UPTEC ES10004
Examensarbete 20 p April 2010
Thermal-hydraulic modelling of Forsmark 1 NPP in TRACE
Validation versus the 25th of July, 2006 plant transient
Lisa Bladh
Teknisk- naturvetenskaplig fakultet UTH-enheten
Besöksadress:
Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0
Postadress:
Box 536 751 21 Uppsala
Telefon:
018 – 471 30 03
Telefax:
018 – 471 30 00
Hemsida:
http://www.teknat.uu.se/student
Abstract
Thermal-hydraulic modelling of Forsmark 1 NPP in TRACE
Lisa Bladh
There is a widespread use of thermal hydraulic codes in nuclear industry. The codes are used to analyse the transient and steady-state behavior of the nuclear power plants. The US Nuclear Regulatory Commission that has long experience of developing such codes are now incorporating the capabilities of their earlier codes into one modern simulation tool, called TRACE. The code is under development and validation work is required especially in the field of BWR applications. Eventually the code is expected to replace similar codes such as TRAC and Relap5.
With this in mind, a TRACE model of Forsmark 1 has been set up to investigate how well it can simulate a plant transient. On the 25th of July, 2006 there was a
disturbance at Forsmark 1 that caused the RPV water level and pressure to decrease.
In this project, plant data acquired during the event are used to validate the model of Forsmark 1. The validation work is focused on comparing measured and calculated water and pressure levels in the RPC during the transient.
The results show qualitatively good agreement with the validation data, however during a period of the simulations there are large discrepancies concerning the pressure and water level in the RPV. In total, 13 simulations are performed, studying the influences of parameters such as simulation time-step size, the feed water flow boundary conditions and the steam line isolation valve characteristics. Based on the results of the simulations, a number of recommendations are made regarding suggestions for further work.
Sponsor: Forsmarks Kraftgrupp AB ISSN: 1650-8300, UPTEC ES10004 Examinator: Ulla Tengblad
Ämnesgranskare: Michael Österlund Handledare: Farid Alavyoon
2 (62)
Sammanfattning
Simuleringsverktyg i kärnkraftbranschen
I detta examensarbete har en modell av Forsmark 1 byggts upp i beräkningsprogrammet TRACE.
Programmet är utvecklat av den amerikanska myndigheten NRC (United States Nuclear Regulatory Commission), som har 30 års erfarenhet av denna typ av beräkningskoder. Vid utförandet av examensarbetet användes inte TRACE av Forsmarks Kraftgrupp AB och examensarbetet sågs som ett sätt att introducera programmet i företaget.
Att på ett korrekt sätt kunna analysera ett kärnkraftsverks funktion både vid normal drift och under störningar är av stor vikt för beslut som rör design, drift och säkerhet på ett kärnkraftverk.
Av naturliga skäl kan man inte i anläggningen testa hur man klarar stora rörbrott eller stopp av viktiga säkerhetssystem. För att ändå kunna analysera om reaktorn kan tas till säkert läge vid denna typ av händelser används simuleringsprogram i vilka modeller av kärnkraftverk sätts upp.
Med hjälp av dessa program kan händelseförloppen vid olika störningar och haverier simuleras varvid parametrar som till exempel tryck, temperatur och nivå i kärnkraftverket kan beräknas. Av Forsmark Kraftgrupp AB används idag ett flertal sådana program och modeller.
Forsmarksincidenten – ett intressant valideringsfall
Att jämföra mätdata från en inträffad störning med simulerade värden är ett bra verktyg för att kvalitetssäkra modellen och på så vis försäkra sig om att den återger anläggningens beteende på ett korrekt sätt. Den 25 juli 2006 inträffade en störning på ett av Forsmark Kraftgrupps
kärnkraftverk. Störningen rönte stor uppmärksamhet i media och från myndigheterna och har kommit att benämnas Forsmarksincidenten eller 25:e juli händelsen. Målet med examensarbetet var att sätta upp en modell av Forsmark 1 i beräkningsprogrammet TRACE och undersöka hur väl modellen kunde simulera 25:e juli händelsen, med fokus på tryck och nivå i reaktortanken.
Störningen hade sitt ursprung i en kortslutning i ett ställverk utanför anläggningen vid ett arbete som utfördes där av Svenska Kraftnät. De spänningsfluktuationer som skapades slog ut viktiga komponenter i anläggningens utrustning för obrytbar strömförsörjning. Som följd av detta startade inte två av fyra dieselgeneratorer som ska säkra strömförsörjningen till anläggningens
säkerhetssystem. Under cirka 30 minuter förblev delar av anläggningen strömlösa tills kontakt med ett lokalt nät hade etablerats. Under förloppet sjönk både tryck och nivå i reaktortanken väsentligt.
Modellbyggnad
Modellen av Forsmark 1 är baserad på indata hämtat från ritningar och annan dokumentation som beskriver anläggningens design och funktion. Modellen innefattar huvudsakligen reaktortanken och dess interndelar, samt anslutande ångledningar och kylvattensystem. Som randvillkor till modellen har mätdata från händelsen använts, så som matarvattenflödet in i reaktorn,
cirkulationspumparnas varvtal och den utvecklade effekten i härden.
Vid uppbyggnad av en reaktormodell finns ett antal komponentmodeller till förfogande i TRACE komponentbibliotek. Pumpar, ventiler, rör och ångseparatorer är exempel på sådana komponenter.
Dessa komponenter har kopplats samman i ett grafiskt användargränssnitt (SNAP). I detta användargränssnitt kan sedan indata såsom initialtillstånd och geometrier matas in vilket
genererar ett de indata som behövs till beräkningskoden TRACE. Simuleringsresultaten kan sedan
studeras med hjälp av ett grafritande program (AptPlot) som kan plotta de beräknade parametrarna såsom tryck, temperatur och flöde.
Resultat
Kvalitativt överensstämmer simuleringarna väl mot valideringsdata. Dock förekommer stora avvikelser mellan simulering och valideringsdata strax efter den inledande händelsen. Till
exempel faller nivån under de första 200 sekunderna avsevärt lägre än vad som mättes upp under störningen. I simuleringarna uppkommer även en tryckstegring i reaktortanken som inte kan återfinnas i valideringsdata. På grund av dessa skillnader mellan simulerade värden och
valideringsdata kan modellen inte ännu anses validerad, utan ytterligare utvecklingsarbete krävs.
Parameterstudier
Ett flertal simuleringar har utförts där olika indata har varierats för att undersöka hur stor inverkan ett fel i indata har på simuleringsresultaten. Fyra olika sådana så kallade parameterstudier har utförts. Simuleringarna visar att olika storlek på de tidssteg som används i beräkningen bara har marginell påverkan på resultaten. Ventilkarakteristiken i ångledningarnas skalventiler har varierats och simuleringarna visar att ju snabbare ventilerna stänger desto högre blir trycket i reaktortanken. Skillnaderna mellan simuleringarna är dock marginella och alla fall uppvisar den inkorrekta trycktoppen strax efter inledande händelse.
I modellen har det matarvattenflöde som uppmättes under händelsen använts som randvillkor.
Detta randvillkor är dock behäftat med en viss osäkerhet då utrustningen som mäter flödet blev spänningslös under tio sekunder innan matarvattnet stängdes av och flödet därmed inte är mätt korrekt under denna period. En parameterstudie har utförts där flödet har varierats under dessa tio sekunder. Simuleringarna visar att nivån inte faller lika lågt under det inledande förloppet då ett större matarvattenflöde under dessa tio sekunder antas. I alla dessa simuleringar faller dock nivån mer än väntat och fel i randvillkoret som föreskriver matarvattenflödet kan inte förklara den hastiga nivåsänkning som uppkommer i simuleringarna.
Det kylvattensystem som fyller upp reaktortanken i slutet av händelsen är modellerat som en funktion av trycket i reaktortanken då systemet är uppbyggt av centrifugalpumpar vars flöde är starkt tryckberoende. Nivån stiger i simuleringarna snabbare än under själva händelsen vilket antyder att flödet är för högt. Det finns inga mätdata av flödet att jämföra med. Parameterstudier har utförts där det tryck som styr flödet har varierats. Simuleringarna visar att nivån stiger med korrekt hastighet om trycket i reaktortanken inte fallit lika snabbt. Detta överensstämmer med att trycket är simulerat något lägre än det tryck som uppmättes vid transienten vilket gör det troligt att härdkylsystemet är korrekt modellerat.
Rekommendationer
Då modellen inte är fullt validerad krävs ytterligare utvecklingsarbete. Delar som kan förbättras är en justering av flödesmotståndet i härd, bypass och ångseparatorer. Den mängd av kallt vatten som injiceras i reaktortanken via de tankar som pressar in styrstavarna modelleras inte här vilket kan implementeras och inverkar på nivå och tryck i reaktortanken. Utöver detta kan modellen utvecklas med mer detaljerade ångledningar samt värmestrukturer i reaktortanken.
4 (62)
List of contents
1 Introduction 6
1.1 Objectives of this study 6
1.2 Background 6
1.3 Method 7
2 Two-phase flows 8
2.1 Two-phase flows in boiling-water reactors 8
2.2 Modelling two-phase flows in TRACE 9
3 The 25th of July event 10
4 Model description 12
4.1 The reactor vessel and its internals 14
4.2 Surrounding systems 26
5 Simulations 38
5.1 Reference simulation 39
5.2 Time-step size 39
5.3 Steam line isolation valve characteristics 40
5.4 Pressure evolution and emergency core cooling system flow 40
5.5 Feed water flow 41
6 Results 43
6.1 Reference simulation 43
6.2 Sensitivity to time step size 46
6.3 Sensitivity to steam line isolation valve characteristics 46 6.4 Pressure dependence of the core emergency cooling system model 46
6.5 Sensitivity to the feed water flow boundary condition 47
7 Conclusions and suggestions for further work 48
Appendix
Appendix 1 Results
Appendix 2 Reactor vessel nodalisation Appendix 3 Pumps
Appendix 4 The emergency core cooling system 323 Appendix 5 The auxiliary feed water system 327 Appendix 6 The residual-heat removal system 321 Appendix 7 The pressure relief system 314
Appendix 8 The 25:th of July event
Appendix 9 Measurements from the 25th of July, 2006
Abbreviations
APRM Average Power Range Monitoring
BWR Boiling-water Reactor
LOCA Loss of coolant accident
NPP Nuclear Power Plant
NRC United States Nuclear Regulatory Commission
POLCA Power On Line Calculation
RPV Reactor Pressure Vessel
SAR Safety Analysis Report
SNAP Symbolic Nuclear Analysis Package
SCRAM Safety Control Rod Axe Man – emergency stop by insertion of control rods
TRACE TRAC/Relap Advanced Computational Engine
UPS Uninterruptible power supply.
6 (62)
1 Introduction
1.1 Objectives of this study
The purpose of this master thesis is to investigate how well a thermal hydraulic model of Forsmark 1 NPP in the system code TRACE [1] can simulate the transient of the 25th of July event at Forsmark 1[2]. A simplified model, with focus on the reactor pressure vessel (RPV), is set up in TRACE. The computational results are compared with the measurements in the plant.
The focus of the work is on the validation of the water level and pressure in the RPV during the transient.
1.2 Background
1.2.1 Safety analysis and simulations at Forsmark
There is a large interest in the nuclear industry in being able to simulate and analyse transient and steady state behaviour of nuclear power plants. The results from these analyses can be used as a basis for decisions concerning plant operation, design and safety.
Due to the nature of nuclear power it is requested that the safety functions of the power plant are functioning at all times. Every safety function depends on a number of safety parameters for which safety limits have been calculated during plant design. To guarantee that a safety function is fully operational, sufficient margins to these safety limits have to be withheld. These limits can concern maximum pressure in the reactor containment or how much water that may be lost during a large pipe break. A vast number of analyses concerning these safety limits are performed, both periodically and when the plant is modified. In safety analyses, conservative assumptions are always used, making sure that margins to the safety limit are maintained even during the worst possible scenario.
At Forsmark, there is a wide range of simulation tools used for deterministic safety analyses.
MAAP [3] is used for severe accident analyses, Matstab [4] for core stability analyses and Bison [5] for dry-out analyses. TRACE [1], Relap 5 [6] and Apros [7] are thermal-hydraulic codes, specialised to perform process analysis and loss of coolant accident analyses.
At the start of this master thesis the software TRACE was not in use at Forsmark. This master thesis is a first step to introduce the program at Forsmark, with the goal to use it for safety and process analyses in the future. Since the conservative principles that distinguish a safety analysis are not used in this thesis, it could not be compared with a safety analysis. However, the
geometries of the model and the lessons learned from this thesis would be of great help in developing future models used for such work.
1.2.2 TRACE
TRACE is an advanced best-estimate thermal-hydraulic code being developed by the United States Nuclear Regulatory Commission (NRC). NRC has a 30 year long experience of developing different codes for neutronic-thermal-hydraulic analyses. At present, the goal of NRC is to
incorporate the capabilities of earlier codes such as Relap 5, TRAC and RAMONA into TRACE as one modern simulation tool [8]. In TRACE, multidimensional two-phase flows, nonequilibrium thermodynamics, reactor kinetics, heat transfer and level tracking can be modelled.
1.2.3 Summary of the 25th of July event
The model built in this thesis is validated against measurement data from the plant disturbance that occurred on Forsmark 1 on the 25th of July, 2006 [2], in this work referred to as the 25th of July event. During the disturbance, a short circuit outside the plant caused severe voltage
fluctuations that in turn lead to disconnection of the plant from the outer grid and a high-voltage transient in the internal power supply system. The voltage peak knocked out components in the uninterruptible power supply system. This sequence of events eventually resulted in a full scram of the reactor, shutting down the power and reducing the recirculation flow. The reactor
containment was isolated and emergency core cooling was carried out along with pressure relief.
The tripped components in the battery backup system were the reason why two of the four diesel generators did not connect during the disturbance. This meant that only two of the four emergency core cooling circuits were working initially. During this event lasting around 30 minutes, the water level fell from 4 meters to a lowest value of 1.9 meters above the top of the core. The pressure was decreased from 70 bar to 6 bar [2].
1.3 Method
The modelling of Forsmark 1 was performed in the software TRACE 5.0 (solver), using the graphical interface SNAP 0.27.4 (pre-processor) and the plotting tool AptPlot 6.1.8 (post- processor).
Drawings and plant design documentation of the reactor vessel and its internals were studied. This information was used as input data in the different component models from the TRACE
component library. By adding simple models of the feed water flow and the main steam lines and its automation systems the model could simulate the reactor vessel in steady state operation, presenting correct vessel pressure and water levels. The core power was modelled as a simple power source, not taking feedback effects into account. The power level was specified as a function of time.
The power plant subsystems that affected the pressure and water levels in the reactor vessel during the disturbance on the 25th of July were listed. Simple models of these systems were developed. Plant data measured during the transient was used in these models to specify mass flows or valve stem positions as functions of time. No automation of these systems was modelled;
hence these models cannot be used to simulate a general transient. Due to the nature of the transient with power loss in some measurement equipment, measurement data for some components were not available. The states of these systems were derived from plant design documentation and the signalling during the disturbance in the power plant. This derivation introduced uncertainties concerning the time when some systems started to operate during the transient.
Several simulations were performed and the RPV pressure and water level from the simulations were compared with measurement data. Finally, the results were analysed.
8 (62)
2 Two-phase flows
2.1 Two-phase flows in boiling-water reactors
Water has a central position in nuclear power plants, enabling energy transport, protecting the core from core damage and providing means to control the power of the plant. In boiling-water reactors, water exists in both its liquid and gas phase and the two phases interact in a complex way. Some important physical phenomena of two-phase flow are presented below.
2.1.1 Two-phase heat transfer and phase changes in a BWR
When feed water enters the reactor vessel it is subcooled more than 100°C. The degree of
subcooling describes the difference between the saturation temperature and the bulk temperature [9]. The temperature is increased in the downcomer as the feed water is mixed with the
recirculated reactor water. As the water enters the core it will be heated to saturation temperature.
Vapor bubbles will start to form along the surfaces of the fuel rods through nucleate boiling. The flow patterns will then transform from bubbly flow,
via slug flow to annular flow as heat is transferred to the fluid. The different flow patterns and the heat transfer regions can be viewed in Figure 2-1.
When annular flow is reached and the liquid film is thin enough the heat will transfer through the liquid and the vaporisation will occur at the liquid vapour interface [9]. If the liquid film evaporates completely, a state called dryout is reached. It is of great
significance to avoid dryout in the core, because it is accompanied by a significant temperature rise in the cladding of the fuel rods that would damage the cladding and allow leakage of radioactive substances.
Water boils when its temperature reaches saturation temperature. The saturation temperature of a liquid varies with pressure. The water in the downcomer of a RPV is subcooled and hence does not boil. However, if pressure relief is applied and the ambient pressure is decreased, the saturation temperature decreases and may reach the temperature of the water in the downcomer. Consequently the water starts to vaporise. This phenomenon is called flashing.
Figure 2-1 Heat transfer regions and flow patterns in convective boiling [9].
2.1.2 Void
Void is defined as the volume fraction of steam in a steam-water mixture [10]. By controlling the void it is possible to control the power of the nuclear power plant. In practice, this is done by varying the mass flow from the recirculation pumps[10]. A decreased recirculation flow results in decreased core cooling, a larger void, less thermal neutrons and thus decreased power.
2.2 Modelling two-phase flows in TRACE
Two-phase flow calculations are much more complicated than single-phase flow. Water and steam interact in a complex way that can often only be described with empirical correlations.
Principles of mass, momentum and energy conservation are still valid, but need to be calculated for both phases. It is also necessary to keep track of mass transfer between the phases and the transfer of momentum associated with this mass transfer [8].
The differential equations being solved by TRACE originate from the Navier-Stokes field equations that describe the conservation of mass, energy and momentum in a fluid system.
Separate equations are set up for the liquid and gas fields, equations that are adjusted with
empirically based correlations to create jump conditions between the two phases. The differential equations describing the system are solved in TRACE through finite-volume numerical methods.
Detailed information about the solving methods can be found in the TRACE theory manual, reference [8].
2.2.1 Simulation time-step size
The simulation time step in TRACE is calculated by a time step selection algorithm. The
computer execution time is a function of the maximum time step size, the number of cells, and the rate with which the thermal-hydraulic phenomena changes [8]. As the differential equations are solved numerically, the time step size is limited by the material courant stability limit that is a convergence condition for the calculations. The time step size Δt is limited as follows:
V t≤ (Δx) Δ
Here, Δx is the cell length and V is the flow velocity. In TRACE, limitation by the Courant limit can be partly avoided by using a stability enhancing numerical method called SETS. However, the method operates at the expense of a higher numerical diffusion [8]. In the simulations performed in this diploma work, the SETS method is used.
10 (62)
3 The 25th of July event
When modelling Forsmark 1 and the 25th of July event, the flows in some of the supporting systems had to be derived from plant documentation and the signalling in the plant during the disturbance. A more detailed description of the event is provided in this section and a second by second list of what happened in the plant is listed in appendix 8 together with the graphs
presenting some of the measurements from the event in appendix 9. At disturbances at Forsmark, a disturbance printer will start saving detailed information from the plant 60 seconds prior to the disturbance and onward. All results in this section and the second by second list that are derived directly from the measurement data saved by the disturbance printer at Forsmark 1 are marked (MD).
The origin of the incident was an incorrect switching operation in the 400 kV switchyard. This lead to an electrical arc and a two-phase short circuit. As a result of the short circuit there was a large voltage drop which in turn led to tripping of both the generator circuit breakers [2]. The plant was now disconnected from the external grid, a condition which automatically leads to a partial scram of the reactor [12]. During a partial scram, half of the control rods are inserted into the core reducing the power to about 30 % and the speed of the recirculation pumps are reduced.
The plant switched to house load operation getting its power supply from its own two generators and dumping excess steam to the condensers ([2] & [13]). At this moment, Forsmark 1 got its power supply from the two plant generators. Starting signals were issued to all diesel generators at the disturbance.
The short circuit in the switchyard and the disconnection from the grid caused a brief but substantial over voltage on the internal electrical network of the power station [13]. The over- voltage transient knocked out both rectifiers and inverters in sub A and B uninterruptible power supply system [2]. The UPS system ensures a no-break supply of alternating current for a period of at least two hours, by connecting back-up batteries to the plant grid [13]. Because the batteries are DC units, they are charged through a rectifier and supplies power through an inverter. Both the rectifiers and inverters in sub A and B tripped due to their component protection as the transient exceeded the UPS design specifications, which disconnected these batteries from the electrical system of the plant [13]. At this time however, the loads connected to the battery-secured grid still got their power supply from the main generators [2].
The next thing to happen was that turbine 1 tripped due to low governing oil pressure ([2] & MD).
Shortly afterwards the same turbine was inhibited to dump its steam to the condenser. As the station was on house load operation, the voltage in the A/C sub of the station was decreasing with the decreasing frequency of the generator. When the voltage had fallen under 90 % there was an automatic changeover to direct supply to the battery secured grid on sub A. During the
changeover, the sub A of the battery-secured grid was without power for two seconds because the grid was not actually connected to the batteries. The battery-secured grid supplies power to much of the instrumentation at Forsmark 1, such as equipment that measures the reactor vessel pressure and water level. This equipment is divided in four subs, with pressure and the reactor water level being measured by each sub. At loss of power the measurement values alternates between high and low values before settling at the lowest value in the measuring range [2]. During these two seconds without power in sub A the level measurement instrumentation incorrectly showed a high water level in the reactor vessel, which tripped the channel A of the emergency shutdown logic train (SS5).
The frequency of the voltage supplied from generator 1 was decreasing to even lower values as the turbine slowed down. At this time the normal supply circuit breakers in subs A and C opened
due to the low frequency and disconnected the generator from the 500V diesel-secured grid to switch to power supply from the diesel generators. The diesel generator in sub C connected.
However, because the connecting of the diesel generators is dependent on the power supply from the battery secured grid, the connection in sub A failed. Instrumentation supplied by the battery- secured grid in sub A was once again without power, which resulted in further fluctuations in the measurements of the pressure and level in the reactor vessel before it dropped to zero.
The second turbine, connected to sub C/D, tripped on high pressure in the condenser about 30 seconds after the first turbine had tripped. Similarly to what happened in sub A, the now sinking voltage (<90 %) from the second generator caused a changeover to direct supply of the Sub B network. The power supply to the instrumentation in sub B was gone for two seconds during the changeover. The loss of power resulted in level fluctuations of the measurement instrumentation values which triggered the emergency stop channel in sub B on high water level. As the
emergency stop had been triggered in two out of four channels these signals were automatically followed by a full reactor scram. The fluctuations also triggered the blow-down chain as well as the reactor containment isolation chain.
The power supply for the 6 kV A/C busbar switched to the 70 kV outer grid as the power from generator 1 had fallen below 5 MW and the generator circuit breaker had tripped. Because the normal supply circuit breakers in sub A were open, the diesel- and battery-backed busbars in sub A were still without power. The situation in the B/D subsystem developed the same way as in subsystem A/C. The normal supply circuit breakers in sub B and D disconnected the diesel- backed busbars to switch to power supply from the diesel generators. The diesel generator in sub D connected, while the diesel generator in sub B failed to do so. There was a power supply changeover for the 6 kV C/D busbar to supply from the local 70kV outer grid instead of the second main generator as it also provided too low power. The battery- and diesel-backed busbars in sub A and B were powerless for 21 minutes before the supply to these systems was manually restored by connecting them to the fully functioning 70 kV local grid [2].
When the scram occurred, the containment valves in feed water and main steam lines were closed and the feed water pumps stopped to isolate the reactor containment. Instead, the core was cooled by the auxiliary feed water system supplied from sub C and D and the pressure was relieved by the pressure relief system that condensates the steam in a condensation pool in the reactor enclosure. The full flow from the auxiliary feed water system was established when the diesel generators were manually connected to the internal power supply system 22 minutes later,
restoring all functional and supervisory facilities powered by sub A and B. The normal water level was then quickly restored.
12 (62)
4 Model description
The simulation of Forsmark 1 during the 25th of July event is performed in TRACE. The model and the input data are set up in the pre-processor and the graphical interface SNAP and thermo- hydraulic components of the full model can be viewed in Figure 4-1 below. The following systems are included in the model:
• Reactor vessel and its internals; including recirculation pumps, steam separators, fuel assemblies and control rod guide tubes.
• Steam piping and pressure control system
• Feed water piping and level control system
• Residual-heat removal system
• Auxiliary feed water system
• Emergency core cooling system
• Pressure relief system
Figure 4-1 The full model of Forsmark 1 as shown in the graphical interface and pre- processor SNAP.
The model is formed by components from the TRACE component library. The different
component types can be viewed in Figure 4-2 and their application areas are presented in table 4-1 below.
Table 4-1 Short description the used components and their functions.
Component Function
PIPE Component used when modelling pipes.
VALVE Component used when modelling valves.
FILL Component used to prescribe flow as a boundary condition.
BREAK Component used to define pressure as a boundary condition.
VESSEL 3D component used when modelling the reactor pressure vessel.
PUMP Component used when modelling pumps.
SEPD Component used when modelling steam separators.
CHAN Component used when modelling the geometry, flow paths and heat conduction of BWR fuel assemblies.
POWER Component used when modelling power production.
Figure 4-2 The different TRACE components used when setting up the model of Forsmark 1.
In the following sections of this chapter the design and functions of the different components at Forsmark 1 will be briefly described. In addition, it will be described how this type of component is modelled by TRACE and finally what modelling decisions was made when setting up the separate components.
14 (62)
4.1 The reactor vessel and its internals
4.1.1 Introduction to the regions in a reactor vessel
In the lower part of the reactor vessel there is a large cylindrical assembly welded to the bottom of the vessel. The cylindrical assembly is called the shroud and separates the flows going downward in the downcomer and the flows going upward to the core. The volume within the shroud is characterized by its upper and lower parts. The lower part is called the lower plenum and contains the control rod guide tubes. There are holes through the shroud at the bottom of the vessel
allowing water to be pumped from the downcomer into the lower plenum. The pumps are mounted at the pump deck in the lower part of the downcomer. The upper part of the shroud encloses the core and its fuel boxes. The volume between the core shroud and the fuel boxes is called the core bypass. At the top of the core shroud there is a grid called the top guide that provides lateral support for the fuel boxes. At the very top of the core shroud is the core shroud head on which the steam separators are mounted. The volume above the top guide and within the core shroud head is called the upper plenum.
The region referred to as the steam dome is the volume within the vessel closure head. The moisture dryer banks are placed below the steam dome within the moisture dryer assembly. The moisture dryer assembly extends downward below the water level in the reactor vessel with a moisture dryer skirt that separates volumes of dry steam from the volumes of steam that is yet to be dried in the moisture dryer banks. There are a number of nozzles in the reactor vessel wall connecting to the feed water system, the emergency core cooling system, the residual-heat removal system and the steam lines.
Figure 4-3 The reactor vessel and its internal parts [21].
4.1.2 The TRACE vessel component
The vessel component is a model of the reactor vessel and is designed to simulate all the flows in and around the reactor internals. The flows are computed using a six-equation two-phase model.
The 3D vessel geometry, structure materials and placement of connecting nozzles can be described in the vessel component.
The vessel component is a meshed cylinder as illustrated in Figure 4-4. The user specifies the number of axial levels, radial rings and azimuthal sectors in the mesh, dividing the cylinder into mesh cells X[1]X. The coordinates of the mesh are chosen so that the mesh cell interfaces coincide with significant levels or surfaces in the reactor vessel to be modelled. The separate models for fuel boxes, steam separators and pumps can then be connected to the correct levels in the vessel component.
Figure 4-4 The vessel component, subdivided by axial levels, radial rings and azimuthal sectors X[1]X.
XFigure 4-5X is a picture of a mesh cell, demonstrating the interfaces resulting from an axial, azimuthal and radial meshing. A one-dimensional hydraulic component can be connected to any mesh cell interface of any mesh cell in the vessel component. However, the connection is always placed in the center of the interface surface of the mesh cell. When modelling reactor vessel nozzles at the correct axial levels, it is important to be aware of this fact.
As described in the TRACE manual X[1]X, the volumes and areas in the vessel are calculated on the basis of geometric mesh spacings. However, to be able to specify the volumes and flow areas that are occupied by structure within the vessel, the flow area fractions and volume fractions are specified for all mesh cells volumes and edges. Setting a flow area fraction to zero is equivalent to modelling an impermeable solid wall.
16 (62)
Figure 4-5 Mesh cell X[1]X.
4.1.2.1 Reactor nodalisation of the Forsmark 1 model
The reactor vessel has been axially divided in 27 levels. The vessel component has then been divided into two radial rings, one representing the center of the reactor vessel containing the core and the steam separators and the other ring representing the downcomer. This results in 54 cells in the reactor vessel. The user should be aware that SNAP and AptPlot do not number the levels of the vessel similarly. In AptPlot the bottom of the vessel is assigned as level one, while level one in SNAP is the top of cell one as illustrated in XFigure 4-4X. In this report, the SNAP convention of numbering levels will be used.
The levels have been chosen, so that the positions of the levels coincide with important structure positions in the reactor vessel. Similar mesh spacing in the vessel was preferred, resulting in extra levels in the downcomer, the lower plenum and the dryer assembly. The full list of axial levels and their positions is given in the appendix 2 and is also illustrated in XFigure 4-6X.
The radius of the reactor vessel model is intentionally made larger than the true reactor vessel.
This is due to the fact that the width of the downcomer varies along the height of the vessel. The larger flow areas and volumes are then adjusted with the flow area fraction and volume fraction to model the true geometries.
Levels 13, 16, 18, 19, 23 and 24 are chosen so that they coincide with the correct elevations of the nozzles of the level instrumentation, the feed water pipes, the steam lines and the residual heat removing system. The emergency core cooling system nozzle is placed slightly off its true position. This should be adjusted if the model should be used in analyses where this is of importance.
Figure 4-6 The nodalisation of the RPV model. The outer, second ring represents the downcomer and contains the model of the recirculation pumps. The inner ring contains the core, the steam separators and the control rod guide tubes. The flow paths in the vessel are marked in the figure. The flow to the core bypass enters through a hole on the side of the control rod guide tubes.
18 (62)
4.1.3 The recirculation pumps
4.1.3.1 The recirculation pumps and the recirculation system – general description
The recirculation system has two purposes. The first purpose is to cool the core by pumping water through it, removing the heat produced by the fuel and ensuring that there are sufficient margins to dryout in the core at all times. The second purpose is to control the thermal power of the plant by providing more or less moderation of the neutrons from the nuclear process X[14]X. A higher mass flow from the recirculation pumps provides better cooling and a better moderation which increases the thermal power.
The reactor water consists of a mixture of the incoming feed water and the water that has been separated from the two-phase flows in the steam separators and moisture dryers. The pumps cause a forced circulation, pumping the water from the downcomer into the lower plenum, through the core plate and the core, through the steam separators and back down through the downcomer. The steam that is produced in the core rises through the fuel bundles to the upper plenum, through the steam separators and the moisture dryers and up to the steam dome. The steam exits the reactor vessel to the turbine through the steam outlet nozzles on the side of the reactor vessel.
The coolant flow through the bypass and the fuel assembly originates from different flow paths in the lower plenum. The bypass flow is provided through flow paths in the control rod guide tubes, entering in the bottom parts of the lower plenum through holes on the sides of the control rod guide tubes and exiting to the bypass through the core plate. The fuel bundle flow finds its way from the bottom of the lower plenum in between the control rod guide tubes and through the core plate to the fuel bundles.
The eight recirculation pumps are mounted at the bottom of the downcomer, with its suction and discharge sides separated by a pump deck X[14]X. The pumps are placed evenly around the core shroud as presented in appendix 3.
According to ref X[14]X, the normal circulation flow at full power is 9700 kg/s. The maximum flow is 11450 kg/s at coast-down. The flow is regulated by the pump speed, which is also measured.
Further characteristics of the pumps are listed in appendix 3.
4.1.3.2 The TRACE pump model
The pump model in TRACE is not a mechanistic model, but simply describes the increase in momentum resulting from the pump action as an addition to the equation at the interface between two cells. There is no motor torque versus pump-impeller speed model in TRACE, instead the pump impeller speed is assumed to be an input to the model. According to the TRACE manual
X[1]X, the head and torque of the pump model are defined as functions of the fluid volumetric flow and pump-impeller speed. These functions are described in homologous pump curves that
describe all operating states of the pump by combining positive or negative fluid volumetric flows with positive or negative pump-impeller angular velocities. The homologous curves are
dimensionless to allow one set of curves to be used for pumps that are similar geometrically. In a pump model, both built-in and user-defined pump curves can be used.
According to the TRACE manual X[1]X, a significant degradation of pump two-phase performance occurs if the volume fractions of the least abundant phase in the pump is over 20 %. It is possible to include these degrading two-phase effects in the TRACE pump model. Degraded pump
performance parameters as head and torque are calculated from a two different sets of
homologous curves, one set for normal single phase performance and one set for fully degraded performance. The different sets are then combined using a weight parameter that is dependent on the volume fractions of the two phases. It is also worth mentioning that friction and form losses aren’t modeled from the cell center before and after the pump.
4.1.3.3 Modeling the recirculation pumps
The eight recirculation pumps at the bottom of the downcomer are all modeled separately. The pumps are in the model connected between level 1 and level 2 in the outer radial ring of the vessel component. The lengths of the pump models are not representing any real pump lengths, but are simply the distance between the two levels. The pump flow areas were taken from an earlier reactor vessel model by Granström X[15]X.The flow area fraction at level 1 radial ring 2 in the reactor vessel component is set to zero to represent the pump deck. The pump deck enables the pumps to raise the pressure from the downcomer to the lower plenum.
Due to the fact that homologous curves for the recirculation pumps in Forsmark 1 had already been developed X[16]X, the option of user-defined curves was chosen. These curves, presented in appendix 3 were developed for the program Relap 5. The curves can easily be translated from one program to the other, but the two programs divide the homologous curves into different segments.
TRACE defines four homologous curves with values from -1 to 1. Relap5 on the other hand divides the four curves into eight curves, four from -1 to 0 and four from 0 to 1. Table 3.3-1 in appendix 3 illustrates how to translate homologous curves from Relap5 to TRACE.
Because there were no fully degraded homologous curves available, the degradation of pump performance in two-phase flow was not modeled. This simplification is estimated to be valid, because the volume fraction of steam at the recirculation pumps in normal operation should be less than 0.20 X[16]X. According to earlier reasoning in section X4.1.3.2X the two-phase performance degradation should be minimal. Further listings and derivations of the input data for the
recirculation pumps are presented in the appendix 3.
4.1.3.4 The recirculation pumps during the 25:th of July event
During the 25th of July event, four of the pumps tripped due to loss of power supply and four pumps just reduced their speed due to the runback feature during a partial scram. The measured pump impeller speeds are shown in appendix 9. The measurement data has been used as input to the different pump models, however in a filtered form to reduce the amount of data. In areas with large gradients, all data has been used.
20 (62)
4.1.4 Steam separators
4.1.4.1 Steam separators at Forsmark 1 – general description
The steam separators mounted on the standpipes on top of the core shroud head separate the reactor water from the steam that is produced in the reactor core. The standpipes and the core shroud head are shown in XFigure 4-3X. The two-phase mixture entering the separators passes through a set of stationary vanes that give the flow a rotating motion. As water has a higher density than steam, the centrifugal force separates the steam and water and hurls the water to the separator barrel wall. The separated water is discharged through perforated holes in the separator barrel wall and is led back to the downcomer to be re-circulated through the core by the pumps.
The steam exits through the top of the steam separators X[16]X.
If steam separators were ideal they would separate steam and water perfectly. In reality, this is not the case. The mass-flow fraction of steam that is re-circulated to the downcomer is called carry under. Similarly, the water still mixed with the steam after the steam separators is called carry over.
There are currently 125 steam separators in Forsmark 1, gathered in groups of five that are surrounded by shroud plates as displayed in XFigure 4-7X. It is specified that the fraction of water in the water leaving the separators must not be greater than that the moisture dryers can reduce the moisture content to 0.1 % by weight. Also, the volume fraction of steam in the water that re-enters the downcomer must be less than 6 % at full reactor power X[16]X.
Figure 4-7 Steam separators X[21]X.
4.1.4.2 The TRACE steam separator model
The separator model in TRACE is treated as a black box, in the sense that there is no modelling of what happens mechanically inside the separators X[1]X. Only inputs and outputs to the separator are of interest.
Figure 4-8 The black box separator model
The flows through the separator are represented by one inflow, IN, and two outflows, EX and DIS, according to XFigure 4-8X. The DIS outflow represents the water that is discharged through the perforated wholes and the EX outflow represents the steam that exits through the top of the
separator. The carry under and carry over quality, defined in the equations below, can either be set to a constant value or as a function of the inlet quality. The separator model will then, according to the TRACE manual X[1]X, trigger a special solution of the field equations in order to fit the results, if possible, with the prescribed carry under and carry over. These calculations are enabled by assuming (i) the flow is balanced and there is no mass build-up or phase change in the
separator, (ii) the flow is homogenous, meaning that the water and the steam have the same
velocities trough the separator. Furthermore, the phasic densities upstream of the separator and the inlet quality need to be known. This information is gathered from the solution of the field
equations that lists calculated pressures and temperatures. If pure water or steam will enter the separator model, it cannot work as intended. The separator component will then switch to a simple pipe component with the same geometries.
Inlet quality
IN g IN l
IN g
IN m m
x m
, ,
,
&
&
&
= +
Carry over quality
EX g EX l
EX l
co m m
x m
, ,
,
&
&
&
= +
Carry under quality
DIS g DIS l
DIS g
CU m m
x m
, ,
,
&
&
&
= +
The indices g and l in the above equations denote gas and liquid respectively.
22 (62)
4.1.4.3 Modelling the steam separators at Forsmark 1
In the separator model the cell where the actual separation takes place, earlier referred to as the black box, can be connected to further piping on both sides. The steam separators at Forsmark 1 are modelled from where the stand pipes attach the core shroud head to the top of the separators.
The geometries represent the real inner measurements of one steam separator. The main pipe of the steam separator model connects to level 15 and 19 in the reactor vessel component. Because level 15 is adjusted to place the feed water nozzle at the correct elevation, the stand pipes are slightly longer than the true stand pipes. The discharge outflow exits through a side arm connected to level 16. The flow area of the side arm is equal to the total area of all the perforated holes. All separators are modelled alike, by simply prescribing that there are 125 separators.
Because no separator performance curves were available, the carry over and carry under was prescribed as constants based on assumptions. The carry under default value in TRACE is a weight fraction of 0.3%. As mentioned in section X4.1.3X, it is specified that the fraction of steam carried back to the downcomer at Forsmark 1 must be less than 6% by volume. This corresponds to a weight fraction of 0.315%, if calculated at 70 bar and 286°C. Therefore it is reasonable to assume a carry under weight fraction of 0.3%. The default value is also used for the carry over, which is specified to be below 10% by weight, so the default value of 5% by weight is used.
According to the TRACE manual X[1]X, great care needs to be taken when specifying the initial conditions for the separator model, concerning velocities and void fractions. Estimations of these values have been made, based on measurements of the steam flow and recirculation flow at normal operation. It is assumed that pure steam exited the top of the separator and pure water exited the discharge holes. It is also assumed, although incorrectly, that the steam and the water had the same velocities when flowing together. Densities for both water and steam are calculated at saturation at 70bar. Furthermore, no heat structures are modelled.
4.1.5 Fuel and fuel boxes
4.1.5.1 Nuclear fuel assemblies
Figure 4-9 A SVEA-96 nuclear fuel assembly X[18]X.
There are 676 fuel assemblies in the core of Forsmark 1. The majority of the fuel assemblies were of the type Atrium 10B at the time of the incident, which is a 10x10 BWR fuel designed by Areva. In the fuel, design features such as water rods, partial length fuel rods and spacers are utilised. A description of the design and function of BWR fuel assemblies are presented in reference X[10]X.
4.1.5.2 The TRACE fuel assembly model – the CHAN component
The CHAN-component from the TRACE component library is used to simulate the BWR fuel assemblies. Note however, that the power of the fuel is modelled by a separate power component.
In the CHAN-component, the geometries of the fuel assemblies are specified to model the thermal hydraulic volumes and to simulate how heat is transferred from the fuel through its cladding to the
24 (62)
in-channel coolant and from the in-channel coolant trough the fuel assembly wall to the bypass flow. The CHAN-component spawns and connects a number of subcomponents that together simulate the functions of a fuel assembly. Some of these subcomponents are listed below.
• The 1D pipe components that simulate the in-channel flow through the fuel assembly.
• The 1D pipe components that simulate the flow through water rods in the fuel.
• The powered heat structures in the fuel rods.
• The non-powered heat structures in the fuel assembly canister wall.
• The non-powered heat structures of the water rod walls.
• The radiation heat transfer in the fuel assembly.
Setting up a CHAN-component, the dimensions of the canister wall, the fuel, its cladding and the gas gap in between is specified. As the materials are specified, the heat transfer coefficients can be determined and the heat transfer can be calculated during the simulation. There is also an optional model to simulate radiation heat transfer, a type of heat transfer that is only significant if the fuel assembly is dried out. Possibilities for detailed heat transfer modelling of the fuel are available through optional gas gap and fuel cracking heat transfer models. The critical heat flux that will limit the heat transfer from the fuel cladding to the coolant can be calculated by either the Biasi or the CISE-GE correlation. Full boiling curves from AECL-IPPE CHF table are used for calculating the heat transfer coefficients for the subcritical states.
Figure 4-10 A nuclear fuel assembly cross section X[1]X.
Different fuel types of fuel assemblies including partial length fuel rods, and water rods of different designs can be specified.
The numbers of CHAN components that are used when modelling a reactor depends on the noding in the core region. A greater number of CHAN components could also be used to simulate different types of fuels and radial power distributions.
4.1.5.3 Modelling the fuel boxes at Forsmark 1
The purpose of the reactor model is to be able to study the reactor water level and reactor vessel pressure. The core should be of correct geometry and transfer the heat to the coolant correctly, in all other aspects the degree of detailing can be kept at a moderate level. As described in section
X4.1.6X the nuclear power is described as a simple heat source.
In 2007, 542 of the 676 fuel assemblies were of the type Atrium 10 B according to the program LADDA. Atrium 10 B is a 10x10 fuel designed by Areva. Being in majority, the geometry from this type of fuel was used when setting up all the fuel assemblies. To simplify the core model, the water rods and leakage flows from within the fuel core assembly to the core bypass has been omitted. Also, it has been chosen not to activate the radiation model. The radiation model is of interest when no coolant is available between the fuel rods.
4.1.6 Power-component
By specifying the materials of the structures in the CHAN-component, TRACE can calculate how heat is transferred through the fuel rods and the fuel boxes. However, the source of power is specified in a special POWER-component. A POWER-component can supply power to one or many CHAN-components. There are different levels of sophistication for the modelling of the total power emitted. The basic level is to define the power in a table. To capture specific features of nuclear power such as feedback due to void and fuel temperature, a point kinetics model could be specified. At the most advanced level, the thermal-hydraulic model in TRACE can be coupled to PARCS, which performs full 3D transient neutronics calculations X[1]X. In the Forsmark 1 model, the power is simply specified in a time versus power table. The plan is to replace this model with a point kinetics model in the future, but it was not feasible to fit this within the timeframe of the master thesis. A power level of 2.89GWt was used for the steady-state calculations, because it was the power level before the disturbance on the 25th of July, 2006. For the transient the
measurements from the APRM monitor were used. The APRM monitor measures the neutron flux in the core, hence it does not capture residual heat power from fission products. The omitting of residual heat in this simulation is not considered to be of great significance.
In the POWER-component it is also possible to specify axial and radial power distribution of fuel rods. If several CHAN-components are used to simulate many different fuel types or regions of different power levels in the core, it can be specified how the power is distributed among these CHAN-components. In this model, only one CHAN-component is used. A simple uniform axial power distribution shape has been specified.
26 (62)
4.2 Surrounding systems
The goal of this project is to model the reactor vessel and then try to validate it versus
measurements from a real transient in the reactor vessel at Forsmark 1. In order to perform this validation, some simplified models of the surrounding systems had to be included. However, most of these models are only intended to be able to simulate the specific transient from the 25th of July and are not supposed to simulate general transients at Forsmark 1.
Most of the data from the event on the 25th of July, 2006 have been available during this project.
However some data concerning the flows in the supporting system have not been available, partly due to the nature of the event with power loss on some sub channels and partly due to the fact that the disturbance printer do not save all the measured variables from the event. These data have been derived from the signalling during the event and from plant design documentation.
4.2.1 The feed water system 415
4.2.1.1 The feed water system at Forsmark 1
The main task of the feed water system is to pressurize, preheat and transport water from the condensers to the reactor vessel X[19]X. The feed water flow is controlled by the feed water control system that regulates the speed of the feed water pumps to achieve a RPV level of four meters above the core X[20]X. The reactor vessel water level is measured by the short range level
instrumentation described in section X4.2.3.1X.
4.2.1.2 Modelling the feed water system
The feed water pipes have been modelled by one single pipe, representing the four pipes in the feed water system that enters the reactor vessel. All four feed water pipes have the same flow area during the first 16 meters of piping. In the model they have been substituted by a single pipe of 16 meters with a flow area four times that of an individual pipe. The different orifices and pressure drops in the feed water piping has been omitted because the pipes are not modelled separately.
The piping from the auxiliary feed water system and residual heat removal system, described in section X4.2.5X and X4.2.6X respectively, connects to the feed water piping according to the plant design.
During the steady-state calculations, the feed water flow is controlled by a level control valve that regulates its flow area to achieve a reactor vessel level of 4 meters. However, it is desired to set the feed water flow as a boundary condition during the simulated transient. Hence, a valve has been added that isolates the level controller from the feed water pipe after 500 seconds of steady- state calculations. Instead, the flow is set according to a table of measurements used as input data to a FILL component from the TRACE component library.
Figure 4-11 The model of the feed water system as shown in the graphical interface SNAP.
The inflow from the residual heat removing system and the auxiliary feed water system connects to the feed water piping. The FILL component number 435 sets the feed water flow to the measured levels after 500 seconds of steady state calculations.
28 (62)
4.2.2 The steam line system 411
4.2.2.1 The steam lines and the RPV pressure controller
The purpose of the steam line system is to transport the steam from the reactor vessel to the high- pressure turbine or to the pressure relief system during disturbances. Eight steam pipes are connected by nozzles to the reactor vessel wall. The steam pipes are equipped with inner and outer containment isolation valves. The piping of the pressure relief system is connected to the steam pipes within the containment. X[21]X
The steam flow to the turbines is regulated by a control valve X[21]X that maintains the pressure in the reactor vessel at 70 bar during normal operation.
4.2.2.2 Modelling the steam lines and the reactor vessel pressure controller A simplified model was used of the steam piping and the turbines. The steam lines have been modelled as one single pipe, representing all eight outgoing steam lines from the reactor vessel steam dome. The flow areas of one steam line are multiplied by eight to represent the summed flow area of the pipes. Elevation changes, pipe bends and pressure drops in the system has not been considered. The valves in the pressure relief system described in section X4.2.7X are connected to the modelled steam pipe. There is a lot of potential to improve this system during further development work.
The reactor vessel pressure controller is modelled by setting up a control valve connected to the steam pipe that regulates the pressure in the reactor vessel to 70 bar by changing its flow area. On the low pressure side of the control valve there is a pressure boundary condition set to 62 bar.
During the transient on the 25th of July the inner containment isolation valves were active and closed off the steam flow through the steam piping. To simulate this, an additional valve has been set up. The position and geometry of the valve is not in accordance with the design of the inner containment valve. However, the closing time of the valve is in accordance with measurements from the plant, documented in reference X[22]X.
Figure 4-12 The model of the steam lines and the turbines set up in the graphical interface and pre-processor SNAP.
4.2.3 Level measurement system 536
4.2.3.1 The level instrumentation at Forsmark 1
The reactor vessel water level is measured indirectly by measuring pressure difference. The pressure at a certain level in the reactor vessel is compared to the pressure at the bottom of a reference tank that is filled with water. At the top of the reference tank there is an opening to the steam dome so the water in the reference tank as well as the water in the downcomer is subjected to the reactor vessel pressure. Hence, the pressure difference will depend on the levels of water in each tank.
The measurement method has the advantage that no equipment needs to be mounted in the reactor vessel and endure its tough environment. A drawback however, is that the density of the reactor vessel water varies and a certain pressure difference can correspond to different water levels depending on the pressure and temperature in the reactor vessel. Also the density of the water in the reference tank and the steam density will affect the measurements. To deal with this problem the measurements are adjusted in a density compensator, taking into account the effects of changing densities.
There are pressure measuring nozzles at three levels in the tank. The lowest nozzles are positioned 0.2 meters above the core and are used for the wide range level instrumentation. At 2.5 meters above the core there is another set of pressure nozzles used for the short range level
instrumentation. At 6.4 meters above the core in the steam dome there is an upper set of nozzles used for both short- and wide-range level instrumentation that are connected to the top of the reference tanks.
4.2.3.2 Modelling the level measurement system
Both the wide- and short-range level measurement system has been implemented in the TRACE model of Forsmark 1. The nodalisation of the reactor vessel has been chosen in such a way that the vertical centers of cell 14, 18 and 23 coincide with the positions of the level instrumentation pressure measurement nozzles. A density compensator model is also included according to the detailed description in reference X[23]X.
4.2.3.3 Level calculation using the TRACE function collapsed water level The water level in the reactor vessel model has also been calculated by the TRACE function that calculates the collapsed water level in a number of cells. The collapsed water level calculations are performed for cell number 14 to cell number 23 in the downcomer. These cells are chosen because their vertical centers mark the positions of the level instrumentation pressure nozzles in the reactor vessel. As described in reference X[1]X, the collapsed water level function in TRACE sums the volumes of liquid water in all the cells in the chosen range of cells. Then the volume of each cell, starting at the bottom, is subtracted from the summed volume of liquid water. This is continued until the remainder of the total liquid water volume is less than the volume of the next cell. A volume fraction of liquid is calculated for the last cell that can only be partly filled with water and this fraction is multiplied with the height of this cell. The collapsed level is then the summed height of the filled cells plus the liquid fraction of the last cell times its height.
30 (62)
4.2.4 The emergency core cooling system 323
4.2.4.1 The emergency core cooling system – function and design
The emergency core cooling system accounts for core residual heat removal at low RPV pressures and levels. The system protects the core from overheating together with the auxiliary feed-water system X[24]X. The system is also used to fill the reactor vessel with water before refuelling the core.
The nozzles of the emergency core cooling system are positioned in the downcomer at a level immediately above the top of the core. The system has four separate pathways with separate pumps, numbered from one to four. The first and third pathway have sleeves entering the downcomer to reduce the thermal stresses X[24]X.
The condensation pool supplies the emergency core cooling system with water. The temperature in the condensation pool can vary between 20 and 95˚C X[24]X. At the start of a disturbance, before steam from the pressure relief system has been dumped into the condensation pool for any longer period of time, the water temperature can be assumed to be approximately 20˚C.
The pumps in the emergency core cooling system are centrifugal pumps, connected to different subs of the stand-by diesel system X[24]X. Centrifugal pumps can only supply a flow of water if the pressure on discharge side is sufficiently low. The pumps in the emergency core cooling system start supplying water when the reactor pressure is below 11 bar X[24]X, and will increase its flow as the reactor vessel pressure is decreasing.
4.2.4.2 Modelling the emergency core cooling system
On the 25th of July, the isolation logic train was initiated due to the fact that the measurement equipment were indicating low level (L4, 1.1m) in the reactor vessel. Among other functions, the logic train started the pumps of the emergency core cooling system and the auxiliary cooling system X[25]X. There are no measurements of when the flow from the emergency core cooling system was initiated and how the coolant flow varied over time. However, tables have been set up of how the mass-flow from the emergency core cooling system varies with pressure, tables that have been produced for SAR-analysis purposes X[26]X. The table in reference X[26]X with best-
estimate values have been used as input data for a FILL-component from the TRACE component library to set the mass-flow as a function of the reactor steam dome pressure. Hence, the time when the emergency core cooling system will start delivering water can not be controlled during simulation, but will depend on the reactor vessel pressure. FILL-components are used to set a certain mass-flow as boundary conditions in the model. The water temperature is set to 20˚C.
Geometrically, the emergency core cooling system has been set up as a single pipe with an area of the summed flow area of the four reactor nozzles in the system. The sleeves in the reactor vessel on pathway one and three, the true lengths of the piping, the pipe bends and the wall roughness has been neglected when modelling the system. These simplifications are assumed to be valid because the mass-flow is set and will not depend on the pressure drop in the piping of the emergency core cooling system.
The input data of the emergency core cooling system model can be viewed in appendix 4.
