ML20028A823

Single Loop Natural Circulation Cooldown

ML20028A823
Person / Time
Site:	Midland
Issue date:	08/31/1982
From:	Carlton J, Tally C BABCOCK & WILCOX CO.
To:
Shared Package
ML20028A821	List: ML20028A820 ML20028A823
References
NUDOCS 8211290086
Download: ML20028A823 (70)

Text

{{#Wiki_filter:_ _ _ I I I I I I SINGLE LOOP NATURAL CIRCULATION C00LDOWN Prepared for CONSUMERS POWER COMPANY I E Midland 1 and Midland 2 Projects l l. l I 'I 8211290086 821029 PDR ADOCK 05000329 E PDR

I I . I SINGLE LOOP NATURAL CIRCULATION C00LDOWN . I Prepared for I CONSUMERS POWER COMPANY Midland 1 and Midland 2 Projects Prepared by: b C.W. Tally 6 Reviewed by: J Carlton lI l l l BABC0CK & WILC0X Nuclear Power Group Nuclear Power Generation Division P.O. Box 1260 Lynchburg, Virginia 24505 I I I

I I I Babcock & Wilcox Nuclear Power Group Nuclear Power Generation Division Lynchburg, Virginia Report BAW-August 1982 Single Loop Natural Circulation Cooldown C. W. Tally Key Words: Cooldown, Natural Circulation, Single Loop, Loop Vents, Boron Mixing I EXECUTIVE

SUMMARY

/ ABSTRACT This report addresses natural circulation cooldown with an isolated 0TSG. Reactor vessel (RV) head and idle loop cooldown rates were calculated to be approximately 2*F per hour with the RV head being the limiting factor for depressurizing the I

RCS. Analysis results suggest that the idle loop can be properly cooled and borated during the RCS cooldown. Three new functional capabilities were identified as major enhancements for plant control during natural circulation. They are: e RV Sead venting e RV head temperature measurement e RCS loop boron measurement l An RV head vent could pennit faster cooldown of the head by venting water and would provide a positive method for eliminating a steam bubble if it should occur. An RV head temperature measurement would provide a direct indication of the saturation conditions in the head. This would be valuable for preventing steam bubble formation. l

I I CONTENTS Page I 1. INTRODUCTION........................... 1-1 I 1.1. Background......................... 1-1 1.2. Scope of Work 1-2 1.3. Sumary and Conclusions 1-3 2. ANALVSES............................. 2-1 2.1. Introduction........................ 2-1 I 2.1.1. Bubble Avoidance..................... 2-1 2.1.2. Bubble Mitigation 2-2 2.1.3. Cooldown Limits 2-3 I 2.2. Analyti cal Methods..................... 2-3 2.2.1. Reactor Vessel Head Cooldown............... 2-3 2.2.2. RCS Loop Cooldown 2-3 I 2.2.3. NSS Simulation...................... 2-4 2-4 2.3. Results of Analyses 2.3.1. Reactor Vessel Head Cooldown............... 2-4 2.3.2. RCS Loop Cooldown 2-6 I 2.3.3. Boron Mixing....................... 2-7 2.3.4. RCS Perfonnance During Single Loop Natural Circulation.. 2-9 l 3. RECOMMENDATIONS 3-1 3.1. Operational Considerations................. 3-1 3.2. General Guidelines..................... 3-5 3.3. Plant Modifications 3-7 4. REFERENCES............................ 4-1 APPENDIX A - Reactor Vessel Head Cooldown Analysis Method A-1 l APPENDIX B - RCS Idle Loop Cooldown Methods B-1 APPENDIX C - NSSS Simulation Modeling C-1 I -I i I - iii -

I I I 1. INTRODUCTION j This report addresses the operational aspects which may occur during a natural circulation cooldown of the reactor coolant system (RCS), focusing on those problems peculiar to single loop operation. It is intended to provide infomation needed to develop operator guidelines for natural circulation which will prevent famation of RCS steam voids or will pemit their timely recognition and control. 1.1.

Background

A vital element in nuclear safety is the maintenance of long term core cooling. The ideal decay heat removal mechanism involves heat transfer from the core to the reactor coolant which is then transported to the steam generators (OTSGs). The heat is transferred across the OTSG tubing to secondary fluid which is flashed to steam and ultimately condensed for recycling as feedwater. Many plant conditions can arise to interrupt this sequence of events or complicate its continuity. Since the objective of this whole evaluation was to be recor:rnendations for operator guidelines, the scope of work could be roughly translated into a series of questions concerning operator action. They are: e How does the operator avoid foming either a RV head bubble or a loop l bubble while progressing through a natural circulation cooldown and l depressurization? e How does the operator know if a bubble exists in the RCS (either loop or RV head)? e If a loop or RV head bubble exists, can the operi. tor still proceed with the cooldown? How should the bubble (s) be eliminated and how long will it take? e e Are there any operational problems associated with bubble removal? If so, how should they be handled? I I l 1-1

I These questions all relate to bubble avoidance or mitigation. Another key area I of concern is reactivity control, especially as it relates to boric acid control in an idle RCS loop. The key questions here are: o Can the idle loop be adequately borated? e If so, how must the plant be operated h achieve proper boration? 1.2. Scope of Work The scope of work included the following items: e Establish a recommended cooldown rate for the single loop natural circulation condition such tat a reactor vessel head bubble will not be formed. o Evaluate loop bubble proolems associated with natural circulation cooldown and depressurization. e Evaluate reactor vessel head bubble problems associated with natural circulation and depressurization. e Evaluate idle RCS loop ambient losses. e Detennine long tenn reactivity requirements and control schemes for boric acid control for long t,enn cooldown. s Assimilate the above aspects of a natural circulation cooldown into an integrated set of recomendations which can be used to develop operator guidelines. Although several of these items represent stand alone analyses, the last one integrates them into a comprehensive evaluation. This evaluation explores the basic voiding ohenomena and provides a basis for meaningful operational guidelines. Specifically, excluded from this scope of work were: e RV head vent (s) (not presently included in the Midland design) l e The impact of loop venting discharges to the reactor building e Specific RV head or RCS loop level measurements for various plant conditions e Prediction of total RCS cooldown ti,mes given various plant conditions e The transition of core cooling from the OTSGs to the decay heat removal system. 1-2

1.3. Sumary and Conclusions The following conclusions can be drawn from this evaluation: e The calculated RV head cooldown rate is between 1.5 and 2*F per hour when all reactor coolant pumps are off. e The calculated average cooldown rate for the idle loop during natural circulation is approximately 2*F per hour, e An RV head bubble has no apparent effect on loop natural circulation flow. (Allowing the bubble to expand into the hot legs was not considered.) e The idle loop is sensitive to flow changes in the good loop, and can l be controlled to some degree to enhance idle loop cooldown and boration. e The time required to reach the decay heat system cut-in point is expected to be between 84 and 130 hours assuming no RV head bubble is allowed to form. This evaluation shows that two loop cold safe shutdown procedures (Reference 4) can be applied to single loop cooldowns if suitable limitations are applied. These limitations involve the following factors: l e Bubble formation in the idle loop e Bubble formation in the RV head e Boric acid mixing between the good loop and the idle loop e Tube to shell t.T in the idle loop OTSG A logic flow chart for controlling a single loop natural circulation cooldown was dweloped (Section 3.2). This evaluation also demonstrates the value of three plant modifications which would enhance the operator's capability to control the plant during a single loop natural circulation cooldown. The first two would be valuable during two loop natural circulation cooldowns. The three are: o Addition of an RV head vent e Addition of an RV head temperature measurement 1-3

e Addition of boron sampling capability from the RCS loops, especially from the hot leg high points or OTSG plena. These modifications are discussed in Section 3.3. I I I ' I 'f I 1-4

2. ANALYSES 2.1. Introduction A variety of analytical techniques was applied to complete the scope of work. These methods are described in Section 2.2. RELAP 4 Mod 6 was used for predicting overall RCS response. Hand calculations were employed as necessary to supplement computer results. 2.1.1. Bubble Avoidance Bubble fonnation can occur in the RV head and in either or both hot legs. Analyses were perfonned to detennine the max'.num allowable cooldown/depressurization rate of the coolant in the RV upper head and in an idle loop during a natural circulation cooldown following full power operation without forming an RV head bubble. All of the mechanisms described below were considered. Cooldown/Depressurization Rate: Voids will not fonn until cooolant in any one region reaches saturation conditions. Natural Circulation Flow: Natural circulation flow will keep the loop in close comunication with the core. Thus, as the cooldown proceeds a naturally circulating loop will track core temperatures. However, a stagnant loop has no convective mechanism for cooling and its temperatures can be expected to diverge from core temperature as the cooldown proceeds. In natural circulation only a small amount of flow reaches the upper side of the RV plenum cover. This flow is at low velocities and does not penetrate into the uppennost part of the RV head dome. The top part of the dome is stagnant and will not track loop temperature during a cooldown. ~ Heat Transfer: Heat transfer in the RV head is from the upper head fluid and metal to the cooler fluid at the top of the plenum cover, the control rod drive (CRD) nozzles, the column weldments, and other cooler surroundings. In an idle loop 2-1

heat transfer is from the coolant to the piping and out to the surroundings through the insulation. The steam generator shells and contained steam effectively insulate the OTSG tubing. Forced RCP Flow: The RV upper head region is approximately equal to the hot leg temperature during forced flow conditions in the RCS. Operating RCPs after a I reactor trip will reduce the RV upper head temperature from the full power hot leg temperature down toward the lower post-trip hot leg temperature. Venting: High point venting from a stagnant region, either in an idle loop or RV head, adds a strong convection heat loss factor which greatly accelerates that region's cooldown rate. (Note: The Midland Units do not have RC head venting capability at this time.) Hot stagnant water is replaced with cooler subcooled fluid, not only removing the hotter fluid, but providing an additional mechanism for removing the latent heat of the piping and other components. 2.1.2. Bubble Mitigation Several mechanisms can mitigate a steam bubble should it occur: Heat Transfer: The bubble will eventually cool and condense. This is a very slow process because of the mass of the affected components and the insulation surrounding them. In addition the steam to wall heat transfer coefficient is small. Mixing: The bubble can be condensed by mixing with subcooled coolant. In natural circulation this is a very small factor. Loop Venting: The bubble may be vented directly from the high point, allowing it to be replaced by subcooled coolant. Condensing: Raising RCS pressure to accelerate bubble condensation is an ineffective mechanism, theoretically impossible in an adiabatic system. Although neither an idle loop nor the RV head region is truly adiabatic, the analyses show the heat losses to be low enough that each region may be considered adiabatic. The analyses demonstrate that all other factors are insignificant compared to direct venting. i I 2-2

~I l 2.1.3. Cooldown Limits Several factors limit the allowable pressure-temperature envelope of the RCS l during cooldown: e RV brittle fracture e Fuel compression e Subcooled margin e RCP seal staging e RCP NPSH e OTSG tube to shell delta T e Decay heat removal system maximum pressure Each of these must be considered when fomulating a strategy for achieving an orderly transition from post-trip hot conditions to cold shutdown. No time constraint has been put on single loop cooldown evolutions by regulatory requirements to date. Thus, no minimum time to reach the decay heat system cut-in point was assumed. 2.2. Analytical Methods 2.2.1. Reactor Vessel Head Cooldown The cooldown of the RV upper head under natural circulation conditions is a result of both heat and mass transfer from the upper head. Heat transfer from the RV upper head is accomplished by a continuation of conduction, convection, and radiation. These factors were considered in a finite difference model constructed to dynamically simulate the heat losses from the RC head fluid and metal. Details of this method are provided in Appendix A. 2.2.2. RCS Loop Cooldown The cooldown rates of the coolant in the idle loop hot leg and 0TSG upper plenum were evaluated using one-dimensional finite difference methods. Steady state hand calculations were applied to the reactor coolant pump casings and the OTSG tubes to estimate heat losses from these components. ' Details of these calculational methods are provided in Appendix B. I 2-3

I 2.2.3. NSSS Simulation The Midland single loop natural circulation analyses used a RELAP 4/ Mod 6 Version 13.1 model. A symmetrical two-loop model arrangement was assembled, composed of 46 control volumes and appropriate flow junctions and heat transfer slabs. The model included the following basic features: l e Two RCS loops e Pressurizer e RV head region e One RCP and cold leg per loop e RV vent valves t e Loop high point vents e High pressure injection and make up control on level e AFW injection at the top of each OTSG e Main steam safety and atmospheric dump valves Details of this model are provided in Appendix C. 2.3. Results of Analyses This section is divided into four subsections addressing the following areas: 2.3.1. Reactor Vessel Head Cooldown 2.3.2. RCS Loop Cooldown 2.3.3. Boron Mixing 2.3.4. RCS Performance furing Single Loop Natural Circulation The evaluation of these results is addressed in Section 3. Reactor Ve'sel Head Cooldown 2.3.1. s The results of the reactor vessel head cooldown analysis are shown in Figure 2-1. The temperatures of the hottest RV head coolant node and the hot leg coolant are shown as a function of time. The maximum coolant cooldown rate in the RV head is 1.70*F/hr. while the primary coolant cooldown rate is 100*F/hr. The analysis assumed the following: I 2-4 I

I e Initial coolant temperature GT 1. < 4 e Initial shell temperature of 604*F e Ambient temperature of 120*F i e Natural circulation flow of about 3% normal flow I The head coolant temperature initially increases due.to the temperature gradient between the shell and the coolant. Therefore, the maximum coolant temperature I occurs at nodes adjacent to the head metal. This is evident in Figure 2-2, which shows the approximate temperature profiles two hours after the cooldown starts. As the cooldown continues, heat is removed from the coolant via ambient losses through the head and by convection as coolint flows through and over the plenum cover. The coolant in the center of the head remains stagnant and loses heat very slow'y. This region will then be hotter than the coolant above it. This is shown in Figure 2-3, which shows the temperature profiles at seven hours. This stratification is due to the model's lack of a free convection model between these nodes. Realistically, natural circulation would occur internal to the I head region. This mixing would increase the cooldown rate as the cooldown progresses. Nevertheless, the maximum calculated cooldown rate of 1.7*F/hr. (which occurs at the beginning of the cooldown) would not change. The analysis assumed that flow up through the plenum ccver affected only the first layer of nodes above' the cover. These nodes cooldown at the same rate as the circulating coolant (100*F/hr.). Hence, if flow were to extend farther above the plenum cover the cooldcwn rate would increase. However, there is no data or evidence to suggest that additional penetration would occur during natural circulation. Flow velocities into the upper head region from below the plenum cover are expected to b less than two feet per second and could not cause appreciable penetration up into the dome region. It should be pointed out that the dome region is a large volume, approximately 458 ft.3, with a plenum cover to dome top distance of about Si feet. Even if I twice the penetration had beer a strad, the results would not change dramatically. a,a n with ~ simple independent hand calculations The model's results wer-which detennined the in 4.ai down rate with the head at 585*F. The agree-ment was good, confinning the model's general accuracy. I 2-5

These results can not be directly supported nor refuted by field data. No B&W I plant has ever perfomed a natural circulation cooldown. In addition, the necessary instrumentation to measure the RV head cooldown rate is not presently installed at any site. In summary the model demonstrates clearly that the RV head region will not cooldown at any appreciable rate with the plant in natural circulation. The model has included all significant and many secondary heat loss mechanisms with the exception of free convection between nodes in the RV head as the cooldown progresses. 2.3.2. TCS Looo Cooldown The original objective of this analysis was to detemine the average cooldown rate of the primary coolant in an idle loop. Because of the complexities involved in calculating the cooldown rate of the coolant in the steam generator tube region and in the reactor coolant pumps, an exact cooldown rate for the entire idle loop was not detemined. However, the cooldown rates of the coolant in the RCS piping (hot leg) and 0TSG head region were detemined using steady-state heat loss calculation results. These steady-state results are maximum cooldown rates. The cooldown rates in these four regions have been used to draw conclusions about the average cooldown rate of an idle loop. The results of the hot leg piping coolant and OTSG head coolant cooldown rates are summarized below. Without Air Gap With t" Air Gao Cooldown Rate of Hot Leg Coolant 1.7 F/hr. 1.3 F/hr. Cooldown Rate of OTSG Head Coolant

• 0.3 F/hr.

0.25'F/hr. In the analytical method section it was stated that the cooldown rates were to be found using various internal convective correlations. However, the analysis showed that this themal resistance is small compared to the total resistance. Therefore, the cooldown rates are independent of the internal resistance. I Internal natural recirculation flow was not modeled. Therefore, the cooldown rates of the hot leg and 0TSG head were due to radial heat transfer only, and

• Note that this is the cooldown rate of the head at its greatest radius. The cooldown rate will increase from this rate to the hot leg coolant cooldown I

rate as the radius of the head decreases. The average rate should equal 1/2(1.7 + 0.3) = 1.0 F/hr. I 2-6

l I do not take into account mass transfer (mixing due to internal recirculation). This internal recirculation will not increase these cooldown rates, but will allow the coolant to mix and cool down more unifonnly. Using the steady-state heat loss calculations, the OTSG tube coolant cooldown rate is 2.0*F/hr., while the RCP coolant cooldown rate is 3.9*F/hr. A " mass-averaged" cooldown rate for an idle loop is found using these cooldown rates and the volumes shown below. Region Volume Cooldown Rate Product 3 RC Piping 743 ft 1.7*F/hr. 1263 ft.3*F/hr. 3 OTSG Heads 554 ft 1.7+0.3 = 1.0*F/hr. 554 ft.3 F/hr. 2 3 OTSG Tubes 1487 ft 2.0 F/hr. 2974 ft.3*F/hr. 3 RC Pump 100 ft 3.9*F/hr. 390 ft.3*F/hr. 3 Total 2884 ft 5181 ft.3.F/hr. Thus, the average cooldown rate is 1.8*F/hr. 2.3.3. Boron Mixing Regulatory Guide 1.'139 specifies that cold shutdown must be achievable following a loss of offsite power event using natural circulation decay heat removal. Although certain assumptions and criteria are addressed in this document, specific guidance on the treatment of single loop natural circulation situations is not provided. Reference 4 demonstrated acceptable procedures for reaching cold shutdown assuming worst case assumptions for equipment and systems availability and core conditions. Only safety grade equipment was used in the evaluation. It is not clear that Regulatory Guide 1.139 requires that these assumptions be maintained while addressing the additional failure (s) associated with single loop operation. The fundamental problem associated with the single loop condition is demonstrating I adequate boration of the idle loop. The concern is that unborated water from the idle loop could at some point be displaced into the core causing a reactivity excursion and violation of the required shutdown margin. The approach to this problem involved two steps. First, steady state calculations were made to determine boric acid concentrations with and without any mixing with I 2-7

l the idle loop to show that they satisfied the required shutdown concentrations. Second, RELAP was used to show that boron mixing occurs in the idle loop during flow oscillations and reversals. The results of this approach show that: 1. Suff.cient boric acid from the Chemical Addition System or the Borated Water Storage Tank and Emergency Boration System can be added to the RCS to satisfy the nost limiting shutdown requirements available (cycle 1 and preliminary cycle 2 data), and 2. Flow oscillations and reversals in the idle RCS loop are sufficient to provide adequate mixing in the idle loop. The first result is inadequate by itself because it does not account for the actual mixing process between the active and idle parts of the RCS. The flow reversals and oscillations are discussed in detail in Section 2.3.4. Contraction volumes available for boron injection were conservatively calculated during RCS cooldown by assuming no cooldown of the idle loop. The available contraction volume was filled with borated water from either the Chemical Addition System (CAS) or from the sequential use of the Emergency Boration System (EBS) and the Borated Water Storage Tank (BWST). _ At regular temperature intervals during I the cooldown, the boron concentration was calculated assuming the boric acid remained in the active parts of the RCS and assuming it mixed completely in the RCS (including the idle loop). These results were compared to the shutdown I requirements for cycle two with the maximum worth rod withdrawn. These are the limiting known boration requirements. Boration from the EBS/BWST provided the smallest margin to the required concen-trations. A comparison of the required boron concentrations and the achievable concentrations is shown on Figures 2-4, 2-5, and 2-6. It is clear that the requirements are satisfied following mixing of the idle loop (at the critical concentration) with the remainder of the RCS. The second part of the boration problem involves the mixing mechanism. The RELAP model contains a simple mixing model which distributes the boron entering each node over the entire node. Thus, it is possible to observe boron movement from one node to another as a transient progresses. l One single loop natural circulation case was extended to observe this phenomenon. The results show boron distributing following an HPI initiation. The idle loop, l 2-8

I l

I displaying flow oscillations and reversals, begins to borate as idle loop fluid I moves in and out of the RV downcomer and upper plenum regions. This effect can be observed in Figures 2-7 and 2-8. The RELAP analyses show that the operator can influence idle loop flow by controlling active loop secondary side parameters such as AFW flow and steam pressure. In addition, short bursts of AFW to the idle OTSG can induce significant idle loop flow rates. These results mean that boron mixing in the l idle loop can virtually be assured by deliberate operator actions. Periodic cycling of flow in the idle loop before and during a cooldown will ensure boron mixing. However, because of the qualitative nature of this evaluation, it has not been demonstrated for all possible plant conditions and transient sequences that adequate boron mixing will occur. This strongly suggests that a positive I means of determining idle loop boron concentration be provided to the operator. I Unsuccessful attempts to induce proper boron mixing via the active loop would indicate a need to vent the idle loop. The additional mass exchange in the idle loop would enhance the boration process. I By inducing mixing in the idle loop, the cold safe shutdown methods recomended in Reference 4 can be applied for single loop conditions. Since no aspect of I these actual procedures are time dependent, a slower cooldown process to prevent l RV head bubble fomation and to allow for mixing in the idle loop can be accommodated. The only seriously affected parameter might be the AFW storage capacity. (This point is addressed later in the evaluation of the RV head cooldown rate.) 2.3.4. RCS Performance During Single Loop Natural Circulation The following general strategy was used in perfoming the RELAP analyses. The model was initialized at full power conditions and a loss of offsite power was l simulated. This resulted in the reactor tripping, the turbine tripping, and all four reactor coolant pumps coasting down. Ten minutes of transient time were run to allow the RCS to approach a steady state natural circulation condition. Main feedwater was teminated to both OTSGs and auxiliary feedwater (AFW) was only introduced to one OTSG. The other OTSG was allowed to dry out. This is referred I to as Case 0 on Table 2-1. After ten minutes, several cases were run from the established RCS condition, I taking advantage of RELAP's restart capability. These cases are tabulated in Table 1. 2-9 l

TABLE 2-1. Summary of RELAP Cases Run # Initial Conditions Details of Run Purpose 0 Full power, four RCPs, Simulate loss of offsite Establish single loop natural steady state. power: reactor trip, four circulation conditions. RCPs trip, loss of MFW, AFW initiated, one OTSG isolated. 1 Quasi-steady state single Begin cooldown (C/0) with Investigate response of idle loop N.C. P0AVs on operating 0TSG. loop to C/D on active loop. Does flow reverse? Reverse heat transfer? N 2 Quasi-steady state single Open PORV and blow down Approximate procedure for OTSG loop N.C. until RV head is voided. tube rupture. Start HPI if hot legs reach saturation. 3 Quasi-steady state single Close PORV. Continue C/D Approximate C/D forced from good loop N.C. with RV head with good 0TSG P0AVs. OTSG. Examine RCS pressure voided. response. 4 Quasi-steady state single Restart AFW to idle loop. Can minimal injections of AFW loop N.C. with RV head to idle OTSG induce significant voided. flow in bad loop? 5 Quasi-steady state single No changes. Evaluate the steady state condition loop N.C., no RV head established at ten minutes by bubble. simply extending it. 6 Quasi-steady state single Restart one RCP. (NOTE: No Investigate effect of RCP reticri loop N.C. with RV head meaningful results obtained on RV head bubble and RCS pressure.

voided, from this run.)

1 7 Quasi-steady state single Vent idle loop hot leg. Determine venting time and RCS loop N.C., RV head voided pressure response, and voids also in idle loop.

I Because large portions of the entire cooldown process would have required very I large amounts of computer time, these studies attempted to examine the basic phenomena instead. As a result, some manipulations of the model do not coincide with good operator practice. This applies most directly to the extremes that certain parameters would be allowed to reach. For example, in Case 1 the plant l cooldown via the power operated atmospheric valves (P0AV) in the good 0TSG was started by opening all P0AVs. Steam pressure dropped below 400 psig in three minutes. In the actual plant, steam pressure transients would generally be controlled in a deliberate fashion. The transient cases will be discussed individually, followed by a discussion integrating these results with those from the RV head model. Case #1 In this run, all of the P0AVs in the good loop were opened for three minutes to .l initiate a cooldown. No voids existed in the RCS and temperatures and pressures were as expected following a loss of offsite power event. This represented the maximum cooldown rate which could have been induced by dropping steam pressure. The results show that RCS pressure remained above 1600 psig because the normal pressurizer level control system (using makeup, not HPI) was sufficient to control pressurizer level. The large steam mass vented from the good 0TSG fg caused a drop of about 600 psig in steam pressure, exceeding 100 psi / minute B early in the venting. The steam pressure decrease was aggravated by high AFW flow rates to the good 0TSG. The sudden cooldown of the active loop caused the reactor coolant flow in that loop to accelerate. The additional flow caused a sufficient increase in the core ~ pressure drop to induce reverse flow in the idle loop. (This is the first time this effect has been demonstrated for a B&W plant.) It was also observed in other runs. As hot water from the reactor vessel began moving into the cold leg, it eventually reached a point where the driving head provided by the good loop was insufficient to sustain more reverse flow. Consequently, the flow stopped, and, because of the new thermal gradients set up in the loop, began i s1 wly fl wing in the forward direction again, eventually dampening out to a E5 stagnant condition. I I 2-11 I

Two major conclusions can be drawn from this trans'ent: 1. Overcooling due to secondary side upsets while in a single loop natural circulation condition tends to be slow acting and milder than when in forced flow with one or both OTSGs in service. 2. Reverse flow in the idle OTSG may be a significant phenomenon which could cause unusual trends in the idle loop temperature measurements. Case #2 This run was started from the quasi-steady state condition established after the ten minute simulation of the loss of offsite power event. One OTSG was drying out; the other was controlling level at 50% on the operate range. The RCS pressure was deliber.ately decreased using the PORV, until saturation conditions were reached in the RV head. After complete voiding of the head was achieved, the PORV was closed. RCS pressure was approximately 1400 psig. A slight amount of voiding occurred in the top of the hot legs. However, the coolant flow rate of 200 to 800 lb. per second was sufficient to prevent steam from collecting in significant volume, even in the idle loop where oscillatory flow was observed. The RV head was also at a somewhat higher temperature than the hot legs, 600*F ,j compared to 585 to 590*F. Thus, the head basically maintained system pressure above saturation for the hot legs. HPI was initiated immediately after the PORV was closed at 760 seconds. RCS pressure imediately began increasing, reaching the normal RCS pressure of 2160 psig in approximately 31 minutes. The pressure increase was smooth, causing both the pressurizer and RV head bubbles to compress as expected. The small amount of bubble fonnation in the hot legs was condensed. Several observations can be made from this run: 1. Opposite trending of pressurizer level and RCS pressure was clear I during the large RCS pressure decrease. 2. HPI actuation at 1600 psig should effectively prevent major bubble I fonnation in the RV head except in the case of severe overcooling. Saturation temperature for 1600 psig is approximately 604 F, at or near the maximum expected RV head temperature at full power. I 2-12 I

I 3. Even with the PORV fully open for nearly three minutes, the RV head was incompletely voided, suggesting that bubble formation occurs at a modest rate. 4. The repressurization of the RCS with HPI occurs at an easily controlled rate. The operator should have time to respond. 5. Natural circulation was not interrupted in the good 0TSG. 6. The boron concentration in the core increased dramatically with the initiation of HPI. RELAP output shows other parts of the RCS borating as well, with time lags due to transport times. Figures 2-9, 2-10, 2-11, and 2-12 demonstrate the above points. Run #3 This run combined conditions from runs one and two to evaluate whether an RV saturation condition would influence the initiation of a single loop cooldown. The transient was begun with no void in the RV head and RCS pressure between 1400 and 1500 psig. The P0AVs in the active loop were opened to initiate cooldown. Partial RV head voiding occurred two minutes later. As in the case of Run #1, the overcooling effect was very modest. T of the cold active loop dropped approximately 50*F as the P0AVs remained open three minutes. Due to the dilution effect of the RV vent valves, core inlet temperature only decreased 15 to 20*F. Pressurizer level decreased very slightly and the RV head bubble volume remained nearly constant. Hot leg voiding did not occur. The model predicted about 5 F subcooling margin in the hot legs. l! t l[ This scenario would not normally occur in the field because the operators would increase the RCS subcooling margin before inducing a cooldown. Nevertheless, the run displays the general stability of the system. Increased voiding did not occur as the cooldown began. RCS pressure and pressurizer level remained stable. Run #4 This transient was run to determine the feasibility of indirectly re-establishing (or temporarily inducing) natural circulation in the idle RCS loop. The run ,g E was initiated with an RV head bubble and RCS pressure at saturation; pressurizer level was off-scale high due to bubble expansion. Only the active OTSG had AFW delivery; it was on level control at 50". on the operate range. 'I 2-13 i

Full capacity AFW was initiated into the idle OTSG and maintained for two minutes. Steam pressure in that OTSG dropped rapidly and the heat removal rate (primary to secondary) increased steadily to approximately 5% of the full power heat flux. RV inlet temperature began dropping, but, as in previous runs, it was wanned by recirculating vent valve flow. Thus, core inlet temperature did not demonstrate as sharp a decrease as the idle loop. While AFW was on, idle loop flow remained high, exceeding the natural circulation flow rate in the good loop. This is shown in Figures 2-13 and 2-14. Note that just prior to initiating AFW (s735 seconds) an MSSV lifted in the idle loop. This caused enough of a temperature decrease to cause the idle loop flow to start in the forward direction. At 770 seconds, AFW was initiated and kept the idle loop flow from decreasing as it had done during earlier MSSV lifts. The prolonged and strong RCS flow in the idle loop confims that AFW injected high into the OTSG is sufficient to induce natural circulation even in the absence of an OTSG pool. This suggests that an idle loop could be effectively " moved" intennittently with short bursts of AFW. The analysis also shows that such bursts would not cause significant RCS pressure or pressurizer level swings. Run #5 This run was an extension of the base run (used to establish single loop natural circulation). Several questions had developed concerning the steady state condition assumed at the 600 second mark. In order to better establish that a reasonable steady state existed at 600 seconds, the run was extended to 1000 seconds and compared. Of particular interest was the RC flow in the idle loop. Previous analyses (References 1, 2) showed that natural circulation is an oscillatory phenomenon and only at high secondary side OTSG levels are oscillations dampened out. Steam pressure or large AFW flow swings can induce oscillations. The RELAP analyses done for this work demonstrated similar behavior in regard to an idle loop. The active loop oscilla1:ed with a long period (s4 minutes) and was probably responsible for some of the lesser oscillations in the idle loop. A result of this behavior is that the idle loop flow was nonzero during most of this and the other transients. It frequently reversed for periods up to 30 to 45 seconds, and remained sensitive to flow in the active loop. This was noted in Run #1. I 2-14

As an additional check of this interaction, Run #5 was restarted with all primary to secondary heat transfer in the idle loop OTSG deliberately set to zero. This eliminated all possible effects of the small remaining secondary side inventory. I The run clearly shows that idle loop oscillations continued. Figure 2-15 shows the idle loop flows at or near zero for nearly a minute only to begin increasing again. Flow reversals during single loop natural circulation conditions in B&W reactors I were noted in the NSAC report concerning the Crystal River Unit 3 incident of February 1980 (Reference 3). This event involved delayed actuation of AFW after a power operated relief valve failure had reduced the RCS to saturation conditions. The operators correctly tripped all reactor coolant pumps and let HPI run. AFW was finally initiated, but to only one OTSG. Within approximately two minutes, RCS flow in the other OTSG had reversed. These results were predicted by a RETRAN model. Neither the magnitude nor the duration of the flow reversal was I reported. The conclusions from this run are that: e the condition from which all other transients were initiated (t=600 seconds) was representative of a steady state condition e flow reversals in the idle RCS loop can be expected to occur during natural circulation Run #7 This case simulated venting of the idle loop. The idle loop hot leg contained approximately 25 cubic feet of steam before venting started. This was achieved by starting this run after the RCS pressure had been deliberately decreased to fom voids in the head and hot legs. Venting was started with no makeup or HPI running. This was intended to accentuate any RCS pressure changes which might occur. RCS pressure was about 1365 psig. 3 The venting proceeded to reduce the bubble at the rate of 0.5 ft /sec, eliminating it in about one minute. As expected, the RCS near saturation experienced little change in pressure. I 2-15

a i Figure 2-1 REACIOR VESSEL HEAD C00LDOWN 600 = 1.7 F/HR TRV HEAD MAX dt 550 l l 177 FA N.C. COOLDOWN 500 INITIAL SHELL TEMPERATURE = 604 F -~ w u. 1 L 450 INITIAL WATER TEMPERATURE = 505 F = r CS AMBIENT TEMPERATURE = 120 F R l NATURAL CIRCULATION FLONRATE = 1141 #/IIR 400 ~ $ = 100*F/HR dt l 350 \\, = 300 8 l 1 2 3 4 5 6 7 8 t(hrs.) l i i e

I Figure 2-2 APPROXIMATE TEMPERATURE PROFILE OF REA:170R VESSEL HEAD TWO HOURS AFTER C00LOOWN BEGINS I I T = 590*F T = 595 F r I l T = 585* F L ~ I T = 400* F I T = 415* F I

I L

T = 385 F 'I I I

2. n I

g

I Figure 2-3 APPROXIMATE TEMPERATURE PROFILE OF THE REACTOR VESSEL HEAD SEVEN HOURS AFTER C00LDOWN BEGINS =5 F = so F T = 585 F l i

f Y T = 310 F T = 310"F I

I T = 310*F i I L .I 'I I 2.i e I I

I I Figure 2-4 80L SHUT 00WN BORON VS. TEMPERATURE FOR CYCLE 2 OPERATION-STUCK R00 1700 I C t 1600 C I 1400 f a. 1200 c I = l SHUTDOWN REQUIREMENT c 1000 I i Ct = CONCENTRATION IN ACTIVE PART OF RCS c 800 l j WITH NO MIXING WITH IDLE LOOP 5 C = CONCENTRATION IN RCS WITH COMPLETE I I"U 600 '! I 400 200 I O 550 450 350 250 150 50 RCS Average Temperature *F -I 2-19 I

I I Figure 2-5 MOL SHUTDOWN BORON VS TEMPERATURE FOR CYCLE 2 OPERATION-STUCK R00 I 1400 - C t 1200 E C a. 3 1000 p I i 8 [o 800 E SHUTOOWN REQUIREMENT E 600 Ct = CONCENTRATION IN ACTIVE PART OF RCS WITH NO MIXING WITH IDLE LOOP 400 C = CONCENTRATION IN RCS WITH COMPLETE MIXING I 0 I I I I ,g E 550 450 350 250 150 50 RCS Average Temperature *F I 2-20 I

~I I Figure 2-6 EOL SHUTDOWN BORON VS. TEMPERATURE FOR CYCLE 2 OPERATION-STUCK R00 1600 Ct = CONCENTRATION IN ACTIVE PART OF RCS 1400 WITH NO MIXING WITH IOLE LOOP C = CONCENTRATION IN RCS WITH COMPLETE MIXING Q. I i C t } 1000 E 800 C 2 O I i 600 3 SHUTDOWN REQUIREMENT I 400 l 200 ,I O 550 450 350 250 150 50 RCS Average Temperature *F !I 2-21 I

M M M M M M M M M M M m' M M M m Figure 2-7 CASE #2 LOOP BORON CONCENTRATIONS CPC0 SINGLE LOOP NAT. CIRC. CASE 2 BORON TRANSPORT 280 y i g i i IDLE LOOP HOT LEG 240

---

ACTIVE LOOP HOT LEG

~ l 1

---

10LE LOOP COLD LEG l

- - - ACTIVE LOOP COLD LEG es s / / i 200 s 1 's / s / / /- 5 160 / e [ 5 / i s y j I l 120 fl~W~~ I 80 f j / / 7-~~~ s / / 40 j,' f,1 z _ _ 0 1 I I I I I I I I l l ) 800 860 920 980 1040 1100 1160 1220 1280 1340 1400 Time (sec) 1

l Fi 2-AS N RC S P E AT AAE CPC0 SINGLE LOOP NAT. CIRC. CASE 2 BORON TRANSPORT l 660 g g g g g y i g '\\ / s I \\ / '\\ / IDLE LOOP ,1 g 540 y / gi l I - - - ACTIVE LOOP / 1 \\ I \\ / / g \\/~s/ 420 f i M i l 3 I I 300 l i ) 7 E I O b I \\ = i [ 180 1 g E I l l 8's I f l i \\ s~me \\g g / \\p g 0 l 60 0 -60 ~ l I A I -180 800 920 1040 1160 1280 1400 Time (sec) I

l M M M M M M M M M M M M M M m m m Figuea 2-9 CASE #2 RCS PRESSURE AND RV HEAD SATURATION PRESSURE l I CPC0 SINGLE LOOP NAT CIRC 8ASEDECK 2400 I I I I I I i l l l 2250 i I 1 a Z 2l00 z E J l 3 1950 '? d. r I E E i = 1800 5 0 i d-1650 j I i N \\ %~~_ 1500 ,s*# 1350 i I I I i i I 600 640 680 720 760 800 840 880 920 960 1000 Time (sec) 4 l 1

l M M M M M M M M M M M M M M M M M Figure 2-10 CASE #2 RV HEAD LIQUID VOLUME CPC0 SINGLE LOOP NAT CIRC BLOWOOWN 520 I I I I I l l l 440 1 C 5 360 a a2 x I E 280 7 e ~ O E = E 200 o 5 .? ] 120 ~ l 40 -401 I I I I I I 1 l 600 640 680 720 760 800 840 880 920 000 1000 i j Time (sec) i l l

M M M M M M M M M M M M Figure 2-11 CASE #2 PRESSURIZER LEVEL 4 I l CPC0 SINGLE LOOP NAT CIRC BLOWDOWN 100 if i ~ i 1 80 o I 60 w b 'f E a-i 40 ) l I 20 ~ l 4 4 O i i i i i I I I I 600 640 680 720 760 800 840 880 920 960 1000 l Time (sec) l

m m M M M M M M M M M m M M M m m m m Figure 2-12 CASE #2 REACTOR BORON CONCENTRATION CPC0 SlHGLE LOOP NAT. CIRC. CASE 2 BORON TRANSPORT 280 g g g g i y y j 240 200 I i i e l 7 E 160 t: a l E n. i 120 1 l 80 40 i i 0 l l I I I I I I I j 8 00 920 1040 1860 1280 1400 Time (sec) 4 1

M M M M M M M M M M M M M M M M M M Figure 2-13 CASE #4 RCS LOOP FLOWRATES, A AND B CPC0 SINGLE LOOP NAT CIRC BLOWOOWN 1200 i i i i i i i 1000 I i i I O 800 IDLE LOOP g 1 C PJ'yW I \\ = 600 I o y E p'" \\ \\

p>

,/ g / p 400 I y A-vsg %, f 'w=yw 200 ACTIVE LOOP 0 i I I I I I I I I -200 600 640 680 720 760 800 840 880 920 960 1000 Time (sec) 1 I i

Figure 2-14 CASE 04 OTSG PRESSURES, A AND 8 CPC0 SINGLE LOOP NAT CIRC BLONDOWN 1200 I I I I I I I l I liSSV LIFT 1125 i AFW INITIATED o /t 'g = / I 1 t 1050 m I ,s ,/ f \\ ',s' \\ \\ ',s I \\ I p/ ~ E 975 ACTIVE LOOP y B = ~ ) W a. I e 900 IDLE LOOP g i 825 i i 750 i l 675 I I I I I I I I l 600 640 680 120 760 800 840 880 920 960 1000 j Time (sec)

.i i M M M M M M M m m m m Figure 2-15 CA$E 5 RCS LOOP FLOWRATES, A AND B l l CPC0 SINGLE LOOP NAT. CIRC. CONTINUE CASES 750 I I I I I I I I i i ACTIVE LOOP ] 600 IDLE LOOP A g 450 i i i h 1 = 300 g 5 l ~ j h \\ ,/ -r \\ / i / ) E 150 l 2 \\, / / 4 %s/'% -T e / g,3 0 I< so p us i -150 I i -300 I I I I I I I I I ) 150 800 850 900 950 1000 l Time (sec) l 1 i

I l i 3. CONCLUSIONS AND RECOMMENDATIONS 3.1. Operational Considerations The primary concerns related to single loop natural circulation cooldown are: e RV head cooldown ,s e Idle loop cooldown e Boration of the idle loop 3.1.1. RV Head Cooldown Analytical work showed that the RV head will cool very slowly with a maximum expected cooldewn rate of approximately 1.7'F per hour. The maximum allowable pressure (as measured from the RCS hot leg) for the decay heat removal (DHR) system is 350 psig (Reference 7). This pressure corresponds to a saturation temperature of 436*F. Thus, the RV head must be cooled from 604*F (a change of 168 F) with no allowance for instrument errors. Assuming a con-tinuous head cooldown rate of 1.7 F/hr, it would require 98.8 hours to reach the DHR cut-in I point. (This also assumes that the 280 F limit on DHR temperature has been satisfied.) I The cooldown time has been plotted as a function of the average cooldown rate (Figure 3-1). The cooldown time for the head to reach 436 F is between 84 and I 120 hours. This is the minimum time required to get on the DHRS. If a lower RCS pressure is desired before initiating the DHRS, the RV head cooldown time increases proportionately. With no way of either inducing flow through the head or venting it, the head cooldown rate cannot be affected. The conclusion from this is that the RV head becomes the limiting component during the RCS cooldown if drawing a bubble in the head is to be avoided. The above cooldown times far exceed those reported in "An Evaluation of Safety Grade Cold Shutdown Capability for Consumer's Midland Units 1 & 2" (Reference 4). This reference showed a maximum cooldown time of six hours for the case in which 3-1 I

I no hot leg temperature instrument errors are assumed and 38 hours for the case I with worst case instrument errors. The prolonged time for supplying AFW for decay heat removal increases the integrated volume requirements for AFW (Figure 3-2). Midland's condensate storage tankage (Reference 6) could probably ] satisfy these large AFW requirements, but only if credit could be taken for i both tanks and assuming they are filled to near c.apacity. Because the cooldown time is slow, ample time would be available to ensure proper procedural compliance with the guidelines presented in Reference 4. It is important to note that the cold safe shutdown guidelines do not have time dependent actions associated with them. Therefore, the sequences apply no matter how slowly they are perfomed. The RELAP analysis also shows that a completely voided RV head does not interfere with natural circulation. RCS pressure control is limited by bubble fomation because bubble expansion limits the depressurization which can be achieved. The I RV head cooldown will not be slowed by the existence of a bubble. 3 Calculations show that approximately 30 hours are required to collapse a 458 ft steam bubble (a full RV head) at 2000 psig durir.g an isobaric process. Dissipation of the bubble could probably be accelerated by deliberately allowing expansion I of the bubble into the upper plenum cylinder and hot legs. However, this evolution may not be well controlled and could lead to partial or total occlusion of a hot leg. It could also involve a rapidly moving steam-liquid interface that could cause themal shocks to the reactor vessel or head. This evolution was not investigated during this analysis. An important point is that there is no temperature indication for the upper head region. Therefore, the operator will have to rely on elapsed time to ensure reasonable cooling of the head assuming a 1.5*F per hour cooldown rate. This rate is a compromise between the fact that the analysis was somewhat conservative and the cooldown rate will decrease as the head cools. The cooldown time can be calculated by detemining the initial head temperature before reactor trip (assuming I head temperature equals hot leg temperature) and subtracting 435 F from it. This difference should be divided by 1.5 to get the cooldown time in hours. Direct head temperature measurement would be preferred. I 3-2 I

\\ I i 3.1.2. Idle Loop Cooldown The idle loop cooldown rate has been shown to be close to that of the RV head. However, the RELAP analysis shows that, unlike the RV head region, considerable I flow continues to exist in the idle loop after the RCPs have coasted down and ' the idle loop OTSG has dried out. The analysts supports the following conclusions: 'g e Nomal flow fluctuations in the good loop cause forward and reverse 5 flow oscillations in the idle loop. e Sudden increases in the heat removal in the good 0TSG (e.g. via steam I pressure decreases) induce large flow changes in both the good and idle loops. e The addition of AFW to the idle OTSG for periods as brief as two minutes can induce large forward flow surges in the idle loop. The result is that the idle loop can be circulated deliberately, @ necessary, to provide mixing of hot water from the idle loop with water in the reactor ~ vessel. As the cooldown of the good loop and reactor vessel proceeds, periodic mixing transients can be inducc1 to accelerate cooling of the idle loop. Thare are three major ways to induce idle loop mixing: I [ e Decrease steam pressure in the good 0TSG. e Rapidly increase the level in the good OTSG. e Provide a short burst of AFW to the idle OTSG. From a mass release standpoint, the first two are clearly preferable. However, periodically cycling AFW into the idle OTSG has the advantage of providing a positive cooling mechanism for the OTSG shell. Because the exact cooldown rate and instantaneous conditions for all possible plant situations cannot be predicted, the operator should take advantage of available instrumen-tation to detemine the best course of action. Specifically, OTSG shell themocouples can be used in conjunction with the idle loop temperature readings to detemine a value for the tube to shell temperature difference. If this i difference does not reach a limit (60 F) for non-emergency events), idle loop mixing can continue via the nomal oscillations in the good loop or by inducing transients in the good 0TSG. lI 3-3 I

I Another way to provide idle loop cooling is by venting from the idle loop high I point. This was not analytically evaluated because it is not an effective cooling mechanism for the idle loop. It has the significant drawback of requiring large discharges of reactor coolant to the reactor building, and is seen as an unnecessary evolution. 3.1.3. Boration of the Idle Loop 1-The following conclusions can be drawn from the analysis: e Adequate boric acid can be injected into the RCS to satisfy the most restrictive shutdown requirements available now (preliminary cycle 'I 2 data). e Adequate mixing of the boric acid into the idle loop is likely to take I place as. result of RCS flow oscillations. The mechanisms discussed in Section 3.1.2. provide the means for borating the I idle loop as well as cooling it down. Although these flow mechanisms have been demonstrated in the RELAP analysis, the mixing process itself is somewhat more time dependent. RELAP is known to overpredict the degree of mixing from node to node. The result is that at any given time the operator will have no positive indication of the boron concentration.n the idle loop. During natural circulation with an 4 i idle loop there is no substitution for direct boron concentration measurement. Assuming such instrumentation were available, the operator could respond accordingly by stopping the cooldown until proper mixing was achieved. A criterion for good mixing is to have the idle loop baron concentration within 50 ppm of the good loop. This is a conservative, but achievable value. Because excess shutdown margin exists, the operator could compare the required shutdown concentration for any given temperature and simply wait for the idle loop to borate to this value. A higher concentration in the good loop would be unimportant. 1 In summary, the cold safe shutdown guidelines provided in Reference 4 can be invoked for single loop cooldown events provided that: e Sufficient care is taken to borate and cool the idle loop, e The cooldown is extended sufficiently to allow for RV head cooldown. I 'I 3-4 I

I e Adequate supplies of AFW grade water are available for decay heat removal until the DHR system is cut-in. 3.2. General Guidelines I RCS cooldown may proceed when the following plant conditions have been achieved: e Idle loop is water solid. e Idle loop boron concentration is within 50 ppm of the good loop. e Idle OTSG tube to shell AT is less than 100"F. e The existence of an RV head bubble has been detennined (yes or no). The results of the analyses suggest a logical sequence for achieving cold safe shutdown with an idle OTSG. This sequence is shown on Figure 18. The sequence utilizes the procedures developed for the two loop cold safe shutdown operation (Reference 4) with added restrictions to account for single loop effects. This procedure assumes no flashing is allowed in the RV head, i.e., the cooldown progresses assuming a head cooldown rate between 1.5 and 2.0 F per hour. This constraint minimizes potential thermal shock concerns in the RV head and upper internals. If an RV head bubble forms, depressurization events are to be avoided until the head bubble is condensed. The lack of temperature instrumentation in the head region prevents direct monitoring of the localized saturation margin. This means that an RV head temperature calculation will be necessary during the cocidown. This can be done by assuming that the head is at the hot leg temper c ure which existed prior to the single loop condition. For example, if the plant had been ,g at full power before the transient, the head would have been at or near 604 F. 5 Its temperature at a later time can be calculated with the following single equation: T(new) = 604 F - (2*F/hr.) (at) where at is the elapsed time from reactor trip. This simple equation can only I give a reasonable estimate if a plant cooldown is initiated within a few hours of the reactor trip. I I 3-5 I

I The procedure begins with boration of the RCS to the hot safe shutdown condition. Reference 4 provides the necessary guidance for this step. The first condition to be mitigated, if it exists, is a bubble in the idle loop (Block 1.2, Figure 3-3). It will be indicated by loop level instrumentation and should be vented. This will return the loop to a solid condition and pemit forward and reverse circulation to occur. The second condition to check for is an RV head bubble (Block 1.4). Control room strip charts should be examined for opposite trending of pressurizer level and RCS pressure. Opposite trending of these parameters is a symptom of voiding in the RCS, although there are circumstances in which they may trend oppositely with no voiding. (The latter situation may occur during slow increases in pressurizer level with pressurizer spray initiated.) If opposite trending is I observed and bubble fomation in the hot legs is not indicated by the hot leg level instrumentation, then an RV head bubble should be assumed. I The next condition to be checked (Block 1.6) is the boric acid mixing between the idle loop and the active part of the RCS. Mixing can be speeded by inducing g 5 flow in the idle loop by: e Decreasing steam pressure in the good loop. e Increasing OTSG le' vel in the good loop. i e Introducing feedwater to the idle OTSG. The analysis d'd not define limits for using the above methods. However, the following guidelines are reasonable: e Decrease steam pressure by 100 psi in the good 0TSG for one minute. Then restore nonnal control. e Rapidly increase OTSG level by 10% on the operate ranges. Then restore normal control. e Inject AFW feedwater into the idle OTSG at approximately 500 gpm for 30 seconds. The above mpnipulations may be repeated (one at a time) until adequate mixing is achie/ed. I 1 , 'l l 3-6 l I

I The boron mixing itself can only be ensured by measurement of the boron concen-I tration in the idle loop. Both cold leg and hot leg measurements of boron concentration are desirable. The hot leg measurement will probably be more indicative of overall loop concentration because of RCP seal injection and possible stratification in the cold legs. Block 1.8 is a test for the OTSG tube to shell AT in the idle loop. If this limit (100*F) is reached, the RCS should be stopped until the shell cools further. If main feedwater is available, small amounts of main feedwater can be introduced to enhance the shell cooldown. Auxiliary feedwater should not be used for cooling the shell. When each of the above required conditions has been achieved, plant cooldown can proceed in accordance with the guidelines for two loop cooldown (Reference 4). The steps shown on Figure should be repeated every 50*F (Tave) during the cooldown or as the limits of Figure are reached. 3.3. ?lant Modifications This evaluation accentuates the potential need for the following additional functional capabilities of the plant: l e Venting of the RV head. e Measurement of the RV head fluid temperature. <g N e Measurement of RCS loop boron concentration. Each of the above capabilities would have a significant impact on plant operation I during a natural circulation cooldown. An RV head vent would permit a substantial increase in the achievable cooldown rate of the head. Calculations for another plant configuration demonstrated a 38*F/hr. RV head cooldown rate with a 50 gpm vent rate (liquid). In addition I an RV head bubble could be imediately eliminated, restoring positive RCS pressure control to the operator. RV head fluid temperature indication would allow the operator to detennine the head region subcooled margin. He would be able to avoid reaching saturation conditions, thus reducing the likelihood of forming a head bubble. If both the head vent and the temperature measurement were av'ilable, the operator would base the amount of venting on the temperature of the head. This would be significant if the venting were to the reactor building. ~I 3-7 I

I Adequate boration of the idle loop can not be guaranteed in the absence of forced I flow. The RELAP results suggest that good mixing is possible with modest to low idle loop flow rates, even if the flow is intermittent. However, boron measurements are necessary, especially from the hot leg region, to ensure that boration is satisfactory. The above three modifications would greatly enhance the operator's control of the plant, and would permit a more timely cooldown following a loss of offsite power. . I I I I I 'I I I I I I I I 3-8 I

r" E E I I Figure 3-1 TIME TO REACH OHR SYSTEM CUT-lN 130 I 120 E ,I E e c b 110 = I 8 5 ,'l 5 100 = Lm a 3 y I 80 I I I 1.2 1.4 1.6 1.8 2.0 RV Head Cooldown Rate, *F/HR I 3-9 I

I I Figure 3-2 REQUIRED AFW FOR COLD SHUTDOWN (TAKEN FROM REFERENCE 5) 140 120 100 80 i= I 60 M E I 5 40 E .g 0 5 10 " ~ 8 NOTE: l 7 TO THIS CURVE, A00 THE AFW REQUIRED 0 6 ADDITION AFW = 139.60(574-T)5ALLONS 4 T=AVERA5ERCSTEMPERATUREAT l 3 ENDPOINT OF INTEREST 2 1 0 t '-i t i O 1 2 3 4 5 6 7 8 9 10" 30 50 70 90 110 130 150 170 Time Since Trip, nours I l 3-10 I

Figure 3-3 LOGIC FLOW CHART FOR SINGLE LOOP NATURAL CIRCULATION C00LDOWN 1.0 START p 1.1 BORATE TO HSSD I 1.3 I* VENT IDLE , I I BUBB E IST ) SOLID LOOP CONDITIONS NO 1.5 1.4 MAINTAIN OR DOES YES INCREASE RCS BUBBLE MIST PRESSURE; AVOID DEPRESSURIZING EVENTS NO>< ' Is 1.6 INDUCE FLOW tot.E OSCILLATIONS N( ACID ONC WITHIN IN IDLE LOOP 50 Prn or cooD YES 7 1.10 1.9 s PROCEED WITH I I NO C00LDOWN YES Tunt/sartt. WAIT ON C/D ASSUMING 2 F/HR aT OK7 7 C00LDOWN RATE (<1oor) IN RV HEAD 3-11 l

I I \\ I 4. REFERENCE! 1. " Consumers NATURAL Code Runs," 32-1123172-00, dated February 27, 1981. 2. " Emergency Feedwater Level Rate Control," BAW-1686, dated August,1981. 3. NSAC-DR-1, "Thermalhydraulic Analysis of Crystal River - Unit 3 Incident," August 1980. 4. "An Evaluation of Safety-Grade Cold Shutdown Capability for Consumer's Midland Units 1 & 2," Doc. No. 86-1134575-00, May 14, 1982. 5. B&W Doc. No. 32-1132908-00, " Required AFW for Cold Shutdown." 6. System

Description:

Condensate Storage and Transfer System, CPCo No. 7220-SD-M-49B, Rev. 2. 7. Plant Limits and Precautions, B&W Doc. No. 67-1005821-00, October 30, 1978. I 4-1 I

l I 1 1 APPENDIX A Reactor Vessel Head Cooldown Analysis Method I I A-1 l

I I The cooldown of the RV upper head under NC conditions is a result of both heat and mass transfer from the upper head. The flow paths are shown in Figure Al and the heat transfer model for the analysis is shown in Figure A2. Heat I transfer from the RV upper head is accomplished by a combination of conduction, convection, and radiation. The coolant rising from the fuel assembly upper end fitting into the upper plenum and column weldments is assumed to be at the same temperature as the RCS hot leg since the bulk of this coolant goes directly to the hot leg. The RV head (dome) temperature was assumed to be 604*F. This is approximately the hot leg temperature for 100% power. The RCS loops, including the hot leg, were assumed to cool down at 100*F/hr. The amount of flow that enters the RV upper head through the plenum cover is derived from previous hydraulic calculations. The distribution of the coolant in the RV upper head (including the amount of mixing before it exits through the plenum cover holes) is based on the conservative assumption that minimum mixing occurs. Other important assumptions used in this analysis are listed below: 1. Convective heat losses from the CRDs were effectively modeled as conductive heat losses. Instead of a Q = hA AT equation, a Q = -kA dT/dx equation was used with the CRD temperature set at 120*F three feet above the RV head. 2. The flow rate for natural circulation was 3% of full flow. Based on operating experience and the NATURAL code, this rate is considered conservatively low. l 3. Where conductive heat transfer exists between adjacent nodes of different materials with different thennal conductivities, the smaller or limiting value was used. 4. Radiative heat transfer was used for the reflective insulation heat losses to the containment atmosphere. An emissivity value of 1.0 was used. I A-2 I

I 5. The ambient (reactor building) temperature was assumed constant at 120*F. Vendor-supplied transference values were used at insulation / air interfaces. The upper portion of the RV and internals was divided into a multinode representation as shown in Figure A-3. A mass transfer model was superimposed on this multinode model as_ indicated by the solid and dotted flow paths. A solid line from one node to another signifies mixing. A dotted line signifies no mixing, such as the case for coolant rising inside the column weldments from the upper plenum to the RV upper head. Figure A-3 shows that mixing is assumed only in the first layer of nodes in the RV upper head. This assumption is critical to the results of the analyses and, as discussed above, is conservative. As shown on Figure A3, each node represents a three-dimensional ring in the analysis. Finite difference equations were then written for each node volume. This set of finite difference equations was then solved simultaneously for each discrete time step. A 20-second time step was chosen based on conventional stability criteria. Future node temperatures were calculated based on the current temperature plus the heat and mass transfer over the time step. The general fom of the finite difference equations is present + at(Q

• O
• O + O )

T =T k h m r future pxC xV p where at = time step, Q = conduction heat transfer = -kA x AT/aX, k Q = convection heat transfer = hA AT, h Q = heat transferred with mass = p x V x C x (T -Tpresent)/at, m p new Q = radiation heat transfer = B x A x (Tpresent) - (Tadj) ' 7 p = density of node material, C = specific heat of node material, p V = volume of node, k = themal conductivity of node material, A-3

A = horizontal cross-sectional area, AT = temperature difference across interface, AT/aX = temperature gradient across node, h = convection coefficient, B = Stefan-Boltzmann constant, adj = temperature of adjacent node T (bothT and T I" O equation are absolute), present adj r T = new temperature of node due to mass transfer, new = (T -Tin) exp (-lh/o x vol)at + Tin' present rh = mass flow rate into node, T = weighted mass average incoming temperature, in T = present node temperature. present Conductive heat transfer was considered at the boundaries of similar media (air-air, steel-steel, water-water, insulation-insulation). Convective heat losses were considered at other boundaries, (air-steel, air-insulation, steel-I water). Finally, radiative heat transfer was considered for insulation-ambient air conditions. The CRD convective heat losses (to the service structure region) were modeled as conductive heat losses. Ambient conditions of 120 F were assumed: I -kA(dT/dx) Q = k where Q = heat transferred by conduction, k k = thennal conductivity of carbon steel, A = horizontal-cross sectional area, dT/dx = linear temperature gradient along CRD length. The additional Q tenn above was added to the finite difference equations for nodes in the RV head (dome) that contain CRD nozzles. The leadscrews and column weldments were modeled similarly. The masses and volumes of these components were distributed among the applicable node rings to take into account the cooling by these components. A-4 I

I All sources of heat - both into and out of each node - were summed and then divided by the mass and C of the node. This term was added to the present p temperature to obtain the new node temperature. The process was carried out I for all nodes before continuing on to the next time step. I I I I I I 'I I !I el l u LI .g 'I A-5 I

E Figure Al UPPER PLENUM AND RV UPPER HEAD E MASS TRANSFER MODEL l I 5 I 10 s K

1. )L ff i j OC fl I

l [9 J

x/

n v M' / A' j (,s* I J / / (pi b / / / ' g n m .~,1. 3 g.4 g et.u.s(c)i ri..i. g ...i - -. i g E . ~n ... n.... i I i si ri., ..i....i_.,i.. o it =,.i n....i_ u 7ag g

i....i

.,i.. i_ I u go,.., i.... .,i... .,i. a n ..i... .,i.. o i. i. ri I i. 3,=.. ......i i. u n,,g _ a a n.i- -...i o. I A-6 I

Figure A2 HEAT TRANSFER N00EL h6 l

• ikA I

1 l4 - eg I i g I Nf?, 2 'O i }/ @/# P' 50 0 I / ... A - ~ f f@ ' ~@~ K. Jl I ( \\ \\ / a / \\ Y' a a / I f~/ a ,/ / / f %b-i (>./,/ / ((, y . = _... _ I (b) Celues Melamaust le tleser Planie 4 (t) ue CaC assale (a) 81 need te a1P llMer Leestatten 6 (g) Insulatten to Centatsiment s E F (t) av seene to Pienen Caver (,) - este. te - --le .ed.e 9 tap av nues to Ceolans to unsee neae I 10 (a) unser need Centent to Plane Ceser 11 (a) see tanese laattaties to Inselatten 12 (b) Prue mettee to Ceeser meses er Colume y mee ) u m) C=aoi.=C-=i-t m I 14 (a) laseisties to Centstammt 15 (a) av meII ta unser Pieme Caelaet I (a) Camerciam (t) Casemetten I te) test attem Cao assale see assumme to to 120P at 3 feet aseve av none. I A-7 I

I I Figure A3 N00 LNG DIAGRAM AXIS OF ROTATION I _ _ 7 _, __ I i i i i i AMBIENT __ _ _ _ _ _ + _ L _L L I L_ _ I 8 _ i_ I_ _ { __ _ _ INSULATION g i_ _7 _ j 1-- -r - T-- - i i E r- ----' STAGNANT [ AIR SPACE l- - p_ _I t-{- Z_J l L__j._ql___L__j___9 + L_ }_ _ p_ _ _ _J g RV UPPE HIAD M RV HEA0 = REGION 1 _ _ y - l -l, l s (DOME) l J _p___ _] _4__l I1 g g_ f! !h f lf ik \\ IPLENUM" '" h lj' l' I I 'I A 'l il ll ll !l ll2 t d RV ABOVE 1 I (, g '[I jl ij l--tr, -j-- 4-J-hId- ---l O HOT LEG l l l h g A n ( n d - PLENUM RV UPPER s l ll d l% A'N. 5-t--- b,H -] CYLINDER L-PLENUM -- T +"- J l l I l ll g REGION l l l OU I .J d JL, E..__m l _.l...l.___ U ? O HOT LEG COOLANT FLOW NOTE: THIS N00 LNG DI AGRAM IS ONLY A ONE-HALF CROSS SECTION OF THE UPPER REACTOR VESSEL AND INTERNALS. IN THE ANALYSIS. THIS N00 LNG OlAGRAM IS ROTATED ABOUT THE LEFT SIDE VERTICAL AXIS S0 THAT THE RV AND INTERNALS ARE MODELED IN THREE DIMENSIONS. EACH NODE THUS BECOMES A RING. A-8

a a u ---+--a.. I I APPENDIX B RCS Idle Loop Cooldown Methods 1 i i l. i i t I B-1

I The cooldown rates of the coolant in the hot leg and 0TSG head were evaluated using one-dimensional (radial), four volume, finite difference (explibit) methods. Figures B1 and B2 show the geometry and dimensions of the hot leg and 0TSG head. There are four types of thermal resistances between the coolant and contair. ment atmosphere: internal convective resistance, pipe or shell conductive resistance, insulation conductive resistance, and the external convective and radiative resistance. Cooldown rates with and without a one-half inch air gap between the pipe (shell) and insulation were determined. The air I gap resistance was incorporated into the insulation resistance using an equivalent thermal conductivity. Also, three types of heat transfer coefficients were used in determining the internal convective resistance: free convection within enclosed spaces, forced convection with V = 60.2 ft/s (100% full RC flow), and forced convection with V =.18 ft/s (.3% full RC flow). Note: the coolant is assumed to cooldown only due to heat transfer affects and not mass transfer affects, i.e., the loop ie considered completely stagnant. Figure B3 is a ir. atrix I summarizing the cases investigated. Figures 81 and B2 show the models for the hot leg pipe and OTSG head cooldown analyses. The future temperature of a volume, i, is calculated from the following equation. in)i at present, 1 I T =T out future, i (pV C )i p where fpresent,i=presenttemperatureofthevolume, = heat flowing into and out of the volume, Qin' Oout at = ti e step, p, C = density and specific heat of the volume (these may be temperature p dependent),and V = volume of the volume, i. B-2

The heat transfer rates Q are found using the present temperatures and in' out the thermal resistances, i-1 i (except for i=1, in whien case, h = 0) h,i = in in 1-1 (except for i=4, in which case, T =T Qout,i = T, - Tg1 gg ambient) R where R = thermal resistance. The themal resistances are calculated by: RA = 1 hA Ag where h = internal convective heat transfer coefficient (calculated from either free or forced convection correlations), and A = internal heat transfer area. A R ' C " 2n K B H I.E ( where (y /y ) = outer to inner radius ratio, g j I.E = thermal conductivity of the insulation or equivalent themal K conductivity of the insulation and 1" air gap, and H = height of the volume 1 R = 4 (h th )A g r 4 h = external free convection coefficient, o 4 4 c(T4 -TA ), and h = radiative coefficient = r (T -T) 4 A A4 = external heat transfer area. I B-3

I The program takes an initial temperature distribution and calculates new temperature profiles with each succeeding time step using the finite difference equation. I . I I I I e l I B-4

I Figure 81 GEOMETRY & O!MENSIONS OF THE ll0T LEG COOLDOWN PROBLEM . I INSULATION ~' CONTAINMENT ATMOSPHERE AIR GAP COOLANT HOT LEG PIPE "i k kg NA TA h g T; o [ 9 N 1.5' H = 1.5' i ARBITRARY

3 SELECTION HH E

Xa ;= =l 4.5" T T4 l 7' T3 N - N .= N N-A-T2 -N = = i INTERNAL PIPE AIR GAP INSULATION EXTERNAL CONVECTIVE CONDUCTIVE CONOUCTIVE CONDUCTIVE CONVECTIVE l RESISTANCE RESISTANCE RESISTANCE RESISTANCE & RADIATIVE RESISTANCE 'I .I ~ I f I I y e.s

I I \\ \\ figure 82 GEOMETRY & OlMENSIONS OF THE OTSG SHELL C00LOOWN PROBLEM I I . I AIR GAP I INSULATION OTSG SHELL COOLANT kA CONTAINMENT ATMOSPHERE T "8 k l iE hi J Ti o 5 4.98' i 5.63' l = e-X g ( . 4.5" ! I [ Tj T T3 T TA h

4 h 2

h I INTERNAL OTSG SHELL AIR GAP INSULATION EXTERNAL CONVECTIVE CONOUCTIVE CONOUCTIVE CONDUCTIVE CONVECTIVE nESISTANCE RESISTANCE RESISTANCE RESISTANCE & RADIATIVE RESISTANCE I I

I . I Figure 83 COOLDOWN RATE MATRIX FOR HOT LEG AND OTSG HEAD C00LDOWN . I P I BASIS FOR INTERNAL CONVECTIVE HEAT TRANSFER COEFFICIENT WITHOUT 1/2" AIR FREE CONVECTION FORCED CONVECTION FORCED CONVECTION GAP WITHIN ENCLOSED V = 60.2 FT/S V =.18 FT/S SPACES WITH 1/2" AIR FREE CONVECTION GAP WITHIN ENCLOSED V = 60.2 FT/S V =.18 FT/S l SPACES

g

I t

I 'I I I I B-7

'I I !.II I lg 'I I APPENDIX C NSSS Modeling I .g I 'I I .I 'I

I I

C-1 I

I I I The Midland single loop natural circulation analyses used a RELAP 4/ Mod 6 1 Version 13.1 model. A syninetrical two-loop model arrangement was assembled, composed of: 46 control volumes 80 flow junctions 40 heat transfer slabs and 2 homologous RC pumps Figure C1 shows the basic noding scheme. Volumes Eight non-equilibrium volumes in the primary system are used to model the pressurizer, hot legs, OTSG inlets, RV head, and the upper RV internals. Equilibrium volumes are used for the remaining twenty-six primary volumes. Six equilibrium volumes are used on the secondary side of each OTSG. Four par OTSG are active steam generating volumes, while the other two act as the downcomer and steam annulus volumes. Eight equilibrium volumes per OTSG (16 total) are used for the tube volumes. Radial tubing is modeled by apportioning them into two columns of 4 volumes each, and one column with 90% flow area and the other with 10%. By controlling the heat transfer from selected volumes, the AFW " shadowing" phenomena can be approximated, i.e., the non-uniform AFW wetting of the tube bundle. Boron transport studies can be made concurrent with other events. The RELAP treatment of boron concentrations is a simplistic averaging within a volume, i.e., there is a constant mixing of the incoming borated stream with the existing liquid in the volume. The detail of the noding thus affects the ability to model transport or hideout of boron. At initialization, the boron concentration in the RCS was taken as an E0L value of 17 ppm. High pressure injection has the borated water storage tank concentration while makeup was used with 17 ppm. I I C-2 I

Junctions Fourteen fill paths of eight types are used to model those flows which do not connect adjoining volumes. Of these, half are controlled inflow (fill) paths: I makeup, HPI-A, HPI-B, MFW-A, MFW-B, AFW-A, and AFW-B. The other seven are controlled outflow paths: Main Steam A and B, letdown, Main Steam Safety Valve (banks 2-4 combined), A and B, and the initialization aspirator flow paths on each OTSG. Eight leak paths use the homogeneous equilibrium model to show choked flow through sized orifices. By this means the following can be modeled: candycane high point vents, A and B, the electromatic relief valve, the pressurizer safety valves, the turbine bypass and atmospheric dump valves, A and B, and the bank 1 Main Steam Safety Valves, A and B. A control on level is modeled for the MU (PZR) and EFW (OTSG). The total liquid mass for these volume (s) is sensed and used to control the fill paths. Dead banks and time-hysteresis are also built into this model. Two junction-based options are inter-related. The enthalpy transport model is used for all junctions that are connected to volumes which actively undergo heat transfer. As an example, an OTSG tube volume will have associated junctions with different enthalpies. The outflow path will.not have the average enthalpy of the volume. Instead, its enthalpy and state are a function of heat transfer g 5 to the volume, flow through the volume, incoming junction enthalpy, and the average volume enthalpy. This option was used for the two core paths and for forty-two OTSG-related junctions. An option which is related to enthalpy transport is the vertical slip model. Vertical slip allows water and steam to separate and displace one another between vertically adjacent volumes. Without it, the code would show a " pancaking" of trapped water over steam and an inability for normal phase separation within a stack of volumes. When larger mass flowrates are involved, i.e., in the OTSGs, use of flowpath pairs which are vertically offset assures the continued I perfonnance of the vertical slip option. The lower junction of a pair would allow water to slump while the upper one of the pair allows steam to rise into the next volume up in the stack. Thus, each of the four active volumes on each OTSG's secondary side is connected to its adjoining stacked volume by a pair of flowpaths. This allows vapors to rise from volume to volume, displacing the water which would otherwise " pancake" in those volumes. I C-3 I

/ ( Slabs I ) Of the forty heat slabs, two are used for the core, two are imersed in the upper I plenum volume, tuen,ty, serve as the OTSG tubes and tubesheets, and the remaining seventeen slabs ar.e crimary piping or pressurizer shell metal. Individual component geometries (can be specified, along with the composition and lamination ' structureoftheilads. The primary metal slabs are used to model heat losses from thoRCS. One input I is an assumed heat loss tenn for the external surface of the piping. The second input is an assumed fraction of total time-zero core output lost through that segment of piping., The code then detennines an outside temperature and varies the overall heat loss rate as fluid conditions "within" the piping vary. Other Features The aspirator model used is actually a set of two dissimilar junction types per OTSG. Each type is used at a different segment of the analysis. During r initialitation and stabilization at 100% power, a negative fill path acts as the aspiration port. The mass and enthalpy thus removed from the OTSG are balanced by inputting a MFW flow which is already saturated and at the total mass flowrate. Once the MFW and this initial aspirator type are ramped back after reactor trip, another aspirator model is used. A junction which is located between the j! downcomer and the steam generating volumes gradually ramps open. This allows an interchange of mass between these volumes, similar to the actual aspiration port. The pressurizer has a group of heater banks capable of mimicking the actual heater banks. Their location, energy input, and setpoints are derived from the plant specifications. r l I I I C-4 I

~ i 3 O T@ m + 3 gi t =4 I g @4 t@ @t i, O c. @- ad G ,~ 8 m g lij -jj ~ / o n l I ( I ~ = T 5 S T i E 2 l @ m e t 9 m o t E / ) e I j@ j Q e = c ( ~ g ( E O q 2: w , n = n n. 6 6 c d <dt_L + e fg I t '1 5 5 u 5 6 E e ~ E @ @b

• il E

S a i-I@ a l@ m ,~ ~ T ~ e g ~ ~ 2@l @t I 5a: e ~ o y - @t @d f @t l@ @4 g o i I I + pg - + i C-5 I --}}