600 MWe gücünde PWR tipi bir nükleer reaktör kalp öndizayn analizi

Abstract

Nükleer güç reaktör kuruluşa kadar çeşitli karmaşık merhalelerden geçer. Bunlardan en önemlisi reaktör kalp dizaynıdır. Kalp dizaynı ise belirli sıra takip eden ve kendine özgü mantık yapısı olan çeşitli adımlardan oluşur. Burada ilk adım bir öndizayn çalışması gerçekleştirmektir. Öndizayn çalışmasında daha önce işletime girmiş reaktörlerden derlenen ve profesyonelce yapılmış tahminlerden oluşan ilk datadan yola çıkarak, reaktör kalbinin parametrik tasarımı gerçekleştirilir. Bu çalışmada nötrönik ve termal-hidrolik kuplaj gözönünde bulundurularak, hesaplamalar başlıca spektrum, difüzyon ve termal- hidrolik analiz safhalarından oluşur.. Öndizayn Çalışması sonucu elde edilen değerler ileri safhalada yapılacak olan önemli ve hassas kalp dizayn hesaplan için bir baz oluşturur. Bu çalışmada ise 600 MWe gücüne sahip bir PWR tipi konvansiyonel bir nükleer güç reaktörü için kalp öndizayn çalışması yapılmıştır. Önce öndizaynın teorik olarak nasıl gelişeceği ve temel prensipleri anlatılmıştır. Daha sonra bu işin önemli kısmı olan simülasyon araçlarının dayandığı teorik metodlar ve çalışma karekteristikleri tanıtılmıştır. Sonunda ise 600 MWe PWR tipi bir reaktör ile ilgili çalışma ve sonuçlan sunulmuştur. Simülasyon çalışmaları ağırlıklı olarak LEOPARD, 2DB ve SCTH kodlan ile yapılmıştır. Kodların koşumuna İ.T.Ü. IBM- VMS ortamında başlanmış, fakat daha sonra Linux işletim sistemi altında devam edilmiş ve tamamlanmıştır.

In many aspect, a nuclear power plant resembles to fossil-fuelled power plants. However, a peculiar component to a nuclear power plant is the reactor core in which a nuclear reaction takes.place to generate energy. Thus, a design responsibility of a nuclear reactor core primarily belongs to reactor physicists or nuclear engineers. This thesis study includes a core design analysis of a LWR reactor. For this purpose a 600 MWe PWR type reactor is taken as a model and its preliminary survey analysis is performed. In this study, any new design feature has not aimed, but rather a conventional reactor model has been chosen. The evolment of a power-reactor system, namely, to accomplish finally the construction and reliable operation of a nuclear power plant needs several stage processes. If the desired goal here can be considered a problem, then the solution of which will need to proceed through a number of logical steps. The initial step is to define the problem in a total way. Since we have already defined - or at least named - this, we can now perform next stage study which is called preliminary design and that is the subject of this thesis project. In a design study, a set of system parameters must be determined such a way that safe, reliable, and economical reactor operation at the rated power level over the desired core lifetime is possible. This requires analysing the effects of various relevant parameters together considering many constraints or restrictions. Obviously, analysis models will involve computer simulations in many stages of design process. A design effort in each stage will not involve in the same level of sophistication with respect to used analytical models and computational tools. As more information is accumulated, the design effort advances to increasingly detailed studies in an effort to narrow in on a reference design with which one can perform trade-off studies in order to determine the best choice of system parameters to achieve the optimum core performance consistent with design constraints. Naturally during these latter stages of the design processes the analytical models used to predict core behaviour become more detailed and expensive. Earlier and later stages of the design studies, however, include many similar logical structures, in contrast to analytical and computational models used in terms of the degree of sophistication. Consequently, while we perform here a preliminary survey analysis of our model pointed out above, we present both solution of our model problem and also provide general design aspects indicating how-to attack a similar problem. In this study, we consider neutronic and thermal-hydraulic behaviourof the core. Between these parts of analysis, there exists a strong interaction. Although structural (or stress) analysis of core components and economic analysis also play a deterministic role in the design, we exclude this part of analysis in our study. We have to now perform design a reactor for given electric power and overall plant thermal efficiency, (hereafter, the word 'design' should tell us that this is a preliminary design). In this study the latter parameter is taken as 0.33 that this means our reactor will have 1818 MW thermal output. In order to start, we should have some initial data. This data can be obtained from some monographs and safety analysis reports of nuclear reactors under operation. For our model, we choose Combustion Engineering (CE) 16x16 batch, PWR type reactor. This provides us core geometry,' approximate number of assemblies and fuel enrichment vs. Since a different power level is assumed, some adjustment in initial data is made. Now our goal is to determine reactor nuclear and thermal-hydraulic parameters with required characteristic constraints. For this purpose, we are expected to determine core nuclear and thermal behaviour. This two aspects of the core analysis is strongly coupled to each other. The nuclear behaviour requires the neutronic analysis of the core. We should determine the core multiplication and flux or power distributions of the core. The core power distribution of the core is of central importance to following thermal analysis and fuel depletion studies of the core. On the other hand, the core multiplication will depend on the reactivity and control analysis. In order to know neutronic behaviour, we need to keep track neutron balance during operation of the reactor. The best description for this purpose is provided by the neutron transport. If we can solve time- dependent neutron transport equation for the reactor core during operation life, our task will be truly too easy. However, the solution of this equation with seven independent variable is not almost practical because of some reasons. Mainly this is due to extreme heterogenous structure of the core and extreme energy and space dependence of the cross sections. Therefore, the neutronic analysis of a core is to be handled by clever and practical manipulations. This necessitates rather the solution of the multigroup diffusion equation (MGD) to determine spatial distribution of neutrons (especially for preliminary analysis). Hence this provides us core multiplication factor and power distribution throughout the core. Diffusion theory is an approximation to the transport equation that it will only yield reasonable accuracy if some conditions meet. This is usually not the case, however diffusion codes, if accurate cross section sets are supplied, can perform few- group detailed spatial power distribution for the core. Since the cross sections depend on sensitively on energy, few-group constants (spatially and energetically collapsed) must be generated accurately. In reactor design, this can be accomplished by the so- called spectrum calculations. On the other hand, initial fuel inventory in the core is always greater than the critical amount. Hence this causes an excess reactivity in the core lifetime and must be compensated by some means. This requires in core design a control analysis. As a result, roughly speaking, instead of directly attacking to the solution of transport equation, the neutronic behaviour is treated by spectrum and VIdiffusion calculations and also control analysis is incorporated into this scheme. Basically, in terms of the involved theory, the spectrum calculations is the most challenging part of this calculations. Many transport theory methods, approximations, and corrections are employed to perform this part of the design analysis. In this thesis project, we employed LEOPARD code for the spectrum calculations. This code zero-dimensional (0-D) pin-cell analysis code, which is widely used LWR design. Choosing 0.625 ev cadmium threshold energy, in its upper and lower energies, it performs fast and thermal spectrum calculations, respectively. The code fast and thermal spectrum calculations performs separately by using 54 group MUFT and 172 group SOFOCATE algorithms. As seen, the energy treatment of the code is very detailed, which is one of the main features of the spectrum codes. On contrary, for spatial treatment, the code homogenizes everything to the cell. In this crude spatial treatment, the LEOPARD treats a pin-cell considering four spatial region: fuel, clad, moderator, and non-lattice region. In a LWR, an actual core consists of many regions that do not consist of pin cells, e.g., burnable poison lattice cells, waterholes (for instruments or control rods), and interassembly water gaps. Thus if the MGD calculation is to include those regions, appropriate cross sections would have to be determined by some auxiliary calculation. This is true whether the MGD calculation is to treat the extra regions explicitly or implicitly homogenizing the extra regions with the unit cells. In fact, the latter case would necessitate knowing the relative flux levels of the unit cell and extra regions in order the collapse the cross sections correctly. LEOPARD improves on this situation by defining a `non-lattice` region which includes the portions of the core which are not unit cells (`lattice regions`). Thus by this approach, non-lattice regions are considered by the LEOPARD and thereby following diffusion calculations. In fast spectrum calculations, an important issue is the treatment of the resonance absorption. MUFT algorithm (in a version used by us) treats resonance integrals by the unbroadened NRIM approximation. In fast region, other important phenomenon is the energetic self-shielding effect of the fuel rod. For this purpose, the code performs a optional o)*- search. According to this search, direct resonance integral calculation results is adjusted until it is in agreement with Strawbridge-Bary metal-oxide correlation. As in thermal spectrum calculations, for resonance absorption, fast fission effects etc. fast spectrum calculations rely on a useful transport method called `collision-probability` method. In MUFT algorithm, fast spectra is generated by solving Bi equations. For this purpose Selengut-Goertzel slowing down scheme is employed, and, total equations are discretized over lethargy variable and numerically solved. Previously calculated resonance parameters are used in this scheme. Once the fast neutron-energy spectrum is obtained, 54-group MUFT calculation is utilized to generate one-fast group constants for the next stage MGD calculations. Thermal spectrum calculations have somewhat different aspects with respect to that of fast spectrum. Here some molecular effects and upscattering also play role and need relatively detailed spatial treatment. Therefore for spatial collapsing, one need to know thermal disadvantage factors of different regions. This procedure in vuSOFOÇATE algorithm is performed by a well-known technique ABH method. This is a crude approximation as compared to integral transport algorithm (THERMOS) which is used in 1-D spectrum codes. The determination of the thermal spectrum in this scheme is performed by the Wigner-Wilkins method. Now the ABH method is performed for the spatial collapsing for 172 group. Then the Wigner-Wilkins spectrum is used to energetically collapse group cross-sections to generate one thermal group MGD constant. In our work for the MGD calculations we used a 2-D finite difference code 2DB. This code is capable of doing global calculations for both coarse and fine mesh. It solves MGD equations with the spatial discretizing. Hence the core is analyzed energetically by a few-group (for 2DB this is either two or four which is exactly what LEOPARD is capable of generating MGD constants) but with a detailed spatial treatment. The code has x-y and r-z run options. Employing core symmetry, the code may be run for 1/4 core or upper half for these cases, respectively. The output of the code includes spatial fast and thermal flux distributions throughout the core. Depending on this output data, more useful output of the code is the power distribution of the core. Hence power maps for each assembly (by x-y run), regional axial power maps (by r-z run), or inter-assembly power maps (by fine mesh run) can be calculated by the code. 2DB also yields core multiplication which is needed by control analysis. Since there is a coupling between neutronic-and thermal-hydraulic behaviour of the core, in order to start spectrum calculations, many thermal parameters are required. This parameters are determined via thermal-hydraulic analysis of the core. For this purpose in our design study, we used flow-channel analysis code SCHT. This code calculates the coolant conditions such that pressure, density, temperature, quality etc. and the fuel rod temperatures in both an average coolant channel and the maximum power coolant channel of a light water reactor. These calculations are performed in each of these two channels. In performing the calculations, the code accounts subcooled, local boiling and bulk boiling regimes for the clad-coolant heat transfer. If İt is required, axial power profile may be input the code. The code evaluates the departure from Nucleate Boiling Ratio (DNBR). For this purpose, the code evaluates DNB heat flux from W-3 heat flux correlation which is developed especially for PWR's. Code also evaluates channel pressure drops in addition to many other thermal and flow condition parameters. Once these computational tools whose properties described above are available, we can now start our design effort. As we mentioned above, first we determined initial data with previous experience and model associated with educated guess. Our main design goals are to obtain power distribution throughout the core as flat as possible, to have appropriate reactivity coefficients and a reasonable cycle length. Most of these data are provided us by the 2DB code. However, to run this code, we must first perform spectrum calculations. Since LEOPARD analyzes pin cell, and even a quarter core have ten thousands pin cell, we run this code for representative pin-cells which are called asymptotic pin-cells. Also it is worth to mention that the spectrum calculations depend on large range of data including vnithermal-hydraulic data. Thus initial data must be set up the description given above. When the spectrum calculations are performed, we next feed homogenized, collapsed cross-sections into 2DB to obtain mainly power maps and core multiplication factor. 2DB run with x-y and r-z options provide us assembly and axial power maps, respectively. The latter power profile is also needed as one of the main input to SCHT code. In second iteration, we are supposed to use SCHT thermal output rather than initially compiled data in LEOPARD, then repeat this process until a convergence is reached in power maps, multiplication factor etc. This iteration is usually repeated maybe hundreds of times. However, while this is the logical structure of the design process, never all aspects of the design evaluation can be performed at one stroke. In other words, loading or fuel replacement strategy, poisson and depletion problems can be solved subsequently. Hence, to start iteration process described above, we have to first decide our loading map. In our model, the core consists of three type assembly with respect to average assembly enrichment. At the same time each assembly contains three different enriched fuel-pins. For the unpoissoned core, we choose the same loading map for each type assembly with respect to different enriched fuel-pins and water holes while preserving İnter-assembly symmetry. Once the core loading is set up, we examine assembly power maps. As we expect, this power with high power- peaking factors will be never flat in the required level. This situation will improve, however, by the model of burnable poisson. In order to both increase the allowable initial core loading and compensate significant portion of the excess reactivity, some materials (poisson) with very high neutron absorption cross section is loaded into core. For this purpose, different choices such that soluble poisson (chemical shim, e.g. soluble boron) or lumped poisson pin (e.g. GdaOs) alternatives are possible. Each choice has its own advantages and disadvantages. In our model, we consider Gd203 pins for burnable poisson. As compared to homogeneously distributed poisson, this model several advantages. Because of self-shielding, lumped poisson pins have better excess reactivity match during a cycle and provides no residual poisson at the end of cycle, otherwise, it would cause shortening fuel lifetime. Unfortunately, the analysis of burnable poisson is quite difficult and requires complex neutronic (transport) codes other than we used in order to obtain results useful for detailed design studies. As in it is sometimes resorted to, we rather used crude models to treat poisson. We used definite number of poisson pins close to the center of the core to compensate the excess reactivity and also to have acceptable power distribution with reasonable power peaking factors for a clean core. Our next task is to analyze one fuel cycle. This requires depletion analysis, our goal here is to have a reasonable cycle length. This part of analysis must include also poisson depletion in addition to the fuel. Since the clean core power distribution can not be maintained after the operation starts, to follow power distribution throughout the cycle is required in order not to exceed the thermal limitations thereby causing fuel failure. In this stage of design, we can proceed further with similar logic to the process of setting clear core design with the exception of details involved. For this IXpurpose we again used LEOPARD code with burnup option. Assigning optional time steps until end of fuel life, this code can generate cross-section sets to be used in the 2DB. This procedure is associated with our poisson depletion model. By this means we obtained inter-cycle and end-of-cycle radial and axial power maps. In our design effort, the last step is to model movable control rods. They serve to power manoeuvring, shim control and reactor scram. For shim control, control rods is desired to compensate minor portion of excess reactivity. Because they are strong absorbers hence they cause major perturbations in the neutron flux in their vicinity. On the other hand they capable of compensating all excess reactivity in emergency situation for shutdown the reactor. Due to their strong absorbing nature also, to model control rods accurately is difficult task. In our work, we used relatively simple scheme to design control rods. Theories and design principles which briefly summarized above are presented in this thesis in detail. Initial, intermediate and final results which belong to the appropriate stages of design steps are given. Here the major emphasis is given to the evaluation and the interpretation of power distribution of the core. The final design data including core, fuel, and thermal and flow condition parameters are determined and presented. Core multiplication factor for different cases and reactivity coefficients together with engineering safety factors are evaluated.