Method for determining asphaltene flocculation conditions in a hydrocarbon fluid

Abstract

The present invention relates to a method for determining flocculation conditions in a hydrocarbon fluid, according to which a flocculation model is applied to the hydrocarbon fluid, as well as a compositional model. Next, flocculation parameters obtained by the compositional model associated with a thermodynamic model and the flocculation model are compared to calibrate the compositional model.

Description

[0001]METHOD FOR DETERMINING ASPHALTENE FLOCCULATION CONDITIONS OF A HYDROCARBON FLUID Technical field The present invention relates to the field of characterizing a hydrocarbon fluid, with a view to optimizing its exploitation. Asphaltenes are the heaviest and most polar components of an oil fraction. A hydrocarbon fluid, also called petroleum fluid, is in fact a continuous sequence of hydrocarbons (alkanes, naphthenes, aromatics). Asphaltenes are defined, according to the French standard AFNOR T 60-115, as the fraction of a petroleum fluid that flocculates (precipitates) in n-heptane and is soluble in benzene. It is therefore a solubility class. These asphaltenes flocculate under the effect of three factors: variation in temperature, pressure or chemical composition. Indeed, the modification of the composition of the fluid can also induce the phenomenon of flocculation.The addition of light alkanes will produce flocculation of asphaltenes while the addition of aromatics will stabilize them in the crude oil. During well production, the hydrocarbon fluid undergoes variations in pressure, temperature and composition. It is therefore likely to flocculate, i.e. to form an aggregate of solid particles. This flocculation will result in a deposition of asphaltenes in the pores of the reservoir, which results in a modification of the porosity of the medium, its permeability and ultimately the production of hydrocarbons. In the ultimate cases, this flocculation can lead to the clogging of pores and cause the well to shut down. It is therefore important to be able to predict this phenomenon. In addition, asphaltene flocculation can represent a risk of plugging wells and production lines. When exploiting asphaltenic petroleum fluids, it is therefore essential to assess the risk of flocculation.This evaluation is based on the determination of the fluid flocculation curve, which represents the amount of flocculated asphaltene as a function of pressure for a given temperature (see example figure 1). Figure 1 illustrates a curve of the amount of asphaltene w precipitated as a function of pressure P for a given temperature. This curve is called the flocculation curve. Starting from low pressure, flocculation starts from a limiting pressure called the low threshold pressure (P. L ). The maximum asphaltene w max flocculated is reached near the saturation pressure of the fluid (P sat ). By continuing to increase the pressure, the quantity of flocculated asphaltene then decreases, until it disappears at the high threshold pressure (P U). Flocculation is at its maximum (and the quantity of asphaltenes in the liquid at its minimum) when the bubble pressure (appearance of the first vapor bubble) is reached. Below this pressure, the appearance of a vapor phase rich in alkanes, and therefore in flocculating products, allows the liquid to become welcoming to asphaltenes again. Prior art The flocculation curve of a hydrocarbon fluid can be determined by means of laboratory experiments, but these experiments can be long and costly. The use of predictive models to determine this curve is therefore an interesting alternative,in particular for software dedicated to reservoir simulation and the transport of petroleum fluids. Patent application FR2836719 (US2003 / 0167157) describes a method for predicting a flocculation curve of a petroleum fluid from homothetic transformations of the flocculation curve of a reference fluid. The flocculation curve of the reference fluid is obtained beforehand by experimental measurements. This method therefore requires experimental data in order to have a flocculation curve. The document: Sanchez, NL A General Approach for Asphaltene Modeling, SPE-107191-MS, Latin American & Caribbean Petroleum Engineering Conference, Buenos Aires, Argentina,April 2007 describes a method for determining a flocculation curve. This method involves decomposing the fluid into seven constituents and fitting the composition of the heavy constituent to experimental flocculation data. This method requires experimental data in order to obtain a flocculation curve. The document: Nghiem, LX; Khose, BF; Farouq Ali, SM Asphaltene Precipitation: Phase Behavior Modeling and Compositional Simulation, SPE-59432, Asia Pacific Conference on Integrated Modelling for Asset Management, Yokohama, Japan, April 2000. describes a method for determining a flocculation curve. This method proposes separating the heavy constituent of the fluid into two fractions, and treating the asphaltene phase as a solid phase. The fluid fitting involves binary interaction parameters between the heavy and light constituents,based on experimental flocculation data. This method requires experimental data in order to have a flocculation curve. The document: Szewczyk, V.; Behar, E. Compositional model for Predicting Asphaltenes flocculation, Fluid Phase Equilibria 185-160 (1999) 459-469 describes a method for representing the fluid according to a standardized representation with 33 constituents, and a calibration of the molar mass and the critical temperature of the heavy constituent based on experimental flocculation data. This method requires experimental data in order to have a flocculation curve. In addition, the models used in this representation are complex, and therefore cumbersome to implement with computer resources (computer memory, processors). Summary of the invention The aim of the invention is to determine asphaltene flocculation conditions of a hydrocarbon fluid, in a precise, simple and rapid manner,without requiring experimental measurements. For this, the invention relates to a method for determining flocculation conditions of a hydrocarbon fluid in which a flocculation model is applied to the hydrocarbon fluid, as well as a compositional model. Then, flocculation parameters obtained by the compositional model associated with a thermodynamic model and the flocculation model are compared to calibrate the compositional model. The flocculation model of the fluid makes it possible to dispense with experimental measurements to calibrate the compositional model. Thus, it is possible to obtain the flocculation conditions simply and quickly. The invention also relates to a method for determining or predicting asphaltene flocculation, as well as a method for exploiting a hydrocarbon fluid. The invention relates to a method for determining asphaltene flocculation conditions of a hydrocarbon fluid. For this method,the following steps are implemented: a) A flocculation model is applied to said hydrocarbon fluid to determine at least one modeled flocculation parameter, said flocculation model linking the composition of a fluid to at least one flocculation parameter; b) A compositional model is constructed of said hydrocarbon fluid, said compositional model decomposing said hydrocarbon fluid into several constituents,and at least one physical property of said hydrocarbon fluid is determined by means of said compositional model; c) a thermodynamic model is applied to said at least one physical property determined by means of said compositional model to determine at least one flocculation parameter; d) said compositional model of said hydrocarbon fluid is calibrated by minimizing a difference between at least one flocculation parameter obtained from said thermodynamic model and said at least one modeled flocculation parameter obtained from said flocculation model; and e) the asphaltene flocculation conditions of said hydrocarbon fluid are determined from said calibrated compositional model of said hydrocarbon fluid. According to one embodiment, the method comprises a prior step of constructing said flocculation model by machine learning,in which the following steps are implemented: i) A learning base of learning fluids is constructed, each learning fluid being characterized by a composition and at least one flocculation parameter, preferably said at least one flocculation parameter of said learning fluids being obtained experimentally; and ii) A flocculation model is constructed by means of a machine learning method trained on said learning base, said flocculation model linking the detailed composition of a fluid to said at least one flocculation parameter. Advantageously, each flocculation parameter is chosen from a low threshold pressure, a high threshold pressure, a saturation pressure, a maximum quantity of flocculated asphaltene. Advantageously, said compositional model is based on a representation in N constituents, N being between 6 and 15, preferably between 6 and 12,and preferably worth 10. According to one aspect, said compositional model is calibrated by means of an iterative method. According to one implementation, said thermodynamic model is an algorithm for calculating the three-phase liquid-liquid-vapor equilibrium of a fluid at a given temperature and at a given pressure. According to one embodiment option, said flocculation conditions correspond to a flocculation curve which relates the quantity of flocculated asphaltene to the pressure of the fluid. According to one embodiment, the at least one physical property is chosen from the molar mass, the critical temperature, the critical pressure, the acentric factor of the constituents of the fluid. In addition, the invention relates to a method for determining or predicting asphaltene flocculation of a hydrocarbon fluid within an underground formation or within a fluid transport pipe or within a conversion reactor,in which the following steps are implemented: a) Asphaltene flocculation conditions of said hydrocarbon fluid are determined by means of the method for determining asphaltene flocculation conditions according to one of the preceding characteristics; b) A pressure and a temperature of said hydrocarbon fluid within said underground formation or within said transport pipe or within a conversion reactor are determined by measurement or predicted by simulation; and c) A possible asphaltene flocculation of said hydrocarbon fluid is determined or predicted as a function of said determined flocculation conditions and the determined or predicted pressure and the determined or predicted temperature. Furthermore, the invention relates to a method for exploiting a hydrocarbon fluid within an underground formation or within a fluid transport pipe or within a conversion reactor,in which the following steps are implemented: d) Asphaltene flocculation of said hydrocarbon fluid is determined or predicted by means of the method for determining or predicting flocculation according to one of the preceding characteristics; and e) Said hydrocarbon fluid is exploited as a function of the possible flocculation of asphaltene of said hydrocarbon fluid, in particular by modifying the temperature, and / or the pressure and / or the composition of the hydrocarbon fluid in the event of asphaltene flocculation and / or by adding solvent. Other characteristics and advantages of the method according to the invention will appear on reading the following description of non-limiting examples of embodiments, with reference to the appended figures described below. List of figures Figure 1, already described,illustrates an asphaltene flocculation curve. Figure 2 illustrates the steps of the method according to a first embodiment of the invention. Figure 3 illustrates the steps of the method according to a second embodiment of the invention. Figure 4 illustrates the steps of the method according to a third embodiment of the invention. Figure 5 illustrates the steps of the method according to a fourth embodiment of the invention. Figure 6 illustrates, for an example, a comparison between experimental pressures, pressures obtained using the flocculation model and pressures determined by an embodiment of the invention. Figure 7 illustrates for the example of Figure 6 asphaltene flocculation curves determined using the method according to an embodiment of the invention. Description of the embodiments The present invention relates to a method for determining asphaltene flocculation conditions of a hydrocarbon fluid,in other words a method for determining flocculation conditions of an asphaltene contained in a hydrocarbon fluid. The hydrocarbon fluid is in particular a petroleum fluid, for example oils, in particular crude oils. For example, it may be a fluid produced by an underground formation and / or a fluid present in a petroleum fluid transport pipe and / or a fluid within a conversion reactor (for example a desalting reactor). The flocculation conditions are the conditions for generating asphaltene flocculation. They may for example include, for a given temperature, pressure conditions (for example the low threshold pressure, the high threshold pressure,the saturation pressure of the fluid) and the maximum quantity of flocculated asphaltene. The low threshold pressure is the pressure from which asphaltene flocculation starts. The high threshold pressure is the pressure from which asphaltene flocculation disappears. The saturation pressure is the pressure at which the quantity of flocculated asphaltene is maximum. Preferably, the asphaltene flocculation conditions can make it possible to define the flocculation curve of the hydrocarbon fluid studied for a given temperature. As a reminder, the flocculation curve links, for a fluid, the quantity of flocculated asphaltene to the pressure for a given temperature. Thus, the method according to the invention makes it possible to define characteristics linked to the physics and chemistry of a hydrocarbon fluid. Advantageously, the composition of the fluid studied can be known. Alternatively,the method according to the invention may comprise a preliminary step of determining the composition of the fluid studied. In addition, the density and bubble pressure of the fluid studied, or other similar characteristics, may also be known. The method according to the invention implements the following steps: 1) Application of a flocculation model 2) Construction of a compositional model 3) Application of a thermodynamic model 4) Calibration of the compositional model 5) Determination of the asphaltene flocculation conditions These steps may be implemented by computer means, in particular by a computer or by a server. These steps will be detailed in the remainder of the description. Figure 2 illustrates, schematically and in a non-limiting manner, the steps of the method according to a first embodiment. For a fluid to be studied FLU,a flocculation model MOD FLO is applied to determine modeled flocculation parameters. A compositional model MOD COM of this fluid FLU is also constructed. A thermodynamic model MOD THE is then applied to the compositional model. The compositional model MOD COM is then calibrated using the modeled flocculation parameters. The calibrated compositional model then makes it possible to determine flocculation conditions CDF. According to one embodiment of the invention, the method may comprise a preliminary step of constructing the flocculation model by machine learning. For this embodiment,the method according to the invention may comprise the following steps: A) Construction of a learning base B) Construction of the flocculation model 1) Application of a flocculation model 2) Construction of a compositional model 3) Application of a thermodynamic model 4) Calibration of the compositional model 5) Determination of the asphaltene flocculation conditions Steps B and 1 to 5 may be implemented by computer means, in particular by a computer or by a server. Steps A and B may be carried out only once beforehand. If it is desired to apply the method according to the invention to several fluids, only steps 1 to 4 are repeated for each fluid. These steps will be detailed in the remainder of the description. Alternatively, steps A and B may be repeated, adding, at each iteration, the last fluid studied to the learning base. Thus, the learning base is completed at each iteration,and the flocculation model can become more precise. Figure 3 illustrates, schematically and in a non-limiting manner, the steps of the method according to such an embodiment. The steps already described for Figure 2 are not detailed again. The method further comprises a step of constructing a BAP learning base. This BAP learning base allows training of an APP automatic learning method to construct the MOD FLO flocculation model. A) Construction of the learning base During this optional step, a learning base is constructed from a plurality of learning fluids. The learning fluids are hydrocarbon fluids. Each learning fluid is characterized by a composition and at least one flocculation parameter for a given temperature. Each flocculation parameter can be chosen from the low threshold pressure denoted P, L , the high threshold pressure noted P U, the saturation pressure noted P sat and the maximum quantity of flocculated asphaltene noted W max. According to one embodiment, the flocculation parameters for at least one training fluid can be experimentally determined. For example, one of the methods described in the document: “Pina et al., Oil & Gas Science and Technology – Rev. IFP, Vol. 61 (2006), No. 3” can be applied. For example, by gravimetry (see paragraph 5.1), by acoustic resonance, by light scattering (see paragraph 5.2) or by filtration (see paragraph 5.3). Alternatively or cumulatively, the flocculation parameters for at least one training fluid can be determined from at least one published documentation, which discloses experimental results. According to one implementation of the invention, each training fluid can be described by a vector of several constituents (for example according to a compositional model as described in step 2). This decomposition allows a good representation of each training fluid.In particular, the compositions of each learning fluid can be described by a vector comprising from 8 to 33 elements, preferably from 10 to 20 elements. For example (non-limiting), the compositions of each fluid can be described by a vector of 14 elements according to the following distribution: • four gases: H2S, N2, CO2, CH4; • five light pseudo-components in the C2 to C6 cut; • an intermediate pseudo-component for the C cut. 7+ ; • the density and molecular weight for the C cut 7+ ; • two heavy pseudo-components, one per fraction in Resins and Asphaltenes resulting from a SARA analysis on section C 20+. SARA analysis is a fractionation-based heavy oil characterization method in which a heavy oil sample is separated into smaller quantities or fractions, each fraction having a different composition. Fractionation is based on the solubility of the hydrocarbon components in various solvents. Each fraction consists of a solubility class containing a range of species of different molecular weight. In this non-limiting example, crude oil is fractionated into two solubility classes Resins and Asphaltenes, among the constituents collectively referred to as SARA (SARA denotes respectively: Saturated Hydrocarbons S, Aromatics A, Resins R and Asphaltenes A). Saturated hydrocarbons are generally iso- and cyclo-paraffins, while aromatics, resins and asphaltenes form a continuum of molecules with increasing molecular weight, aromaticity and heteroatom content.Asphaltenes may also contain metals such as nickel and vanadium. This method is sometimes referred to as asphaltene / wax / hydrate deposit analysis. According to one aspect, the learning base may comprise a number of learning fluids at least equal to the number of variables (the number of variables may correspond to the sum of constituents of the compositional model and the number of physical parameters). Advantageously, the number of learning fluids may comprise a number of learning fluids at least equal to twice the number of variables, in order to allow good representativeness of the different fluids and to partially overcome the problem of overfitting, to ensure good reliability of the flocculation model. B) Construction of the flocculation model During this step, which is also optional, a flocculation model is constructed which associates at least one flocculation parameter with a fluid composition.In other words, the flocculation model makes it possible to determine at least one flocculation parameter for a hydrocarbon fluid as a function of its composition and its temperature. During this step, the flocculation model is constructed using a machine learning method trained using the learning base constructed as described in step A. During this step, the flocculation model can be constructed by machine learning, preferably by supervised machine learning, preferably by supervised regression machine learning. According to an exemplary embodiment, the flocculation model can be written: ^^ ^^ ^^ ^^ ^^. ^ ൌ ^^ ^ ^^ ^^ ^^ ^^, ^^ ^ with f a function, ^^ ^^ ^^ ^^ ^^ ^a flocculation parameter (e.g., low threshold pressure, high threshold pressure, saturation pressure, or maximum amount of flocculated asphaltene), T the fluid temperature, and Comp the fluid composition. The function f can be determined from a supervised learning algorithm without variable selection. The models of such an approach are linear models, since each element of the vector is associated with a weighting coefficient, and for an oil, the sum of the products of the elements of the vector by their respective coefficients leads to the predicted value. Alternatively, the learning algorithm can be a support vector machine (SVM), a neural network, a random forest, or a combination of these methods.For the embodiment, for which the function f can be determined from a supervised learning algorithm with variable selection, the models can be non-linear approaches that can be obtained by genetic programming (MGGP, Multi-Gene Genetic Programming in English). Such models can be constructed by combining elements of the vector with mathematical operators (addition, subtraction, division, product, etc.) according to a process of genetic evolution (model selection, crossover, mutation, etc.). For both approaches, the models can be trained on one portion of the database, and tested on the other portion of the database (data external to the training process). According to one aspect, this step can include a validation method, preferably a cross-validation method, in particular a k-fold cross-validation method.This cross-validation method helps reduce overfitting problems and improve model accuracy. of a flocculation model In this step, a flocculation model is applied to the hydrocarbon fluid to determine at least one modeled flocculation parameter, also called the first flocculation parameter. The flocculation model may be known beforehand or, if necessary, may be the one constructed during the optional steps A and B. The flocculation model links the composition of the hydrocarbon fluid to at least one flocculation parameter for a given temperature. In other words, in this step, at least one flocculation parameter of the hydrocarbon fluid is determined by means of a flocculation model, which takes the composition of the fluid and the temperature as input. The use of such a model makes it possible to determine at least one flocculation parameter of the hydrocarbon fluid without experimenting with the hydrocarbon fluid. Thus, determining flocculation conditions is simpler and faster.Preferably, the flocculation model may be a data-trained model (as is the case for the model constructed in the optional steps A and B). Thus, such a model does not require a priori knowledge of the physical and / or chemical phenomena for the determination of the flocculation parameters. Each flocculation parameter may be chosen from the low threshold pressure denoted P. L , the high threshold pressure noted P U , the saturation pressure noted P sat and the maximum quantity of flocculated asphaltene noted W max . 2) Construction of a model In this step, a compositional model of the hydrocarbon fluid is constructed. The compositional model decomposes the hydrocarbon fluid into several constituents. Thus, such a model allows a simplified and structured representation of the hydrocarbon fluid, which simplifies the determination of the flocculation conditions. Preferably, the compositional model of the hydrocarbon fluid can be constructed by means of a SARA analysis, as described for step A. In addition, in this step, at least one physical property of the hydrocarbon fluid is determined based on the decomposition of the hydrocarbon fluid. Indeed, certain physical parameters are directly linked to the composition of the fluid. Each physical property can be chosen from the molar mass, the acentric factor, the critical pressure and the critical temperature of the constituents of the hydrocarbon fluid.It is recalled that the critical temperature and the critical pressure are respectively the highest temperature and the highest pressure at which there can be a liquid-vapor equilibrium. The acentric factor is a number used in the description of matter in thermodynamics. For the embodiment for which parameters of the constituents of the hydrocarbon fluid are known, the initial fluid can be calibrated using these parameters, for example the bubble pressure. According to one embodiment of the invention, the compositional model can be based on a representation in N constituents. Advantageously, N can be between 6 and 15. Indeed, a decomposition into 6 constituents allows a representation with a gas, a light cut and the SARA components (from a SARA analysis). Thus, if fewer than 6 constituents are considered, the representation risks being less precise.Furthermore, with a decomposition beyond 15 constituents, the precision is improved at the expense of the complexity of the representation, and consequently, at the expense of the complexity of calculations, time and computer memory used to implement the invention. Preferably, N can be between 6 and 12, advantageously between 8 and 12, and can be 10. These values ​​allow a good compromise between precision of the model, and time and computer memory necessary for the calculations. For the exemplary embodiment, for which N is 10, the representation in 10 constituents can be as follows: ^ 4 gases: N2, H2S, CO. 2, CH4^ 1 light pseudo-component in the C2 to C6 cut^ 1 intermediate pseudo-component in the C cut 7+ to C 20 ^ 4 heavy pseudo-components, one per fraction (on section C 20+344 °C) in Saturates, Aromatics, Resins, Asphaltenes (SARA) The pseudo-components SARA are considered in this approach as representative of the heavy fraction C 20+, with a number of carbon atoms nC greater than 20. In the case where there are no heavy components with nC > 20 in the initial fluid, it is considered that there is no risk of asphaltenic deposit, and it is not useful to implement the other steps of the method according to the invention. A non-limiting example of implementation of such a compositional model with 10 constituents is described in the remainder of the description. Other approaches can be implemented, for example the method described in the following document or any analogous method: Szewczyk V., Béhar E., Compositional model for predicting asphaltenes flocculation, Fluid Phase Equilibria, 156-160, p.459-469, 1999 The first step in this implementation of the compositional model can consist of grouping all the m heavy components, whose carbon number nC is greater than 20, within a single pseudo-component hComp.The average critical properties of this pseudo-body are calculated using the Montel-Gouel clustering method. This step gives an intermediate fluid denoted φ1 which is composed of (n - m +1) constituents: (nm) light and intermediate components, and 1 heavy pseudo-component hComp. It should be noted that there is possibly a part C. 20- in this pseudo-body hComp: this is the case where the initial fluid given by the user does not have a cut in C 20 and there are some components among the m constituents having a lighter part with nC < 20. Example: we consider an initial fluid which comprises, apart from light and intermediate components (nC < 20), two heavy components, one in the C section 12 C 25 (with nC = 22) and the other in the C cut 25+ (with nC = 32). The goal is to convert these two components to C 12 C 20 and SARA. In the first step, the pseudonyms C 12 C 25 etc25+ are grouped into a single pseudo C 12+ . In the second step of this implementation of the compositional model, the pseudo-component hComp of the intermediate fluid φ1 obtained via the previous step is cut to introduce the 4 pseudo-components SARA. The challenge therefore comes back to the decomposition of the hComp into two parts: a part C 20+ corresponding to the SARA fractions and a lighter part C 20- , here noted pC 20- . In the example, you have to divide the pseudo C 12+ in C 12 C 20 and SARA. This action is done based on the conservation of mass between the two representations. The physical properties (M w , T c , P c , and ω) of the mixture of 5 new pseudo-constituents must be equivalent to those of the pseudo body hComp (C 12+) from step 1. For this, an iterative approach can be used. This is a constrained optimization with 4 unknowns: the molar mass of the pseudo-compound pC 20 - and those of the three components Saturated, Aromatic and Resin (SAR). The molar mass of asphaltenes is fixed at 1.0 kg / mol. The constraints to be respected for this iterative approach can be: ^ Inequality constraints: M w, Saturés < M w, Aromatiques < M w, Résines < M w, Asphaltènes = 1.0 kg / mol ^ Equality constraints (4 constraints): With i denoting a compound of the SARA group or a pseudo-compound pC 20 -, M w the molar mass, Tc the critical temperature, Pc the critical pressure, ^^ the mole fraction of the component. In these equations, the mole fractions x i can be calculated based on a SARA analysis, and the closure condition: o ^^ ^^ଶ^ି ^ ∑ ^∈ௌ^ோ^ ^^ ^ൌ 1.0 With wti the mass fraction of SARA compounds. The limits to be respected for this iterative approach can be the following: ^ M w , pC20- belongs to the interval [0.045 , MW_C20] where 0.045 kg / mol is the minimum valid molar mass for pseudo components and MW_C20 (282.55E-3 kg / mol) corresponds to the molar mass of n-eicosane. ^ M w, Saturés , M w, Aromatiques , M w, Résines belong to the interval [MW_C20,1.0]. At the end of this optimization process, we replace the pseudo-component hComp with the set of new components (SARA and possibly pC 20- ). The intermediate fluid φ2 is therefore composed of (n-m+4) or (n-m+5) components, corresponding respectively to the case without and with pC 20-. The third step of this implementation of the compositional model can be to create the asphaltene fluid of 10 components by referring to the components of the standardized representation of the target fluid by the components of the fluid φ2. First, the gases (N2, CO2, H2S, CH4) are brought back into the final fluid if they are present in the fluid φ2, or failing that, zero mole fractions are identified. Then, the light and intermediate hydrocarbons are classified according to their number of carbon atoms and grouped into two pseudo-bodies C2C7with the number of carbon atoms varying between 2 and 7, and C 7+ C 20with the number of carbon atoms between 7 and 20. Finally, the SARA components characterized in the second step are brought back into the final fluid. Regarding the constitutional grouping, the descriptions for gases and for SARA fractions are already defined. The descriptions for the other two pseudo-bodies can be considered as two groups CH3 and (nC i – 2) CH groups 2, where nC i is the number of carbon atoms in each pseudo-body. Then, at least one physical property of the hydrocarbon fluid can be determined by means of decomposition and group contribution method-based approaches. For this, a classical Montel and Gouel clustering method can be implemented for the light and intermediate pseudo-components (C2C6and C 7+ C 20) and an Avaulée approach for heavy fractions (SARA pseudo-compounds). Other approaches can be considered. The classical Montel and Gouel clustering method is described in the following document: Montel, F. and Gouel, PL1984. A New Lumping Scheme of Analytical Data for Compositional Studies. In SPE Annual Technical Conference and Exhibition. Society of Petroleum Engineers. This method allows the calculation of critical parameters such as the critical temperature T c , the critical pressure P c , the molar mass M w , and the acentric factor ω. This method can define the following relationships: with i and j the constituents of the composition. The Avaulée approach, also called Avaulée correlation, is described in the following document: Avaullée, L.1996. Development of methods for the thermodynamic characterization of petroleum fluids with a view to predicting their volumetric properties and gas injection experiments in reservoir oils, thesis from the University of Aix-Marseille, Aix-en-Provence, France, Thesis number 96 AIX30050 In this step, a thermodynamic model is applied to the compositional model determined in step 2. The thermodynamic model allows thermodynamic balancing of the different phases of the hydrocarbon fluid, in order to determine a flocculation parameter from the compositional model. The thermodynamic model has in particular as input at least one physical property (preferably all the physical properties) from the compositional model, and has as output at least one flocculation parameter, also called second flocculation parameter. According to one embodiment, a thermodynamic model can implement a thermodynamic equilibrium calculation (called flash algorithm) capable of managing the simultaneous existence of a vapor phase, a hydrocarbon liquid phase, and a second asphaltenic liquid phase. It can then be a three-phase liquid-liquid-vapor equilibrium calculation.For non-limiting example, the flash algorithm used may be that proposed by Michelsen in the following documents: Michelsen, ML, Fluid Phase Equilibria, 9 (1982), 1-19 Michelsen ML, Fluid Phase Equilibria, 9 (1982), 21-40 The thermodynamic model used for this calculation in this algorithm may be the Peng & Robinson cubic equation of state coupled with the Abdoul-Péneloux mixing law, as described in particular in the following documents: Abdoul, W., 1987, A group contribution method applicable to the correlation and prediction of thermodynamic properties of petroleum fluids, Thesis of the University of Aix-Marseille III, France; and Péneloux, A., Abdoul, W. and Rauzy, E., 1989, Excess functions and equation of state, Fluid Phase Equilibria, 47, 115-132. Such a pattern can be written: ^^ ൌ. ோ் ^ ^ ^்^ ௪ ^ି^^െ௪^^௪^ାఊ^^^Where T and P are the temperature and pressure of the system respectively, R is the universal gas constant (R ^ 8.314 kPa⋅L / mol⋅K), ^^ ൌ 4.82843 is the characteristic constant of the equation, ^^ ^ indicates the uncorrected molar volume, a i (T) is the interaction parameter and b i denotes the covolume of component i. These parameters are calculated for the critical temperature T c,i , the critical pressure P c,i, and the acentric factor from the compositional model. Furthermore, The Abdoul-Péneloux approach considers the volume translation, where the "true" molar volume ^^ ^ is corrected by a term ^^ ^ : Where Z Ra,iis the Rackett compressibility factor. These equations are valid for pure substances. For a mixture, the entire mixture can be considered as a single constituent following the same formalism. The Abdoul-Péneloux mixing rule is applied to the critical properties of pure substances to obtain the global mixture parameters a, b, c. At each given temperature, a series of three-phase flash calculations (in other words, several flash algorithms are applied) can be performed by varying the pressure to obtain the entire flocculation curve as shown schematically in Figure 1. This approach also allows determining key pressures such as the saturation pressure P sat and low threshold pressures P L and high P U . In this step, at least one physical property of the compositional model determined in step 2 of the hydrocarbon fluid is calibrated, by minimizing a difference between at least one flocculation parameter obtained from the thermodynamic model in step 3 (second flocculation parameter) and the at least one modeled flocculation parameter determined in step 1 (first flocculation parameter). In other words, the compositional model is adjusted so that it behaves in a similar manner to the flocculation model. This adjustment concerns the setting of at least one physical property of the compositional model so that a flocculation parameter from the compositional model comes as close as possible to a modeled flocculation parameter from the flocculation model. This calibration step makes it possible to improve the quality of prediction of the asphaltene flocculation conditions.According to one embodiment of the invention, the calibration can implement an iterative method, in which at least one physical property is modified at each step until a minimum is obtained between the flocculation parameter from the compositional model and the modeled flocculation parameter. For this iterative method, it is possible to seek to minimize an objective function which includes a term measuring a difference between the calculated values ​​of the flocculation parameters and the modeled flocculation parameters. According to one implementation of the invention, the physical properties to be calibrated can be the molar mass and the critical temperature. The other physical properties of the asphaltene constituent (for example the critical pressure and the acentric factor) can be determined as a function of the molar mass and the critical temperature by the Avaullée approach.This approach provides a better characterization of very heavy components compared to other approaches. 5) Determination of asphaltene flocculation conditions In this step, the flocculation conditions of the hydrocarbon fluid are determined from the flocculation parameter from a thermodynamic model applied to the calibrated compositional model of the hydrocarbon fluid. Thus, the adjusted decomposition of the hydrocarbon fluid makes it possible to determine the implementation of the flocculation, for example to determine a flocculation curve of the hydrocarbon fluid at a given temperature. In addition, the invention relates to a method for determining or predicting asphaltene flocculation of a hydrocarbon fluid within an underground formation or within a fluid transport pipe or within a conversion reactor (for example for a desalting process), in which the following steps are implemented: a.Asphaltene flocculation conditions of the hydrocarbon fluid are determined by means of the method for determining asphaltene flocculation conditions according to any of the variants or combinations of variants described above; b. A pressure and a temperature of the hydrocarbon fluid within the underground formation or within the transport pipeline or within a conversion reactor are determined or predicted by simulation or measurement; and c. Possible asphaltene flocculation of the hydrocarbon fluid is determined or predicted as a function of the determined flocculation conditions and the determined or predicted pressure and the determined or predicted temperature. Thus, by means of this method, possible asphaltene flocculation can be determined during the operation of such a fluid, and possible asphaltene flocculation can also be simulated within a simulation of an operation of such a fluid.The exploitation of a hydrocarbon fluid may in particular concern the recovery of hydrocarbons within an underground formation, the transport of hydrocarbons from the underground formation or the transport of a petroleum fluid, the treatment of a hydrocarbon fluid within a conversion reactor (for example for a desalting process). Thus the simulation concerned may be a simulation of transport of hydrocarbon fluid within an underground formation or a simulation of transport of hydrocarbon fluid within a transport pipe or a reaction of the hydrocarbon fluid within a conversion reactor. Advantageously, steps a) and c) may be implemented by computer means, in particular a computer or a server. Advantageously, the transport pipe may be a pipe in a hydrocarbon production well, or a pipe of a hydrocarbon production line.Furthermore, when the method allows simulation, step b) can also be implemented by computer. Figure 4 illustrates, schematically and in a non-limiting manner, the steps of this method according to one embodiment. The steps already described for Figure 2 are not detailed again. The method comprises an additional step of determining the prediction DET of a possible flocculation of asphaltene FDA, which depends on the flocculation conditions CDF, and the operating conditions COP, in other words the temperature and pressure determined or predicted. This figure does not represent the possible step of measuring or simulating the operating conditions COP. Furthermore, the embodiment of Figure 4 can be combined with the embodiment of Figure 3.Furthermore, the invention relates to a method for exploiting a hydrocarbon fluid within an underground formation or within a fluid transport pipeline or within a conversion reactor, in which the following steps are implemented: 1. Asphaltene flocculation of the hydrocarbon fluid is determined or predicted by means of the method for determining or predicting flocculation according to any one of the variants or combinations of variants described above; and 2. The hydrocarbon fluid is exploited depending on the possible asphaltene flocculation of the hydrocarbon fluid, in particular by modifying the temperature, and / or the pressure and / or the composition of the hydrocarbon fluid and / or by adding a solvent in the event of asphaltene flocculation.Thus, if a risk of asphaltene flocculation is determined or predicted, the hydrocarbon fluid exploitation process can be adapted, and thus asphaltene flocculation can be avoided or at least asphaltene flocculation can be limited. In this way, control of the exploitation of a hydrocarbon fluid can be ensured. The exploitation of a hydrocarbon fluid can in particular concern the recovery of hydrocarbons within an underground formation, the transport of hydrocarbons from the underground formation or the transport of a petroleum fluid between two installations or two sites, a de-alphatation process in a conversion reactor. For this, it is possible in particular to increase or reduce the temperature of the fluid, and / or to increase or reduce the pressure of the fluid and / or to adjust the composition of the fluid and / or to add a solvent. Figure 5 illustrates, schematically and in a non-limiting manner, the steps of this method according to one embodiment.The steps already described for Figure 4 are not detailed again. The method includes an additional step of EXP exploitation of the hydrocarbon fluid as a function of the possible flocculation of FDA asphaltene. This figure does not represent the possible step of measuring or simulating the COP operating conditions. In addition, the embodiment of Figure 5 can be combined with the embodiment of Figure 3. Application example The characteristics and advantages of the method according to the invention will appear more clearly upon reading the application example below. For this example, the aim is to determine the flocculation curve of a fluid considered and noted Oil_9 in the following document: Sullivan, M., et al., 2020, A Fast Measurement of Asphaltene Onset Pressure. SPE Reservoir Evaluation & Engineering 23 (03), 962-978.The experimental data of this document, such as the saturation pressure and the low threshold pressure at three different temperatures 75°C, 100°C and 125°C, are used for the validation of the models according to the invention. For this example, the steps of the method according to the invention are applied. In particular, the flocculation model is built by machine learning, from a learning base that includes 53 fluids. In addition, for this example, a compositional model is applied with a decomposition of the fluid into 10 constituents: ^ 4 gases: N2, H2S, CO. 2, CH4^ 1 light pseudo-component in the C2 to C6 cut^ 1 intermediate pseudo-component in the C cut 7+ to C 20 ^ 4 heavy pseudo-components, one per fraction (on section C 20+344 °C) in Saturates, Aromatics, Resins, Asphaltenes (SARA) The calibration of the asphaltene constituent of the compositional model is carried out on the saturation pressures and the low threshold pressures from the flocculation model. In addition, the thermodynamic model used for this calculation can be the Peng & Robinson cubic equation of state coupled with the Abdoul-Péneloux mixing law. Figure 6 illustrates, for this example, the pressures P in MPa as a function of the temperature T in K. In this figure, the Calc index indicates calculated (i.e. obtained by the thermodynamic model and the calibrated compositional model), the Exp index indicates experimental (from the cited document) and the ML index indicates obtained by the flocculation model of the process according to the invention. P sat denotes the saturation pressure and P Udenotes the low threshold pressure. The calculated pressure curves Calc reproduce reference experimental data, which makes it possible to validate the calibration implemented in the process according to the invention. These curves are generated by considering a wider temperature range to visualize the evolution of key pressures (between 0°C and 250°C). In this figure, we note that the pressures obtained using the flocculation model are consistent with the experimental data presented in the document (Sullivan et al., 2020). Thus, the flocculation model allows for precision in the determination of the flocculation parameters. Then, the flocculation curves are determined using the thermodynamic model. Figure 7 illustrates, for this example, the quantity of flocculated asphaltene W as a function of the pressure P in MPa for the three temperatures 75°C, 100°C and 125°C.In this figure Calc indicates calculated using the method according to the invention, and Wmax-ML indicates the maximum amount of flocculated asphaltene determined by the flocculation model of the method according to the invention. The Wmax points calculated by the flocculation model, which are not used in the calibration, are also in very good agreement with the flocculation curves. It can be noted that there is no asphaltene precipitation data in the article (Sullivan et al., 2020) to experimentally validate the shape of the entire flocculation curve. The results presented in this example therefore make it possible to demonstrate the performance of the method according to the invention for determining the flocculation curve of an asphaltenic fluid in a scenario where experimental data are absent.

Claims

Claims 1) Method for determining asphaltene flocculation conditions of a hydrocarbon fluid, characterized in that the following steps are implemented: a) A flocculation model (MOD FLO) is applied to said hydrocarbon fluid (FLU) to determine at least one modeled flocculation parameter, said flocculation model (MOD FLO) linking the composition of a fluid to at least one flocculation parameter; b) A compositional model (MOD COM) is constructed of said hydrocarbon fluid, said compositional model (MOD COM) decomposing said hydrocarbon fluid into several constituents, and at least one physical property of said hydrocarbon fluid is determined by means of said compositional model; c) A thermodynamic model (MOD THE) is applied to said at least one physical property determined by means of said compositional model (MOD COM) to determine at least one flocculation parameter;d) said compositional model of said hydrocarbon fluid is calibrated (CAL) by minimizing a difference between at least one flocculation parameter obtained from said thermodynamic model (MOD THE) and said at least one modeled flocculation parameter obtained from said flocculation model (MOD FLO); and e) the asphaltene flocculation conditions (CDF) of said hydrocarbon fluid (FLU) are determined from said calibrated compositional model of said hydrocarbon fluid. 2) Method according to claim 1, wherein the method comprises a prior step of constructing said flocculation model by machine learning, in which the following steps are implemented: i) a learning base of learning fluids (BAP) is constructed, each learning fluid being characterized by a composition and at least one flocculation parameter, preferably said at least one flocculation parameter of said learning fluids being obtained experimentally;and ii) a flocculation model is constructed using a machine learning method (APP) trained on said learning base, said flocculation model (MOD FLO) linking the detailed composition of a fluid to said at least one flocculation parameter. 3) Method according to one of the preceding claims, in which each flocculation parameter is chosen from a low threshold pressure, a high threshold pressure, a saturation pressure, a maximum quantity of flocculated asphaltene.; 4) Method according to one of the preceding claims, in which said compositional model (MOD COM) is based on a representation in N constituents, N being between 6 and 15, preferably between 6 and 12, and preferably being 10. 5) Method according to one of the preceding claims, in which said compositional model is calibrated (CAL) by means of an iterative method. 6) Method according to one of the preceding claims, in which said thermodynamic model (MOD THE) is an algorithm for calculating the three-phase liquid-liquid-vapor equilibrium of a fluid at a given temperature and at a given pressure. 7) Method according to one of the preceding claims, in which said flocculation conditions (CDF) correspond to a flocculation curve which relates the quantity of flocculated asphaltene to the pressure of the fluid.8) Method according to one of the preceding claims, in which the at least one physical property is chosen from the molar mass, the critical temperature, the critical pressure, the acentric factor of the constituents of the fluid.9) Method for determining or predicting asphaltene flocculation of a hydrocarbon fluid within an underground formation or within a fluid transport pipe or within a conversion reactor, in which the following steps are implemented: a) Asphaltene flocculation conditions (CDF) of said hydrocarbon fluid are determined by means of the method for determining asphaltene flocculation conditions according to one of the preceding claims; b) A pressure and a temperature of said hydrocarbon fluid (COP) within said underground formation or within said transport pipe or within a conversion reactor are determined by measurement or predicted by simulation; and c) A possible asphaltene flocculation (FDA) of said hydrocarbon fluid is determined or predicted (DET) as a function of said determined flocculation conditions (CDF) and the determined or predicted pressure (COP) and the determined or predicted temperature.10) Method for exploiting a hydrocarbon fluid within an underground formation or within a fluid transport pipe or within a conversion reactor, in which the following steps are implemented: a) Asphaltene flocculation (FDA) of said hydrocarbon fluid is determined or predicted by means of the method for determining or predicting flocculation according to claim 9; and b) Said hydrocarbon fluid is exploited (EXP) as a function of the possible asphaltene flocculation (FDA) of said hydrocarbon fluid, in particular by modifying the temperature. and / or the pressure and / or the composition of the hydrocarbon fluid in the event of asphaltene flocculation and / or by addition of solvent.

Images