ML20207L418
| ML20207L418 | |
| Person / Time | |
|---|---|
| Site: | Peach Bottom  |
| Issue date: | 10/11/1988 |
| From: | Fogarty E PECO ENERGY CO., (FORMERLY PHILADELPHIA ELECTRIC |
| To: | Butler W Office of Nuclear Reactor Regulation |
| References | |
| NUDOCS 8810170288 | |
| Download: ML20207L418 (29) | |
Text
i N.
PHILADELPHIA ELECTRIC COMPANY 2301 MARKET STREET P.O. BOX 8699 PHILADELPHI A, PA 19101 (ais) sai soao
'
- b A.$f *
- October 11, 1988 ftW C LE AR SU PPOR T OtvittON Docket Nos.
50-277 50-278 Mr. W. R. Butler, Director Project Directorate I-2 Division of Reactor Projects I/II U. S. Nuclear Regulatory Commission ATTN:
Document Control Desk Washington, D.C.
20555
SUBJECT:
In-House Reload Licensing for Peach Bottom Atomic Power Station Units 2 & 3 i
REPERENCE: Letter from R. E. Martin, NRC, to W. M. Alden, PECo, dated July 28, 1988, Request for Additional Information
Dear Mr. Butler:
This letter responds to the referenced NRC letter requesting additional information regarding Philadelphia Electric Company's topical report PECO-FMS-0004, "Methods for Performing BWR Systems Transient Analysis".
The attachment to this letter
(
provides PECo's response to the NRC requests.
If you have any questions or require additional information, please do not hesitate to contact us.
Very truly yours, q
]
W Attachment cc:
Addressee W. T. Russell, Administrator, Hegion I.
USNRC l
R. E. Martin, USNRC Project Manager T. P. Johnson, USNRC Senior Resident Inspector l
T.
E. Magette, State of Maryland J.
Urban, Delmarva Power J.
T.
Hoottger, Public Service Electric & Cas H.
C.
Schwems, Atlantic Electric i
f00I B810170200 881011 DR ADOCK 050CO2 7 il g 4
s Dockst Nos. 50-277 50-278 Page 1 of 22 Response to Request for Additional Information on PECo-FMS-0004 1.0 Qualification of RETRAN Computer Model NRC Request No. 1:
r For the PECo version of the RETRAN computer model intended for Peach Bottom 2 and 3 reload analysis, justify: (1) the plant nodalization on a transient by transient basis; (ii) the use of the algebraic slip option for modeling void distribution in the core; (iii) the adequacy and accuracy of the use of the non-equilibrium pressurizer model; and (iv) the criteria used to determine the set of boundary conditions used to initialize the input deck.
In addition, provide a summary and results of at least one operational transient which thoroughly exercises all components of the model, demonstrating that the RETRAN plant model is an appropriate best-estimate model and where appropriate, demonstrating that use of code models is conservative.
In conjunction with the foregoing justification. (1) provide and describe in depth all parametric studies performed to justify the conclusion that the nodalization presented in Section 2.0 of PECO-FMS-0004 is either a best-estimate or a conservative i
representation of the plant, as the case requires; and (ii) since the algebraic slip model was not approved in the RETRAN SER, (1) provide an independent qualification of this model by comparison to applicable experimental data in both steady state and transient flow regimes; or (2) explain in depth the intended use of this model in licensing and operational transient analysis to assure conservative results despite the fact that it has not been quallfled.
Response
The adequacy of the Philadelphia Electric Company RETRAN model nodalization is discussed in Ref.
1.
Purther discussion and the results of various parametric studies are presented in Response 8.
In response to earlier NRC questions (Ref. 1), Philadelphia Electric Company has described the use of the RETRAN subcooled void and algebraic slip models.
As noted in Tables 1 and 2 of Ref.
1, Philadelphia Electric Company plans to ensure conservatism in the licensing applications of the RETRAN model on a transient-by-transient basis.
Included as part of this conservative methodology will be a parametric study of key RETRAN i
input parameters which have been identified as having the most
Docket Nos. 50-277 50-278 Page 2 of 22 significant impact on the results of the limiting Peach Bottom licensing events (load rejection without bypass and feedwater controller failure).
Uncertainties in the RETRAN subcooled void and algebraic slip models (and the other RETRAN inputs listed in Table 2 of Ref. 1) will be statistically evaluated in the parametric study to ensure conservatism in licensing application.
The results of the evaluation will be presented in PECo-FMS-0006, "Methods for Performing Reload Safety Analysis", to be submitted to the NRC at a later date.
As a further qualification of the subcooled void and algebraic slip models, Philadelphia Electric Company has analyzed the three FRIGG testa presented in Section 4.0 of Ref.
2.
The RETRAN predictions (see attached figures) are compared to the test data and to the predictions of the void correlation on which the algebraic clip and subcooled void models are based (Ref. 3).
The results indicate that the models adequately predict the measured test data and are implemented in R ETR AN in a manner consistent with the baue correlation.
The adequacy and accuracy of the use of non-equilibrium pressurizer model is discussed in Response 3.
The criteria used to determine the set of boundary conditions used to initialize the HDTRAN model input deck are described in Response 5.
No one transient event exercises all components of the plant model and results in all the desired trends.
Ilowever, two events analyzed and presented in Section 3.0 of PECo-FMS-0004 together exercise all the major component models.
These are the two M-G trip test (Section 3.1.3) and one of the turbine trip tests (Section 3.3, TT2 is chouen).
A detailed description of the tests and the response of the affected component models is presented below.
Two M-G Trip Test:
The two M-G trip event is initiated by a loss of power to the recirculation M-G set drive motors.
The consequent loss of driving torque results in an initially rapid decrease in the speed of the recirculation pumps.
As the pumps slow down, the stored rotational energy in the components of the M-G nets provides a limited amount of driving torque resulting in a more gradual coastdown.
The coastdown rate is governed by the rotational inertia of the M-G sets and by the recirculation pump frictional and hydraulle torque (as defined by the pump head / flow / torque properties) characteristics.
The accucate prediction of the transient recirculation pump flows (Figo. 3-20
& 3-21) demonstrates the adequacy of this portion of the plant model.
Further information regarding the predicted pump flow is provided in Response 13.
The decrease in recirculation pump speed (and head) results in a reduction in the driving head and
Docket Nos. 50-277 50-278 Page 3 of 22 flow to the reactor jet pumps.
This reduction leads to a decrease in the jet pump (and hence core) flow.
The accurate prediction of the core support plate precoure drop (Fig. 3-16) demonstrates the adequacy of jet pump modeling.
In general, the resulta discussed above demonstrate the accuracy and adequacy of the overall reactor recirculation modeling including the dynamic characteristics of the recirculation M-G sets and pumps, the jet pumps, and hydraulic loeses through the entire recirculation path (which consists of the recirculation loops, reactor core, steam separators, downcomer and jet pumps).
The reduction in the reactor core flow results in an increase in core void.
Due to negative void reactivity feedback, the core power decreases.
As the core power decreases, the void content equilibrates and eventually a new steady-state power is approached.
The accurate prediction of the core power (Fig. 3-15) demonstrates the adequacy of the core neutronics (point kinetics) and hydraulics (which provides the void feedback) modeling on a core average basis.
The decrease In core power results in reduced reactor steam flow and a consequent reduction in reactor pressure.
The turbine Elle system responds by partially closing the turbine control valves to maintain pressure.
As the core power equilibrates, the reactor pressure equilibrates at a value lower than the initial.
The new steady-state presuure is lower d to the reduced steam flow which resultn in a cualler steam dome to turbine inlet pressut - drop.
The accurate prediction of the reactor pressure (Fig. 3-17) demonstrates the adequacy of the turbine EllC model.
The increase in core void content which results in the power reduction also results in a reactor water level increase in the downcomer.
The predicted reactor water level is presented in Fig. 3-18.
The water level is accurately predicted from 0-13 seconds.
After 13 seconds, the predicted level divergen from the data.
This divergence is discussed in fleu ponse 12.
The change in the reactor water level (and to a smaller degree the change in the steam flow) influences the feedwater system.
In response to the level increase, the teedwater system reduces flow in an attempt to bring the water
)
level bacx to the initial value.
The accurate prediction of the feedwater flow (Fig. 3-19) demonstrates the adequacy of the feedwater systen. model.
The analysia of the two M-G trip test exercises and qualifies many of the component models that make up the plant model.
Ilowever, the component models that are important in the prediction of rapid reactor pressurization have not been exercised fully.
This is accomplished in the analysis of the t el 10uir.; r c.d.
Docket Nos. 50-277 50-278 Page 4 of 22 Turbine Trip Test (TT2):
The turbine trip event is initiated by a rapid closure of the turbine stop valves (in approximately 0.1 seconds).
The complete cessation of steam flow immediately upstream of the stop valves results in a rapid increase in pressure due to momentum effects in the steam lines.
The pressure increase propagates down the steamlines at sonic velocity (approximately 1640 ft/sec) and causes a rapid pressurization in the reactor vessel.
The magnitude of the pressure wave is mitigated somewhat by the rapid opening of the turbine bypass valves (approximately 0.05-0.85 seconds).
The pressure wave is reflected several times between the reactor vessel and the closed stop valves.
To accurately predict the system pressure response requires that: 1) the steam line dynamic characteristics are adequately represented by the steam line model, 2) the turbine bypass system dynamic characteristics are adequately represented by the bypass model,
- 3) the dynamic coupling between the steam dome and reactor core is adequately represented b) the steam separator model, and 4) the interaction between the phases at the steam-water interface in the upper downcomer is adequately represented by the non-equilibrium pressurizer model.
A more detailed discussion of these models is presented in Response 8.
The pressurization in the reactor vessel results in the densification of the core moderator (reduction in core void content).
This results in a rapid increase in the core neutron flux due to positivo void reactivity feedback.
The moderator densification peaks at approximately 0.7 seconds and then diminishes as the pressure wave is reflected out of t he reactor vessel and back into the steam lines.
This, in conbination with negative doppler reactivity feedback due to increased fuel temperature, results in a reduction of the neutron flux.
Shortly thereafter (approximately 0.75 seconds), the control rods are inserted and shut down the core before the pressure wave is reflected back into the reactor vessel.
To accurately predict the core neutronic response (assuming that the system pressure response is predicted accurately) requires that: 1) the core neutronics input (1 " kinetics) accurately represents the core, and 2) the core tht< mal-hydraulics modela accurately predict the reactivity feedback variables (moderator density and fuel temperature).
The generation and qualification of the 1-D kinetics input will be described in detail in PEco-FMS-0006.
The accurate predictions of the LPHM flux responses (Figs. 3-46 through 3-S0) demonstrate the adequacy of the 1-D kinetics input and the core thermal-hydraulics models, principally the algebraic slip and subcooled void models.
Purther ir formation regarding these models was provided earlier in this. Response.
The anal / sis of the turbine trip test exercises and qualifies those component models not fully exercised by the two M-G trip test.
These two test predictions, together wi.th the other test predictions presented in Section 3.0, demonstrate that the whole plant model is an
=
Docket Neo. 50-277 50-278
)
Page 5 of 22 adequate and accurate representation of Peach Bottom Atomic Power Station.
i l
NRC Hequest No. 2:
In using the built-in RETRAN separator model, carryover
~
carryunder fractions are held constant at levels based upo..
manufacturer's data.
Justify the use of constant values for
[
transient analysis.
In addition, define what adjustments were necessary in order to use manufacturer's data in the RETRAN model (including steam separator carryover and carryunder, especially j
at the initial startup and operational transient simulation).
Response
l The RETRAN s9parator model contains def ault curves for carryunder and carryover as a function of separator inlet quality and exterior water level.
The curves are based on BWR/6 da'a and are not applicable to Peach Bottom which is a BWR/4.
These curves are replaced using RETRAN input to result in constant values of carryover and carryunder.
The nominal carryover fraction is small (0.00075) and is not a strong function of quality or level.
l Utilizing a fixed value has essentially no impact on transient results.
The nominal carryunder fraction is 0.002.
The dominant effect of the carryunder is its influence on core inlet enthalpy.
i l
At nominal reactor conditions, it contributes approximately 1.3 j
BTU /lbm to the inlet enthalpy.
However, there is approximately an 11 second transport time from the separators to the core 1
i inlet.
This is the time it would normally take for any change in carryunder to influence the reactor core.
At lower reactor flow i
rates, this transport time increases.
Thus, for the majority of transients, especially the rapid pressurization events which are
{
the limiting licensing-type events, the critical portion of the r
transient is over before a change in carryunder can influence the q
]
core.
Theretore, the use of a constant carryunder fraction has j
essentially no influence on most transient analysis results.
l Occasionally a transient is analyzed which results in very low or zero separator inlet quality (e.g. loss of feedwater flow).
To l
assume a positive carryunder fraction under these conditiona can l
l be non-physical.
Thus, when analyzing a transient of this t
nature, the carryunder fraction is linearly ramped to zero as the inlet quality goes to zero.
The initial carryover fraction is specified on the separator r
input card.
The initial carryunder fraction is specified by selecting a core inlet enthalpy consistent with an overall heat balance which assumes a 0.002 carryunder fraction (see Response i
S).
I I
t l
1 4
y-..,_
,__..,,y
__..m-_--_,
7_,. - _ _.. _
2 4
l
\\
Docket Non. 50-277
)
50-278 Page 6 of 22 Request No. 3:
,)
In using the non-equilibrium pressurizer model to account for non-equilibrium effects in the upper downcomer volume, a user 4
limitation was placed on the use of the model that the ateam-water interface must not croan either boundary of the volume during the transient being examined.
The placing of this limitation does not address qualification of the model nor doca it demonstrate that the model'a results will be either best-estimate or conservative during such use.
Demonstrate that the proposed une of the non-equilibrium pressurizer model for the l
upper downcomer volume (including the particular selection of input annociated with that model) in either best-estimate or i
conaervative as the case requires.
Response
l The stanctard RETRAN treatment (llEM) of a separated two phase f
mixture in a node or volume (as in the apper downcomer) is to t
assume that the mixture can be regarded an a single, homogeneous i
4 l
fle'd (i.e. equal temperaturen for both phases).
During a presourization, the homogeneous assumption allows the rapid l
transfer of energy f rom the vapor phase to the liquid phase resulting in the underprediction of the preocurization rate.
]
This has been demonstrated by reanalyzing TT2 and TT3 without specifying the RETRAN non-equilibrium option in the upper 1
downcomer.
The reanalysin resulta in the underprediction of the I
first ateam dome prenaure oscillation by 21% of the measured I
increase and the underprediction of the peak neutron flux by L
approximately 401 for both tents.
The homogeneous assumption in not appropriate for the upper downcomer region because the actual j
heat transfer path between phases in the plant la across the 11guld-vapor interface, which providen very little heat transfer i
area.
It resulta in large, non-conservative biases.
The non-r j
equilibrium model in chosen for this region to appropriately t
j model the interaction between the phasen during transient evento.
In addition, the prediction of pressurization rates is incenaltive to the inter-region heat transfer coefficient, and 1
i thus it in conservatively input in the model an zero.
The j
results presented in Section 3.0 of PEco-FMS-0004 demonstrate 1
i that the une of the non-equilibrium model in the upper downcomer (and ateam separatora) results in accurate (beat-entimate) 4 l
predictions of pressurization (and depreocurization) raten during l
j various tranaient events.
l l
1 i
i l
3 t
i
}
l i
l
~--
i Docket Nos. 50-277 50-278 Page 7 of 22 Hequent No.
4.
l i
i Model qualitication presented in the subject topical report
)
consists of three sets of startup tests to check best-estimate
{
control system modeling.
Without qualifying the best-estimate whole plant model, this best-estimate model was used to perform a l
licensing-type calculation.
Explain what changes must be made to 4
adapt the best estimate plant model (assuming it has been qualifted as such) to the performance of the various licensing l
].
Henponset t
4 Section 3.1 of PECo-FMS-0004 presents the results of model j
i qualification against plant start-up tests.
Most of the analyues t
in Sections 3.1.1 and 3.1.2 qualify best-estimate control systems i
models (feedwater & turbine EllC) independently of the whole plant model.
Ilowever, the feedwater transient at 60% power was also analyzed using the whole plant model, which incorporates the feedwater control system model, verifying that it interfaces with i
the rest of the model accurately via the reactor water level and
)
uteam flow feedback signals.
Only the resultn of the reactor water level response were presented for this analysis as the 1
changes in other reactor parameters were not algnificant.
In
]-
addition, the analyses presented in Section 3.1.3, while done primarily to qualify the reactor recirculation system model, were performed using the whole plant model of which the aforementioned f
j control system models are a part.
The M-C trip test analyses exercised moet components of the whole plant model, including the j
turbine EllC model for pressure control, the teedwater model for i
l level control, the reactor power calculation, etc.
A detailed i
discussion of the two M-G trip analysis is presented in Response i
1.
The SRV analysis presented in Section 3.2 further qualifies i
i the turbine EllC model as part of the whole plant model.
The results of the analysis presented in Sections 3.1 and 3.2 demonstrate the accuracy of the control system models as incorporated into the whole plant model.
Altogether, the l
l anal,"ses presented in Sections 3.1, 3.2 and 3.3 (for a detailed diucussion of the TT2 analysis see Response 1) demonstrate that l
the whole plant modol is an adequate and accurate best-estimate i
representation of the actual plant.
The application of the best-j estimate plant model to the performance of licensing transients la discussed in Hesponse 1 and Response 19.
i j
NRC Hequest 5:
1 In Section 2.0 of PECo-FMS-0004, it is stated that the model i
presented is applicable to a wide range of transients with only
}
l f
Docket Nos. 50-277 l
}
50-278 I
Page 8 of 22 "minor modifications to the model input" which are transient dependent.
Provide a detailed list of what changes are necessary for each licensing and operational transient to be analyzed and a thorough explanation of why each such change is appropriate.
i
Response
The model described in Section 2.0 of PEco-FMS-0004 is nominally configured to represent the plant at rated operational conditions in the steady-state mode.
Typically, to analyze a transient event requires that only a trip card be changed to initiate the event.
Occasionally an event is analyzed at off-rated conditions which may require changing various initial input parameters.
The i
Input parameters, or boundary conditions, which define the i
initial conditions are based on measured or assumed initial values, detailed plant heat balance data, and other component j
data.
They include the following:
1.
Total Core Powcr (MW) - initial value (also distribution for i
j point ilnetics) 2.
Total Core Plow (lbm/sec) - initial value
?
3.
Steam /Feedwater Flow (Ibm /sec) - initial value 4.
Core Bypass Flow (lbm/sec) - initial value determined by l
l steady-state hydraulics calculations 5.
Pressure Distribution (pala) j a)
Steam Dome - initial value b)
Upper Plenum - initial value based on manufacturer
)
equations relating separator and dryer pressure drop to 1), 2) and 3) above and a) i c)
Lower Plenum - initial value based on core pressure drop 1
determined by ateady-state hydraulics calculations and a) and b) i 6.
Feedwater Temperature (Deg. P) - initial value based on heat
[
balance data consistent with 1) and 3) l i
7.
Core Inlet Enthalpy (HTU/lba) - initial value determined by a t
]
heat balance consintent with 1), 2), 3), 6) and a carryunder
{
fraction of 0.002.
8.
Recirculation Parameters - initial recirculation pump speed
[
(rpm) and flow (Iba/nec) are set to be consistent with 2) and g
5) f 9.
Control system models - all control system model initial j
steady-state conditions are set to be consistent with 1) j through 8) i l
Changes in fuel bundle design may require changes to input i
)
related to the reactor core (volumes, junctions, heat conductors, etc.) on a cycle-by-cyc?e basis.
Once the above inputs have been i
j determined, a RETRAN steady-state initialization and null-1 1
i
Dockot Nos. 50-277 j
50-278 i
Page 9 of 22 4
transient are performed to verify that the inputs are consistent I
with each other and implemented correctly.
In addition to the above, certain key parameters are set to conservative values when performing licensing calculations.
These parameters include:
1)
Scram actpointo at technical specification limits.
2)
RPS logic at technical specification limits.
l 3)
Helief valvo capacities at minimum specified values.
I 4)
Helief valse setpoints and response at technical i
specification setpoint limito and maximum specified l
response times.
4 S)
MSIV, turbine control valves, turbine stop valves and i
turbine bypass valves stroke times at minimum (or j
maximum) of deolgn valves.
1 l
6)
No allowance for recirculation H-G net shed or turbine i
l bypass valve actuation when analyzing a generator load reject.
j Occasionally, a transient which results in a large loss of liquid inventory in the downcomer region, such as a loss of feedwater i
flow event, is analyzed.
To prevent the upper downconer volume i
from emptying during this event, and thus invalidating the l
i analysis, volume (and hence inventory) in transferred from the l
j middle downcomer to the upper downcomer in the model.
This is
]
not a change of nodalization per se as no volumes are added or
)l deleted from the model and no volume boundaries are changed.
The overall downcomer volume and initial liquid inventory are j
conserved s.th aufficient initial inventory in the upper l
3 downcomer volume to predict the transient loaa of inventory i
without emptying.
An analysis of the loss of feedwater flow
{
event will be presented in PEco-FMS-0006.
4 NHC Request 6:
1 Explain the process used to determine the gap conductivity.
1 i
f Hesponses f
I Core average gap conductivity (and hot-channel gap conductivity) l in determined using the FROCSTEY computer code.
Qualification of l
FROSSTRY 10 documented in PECo-FMS-0003, ' Steady-State Fuel i
l Performance Methods Report',
NHC Hequest 7:
l In reference to the computer codes used for the report, the licensee's versions of the codes (RETRAN, SIMULATE-E and SIMTRAN-1 i
)
J i
a _: 1 1
Docket Nos. 50-277 50-278 Page 10 of 22 E) were reportedly used.
Document the differences between the PUCo versions and the officially distributed versions.
Hesponset i
j The differencen between PECo's version of RETRAN-02 MODG4 and the i
officially distributed version are documented in Ref. 4.
As l
)
related to transient analysis, changes made to the officially l
distributed version of S1MULATE-E consist primarily of input / output enhancements, accommodations for installation onto the PECO IBM operating system, and the expansion of coding dimensionality to allow for larger problem sizes.
These changen j
have no impact on the generation of physics input for RETRAN.
A i
description of the changes made to SIMULATE-E by PECo is contained in PEco-FMS-0005, "Methods for Performing BWR Steady-
)
State Heactor Physics Analyses".
The PECo SIMTRAN-E code is l
based on a pre-r*1eaue (MOD 008) of the officially distributed 1
vcraion of the code.
Changes made to the code are similar to t
I those described above for SIMULATE-E.
In addition, two more i
substantial modifications were made to the code.
These I
modifications were designed to result in the consistent I
prediction of the core reactivity components and initial power i
distribution between the 3-D simulator model and the 1-D RETRAN f
l model, and are thus normalization options in the code.
A i
detailed discussion of these code modifications will be presented in PECo-FMS-0006.
i i
NHC Hequest H:
7 Since only a limited amount of startup tests and operational j
transients were used for the model qualification (and much of them are for the control system benchmark), qualify the steam separator model, liquid level model, steam line model, bypass i
valve sizing, isolamion condenser, etc., and assess and justify the uncertainty level (or blas) associated with each one of these models.
/
Hgsponset
(
l Qualification of the various component models that make up the Peach Hottom plant model is accomplished in a diverse fashion.
(
l Certain component models such as pumps and control system models j
j can be easily qualified individually against separative effects j
test data.
Other ccaponent models, for which no separate test q
data may exist, such as the steam separator model and the non-1 equilibrium pressurizer nodel (for BWR application), can be qualified only in an ' integrated' tashion as part of the whole l
plant model prediction of test data.
Below is a general i
discussion of the qualification of the various components models i
{
that make up the whole plant model.
1 1
l i
l 4
1 1
Dockot Nos. 50-277 50-278 Page 11 of 22 centrifugal Recirculation Pumps and Jet Pumps:
Qualification of the recirculation pump model and the jet pump model is presented in Section 3.1.3 of PEco-FMS-0004.
Qualification of these models is demonstrated by the accurate prediction of recirculation pump coastdown rate and flow and the accurate predic'. ion of core flow (via core plate pressure drop).
Purther elaboration on the predicted results is presented in response to Hequests 1 and 13.
The validity of these built in RL7RAN models is discussed further in Response Ib of Ref. 1.
Uncertainties associated with various it. puts of the recirculation system ccmponents will be conservatively addressed as described in Hesponse 1.
Turbine flypass Valves and Lines:
The turbine steam bypass system model capacity and pressure drop characteristics are based on manufacturer supplied data.
Qualification of the hypass system is demonstrated in Section 3.3 of PECo-FMS-0004 by the accurate prediction of the turbine inlet pressure for TTI, TT2 and TT3.
When performing liccnsing-type calculations, a conservative bypass valve stroke time (maximum design opecification) is utilized.
Por the limiting Peach Bottom event (load rejection), the bypass system is assumed to fall.
Steam Line Nodalluatitin The 43S ft main steam lines are modeled using six nodes with an average unit length of approximately 73 ft.
Por rapid pressurization events, with a sonic velocity of approximately 1640 ft/see and a pressure wave frequency of approximat.ely lilZ,
one node represents 4.5% of the wavelerigth and thus provides adequate resolution to predict the pressurization rate.
Qualification of the steam line nodalization is demonstrated in Section 3.3 ot PEco-FMS-0004 by the accurate prediction of the propagation of the pressure wave down the steam linen to the steam dome for TT1, TT2 and TT3.
To further demcnstrate that the steam line nodalization is adequate, TT2 and TT3 were reanalyzed utilizing a ten node steam line model.
The repredicted neutron flux peaks were 3.21 lower and 2.4% higher for TT2 and TT3, respectively.
The predicted first pressure oscillation peaks were 0.5 psi (1.3%) and 0.8 pai (1.7%) higher for TT2 and TT3, respectively.
These differences are small and within the measurement uncertainty.
This demonstrates that the 6 vo'ume nodalization provides adequate resolution for the accurate prediction of precoure wave propagation.
Steam _ Separator Models llecause the highly complex, multi-dimensional fluid behavior in a steam separator precludes the development of a fully mechanistic separator performance model, the modeling approach taken is to approximate the separator performance characteristic by utilizing
Docket Nos. 50-277 50-278 Page 12 of 22 available atuady state correlative data.
The two-phase separator model employo quasi-atatic mass and energy uxchange balance equations to describe the separation of the liquid and the vapor phaseo.
The nominal steady-state fluid traasit time through the separatora 10 a few tenths (0.2-0.3) of a second.
The transients analyzed with the Philadelphia Electric Company RUTRAN model do not result in significant reductions in the separator transit time and thus the separator model is used in the range of validity specified in the RETRAN SER.
The moat important aspect ot the steam separator modeling is the prediction of the dynamic coupling between the ateam dome and the core upper plenum.
i Various input. associated with the separator model define the dynamic characteristics of the separators.
These include:
1)
Separator Carryover and Carryunder Practions - The separator carryover and carryunder are defined as described in Response 2.
2)
Separator Pressure Drop - The separator hydraulic loss charact eristic (separator inlet losa coef ficient) is based on a vendor equation correlating steady-state pressure drop data to separatur inlet flow and quality.
3)
Separator Inlet Inertia - The separator inlet inertia 10 based on vendor data correlating separator inertia to inlet quality.
No separative effects transient test data 10 uvallable for qualification of the steam separator model.
Qualification of this modo! la demonstrated in Section 3.3 of PECo-FMS-0004 by the accurate prediction of the steam dome and core upper plenum pressures for TT1, TT2 and TT3 (exceptions regarding TTl are discuaced in Section 3.3 and Response Ib of Ref. 1).
Uncertaintica associated with the inputs to the steam separator model (inlet inertia, pressure drop, etc) will be conservatively addrenced an described in Response 1.
The validity of this built in RETRAN model is discussed further in Response Ib of Ref 1.
Non-equilibrium Pressurizer Modei_t A discunnion regarding the qualification of the non-equilibrium pressurizer model is presented in Response 3.
Dulkwater Nooallzatlon:
The bulkwater nodalization describes the regions of the reactor vessel typically filled with subcooled fluid.
These regions consist of the reactor downconer, the recirculation loups and the lower plenum.
Hecause of the very high sonic velocity in the fluid (approxinutely 3500 ft/sec), the prediction of pressure wave propagation is ', sensitive to the nodalization in these regions.
Therefore, other phenomenon are considered in the development of the nodalization.
The downcomer region is described by three nodes or volumes.
The upper volume models the naturated fluid above the feedwater apargers and contains the
i i
1 Docket Nos. 50-277 i
50-278 I
Page 13 of 22 1
l l
steam-water interface.
The non-equilibrium pressurizer option is chosen for this volume to appropriately model the interaction j
between the 1Iquid and vapor regions (see Hesponse 3).
The middle volume modelb the region where the separator liquid reture j
flow and the reactor feedwater flow mix.
The lower volume models t
the narrower region around the core shroud.
The non-equilibrium J
pressurizer option is not chosen for the middle or lower downcomers.
Stacking a non-equilibrium volume above another may lead to a non-physical ' pancaking' effect in which the liquid region of ore non-equilibrium volume is above the vapor region of i
4
)'
the other.
This effect may lead to unrealistic results.
The upper downcomer region was modeled using only one volume for this i
reason.
The two recirculation loops are described by three f
f volumes each, pramarily to distinguish between the pressures in
(
the auction pip'.ng, the recirculation pump, and the discharge
[
piping, and to accurately predict the dynamic influence of the recirculation pumps on flow.
The lower plenum region, which has l
a large flow..rea and thus very little inertia effects, is 1
j modeled as a single volume.
Core Nodalizations i
The core region is modeled using 24 volumes to describe the active core and one volume to describe the bypass region.
The
{
active core nodalization corresponds to the axial nodalization in i
the 3-1) simulator used to develop 1-D and point kinetics input for HETRAN.
Past experience and the results of test predictions l
(e.g. TT tests, etc.) indicate that the nodalization is adequate I
1 to predict the principal phenomenon occurring in the core region.
d A discussion of the algebraic slip model utilized in the core
)
region is presented in Response 1.
l Qquid I.evel Model:
Qualificalion of the liquid level model is demonstrated by comparison of predicted va actual level response for three l
transient tests.
The analysis of the feedwater system test at i
l 60% power (Section 3.1.1 of PFCo-FMS-0004) indicates excellent l
agreement between predicted and actual water levels.
Analysis of the one M-G trip test (Section 3.1.3) also indicates excellent agreement between predicted and actual water levels.
Analysis of the two M-C trip test indicates good agreement.
harther discussion of the predicted water level for the two M-C trip test 1s presented in Response 12.
Acceptance _ Criteria:
J The establishment. of fixed acceptance criteria for evaluating the accuracy of HUTRAN predicted results va actual measured data can i
be somewhat subjective.
Therefore, the conclusions stated in l
PECo-FMS-0004 are based primarily on sound engineering judgement.
In order to provide a basis for judging the accuracy of the model, an evaluation of the RETHAN predicted results was
]
1 i
t Dockot Nos. 50-277 J
50-278 j
Page 14 of 22 I
i performed using the independent acceptance criteria established in Ref.
5.
Appendices 1 and 2 are excerpts from Ref. 5 which
{
define the acceptance criteria.
Appendix 3 presents the rating of the RETRAN predictions for each test in Section 3.0 based on the criteria.
The overall rating supports the conclusion thgt j
the RETRAN model is an adequate and accurate representation of l
the plant.
i
}
i 2.0 startup Tests
{
i i.
2.1 Peedwater System Transient Tests l
NHC Hequest__9:
f Explain how the whole plant model is qualified on the Lasis of the results obtained by using a stand-alone model of the turbine-l pump dynamic logic.
Hesponse:
I The stand-alone model used to qualify the feedwater turbine-pump I
)
dynamics is an exact duplicate of the model incorporated in the 1
whole plant model.
It was used to demonstrate that given the j
1 same inputs as the actual plant system, it would accurately
]
reproduce the output, feedwater flow.
The norm 41 interaction of 1
the feedwater model with the rest of the plant model is through f
I the reactor water level and steam flow feedbacks.
This 3
interaction was qualified in Section 3.1.1 of PEco-FMS-0004 by i
j analyzing the 60% power feedwater test using the whole plant I
j model (see Response 4).
This interaction was further qualified i
in Section 3.1.3 in the analyses of the recirculation M-G set l
trips where the feedwater system model accurately responds to the changing reactor water level and steam flow.
These test analyses i
(and others in Sections 3.2 and 3.3) support the conclusion that
{
j the whole plant model, of which the feedwater model is an i
integral part, is an accurate representation of the plant.
NHC Hequest 10:
l l
Explain why the 60% power Leut data for the change in NR level l
l shows the hysteresis effect yet the METaAN result does not.
Heaponse:
l An examination of the ateady-utate data prior to the feedwater
(
i Lust initiation indicates a kaximum peak-to peak change in the
{
measured reactor water level of approximately 0.75 inches.
During normal operation, a power reactor tends to exprience a great deal of fluid turbulence and mechanical vibration.
These disturbances can influence the sensitive instruments used to measure a variety of parameters, including the reactor water level.
The apparent hysteresis in the measured data is actually random variation or ' noise' commonly seen in measured parameters l
I
Dockot No3. 50-277 50-270 Page 15 of 22 during normal plant operation.
The idealized RETRAN nolution does not predict thin random variation.
2.2 Turbine Electro-Hydraulic Control Transient Testa NRC Ijeguest 11:
Explain why agreement between RETRAN and tent data la much better at 100% power than at 251 power.
Explain further why RETRAN control logic did not predict the hystereola effect exhibited (Fig. 3-13) by the test data.
Relate the response to the previous Question 10, if applicable.
i
.ljesponnes Piga. 3-7 and 3 9 in Section 3.1.2 of PEco-FMs-0004 indic0te that the maximum difference in the predicted va actual control valve i
error algnal for the :St power tenta la approximately 0.25V.
The maximum difference in the predicted va actual control valve error algnal for the 100% power test data, which appears to be azaller in Piga. 3-11 and 3-13 due to the larger acale of the ordinate, ia approximately 0.40V.
More important in the prediction of the control valve position which governa the ateam flow to the turbine and influences the pressure behavior in the steam linec 1
and reactor vennel.
110th nota of calculations reault in approximately the same relative difference in the predicted vs.
l measured change in control valve position.
ThO RETRAN prediction of the preasure actpoint increase test at 1001 rower fails to predict an apparent hysteresia effect in the control valve error algnal data.
A clone examination of the data and the analyala annumptions revealed two principal reasons for i
the difference in the predicted va meauured algnalu.
It was i
diacovered that an error of 0.02 acconds was made in the nelection of the transient initiation time.
Although this error may appear to be insignificant, the change in the control valve error signal is extremely rapid and 0.02 neconda representa 13%
of the time to peak algnal change.
In addition, the analyals was pericrmed assuming an initial control valve position which resulted in control valve operation in a region whnre the alope of the diode function generator (which relates prJssure control unit output to control valve demand) in changing rapidly, affecting the overall control valve pooltioning loop gain by au much as 601.
The actual initial control valve position was not provided in the data and was encimated for the analyals.
A reanalysin of thin test was performed annum 5ng an initial control valve position of only 0.5% more open than previcualy.
The transient initiation time waa also corrected by 0.02 secondo.
The original analyala was performed uutng the average control valve servo gain an determined from the combined test data 4
(0.15/second).
The reanalysin was performed utiliring the actual control valve nervo closing gain as determined from the data for 1
thin tent (0.1/accond).
The reanalyala resulted in a much i
)
Dockot No3. 50-277 50-278 Page 16 of 22 improved prediction of the control valve error signal.
More i
important, however, la the fact that the prediction of the measured control valve position was improved only slightly.
Because the actual control valve position provides feedback to the control loop, it is more important to accurately predict the control valve demand signal than the control valve error signal.
l 2.3 Reactor Hocirculation Transient Tests I
i NRC Requent 12:
[
t Explain the divergence of the water level prediction from the f
data presented in Pig 3-18 for the M-G set motor trip test.
l f4xplain further why the oscillatory behavior was no' predicted by BETRAN.
I l
Bupponses l
The reactor water level prediction for the two M-G set trip t
analysis in Section 3.1.3 of PEco-FMS-0004 la very accurate op to approximately 13 seconds into the transient.
At that time, the
(
measured data begins to oscillate rapidly.
Generally, there are l
two phenomena that result in a change in reactor water level.
A j
change in the void content in and above the reactor core can j
displace fluid into (or out of) the downconer.'egion resulting in a change in water level.
This is the principal mechanism driving the level increase during the first phase (0-10 seconds) of the l
Ilowever, a change in core void sufficiently large to
[
result in a change in water level of several inches also rcsults in a significant change of core pcwer.
The core power data shows no significant oscillations between 13 to 25 seconds, therefore, it is unlikely that a change in core void is responsible for the
[
level oscillations.
A mismatch between the reactor feedwater flow and steam flow can also cause a change in level by changing
[
the reactor liquid inventory.
The reactor feedwater flow data j
nhows a steady and gredually changing flow.
To achieve the indicated level oscillations by this mechanism would require l
extreme oscillations in the reactor steam flow which would result in significant changes in reactor pressure.
The teactor pressure data shows no significant oscillations between 13 to 25 seconds, therefore, it is unlikely that a mismatch between feedvater flow and steam flow is responsible for the level oscillations.
The 3
cause for the indicated level oscillations la uncertain.
A J
calculation was performed using the utand-alone feedwater system model in which the recorded water level data was used to estimate the feedwater controller output and drive the model.
The predicted feedwater flow matched the measured data well up to the l
time of the leval oscillations (approximately 13 seconds).
After j
that time, the calculation predicted a large change in the feedwater flow (due to the level oscillations) not indicated by
)
the data.
This suggests that another level instrument (there are four) which did not sense large level oscillations may have been providing the input signal to the feedwater system.
Thus, *,here l
(
Dockot Nos. 50-277 50-278 Page 17 of 22 4!
4 1
la nome uncertainty as to the accuracy of the measured water 1cvel data after 13 secondo.
Por licensing-type calculations, i
uncertainties are accounted for by using conservative level notpointo.
i l
]
NHC Hequest_13 Justify the fact that the HETRAN predicted recirculation drive flow (Fig. 3-20) in diverging in the non-conservative direction.
Heaponse
]
The aualyain of the two M-G trip test in Section 3.1.3 of PECo-PMS-0004 was performed assuming symmetric recirculation loop behavior (i.e name initial conditions for both lovps).
A close j
examination of the data indicates an apparent asymmfsry in the recirculation loop responses falso seen in Figo. 3-20 and 3-21).
This may be caused by an actual difference in initial l
recirculation loop conditions or by a misinterpretation of the j
actual initial loop flows on the recording charts due to measurement noise.
The uncertainty in the measured recirculation flows is approximately 5% of the initial value.
The average i
divergence of t he predicted flows in Pign. 3-20 6 3-21 is i
approximately 3% of the initial value which is within the i
I measurement uncertainty.
In addition, Philadelphia Electric Company has performed other calculationc that indicate the I
predicted critical power ratios are insensitive to recarculation coast down rate.
)
I
]
2.4 Peach _Hottom Safety /Helief Valse Test 1
NHC Hequest 14:
\\
]
Explain why in a bent-entimate base model which reportedly represents the typical plant parameters a connervatism was built in for the SHV flow.
Were the valves not claed to yield the rated flow?
Hgaponset l
l 6
i The plant model presented in Section 2.0 of PI:Co-FMS-0004, while
[
described an a best-estimate model, actually contained some conservative inputs, including those for the safoty/ relief valve i
i characterlatica.
To perfcts a best-estimate calculation for the SHV tect, these conservatisms were removed (by overlay) from the model.
The model has recently been modified to replace all l
conservative inputa with best-estimate values.
When performing licensing-type calculations, certain conservative inputs are utilized as described in Response 5.
i l
l i
Dockct Nos. 50-277 50-278 i
Page 18 of 22 NRC Requent 15:
l Explain why, although steam dome pressure was predicted well in the PH2CS SRV lift tect, the second peak in core flux was I
overpredicted by a factor of 2.
Discuno, in depth, how the measured core flu en differ from predictions and the relationship to the use of the qgebraic alip model.
Henponses A cloao examination of the SRV tent results indicated that the i
I predicated peak-to peak pressure rise for the first S4V closure (approximately 14-16 seconds) accurately matches the data (5.7 psi vu S.8 pul) resulting in an accurate prediction of the first flux increase.
The predicted peak-to-peak pressure rise for the second SRV closure (approximately 18-20 seconds) la approximately 201 greater than the data (4.7 pai vn 3.9 pal) resulting in the observed flux overprediction.
This may be due to the lack of data on the actual SRV closure timen and characteristics.
The clonure time for each SRV was inferred from the pressure data and both SHVa were assumed to have exactly the came capacity.
The overprediction of' the flux response 10 not believed to be related to the une of the algebraic slip model, but rather to the overprediction of the pressure increase (for further discunolon of the algebraic allp model see Response 1),
l 2.S Peach Bottom Turbine Trip _Tento i
tJHC Request 16:
I Since plant data were used ar input, how is rapid feedwater flow excurcion almulated?
Heaponne The rapid feedwater flow excursions during the three turbine trip tests were caused by the opening of the feedwater turbine high i
pres-'re steam admlasion valves which are not modeled in the current HETRAN feedwater system model.
Phtladelphia Electric Company plans to upgrade the feedwater system model in the next model revit ion to include the high pressure utcam source to the feedwater turbines.
It should be noted, however, that this aspect of the teedwater model has no impact on predicted operating limits when performirg licensing-type analyses since the opening of the high pressure steam admionion valves occura after the critical portion of the limiting transienta.
NRC Hegue.at 17:
How did the measured control rod scram time and speeds used in the analysia differ from taose normally modeled in the RETRAN base plant model generally used for the report?
Dockot Noa. 50-277 50-278 Page 19 of 22 Responnes The normal neram times and speeda used in the base plant model provided by the fuel vendor and are the ntatistical average a-a of a large quantity of measured data.
The measured valuco used when evaluating the turbine trip testa were nearly identical to the normal valuco.
When performing licensing-type analyses, e acervative scram delay times are utiliced (ace Hesponse 5).
In addition, uncertaintico annociated with scram r.peeds will be ntatiLticalla consideret as part of the conservative licensing methodology discussed in Response 1 and Huuponse 19.
I NRC Hequent 18:
Juutify the tact that although the pressure distribution was predicted well for TT1, all the local neutron fluxen were overpredicted by as much as a factor of 2.
Discuns, in depth, how the measured core fluxes differ from predictions and the relationship to the use of the algebraic slip model.
l Rgsponnes l
The fact that the pressure ' distribution' wau predicted well for TTI in mialeading with regard to the prediction of the neutron flux responsen.
The increase in rieutron flux during the transient in caused by the densification of the core moderator 4
1 which la a consequence of the 11 crease in reactor pressure.
At the time of peak moderator densification (and peak reactivity insertion), the reactor (upper plenum) pressure 10 overpredicted by apr.roximately 2.9 pai or 11% of the measured increase.
This resulta in an overprediction of the peak reactivity insertion of approximately 12% which in turn leads to an overprediction of the nei. tron flux by 47%.
A calculatlun wan performed and presented ir. Section 3.3 of PECo-FMS-0004 which demonutrated that given the correct preasure increane, the neutron flux la accurately l
predicted at all levela in the core.
These results, along with the renultu of the TT2 and TT3 predictions, demonstrate that Hl:THAN accurately predicta the axially dependent change in 4
2 moderatar density (uning the subcooled void and algebraic slip modela) and the neutronics response (uning the 1-D reactivity models).
For further discussion of the algebraic slip model see j
j Heaponne 1.
3.0 1.icensing_Haulu Trannient Analynin A turbine trip without bypaan transient analyuis which wan identifled by the NRC in presented.
The results are compared wit h those obtained by Hrookhaven National baboratory using HELAP-3H during their audit of CE's ODYN computer code.
Provide the following information on this licensing basic transient analysia.
/
Dockot Nos. 50-277 50-278 Page 20 of 22 3.1 Turbine Trip Wit'.out Bypa_s s NHC Hequent 19:
What are the conservative assumptions incorpe 2d into the best-estimate model presented in Section 2.0 of Fu PMS-0004 to create a model cultable for a licensing-type calculation for this tranulent?
Por example, are there any modifications to the cont rol system logic t.o convert the best-estimate (based upon the actual plant data) to a licensing-type model?
Are these licensing conditions different from thune normally used for Peach Hottom?
E"EP90GS:
When performing the NHC tout problem analyalu (Section 4.0 of PEco-FMS-0004) certain key parameter's in the best-estimate model were replaced with conservative values at values consistent with those annumed by GE and DNL (Hef. 7).
These parameters include:
1)
Initial power level at 104.5% of rated (3440 MWL).
2)
Scram setpointa at technical specification limita.
3)
HPS logic delays at technical specification limita.
4) control rod insertion times at technical specification limits (678).
5)
Helieve valve capacitics at values specified in Hef. 7.
6)
Helleve valve setpointo and response timca at the valuca upecified in Hef. 7.
7)
Turbine stop valve stroke time at minimum of design value.
H)
No a1lowance for bypass vaIve actuatlon.
When performing actual PHAPS licensing 6.alculations, a connervative ntatistical approach (almilar to that used by the r
fuel vendor) will be used rather than a strictly deterministic ap-).'oach (au appearn to be the case f or the NHC test problem).
Au part of the overall statistical approach, certain key parameteru are net to conservative values as described in Hesponse 5.
The only changea made to control ayatem modelu for I
licensing application arc in the turbine EllC model to leplement i
the conservative valve stroke times identified in Hesponse 5.
j This conservative licensing methodology is outlined in Heaponso 2 i
of Hof. I and will be described in detail in PEco-FMS-0006.
i i
NBC_ HOju rg_t_ 20 :
1 Explain why the initial core inlet oubcooling is different from
)
the other calculations (apparent in the lower portisn of the core j
void profile in Pig.
- -3).
1 l
Docket Nos. 50-277 50-278 Page 21 of 22
Response
Fig. 4-3 compares the RETRAN initial core void distribution to the GE and BNL distributions in two forms.
The first RETRAN distribution is labeled 'T/H VOID' and represents the thermal-hydraulic void only (i.e no subcooled void).
The second RETRAN distribution, labeled 'NEUT VOID', includes subcooled vo. ids and is intended for comparison to GE and BNL.
The RETRAN initial subcooling dif fers f rom the GE and BNL values by only 1.2 DTU/lbm due to heat balance considera. ions.
NRC Request 21:
Explain why, although RETRAN had similar initial axial void distributions to the other codes used for analysis of the NRC specified test problem (Fig. 4-3), the transient data differed from both other codes (Fig. 4-6) while the core power was between that of the other two codes (Fig. 4-4).
These differences, together with the differences in flux noted in Q.15 and 0 18 indicate that the local void distribution predicted by the RETRAN algebraic slip option may be inaccurate.
Response
The use of the subcooled void and algebraic slip models has been discussed in response to Request 1.
In particular, Philadelphia Electric Company has presented the results of RETRAN predictions of the three FRIGG tests described in Section 4.0 of Ref.
2.
In its qualification of the PBAPS RETRAN model, Philadelphia Electric Company has presented analyses of actual plant tests in which initial plant conditions and plant transient response were accur tely recorded.
These test analyses exercised both the integrated system model as well as the individual sub-system component models.
Comparison of RETRAN calculational results of analytic test problems to those produced by other organizations utilizing different computer codes is difficult and often Icada to ambiguities, especially if the test proolem is not well defined as is the case with the NRC test problem.
Therefore, Philadelphia Electric Company can only conjecture as to the assumptions employed in the other analyses and as to the exact definition of the output parameters edited by the different codes.
In nummary, it in Philadelphia Electric Company's intent to qualify the RETRAN model primarily by comparison to actual plant measured data.. Comparison to analytic calculations performed by other organizations serves only as a general check of expected trends in plant response and parametric relationships.
Docket Nos. 50-277 50-278 Page 22 of 22 References 1.
Letter from J. W. Gallagher, PECo, to W. R. Butler, NRC, dated June 6, 1988, Response to Request for Additional Information.
2.
E.I.
Incorporated, "RETRAN A Program for Transient Thermal-Hydraulics Analysis of Complex Fluid Flow Systems, Volume 4: Applications", EPRI-NP-1850-CCM, Vol.
4, January 1983.
3.
G.S.
Lellouche, B.
A.
Zolatar, "Mechanistic Model for Predicting Two-Phase Void Fraction for Water in Vertical Tubes, Channels, and Rod Bundles", EPRI NP-2246-SR, February 1982.
4.
Letter from L.
P. Rubino, PECo, to L. J. Agee, EPRI, dated January 4, 1988, RETRAN Modifications.
5.
J.
F. Harrison, et.al., ' Qualification of RETRAN for Simulator Application', EPRI-NP-5840, July, 1988.
6.
General Electric, ' Qualification of the One.-Dimensional Core Transient Model for Boiling Water Reactors', Volume 1,
NEDO-24154, October, 1978.
7.
M.S.
Lu, et.al., ' Analysis of Licensing Bacis Transients for a BWR/4 'BNL-NUREG-26684, September, 1979.
i
Docket Nos. 50-277 50-278 Page 1 of 1 Appendix 1 The overall approach for rating each prediction involves the assignment of one of three grades to the prediction of each parameter that has been compared.
Three separate measures of fidelity are considered; magnitude, trend and timing.
A grade of plus (t) will be assigned to the results of a comparison if it is judged to be completely acceptable.
This comparison will probably be within the instrument error for the parameter being compared.
The comparison will probably be given a plus (+) grade if for most of the transient it is within the j
error guidelines shown in Table 2.0-1 for PWRs and Table 2.0-2 I
for BWRs, but is outside for some small time period.
In many transients the prediction and the data deviate when the rate of change of the parameters is the greatest, then come back together as the transient begins to trend to the final condition.
These differences do not generally have an impact on the interpretation of the event by the operator, but can sometimes mean the difference between having an automatic action and not having one.
A plus(+) will be assigned to the trend and timing ratings if the trends and timing of events and major inflection points are consistent with the data and any difference would not affect the interpretation or course of the event.
A grade of zero (0) will be assigned to the results of a comparison if it is judged to be generally acceptable but not as good as one would expect.
The grade of zero (0) is indicative that the difference between the prediction and the data would probably not cause any difference in the interpretation of the event.
The grade of zero (0) will be generally indicative of an error near but on the high side of the instrument error bounds.
Slight differences in the trends or in the timing of events or inflection point will result in a grade of zero (0) for the parameter being examined.
These slight differences would not be in a time frame that would cause difficulties in the interpretation of the event.
A grade of minua L-) will be assigned to the results of a comparison i f ~l t
.s judged to be unacceptable, could certainly
~
cause a mis-interpretation hy the plant operator, have a good possibility of causing an inappropriate automatic actuation or is well outside the normally expected instrument error boundc.
The grade of (-) will be assigned to the trend for a certain parameter if it in counter to the data for a period of time that could cause a problem in interpretat'.on of the event.
Major difference in timing will probably go hand in hand with problems in trend and magnitude.
l I
Docket Nos. 50-277 50-278 Page 1 of 1 Appendix 2 Guidelines for Grading the Error in Magnitude for BWR Pacameters l
Parameter Rating I
(+)
(0)
(-)
Steam Dome Pressure
<10 psi *
<2Gpsi
>20 psi Downcomer Level
< Sin
<10in
>10in Steam Flow Rate
<Si
<10%
>10%
Peedwater Flow Rate
<5%
<10%
>10%
Recirculation Loop Plow Rate <5%
<10%
>10%
Core Flow Rate
<S%
<10%
>10%
Reactor Power
<3%
<6%
>6%
- < 5 psi for Turbine Trip Tests l
l l
1 j
Docket Nos. 50-277 50-278 Page ! of 1 Appendix 3 Rating of RETRAN Plant Predictions Event Parameters Steam Dome Reactor Steam Feedn'ater Recirculation Core Reactor Pressure Level Flon-Flon Flow Flo.
Power FW Test 1 N/A N/A
+
N/A N/A N/A
+
FW Test 2 N/A N/A N/A N/A N/A N/A
+
FW Test 3 N/A N/A N/A
+
N/A N/A N/A FM Test 4 N/A N/A N/A
+
N/A N/A N/A FW Test 5 N/A N/A N/A
+
N/A N/A N/A EHC Test 1 N/A N/A
+a N/A N/A N/A N/A EHC Test 2 N/A N/A
+a N/A N/A N/A N/A EHC Test 3 N/A N/A
+a N/A N/A N/A N/A EHC Test 4 N/A H/A
+d N/A N/A N/A N/A T o M-G Trip
+/o N/A
+/o
+
+
+
+
+b N/A N/A One M-G Trip
+
+
N/A
+
SRV Test N/A N/A N/A N/A N/A
+/o
+
TT Test 1 o
N/A N/A N/A N/A N/A
+/ C TT Test 2
+
N/A 14/A N/A N/A N/A
+
TT Test 3
+
N/A N/A N/A N/A N/A
+
d TT Test I N/A N/A N/A N/A N/A N/A
+
N/A - N3t Applicable a - expected based on predicted control valve position b - based on recirculation pump speed c - + for timing and trend, - for magnitude d - measured steam uome pressure as boundary condition
Docket Nos. 50-277
'50-278 Figure 1 l
l C
i l
l l
l l
I I
l l
C DATA
-o EPRI VOIP RETRAN l
c n.
-o o
l
.P
-o c
=e
- c. -
Jo c
O
-o en t
e i
D N
-c
-c o
0.0 0.j 1 0.l 2 0.l 3 0.j 0.l5 0.6 0.7 0.8 0.l9 1.0 O
I I
l 1
8 t
j i
j i
j i
4 4
I (REL AT I VE)
FRIGG FT-36R 313-019
Docket Nos. 50-277 50-278 1
l l
Figure 2 c
l l
l l
l I
l l
I c
8 DATA
-o EPRI VOID F.ETFW1 e
m
-c e
_e c
--- A
- c. -
._JC C
C r
_=
en
_e N
-e c
P c
i i
1 i
i i
i l
i i
0.0 0.1 0.t 2 0.l 3 0.i '4 0.5 0.6 0.7 0.8 0.9 1.0 e
i t
i i
i i
Z (RELATIVE)
FRIGG FT-36A 313-016
Docket Nos. 50-277 50-278 Figure 3 1
C l
1 l
l I
l I
l l
C DATA C
EPRI VOID RETRXi C
n C
4 C
y c
=e c-JC C
=
C i
m fJ C
N C
l l
)
C i
I I
I I
I l
6 i
i I
I 8
i 8
i 8
0.0 0.1 0.2 0.3 0.4 0.5 0.5 0.7 0.3 0.9 1.0 C
i I
(REL AT I VE)
FRIGG FT-36A 313-029