Method for determining the characteristics of a target nuclear power plant for the management of nuclear resources and waste
The data processing method efficiently determines nuclear fleet characteristics at equilibrium by translating cycle diagrams into reactor-plant systems and resolving non-linear balance equations, addressing the computational and precision issues of existing methods.
Patent Information
- Application Number
- FR2023005211
- Authority / Receiving Office
- FR · FR
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-25
- Publication Date
- 2024-11-29
AI Technical Summary
Current methods for determining the characteristics of a nuclear fleet at equilibrium are computationally intensive, time-consuming, and imprecise, often requiring substantial engineering resources and expertise, and do not guarantee convergence to a balanced state.
A data processing method that involves generating input data, translating a cycle diagram into reactor-plant systems, constructing balance equations, selecting meta-models, and resolving a system of non-linear balance equations using meta-modeling techniques such as neural networks to determine the characteristics of a target nuclear fleet.
This method allows for rapid, reliable, and precise determination of nuclear fleet characteristics at equilibrium, reducing computational and engineering time, and ensuring accurate material flow balancing without temporal considerations.
Smart Images

Figure 00000026_0000 
Figure 00000026_0001 
Figure 00000027_0000
Abstract
Description
Title of the invention: Method for determining the characteristics of a target nuclear park for the management of nuclear resources and waste Technical field
[0001] The present invention relates to the modeling of the flows of recoverable materials from an advanced nuclear fleet at equilibrium. Prior art
[0002] A nuclear fleet is made up of all the facilities implemented during the nuclear fuel cycle, ranging from the extraction of uranium ore to the storage of ultimate waste, including the manufacture of nuclear fuel and its installation in the reactor. These facilities include spent fuel reprocessing plants (96% of which is still recyclable materials in the case of standard UOX PWR fuels) and plants for conditioning and storing ultimate waste (non-recyclable materials). “UOX” means a fuel based on natural uranium oxide isotopically enriched in uranium 235.
[0003] Reprocessing plants allow all reusable materials to be recovered for recycling in reactors, and at the same time reduce the volume and toxicity of waste.
[0004] A nuclear park is characterized by flows of recoverable materials which pass through the fuel cycle installations.
[0005] A nuclear fleet is considered to be "in equilibrium" when there is no longer any change over time in the flows of recoverable nuclear materials entering and leaving the fuel cycle. A nuclear fleet in equilibrium is generally represented in the form of a flowchart, called a cycle diagram, describing the stages of the cycle and the flows of materials entering and leaving each stage.
[0006] Based on the definition of an initial state of the nuclear fleet and its evolution, nuclear scenario studies aim to provide an overview of the influence of strategic decisions (deployment of a new type of reactor / fuel, evolution of installed power, etc.) on the nuclear fuel cycle and to quantify the impact of a given change in strategy, according to different evaluation criteria formulated on the observable quantities of the cycle: evolution of material flows, use of recoverable materials, plant capacities, waste production, etc. Nuclear scenario studies make it possible, in particular, to evaluate the impact of the integration of different technologies into the future fleet of nuclear reactors and to study the associated fuel cycle. They can therefore provide elements of response to the problems linked to the evolution of a nuclear park and thus support decision-making.
[0007] These studies are based on simulations of the nuclear fuel cycle modeling all the operations carried out on nuclear fuel from the mine to final storage (extraction, enrichment, manufacturing, irradiation in the reactor, cooling, reprocessing, etc.). The simulation of nuclear scenarios is carried out using a nuclear scenario code, which is dynamic in nature, i.e. the simulated nuclear fleet evolves over time (order of magnitude decades to hundreds of years, in line with the relatively long lifespan of industrial nuclear installations).
[0008] Determining the material flows in a nuclear fleet at equilibrium (in steady state, i.e. without evolution) is of interest for defining future target fleets that can meet the need for managing nuclear resources and waste. Indeed, static calculation - at equilibrium - makes it possible to determine the characteristics of a fleet that could constitute an interesting target during a dynamic scenario study. A fleet at equilibrium is generally represented in the form of a flowchart describing the stages of the cycle and the material flows entering and leaving each stage.
[0009] As an example, [Fig.l] represents the cycle diagram associated with a fleet consisting solely of isogenerator RNRNA (sodium-cooled fast neutron reactors) operating in a closed cycle (reusing the recoverable materials from the spent fuel). The “equilibrium fleet” object, the subject of the simulation tool, is therefore, in this example illustrated by the cycle diagram in [Fig.l], a set of nuclear reactors with the associated cycle installations, for which the equilibrium will relate to mass balances.
[0010] Beyond this illustrative example, many other parks can be imagined (by defining their composition in terms of the number of reactors of each type and the way of recycling the recoverable materials in the downstream cycle plants) for which the search for a balanced situation is interesting. We can cite symbiotic parks comprising 2 types of reactors, one producing plutonium and the other consuming, or even minor actinide transmuting parks or even those based on the natural thorium resource.
[0011] In the field of scenario simulations, France is particularly active because the choice to partially close the cycle by reprocessing the spent fuel to extract the recoverable materials within its industrial recycling plant at La Hague leads to the consideration of complex scenarios which require the use of calculation tools.
[0012] There are different families of calculation tools used to describe situations balance: - Dynamic simulation codes; - Semi-dynamic simulation processes known as turn-based approaches, based on neutron simulation codes; - Automated search processes for fleet compositions (number of reactors of each type for a given cycle diagram, whose plants and destination of materials at each stage are fully defined) with dynamic scenario codes.
[0013] The nuclear scenario codes commonly used in the international community are dynamic simulation codes, i.e. they model nuclear fleets whose composition may change over time. They calculate the evolution of inventories of elements of interest and flows of recoverable materials at each moment in each unit of the cycle (mining and conversion facilities, enrichment and manufacturing plants, reactor, cooling pool, spent fuel treatment plant (or reprocessing plant), geological storage centers for ultimate waste packages, etc.), under neutron flux in the reactor or outside flux (during cooling of spent fuel for example).Dynamic scenario codes can be based on mathematical models to determine the composition of new fuel and its fraction of recoverable materials (equivalence model), and to calculate the evolution of its composition under irradiation (irradiation model).
[0014] The COSI code has been widely used for scenario studies for over 30 years. The latest version 6 of this code is notably described in the document [COSI6: a tool for nuclear transition scenario studies and applications to SFR deployment scenarios with minor actinides transmutation, Nucl Tech. 192, 91-110]. It is the reference code in the field (the most detailed, the most precise). However, the dynamic simulation of complete nuclear fleets consumes a lot of computing resources: for example, the simulation of a scenario for the evolution of the French fleet developed by several companies for the national strategy requires the mobilization of around twenty processors over around ten hours and the occupation of around one hundred GB of RAM.
[0015] In the field of determining the characteristics of nuclear power plants in equilibrium, the search for solutions is iteratively carried out by trial and error. This is a process that is all the more tedious as the cycles studied are complex. It has happened that no perfectly satisfactory solution from the point of view of the equilibrium of fuel flows has been found after several years of study, for systems that nevertheless include an infinite number of solutions.
[0016] The time parameter is the main cause of this limitation.
[0017] At each time step, or at each significant date on which an event occurs, the scenario code performs calculations of the evolution of material under flow (irradiation) or outside flow (radioactive decay), and / or transfers materials from one station in the cycle to another. Beyond these operations, no code to our knowledge seeks to balance the flows of recoverable materials globally, on the scale of the complete cycle.
[0018] The evolution as a function of time adds a dimension to all the parameters of the problem, which themselves may be led to evolve as a function of time, and requires adjustments to the parameterization since we do not have a priori knowledge of the characteristics of the target fleet at equilibrium (composition of the nuclear fleet in reactors of each type, cooling time of the spent fuels and recycling of the separated materials, destinations of these materials between recycling and waste, and compositions of the recycling fuels). Cooling time means the transit time of a spent fuel between its unloading from a reactor and its reprocessing. At each parameterization, the calculations must be restarted until the input parameterization allows convergence towards a satisfactory result.
[0019] In order to simplify this step, approaches to searching for nuclear parks in equilibrium have been developed.
[0020] The most common approach in the literature, the "round approach" is disclosed in the document [G. Youinou, and A. Vasile. "Plutonium multirecycling in standard PWRs loaded with evolutionary fuels." Nuclear science and engineering 151.1 (2005): 25-45]. This approach is mainly developed to calculate the composition of a fuel at equilibrium of a multi-recycling strategy. A "round" then designates the irradiation in the reactor of the "new" fuel constructed from the materials extracted from the spent fuel during the recycling step during the previous "round" and possibly, from an external supply of fissile material with a fixed composition. The parameters associated with the cycle, such as the cooling time of the spent fuel and the manufacturing time of the new fuel are defined as input to the calculation.This approach does not rely on the development of a dedicated code but on the iterative use of a neutron modeling code, often at the assembly scale. Originally, given the computational capabilities available when implementing this method, only a limited number of "turns" (less than 10) were modeled. Convergence to an absolute equilibrium state was therefore not guaranteed. Still relying on the iterative use of a neutron modeling code, an estimator of the convergence to an equilibrium solution of the fleet is calculated. Several dozen "turns" are then modeled until an acceptable convergence is reached, for which the variation in the fuel composition . nine from one round to the next is then considered negligible. This typically takes tens of days, in both engineer and machine time. This more robust method is still not implemented in a dedicated equilibrium scenario code, but rather constitutes an iterative manual procedure using a neutron modeling code at one of its steps.
[0021] It is difficult to generalize, since it depends on the type of reactor and fuel modeled, and therefore on the neutron modeling code on which it is based.
[0022] It requires developing a calculation procedure specific to each cycle studied. This round-based approach is heavy in computing resources.
[0023] It requires a lot of neutronic calculations to be carried out sequentially.
[0024] In addition, this approach is very heavy in terms of engineer resources because it uses a manual, non-automated process, carried out on a case-by-case basis. In addition, it is a very expert approach that requires knowledge of the fuel cycle and extensive experience in neutron simulation.
[0025] Finally, it is an imprecise approach which requires carrying out a large number of rounds to converge towards an equilibrium situation and simulations based on reference neutron calculations, which are costly and therefore limit the number of rounds carried out: this generally results in a poorly converged and therefore imprecise result.
[0026] More recently, a method for determining the characteristics of nuclear fleets at equilibrium based on the use of dynamic scenario codes has been designed. This method is based on a large number of nuclear scenario simulations (several thousand) for which different values can be assigned to the input parameters (composition of the fleet, cooling and manufacturing time, etc.). Analysis of the results of these simulations makes it possible to identify the characteristics of the nuclear fleets (and therefore the values of the input parameters) satisfying the conditions of equilibrium of the material flows. This approach requires defining a time horizon at which the dynamic simulation is stopped and the characteristics of the final fleet recorded. Since this choice is arbitrary, it is not guaranteed that the equilibrium state observed at this time is absolute.Furthermore, substantial computing resources are required to implement this method. Indeed, it is very heavy in computing resources because many scenario calculations must be launched to pave the space of possibilities. In addition, it requires in-depth knowledge of neutron simulation. Finally, there is no certainty of finding the solution due to the slowness of the process, a large number of parameters whose variation ranges defined a priori (when defining what is commonly called a "design of experiments" in simulation) may be outside the desired solution and the disruptive nature of the dynamics on the static solution.
[0027] It is thus necessary to develop a data processing method that is fast, reliable and accurate, so as to balance the flows of nuclear materials (heavy nuclei) so that a given fleet (characterized by a fuel cycle, fuel management in the reactor and the associated cycle facilities) can be operated sustainably. By "sustainably" is meant the possibility of operating a fleet in a steady state without the characteristics changing over time. In order to at least partially address this need, the data processing method that is the subject of the invention has recently been implemented in a static nuclear fleet simulation platform called SEPAR, which makes it possible to perform equilibrium fleet calculations.This platform is based on object-oriented programming using the C++ language, which incorporates meta-modeling methods using neural networks and algorithms for solving nonlinear equation systems. The data entry uses the standard XML format. A preliminary simplified version of the "equation entry and resolution of equilibrium nuclear fleet calculations" section was tested before the development of the overall method. The first equilibrium fleet calculations thus carried out on a case-by-case basis have shown promise in the document [G. Martin, Study of a mixed fleet of breeder SFR and EPR supplied with LEU and MOX fuels to balance the plutonium inventory, Proceedings of ICAPP, Paper 23859, 2018].
[0028] These preliminary calculations motivated the development of SEPAR which integrates the global data processing method and the results obtained were disclosed in the paper [H. Barale, et al. "A new tool for the simulation of different nuclear fleets at equilibrium." EPJ Nuclear Sciences & Technologies 8 (2022): 1] and in the paper [H. Barale, et al. “An approach for evaluating actinide production in the equilibrium simulation code SEPAR.” Proceeding in GLOBAL 2022 International Conference on Nuclear Fuel Cycle (2022)]. Statement of the invention
[0029] The present invention is based on a data processing method which aims to determine the characteristics of a target nuclear park for the management of nuclear resources and waste.
[0030] Thus the invention relates to a method for determining the characteristics of a nuclear fleet at equilibrium, represented by a cycle diagram, for the management of a nuclear fuel, said nuclear fuel forming spent fuel after exploitation by said nuclear fleet, said spent fuel being capable of being reprocessed to constitute a recycling fuel and / or separate materials, said determination method comprising: - a step of generating input data, called class 1 data, such as nuclear data, plant data or reactor data; - a step of translation by a user of a nuclear fleet cycle diagram, into so-called class 2 data; - a step of transforming said cycle diagram into reactor-plant systems from class 1 data and class 2 data; - a step of constructing balance equations from the reactor-plant systems, said balance equations forming a system of non-linear balance equations; - a step of selection of meta-models of composition of the recycling fuel; - a step of computer resolution of the non-linear balance equation system from the meta-models selected for the determination of characteristics of said nuclear fleet, called class 3 data.
[0031] The invention thus makes it possible to improve the management of nuclear resources and waste by proposing a loop-free cycle diagram, as is particularly visible in [Fig.10]. This [Fig.10] is the result of a transformation of [Fig.9]. This transformation allows the duplication of the start and finish box of the loop into an arrival box at the start of the line and an arrival box at the end of the line in order to avoid the representation in the form of a loop.
[0032] The succession of steps of the method described here is more particularly shown diagrammatically in [Fig.4].
[0033] Thus, the method can support preliminary studies for dynamic scenario studies, or be directly included in studies on the viability of advanced nuclear parks. It makes it possible to propose innovative scenarios for the development of nuclear parks by taking into account the consumption of natural resources and the production of non-recoverable radioactive materials from the cycle, in an approach of non-accumulation and sustainability of the nuclear park.
[0034] Being a method which resolves a stable state, it could be useful in particular in the industrial context of sizing reprocessing and enrichment plants to their nominal operating state, or even to help in the sizing of storage centers for example.
[0035] Moreover, although this process has only recently been implemented in SEPAR (in 2020), it has already been used for studies carried out within the laboratory. The publications around SEPAR focus on comparing the results obtained with similar results obtained using the dynamic COSI code (see
[0016] ), without explanation of the process however. The equilibrium state of the nuclear fleet obtained with this process is characterized by knowledge of all the characteristic data of the nuclear fleet (class 3).
[0036] The method makes it possible to determine these parameters so that a new nuclear park operates according to a fuel cycle in steady state, without taking into account the transition phase to this new nuclear fleet. It therefore has no time dimension.
[0037] Its calculation time is very short, and this makes it a very useful method for exploring and comparing a large number of nuclear parks.
[0038] It also consumes little engineer time.
[0039] It is easy to access, insofar as it is limited for the user to defining a cycle diagram, by means of the recording of reactor, plant and nuclear data (class 1 data recorded at stage E0 prior to the use of the code) in advance by competent engineers.
[0040] It is reliable to the extent that it is capable of solving particularly complicated cycles.
[0041] With regard to its accuracy, it is acceptable compared to approaches using dynamic or semi-dynamic simulation processes, time being a disturbing factor in the balance of flows. In addition, particular embodiments can be declined according to the definition of the input and output data (class 3 data).
[0042] In addition, the step of solving the system of non-linear equations makes it possible to obtain one or more fractions of the used fuels to be sent to the reprocessing plant.
[0043] This step of solving the system of non-linear equations makes it possible to obtain the fraction of separated materials (originating from the treatment-recycling of spent fuels) to be recycled in the reactor in the form of fuel intended for recycling. The equilibrium state of the nuclear fleet with this process is defined by knowing the fractions of separated materials used for the manufacture of recycled nuclear fuels, and the fractions of stored separated materials.
[0044] The step of solving the system of non-linear equations also makes it possible to obtain one or more residence times between the step of unloading the irradiated fuels outside the reactors and the step of their recycling in a reprocessing plant (also referred to here as a treatment plant). The equilibrium state of the nuclear fleet with this method is defined by knowing the cooling times of the spent fuels before treatment.
[0045] The step of solving the system of non-linear equations makes it possible to obtain one or more residence times between the reprocessing step and the recycling in the reactor of separated materials originating from nuclear waste. The equilibrium state of the nuclear fleet with this method is defined by knowing the aging times of the separated materials before reintroduction into the reactor. By "aging time" is meant the transit time of a recoverable separated material originating from spent fuel between its exit from the reprocessing plant and its reintroduction into a reactor.
[0046] The step of solving the system of non-linear equations also makes it possible to obtain the composition(s) of recycled fuels originating from spent fuels. The equilibrium state of the nuclear fleet with this method is defined by knowing a composition of the recycling fuels. Fraction in fissile isotopes and elemental contents can be defined according to several conventions, mass, or molar, or atomic.
[0047] In a particular embodiment, the computer resolution step comprises processing by neural networks of the selected meta-models.
[0048] In a particular embodiment, the computer resolution step comprises processing of the meta-models selected from kriging models.
[0049] In a particular embodiment, the computer resolution step comprises processing of the meta-models selected from chaos polynomials.
[0050] In a particular embodiment, the computer resolution step allows a determination of an organization of the nuclear fleet, said organization comprising a list of several reactors associated with a fraction of use for each reactor, the sum of the fractions associated with the different reactors being equal to 1 so as to limit a number of unknowns in the system of non-linear balance equations.
[0051] In a particular embodiment, the class 3 data comprises at least one of the characteristics selected from the list of following characteristics: - a destination for the spent fuel;
[0052] - a destination of separated materials from the reprocessing of spent fuel; - a transition time for separated materials from the reprocessing of spent fuel; - a composition of recycled fuel from the reprocessing of spent fuel.
[0053] The present invention will be better understood on reading the detailed description of embodiments taken as non-limiting examples and illustrated by the appended drawings, in which: - [Fig.l] schematically illustrates a park of RNRNA operating in a closed cycle according to the prior art; - [Fig.2] schematically illustrates all the parameters describing a target nuclear park in equilibrium for the management of nuclear resources and waste; - [Fig.3] is a general diagram of a reactor-plant system for which the upstream reactors are numbered from 1 to r, and the treatment processes (plants) are numbered from 1 to p, a particular reactor-plant system being a rectilinear graph of objects obtained following the application of step E2 on a particular data set; - [Fig.4] illustrates the different stages of the process of determining all the parameters grouped in [Fig.2]; [Fig.5] illustrates the cycle diagram of a target nuclear park for the management of nuclear resources and waste according to a first application case, the dotted lines specifying fractions of material flows between reactors and plant not defined by the user; - [Fig.6] schematically illustrates the reactor-plant system resulting from the graph analysis of step E2 of the cycle diagram given in [Fig.4], the dotted lines specifying fractions of material flows between reactors and plant not defined by the user; - [Fig.7] illustrates the cycle diagram of a target nuclear park for the management of nuclear resources and waste according to a variant of the first application case (see cycle diagram of [Fig.5]), the solid lines specifying fractions of material flows between reactors and plant defined at 100% (all the materials are transferred, destination defined by the user); - [Fig.8] schematically illustrates the reactor-plant system resulting from the graph analysis of step E2 of the cycle diagram given in [Fig.7], the solid lines specifying fractions of material flows between reactors and plant defined at 100% (all the materials are transferred, destination defined by the user); - [Fig.9] illustrates the cycle diagram (cyclic: recycling fuels are themselves recycled at least in part) of a target nuclear fleet for the management of nuclear resources and waste according to a second application case, the solid lines specifying fractions of material flows between reactors and plant defined at 100% (all the materials are transferred), and the dotted lines of the transferred fractions corresponding to the percentage indicated near the line, of 5% or 10%, of the mass flow, the remainder leaving the cycle to be stored (material destinations specified by the user); - [Fig. 10] schematically illustrates the two rectilinear reactor-plant systems resulting from the graph analysis of step E2 of the (cyclic) cycle diagram given in [Fig.9], the solid lines specifying fractions of material flows between reactors and plant defined at 100% (all the material is transferred), and the dotted lines of the transferred fractions corresponding to the percentage indicated near the line, of 5% or 10%, of the mass flow, the remainder leaving the cycle to be stored (material destinations specified by the user).
[0054] The invention is not limited to the embodiments and variants presented and other embodiments and variants will become clear to those skilled in the art.
[0055] The times shown on the arrows in Figures 5 to 10 correspond to the cooling and aging times defined in paragraphs
[0018] and
[0046] respectively. tively. Cooling and aging times can of course take different values.
[0056] The originality of the invention, which therefore has as its object the determination of the characteristics of nuclear parks at equilibrium, lies in the succession of six stages E0 to E5.
[0057] Step E0, disclosed in the document [Fanny COURTIN, Study of the incineration of plutonium in MOX PWR on enriched uranium support with the dynamic simulation code of the CLASS cycle, Doctoral dissertation, Université Bretagne Loire, 2017] consists of defining the generic input data set in the field of nuclear materials and installations. By "MOX", we mean a fuel based on a mixture of uranium oxides depleted in uranium 235 and plutonium. These input data, called class 1, are introduced into the code by experts in the nuclear field, prior to the implementation of the procedure by a user.
[0058] They include nuclear data: inventory of actinides to be taken into account in the calculation and associated physical quantities (decay chain, decay constant, branching ratios, etc.). Only isotopes with a half-life of less than 1000 years are considered in the application cases presented below, knowing that this duration is very long compared to the stages of the cycle and the operating times of nuclear installations (a few decades), which means that the accuracy of the calculations is not impacted by this approximation. The isotopes are grouped by type of element, and no longer distinguished as isotopes, since the material flows considered during the treatment-recycling of spent fuels will concern the elements.Americium 241 is associated with the element plutonium due to its presence, generally in non-negligible quantities, in plutonium recycling fuels, and the importance of taking it into account in neutron simulations. In step E0, the expert indicates representative compositions of the recycling fuels as default initialization values. The calculated fleet characteristics do not depend on these values, which only serve as a starting point for the non-linear equation resolution procedure implemented in step E5.
[0059] They also include reactor data: characteristic quantities of a given reactor, in terms of inlet and outlet material flows (in the form of a data matrix whose dimension depends on the number of possible compositions retained for the fuel considered).
[0060] They also include plant data: fractions of materials transferred (called transfer functions) during treatment / recycling for each element entering into the composition of the spent fuel. Data linked to treatment plants or reprocessing plants make it possible to describe process losses by element, knowing that the flows transferred and recoverable in recycling fuels are the complement to 1 of these losses.
[0061] Step E1 corresponds to the definition, by the user, of the case study by means of class 2 input data. These data define the case study and correspond to a translation, in xml file, of the cycle diagram (as an example is given in [Fig.l]) describing the flows of nuclear materials between the reactors and the reprocessing plants. In order to simplify the procedure, all the upstream and downstream plants of the cycle are grouped under the term "processing plants" with the associated transfer functions. Step 1 also uses class 3 input data, which correspond to the characteristics of nuclear parks that one seeks to determine by the procedure, either by fixing them a priori (input data), or by calculating them (output data). Indeed, they are too numerous to be all determined by the calculation and some of them must be fixed by the user according to the problem that one has defined (cf.step E3). These class 3 data, which are of five types, are broken down into nuclear fleet characteristics, into the share of each of the reactors present in the cycle diagram (the percentage that each type of reactor occupies in the fleet in number of units), into the fraction of spent fuel to be sent to the reprocessing plant; into the fraction of separated materials (after reprocessing of the spent fuel) to be recycled in the reactor in the form of fuel intended for recycling, into the composition of the fuels intended for recycling, into the transit time of the materials between the different installations.
[0062] Step E2 is a step of analyzing the consistency of all the data corresponding to a single nuclear fleet to ensure that there is no zero flow of material upstream of the plants and reactors, and that the cycle diagram is indeed representative of a single nuclear fleet. This step, based on a graph analysis of all the class 2 data (potentially cyclic cycle diagram in the case of partial or total recycling of recycling fuels), makes it possible to identify and count the class 3 data, as well as the number of equations of the problem to be solved, in order to evaluate, by comparison (between the number of unknowns and the number of equations), the number of class 3 data to be set as input data. The choice of the nature of these input data to be set depends on the type of problem that one wants to solve.The result of this graph analysis can be schematized by rectilinear reactor-factory systems which are representations in terms of programmatic objects (in the case of SEPAR in C++) of the balance equations representative of the equilibrium, and which are constructed according to the problem posed. The determination of these systems amounts to applying a classical analysis - via graph theory - to the cycle diagrams (class 2 data) specified by the user, which can be cyclic by definition (as in the example in [Fig.l]). The objective is to . convert into “upstream reactors - treatment processes (or plants) - recycling reactor” systems, subsequently called reactor / plant systems. Each system thus constructed is associated with a single recycling reactor, so that several systems can be constructed for a single coherent cycle diagram. A reactor-plant system is generally represented in [Fig.3].
[0063] Step E3 consists of putting the reactor-processing plant systems into equations at equilibrium, said equations forming a single system of non-linear equations which can be solved quickly by well-known numerical methods. These equations are established by scanning the linearized reactor-plant systems, from the reactors upstream of the treatment to the recycling reactor, and by constructing the material balances at each stage, according to the class 1 and class 3 input data. Among the class 3 data, those specified by the user have a defined value used to evaluate the material flow in the balance equations, while those which are not specified constitute the unknowns of the equations thus constructed, and the determination of which makes it possible to know all the characteristics of the nuclear fleet studied.Class 1 data provided in matrix form requires an intermediate meta-modeling step to be usable (step E4). An equation of the type described by this step E3 to characterize the inputs and outputs of the processing plants is disclosed in [G. Martin, Study of a mixed fleet of breeder SFR and EPR supplied with LEU and MOX fuels to balance the plutonium inventory, Proceedings of ICAPP, Paper 23859, 2018].
[0064] Step E4 concerns the identification of suitable meta-models, said meta-models being usable via neural networks. This type of meta-modeling is classic in the field of dynamic scenarios, it is disclosed in the document [Fanny COURTIN, Study of the incineration of plutonium in REP MOX on enriched uranium support with the dynamic simulation code of the CLASS cycle, Doctoral dissertation, Université Bretagne Loire, 2017]. This involves identifying meta-models suitable for solving the system of non-linear equations. The advantage in functional terms of this step is to be limited to the construction of non-existing meta-models.
[0065] In step E5, the resolution of the system of nonlinear equations makes it possible to determine the characteristic quantities of the equilibrium situation. The advantage in functional terms of this step is the rapid resolution of the equilibrium problem thanks to meta-modeling. Step E5 makes it possible to obtain the characteristics of the nuclear fleet studied (class 3 data) in addition to those fixed as input data, said characteristic quantities comprising a list of several reactors associated with a fraction of occurrence for each reactor, the sum of the fractions associated with the different reactors being equal to 1 so as to limit the number of unknowns in the system of nonlinear equations. This step is classical, the computer resolution of systems of nonlinear equations being based on mathematical techniques disclosed in [Lasdon, LS, Waren, AD, 1983. Large scale nonlinear programming. Comput. Chem. Eng. 7, 159].
[0066] The application cases defined below allow for a better understanding of the sequence and content of the different stages.
[0067] It is appropriate, firstly, to describe step E0 which is common to the different application cases and therefore, to define the class 1 input data.
[0068] Nuclear data: - U element - U234, parent isotope Pu238, initialization value 0.002 - U235, fissile isotope, initialization value 0.008 - U236, initialization value 0.005 - U235, initialization value 0.985 - Pu Element - Pu238, half-life 87.69805 years, branching ratio 1.0, parent isotope Cm242, initialization value 0.0259 - Pu239, fissile isotope, parent isotope Cm243, initialization value 0.552 - Pu240, parent isotope Cm244, initialization value 0.2585 - Pu241, fissile isotope, half-life 14.32999 years, branching ratio 0.999975, initialization value 0.0727 - Pu242, initialization value 0.0787 - Am241, half-life 432.7991 years, branching ratio 1.0, parent isotope Pu241, initialization value 0.0 - Np element - Np237, parent isotope Am241 - Element Am - Am241, half-life 432.7991 years, branching ratio 1.0, parent isotope Pu241 - Am243 - Element Cm - Cm242, half-life 0.4460796 years, branching ratio 1.0 - Cm243, half-life 30.0 years, branching ratio 0.9976 - Cm244, half-life 18.00001 years, branching ratio 0.999999 - Cm245 - Cm246
[0069] Reactor data: Here all fractions and contents are expressed in mass values: - Case of UOX REP fuel: • Mass of fuel in the core: 129 tmli • Cycle length: 515 JEPP • Unloading combustion rate: 55,000 MWj / tmli • Number of cycles: 3 • Average availability factor: 0.83 • Composition of the fuel at the outlet: • Uranium: mass content 0.929326E+00, isotopic mass contents U234 0.170912E-03 / U235 0.808079E-02 / U236 0.679790E-02 / U238 0.984950E+00 • Plutonium: mass content 0.124412E-01, isotopic mass contents Pu238 0.297633E-01 / Pu239 0.485115E+00 / Pu240 0.255289E+00 / Pu241 0.145723E+00 / Pu242 0.795322E-01 / Am241 0.457791E-02 • Neptunium: mass content 0.845862E-03, isotopic mass contents Np237 0.883297E-00 • Americium: mass content 0.296812E-03, isotopic mass contents Am241 0.191888E-00 / Am243 0.803925E+00 • Curium: mass content 0.148814E-03, isotopic mass contents Cm242 0.187104E+00 / Cm243 0.557436E-02 / Cm244 0.746824E+00 / Cm245 0.533736E-01 / Cm246 0.699705E-02
[0070] As illustrated in [Fig.5], the data is graphically represented in a box simply named UOX. - Case of MOX REP fuel, recycling fuel from reprocessing, (URE REP management is described in a similar manner): • Mass of fuel in the core: 129 tmli • Cycle length: 515 JEPP (for Full Power Equivalent Days) • Unloading combustion rate: 55,000 MWd / tmli (for MW x day / tonnes of initial heavy metal) • Number of cycles: 3 • Average availability factor: 0.83 • Composition of the fuel at the outlet: It depends on the isotopy of the plutonium used for the production of MOX (MOX being a plutonium recycling fuel supported by depleted uranium). Many calculations have been carried out and the results are recorded by line in a *.dat file. Below are 3 with the presentation of the elemental contents followed by the letter "C" in the header, and the inlet compositions ending with "i" and those leaving the reactor with "o":
[0071] Pu238i Pu239i Pu241i Pu242i Am241i Pu240i PuCi U232o U234o U235o U236o U238o UCo Pu238o Pu239o Pu240o Pu241o Pu242o Am241o PuCo Np237o Np238o Np239o NpCo Am241po Am242mo Am243o AmCo Cm242o Cm243o Cm244o Cm245o Cm246o Cm247o Cm248o CmCo
[0072] 3.174314E-02 6.518413E-01 1.419446E-01 1.366310E-01 1.808024E-02 1.975971E-02 7.249880E-02 2.712770E-10 5.887010E-05 1.085280E-03 3.151180E-04 9.985410E-01 8.919850E-01 4.928650E-02 3.909880E-01 2.047680E-01 1.365190E-01 2.040720E-01 1.436620E-02 4.720220E-02 6.979450E-01 1.570200E-03 3.004850E-01 2.645060E-04 2.222890E-01 6.961910E-03 7.707500E-01 3.050610E-03 1.040730E-01 5.993070E-03 7.879660E-01 9.261630E-02 9.120010E-03 2.149790E-04 1.673980E-05 2.065520E-03
[0073] 6.631992E-03 5.442835E-01 3.552228E-02 1.333906E-01 2.120325E-02 2.589684E-01 1.347760E-01 3.423070E-10 4.369150E-05 1.417490E-03 2.934820E-04 9.982450E-01 8.332710E-01 1.990210E-02 3.790990E-01 3.025060E-01 1.255660E-01 1.548310E-01 1.809600E-02 1.050420E-01 7.506090E-01 1.159390E-03 2.482310E-01 2.591440E-04 3.681350E-01 1.396940E-02 6.178960E-01 5.163460E-03 1.368010E-01 8.046270E-03 7.572770E-01 9.262480E-02 5.130680E-03 1.142550E-04 6.351560E-06 2.255790E-03
[0074] 8.377948E-03 4.474735E-01 1.320619E-01 4.309932E-02 1.067962E-02 3.583077E-01 1.162200E-01 3.281480E-10 3.843540E-05 1.362610E-03 3.003210E-04 9.982990E-01 8.510790E-01 2.116840E-02 3.301470E-01 3.608460E-01 1.787250E-01 8.783090E-02 2.128280E-02 8.915190E-02 7.425810E-01 1.227740E-03 2.561910E-01 2.613650E-04 4.824220E-01 1.678660E-02 5.007920E-01 3.933070E-03 2.146380E-01 1.052030E-02 6.834020E-01 8.632760E-02 4.998350E-03 1.070520E-04 6.067460E-06 1.539480E-03
[0075] As illustrated in [Fig.5], this data is graphically represented in a box simply named MOX.
[0076] Similarly, a URE REP fuel (enriched reprocessed uranium) is defined, which requires an enrichment step, with a U235 content fixed in relation to the isotopy of the reprocessed uranium, and a composition of the output fuels dependent on the input composition thus defined.
[0077] Factory data: - Uranium flow: losses 0.001 - Plutonium flow: excluding Am241 losses 0.001, Am241 losses 1.0 After defining the input data in step E0, the method is applied below: below to different cases for examples.
[0078] Application case no. 1:
[0079] A first application case concerns a plutonium mono-recycling park in PWR defined by the cycle diagram of [Fig.5], entered by the user during step E1. The cooling times and the aging times are respectively defined here at 20 and 2 years.
[0080] In a step E2, this cycle diagram is transformed into a subsystem with the non-explicit variables which are defined: - The fraction of spent fuel UOX sent to the reprocessing plant is not defined, the procedure assumes that it is a free variable VAR1. - The fraction of materials separated by the reprocessing plant and used for the production of MOX is not defined, the procedure considers that it is a free variable VAR2.
[0081] The management characteristics indicate that the MOX fuel is manufactured from a separated plutonium of undetermined isotopy, the flow of separated materials between the reprocessing plant and the reintroduction of the MOX into the reactor is defined as being Pu.
[0082] The nuclear data entered defines plutonium as comprising 2 fissile isotopes, Pu239 and Pu241, the sum of the fractions of which defines the fissile quality; this adds 1 free variable VAR3, the fissile quality of plutonium.
[0083] MOX is not recycled: all spent MOX fuel is therefore sent to spent fuel storage.
[0084] There are only 2 distinct fuel managements, the distribution is a free variable, VAR4 being the fraction of UOX, (1-VAR4) the fraction of MOX.
[0085] [Fig.6] illustrates the resulting system which explains everything that is not included in the cycle diagram, the dotted lines indicate several flows of the same material.
[0086] At the end of step E2, it is identified that the equation setting would consist of balancing the mass flows on the Pu element (total and fissile). The constructed equation system contains 2 equations but 4 unknowns (VAR1 to VAR4), so that 2 variables must be fixed in order to be able to use techniques for solving non-linear equation systems that are the most efficient and rapid while having a fully posed problem. The procedure therefore returns to the definition of the data, following the condition described after step E2 in accordance with the progress of the procedure described in [Fig.3] for this application case, and which returns to step El (iterative process).
[0087] Application case no. 1 bis, by fixing 2 variables in addition to in case no. 1. The cooling times and the aging times are also respectively defined at 20 and 2 years.
[0088] The user therefore defines a variant of this first case concerning a plutonium mono-recycling park in PWR. The data set is here defined by the cycle diagram of [Fig.7] with solid lines indicating specified transfers of 100%. The user has fixed during a new step E1 two more variables than in the previous case so that it can produce results: these are the fractions of spent fuel and separated materials, at 100%.
[0089] In a step E2, this data set is transformed into a subsystem with the non-explicit variables which are defined ([Fig.8]).
[0090] The management characteristics indicate that the MOX fuel is manufactured from separated plutonium of undetermined isotopy. The flow of separated materials between the reprocessing plant and the reintroduction of MOX into the reactor is defined as plutonium.
[0091] The entered nuclear data defines plutonium as having 2 fissile isotopes, Pu239 and Pu241, the sum of whose fractions defines the fissile quality. This adds 1 free variable VAR1, the fissile quality of plutonium.
[0092] MOX is not recycled: all spent MOX fuel is therefore sent to spent fuel storage.
[0093] There are only 2 distinct fuel managements, the distribution is a free variable, VAR2 being the fraction of UOX, (1-VAR2) the fraction of MOX.
[0094] The resulting system, which explains everything not included in the cycle diagram, is illustrated in [Fig.8].
[0095] At the end of step E2, it is identified that the equation setting at a later step E4 will consist of balancing the mass fluxes on the Pu element (total and fissile). The single system of equations thus constructed contains 2 equations and 2 unknowns (VAR1 to VAR2). It is viable and the construction of the equations can therefore take place. Note that the decay chains are taken into account and the consequences on the isotopic balances are declined as a function of the cooling and aging times, fixed here. The cooling and aging times are respectively fixed here at 20 years and 2 years.
[0096] In a step E3, for the resolution of the system of equations, it is necessary to have a meta-modeling by neural networks of the composition at the input of the MOX reactors as a function of the isotopy of the plutonium actually recycled. This step is carried out by relying on the input data, results of neutron calculations making it possible to determine this on different isotopies scanning a certain range, which we will designate here as being the domain of isotopic validity of the new managements. It is also advisable not to re-train an existing neural network from the same data set if it already exists, in order to save time computation, the training of neural networks being clearly the limiting step in the application cases described here, with associated times of the order of a minute for each neural network when the rest of the code up to resolution generally executes in less than a second.
[0097] As an example, a multi-layer perceptron type neural network is indicated here used in solving the application cases presented here. This is the one evaluating the plutonium content of the MOX fuel as a function of the input Pu isotopy: • One hidden layer architecture: 5 neurons - 5 neurons - 1 neuron • Validity range: 14134e-05 < Pu238 < 0.0449874 / 0.350006 < Pu239 < 0.793175 / 0.000133562 < Pu240 < 0.199941 / 6.47882e-05 < Pu241 < 0.149995 / 0.00360508 < Pu242 < 0.50403 / Am241 variable not explained since deduced from the other isotopic mass fractions (unit sum) • Analytical expression: scaled_Pu238i=2*(Pu238i-1.14134e-05) / (0.04498741-1.14134e-05)-1;
[0098] scaled_Pu239i=2*(Pu239i-0.3500064) / (0.7931746-0.3500064)-1;
[0099] scaled_Pu241i=2*(Pu241i-0.0001335621) / (0.1999411-0.0001335621)-l;
[0100] scaled_Pu242i=2*(Pu242i-6.47882e-05) / (0.1499952-6.47882e-05)-l;
[0101] scaled_Pu240i=2*(Pu240i-0.003605082) / (0.5040303-0.003605082)-l;
[0102] y_l_l=tanh(-0.08517006143
[0103] -0.00407497733 l*scaled_Pu238i
[0104] +0.2230958367*scaled_Pu239i
[0105] +0.3139601793*scaled_Pu241i
[0106] +0.547924886 l*scaled_Pu242i
[0107] -0.1006251419*scaled_Pu240i);
[0108] y_l_2=tanh( 1.282340378
[0109] +0.09213480904*scaled_Pu238i
[0110] +1.824382626*scaled_Pu239i
[0111] +0.9519577864*scaled_Pu241i
[0112] +0.1633116632*scaled_Pu242i
[0113] +0.7685069638*scaled_Pu240i);
[0114] y_l_3=tanh(0.2964420625
[0115] -0.04221592342*scaled_Pu238i
[0116] +0.07093392605*scaled_Pu239i
[0117] +0.2466695046*scaled_Pu241i
[0118] -0.1356803758*scaled_Pu242i
[0119] -0.5887998225*scaled_Pu240i);
[0120] y_l_4=tanh(0.3446033673
[0121] +0.003124708328*scaled_Pu238i
[0122] -0.1637231303*scaled_Pu239i
[0123] -0.05698609715*scaled_Pu241i
[0124] +0.1082686964*scaled_Pu242i
[0125] -0.1087859567*scaled_Pu240i);
[0126] y_l_5=tanh(-0.3169908761
[0127] +0.007556951657*scaled_Pu238i
[0128] +0.3506223687*scaled_Pu239i
[0129] +0.06484478921*scaled_Pu241i
[0130] -0.4343194958*scaled_Pu242i
[0131] +0.1064831165*scaled_Pu240i);
[0132] scaled_PuCi=(0.3784407742
[0133] -0.5220428075*y_l_l
[0134] -1.760105089*y_l_2
[0135] +0.34330038*y_l_3
[0136] +0.2569349883*y_l_4
[0137] -0.6959011735*y_l_5);
[0138] (PuCi) = (0.5*(scaled_PuCi+1.0)*(0.201063-0.0636912)+0.0636912);
[0139] In a step E4, the balance equations are put into effect by passing the fixed values from the input data and the meta-models constructed in the previous step E3 if they do not exist, leaving the class 3 data to be determined free.
[0140] In a step E5, the resolution is extremely fast (less than one second).
[0141] Results of application case no. 1 bis: UOX VAR2 management fraction: 0.913875 / MOX management fraction: 0.086125 Composition of new MOX fuel: Plutonium content: 0.119621 Fissile quality of plutonium (sum of Pu239 and Pu241 fractions) VAR1: 0.590292
[0142] Description of application case no. 2:
[0143] A second, more complex application case concerns a recycling park for plutonium and uranium from UOX to PWR, with partial reintroduction of the spent MOX (15%) and URE (5%) fuel flow into the cycle. We pass the possible iterative steps of setting up a problem with as many equations as there are unknowns. The data set is here defined by the cycle diagram in [Fig.9] (after a step E1 of defining the case study). The different cooling and aging times are indicated in [Fig.9]. The cooling time here is 5 years and the aging time is 2 years.
[0144] During a step E1, the fractions of spent fuel URE and MOX sent to the storage are set by the user as 90% and 95% respectively (these are class 3 material destination type data). "URE" means a reprocessed uranium oxide fuel, isotopically enriched in uranium 235.
[0145] During a step E2, the management characteristics indicate that the MOX fuel is manufactured from a separated plutonium of undetermined isotopy, the flow of separated materials between the reprocessing plant and the reintroduction of the MOX into the reactor is defined as being Pu.
[0146] The management characteristics indicate that the URE fuel is manufactured from enriched reprocessed uranium. The flow of separated materials between the reprocessing plant and the reintroduction into the reactor of PURE is defined as PU. An enrichment plant is added between the flow of U and the reintroduction in the form of URE fuel into the reactor.
[0147] The nuclear data entered defines plutonium as comprising 2 fissile isotopes, Pu239 and Pu241, the sum of the fractions of which defines the fissile quality; this adds 1 free variable VAR1, the fissile quality of plutonium.
[0148] The nuclear data entered defines uranium as having 1 fissile isotope, U235, this adds 1 free variable VAR2, the U235 content of the reprocessed uranium.
[0149] There are three distinct fuel managements, the distribution is carried by two free variables, VAR3 and VAR4, the last fraction being deduced as being 1 minus the sum of the other two.
[0150] At the end of step E2, two systems are then constructed, with object indexing (all at "1"), which indicate that the fuel managements in the two systems are the same, just like the reprocessing plant. It would indeed be possible to use managements of the same nature but distinct at the cycle diagram level by differentiated indexing. [Fig. 10] illustrates the two reactor-plant systems for the equation setting in the second application case, at the end of step E2. It is identified that the equation setting at a later step E4 will consist of balancing the mass flows on the Pu element (total and fissile) and the U element (total and fissile). The single system of equations thus constructed contains 4 equations and 4 unknowns (VAR1 to VAR4), it is viable and the construction of the equations can therefore take place.Note that the decay chains are taken into account and the consequences on the isotopic balances are broken down according to the cooling and aging times, fixed here. The cooling time here is 5 years and the aging time 2 years.
[0151] In a step E3, for the resolution of the system of equations, it is necessary to have a meta-modeling by neural networks of the input and output compositions of the reactors based on the isotopy of the materials actually recycled. This step is carried out based on the input data, results of neutron calculations allowing this to be determined on different isotopies covering a certain range, which we will designate here as the isotopic validity domain of new management. It is also advisable not to re-train an existing neural network from the same data set if it already exists, in order to save computing time, the training of neural networks being clearly the limiting step in the application cases described here, with associated times of the order of a minute for each neural network when the rest of the code up to resolution generally runs in less than a second.
[0152] In a step E4, the balance equations are put into effect by passing the fixed values from the input data and the meta-models constructed in the previous step E3 if they do not exist, leaving the class 3 data to be determined free.
[0153] In a step E5, with a recycling park for plutonium and uranium from UOX to PWR, with partial reintroduction of the flow of spent MOX (15%) and URE (5%) fuels, the calculation is carried out in less than one second.
[0154] Results of application case no. 2:
[0155] - Fraction of UOX VAR3 management: 0.816575 / fraction of URE management: 0.090426 / fraction of MOX VAR4 management: 0.092999
[0156] - Composition of new MOX fuel: • Plutonium content: 0.110694 • Fissile quality of Pu VAR1: 0.605268 • Fissile content (U235 fraction) of reprocessed uranium before enrichment VAR2: 0.007492
[0157] - U235 enrichment of new URE fuel: 0.0525327
[0158] In both application cases, it can be verified that the Pu fissile qualities of the new MOX are within the validity domain of the experimental plan at the origin of the meta-modeled data, which is preferable to avoid extrapolations. Otherwise, a message would be returned by the code to indicate that the reactor data are not sufficiently complete to allow the fleet balance in question to be resolved.
[0159] The procedure for validating the results, disclosed in [Barale et al. "A new tool for the simulation of different nuclear fleets at equilibrium." EPJ Nuclear Sciences & Technologies 8 (2022): 1], is as follows: once the equilibrium state is defined, it is a matter of reproducing it in the reference dynamic scenario code, which is difficult to achieve. The underlying models should be the same as far as possible, in which case deviations can also be attributed to model differences. Moreover, as described in the prior art, fleet characteristics fluctuate by essence as a function of time in the dynamic simulation and are therefore fluctuating. It is therefore a question of establishing a sufficiently stable evolution over time so that, on average, these characteristics of the evolving nuclear fleet are identical to the static case characterized by the present procedure.
[0160] The automated procedure for determining nuclear parks at equilibrium of the invention provides the following advantages: - calculations requiring few computer resources and generally less than a minute; - the method relating to a system requires encapsulating reliable data on the elements constituting it, this is the minimum; - it is sufficient to define a cycle diagram to characterize a system; - one or more equilibrium situations will be found, if they exist; - procedures adapted to equilibrium systems, without time-related disturbance.
Claims
Claims
1. Method for determining the characteristics of a nuclear fleet at equilibrium, represented by a cycle diagram, for the management of a nuclear fuel, said nuclear fuel forming spent fuel after exploitation by said nuclear fleet, said spent fuel being capable of being reprocessed to constitute a recycling fuel and / or separated materials, said determination method comprising: - a step (EO) of generating input data, called class 1 data, such as nuclear data, plant data or reactor data; - a step (El) of translation by a user of a cycle diagram of the nuclear fleet, into so-called class 2 data; - a step (E2) of transforming said cycle diagram into reactor-plant systems from the class 1 data and the class 2 data;- a step (E3) of constructing balance equations from the reactor-plant systems, said balance equations forming a system of non-linear balance equations; - a step (E4) of selecting meta-models of the composition of the recycling fuel; - a step (E5) of computer resolution of the system of non-linear balance equations from the meta-models selected for determining characteristics of said nuclear fleet, called class 3 data.;
2. A determination method according to claim 1, wherein the computer resolution step (E5) comprises processing by neural networks of the selected meta-models.
3. A determination method according to claim 1 or claim 2, wherein the computer resolution step (E5) comprises processing the meta-models selected from kriging models.
4. Determination method according to any one of claims 1 to 3, in which the computer resolution step (E5) comprises processing the meta-models selected from chaos polynomials.
5. Determination method according to any one of claims 1 to 4, in which the computer resolution step (E5) allows a deter-
6. mination of an organization of the nuclear fleet, said organization comprising a list of several reactors associated with a fraction of use for each reactor, the sum of the fractions associated with the different reactors being equal to 1 so as to limit a number of unknowns in the system of non-linear balance equations. Determination method according to any one of claims 1 to 5 in which the class 3 data comprises at least one of the characteristics selected from the list of following characteristics: - a destination of the spent fuel; - a destination for separated materials from the reprocessing of spent fuel; - a transition time for separated materials from the reprocessing of spent fuel; - a composition of recycled fuel from the reprocessing of spent fuel.
Citation Information
Patent Citations
METHOD FOR PROTECTING A NUCLEAR REACTOR AND CORRESPONDING NUCLEAR REACTOR
FR3085788A1
Method for determining uncertainties associated with a model of a sedimentary basin
WO2022112065A1
