ML20212J971
| ML20212J971 | |
| Person / Time | |
|---|---|
| Site: | Calvert Cliffs  |
| Issue date: | 02/24/1987 |
| From: | Tiernan J BALTIMORE GAS & ELECTRIC CO. |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| NUDOCS 8703090186 | |
| Download: ML20212J971 (36) | |
Text
\\
I BALTIMORE l
GAS AND l
ELECTRIC CHARLES CENTER P. O. BOX 1475 BALTIMORE, MARYLAND 21203 JosEPN A.TIERNAN VM:t ParslOENT NucttAn Ehrnov February 24,1987 U. S. Nuclear Regulatory Commission Washington, D. C. 20555 ATTENTION:
Document Control Desk
SUBJECT:
Calvert Cliffs Nuclear Power Plant Unit Nos.1 & 2; Docket Nos. 50-317 & 50-318 RETRAN Review - Submittal of Additional Information
REFERENCE:
(a)
Letter from Mr. S. A. McNeil (NRC), to Mr. 3. A. Tiernan (BG&E),
dated November 20, 1986, Request for Additional Information -
RETRAN Review Gentlemen:
In Reference (a), you requested additional information regarding Topical Report A-85-ll (RETRAN Computer Code, Reactor System Transient Analysis Model Qualification) which was submitted in January 1986. Our response is provided as an attachment to this letter.
Should you have any questions regarding this matter, we will be pleased to discuss them with you.
Very truly yours, hD PD P
JAT/LSt./ dim Attachments
[J2: [ d h ck r [d8 cc:
D. A. Brune, Esquire I
- 3. E. Silberg, Esquire 08l ig A. C. Thadani, NRC
I
- 5. A. McNeil, NRC (3 Copies)
T. E. Murley, NRC T. Foley/D. A. Trimble
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION The following responds to your request for additional information regarding Topical Report A-85-Il (RETRAN Computer Code, Reactor System Transient Analysis Model Qualification).
- l..
BG&E used 'the RETRAN-02/ Mod 03 computer code which has not been generically approved by the NRC.
Discuss the differences between this code and the RETRAN-02/ Mod 02 code (which is the NRC approved version) and the impact of these l
differences on transient and accident analyses.
Response
l The NRC Safety Evaluation Report (SER) (Reference (1)) for RETRAN states that RETRAN-02/ Mod 02 was the code reviewed. As a result of code errors identified during the review process, the SER for RETRAN-02/ Mod 02 states:
l "The staff requires that these errors be corrected in the approved versions of
'RETRAN-01 and RETRAN-02 prior to safety related application of these codes. Code modifications should be corrected according to established error reporting and control change procedures. No analyses in support of licensing actions should be submitted utilizing uncorrected versions of the code."
RETRAN-02/ Mod 03 is the Electric Power Research Institute (EPRI)/ Utility Group for l
Regulatory Action (UGRA) revised version of RETRAN-02/ Mod 02 with the identified errors corrected. The NRC was notified in letters from both Mr. L. 3. Agee (EPRI) and Mr. T. W. Schnatz (UGRA) to Mr. C. O. Thomas (NRC) (attached References (2) and (3)).
As stated in Mr. Schnatz's letter, no significant differences were noted when 10 sample problems were' rerun with RETRAN-02/ Mod 03 and compared to the results of RETRAN-02/ Mod 02 (see EPRI NP-1850 CCM-A, Vol. 3). Finally, the RETRAN-02/ Mod 02 document, EPRI NP-1850-CCM, was modified to EPRI NP-1850-CCMA as required by NUREG-0390 and represents RETRAN-02/ Mod 03.
The above status of RETRAN-02/ Mod 03 is also discussed in attached References (4) and (5).
2.
Provide the data and comparison which support BG&E's conclusion that the use of a single node for the secondary side of the steam generator will produce acceptable results. In addition, provide and justify the types of transients for which such a single node secondary volume is valid.
Response
The statement of acceptable results for a single node steam generator (SG) secondary side l
1s made on page 13 of the BG&E Topical Report (Reference (6))in discussing the RETRAN one loop model. It should first be noted that the two loop and four loop models, discussed on BG&E Topical Report pages 16-20, utilize a four node recirculating SG secondary side model.
T ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION The BG&E Topical Report presents two analyses using the one loop model with a single node SG secondary. The first analysis, four pump coastdown from 20% power (Sec. 4.2), is not affected by the secondary side. The dominant parameter of interest is transient primary coolant system flow rate. The second analysis, cooldown to residual heat removal entry (Section 5.2), compares primary pressure and temperature and secondary side pressure for 11500 seconds with a similar TRAC model calculation. RETRAN results closely follow the TRAC calculation results. The TRAC model uses 26 calculation mesh cells for the SG secondary side. This excellent comparison between RETRAN and TRAC substantiates the adequacy of the RETRAN single node SG secondary side.
In EPRI NP-1850-CCM Volume 4 (the RETRAN-02 EPRI report that was reviewed by the NRC prior to issuing the general RETRAN SER), an analysis of the Prairie Island Steam Generator Tube Rupture event with RETRAN is presented on pages VI-51 to VI-65. The Prairie Island RETRAN model of this two loop Westinghouse U-Tube SG Plant used a single node SG Secondary. This analysis shows good agreement between RETRAN and measured plant data. for Pressurizer and Steam Generator Pressure, and Hot and Cold Leg Temperature.
The Virginia Electric and Power Company (VEPCO) submitted a RETRAN topical report to the NRC in 1981. This report (Reference (7)) was accepted and VEPCO received an SER (Reference (8)) in 1985. Included in this report is a cooldown and safety injection transient for North Anna (this section is attached). VEPCO used a single loop, single node SG secondary RETRAN model to analyze a stuck open condenser dump valve causing Reactor Coolant System (RCS) depressurization and safety injection. Close agreement between RETRAN and plant data for SG Pressure, Hot and Cold Leg Temperature, Pressurizer Pressure and Level was shown for this event.
The major disadvantage of a single node secondary volume is that a discrete mass and energy distribution is homogenized. This distorts the SG downcomer level and riser region heat transfer coefficients. Instead of a subcooled downcomer fluid approaching saturation in the riser region, the single node models the entire secondary side as a saturated fluid.
Also, tube bundle region velocity is not correctly simulated in a single node bemuse of recirculation. Even adjusting flow area to account for the recirculation ratio would only be valid at a single power level because the recirculation ratio varies with power.
Therefore, a single node SG secondary is valid whenever accurate level prediction is not necessary or, very large changes in secondary inventory and changes in heat transfer mode i
on the tube bundle (i.e., counter current flow, primary side condensation) are not expected.
3.
Justify the use of a single node non-equilibrium pressurizer regarding rapid insurge and outsurge transients.
Response
The RETRAN non-equilibrium model is used in the single node representation of the pressurizer.
This model effectively breaks the node into two mass and energy conservation regions (liquid and vapor) which are not required to be at the same ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION temperature. This allows the model to effectively track a varying mixture level. During insurge transients, the liquid region will subcool due to the incoming hot leg fluid.
Although the basic single node non-equilibrium model.(which assumes complete instantaneous mixing of the liquid region) may not be accurate for a rapid insurge/outsurge transient, a longer term insurge or outsurge transient can be acceptably modeled. This is because homogeneous mixing of the incoming fluid with the pressurizer fluid does occur for the longer term transient.
To overcome the limitation of the single node model described above, RETRAN can use a Temperature Transport Delay Time Model (TTDT) in conjunction with a single node non-equilibrium pressurizer to properly simulate a rapid insurge and outsurge transient. This model tracks the movement of a temperature front to the pressurizer volume and allows the user to divide a single node into multiple mesh intervals for simulating a slug of lower / higher temperature fluid moving within a node. A more detailed explanation of the TTDT Model can be found in the RETRAN-02 Report, EPRI NP-1850-CCMA Vol. 3, page IV-52 and Vol.1, Rev. 2, page VII-13.
Finally, as indicated on page 27 of the BG&E Topical Report, a RETRAN two-node pressurizer without TTDT, as compared to a single node pressurizer without TTDT, had no significant effect on Pressurizer Pressure and Level response for a typical rapid insurge/outsurge transient.
4.
For the Multiple Secondary Side Malfunction (MSSM) event, BG&E assumes three heat transfer slabs transferring thermal energy to a single hydraulic node in the pressurizer.
Discuss the effect of not subdividing the hydraulic node in the pressurizer in light of the temperature stratification of the fluid in the pressurizer.
Response
The selection of three heat transfer slabs in the pressurizer was based on R'1TRAN input requirements for a minimum number of " stacked" heat conducters in a single volume, and engineering judgement on heat transfer modeling of these conductors. For the Multiple Secondary Side Malfunction event, the single hydraulic node was subdivided into two nodes as part of the sensitivity study (Sec. 4.1.4 of the BG&E Topical Report) with no significant effect on Pressurizer Pressure or Level response (Item 3 on page 27 on the BG&E Topical Report). The large flow area, small pressure loss coefficient, and small inertia of the junction connecting these two nodes allow almost instantaneous temperature changes for both nodes.
However, using the TTDT model (see answer to question 3) allows for effective tracking of a temperature front as it enters the pressurizer without homogeneously mixing it throughout the node.
The MSSM event was rerun using the TTDT model with 20 mesh intervals (recommended in the RETRAN-02 input manual). Key results were compared with the case reported in the BG&E Topical Report (without the TTDT model) and measured plant data and are summarized below. - -.
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION Parameter Plant Data Topical Report Case TTDT Model Case Peak Pressurizer 2306 2338 2310.7 Pressure (psla)
Time of Peak (sec.)
48 51 51 Peak Pressurizer 254 244.4 246.3 Level (inches)
Time of Peak (sec.)
60 66 66 The above data shows that the TTDT model reduced peak pressure from 2338 psia to 2310.7 psia which is closer to the measured 2306 psia. Also, the TTDT modelincreased peak level by about two inches closer to plant data. A comparison of pressurizer liquid space temperatures (during the insurge segment of the transient) between the two RETRAN cases shows that the TTDT case was a maximum of 0.5 F cooler (at 73 seconds) than the base case without the TTDT model. BG&E plans to use the TTDT model whenever a similar phenomena is expected to occur in the pressurizer.
5.
Based on our review of Se: tion 4.1 of the topical report, we have identified several anomalles. Provide explanations for the following anomalles:
(1)
BG&E had postulated that "possibly higher run-out auxillary feedwater (AFW) pump flow rates and lower post trip main feedwater (MFW) temperatures caused more rapid cooldown af ter 160 seconds in plant data." However, we note that the plant data show the SG21 level pegged at about 200 sec and that differences in !cvel in SG21 existed from about 70 seconds, not 160. The RETRAN calculation of level in SG21 lagged the actual refill data by roughly 20 seconds throughout the transient; (2)
The slope dif ference in the Steam Generator Level between the RETRAN calculation and the plant data for SG22 resulted in a difference in dry out time of about 40 seconds.
(3)
The effect on the transient of the broader spike in RETRAN analysis.
(4)
What happened in the plant data at approximately 140 seconds following the event?
Why was this transient curve not followed by the RETRAN calculations? Is it related to dryout of the steam generator? Would a more detailed steam generator secondary side nodalization give better results?
4
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION Responses (1-2)
Steam Generator Level is based on wide and narrow range taps on the SG downcomer which are connected to a reference leg. The RETRAN model explicit-ly includes the actual tap locations in calculating SG Level. RETRAN calculates SG Level by applying a correction factor to the calculated differential pressure using the Bernouill Equation, and by accounting for the applicable downcomer fluid density, tap elevation, flow rates and flow area.
Since the four node RETRAN SG secondary model has only one node for the downcomer, the average downcomer thermodynamic propcrtles are used in this correction factor.
I The reported SG Level plant data is actually provided by one of six separate instrument channels for each SG. Combustion Engineering (CE) has identified significant differences between Individual channel level indications and has postulated that this channel discrepancy may be attributable to:
A. Flow perturbations from uneven local downcomer flow rate distribution and currents (including the effect of asymmetric SG shroud cutouts for hand-eddy / tap locations).
hole B.
Inconsistent / erroneous individual instrument channel corrections.
l C. Variations in reference leg temperatures which affect the correlation of reference leg level to SG Level.
Combustion Engineering has determined that the measured SG Level must be corrected to account for power dependent velocity head (from constricted shroud.
to-shell area near the taps), contraction friction losses, and power dependent SG pressure. CE has only provided corrections for steady state power operation at different SG Pressures, power level, and reference leg temperature.
The dif ference in dryout time is really only a difference in reaching the lower tap of the SG Level measurement Instrumentation.
This tap is well above the tubesheet.
in summary, the plant measured SG Level may significantly differ from actual level due to numerous instrumentation errors (channel differences) and the presence of physical phenomena in the downcomer and reference leg that are not accounted for by the level Indication. Some of these phenomena are not modeled by RETRAN. A more detailed nodalization of the SG downcomer region may improve RETRAN calculated level.
However, in light of the uncertainty of measured level and the significant increase in computer run time previously experienced with more detailed SG nodalization, a more detailed SG nodalization would not be practical.
i i
-S-
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION (3)
The broader SG pressure spike in the RETRAN analysis results in a slightly larger steam flow rate out the Atmospheric Dump Valves (ADVs) and Turbine Bypass Valves (TBVs). Up until about 50 seconds, RETRAN SG Pressure is lower than plant data. From about 70 seconds to 120 seconds, RETRAN SG Pressure is 4
greater than plant data. The maximum difference of about 50 psi (out of 900 to 1030 psla) and relatively short time for this difference does not significantly affect primary coolant system thermal-hydraulic response as evidenced by the l
excellent comparison of this calculation to plant data.
~
At approximately 140 seconds, SG Pressure dropped to 870 psia, the setpoint at (4) which all TBVs should have closed, but one stuck open. No other information on the position of this stuck-open TBV was available except that it was identified and l
Isolated at about 300 seconds. As discussed on pages 24 and 25 of the BG&E l
Topical Report, the RETRAN model assumes that the TBV was half open. Some RETRAN sensitivity studies and examination of SG Pressure response indicated that the stuck-open TBV position was more likely varying between 140 and 300 seconds. The complexity and the consuming nature associated with arbitrarily using different time dependent TBV areas to better " match" plant data was deemed impractical in light of the reasonably close agreement from the half open TBV assumption.
4 6.
The statement that the " difference in SG pressure response accounts for differences in
)rimary coolant systems cooldown af ter 160 sec"is not exactly accurate. The differences
.n secondary side pressure began at roughly 70 seconds and the large change in slope in secondary pressure data at roughly 110 seconds was not matched by the RETRAN calculation of the SG Pressure although the general characteristics of the SG Pressure was followed. Explain the sources of all the significant deviations and justify the differences j
between the RETRAN calculation and the plant data.
Response
Steam Generator Pressure response during this event is due to the fo!!owing factors:
1.
Pre-trip reactor power level and steam flow rate to the turbine.
2.
Timing of the TSVs closure and the TBVs and ADVs opening.
3.
The magnitude and enthalpy of the MFW.
4.
The stuck-open TBV area.
I 5.
Additional steam flowpaths.
s I
L i
{ I
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION Of the above factors, only reactor power level and MFW flow were measured parameters.
All other factors were accounted for by deducing their values from other data and plant experience. Uncertainties in these factors can affect RETRAN calculated SG Pressure well in excess of the relatively small differences shown in Figure 4.1-7 of the BG&E Topical Report. It should also be noted that the differences between RETRAN calculated SG Pressure and plant data are bounded by the instrument tolerance delineated in Table 4.1-2 of the BG&E Topical Report.
7.
Expand your discussions of the RETRAN model sensitivity study to support your conclusion stated on page 27 of the report. In particular, did BG&E perform sensitivity studies to determine what happened on the secondary side and its impact on the primary side? If not, then discuss the secondary side flow data and the RETRAN modeling thereof, and Justify any differences by discussion of their impact upon the overall transient results.
Response
The conclusions on page 27 of the BG&E Topical Report were specifically made regarding pressurizer modeling sensitivity studies and their effect on transient Pressurizer Pressure and Level response.
A sensitivity study of secondary side effects on the primary side was performed in examining the performance of the Turbine Stop Valves (TSVs), TBVs, ADVs, and Turbine Governor Valves (TGVs). These four secondary side valve systems directly affect the steam flow rate and associated primary system pressure, level, and temperature response. During the early stages of this analysis, a number of RETRAN runs were made in which the timing and/or position of these valves were varied.
Prior to the reactor trip on low SG level at 62 seconds, the transient TGV steam flow rate (as the operators attempted to match decreasing reactor power down to 70%) had a strong effect on the magnitude and time of peak Pressurizer Pressure, and Level, SG Pressure, and Hot and Cold Leg Temperature. As stated on page 24 of the BG&E Topical Report, the measured first stage turbine pressure and its relationship to turbine inlet steam flow rate was used to deduce the TGV position.
Upon reactor trip, the TSVs close and the TBVs and ADVs open. The time delays of both the TBVs and ADVs opening and the TSVs closing were found to directly affect the slope of decreasing Primary Pressure, Temperature, Pressurizer Level, and SG Pressure and Level during the first 120 seconds. Best estimate time delays were used for the final analysis reported in Figure 4.1-3 through Figure 4.1-8. At 120 seconds, SG Pressure dropped to 870 psia, the programmed value below which all TBVs should have closed. Af ter this time, the position of the stuck-open TBVs affected the primary system pressure, level, and temperature response as well as the SGs Pressure and Level.
Without any direct measurement of the stuck-open TBV position, a large number of possible transient positions could be postulated. The final analysis of the BG&E Topical Report assumes a half-open TBV until isolation at 300 seconds, i
4 7
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION 8.
What input loss coefficients were adjusted to establish initial conditions in the transient of four pump coastdown from 20% power? Were any reverse flow loss coefficients used for the reactor coolant pumps? If the reverse flow coefficients were used, describe what caused the steep changes in flow data at roughly 45 sec? Were the same coefficients used in Section 4.37 Was there a change in flow regime which caused the flow meter to give a different correlation? If the plant data is correct as plotted, why was the slope change not followed by RETRAN? Finally, provide estimates of the natural circulation that is predicted by RETRAN and compare RETRAN's estimates with the expected values.
Response
Input loss coefficients were not used to estab!!sh initial conditions. Rather, the initial conditions for the four pump coastdown from 20% power were established by the use of the initial pressure in each designated volume of the primary coolant system. When initial pressures are input, RETRAN calculates loss coefficients. The pressure drops are based on four pump operation at 100% power. The difference in RCS density between 100% and 20% power, which affects the total primary loop pressure drop by < 2%, was accounted for by appropriately adjusting the overall loop pressure distribution. This best estimate pressure distribution was provided by CE to BG&E for the Pressurized Thermal Shock (PTS) project and was used by the Los Alamos National Laboratory in the TRAC model of Calvert Cliffs.
Although no forward pressure loss coefficients were input for the primary system, the RCP reverse flow loss coefficient provided by CE was used and is the same loss coefficient used in Sections 4.3 and 4.4.
This is discussed further in the response to question 9.
We have concluded that the steep change in flow data at 45 seconds shown in Figure 4.2-1 is an error in the measurement data. This is based on discussions with both CE and Byron-Jackson (the RCP manuf acturer) as well as an evaluation of the four pump coastdown data for other plants.
This steep flow change can not be justifled by the basic physical phenomena affecting flywheel equipped centrifugal pump coastdown.
The RETRAN estimate of natural circulation is addressed in Section 4.5 and further discussed in the answer to question 11.
9.
BG&E states that sensitivity studies were performed to adjust the downcomer cross-flow and RCP reverse flow coefficients. Although the cross-flow takes place across a large area, perhaps the flow modeling is inaccurate and/or incomplete due to failure to model cross-flows in the above core region. Explain how the flows in the above core region were modeled; and justify BG&E's modeling of the cross-flow. Provide detailed results from sensitivity studies of variation of cross-flows and reverse flow coefficients. Provide plots to accompany the analysis and discussions on comparison of RETRAN results with the measured plant data for each of five combinations of RCP operation (4, 3, 2 in the same loop, 2 in opposite loops, and 1 RCP running). Explain how flow rates of various cross-flows were determined for the calculations, including the bypass flows which " flows circumferencially around the downcomer and out the cold leg nozzle (s) in the reverse direction through the shutdown RCP(s) to the steam generator outlet plenum."
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION Compare the predicted flows in these various combinations with vendor computations and explain any_ significant. differences. Provide the data that are available to verify the
" actual flow path" for reduced RCP operation configurations. Describe how the RCP locked rotor reverse flow coefficient was determined.
Response
Flows in the "above core regions" are modeled by dividing this region of the Reactor Vessel into four volumes. These four volumes and their respective RETRAN model node numbers (BG&E Topical Report Figure 3.3-1 or Figure 3.4-1) are: Upper Plenum A (25),
Upper Plenum B (26), CEA Shrouds (27) and Upper Head (28). RETRAN junction flow paths are provided to allow flow from each Upper Core (19 and 20) and Core Bypass (8) volume to the two upper plenum and CEA shroud volumes. Also, cross-flow between the two upper plenum volumes is provided in the RETRAN model. Finally, the upper head can receive (and provide) flow to both the upper plenum and the CEA shroud volumes. Thus, the "above core region" cross-flow is modeled by the following flow paths:
1.
Cross-flow between two separate upper plenum volumes.
2.
Flow from both upper core volumes to the CEA shrouds which flow to the upper head. The upper head then has flow paths to both upper plenum volumes.
Both core bypass and upper plenum bypass (i.e., CEA shrouds) flow rates were also provided by CE during the PTS project. Using the known pressure distribution provided by CE (see answer to Question 8) and the flow rates, RETRAN calculated all pressure loss coefficients in the "above core region" cross-flow junctions except the upper plenum cross-flow junction.
This junction, initially set at zero flow, requires an input loss coefficient. The input loss coefficient was calculated using the geometry for flow across a bank of tubes (i.e., CEA shrouds). This is described in the Handbook of Hydraulic Resistance by I. E. Idel'Chlk (Reference (9)).
The upper plenum cross-flow loss coefficient was varied by 50% with no significant effect on the resulting loop flow splits. As stated on page 51 in the BG&E Topical Report, varying the downcomer cross-flow loss coefficient from.02 to 13.2 and varying the RCP locked rotor reverse flow coefficient by 14% had no significant (i.e., less than 1%) effect on the magnitude of the flow split.
No plots are provided of results because the measured plant data for the flow through each cold leg (flow split) is a steady state value (provided in Table 4.3-1 of the BG&E Topical Report). The RETRAN calculations require 50 to 140 null transient seconds (depending on the particular case) to reach a steady state when initialized with no pumps running and followed by a start of one to three pumps.
Flow rates for various cross-flow paths are directly calculated by RETRAN and are provided as specific junction flow rates. For example, the core bypass flow rate is provided in the downcomer cross-flow junction No. 5 of the RETRAN Calvert Cliffs Nuclear Power Plant (CCNPP) model.
+
~
r l
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION CE has informed us that no vendor computations are available for comparison for these RCP flow split tests. The data that substantiates " actual flow path" is the flow direction for each cold leg reported in the test results.
The RCP locked rotor reverse flow coefficients were provi6d by CE.
These flow coefficients were determined from predicted "K" factor curves based upon 1/5 scale model test results and homologous pump data. These coefficients have also been compared a full scale test data.
CE uses these same coefficients in analyzing the locked reto/
accident in the Calvert Cliffs FSAR for reload analysis.
- 10. With regard to the transient of one pump coastdown from 80% power, discuss the basis for adjustment of reactor coolant pump (RCP) moment of inertia and torque including the uncertainty of such pump information. Were any cross-flow sensitivity studies performed for the one RCP coastdown transient? The transient computation seems to substantially over-predict the flow reduction in the shutdown loop. Identify and describe the likely sources of over-prediction. What results are obtained without cross-flow and with nominal reverse flow coefficients? Why does plant data show nearly instantaneous pressurizer response while RETRAN seems to show no response for approximately 5 sec7 Describe why, even though both the Hot Leg Temperature and the SG Pressure increase for about 5 seconds immediately following the RCP trip, and the Cold Leg Temperature remains constant, neither the primary nor the secondary side pressures immediately increase in the RETRAN calculations. Describe and quantify to the extent feasible the differences in the results between the case in which BG&E adjusted the RCS total flow to match Hot and Cold Leg Temperatures, and the case using the initial conditions described in Sections 4.3 and 4.4.
Response
The RCP moment of inertia and torque were varied by 10% to determine the effect of the parameters on RCS flow coastdown. The RCP manufacturer, Byron Jackson, has stated that the nominal values for Calvert Cliffs' RCP moment of inertia and torque are calculated (by Byron Jackson) from RCP design data and are expected to have an uncertainty of no more than 5% Therefore, the 10% variation of these parameters was intended to envelope actual uncertainties in these values provided by the RCP vendor.
As presented in Table 4.4-1 of the BG&E Topical Report, the base case without the downcomer cross-flow coefficient and a realistic RCP reverse flow loss coefficient was compared to a case which used these coefficients.
Downcomer cross-flow and RCP reverse flow had no significant effect on RCS flow except at 18 seconds when these refinements resulted in a higher flow (79.1% versus 78% for the base case and 80% for plant data) which compared more closely with measured data. Varying the downcomer cross-flow loss coefficient had no significant effect on the RCS flow coastdown.
The RETRAN transient computation, for the base case with downcomh cross-flow and RCP reverse flow model coefficients, under-predicts RCS total flow (i.e.. over-predicts flow reduction in the shutdown loop) by 0.4% to 3.9% During the M-second period of interest, RCS flow measurement accuracy is 12% in addition, the plant Ata presented for this event was taken from a plot in which visual data discrimination due to the plot l -
i ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION grid size is approximately 11%
in light of the aforementioned uncertaintles, the dif ferences in predicted flow coastdown is not considered " substantial." Possible sources of this difference in flow coastdown include differences in actual RCP moment of inertia
/ ^,
and torctic, measurement instrumentation accuracy, and time lag.
Results without cross-flow and with nominal reverst: flow coefficient are reported as the
" base case" in Table 4.4-1 fhe five second difference between RETRAN pressure / temperature results and data is due
/ to the Imnosition of a five second delay of the Ti;Vs and ADVs opening in the RETRAN model This delay was based on previous experience with the TBVs and ADVs performanc2. A f ew case was run with instantaneous opening of the TBVs and ADVs. The results are oscrlayed on Figures 4.4-2, 4.4-3, and 4.4-4, from the BG&E Topical Report htt;nhed Figures A10-1 through A10-3). These figures show that faster T8V and ADV operlag results.n RETRAll Pressurizer Pressure and Level, and Hot and Cold Leg
,Tr.nperatures compring closer to plant data in both magnitude and timing.
Another case with instantaneous TSV closing, but ti<e second delayed TBV/ADV opening, was run and also showed a closer comparison with SG Pressure (overlayed on Figure 4.4-5 or tiv BG&E Topical Report and shown on attached Figure A10-4), it should be noted that theM.TRAN simulation of this event did not allow a reactor trip (and concurrent turbine trip) ontil actuated by the low RCS flow signal which did not occur until about 2.5 seconds.
Any SG Pressure variation prior to 2.5 seconds would not be calculated.
In those cases in which RCS flow was adjusted to match liot and Cold Leg Temperature, the flow adjustment was less than 1% of the Initial conditions described in Section 4.3 and 4.4 and had no significant effect on othat parameters.
/
6 11.
It appears that the RETRAN calculation for a totalloss of flow transient from 40% power was inithtoi with a slightly c'If.etent }!st Leg Temperature from the plant data. Was there any dif ficulty in it'.:lalizit g the calculation at 40% power?
Were any loss coefficients adjusted in s :tting up initial conditions? If so, how did they compare to adjustments in Sections 4.3 and 4.47 What happened in the steam generator at 10 seconds (see Figures 4.5-2, 3, and 5) and why wasa't it modeled in RETRAN? (We note that there was relatively pt.or transient :omputation of temperatures between 10 and 60 seconds because of these dif ference:..) Since the time between 60 seconds to 130 minutes is considered to be the more important period, provide transient curves for that period. The transient curves of' primary flow and prastwe are in agreement but do not agree for the Hot and Cold Leg Temperatures and Se.; secuadary side pressure. What is the reason to co"Se Ahls ejfect and v.hy did %8 secondar/ side.wt have a stronger coupling to the primary side?
- !1 -
MTACHMENT t-RETRAN REVIEW - ADDITIONAL INFORMATION
)
If
Response
The measured total loss of flow test initial Hot Leg Temperature was 555 F; whereas, the RETRAN calculated initial Hot Leg Temperature was 554.4 F.
Small differences in temperature, flow or power instrumentation accuracy (see Table 4.1-2 of the BG&E Topical Report) could account for this difference. Further refinemeiit of this model, to exactly match the measured Hot Leg Temperature, was deemed to be unnecessary because this small difference does not significantly affect the important results of this test (i.e.,
the onset and magnitude of natural circulation flow).
Loss coefficients were not adjusted in setting up initial conditions since the CCNPP model was initiated with a pressure distribution obrained from CE (discussed in our response to question 8). This' pressure distribution, based on 100% power, was adjusted to account for density differences associated with 40% power.
Performance of the TBVs and ADVs was not available in the startup test report. RETRAN assumes normal full power operation of these valves (i.e., quick opening TBVs and ADVs with the ADVs closing at 870 psia). In the past, slow opening times for TBVs have occurred at Calvert Cliffs. A slow opening TBV would cause the higher peak pressure (at 10 seconds) shown in Figure 4.5-5. Without the test specific ADV/TBV performance data, it was impossible to more accurately simulate SG Pressure response.
No transient test data plots were available from the startup test report for the time period of 1 to 130 minutes. The test indicated that natural circulation was confirmed at about five minutes because Hot Leg Temperature decreased at this time. RETRAN calculated Hot Leg Temperature also decreased at this time.
As previously discussed, a lack of information on TBV and ADV performance affected RETRAN's results of Hot and Cold Leg Temperature and SG Pressure. However, during the 60 second post-trip period, there is a strong coupling between the secondary and primary side data. The higher SG Pressure peak shown in plant data is reflected in the faster hot leg cocidown since a larger ADV and TBV flow would result from higher SG j
Pressure. Also, the delayed SG Pressure peak in plant data (as compared to RETRAN)is j
reflected in a delayed Cold Leg Temperature peak.
It should be noted that although other parameter results for 60 seconds were presented for this test, the principal objective of this test and the reason for performing a RETRAN analysis was to compare the onset and magnitude of natural circulation after a trip from 40% power. RETRAN's calculation of natural circulation showed the onset at about the same time as data (five minutes) and the magnitude (i.e., ratio of natural circulation flow i
1 rate to reactor thermal power)in close agreement with measurements.
- 12. One of the objectives of this BG&E Topical Report is to present BG&E's capability in modeling control systems. Since it is apparent that the major reason for disagreement 4
between TRAC and RETRAN results in the differences in the ADV modeling, BG&E should match the ADV control system used in TRAC and rerun the calculation..
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION BG&E states that in spite of the fact that " primary pressure compares very well, RETRAN predicts the pressurizer to empty while TRAC does not" and that "this is consistent with previous comparison." If this difference is present in the revised calculation, then explain vihy one code predicts the pressurizer to empty and the other does not.
Response
There are two important differences between the results of the RETRAN and TRAC calculations. The first is that RETRAN calculates the pressurizer to empty and TRAC does not. The second is that RETRAN calculates a more restrictive cooldown than TRAC.
RETRAN predicted the pressurizer to empty because charging was not restored to maintain level. The assumption (for RETRAN) was that charging would be permanently realigned to provide for auxiliary pressurizer spray. However, in the TRAC analysis, charging was assumed to be restored for level control. RETRAN was subsequently rerun assuming restoration of charging and the results accurately matched the TRAC Pressurizer Level performance (i.e., the pressurizer does not empty).
The more restrictive cooldown is explained in the BG&E Topical Report. For the purpose of responding to the above question, a less restrictive ADV model (geometry) was substituted for the best-estimate ADV model and consequently, TRAC calculated temperatures were reproduced. It should be noted that no changes to the ADV control system were required to resolve the differences between the TRAC and RETRAN results.
- 13. Provide detailed transient curves of RETRAN and TRAC secondary flows, pressures, and temperatures. Correlate to the transient curves for the secondary side with parameters on the primary side. Explain why the PORV reaches its setpoint about 500 seconds earlier than the TRAC predictions. Explain why RETRAN predicts that the pressurizer water level will rise to the top of the pressurizer 1250 seconds before the TRAC predicted time.
In addressing the differences between results, explain the impact that the noding may have in computing downcomer temperature. It seems unlikely to us that these differences are caused by the coarse RETRAN noding in the downcomer (2 vs 54). Discuss the cross-flow between nodes in the TRAC nodalization, and provide the theoretical foundation for the statements that imply cross-flow 'vould increase with increasing nodalization. In addition, explain why increasing the number of nodes in the secondary side of the SG was not considered as an option to reduce the void formation in the upper tube volumes.. -. -
f ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION
Response
Reference 35 (NUREG/CR-4109) from the BG&E Topical Report only presents TRAC calculated results for SG mass inventory and steamline flow rate for this transient.
Therefore, to provide a comparison of SG secondary parameters, RETRAN calculated results are overlayed on these figures from NUREG/CR-4109 (attached Figure A-13-1).
The RETRAN calculated SG mass inventory closely agrees with TRAC results with the overfed SG inventory rising to about the same peak value and the other SG inventory dropping to the same lower value of inventory. Steamline flow rates also closely agree j
with RETRAN slightly over-predicting peak flow, but following the same transient behavior and reflecting the initial runaway MFW until 303 seconds when MFW pumps trip off.
~
i Cooling due to the high steamline flow rate between about 100 and 400 seconds caused the large Pressurizer Pressure and Level, Hot Leg Temperature and Downcomer Temperature drop during this same time period. Similarly, the rapid drop in steamline flow rate af ter 400 seconds resulted in the later increase of these parameters.
The earlier PORV setpoint pressure and Pressurizer Level rise to the top are, attributed to two factors. First, the TRAC model of Calvert Cliffs did not allow backup pressurizer heaters to reactivitate af ter pressurizer liquid level returned above a programmed level setpoint even though the actual CCNPP control system has this feature. Los Alamos National Laboratory acknowledged this error on page 23 of NUREG/CR-4109. The BG&E RETRAN modal does have this feature which became activated at about 450 seconds for I
this transient. Therefore, the RETRAN simulation included a 1200 kw backup heater input to the preiurizer after 450 seconds which was erroneously omitted from the TRAC model. This additional input increased Pressurizer Pressure.
l The second factor which caused the more rapid RETRAN Pressurizer Pressure and Level rise is the large difference in High Pressure Safety Injection (HPSI) flow rate between RETRAN and TRAC. As shown is Figure 5.3-1 of the BG&E Topical Report, RETRAN calculated Pressurizer Pressure is lower than TRAC Pressurizer Pressure until about 1000 seconds. During this time period, RETRAN Pressurizer Pressure is as much as 300 psi lower than TRAC _ due to the selection of a large value of IHTC. This lower pressure results in a much higher HPSI flow rate until the 1275 psia HPSI shutoff head is reached at about 900 seconds in the RETRAN analysis. The higher HPSI flow rate which occurs over this 800 second time period is responsible for the more rapid Pressurizer Level rise.
Since the TRAC model uses six circumferential cells (one for each reactor vessel nozzle) and nine axial levels as compared to RETRAN's two nodes, a much larger combination of mixing flowpaths are available in the TRAC model. This transient is asymmetric since one loop experiences a much more rapid cooldown and higher flow rates than the other. This difference in loop flow and temperature (Figures 5.3-4 and 5.3-5 of the BG&E Topical Report) will create a significant cross-flow within the downcomer.
Several factors differentiate the cross-flow calculated between TRAC and RETRAN.
TRAC uses a three dimensional flow solution (r,9,z) method as compared to RETRAN's one dimensional flow method. TRACs flow solution method coupled with a larger number of nodes allows greater mixing. It was also apparent in performing this comparison that l
_ ~,
~
1 ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION the TRAC report (NUREG/CR-4109) does not identify which downcomer node temperature is reported. For PTS purposes, a node adjacent to the highest neutron flux core midplane may have been selected which is several levels down from the cold leg nozzle location.
With the finer TRAC nodalization, a downcomer node location several levels below the nozzle allows cross-flow mixing from the adjacent warmer downcomer and cold leg nodes. This detail was not included in the RETRAN model. Finally, the TRAC model allows leak flow from the reactor vessel upper head to the upper downcomer. This flowpath, which would increase downcomer temperature, was not included in the RETRAN model.
Increasing the number of nodes in the secondary side of the SG was not considered because the extremely large computer run time for this transient (greater than 10 cpu hours) greatly restricted the number of sensitivity studies and model changes.
Previous experience with increasing SG secondary nodalization has also resulted in further increases in computer run time. Since the model changes in the SG tube volumes allowed RETRAN to simulate SG reverse heat transfer and upper tube voiding, the cost versus benefits of further SG nodalization studies was not justifiable.
14.
Describe any differences between the BG&E RETRAN model and the RETRAN model used in initial analysis of L6-1 and L6-3 by EPRI as presented to the ACRS on January 14,1981.
Response
As a contractor to EG&G Idaho, Intermountain Technologies Incorporated (ITI) developed pretest RETRAN models for the L-6 series of LOFT tests (Reference (10)). Energy Incorporated (EI), as a contractor to EPRI, took the ITI models and modified them to match actual plant initial and boundary conditions, and then performed post-test analyses. The results were presented to the ACRS by EPRI on January 14,1981. Some further refinements were made to the models by El and the results were printed in the May 1983 issue of Nuclear Technology (Reference (11)).
As stated in the BG&E Topical Report, the model which served as the starting point for BG&E was an L6-5 RETRAN-01 model supplied by EI. This was the original RETRAN LOFT model created for EG&G (documented in Reference (10)). When comparing the BG&E model to the El model, we are really only comparing the modifications BG&E and El l
each made to the original L6-5 model. The models were found to be similar, with only a i
few significant differences. The basic geometry and material detail are nearly identical.
l The only important exception is the inclusion of heat slabs in the BG&E model pressurizer (an option not available to El at the time they modified the ITI models).
The control systems are also similar. Although the ordering of the control cards is somewhat different and some trip actions have slightly different set points, the fundamentals of the control systems are the same. (Additional detail will be given on secondary controls in our answer to question 15.)
l l
1 l. _ _ _
.-=
' ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION Boundary conditions and initialization schemes are siightly different. BG&E and EI both chose to input primary system pressure in the pressurizer (El used 2144 psia and BG&E used 2142 psia). The BG&E model input primary pressure and temperature information into the Hot Leg volume 31; whereas, the EI model input temperature into the Cold Leg volume 46. Both models had secondary pressure and temperature input to the Steam Dome volume 61 (BG&E used 779 psia and 610 F with a mixture level of 2.0 feet, while El used
'779 psia and 600 F with a mixture level of 1.7 feet). No steady state equilibrium outputs were available to BG&E using the EI input. The BG&E method results in a steady null transient equilibrium response.
1 e
- 15. Verify the slip impact on secondary swell level by plotting results with an infinite bubble rise model in the steam generator secondary side. Compare the results from an infinite bubble rise model with. results from a single node secondary side model.
Explain significantly greater depressurization and why the main steam flow control valve (MSFCV) was open 40% longer than the test, and provide the transient curve of experimental and i
predicted flowrates. Describe the control system which governs the MSFCV operation in
'the experiment and how this was simulated by RETRAN. The depressurization computed l by RETRAN was 50-60% larger than the test result, therefore, it appears to be not merely due to 40% longer opening of the MSFCV.
Response
l-The L6-3 transient was rerun without slip and used only the bubble rise models to provide phase separation. The level calculated is presented as Figure A15-1. The liquid level is calculated as an equivalent liquid level for. the composite of volumes 75 and 61. For the slip case, after MSFCV closure, fluid drains rapidly from the steam dome volume 61 to the riser volume 62.. The no-slip case does not allow the fluid to drain properly and so liquid mass is retained in volume 61. The equivalent liquid level calculation finds less mass in the composite volume for the slip case and results in a lower (and more accurate) level calculation. A single node secondary model does not accurately predict the level.
\\
j.
The significantly greater primary depressurization seen in the L6-1 simulation is principally due to the pressurizer modeling and not to any effects from the seconpa The L6-1 pressurizer model uses a' very high value for the IHTC (27500.0 BTU /HR-FT gy.
F) in order to correctly calculate the rising pressure, liquid level, and spray effectiveness.
- As stated in the BG&E Topical Report, this modeling gives good results for rapid insurge transients. Unfortunately, the L6-1 transient has a rapid pressurizer insurge followed by an equally ryid outsurge. For rapid outsurge transients, small values of IHTC (e.g.,400.0 BTU /HR-FT - F) are more appropriate. RETRAN does not allow a varying IHTC.
The large value of IHTC used to properly calculate the insurge conditions caused excessive energy transfer from the steam space. This resulted in a lower pressure which caused too much pressurizer liquid to flash during the depressurization/outsurge portion of the transient. This resulted in a greatly depressed Pressurizer Level and Pressure. Thy effect was verified by rerunning the L6-1 transient with a value of 10.0 BTU /HR-FT - F for IHTC. Pressurizer Pressure and Level results have been plotted and are attached as Figures A15-2 and A15-3. As shown in these figures, pressure and (to a lesser extent) level more closely track plant data during the outsurge portion of the transient than the original -
j ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION I
2 calculation which used an IHTC of 27500 BTU /HR-FT _oF.
However, with the lower -
values of IHTC, pressure and level are poorly modeled during the insurge, causing a reactor trip on high pressure to occur at 11.6 seconds (about 10 seconds earlier than actually seen).
The model used for the BG&E Topical Report represents a compromise which, because of its importance, was biased toward accurately predicting the time of reactor trip. A model using a modified MSFCV control system was run to establish the effect of the 40% longer opening of the MSFCV on the depressurization.
The result indicated no change in i
. calculated pressure when the difference in MSFCV open time was reduced by 50%
The principal element of the LOFT secondary side model is the MSFCV. The MSFCV
- operates automatically to provide SG overpressure protection and to limit primary system cooldown after a reactor trip. The valve opens and closes at a stem rate of 5% per second. The valve will begin to close should pressure drop below 920 psig and a reactor I
trip has occurred. The valve stops closing when pressure climbs to 930 psig. If the steam pressure reaches or exceeds 102': psig, the valve will start opening. The valve will stop opening when pressure drops below 1010 psig. The control system used in the BG&E RETRAN LOFT model accurately reproduces this design behavior.
Stem position is converted to flow area using data supplied in the base L6-5 deck. This relationship is also shown in Table 2 of Reference (10).
As stated, the MSFCV position is driven by a simple open or stop-open, close or stop-close i
logic based on reaching certain secondary pressure setpoints. The data reported in our BG&E Topical Report was based on using design setpoints from all our control systems. In a
many cases, the use of a setpoint slightly lower or higher than the design value, but still 4
within the range of instrument or measurement uncertainty, will result in substantial L
differences in the sequence and timing of events.
The sensitivity of the LOFT results to variations in setpoints was not examined for the 1
l BG&E Topical Report. However, by carefully examining the figures for MSFCV valve position and secondary pressure as given in BG&E Topical Report Reference 40, doubt is cast on the design values for MSFCV operation. It appears that the " actual" setpoints were as follows: start opening 1017 psia, stop opening 984 psia, start closing 958 psia, stop closing 966 psia. L6-1 was rerun with these setpoints and secondary pressure is shown in 4
i Figure A15-4. Since all the setpoints are for lower pressures, the sequence of MSFCV actions have been accelerated in time.
4 Although no further sensitivities were examined, we believe that RETRAN would better s
reproduce the LOFT secondary pressure curves by using setpoints between the original design values and the above " actual" values. The best result would be obtained by lowering the start opening setpoint sufficiently to show the reopening of the MSFCV seen in the test data. This is what El did for their LOFT analyses presented in Nuclear Technology (attached Reference (11)). As requested, the mass flow rate through the MSFCV is plotted against test data for the original L6-1 analysis and is shown in Figure A15-5.
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION
- 16. (LOFT-6-3) Modeling of the secondary side appears to be inadequate (see Figure 6.4-1,2, and 5), causing primary side behaviors to differ from the test data. Explain and justify the secondary side modeling including a thorough explanation of why RETRAN calculates "somewhat higher terminal steam flow and feedwater flow."
Response
RETRAN very accurately calculates Hot Leg Temperature as seen in Figure A16-1. BG&E Topical Report Figure 6.4-5 gives an unrealistic impression of the accuracy of the RETRAN calculated L6-3 primary temperatures. Steam generator primary side inlet plenum temperature was presented because it was the first primary side temperature presented in the LOFT data book (BG&E Topical Report Reference 40). There are many possible influences on the SG inlet plenum temperature which could cause it to lag behind Hot Leg Temperature. The placement of the measuring instrument in the LOFT facility is one example.
Another possibility is the local effects due to the proximity to the downcomer and tube regions.
Pressurizer Press.ce was improved by adding the TTDT option (20 mesh intervals). The revised Pressurizer Pressure calculation is shown in Figure A16-2. Given the accurate calculation of Hot Leg Temperature and Pressurizer Pressure, we conclude that the RETRAN model adequately calculates primary system parameters.
The L6-3 SG level is also predicted quite well. The BG&E Topical Report Figure 6.4-2 was plotted by hand from RETRAN output data. The spike at 50 seconds is explained in detail in the BG&E Topical Report.
The figure exaggerates the spread in the data by representing it as a crisp line that swings down and up. A more accurate representation of the data has been plotted and presented as Figure A16-3. As can be seen, there are many possible curve fits to this data.
In the BG&E Topical Report calculation for L6-3, MFW and steam flow were allowed to attain the highest possible terminal value indicated in the LOFT data report (Figures 35-3 and 5C-1 of BG&E Topical Report Reference 40) - about 4 lbm/s of feedwater and 1.5 i
Ibm /s steam at 200 seconds. It was recognized af ter the calculation that this feed flow value may not have been realistic since the feed pump discharge pressure (Figure SS-4 of Reference 40) was given as zero. L6-3 was rerun using a more probable 0.5 lbm/s feedflow with results indicating no change in system response. This was expected since steam flow controls secondary pressure and primary to secondary heat transfer. The true steam flow is unknown since the flow indication seems to have pegged low and the MSFCV position (Figure SS-1) is obscured by its uncertainty band. Rather than using an iterative process l
to find an MSFCV position which will give an exact reproduction of LOFT results, we j
simply state that reduced steam flow in the RETRAN calculation would increase the rate l
of secondary repressurization, better matching the LOFT data.
I 1. - - - -
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION
- 17. With respect to BG&E general analysis of the LOFT tests, BG&E selected two transients that were initiated on the secondary side. Since BG&E did not accurately model the secondary events, the primary response may be incorrectly computed. Develop secondary controls and input which accurately represent the transients conducted in the LOFT test.
Rerun these computations to obtain a valid comparison of RETRAN results with LOFT test data.
Response
The answers to questions 15 and 16 include discussions of additional sensitivity runs for both transients L6-1 and L6-3. Any improvements in calculating pressures, temperatures, or levels were presented in the answers to those questions. Attached as Tables A17-1 and A17-2 are revised sequences of events for the LOFT transients which were generated during the aforementioned sensitivity runs. Improvements to those sequences of events can be attributed to minor changes in the RETRAN control setpoints l
l i
{
l l
l I
1 L
ATTACHMENT RETRAN REVIEW - ADDITIONAL INFORMATION REFERENCES 1.
Letter from Mr. C. O. Thomas (NRC), to Mr. T. W. Schnatz (UGRA), dated September 4, 1984, Acceptance for Referencing of Licensing Topical Report EPRI CCM-5, "RETRAN -
A Program for One Dimensional Transient Thermal-Hydraulic Analysis of Complex Fluid Flow Systems" 2.
Letter from Mr. L. 3. Agee (EPRI), to Mr. C. O. Thomas (NRC), dated December 20,1984 3.
Letter from Mr. T. W. Schnatz (UGRA), to Mr. C. O. Thomas (NRC), dated February 4, 1985, RETRAN - A Program for Transient Thermal-Hydraulic Analysis of Complex Fluid Systems 4.
Lance 3. Agee (EPRI), "RETRAN Overview," Nuclear Technology, January 1987, Vol. 76.
5.
Thomas L. Temple (Middle South Services), "RETRAN Generic Review - A Retrospection,"
Nuclear Technology, January 1987, Vol. 76 6.
BG&E Topical Report A-85-11, RETRAN Computer Code Reactor System Transient Analysis Model Qualification, T. L.
Cook, S.
M. Mirsky, and G. R. Wisniewski, January 31,1986 7.
N. A. Smith, VEP-FRD-41, "VEPCO Reactor System Transient Analyses using the RETRAN Computer Code," March 1981.
8.
Letter to W. L. Stewart (VEPCO) from Cecil O. Thomas (NRC), dated April 11, 1985, Acceptance for Referencing of Licensing Topical Report VEP-FRD-41 9.
I. E. idel'Chik, Handbook of Hydraulic Resistance, AEC-TR-6630,1960 10.
EG&G Idaho, EGG-LOFT-5161; Best Estimate Predictions for LOFT Nuclear Experiments L6-1, L6-2, L6-3, and L6 Experiment Prediction Analysis Report, October 31, 1980 11.
Miller, et al "RETRAN-02 Calculations of Operational Transients in the Loss-of-Fluid Test Facility," Nuclear Technology, May 1983, pp.181-192 i
f
_._l
Table A-17-1 Sequence of Events for Experiement L6-1*
Time After Experiment Initiation RETRAN lime RETRAN Time Event seconds (seconds)
Sensitivity Run MSFCV Closing Initiated 0.0 0.0 0.0 Pressurizer Backup Heaters Off 6.1 + 0.1 7.07 7.07 Pressurzer Spray On 9.1 1 0.1 9.47 9.47 MSFCV Closed 11.6 1 0.2 11.61 11.61 Reactor Scrammed 21.8 1. C. 2 17.67 18.8 Maximum PCS Pressure Reached 22.0 1. 0. 2 18 19.0 MSFCV Opened 22.2 1 0.2 18.61 22.1 0
Q 26.5 +
SG core
- 0.2 25 27.5 Pressurizer Spray Off 30.4 + 0.2 21.46 21.46 Pressurizer Backup Heaters On 32.5 1 0.1 23.8 23.8 MSFCV Closed 40.6 + 0.2 43.23 42.0 MSFCV Opened 91.2 + 0.2 MSFCV Closed 104.4 + 0.2 Only the first 200 seconds are presented.
Pressurizer liquid level at maximum.
RETRAN did not predict this event.
?
Table A-17-2 Sequence of Events for Experiement L6-3*
RETRAN Time Event LOFT Time RETRAN Time Sensitivity Run MSFCV Opening 0.0 0.0 0.0 Feedwater Flow Increased 1.4 12 0.0**
0.0 Pressurizer 8ackup Heaters On 10.2 1 1 0.0 0.0 Maximum Reactor Power 15.6 1 2 14.18 15.53 Reactor Scrammed 15.6 i.2 14.18 15.53 Feedwater Fl ow Terminated 16.6 1 2 15.0 16.6 MSFCV Start Closing 17.8 12 15.0 15.53 HPS! Pump A On 26.4 1 2 18.5 24.13 HPS! Pump B On 26.6 1 2 18.5 24.13 Minimum PCS Pressure Reached 26.8 12 30.5 31.0 0
0 33 +
SG core
- 1 32.25 33.0 MSFCV Closed 36.2 12 34.00 36.0 HPSI Pump A Off 48.6 1,.2 44.14 48.15 HPSI Pump B Off 50 1 2 44.14 48.15 Pressurizer Backup Heaters Off 105.4 11 Pressurizer Cycling Heaters Off 154.9 1 1 Only events occurring in the first 200 seconds of the LOFT Test are presented.
RETRAN Feedwater was delivered by a fill junction designed to match the LOFT data.
Pressurizer 11guld level at minimum.
RETRAN did not predict this event.
em a - * -
amee..
t 9
E
/
f e
e@
o 0
88ef
@s
,I
- e.8 o e
E e -
- g z$
i
=c=tlWe@s fl!
<<<5=e g
o V
/
J q
s I
O
-- h o aG
-l
/
g 90 ll f
O 4A
- I U
7 8
l R*
g
/
1 DD
/
I w
Sk
/
/
W e
/
/
3 E
. /,/
/
z
.i
/
/
/
-l 8
'/
p i
.{
//
E I
i s
o a
=
\\'
C
\\
./
'N 3
\\
i m
m e
e e
m m
(a.)annivvaawal acol s3s - _
l 1
i l
1 I
FIGURE A-IO-2 l
ONE PUNP COASTDOWN PRESSURIZER PRESSURE RESPONSE i
(
2500 i
PLANT DATA 2400 RETRAN 2300
__ RETRAN REVISED 4
-m ADV/TBV/TSV
'm-4 5' ', N I
2200 E
' ~ s. 5. % '. h 5:
2100 g
i m,~% m. _
2200 1900 l
0 10 20 30 40 50 60 TIME (SECONDS)-
)
i i
l i
l 1
1 i
t J
1
1 l
FIGURE A-10-3 ONE PUHP COASTDOWN PRESSURIZER LEVEL RESPONSE
.1 300 PLANT DATA 250 RETRAN
^
RETRAN REVISED i
l0
^*/*/*
200 z
.s %.%.D 3
v
~ ~ _ '
1 z
l, C
150
~~~..m.---.-._
a i
m
~~___
- _____ _.2 100 g
4 50 l
0 0
10 20 30 40 50 60 IlME (SECONDS) l
FIGURE A.IO-4 ONE PUNP COASTDOWN STEAM CENERATOR SECONDARY SIDE PRESSURE RESPONSE 950 PLANT DATA RETRAN 925 -
s
' N.\\
RETRAN REVISED b
\\}e%
G 900
._._.__w__'~~~_
5 ((
l N=
gL.
Y
=,
875
=
o e
m M
Ie i
850 i
/
j 825 >*
800 0
10 20 30 40 50 60 TIME (SECONDS)
FIGURE A-13-1 25 coco 50000o 180P A 20000o-
-- Laer e 6-
- --- RETRAN
-*o000 i
s j
s's 15 coco-I n
8 I,
-300000 $
j NOTE: These transients l
2 assumed multiple 2
20000-operator / equipment f allures. See TABLE I.
-200000 i
y
[
.,,,,,~'..-
..gogog sooog.
o o
o moo 2000 J000 4000 5000 sooo 7000 8000 Trne (s)
Fig. B.9.
Steam generator mass inventory--Runaway MFW to one SG from FP.
1000 goo. n
. -200o l
soo-.
-.oso Il
RETRAN 700-.
0
-.soo l
{
(I 11
~
8"-
6 e
3 I
NOTE: These transients
~
8"~
d l
ossumed multiple a
l operator / eculpment d
- M-j g;
j f ailures. See TABLE I.
.mo Joo-l I
l l
- -soo 2M-l f
J mo.
.-250
,i.
L k-g.t..............
o.
wg
..e l
V v l
-mo i
o Soo 100o 1600 2000 2500 J000 3500 4000 4600 5000 5600 000o
(
h (8)
Fig. B.10.
Steamline mass flow-Runaway MFW to one SG from FP.
245
a
<----e
---m
-a a
O
~'
FIGURE A-15-l LOFT TEST L6-3 LIQUID LEVEL IN STEAM GENERATOR SECONDARY SIDE o.5
,3 i\\
.i i WIDE RANGE RETRAN l
1 1
Oooo NO SLIP
.I 1
l 6
0.0 I
}
o
\\
G
\\.
\\
C t
'J 9
3
\\.
-1 a
,}
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0-s i
3,
,3 j
Y
\\
l
-0.5 a
I 1
I 1
-2 li i
I
.i k.
l
\\. \\ '\\
N l}'
\\.~.-
l
--____....:- -3 l
l
-1.0 l
-so o
50 100 150 200 Time After Initiation (s) l l.
- - - - - - - - - - - - - - - - - - - ' - - ^ ~ - - - - ~ ' ' '
FIGURE A-15-2 LOFT TEST l.6 - 1 PRESSURE IN PRESSURIZER 16.0
,\\
TEST DATA 2300 I
l RETRAN l\\
i i
i.o l
4.
AT 100 SECONDS
'l Il i l[ gj 2250 15.5 Ii fi
}
hih I I 11 2200 i
!)
i
\\\\\\
15.0 j
)'
il a
b,,A t.# k b
.m.
Il 2150
-T,.-
}
e.
2 it
!l E
E l
2 1
2 II
'~
1 14.5 2100 1
I l
?
2050 I
/
j 14.0
././
\\-\\.
,.s s
I.
/
2000
\\.,,_
. '",././
\\
y 13.5 e
t 1930
-50 0
50 100 15 0 200 Time After initiction (s)
- 118 -
l
FIGURE MS-3 LOFT TEST L6-1 LIQUID LEVEL IN TRESSURIZER 1.5 TEST DATA 4.s i
RETRAN Il lHTC 10.0
[l TRANSIENT STOPFE
.i ATl00 SECONDS fI 4.6 1.4 1
I i.
I E
I O
4.4
}
I 3
5-p
\\
O
. }
l)\\ \\
I 1.3 a.,
I I
1 z
i \\.
1, I; I
I
\\j s.0
\\l
.I y
1 1.2
, ~
s.
I 3.s 1.1 l
-50 0
50 100 150 200 Time Af ter initiation (s) 119 -
l t-
FIGURE A 15 4' LOFT TEST L6-1 PRESSURE IN STEAM GENERATOR DOME a
TEST DATA RETRAN
~
ACTUAL SECONDARY SETPOINTS USED
- ONLY 100 SECONDS-I t th 1
l
\\
7 l
s 1000 g
/e','s /
I
.\\ ll
[
J m
g i
/
o 1,//
950 E.
e j
10 3
I I
3 E
[
900 I
.j e
i
)
1 l
i e
i e
3
-50 0
50 100 15 0 200 Time After Initiation (s)
- 116 -
l l
l _.-.
. ~....
L6-1 20 44.27 FIGURE A-15-5 j
l l
I I
I I
I I
I I
I I
I I
RI F T-P004 -012 ---- torr
_ _ 40 RETRAN l
_ p_ _
Range
=0.
to 40.00 (kg/sl Uncert
=
.800 (kg/si 15 I.
~
s' l
_T,_
m u
i 3
s I
- 30 t_
1 O
x r
E
~
l
\\
2 I
I 10 g
I
- j-r-
i i
2 I
20 o
r.
,i_
.a l__
m p
w
,L u
i
<r y
i r
L m
i
--- p 1"
5 1
i E
.j_
- 10 1
~
I I
t
..j-
. fi, _ _ _
y
.g -
i O
d
=-rv w 3
% g
% %- 23.
}
-50 0
50 100 150 200 0
t l
TIME AFTER INITIATION is) 4
)
Figure 3S-4 Mass flow rate at condenser inlet (FT-P004-0121 i
,fqualified).
I 1
}
1 4
30 4
i l
I i
I
%~
(doi 3dn1VB3dW31 INVlOOJ o
o o
o o
e to n
m m
e to to to to I
l I
l l'
O
~
o I
c L
I I
m Y
I i
4 I
g g
~
kE Q
l
~
ta i
e a
i o
-v
~
m o
on jogj o I
i s i
I I
e
.c o I
J l
I I
bg8J -
a.
"-x l
I L
i I
om "i
e_
o l
I l
l I
i l
l l
oo i
i i
g i
i i
o
~
i
-o o
-m-l LO l
l I
i@
l i
i tD um0 i
50 o O
I i
i j.is i z
~'"
Ny mmm I
i A.
I a
ce-
._ n e oom-1, 1 o,<
C CL -
i i to o i
il i
i OC" l
l 1
o~
mm
~
--e't l
l Il I
i V
g
~
9 00- O@
l l
l
)
l eH m h' ~~EEk$
I I
I O
I i
$~o 2
,LJ l
l l
/g l
1
- u Q.
"C2 c
m hc l
I i
l l
b i
I w
LT c e--
oct
<O i
i i
it i i
n u.
- c. o I
i
+1i e
e l
ti l
l r
I i
i i
i
~
e W
l l
1 l
l l
l 8 i E
~
~~"
l l
l l
l I
l l
( ).
I o ~,-
a u e-oc-5 1
I I
i i
i h\\
l m
-cm I
hI oea w
oC CT E
i l
1%
s uu-l l
l T
a o
i o
N l
h I
l T
e l
l e
I I
D i
WstEEL
+m
_r=
1 e
7 o
a i
l T
I I
I I
e o
o o
o I
o e
tn v
u.
to to to to
( 'A i 38niv83dW31 INv1003 1
'-T ee--e.
FIGURE A-16-2 LOFT TEST L6-3 PRESSURE IN PRE 55URIZER 16 TEST DATA
-- -- RETRAN
TEMR TRANSPORT DELAY MODEL 2200 15
_.m s
g sy//,.
8 1
n s
4k
~
2100 e
//
2 2
f/
3
//
e I
14 g;
\\\\
il
\\
il 2000
'k t) 1900 13
-50 0
30 100 150 200 Time After Initiotion (s)
- 130 -
FIGURE A-16-3 1.0FT TEST L6-3 3
LIQUID LEVEL IN STEAM GENERATOR SECONDARY SIDE 0.5
,a i\\
ji WIDE RANGE RETRAN
=--
.I l
AAaa OATA POINTS 1
I 1
.I i.
I
\\
i 0.0 I
0 g
E g
g
\\
~
t 7>
1 j{
\\
1 3
m g
3 l
a s
I
- 0.5 i.
I
-2 ap.&
g 7
^%"..
^
--_ -_.._..._. : -3
- 1.0 l
-50 0
50 100 150 200 Time After Initiation (s) l "o -
.