ML20050B528

Forwards Rept on Time to Loss of Offsite Power Following Turbine Trip for Sys 80 Plants,Per Util 820212 Submittal to Closeout SER Confirmatory Items Re Cessar FSAR Chapter 15 Analyses

ML20050B528
Person / Time
Site:	05000470
Issue date:	03/31/1982
From:	Scherer A ABB COMBUSTION ENGINEERING NUCLEAR FUEL (FORMERLY
To:	Eisenhut D Office of Nuclear Reactor Regulation
References
LD-82-040, LD-82-40, NUDOCS 8204060004
Download: ML20050B528 (17)

Text

• e

. C-E Power Systems , Tel 203/688-1911 8

Cornbustion Engineering inc Telex 99297 1000 Prospect Hill Road Windsor. Connecticut 06095 POWER SYSTEMS Docket No.: STN-50-470F March 31, 1982 LD-82-040 mi9 es 9 Mr. Darrell G. Eisenhut, Director le % 'gs Division of Licensing 9- 4pp 70'Jify ,

U. S. Nuclear Regulatory Commission  ; t: m . O _

Washington D.C. 20555 q8233 12

'lQ

Subject:

Turbine Trip Time Delay

' ~

/,h 7w A\f

Reference:

A. E. Scherer to D. G. Eisenhut, LD-82-016, dated FebruaryiclQ9,82

Dear Mr. Eisenhut:

Combustion Engineering submitted information in the reference to close out several Safety Evaluation Report confirmatory items regarding CESSAR FSAR Chapter 15 analyses. Subsequently, the staff informally requested additional information to justify the time delay between turbine trip and subsequent loss of of fsite power which was credited in the CESSAR FSAR Chapter 15 analyses.

Enclosed is a report which demonstrates that the time to loss of offsite power following turbine trip is at least 3.1 seconds for System 80 plants. The enclosure is transmitted for the Staff review of CESSAR FSAR.

If I can De of any additional assistance in this matter, please contact me or Mr. G. A. Davis of my staff at (203)688-1911, Extension 2803.

Very truly yours, COMBUSTION ENGINEERING, INC.

, 2CH L A. E. Scherer 'u Director Nuclear Licensing AES/cw 00 3 Enclosure cc: C. I . Grimes r204060004 820331

{DRADOCK 0500047o PDR

TIME TO LOSS OF 0FFSITE POWER FOLLOWING TURBINE TRIP FOR SYSTEM 80 PLANTS ABSTRACT Infrequently, the loss of a power generating unit (e.g. the loss of a nuclear power plant due to turbine trip) on an electrical power grid will result in loss of offsite power to tnat unit. Given that loss of offsite power does occur, this report provides an analysis of the actual time delay between turbine trip and loss of offsite power for System 80 plants. The results are based on grid frequency response versus time plots for various generation deficiencies and indicate a time delay of at least 3.1 sec.

3 l

l _

TABLE OF CONTENTS SECTION TITLE PAGE NO,.

Abstract 3 Table of Contents 4 List of Figures 5 List of Tables 5 1.0 PURPOSE 6 2.0 SCOPE 6

3.0 REFERENCES

6

4.0 BACKGROUND

6 5.0 DISCUSSION 8

6.0 RESULTS AND CONCLUSIONS 12 APPENDICES A List of References A-1 4 -

LIST OF FIGURES Figure Titl e Page No.

1 Florida Frequency Response witn Load 10 Shedding 2 Florida Power and Light System Frequency 11 Following Loss of Turkey Point Unit Four 3 Island Frequency Response Without Load 14 Shedding 4 Time to Loss of Offsite Power vs. Generation 15 Loss of Percent

i I

LIST OF TABLES Table Titl e Page No.

1 Plant Regional Information 13 5

1.0 PURPOSE The purpose of this report is to provide a conservative evaluation of the time to loss of offsite power following a turbine trip for System 80 pl ants. The results udll be used in CESSAR Chapter 15 Accident Analyses.

It should be noted that this method could also be used to evaluate time to loss of offsite power for other C-E nuclear power plants provided the necessary plant / grid parameters are obtained and applied to the model.

2.0 SCOPE Under certain conditions the loss of a power generating unit (e.g. the loss of a nuclear power plant due to turbine trip) on an electrical power grid will generate frequency deviations such that the resulting electrical system instability will cause loss of offsite power to that unit (1).

Generation deficiencies caused by the loss of a System 80 plant are applied to a model of the Florida electrical system frequency response to determine time to loss of offsite power.

3.0 REFERENCES

A list of References is provided in Appendix A.

4.0 BACKGROUND

All U.S. bulk power grids normally operate at a frequency of 60 Hertz.

When a sudden load mismatch occurs (i.e. loss of generating capability or loss of load) the power imbalance results in frequency oscillations or transients. The degree of instability experienced by the grid is 4

characterized by the " electrical island" (Reference 3) that forms and by (1) For this report, loss of offsite power to a plant will refer to a grid instability in terms of frequency degradation. Failure of a System 80 plant to fast transfer to offsite power or a pre-existing loss of offsite power will not be addressed in this report.

6

the magnitude and rate of frequency change. Tne frequency change is related to tne magnitude of the load mismatch and the physical parameters of tne island, i.e. tne available rotating inertia, spinning reserve (l),

and the ability of the grid to control frequency oscillations through load damping. (A " stiff" grid will have many interconnections to other systems wnich nelp the surrounding network to absorb or damp out system disturbances.)

The initial response of a grid to a generation loss is primarily dependent on available rotating inertia or the inertia resulting from all rotating masses on the interconnected network. The frequency of a large system with many connections to other systems will respond at a slower rate to a specific generation deficiency than a smaller system with few ties to other networks. During a frequency transient the rotating inertia of a system acts as a stabilizing force to return the grid frequency to normal. Under-frequency load shedding is also utilized to restore the balance between load and powr generation in order to return grid frequency to 60 Hz.

Infrequently a power generation deficiency will occur at a time when system inertia and spinning reserve are low enough to initiate and perpetuate grid frequency decline. If incoming transmission lines from otner major electrical systems cannot makeup the generation loss, an electrical island will form. Usually load shedding will arrest the decline and return the frequency to normal . However, when load shed does not occur early enough in the transient (e.g., automatic load shed fails or is manually prevented), or the load shed does not have sufficient. magnitude, or the rate of frequency decline is very rapid (usually associated with large generation deficiencies), then loss of offsite power can occur to a plant as a direct result of that plant triping off line, e.g. loss of offsite power to a System 80 nuclear power plant following a turbine trip. This (1) Spinning reserve is generating capacity operated at less than full capability so as to be immediately available to support additional load.

7

type of event is assumed to have a probability of 10-3 (Reference 1).

Therefore, given that a plant is assumed to lose offsite power (in accordance with Standard Review Plan Section 15 of NUREG-75/087), the question is how soon will it lose offsite power following a turbine trip.

S.0 DISCUSSION CESSAR Chapter 15 addresses loss of offsite power to a System 80 nuclear power plant following turbine trip. Currently, the loss of offsite powr is assumed to occur with a probability of 10-3 at the instant the turbine trips. This assumption has led to difficulty in demonstrating compliance with radiological rele:se acceptance criteria for some events. C-E believes that a more realistic but still conservative assumption would be to model a time delay betwen turbine trip and loss of offsite power.(1)

Since tne rate of frequency decline following a given generation deficiency is dependent on the characteristics of the island created, it was decided to apply 1979 frequency response characteristics of the Florida network and superimpose the effects of a generation deficiency caused by tie loss of a System 80 plant (1300 MWe) onto curves of Florida grid frequency response versus time (Reference 5). Past operating data (Reference 2) has shown the Florida Peninsula to nave more power system disturbances than any other area of the United States. This is largely due to the fact that Florida is geographically unable to tie into any major powe system other than Georgia and Alabama. The transmission lines to those states cannot make up a generation loss greater than 400 MWe so the Florida Peninsula becomes an electrical island for a generation deficiency caused by a loss of a 1300 t MWe System 80 unit. This situation is unique to Florida and tends to allow l

l greater frequency oscillations or a more severe rate of grid frequency decline to occur relative to other electrical systems in the U.S. network.

(1) Tne time delay evaluated in this report is not a fixed result but is intended as a conservative lowr bound for System 80 plants. Depending on the assumptions used and plant specific information, time delays for other C-E nuclear power plants could also be obtained.

3

Figure 1 depicts the response of the Florida grid to various generation losses. This graph was obtained from Reference 5 and assumes island conditions (Florida is isolated from the rest of the eastern interconnected transmission grid system) and no spinning reserve. Howver, credit is taken for load shedding starting at 59.7 Hz. It can be noted that the effect of load shedding is not always immediate and the benefits decrease as the percentage generation loss increases (l). Figure 2 shows an actual grid response to loss of the Turkey Point 4 nuclear gsnerating station (Reference 5). Turkey Point 4 represented about 6% of the Florida island generating capability during the transient.

Most plants are instantaneously disconnected from the grid betwen 56 and 57.6 Hertz to prevent underfrequency damage to plant components. The instantaneous underfrequency setpoint for a specific plant is chosen at the utility's discretion and is based on their individual protection policy for the turbine-generator. When a nuclear power plant experiences a turbine trip due to internal events (e.g. reactor trip) the plant auxiliary equipment switches to offsite power and the unit becomes load on the grid rather than a power source. Loss of offsite power to the unit can then be equated to the time point at which grid frequency has decayed enougn to cause other plants to isolate their turbine-generators from the grid, i.e.

additional loss of power generation to the grid. Because underfrequency setpoints usually vary from plant to plant, loss of offsite power can be conservatively assumed to occur when the plant with the highest underfrequency setpoint separates from the grid. The higher the setpoint, tne sooner plant separation will result and the sooner loss of offsite power to the disabled unit will occur. 57.6 Hertz was the highest instantaneous underfrequency setpoint that could be determined for any l U.S. grid and will be used to define the loss of offsite powr frequency for all grids connected to System 80 plants.

(1) The accuracy of the frequency versus time curves used in this evaluation is supported by plots in Reference 4 obtained from the Northeast Power Coordinating Council. The grid model in Reference 4 could have been used in this analysis and would have provided similar conclusions. Hoever, the i Florida grid model was felt to be more conservative.

9 i

Florida Frequency Response with Load Shedding

/ -

60.0; '

g j LOAD SHED _ Y 'g'

[g / 2.5 % ~

,/

IllITIATED Lg -

,th Jg b\

N g.10.0 / % 5.0 % ' ' --- '

59.0 s ,

\

i '/

LOAD SHED g '~'h Generation Loss in %

COMPLETE 1, / 20.0% , ,

\\

/

, 58.0 \ l '

/~

'\' \i.

/-

40.0 % -

f

' /*

I 57.0 -

/

> '/

E 60.0 %

System Characteristics (1979) 56.0 -

System H (Inertia)4.5%

System D (Damping) 1.0%

Spinning Reserve 0.0%

Load Shedding:

55.0 -

Freq. Amount 59.7 2.9 %

59.2 5.5 %

53.1 5.6%

54.0 - 59.0 5.5 %

58.8 8.2 %

l 58.7 8.2 %

58.6 12.8 %

58.5 12.9 %

53.0 -

52.0 t i I l- 1 0 2 4 6 8 10 Time in Seconds FIGUP.E 1

,e

FIGURE 2 -

l . ,

"" Rorida Power & Light.

System Frequency Following o Loss of Turkey Point Unit Four 60.0 Loss of Turkey Point a March 17,1979 2:11 AM (t= 0 sec) -

(6% Generation Loss) Frorn Ranch Oscillograph 59.9 -

59.8 ---

1-l .

, initiation of Stop 0 59.7 -- -

9 -A pproximate Seperation of Peninsular Florida 3

U U 59.6 L .-- .-

E

?r

- ~ ~

~

+ Slope .56HzIsec 59.4 ---

59.3 -

Slope .2Hz/se . -j Underfrequency Load Shedding Step i

+ ' "

59.2 - - - - - - - - - ' - - - - - - -

\V '

59,3 - -- . . . .

..-..-l-- ... -_

i Initiation of Underfrequency Load Shedding I

i M.0 1 2 3 4 5 6 7 8 5 . 10 Timoin Seconds

Plant regional information obtained from References 6 through'9 and the National Electric Reliability Council (Reference 10) is presented in Table 1. The percent generating capacity of each plant can be readily calculated (per the respective islands) by dividing Plant Generating Capacity by the Installed Grid Generating Capability. This calculation is performed for both the current Grid Generating Capabilities and the projected Grid Generating Capabilities at plant start-up for comparative purposes. In each case the loss of a System 80 plant was felt to be conservatively modelled by assuming a 10% generation loss to the grid upon turbine trip.

6.0 RESULTS AND CONCLUSIONS The island frequency versus time curves shown in Figure 3 are duplicates of the curves in Figure 1. The horizontal line drawn at 57.6 Hertz represents the underfrequency value at whicn System 80 plants will be assumed to lose offsite power. The next conservatism assumes that grid frequency decline is not arrested by load shedding. By extending the slope of each curve tangent to the point prior to load shed initiation (59.7 Hz.), the model will represent grid frequency response with no load shedding. Since loss of offsite power is modelled to occur at 57.6 Hertz, the time to loss of offsite power for a given generation loss can be obtained by noting the timepoint at which the frequency decays to 57.6 Hertz. Figure 4 is a plot of time to loss of offsite power versus % generation loss for an underfrequency setpoint of 57.6 Hertz. For System 80 plants the generation loss is modelled as 10%, therefore, the time to loss of offsite power is 3.1 sec. The following list of assumptions and conservatisms should be noted:

o The response characteristics of the Florida island are used as a generic model for all System 80 grids.

12

TABLE 1 PLANT REGIONAL INFORMATION Generating Unit PVNGS-1 Cherokee /Perkins WPPSS-3 Yellow Creek 1 Plant Generating Capacity (MWe) 1238 1280 1242 1285 Start-up Date 1983 1988 1985 1985 ,

Grid System or Island I SCE-AZ-NM-WT VACAR NWPP TVA Current Installed Grid 29941 37233 35813 27132 Generating Capability (MWe)

(Sumer 1981)

C Installed Grid Generating 33299 47120 42273 31459 Capability at Start-up (MWe)

Plant Generating Capacity 4.1% 3.4% 3.5% 4.7%

(% of Current Grid)

Plant Generating Capacity 3.7% 2.7% 2.9% 4.1%

(% of Grid at Start-Up)

% Generation Loss Assumed 10% 10% 10% 10%

1 SCE-AZ-NM-kT - Southern California Edison- Arizona-New Mexico- NWPP - Northwest Power rool Area West Texas Extra High Voltage Grid TVA - Tennessee Valley Subregion VACAR - Virginia, Carolinas Subregion NOTE: During a regional electrical system disturbance the grid tends to physically separate into these "sub-regional" islands. (Reference 10)

._ __ m a JSLAND FREQUENCY RESPONSE WITHOUT LOAD SHEDDING 60.0

/ ./

7 l ,

Load Shed --e , p 2.5% ,

/

'[/~

Initiated N ,, f'

\  %, / 5.0% ~ ' ' '

59.0 J F 0%

\s/ Generation Loss in % -

l

.0 % /

\} . *--

Grid Frequency Decline /'/

58 0 i L,

is not arrested y' Underfrecuency setnoint 57.6 Hertz = f 40.C  % ,,.- for instantaneous plant tripe

' /

W 57.0 -

/

5 \ /

@ 60.0%

o S System Characteristics o 56.0 -

E System H 4.5% '

System D 1.0%

Spinning Reserve 0.0%

55.0 -

54.0 -

53.0 -

e For a 10% generation loss, grid frequency declines to 57.6 Hz. at t = 3.1 seconds.

52.0 . I i fI . l . 1 0 2 4 6 8 10 Timo in Seconds FIGURE 3

_ 14 _ _ _ _ _

+-

--i

. - - - - -.-- . --t 1

a .-

(

1

, 1 ,_..

e4 I I 1

w s_ - 4 _ .

C  % .2

. - .5 :z u.

m . ,. --

'O # - ay- e

. f N A__r -4

..w C an u:

,,s

.m

c::

c, 1 ., 2 --

- ~O m_ .

. _6

.m.m ~

w : --: ; -  :

a-a >1o w

-.; .m Tva - .

-- e aa o

1

-rx ~ s ,

w--,g- ,_,

n m m.w -- _

m -. 1

~

M:w Z -EE:

e =;_-- 9 +---

g v . r. w- _

m , m d a.-

nwi.12 A

rG E

--. lu c;; .

,u,

.1 l e:

y; -o.

w- --

(

, ,m u-_

I t_=: .

e.

- ih w 1 1 w.

D*

j -

=

g -, F " - ~ ^

r N 2 a IOu I'+-

Ys

., s,.

7tP

,,/

~

o d _-

Fw - r-

.. g l$=

LAJ

/ '

X

. .- ... a -_1..-...---.-i----s--  :-a ---

c' _x ~ ~~ M ~ $' g-d j b d

4 --

2 e se a _ _

u m wo. .~, .

7

( a -

._._,--...-+__._.1--_-i_._5 2

1

- J2 >.j--t21UQ.m ki bss U::1U t:htLi:i!L =:n'

-a M y . _

( -

~*~  ;

I i  ;

t 4 d.-.-ee

o No credit is taken for load shedding programs designed to arrest frequency decline. (i.e. initial rate of frequency decline is '

assumed.

o No credit is taken for spinning reserve.

o The Florida model assumes " island" conditions (No support from neignboring grid systems, i.e. lower rotating inertia).

o Loss of offsite power occurs instantly at a grid frequency of 57.6 Hertz for all System 80 plants.

o The loss of a System 80 plant is modelled as a 10% generation loss. [The percentages given in Table 1 are based on Installed Grid Generating Capability. During off peak hours loads can drop to 50% of the annual peak load which vdll in effect double the values presented for Plant Generating Capacities (in % of Grid).

Therefore. given that each nuclear plant is operating at 100%

capacity, a 10% generation loss is assumed.]

o No operator action that would either cause or prevent a unit from tripping is considered.

o All grid systems are standardly operated and specifically designed against unstable frequency oscillations resulting from loss of the largest generating unit on that grid.

16 1

a

& 0 i 4 APPENDIX A LIST OF RETERENCES

List of References

1. U. S. Nuclear Regulatory Commission, Reactor Safety Study, An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants, Appendix III, NUREG-75/014 (WASH 1400), October 1975.

2. Power System Disturbance / Outage / Incident Reporting Procedures, DOE / ERA-0055

3. G. J. Bartok, " Frequency Behavior and Event Prediction of an Electrical Island, Northeast Utilities Service Co.." 76 117-2, January 1976.

4. Underfrequency and Undervoltage Relay Applications to Large Turbine-Generators, National Electric Reliability Council, July 1978.

5. Analysis of 12 Electric Power System Outages / Disturbances Impacting the Florida Peninsula, DOE /RG/06359-1, December 1980.

6. Palo Verde FSAR, Chapter 8.

7. Yellow Creek PSAR, Chapter 8.

8. WPPSS PSAR, Chapter 8.

i

9. Cherokee /Perkins PSAR, Chapter 8.

l l 10. Electric Power Supply ar.d Demand 1981 - 1990, National Electric Reliability Council, July 1981.

l l

l A-1 1- _ _ _ _ _ _ - --- -