon (apostse e.

Y MDM r E e

r*,._ r n*

e 1-1

c TRENDS AND DEVIATIONS PLANT NAME

'*9e* Stati$"cd si ance 9

Medstn n s

?

=

PEER GROUP NAMC toe 1 Plant Devictions From Self-Trend Peer Group Medicn Declined improved Below Above OPERATONS Automatic Scroms Whde Critied -

_oj 4

-o2 Safety System Actuotions -

-0.6

-. -os 1,

)

[' {aj significant Events -(g}{js g

-o4

-ts r-Sofety System Feitures -

.g...... -jj

-o s

-cil q

t Cause Codes (ALL LERs)

. -,. - u,, ~ ~.

4,-

m. Lacreed Operate N.

at 3' w s 54

c. opw Pa w Em'-

-ul

~

-88 e - ~ ~ g.o

- M _o

. w - w - ~ ~ - m.u

_uC f, unes<m.m _

_,3 l SHUTDOWN Sofety System Actations -

o o

Sipficent Events -

o o

f Sofety System Foikses -

c3

-03 Cause Codes (ALL LERs)

]u

]u aa,,. - c-** * = _

o.,~~_

L otPW Preens (pror=

u l6J gu

]a e-*_

. w-~ ~ ~~ _

gu l

i

..- ~ _

1 FORCED OUTAGES roceed outoge note * - gg -to

-o.7 t

~[-

tyircagster-es 63,,,, o.o o's to

-io 65 o.o - o's

- to

-io Not Cacubted for Operctional Cyce p

p.a-,,-. eian e

1-2 1

F 1

if PLANT NAME' f

Year - Calender Ooorter Tvre P5ase

.E9-4 90-1 90-2 90-3 90-4 91-1 91 91-3 91 92-1 93 p2-3

-l S RAM Operatione 0

-0 0

0 0

1 0

2-0 0-Fre* Refueling C

.0 0

1 0

0 0

0' O

O 5tartup D

0 0

.1 0'

O O'

O 0

0

. l l

SSA Operatione 1

0 0

1

'O 2

0.

0 Fre-Refueling 0

0 0

0' O

O O-0'

'O O

O O.

Startup 0

0 0

'O O

O 0

0

' 'I Refueling 0

0 0

-0 0'

-0 Non-Refueling C

01 0

0 0

0-0 0

0-0

.)

SE Operatione 0

0 1

0 0

0 1

'0, i

Startup 0

0 0

0 1 0.

0 0

0 e

Fre-Refueling 0

0 0

0 0-0

.j 0

'0' Refueling 0

0-0 0

0 0-0 0

0 0-.

0 0

1 Non-Refueling 0

0 0

0 SSF Operatione 1

2 2

0 0

0 1

1 1

~ 0 1

-. 0 :

Pre-Refueling 0

0 0

'0 0

0' 0

0 I

Startup D

D 0

0 0

2 0

0 0

0 0'

0?

Refuelang 0

0 0

0 1

1 0'-

0 0

'O O

O Non-Refueling 0

0.

1 0

0 0

0 1

0 0

1 2

e FOR (1) 3 0

25 0

0 10 40 17 7

9-65 0

(

EQ. FOR 0.46 0.00 0.00 0.00 0.00 1.34 0.00 1.05 0.00 0.99 2.01

0. 00 --

}

CRI2. ERS 2161 2160 1672 2115 0

747 1338 1898 2078 ~ 2021 497 0;

~

RAD 471 168 32 52 500 117 33-61 177 27 49 EA j

CAi;SE CODES:

Admin.

Operatione 4

3 2

1 0

0 0

1 C

2 0

0 Pr e-Re f ueLirig 0

0 0

3 0

0 0

0 l

Startufng '

O O

O 3

0 0

Refuel 0

0 0

0 3

3 0

0 0

0 I

Non-Ref ueling

(

0 0

0 1

1 0

0 4

1 Lic. Oper.

Operations 0

0 0

0 0-0-

0 0

Pre-Refueling C

0 0

0 Startup 0

0 0

0 Refueling 0

0 0

0

' O O

-0 0

j Non-Refueling 0'

0 0

0 1

0 y

1 2

0 0"

OtA. Per.

Operatione 2

0 1

3 0

1 1

2.

.0 0.

0 0

Pre-Refueling 0

0 0

0.

0 0

Refueling -

0 0

4 1

0

-0 ~

0 0

0-0-

Startup 0

0 0

1 0

0 0

0-

.I Non-Refueling 0

0 1

0 0

0 1

1

+

Maint.

Operatione 4

2 4

3 0

1 5

5 3

7 2

0 i

Pre-Re fueling 0

0 0

5 0

0 0

0 l

Startup 0

0 0

3 0

0 0

0-Y Refuellns 0

0 0

0 12 3

0 0

i Han-Refueling 0

0 1

0 0

0 2

1 0

0 5

2

?

Design Operatione i

1 2

2 0

0 1

2 0

1 3

0

.I Pre-Refueling 0

0 0

0 t

Startup D

0 0

1 0

0 0

0 Refueling 0

0 0

0 1

1 0

0 0

0 Non-Re f ueling C

0 0

0

.2 2

[

Misc.

Operatione 0

0 0

0 1

l' 0

0

.. I Pr e-Ref ueling 0

0 0

1 0

0 0

0

i Startup 0

0-0 0

0 0

Refueling

.0 0

0 0

0 0-f Non-Refueling 0

0 0

0 1

0 0

0 I

)

Phase Pbese type Start End 1.enath Operation Operation 10/01/89 05/21/90.

233

)

hon-Refueling Shutdown 05/22/90 06/12/90 22

)

Operation ration 06/13/90 09/03/90 83 t

l Pre-Refueling retica 09/04/90 09/28/90 25 e

Refueling uteaum 09/29/90 02/21/91 146 l

Start up Operation 02/22/91 03/18/91 25

- Operation ration 03/1 91 03/29/91 11

)

a utdoma 03/3 91 05/05/91 37 Non-Refueling Operation Operatica' 05/0 91 07 18/91 74 Han-Refueling Shutdoun 07/19/91 07 23/91 5

t Operation Operation 07/24/91 00 02/91 41 i

6 Non-Refueling Shutdown 09/03/91 09/06/91 4

Operation Operation 09/07/91-10/15/91 39 I

Non-Refueling Shutdown 10/16/91 10/20/91 5

'fi Operation Operation 10/21/01 04/21/92 184 Non-Refueling Shutdown 04/22/92- 09/30/92 162 k

21me used in calculatione

+

Trend Celeulations Deviation Celeulations

@SiD 10/01/91-04/21/92 199 days (incl.

O days e/u) 03/25/90+04/21/92 540 doye (incl.

25 days stu) 07/03/92-09/30/92 90 days (incl.

O days rei)

S7D 05/02/91-09/30/92 100 days.(incl.

O doye ref)

FOR 01/04/92-09/30/92 270 FOR 04/09/91-09/30/E2 540 E

f r

1-3 9

t

. }

. i r

t

' tt I

i

?

' t i

. i e

3 ATTACHMENT 2 PEER GROUP AND OPERATIONAL CYCLE 1

ENHANCEMENTS TO THE PERFORMANCE INDICATOR REPORT ~

i l

s 5

h f

1 j'

r e

5 i

i I

t a

6 4

t l

w w

I I

I 1.

INTRODUCTION In February of 1989, after a six-month trial program, the staff recommended the inclusion of cause code data in the Performance Indicator (PI) Program as a separate indicator. The cause code indicator identifies the cause(s) of each reportable event in six programmatic areas.

In July of 1989, the staff recommended that cause codes be displayed in a manner similar to that used for other performance indicators except that the appropriate Nuclear Steam Supply System (NSSS) average would be used in place of the industry older plant 4

average in the " Deviations from Older Plant Means" chart. The Commission approved the inclusion of cause code trends in the NRC quarterly Performance Indicator Report. However, the Commission did not approve the display of cause code deviations.

Rather, the Commissicn requested that the staff address how the sets of possible reportable events differ among licensees.

Specifically, the staff was asked to consider differences between older and newer plants due to complexity or number of technical specifications, between two, three, and four loop plants, etc., prior to the Commission's approving cause code deviations. The staff therefore initiated a study at the Oak Ridge National Laboratory (0RNL) to determine appropriate " peer groups" for comparing reportable event data among licensees.

l Early in the peer group study, it was recognized that cause code data are cyclic in nature, with a period approximating that between refueling outages.

Further investigation of this behavior led to the conclusion that differences in operating phase (startup, power operations, refueling, etc.) could have as much or more effect upon the occurrence of reportable events as differences due to peer groups. Therefore, the staff initiated a study at the Idaho National Engineering Laboratory (INEL) to identify those phases of an i

operating cycle in which the frequency of occurrence of reportable events varies significantly.

To incorporate the results of the peer group and the operating cycle phase studies into the PI Program, the staff undertook a joint effort in cooperation with both labs to develop new calculational and display techniques. The goal of this task was to provide more meaningful plant performance information in a manner that could be easily understood.

]

This document discusses the results of these efforts.

1 i

2-1 3

l

i b

i s

2.

RESULTS:

THE ENHANCED PERFORMANCE INDICATOR REPORT l

2.1 Background

The Performance Indicator Report is published in two parts.

Part I includes a two page display for each plant, the left page showing the plant's quarterly data and the right page showing the plant's self-trends and deviations from I

its peer group median. The changes to these displays are designed to (1) show performance data from operational and shutdown periods separately, (2) show performance data relative to a group of similar plants, and (3) indicate the statistical significance of the calculated trends and deviations. These modifications allow substantially more information to be conveyed clearly and concisely. Also included in Part I are Srief descriptions of each of the PI events for every plant for the past four quarters.

In addition to plant specific data, Part I contains an executive summary which presents industry average performance by quarter and, in each fourth quarter report, by year.

[

The executive summary has been modified to include comparisons among peer groups.

Part Il contains tabulated PI data, including a one page breakdown of the data by operating cycle phase for each plant. This section describes the important features of the new data presentations in both parts of the enhanced PI Report.

2.2 Part I a

2.2.1 Quarterly Data - Figure 1 presents a typical quarterly data display. It is similar to the currently approved display with the enhancements noted. The plant name and peer group are shown in block A.

In block B is the legend, which identifies the coding used to differentiate between phases and trend lines. The two blocks labeled with the letter C show the plant's operating profile, indicating periods of time when the plant was operating or shutdown (refueling outages are indicated by the letter R). This information replaces the critical hours curve in the currently approved report.

It is repeated above each column of PI data to facilitate comparison of the data with the plant operational condition. Note that the most recent 12 quarters are shown rather than the 8 quarters displayed in the current report. This is consistent with the maximum time limit allowed for calculation of trends and deviations (see section 5.2).

The four types of PI events - automatic scrams while critical, safety system actuations (SSAs), significant events (SEs), and safety system failures (SSFs)

-are displayed in blocks D.

The presentations are similar to the currently approved report format. The number of events per quarter is shown in a stacked bar format, however, with the bars shaded to indicate whether the events occurred in startup, operation, or shutdown. A peer group long term l

trend line has been added so that plant performance can be compared with the i

average performance of both its peer group and the industry (or, for SSFs, with the appropriate PWR or BWR average). The three long term trend lines (for the plant, its peer group average, and the industry average) are l

2-2 P

.e I

calculated using a linear regression rather than the moving average used in the current report. The most recent 12 quarters are shown.

The three blocks labeled E contain the forced outage rate (FOR), equipment forced outages per 1000 comercial critical hours (EFO), and collective radiation exposure (CRE) indicators.

These PIs are almost the same as in the currently approved report. Data for these three PIs are not separated by operating phase because forced outages can occur in any phase, and radiation exposure is reported as a quarterly value without regard to plant mode. The peer group long term trend line has been added and the most recent 12 quarters are displayed.

Block F shows the display for the six programmatic cause codes. These are collectively treated as one indicator. This display is very different from the currently approved format, which shows only the long term trend line.

Stacked bars identify by quarter the numerical value of each cause code and i

the phase of plant operation in which the event occurred. Trend lines are not shown because they clutter the graphs. Short term cause code trends are provided in the trends and deviations display on the right page.

2.2.2 Trends and Deviations - A typical trends and deviations display is shown in Figure 2.

It appears quite different from the currently approved display, with most PI trends and deviations calculated separately for operations and shutdown. Standard statistical tests are used to determine the probability that an observed trend or devii.lon represents a real change in performance rather than random variation in the frequency of events.

Cause code deviations from the peer group median have been added. Differences between the two displays are noted below.

The plant name and peer group are shown in block A.

In block B is the legend, which identifies the coding used to indicate the statistical significance of the calculated trends and deviations.

Column C shows the plant short term self-trends for the operations and shutdown phase types, and for the FOR and EF0 indicators that are independent of phase. Automatic scrams while critical are not shown for the shutdown phase, since the reactor is subtritical. The bars represent the trend in y

plant performance (declining performance to the left, improving performance to the right) for each PI over the last six months to one year. These trends are i

calculated using a linear regression rather than comparing averages values, as is done in the current methodology. The statistical significance of each trend is indicated by the shading of the bar.

Column D shows the plant deviations from its peer group median performance for the operations and shutdown phase types, and for the FOR and EF0 that are independent of phase. The bars represent the deviations (performance worse than that of the median plant to the left, better than that of the median plant to the right) for each PI over the last 18 months to 3 years. These l

deviations are calculated by comparing the plant's value for each PI to the median value of the peer group for the same Pl.

2-3

2.2.3 Event Descriptions - The currently approved PI report provides brief descriptions-of each of the PI events for the most recent four quarters in Part II. Because of the importance of understanding the significance and the underlying causes of the events, these descriptions have been moved to Part I.

2.2.4 Executive Summary - The current PI report shows industry average values for each PI by quarter, including six quarter moving average trend lines. Each fourth quarter report also contains annual industry average values for each PI. These figures have been retained, although the six quarter moving averages have been replaced by linear regression trend lines and the display has been expanded from 8 quarters to 12 quarters. A typical quarterly industry average display is shown in Figure 2.

In addition, because deviations are new calculated relative to each plant's peer group instead of to the industry, comparisons among peer groups and between each peer group and industry, BWR, and PWR averages for each PI, for both operations and shutdown phase types, have been added. A typical comparison is shown in Figure 4.

2.3 Part II Part II of the proposed PI report is shown in Figure 5.

This format is different from that of the currently approved report. The data are tabulated to show the quarterly PI values for each of five p~ ases. The plant's n

operating history is also presented in the form of start and end dates and duration for each of the five phases.

In addition, the time interval used in

^

the trend and deviation calculations for operations and shutdown phase types and for FOR are provided (the EF0 indicator uses the same time interval as the operations phase type).

1 i

n h

L I

2-4 l

l P

FIGURE 1 QUARTERLY DATA r

Legenc:" " ' " ' "

PLANT NAME A

Feer Avg Trere Opercdona B

maastry Avg Trend

%,3,,,, m Gefuehng R

NDt $hDwn U$ing DD. Cy2 M s.4 j

.M R

S C

C 5'8 5/*

w-i w,-2 sa 5.->

s;-i u-a w- >

m-2 a

w>

s2-i u-3 year - Oster Year - Date-Automatic scrcrns While Critied D

safety system Actu2tions D

4 4

E F

e

$3-

,N J -

X T>2-g g

2-t si

{,-

M q,

ps f$

2,-

rsi n ra E

m I ---

~

0

~

Id 0

23 i

w- >

w-a su s->

w- >

u-a i

w- >

w-a 2su w>

w- >

u-a Year - Osorter year - Guarter D

D Sipificant Events safety system rdiures E

Ij u-j gs-3 3

_t

~

1?R l

,5 i

c.5 -

!>d t-Wa

,6 i

,a

.E sj;j g

i-9-

o c"a w-a swt,!;ws s2-3 u-2 i

w-s w-2 6%

w>

s2-.

u-a w-i Year - Quarter Year - Quarter

+

E Ecp;rvent forced Outenes/

E MX) Commercid tburs rarced Outope Rote (s)

s

_ to - -

g 3

I so 5

es 74-

} eo _

g w

d 40 40

.E os b

m

,,g 1o3 kMS e.,

u

~

M' WRMMM%

Wi-SY ah o_

c w-i w-3 sm w>

w-s n-3 i

w- >

m-3 6%

ws w- > ua Year - Oaarter Year - Oster l

E casse cas y

t Wien Eposwe

c. seman b t.ic Oper

c. Otter Per F

5 m

c.

e, s.

y

ro -

j

ER s,

s, s.

e 3L m

c O

AO"

,=

CE IY "E 00 -

S 0

9 e

M l

N

d. M2nt

e. Design

1. M se.

t

.n3 w a

i w-i w-;

swd wa s2-s n-2 s.

_7 h:

s.

s, g-Year - Oster

"M U

. Site Ave oge Ros::teon Exposse 2-5

FIGURE 2 TREliDS AND DEVIATIONS PLANT NAME

'ese'd s = $": =' si;

coce A

b PEER GROUP NAME L0w i C

D

+

Plant Deviations From Self-Trend Peer Group -

Mecian Decnned W oved Selow Abwe OPERATONS Automotic Scrcms while Csitico! -

-.o 2

-o2 Sofety System Actuations os

-os Significant Events -

j$[@j -e

-o4 Sofety System F6kses -

-ce

-0.1 Couse Codes (ALL LERs) a ae m uot= cea's *mara-o

-a i[

,, %,wy

e. over ewaarre t"* -

-uj M es I

a wwwww== ar===

3,

- g.es

. mi

.-- - m.,,

,c t ur r=== -

se

-o2 l SHUTDOWN Sofety System Actuotions -

o o

Significent Events S

02 4

-0.3 mi Sofety System Tdkres -

w, ed Couse Codes (ALL LERs)

*== vat c,.rw **-"-

]u

]u 0 Leysed Operata Pm -

L otNy Persers (nor,,

gy l g3 u

]u 1

% f i e - *.- -

gu

f. une -

o i

FORCED OUTAGES Torced Outoge Rote *

.to

-o.7 wuma: y w~.

-. c

- to

-d5 0'.0 0'S to

-1.0

-05 00 0'5 1.o

. Not Cciculated for Cheretiono! Cycle r varn a, w e retornum me e

f f

2-6

i FIGURE 3 i

indtstry overoge number of events Average no. of scrams. while critica!

Average no. of safety system actuations y. tD 1

e to

(

.e E f c8-I b

C8-

"g c.s g* cs et tt co-fkh h ES!N I

h hNf:E$i Wi 4% Wisp n

'~

db

} c2 dglEi usa M@$lihidijdEEl i

c2 -

't 1

4 ue w-a is w e 1, wa w-s u2-a i

w-s w-a aw w s swa w-s 62-2 Year - Quarter Year - Ou:rter Average no. of significant events Average no. of safety system failures a

g 2s y $es-w g u-f $'

I-dt gE 06 E

I A

4 "e to -

i:

.. +.

as

.s 1e D*-

4.A h

c.2 f

DS-fl3

.. - @g[

fhff$ffb $g

/

E'S' a w-s I

Y' I ""i""'s

' N -

i -s sw>

w-s w -J i

w-s w-J isw a F

o.o om s o-u2-a 1

w-w-J Year - QJarter Year - Quarter 1

Averah,0 commercial critical hours / -

e equipment forced outages l

Average forced outage rate (x)

K m

10 3

8*

I Eh I

, om l

e I 20 O>

R4 b* c6-b5 No n -

->yj 3p :3g w

- [-.

- g I

m-~

((1 vp *-

'l kI

[

l

$h hfb!INl w

i w-s w-J sw.

i, wy w_s 62-J i

w-s w-3 as s a o n->

w-s 62-a Year - Ow ter Year - Oayter 2

Average collective radiation crposure

_2 200 g g TC 3E

g. C":::2 hostry Averop no.

p of Events or Rotes

., g 2 -

+

)f"-

lji i9M,;

Pj L:

E m

hdJbtfy AWerDK s i '"5 s..

0 yJi

' $ 4/ipW ;A Q ! *

@M SEE COLPUTATONAL fJOTES i

w-s es is u s 6wa w-i a-3 3

Year - osyter P

2 2-7

FIGURE 4 Peer goup overoga number of scroms while critical during Operations cod Sto' tup (f4crmalized for the number of days in Operations and Startup) t Legend:

Peer Average PWR Aweroge PWR Ave'Oge hdJ$try Average D p

1. General Dectric Ite-TM1

2. General Dectric Post-TMI 2D 2D e

g o.

y o.

E

-*w.

u.

em Die

}

['

t o 7s h os.

04S

@N es.,

on

-?

as i

wi wa w i- >

w>

w-i w-J

- i w-s w-3 wv w>

w-s la-a Year - Oster Year - Oater

3. Srnall Westinghouse

4. Older Westinghouw 3-Loop i

k_

k bu g u.

tt M

to.

J,0, S$

to.

oss o.71 I R$

1 f

p,^

A W

fes.

8 c,

cs.

e 42.c.42 0 44 c.,

o.c o.o u,h N

$,Y, n,

$ k, h

co w,

, +,

w, on w

- i.,

w w,

w-,

a

w..

Year - Dater Year - Quarter S. Older Westinghouse 4 --Loop

6. Westinghouse New 3. and 4-Loop

,s ta a

e

[

I, g o-g o.

1

.a

,, o 1,

k j'

Tj Sj

" to -

oss 10 -

1 31 $i b"'l i$$

5'.

3 yM' c..,

$'* o,,c,2 33 r.

e h os h os.

-.; m : a^

7 o *'

!!Yi :%

a o.a

g>;q, ;- -

b hk"'3 N

h Ik-

$ ** $ or h N 74 s

^

'S

^

kw ;,;, Q G i m,; __

i-

e av 'y.;>;b 4

. u f

?.?? 5 & he m 44 omow t

s; mr-rs w w>

w-

.a.a 1

i

w..

w_3 wi wa w.

a.c

- i w,

w.3 w>

Year - Oster Year - Oster j

P

7. Cornbustion Engineering w/o CI'C

8. Cornbustion Engineering with CPC k

k.

t1 bu N u_

'i, e

~

~m

-* to _

to.

4 y

o 77 e y, r$,,,

~

e os h

4&

r"dim a3?SBnaim;_aWsinsm,

os o ss rwn o,

i ws we we wa w-s w-3 i ws us w i- >

w>

w-s wa Year - Quarter Year - Oater

9. Bater ck and Wilcox

10. Reactor Type and ladustry 2b 2D l

s.

I y u_

g u_

E I

i to _

oss to _

N ces I

1 DM c er T'"RRT".".>

W n;4. W 'S.!. w'l0b i

Mmm -

j.3_

j e3. y@

ca oo o,. o..n a p r

a;_

cn g, - i w-w_3 w.

o-3 i

w. s w-3 in->

wy

w.,

w_3 Year - Quarter Year - Os ter i

i 2-8 7..

FIGURE 5 PLANT NAME Year - Calender Overter Type Prese 89+4 90-1 90-2 90-3 90-4 91-1 91-2 91-3 91*'

92-1 92-2 92-3 S$1AM Operatione 0

0 0

1 0

2 C

0 Pr e-Re f ueling 0

0 0

1 0

0 0

0 Startup D

0 0

1 0

0 0

0

$$A Oper atione 1

0 0

1 0

2 0

0 Pre-Refueling 0

0 0

0 Startup D

0 0

0 Refueling C

0 0

0 Non-Refueling 0

0 0

0 SS C4*retions 0

0 1

0 0

0 1

0 Pre-Refueling 0

0 0

0 Stortup 0

0 0

0 Refueling 0

0 0

0 Non-Re f ueling 0

0 0

0 SST Operations 1

2 2

0 0

0 1

1 1

0 1

0 Pre-Refueling 0

0 0

0 5tartup D

0 0

2 0

0 0

0-0 0

Refueling 0

0 0

0 1

1 0

0 0

0 Non-Re f ueling C

0 1

0 0

1 0

0 1

2 TOR (1) 3 0

25 0

0 20 40 17 7

9 65 0

t EQ. FOR 0.46 0.00 0.00 r.00 0.00 1.34 0.00 1.0$

0.00 0.99 2.01 0.00 CRIT. ERS 2161 2160 1672 2116 0

747 1338 1898 2078 2021 497 0

RAL 471 188 32 52 500 117 33 61 177 27 49 NA CAUSE COCES:

Admin.

Oper ati one 4

3 2

1 0

0 0

1 0

2 0

0 Fre-Refcaling 0

0 0

3 0

0 0

0 Startup 0

0 0

3 0

0 0

0 Refueling 0

0 0

0 3

3 0

0 0

0 Non-Re f ueling 0

0 0

0 1

1 0

0 4

1 Lic. Oper.

Operatione 0

0 0

0 Pre-Re f ueling 0

0 0

0 Startup 0

0 0

0 Refueling C

0 0

0 Mon-Refueling 0

0 0

0 1

0 Oth. Fer.

Operations 2

0 1

3 0

1 1

2 1

2 0

0 Pre-Refueling 0

0 0

0 Startuo 0

0 0

1 0

0 0

0 Refueling 0

0 0

0 4

1 0

0 0

0 i

Non-Refueling o

0 1

0 0

0 1.

1 Maint.

Operatione 4

2 4

3 0

1 5

5 3

7 2

0 Fre-Refusiing 0

0 0

5 0

0 0

0 Startup 0

0 0

3 0

0 0

0 Refueling 0

0 0

0 12 3

0 0

Non-Re f ueling C

0 1

0 0

0 2

1 0

0 5

2 Design Operatione 1

1 2

2 0

0 1

2 0

1 3

0 Pre-Refueling 0

0 0

0 Startup C

0 0

1 0

0 0

0 RefuelsM 0

0 0

0 1

1 0

0 0

0 Non-Ref ueling 0

0 0

0 2

2 Misc.

Oper ati one 0

0 0

0 1

1 0

0 fre-Refueling 0

0 0

1 0

0 0

0 Startup D

0 0

0 Refuellns 0

0 0

0 Noc-Re f ueling 0

0 0

0 1

0 0

0 Ptose Phase type Start End Leneth Opermiten ration 10/01/89 05/21/90 233 i

Non-Refueling teown 05/ Z2 /90 06/12/90 22 Oper ation C4*retion 06/13/90 09/03/90 83 Pre-Refueling N retion 09/04/90 09/28/90 2S Refueling

%ubaoun 09/29/90 02/21/91 146 Start up Operation 02/22/91 03/18/91 25 Operation ration 03/19/91 03/29/91 11 N on-Re f ueling

, toown 03/30/91 05/05/91 37

+

Operation ration 05/06/91 07/18/91 74 Non-Refueling utdown 07/19/91 07/23/91 i

Dperation ration 0//24/91 09/02/91 41 N on-Re f ueling utdown 09/03/91 09/06/91 4

Operation ration 09/07/91 10 IS/91 39 Non-Re f ueling

utdown 10/16/91 10 20/91 5

r ati on erstion 10/21/91 04/21/92 164 N n-Refueling utdown 04/22/92 09/30/92 162 Trond Celeulatione Deviation Celeulatione On 10/01/91-04/21/92 199 days (incl.

O days e/u) 0, 03/25/90-04/21/92 $40 daye (incl.

25 days s/u)

S7D 07/03/92-09/30/92 90 doye (incl.

O days ret)

S}D 05/02/91-09/30/92 180 days (incl.

O days ret)

FOR 01/04/92-09/30/92 270 FOR 04/09/91-09/30/92 $40 r

f e

2-9

+

i

3.

PEER GROUPS

3.1 Background

In the Sec*etary's memorandum of August 10, 1989, on SECY-89-211, the Commission requested that the staff assess the validity of comparing individual plants to their NSSS average in light of the plant-to-plant variations within NSSS groups which could cause the sets of possible reportable events to differ among licensees.

In response to this request, the staff initiated a study at ORNL to detemine appropriate peer groups for comparing reported event data among licensees. This section describes the development of peer groups to be used in the presentation of PI cause code data derived from all reported event data (see references 1, 2, 3, 4, and 5).

3.2 Considerations for Peer Grouping More than 80 factors were identified which could affect the rate of-occurrence of reportable events. These factors generally fell into the four categories of design, regulatory issues, organization, and plant operating mode (startup,-

power operation, or shutdown). Upon looking into each of these areas, it was concluded that a plant's operating mode could have as much or more effect upon event reporting as all of the other categories together. This issue was addressed separately in a study at the INEL to identify those phases of plant operation in which the frequency of reportable events varies significantly (see Section 4). The three categories of design, regulatory issues, and organizational issues are discussed below.

To provide confidence in the group statistics, it was necessary to establish a minimum number of plants in a peer group.

In addition, it was important to solicit input from others in the industry, particularly those who had already constructed plant groupings for other purposes. The results of both of these efforts are also described below.

3.2.1 Design Characteristics - Design characteristics were determined to be the most important factor when constructing peer groups for PIs. They provide stable groups which can be used to compare how different licensees, given similar plant designs, perform relative to each other.. Many design issues were considered, such as Nuclear Steam Supply System (NSSS) vendor; vintage, or product line; generating capacity; and design of plant systems, particularly the protection, containment, emergency core cooling, and feedwater systems. These characteristics range from the very general- (NSSS vendor) to the quite specific (number and type of feedwater pumps). Use of the NSSS vendor alone to form peer groups would result in four groups. The Commission requested the staff to consider more than these four. The use of very specific design characteristics would result in many groups with only a small number of plants in each group. To meet the limitations on minimum group size, only a few of the more general design features could be

)

incorporated, such as NSSS vendor and vintage.

2-10

i 3.2.2 Regulatory Issues - Regulatory requirements and programs include the type of technical specifications (standard or custom), licensing date, required programs such as environmental qualification, mandated or voluntary (in response to NRC concerns) review programs, and other NRC programs such as Headquarters and Regional inspections. Many regulatory issues tend to change over time, which limits their usefulness in develcping peer groups. However, many of them, such as backfits arising out of the lessons learned from the event at Three Mile Island (THI), are a function of NSSS vendor or other aspects of plant design, and therefore can be accommodated in peer groups based upon desigr. considerations.

In addition, certain regulatory issues that are recognized to play an important role in the frequency of occurrence of reportable events, such as the type of technical specifications, are closely related to a plant's licensing date relative to the event at TMI.

3.2.3 Organizational Issues - Organizational issues, such as policies, programs, and training, are perhaps the most important factors affecting the frequency of reportable events. However, every plant has its own set of organizational issues, most of which are subject to change over time. More importantly, it is the effect of these issues upon the frequency of reportable events that the Pls attempt to measure; consequently, it is not appropriate to include them when developing peer groups.

3.2.4 Peer Group Size - The effect of peer group size was studied to determine the minimum size acceptable to provide statistically valid data for performing PI calculations. Those calculations include certain parameters for each peer group.

It is important that these parameters accurately represent group performance. A method often used to quantify the accuracy of an estimated parameter is to place a confidence interval on the parameter. The more narrow the confidence interval, the more precisely the estimate represents the true value of the parameter. Confidence intervals, as a function of the number of plants in a peer group, were evaluated to determine the minimum number of plants in a group.

It was learned that confidence intervals grow quite large with fewer than six plants in a group. Therefore, peer groups should have six or more plants. This was an important consideration when constructing peer groups, since the smallest number of currently operating plants designed by one NSSS vendor (Babcock & Wilcox) is seven.

3.2.5 Input from Other Organizations -

Input from others within the industry was solicited to minimize uncertainty and potential dissimilarities in the peer groups. This input was useful in defining the criteria to be used to establish peer groups, as well as in understanding the difficulties to be expected when comparing plants because of the diversity in design implementation. The NSSS owners' groups stressed the effect of age and type of technical specifications upon the frequency of reportable events.

3.3 Peer Groups Selected

+

A set of nine peer groups was constructed based upon NSSS vendor and vintage, modified slightly in consideration of several unique plants and certain regulatory issues (see Table 1). This grouping incorporates the important 2-11 i

effects of both plant design and regulatory concerns: NSSS vendor, generating capacity, product line, age, type of technical specifications THI backfit programs, and licensing date. While this set of peer groups was developed for the specific purpose of comparing cause codes among licensees, they will be used for all PIs. This set, tseed upon these design and regulatory issues, are appropriate for comparing the c erall performance of licensees operating similar plants in a similar regulatory environment.

E 2-12

t l

TABLE 1-PEER GROUPS Combustion Engineering Plants Babcock and Wilcox Plants w/o Core Protection Calculators Arkansas 1.-

i Calvert Cliffs 1 and 2 Crystal River 3 Fort Calhoun-Davis-Besse Maine Yankee.'

Oconee 1,.2, and 3' l

~

Millstone 2 Three Mile Island l' j

Palisades St.Lucie 1 and 2 Westinghouse Small Plants l

Ginna a

Combustion Engineering Plants Haddam Neck j

w/ Core Protection Calculators Kewaunee.

Arkansas 2 Point Beach I and 2~

Palo Verde 1, 2, and 3 Prairie Island I and;2 I

San Onofre 2 and 3 San Onofre 1-Waterford 3 Yankee Rowe-General Electric Plants - BWR/1, Westinghouse.0lder Three Loop Plants 1

BWR/2, BWR/3, and Older BWR/4 Beaver Valley 1 l

_ Big Rock Point Farley I and~ 2 i

' ' Browns Ferry 1, 2, and 3 -

North Anna 1 and 2' Brunswick I and 2 Robinson 2 Cooper-Station Surry I and-2 Dresden 2.and 3 Turkey Point 3 and 4 Duane Arnold i

FitzPatrick Westinghouse Older Four Loop Plants.

Hatch I and 2 D.C.~ Cook I and 2 Millstone 1 Indian Point 2-

-i Monticello Indian Point 3 Nine Mile Point 1 Salem 1 and 2 Oyster Creek Trojan Peach Bottom 2 and 3 Zion.1 and 2 Pilgrim Quad Cities 1 and 2 Westinghouse Newer Plants

-l Vermont Yankee Beaver Valley 2 1

Braidwood I and 2 General _ Electric Plants - BWR/5, Byron 1 and 2 l

BWR/6, and Newer BWR/4 Callaway -

Clinton Catawba 1:and 2 i

Fermi 2 Comanche Peak'I i

Grand Gulf 2 Diablo Canyon 1 and~ 2 i

Hope Creek Harris

-l LaSalle. I and 2 McGuire I and 2 j

Limerick'I and 2 Millstone 3 1

Nine Mile Point 2.

Seabrook Perry-Sequoyah 1,and 2' River Bend South Texas 1 and 2 j

Susquehanna 1 and 2 Summer

l Washington Nuclear 2 Vogtle 1 and 2 Wolf Creek i

i 2-13 1

~!l i

4.

OPERATING CYCLE PHASES

4.1 Background

During the Peer Group study conducted by the ORNL, it was noted that cause code data are cyclic with a period approximating that of the refueling interval (see reference 6). Upon further investigation of this phenomenon, it was concluded that differences in operating phase (startup, power operations, refueling, etc.) could have as much or more effect upon the frequency of occurrence of reportable events as differences due to peer groups.

This issue was addressed separately in a -tudy at the INEL to identify those phases of plant operation which affect the frequency of occurrence of reportable events (see references 1, 7, 8, and 9).

4.2 Considerations for Operating Cycle Phases The aperating cycle profile was determined to be reasonably well characterized by five phases. These are described below.

4.2.1 Startup - The startup phase consists of the first 25 days of operation after leaving cold shutdown following a refueling outage. Any outages over three days in length which occur during this period will stop the 25 day clock and will not be counted in the startup phase (see definition of non-refueling outages below).

Following completion of such outages, the 25 day clock will continue the count from where it stopped. The startup phase comprises those operational and maintenance activities unique to return to power operations after refueling. This includes physics testing, preconditioning of fuel, and testing of equipment after maintenance and/or an extended shutdown. The time period for performance of these activities was judged to be between 15 and 35 days. As a typical value, 25 days was selected as the duration of the startup period.

4.2.2 Power Operations - The power operations phase consists of the time between the completion of the startup phase and the beginning of the pre-refueling phase, excluding non-refueling outages greater than three days long (see definition of non-refueling outages below).

4.2.3 Pre-Refueling - The pre-refueling phase consists of the last 25 days of operation prior to entering cold shutdown for a refueling outage. Any outages over three days in length which occur during this period will stop the 25 day clock and will not be counted in the pre-refueling phase (see definition of non-refueling outages below). Following completion of such outages, the 25 day clock will continue the count from where it stopped. The pre-refueling phase consists of those activities undertaken during the time i

immediately prior to, and in preparation for, a refueling outage. The time period for performance of these activities was judged to be approximately three weeks to one month. As a typical value, and to be consistent with the definition of the startup phase, 25 days was selected as the duration of the pre-refueling period.

2-14

k 4.2.4 Non-Refueling Outages - The non-refueling outage phase consists of the total number of days spent in non-refueling outages of more than three consecutive days duration during the operating cycle. Non-refueling outages of short duration were considered to involve plant activities associated with a quick return to power, activities that would not be very different from those of the power operations phase. At some point in a non-refueling outage, plant activities begin to involve more extensive maintenance efforts. This is the point at which a non-refueling outage begins to look more like a refueling outage. This point was judged to be about four days.

P 4.2.5 Refueling Outage - The refueling outage phase consists of the time between achieving cold shutdown for a refueling outage and. entering the startup phase.

4.3 Selection of Operating Cycle Phases for PIs The evaluations used to determine the definitions of each of the five phases -

suggested that there are differences in the frequency of reportable events between the phases. To confirm this, event rates for each of the PIs and cause codes as a function of phase were examined. The data show that the i

frequency of reportable events differs among the five phases, but that, for many indicators, the operations, startup, and pre-refueling phases are similar, and the non-refueling outage and refueling outage phases are similar.

The proposed five phases provide the capability to monitor and analyze various aspects of plant performance, such as the length of refueling outages, the' number of days of non-refueling outages in a cycle, performance during startups, problems during pre-refueling, etc.. These subjects can be evaluated by examining and trending the event data by phase, which is provided in Part 11 of the report.

But for the calculation of plant trends and deviations, it is neither necessary nor practical to treat the five phases. individually.

Therefore, the five phases were combined-into two phase types - operations, consisting of the startup, power operations, and pre-refueling phases; and shutdown, consisting of the refueling and non-refueling outage phases.

(Because startup phase events are of particular interest, however, they are identified separately in the quarterly data display, yet included in the operations phase type for calculation of trends and deviations.) To accomplish this construction of operations and shutdown phase. types, each plant's operating history must be partitioned as shown in Figure 6.

All refueling outages, and non-refueling outages more than three consecutive days in length, are removed from the time sequence and, retaining their chronological order, are combined into the shutdown phase type. The remaining days, consisting of startups, power operations, and pre-refuelings, are brought together in chronological order to form the operations phase type.

These two phase types are then treated as continuums for the purpose of calculating trends and deviations.

Each of these continuums, while arranged chronologically, does not necessarily consist of consecutive calendar days, depending upon the plant's operating history. This method preserves the important aspects of the operating cycle effects and allows a ready comparison of plant performance during shutdowns with that of power operations.

2-15 i

FIGURE 6 t

N

."'2'"

i Operations Phase Type (con w dtrwn 9m M opweno NeerA 1~A A

A A

i b

-M

_y g,= - +

95ar.)

V Y

V j jf gg g[

Hypothetical (s p

sequence of M

x plant

$WN}

(

[

gp((f Ek

$lEA$[.j esu A

A l

4",,84 >

I m

W

---swee rua c5 *

r.

W 4

Y Y

Y

{h %

Shutdown Phase Type neeser i

4

% paw I

"W k

b 5

i 2-16 l

l l

1

i B

5.

CALCULATIONAL METHODS 5.1 Background-j To incorporate the peer groups and the operating cycle effects.into the PI l

Program required development of different calculational and display. techniques from those currently.used.

Improved techniques were established which employ i

standard statistical methods. They are collectively termed the Operating Cycle Method (OCM) and are described in.this section.

l 5.2 Calculational Methods l

The operations and shutdown phase type continuums described in Section 4.3, while arranged chronologically, do not necessarily consist of consecutive calendar days. All trends and deviations are measured over specified intervals of these histories. Nominal intervals were chosen to provide both short term and long. term perspectives on plant performance (see Table 2).

Plant self-trends are calculated to provide short term perspectives, while deviations from the peer group median and quarterly data trends are calculated' to provide long term perspectives. Maximum calendar time. limits for each l

nominal interval were established to ensure that older data which are less-i representative of current plant performance are excluded from the calculations. Minimum intervals were established to ensure that sufficient experience has occurred for meaningful results.

If.the minimum interval'is not met, the trend or deviation will be N/A.

l NOMINAL MAXINUM MINIMUM-

[

INTERVAL TIME INTERVAL PERSPECTIVE-QUARTERLY PLANT-LONG TERM 3 YEARS 3 YEARS 3 YEARS DATA PEER GROUP-LONG TERM 3 YEARS 3 YEARS 3 YEARS

[

TRENDS INDUSTRY LONG TERM 3 YEARS 3 YEARS 3 YEARS OPERATIONS SHORT TERM 270 DAYS I YEAR 90 DAYS PLANT FOR & EF0 SELF-TRENDS SHUTDOWNS SHORT TERM 90 DAYS.

18 MONTHS

30 DAYS OPERATIONS LONG TERM 540 DAYS 3 YEARS 90 DAYS.

DEVIATIONS FOR & EF0

]

SHUTDOWNS LONG TERM 180 DAYS 3 YEARS 30 DAYS TABLE 2 2-17 i

t 5.2.1 Trend Calculations - For each trend calculation, ~ the appropriate interval is divided into segments of equal length, PI values for each segment are determined, and a standard linear regression is performed on the data points. The linear regression technique is used because-(1) it is widely accepted and understood, (2) it is a straightforward technique with minimal manipulation of the data, and (3) it is unambiguous and easily interpreted.

Each quarterly data trend line is the straight line best fit (least squares estimate) to the data points by the linear regression.

5.2.1.1 Quarterly Data Trends - There are three trend lines displayed on the quarterly data page - the plant trend, the plant's peer group trend, and the industry trend.

In each case, the time interval for calculation of the trend is three years, divided into twelve segments of one calendar quarter each regardless of operational phase type. This was done to provide a chronological comparison of performance among the plant, the peer group, and the industry.

It is not possible to perform such a comparison by operating-phase, since all plants are not in the same phase at the same time. The three year interval provides a long term perspective on performance. Over that interval (about two operating cycles for most plants), the effects of operating phase would tend to cancel, so that the comparison is valid and useful. The displayed plant trend is the straight line best fit to the plant's twelve data points over the three year period. The peer group and industry trends use the average value of the respective group performance for each indicator each quarter.

Each displayed trend is the straight line best fit to the applicable twelve data points over the three year period.

5.2.1.2 Plant self-Trends -

For all plant self-trend calculations, the nominal interval is divided into nine segments of equal length, which provides a reasonable number of data points for trending. With intervals of less than nominal length in the maximum time period, the number of days in a segment remains constant and the number of segments decreases to the largest possible integral value. The minimum number of segments is three, which corresponds to the minimum interval for each trend calculation.

Each displayed plant self-trend is the slope of the straight line obtained from the linear regression divided by the scaling factors described below.

The nominal interval for plant self-trend calculations for the operations phase is 270 days, divided into 9 segments of 30 days each. This interval was chosen to provide a short term measure of plant performance trends since the last Senior Management Meeting (SMM), including about one quarter of overlap with the data presented at the last SMM to provide a frame of reference. The maximum time limit is one calendar year, and the minimum interval is 90 days.

The nominal interval for plant self-trends in the shutchwn phase is 90 days, divided into 9 segments of 10 days each. This interval was chosen to provide a short term measure of plant performance trends over about the last year.

The maximum time limit is 18 months, and the minimum interval is 30 days.

l The nominal interval for the plant forced outage rate (FOR) self-trend is 270 days, excluding scheduled outage time, divided into 9 segments of 30 days each. This interval was chosen because (1) it provides a short term measure of the FOR trend since the last Senior Management Meeting (SMM), including 2-18

i i

about one quarter of overlap with the data presented at the last SMM as a I

frame of reference; and (2) since scheduled outage hours are net used in the calculation of FOR, including them in the interval could severely restrict the generator on-line hours used in the calculation, which would produce artificially high rates. The maximum time limit is one calendar year, and the minimum interval is 90 days.

The nominal interval for the plant equipment forced outages per 1000 commercial critical hours (EF0) self-trends is 270 days of operations phase time (shutdown phase time is excluded), divided into 9 segments of 30 days each. This interval was chosen because (1) it provides a short term indication of the EF0 trend since the last Senior Management Meeting (SMM),

including about one quarter of overlap with the data presented at the last SMM as a frame of reference; and (2) including shutdown phase time in the interval could severely restrict the critical hours used in the calculation, which i

would result in artificially high rates. Since equipment forced outages can occur during any phase (including shutdown), all such outages occurring between the start date and the end date of each segment will be counted and included in the calculation of EF0.

The maximum time limit is one calendar year, and the minimum interval is 90 days.

5.2.2 Deviation Calculations - All deviations are measured over a

)

specified interval, ideally the nominal interval.

Each nominal interval is twice as long as its corresponding plant self-trend nominal interval to provide a longer term indication of plant performance relative to its peers.

i For a plant whose history does not contain the nominal interval within the maximum calendar time, calculations will be performed using the available interval within the maximum calendar time as long as it meets or exceeds a j

minimum value.

For each deviation calculation, the performance of the plant and each of its peer group plants are summed over the selected interval. The j

plant value is subtracted from the median value of the peer group plants. The median value is used because (1) it is more representative of the typical performance of a group with a highly skewed distribution (such as is the case with the PIs), and (2) it is not as sensitive to extreme outliers as is the group average (mean). The displayed plant deviations are the differences calculated for each indicator, divided by the scaling factors described below.

The nominal interval for deviation calculations in the operations phase is 540 days. The maximum time limit is three calendar years, and th.e minimum interval is 90 days.

The nominal interval for deviation calculations in the shutdown phase is 180 days. The maximum time limit is three calendar years, and the minimum interval is 30 days.

)

The nominal interval for calculation of the FOR deviation is 540 days, i

excluding scheduled outage time. The maximum time limit is three calendar years, and the minimum interval is 90 days.

]

The nominal interval for calculation of the EF0 deviation is 540 days of operations phase time (shutdown phase time is excluded). The maximum time i

limit is three calendar years, and the minimum interval is 90 days.

2-19 1

5.2.3 Scaling Factors - Each of the plant self-trends and deviations described above in Sections 4.2.2.2 and 4.2.3 have been divided by scaling factors. The purpose of the scaling factors is to adjust all calculations to the same scale so that (1) each displayed trend and deviation has comparable value relative to every other trend and deviation, (2) displays won't be compressed because of one or more extreme outliers, and (3) only a few trends and deviations will exceed full scale.

5.2.4 Significance Tests - Tests of the statistical significance of each plant self-trend and deviation assist in distinguishing between those patterns that are likely to occur randomly and those that are unlikely to occur randomly. The tests determine the probability that an observed pattern is a random occurrence rather than a real performance trend or deviation.

l The significance tests use Monte Carlo simulation techniques to determine the

(

probability that an observed pattern is random. Simulation is a common l

l technique in statistical analyses to identify outliers. This method is more i

appropriate than the standard t test for significance since the event rates are not normally distributed. Trend and deviation patterns with a probability of being random of greater than 0.200 are of low statistical significance and are shaded white. Those with a probability of greater than 0.025 but less than or equal to 0.200 are of medium significance and are shaded gray.

Patterns with a probability of less than or equal to 0.025 of being random are j

l of high statistical significance and are shaded black. These levels were selected to (1) provide visual indication of those trends and deviations that are most likely to be indicative of real performance, and (2) capture enough plants to be reasonably certain that most real performance trends have been identified.

i 2-20

6.

REFERENCES i

1.

H. M. Stromberg, et. al., " Peer Groups and Operational Cycle

{

Enhancements to the Performance Indicator Report," EGG-P0A-10553, December 1, 1992.

2.

ORNL Letter Report I for Development of Plant Peer Groups, January 19, 1990.

3.

ORNL Letter Report 2, Development of Plant Peer Groups, February 16, 1990.

4.

ORNL/NOAC-261, " Development of Commercial Nuclear Plant Peer i

Groups for Presentation of Performance Indicator Data," May 1990.

l 5.

ORNL/NOAC-265, " Development and Evaluation of Proposed.

Modifications to the Nuclear Regulatory Commission's Performhnce Indicator Program," July 1991.

6.

ORNL/NOAC-254, " Development of an Analytical Tool for Investigating Plant Peer Groups," November 3, 1989.

7.

H. M. Stromberg, et. al., " Operational Cycle Effects on the Performance Indicators," EGG-EAST-9107, June 1990.

8.

H. M. Stromberg, et. al., " Operational Cycle Adjustments to the Performance Indicators," EGG-EAST-9445, February 1991.

9.

H. M. Stromberg, et. al., "Further Development of Operational Cycle Trend and Deviation Measures for the Performance Indicators, EG&G Idaho, Inc., January 1992.

r r

I 6

t h

2-21 j

i

ML20034D515

Contents

Text

Purpose:

Contact:

2.1 Background

3.1 Background

4.1 Background

Navigation menu

ML20034D515

Text

Purpose:

Contact:

2.1 Background

3.1 Background

4.1 Background

Navigation menu

Search