PREDICTIVE DETERMINATION OF EXCESS CHEMICAL POTENTIAL

Description

[0001] The invention relates to the field of predictive determination of the physicochemical properties of solutions.

[0002] The predictive determination of the physicochemical properties of a solution is of great industrial interest. Indeed, since an industrial installation is required to process large volumes, it is important to have the best possible control over the various parameters of a manufacturing process implementing a solution, in order to ensure that the installation is properly matched to the process calibration. For example, some installations will not be able to handle excessively large volumes, others will only be able to handle liquids and it will therefore be necessary to ensure that no boiling occurs, and still others will be able to handle gases but only within a given pressure range beyond which safety problems will arise.Similarly, it may be interesting to know precisely the quantity of heat energy to be supplied to a product for cooking or processing in order to avoid other unwanted reactions and overconsumption of energy.

[0003] Determining parameters and calibrating industrial installations is therefore of major importance. Although it is sometimes possible to carry out this calibration by trial and error, this method is not always satisfactory because it does not ensure that optimal conditions are in place. Therefore, the use of predictive determination methods, for example implemented by simulation software, is particularly interesting. These predictive determination methods make it possible to model the activity coefficients or chemical potentials of molecules.

[0004] In this respect, it is known from documents WO2015175387 and WO2012051242 to model the activity coefficient of a molecule in a solution of interest by approaching each molecule using an empirical algorithm for averaging the charge density profile (or molecular histogram of charge densities), similar to the method described in the article Klamt et al., J. Phys. Chem. A, 1998, Volume 102, No. 26, pages 5074-5085, this method being known as the “sigma-averaging method”. These models use a purely physical and combinatorial approach, which can be satisfactory in the case of certain simple and dilute aqueous solutions but which gives results very far from reality when one seeks to model concentrated solutions within which chemical equilibria are in place, in particular equilibria of chemical associations (complexation, solvation, hydration, etc.).) deduced from the properties of the bonds between the molecules composing the solution. Indeed, these models do not allow us to account for the chemical interactions between the different molecules. This is particularly the case for concentrated aqueous solutions of sugars and / or polyols and / or carbohydrates.

[0005] However, in an industrial context, the components of interest within a solution are generally the solutes and not the solvent and the purely physical and combinatorial approach is rarely adapted to these concentrated aqueous solutions.

[0006] There is therefore a need for a method for predictive determination of physicochemical equilibrium quantities in an industrial context, particularly in the case of concentrated aqueous solutions or those with a very high dry matter content. Indeed, attempts to model aqueous solutions have already been made. To this end, Catté et al. (Fluid Phase Equilibria 105, 1-25., 1995) combined the UNIFAC physical model with a chemical model of fixed hydration of sugars by water in order to model rather dilute aqueous solutions. However, the Applicant was able to observe that this type of model does not work for concentrated aqueous solutions.

[0007] The invention achieves this by means of a method for predictively determining an excess chemical potential of an apparent species comprising the steps of claim 1.

[0008] A true species is a chemical entity present in solution and not always isolable. For example, in the case of an aqueous solution of sorbitol, we will distinguish isolated sorbitol, sorbitol associated with one water molecule, sorbitol associated with two water molecules and sorbitol associated with three water molecules as true species.

[0009] Apparent species are those species in solution that can be isolated and whose properties can be easily measured. In the case of an aqueous solution of sorbitol, for example, this would be water on the one hand and sorbitol on the other.

[0010] The concepts of true species and apparent species are known in themselves. For example, Prausnitz et al. have already described them in the reference work Molecular thermodynamics of fluid-phase equilibria, Prentice-Hall international series in the physical and chemical engineering sciences, Prentice-Hall PTR., 1999, page 353. They also appear in various publications such as those of Toure et al. (The Canadian Journal Of Chemical Engineering 93, 2015, page 445 as well as Fluid Phase Equilibria, 2016, 424, page 92). Achard et al., (AIChE journal, 40(7), 1994. 1210-1222) and the publication of Catté et al. previously cited also refer to the respective properties of the true species when they speak of "true concentration or mole fraction", and of the apparent species when they speak of "apparent concentration or mole fraction".Alternatively, a true species may also be called a molecular entity and an apparent species may be called a chemical species (or set of molecular entities) as defined by IUPAC.

[0011] The excess chemical potential is determined from a partial molar property, leading after comparison between different solutions to the chemical potentials.

[0012] Preferably, the method for predictively determining an excess chemical potential according to the invention further comprises a step consisting of carrying out a tiling so as to approximate each true species determined in step a by a combination of Platonic solids so that it is associated with a shape having an exterior surface consisting of a combination of planar surface segments of the same surface.

[0013] The determination of an excess chemical potential, also called excess free enthalpy, is particularly interesting because it allows the determination of most of the physicochemical properties of the solution. Indeed, the excess chemical potential of the solution helps to account for the availability of molecules.

[0014] By chemical species, we do not mean to be limited to a given molecule. The same molecule can correspond to several different chemical species depending on whether or not it interacts with one or more other molecules. Thanks to the fact that the typology is done by chemical species, and not by molecule, the step of carrying out a typology of the chemical species present makes it possible to account for the interactions within the solution. During this step, we can also take into account the probability of the presence of a given species, in order to account for various physicochemical equilibria. In other words, the "molecule" is an "apparent species" present in the mixture and the "chemical species" is a "true species" actually present in the mixture. For example, the molecule could be the sorbitol and the chemical species corresponding would be the different hydration states of sorbitol.

[0015] After determining the typology of the different true species present, it is necessary to first generate a certain amount of input data. This input data is extracted from a quantum computation result, for example in the form of a COSMO file for each chemical species. An example of visualizing the contents of a COSMO file is illustrated in the Figure 3a where the color is a representation of the value of the content of the physical quantity vector associated with a true species. As a reminder, a COSMO file includes, for each of the segments forming the external surface of the chemical species, the electrostatic potential, the surface, the charge and the surface charge density.

[0016] Preferably, and in particular when the vector of physical quantities comprises surface charge densities, each identified chemical species is then approximated by a combination of identical Platonic solids. A Platonic solid is a regular, convex polyhedron having identical external surfaces on each of its faces, which allows it to seamlessly tile a three-dimensional space.

[0017] The use of identical Platonic solids is then a major innovation. Indeed, at the output of quantum calculations, the segments forming the external surface of a molecule have different sizes; the usual methods of modeling activity coefficients (in particular those of the models mentioned above) consist of empirically averaging the surface charge densities of said segments by an identical flat surface (generally square or circular) of suitable size. By using identical Platonic solids, the entire space representing the volume of the solution is compactly tiled and each molecule has an external surface consisting of a combination of identical surfaces, real surface segments, which makes it possible to compare molecules with each other and to model the interactions between different segments in order to be able to coherently combine a physical model with a chemical model.This makes it possible in particular to model the physicochemical properties of the above-mentioned concentrated aqueous solutions.

[0018] For this purpose, at least one vector of physical quantities is associated with each determined true species. If a Platonic solid tiling has been used, one vector of physical quantities is preferably generated per surface segment. The term vector is understood here in its algebraic sense: it is a table grouping together a certain number of physical quantities relative to the surface segment with which it is associated.

[0019] When each vector is attached to a surface segment of the same dimension, each vector has the same weight.

[0020] The set of vectors therefore constitutes a good description of the solution from which it is possible to predict a chemical potential of excess of an apparent species, which is particularly useful in the context of an industrial installation because it makes it possible to check the adequacy of the predictive method with reality by means of a measurement.

[0021] The determination of the respective excess chemical potentials of the true species is done using a physical vision of the interactions between molecules and / or between chemical species. A judicious choice of the typology of the true species thus makes it possible to coherently associate with this physical vision of the interactions a chemical model of chemical associations, and therefore to carry out a recomposition of the solution into equivalent species to construct the prediction of the respective excess chemical potentials of the apparent species.

[0022] The solution is preferably an aqueous solution of sugars and / or polyols and / or carbohydrates. Indeed, the model according to the invention is particularly suitable for this type of solution which can be very concentrated and present many different chemical species, in particular due to the many water molecules which can be adsorbed by the molecules.

[0023] Carbohydrates include sugars and polyols. The terms “sugars,” “polyols,” and “carbohydrates” are well known to those skilled in the art. Polyols are also known as “alditols.” For each of these terms, those skilled in the art may refer to the International Union of Pure and Applied Chemistry (IUPAC) definitions found in Moss et al., Pure Applied Chemistry, 1995, 67, 1307, Glossary of class names of organic compounds and reactivity intermediates based on structure (IUPAC Recommendations 1995).

[0024] Carbohydrate solutions can be obtained from starch, for example. Aqueous sugar solutions can be maltodextrin solutions, glucose syrups, or fructose syrups.

[0025] Examples of sugars include glucose, arabinose, xylose, fructose, psicose (also called allulose), mannose, ribose, galactose, trehalose, cellobiose, gentiobiose, isomaltose, isomaltulose, kojibiose, laminaribiose, maltose, galactose, lactose, maltulose, nigerose, sucrose, and sophorose.

[0026] Carbohydrates other than the above-mentioned sugars may be saccharides with a degree of polymerization greater than or equal to 3. They may be oligosaccharides with a degree of polymerization greater than or equal to 3, in particular oligosaccharides having a degree of polymerization ranging from 3 to 20 and in particular oligosaccharides such as maltotriose, isomaltotriose, panose, raffinose, maltotetratose and cyclodextrins. In general, polyols may be any of the above-mentioned hydrogenated saccharides. They may, for example, be polyglucitol, i.e. a solution of glucose syrup or hydrogenated maltodextrin. Polyols may be polyols comprising from 3 to 24 carbon atoms such as glycerol, erythritol, threitol, arabitol, xylitol, ribitol, mannitol, sorbitol, galactitol, fucitol, iditol, inositol, volemitol, isomalt, maltitol, lactitol, maltotriititol and matotetraitol.Preferably, the polyols are chosen from sorbitol, mannitol, xylitol and maltitol.

[0027] The combination of Platonic solids is, for example, a combination of icosahedrons. The icosahedron has twenty triangular faces. It is the type of Platonic solid with the largest number of faces and therefore allows for a tiling that best approximates each chemical species present. Alternatively, the Platonic solid is a dodecahedron made up of twelve faces of identical regular pentagons.

[0028] Preferably, at least one of said vectors of physical quantities contains a physical quantity selected from charge density, propensity to form a hydrogen bond as a hydrogen atom donor, propensity to form a hydrogen bond as a hydrogen atom acceptor, and electrostatic potential. Instead of charge density, it is of course also possible to use charge and / or surface area as the physical quantity.

[0029] These physical quantities make it possible to best describe the interactions in the solution and to derive its properties, and in particular the various relevant chemical potentials, in a precise manner. Depending on the physical quantities used, the equations used to derive the chemical potential will not be the same.

[0030] Depending on the type of solution used, the interactions between chemical species may be different and it may therefore be interesting to use other, more suitable physical quantities.

[0031] The typology of the chemical species present distinguishes, for example, different complexation states of the same molecule.

[0032] Preferably, the typology of the chemical species present distinguishes different solvation states, in particular hydration, of the same molecule. It is understood that solvation is a particular mode of complexation by the solvent, hydration being a solvation by water. This typology of chemical species thus makes it possible to associate with the aforementioned physical vision of the determination of the excess chemical potentials of the chemical species, a contribution of chemical associations deduced from the properties of bonds between the molecules composing the solution. As a result, the chemical species actually present in the mixture have different physical properties from the individual molecules forming the apparent mixture. Consequently, a recomposition of the solution into equivalent species makes it possible to construct the prediction of the partial molar properties leading to the determination of the excess chemical potential of the molecules (or apparent species).

[0033] The method for predictive determination of an excess chemical potential according to the invention may further comprise a step c' consisting of producing a molecular histogram of at least one of said physical quantities.

[0034] Such a histogram allows for statistical processing of the solution. A sigma-profile is an example of such a histogram, applied to the charge density profile of a solution.

[0035] The method for predictively determining an excess chemical potential according to the invention may further comprise a step consisting of determining a quantity deduced from the excess chemical potential of the molecules (or apparent species). Knowledge of the excess chemical potential makes it possible to translate the deviation from the ideality of the solutions and to deduce all the thermodynamic properties (water activity aw, activity coefficients γ i of the constituents, osmotic coefficient) and equilibria between phases (liquid-liquid, liquid-vapor and liquid-solid equilibria), from the general relations of thermodynamics. For example, the liquid-vapor equilibrium is characterized by the notion of boiling temperature of the mixture. Preferably, this quantity deduced from the chemical potential is chosen from an activity coefficient of a chemical species or molecule present, a boiling temperature of the solution, and a solubility of a chemical species or molecule present.

[0036] The excess chemical potential of the solution makes it possible to find a large number of quantities of interest in the solution, which can be used for the purpose of this or that installation controlled by the method according to the invention.

[0037] The solution preferably has a solvent mass content of less than 30% relative to the total mass of the solution, advantageously less than 20%.

[0038] Indeed, although the predictive determination method according to the invention is suitable for any type of solution, the precision of the results obtained is particularly interesting in the field of concentrated solutions for which there is no simple alternative.

[0039] The chosen Platonic solids have, for example, a characteristic dimension between 0.0005 Å and 100 Å, preferably between 0.0005 Å and 6 Å. The characteristic dimension of a Platonic solid corresponds to the size of its edge. A characteristic dimension of this order allows a sufficiently precise rendering of the properties of the solution.

[0040] The invention also relates to the use of a method according to the invention to calibrate an industrial installation. Any type of industrial installation requires calibration. Given the volumes involved in an industrial installation, carrying out this calibration predictively, without having to sacrifice part of the production for this purpose, is particularly advantageous.

[0041] The industrial installation is preferably an installation for the production of sugars and / or polyols and / or carbohydrates. Indeed, as indicated above, the method according to the invention is particularly suitable for solutions of sugars and / or polyols and / or carbohydrates more generally.

[0042] The invention also relates to a computer system comprising a processor, said processor being configured to implement a method according to the invention when it receives specific instructions.

[0043] The invention also relates to a computer program product comprising a code configured to carry out a method according to the invention when it is executed by a processor or an electronic control unit.

[0044] The invention may be better understood with the aid of the non-limiting exemplary embodiments described below, and by examining the attached drawing in which: there figure 1 is a representation of the three-dimensional structure of sorbitol and its different hydration states, the figure 2 is a diagram representing the overall structure and data necessary for the implementation of a method according to the invention, the figure 3 represents different steps in the generation of a Platonic solid tiling of a molecule and the corresponding charge density histograms, the figure 4represents a comparison of the molecular histograms of the charge densities (σ-profile), for the representative case of the water molecule, obtained using the aforementioned method of Klamt et al. and the method of the invention, the Figure 5 represents a diagram schematizing a resolution method making it possible to calculate the different excess chemical potentials of a solution by implementing the method according to the invention, the figure 6 is a comparison of the water activity data obtained by implementing the method according to the invention, using a method according to the prior art and calculating them from the boiling temperatures at atmospheric pressure obtained by experimental measurement, the figure 7is a comparison of boiling temperature data obtained using the thermodynamic model developed in this invention, those obtained using state-of-the-art methods and those obtained by experimental measurement. figure 8 is a schematic representation of a wiped film thin film evaporator.

[0045] The following examples will allow a better understanding of the present invention, without limiting its scope.

[0046] In order to carry out a typology of the different chemical species present, it is necessary to determine the different physicochemical equilibria likely to occur within the solution.

[0047] In the case of this non-limiting example, we are interested in the case of sorbitol. Sorbitol is capable of adsorbing 0, 1, 2 or 3 water molecules depending on the quantity of water available. Five chemical species are therefore likely to be present; these are the species illustrated in figure 1 .

[0048] After determining the typology of the different true species present, it is necessary to first generate a certain amount of input data. This input data is extracted from a quantum computation result, for example in the form of a COSMO file for each chemical species. As a reminder, a COSMO file includes, for each of the segments forming the external surface of the chemical species, the surface, the charge, the charge density and the electrostatic potential. An example of visualization of the contents of a COSMO file is illustrated in the Figure 3awhere color is a representation of the value of the content of the physical quantity vector associated with a true species.

[0049] These COSMO files are part of the input data of the physical model as illustrated in figure 2. In addition, operational data (temperature, pressure, composition of the different molecules or apparent species) and universal parameters of the physical model are also used to determine the chemical potentials of the true species by using in the chemical part the equilibrium constants of chemical associations between the molecules (solvation) in the chemical model. A recomposition of the solution into equivalent species thus makes it possible to determine the excess chemical potentials of the molecules (or apparent species) from the knowledge of the true compositions and the respective excess chemical potentials of the chemical species actually present in the solution.To do this, the molecular surface of each chemical species is first paved by a combination of Platonic solids so that it is associated with a shape having an exterior surface made up of a combination of planar surface segments of the same surface, and a vector of physical quantities is associated with each surface segment.

[0050] A Platonic solid is a regular, convex polyhedron having identical external surfaces on each of its faces.

[0051] Preferably, the Platonic solid is a dodecahedron made up of twelve identical regular pentagon faces, most preferably an icosahedron made up of twenty identical equilateral triangular faces ( Figure 3b ). Since these solids are closer to the surface of a sphere, the tiling of the surface is closer to the surface value resulting from the quantum calculation and contained in the COSMO file ( Figure 3a ).

[0052] Preferably, the value of the edge size (lp) of these Platonic solids is less than or equal to 6 Å. For each chemical species, the molecular sigma-profile, i.e. a histogram giving the charge density profile, 310 (p(σ)) is generated by using a tiling of the surface by an integer number of Platonic solids and by assigning to each face, or surface segment, a value that is stored in a vector.

[0053] This generation makes it possible to avoid using a sigma-averaging method and therefore to remain as close as possible to quantum simulations.

[0054] There Figure 3c illustrates the obtained sigma profile 310.

[0055] For each surface segment, that is to say each face of Platonic solid, corresponding to a part of chemical species likely to react as a donor (σd) of hydrogen bond (HB) of a molecule, the donor sigma-profiles 320 (pHB(σd)) are generated, these sigma-profiles being generated by using a tiling of their surface by an integer number of Platonic solids, this tiling being preferentially identical to the previous one.

[0056] Similarly, for each hydrogen bond (HB) acceptor segment (σa) of a molecule, the 330 acceptor sigma-profiles (pHB(σa)) are generated, these sigma-profiles being generated using a tiling of their surface by an integer number of Platonic solids, this tiling being preferentially identical to the previous one.

[0057] Each surface segment is therefore assigned three values ​​in a vector corresponding to each of the profiles 310, 320, and 330.

[0058] The said Platonic solids used in the different tilings have identical edges and are therefore Platonic solids of identical surface area and volume.

[0059] As shown in the Figure 4 , the sigma-profile 41 thus obtained is different from that 42 obtained by the methods of the prior art using sigma-averaging. The figure 4 represents the molecular sigma-profile as a function of the surface charge density in electrons per square Angstrom.

[0060] The sigma profile of the mixture is the mole fraction-weighted sum of the sigma profiles of each molecule.

[0061] In the following, the sigma-profile is therefore used as a descriptor of the tiled chemical species.

[0062] In the physical part of the method according to the invention, as illustrated in the figure 2, after the generation of sigma-profiles of chemical species, the step of obtaining the excess chemical potential of a molecule in a mixture involves obtaining three different contributions.

[0063] The combinatorial contribution takes into account the differences in size and shape between the molecules present in the mixture and accounts for the probability that each face meets another.

[0064] The electrostatic contribution, called the “misfit contribution”, is generated in such a way as to take into account the fact that the surfaces in electrostatic interaction are faces of the Platonic solid.

[0065] The so-called "HB" contribution is obtained by considering that it results solely from the interactions between the donor parts and the hydrogen bond (HB) acceptor parts contained respectively in the acceptor and donor sigma-profiles of the mixture. Combinatorial contribution

[0066] To account for size and shape differences between molecules in the mixture, the combinatorial contribution of the excess chemical potential is calculated using the following formula: μ i E , combi = RT . ln γ i Combi ; Or ln γ i Combi = ln Φ i x i + 1 − Φ i x i + q i r i ⋅ ln θ i Φ i + q i r i ⋅ Φ i x i − θ i x i

[0067] This formula is innovative in the state of the art. It is an adaptation of the Staverman Guggenheim (SG) type formula and allows for the inclusion of an entropic contribution that accounts for differences in the external volumes and surfaces of the molecules. Φ i = x i ⋅ r i ∑ j = 1 n oral x j ⋅ r j θ i = x i ⋅ q i ∑ j = 1 n oral x j ⋅ q j q i = A i s icosa è dre r i = V i v icosa è dre Misfit Contribution

[0068] The misfit contribution of the excess chemical potential results from the following formula: μ i E , misfit = μ i misfit , S − μ i misfit , ref

[0069] Or μ i misfit , S denotes the misfit contribution of the chemical potential of the molecule in the solution and μ i misfit , ref its chemical potential in a reference state chosen by the user. This reference state can be chosen from the pure body reference state for each of the constituents or from the infinite dilution reference state in the majority solvent (e.g. water) for each of the constituents, except for the majority solvent which is in a pure body reference state. μ i misfit , S is the sum, weighted by the number of segments represented in the sigma-profile pi ( σ ), activity coefficients of the surface segments constituting the molecule: μ i misfit , S = ∫ p i σ ⋅ μ misfit , S σ dσ

[0070] The chemical potential of the surface segment µ misfit,S< (σ) depends on the electrostatic interaction energies between all surface segments. μ misfit , S σ = − k ⋅ T ⋅ ln ∫ p s σ ′ ¯ ⋅ exp μ misfit , S σ ′ − E misfit σ , σ ′ k ⋅ T dσ ′

[0071] This interaction energy can be calculated by the following formula: E misfit σ , σ ′ = α ′ ⋅ e 0 ⋅ 10 20 σ + σ ′ 2 4 ⋅ ε 0 ⋅ s p ⋅ 10 − 10 3 π

[0072] Where α' denotes one of the universal parameters of the physical model, as illustrated in the figure 2 . HB Contribution

[0073] The HB contribution of the excess chemical potential is calculated using the following formula: μ i E , HB = μ i HB , S − μ i HB , ref μ i HB , S being the contribution HB of the chemical potential of the molecule in the solution and μ i HB , ref being its chemical potential in a reference state chosen by the user. μ i HB , S is the sum of the donor and acceptor contributions. μ i HB , S = ∫ p i HB σ d ⋅ μ HB , S σ d dσ d + ∫ p i HB σ a ⋅ μ HB , S σ a dσ a

[0074] The first term of this summation, being the HB donor contribution of the chemical potential, is calculated as the sum, weighted by the number of HB acceptor segments represented in the HB acceptor sigma-profile, of the chemical potentials of the HB donor surface segments constituting the molecule.

[0075] The chemical potential of the donor surface segment HB depends on the HB interaction energies between all donor surface segments and acceptor surface segments. μ HB , S σ d = − k ⋅ T ⋅ ln ∫ p HB , S σ a ¯ ⋅ exp μ HB , S σ a − E HB σ a σ d k ⋅ T dσ a

[0076] This interaction energy can be calculated by the following formula: E HB σ d σ a = c HB ⋅ e 0 2 ⋅ σ d ⋅ σ a 4 ⋅ π ⋅ ε 0 ⋅ S p 2 r da ⋅ 10 − 10 r da is the minimum approach distance between the HB bond donor and acceptor. c HB and r da denote universal parameters of the physical model, as illustrated in figure 2 .

[0077] The second term of this summation, being the HB acceptor contribution of the chemical potential, is calculated as the sum, weighted by the number of HB donor segments represented in the HB donor sigma-profile, of the chemical potentials of the HB acceptor surface segments constituting the molecule.

[0078] The chemical potential of the acceptor surface segment HB also depends on the HB interaction energies between all donor surface segments and acceptor surface segments. μ HB , S σ a = − k ⋅ T ⋅ ln ∫ p HB , S σ d ¯ ⋅ exp μ HB , S σ d − E HB σ a σ d k ⋅ T dσ d

[0079] The two contributions HB donor and HB acceptor are interdependent and are therefore calculated simultaneously in an iterative manner.

[0080] This formulation of the "hydrogen bond" contribution is entirely new. This contribution is, for example, called the "first chemical contribution."

[0081] The resulting model simultaneously takes into account this first chemical contribution, the combinatorial contribution and the electrostatic contribution.

[0082] When a carbohydrate such as sorbitol is dissolved in water, the carbohydrate hydrates one or more times, forming hydrated species ie new chemical species containing a carbohydrate molecule strongly bound to one or more water molecules by hydrogen bonds, particularly in the first hydration sphere. The phenomenon is illustrated in figure 1where the typology of chemical species present in a solution of sorbitol and water (apparent species) is carried out. The true species thus determined are water, anhydrous sorbitol, sorbitol hydrated once, twice and three times. In addition, the colored surface surrounding each true species makes it possible to illustrate in a pictorial manner the content of the vector of physical quantities associated with said true species.

[0083] A chemical model is therefore necessary to take into account the contribution of chemical associations (solvation, complexation, etc.) deduced from the properties of bonds between the molecules composing the solution such as the chemical interactions between sorbitol and the solvent, in particular the successive hydration reactions of sorbitol describing the chemical formation of the hydrated forms of the carbohydrate. This contribution is for example called the "second chemical contribution".

[0084] These reactions, and therefore the composition of the different true anhydrous or hydrated species present in the mixture, are characterized by hydration equilibrium constants, themselves dependent on the chemical potential of each of the true species present in the mixture.

[0085] A method for resolving physicochemical equilibria is therefore necessary to determine the respective excess chemical potentials of the true species (free solvent, anhydrous species and hydrated species). Such a resolution method is illustrated in Figure 5 .

[0086] To this end, the generation of the COSMO file made in the first step of the process must be carried out for each chemical species, in the sense of each true species, present in the mixture in order to make all the data generated consistent.

[0087] This method, which is similar to a chemical model, makes it possible to determine the equilibrium properties of the mixture using the resolution method indicated on the Figure 5 to determine the excess chemical potentials of the true species present in the mixture and to deduce all the physicochemical properties deduced from the knowledge of this physicochemical reality of the interactions taking place in the mixture.

[0088] For this purpose, as illustrated in the figure 2 , the user simply needs to indicate the operating conditions (or data) such as the apparent composition, temperature and / or pressure to use the thermodynamic model developed in this invention in order to predict the chemical potentials of the chemical species, therefore the composition of the true species, present in the real mixture and the equilibrium properties of said mixture.

[0089] A comparison is made between the water activity values ​​in a concentrated sorbitol solution obtained by a prior art method (in this case according to the UNIFAC group contribution model) and the method of the invention. Curves 61 (prior art) and 62 (invention) of the method are obtained respectively. figure 6 which represents the water activity as a function of the mass fraction of sorbitol in the solution. Experimental data are superimposed on it and it can be seen that the invention makes it possible to obtain data much closer to the measured parameters.

[0090] A comparison is made between the boiling temperature values ​​at atmospheric temperature in a concentrated sorbitol solution obtained by a method of the prior art (in this case according to the UNIFAC group contribution model) and the method of the invention. Curves 71 (prior art) and 72 (invention) of the method are obtained respectively. figure 7which represents the boiling temperature in °C as a function of the mass fraction of sorbitol in the solution. Experimental data are superimposed on it and it can be seen that the invention makes it possible to obtain data much closer to the measured parameters.

[0091] In the design of final products such as sorbitols marketed as concentrated liquid solutions, knowledge of the control parameters of equipment such as an evaporator is essential. The same is true for sorbitol powders, which are generally manufactured using molten sorbitol solution, i.e. having a very high dry matter content, which can exceed 90% or even 99%.

[0092] Evaporation is the process in which dilute solutions are concentrated by converting the liquid solvent (e.g., water) into a gas. It is one of the most energy-intensive processes in the chemical and food industries, leading manufacturers of this type of equipment to offer a wide range of technologies and systems to adapt to product characteristics, dry matter constraints, and energy costs. It is therefore particularly advantageous to be able to optimize process costs or simulation. Example: Evaluation of wiped film evaporation technology for the production of continuous molten sorbitol from aqueous sorbitol solution

[0093] The operation of a scraped film thin film evaporator as shown in Figure 8is simulated. Stream A represents the flow of solution to be evaporated (aqueous sorbitol solution). The heating steam is introduced into the double jacket of the evaporator by Stream D. During the evaporation step, the water evaporated from the sorbitol solution is extracted from the evaporator by Stream B. The heating steam as well as the condensates of the heating steam are extracted from the double jacket by Stream E. The concentrated solution (molten sorbitol) is thus extracted from the evaporator by Stream C.

[0094] A wiped film thin film evaporator has a high overall heat exchange coefficient, which makes it possible to evaporate very concentrated solutions, which may have a high viscosity, very efficiently.

[0095] To carry out this simulation in which the apparent species are water and sorbitol, we first carry out the typology of the true species present as illustrated in the Figure 1(water, anhydrous sorbitol, single-hydrated sorbitol, double-hydrated sorbitol, and triple-hydrated sorbitol).

[0096] Quantum calculations are carried out for each of these true species using the TURBOMOLE software which allows the generation of COSMO files associated with each true species.

[0097] From these COSMO files are extracted the surface charge densities used in the physical quantities vector.

[0098] The outer surface of each true species is tiling using icosahedral Platonic solids with a characteristic dimension of 1.12 Å.

[0099] As illustrated in the Figure 3c for water, for all surface segments composing the true species, the sigma-molecular 310 (p(σ)) are generated and these are stored as values ​​in the physical quantity vector of the corresponding true species.

[0100] Similarly, for all surface segments composing the true species, the hydrogen bond (HB) donor sigma-profiles 320 (pHB(σd)) are generated and these are stored as values ​​in the physical quantity vector of the corresponding true species.

[0101] Similarly, for all surface segments composing the true species, the hydrogen bond (HB) acceptor sigma-profiles 330 (pHB(σa)) are generated and these are stored as values ​​in the physical quantity vector of the corresponding true species.

[0102] The method of resolving physicochemical equilibria illustrated in Figure 5 was implemented in the ProSimPlus software (marketed by the company Prosim) to determine the respective excess chemical potentials of the true species (free solvent, anhydrous species and hydrated species).

[0103] To simulate the evaporation process, the following operating conditions are also entered into the ProSimPlus software: Apparent species composition of the solution to be evaporated: 70% sorbitol and 30% water; Inlet temperature of the solution to be evaporated: see Table 1; Inlet flow rate of the solution to be evaporated: see Table 1; Absolute pressure in the evaporator: 50 mbar; Apparent species composition of the concentrated solution (molten sorbitol): see Table 1; Exchange surface of the evaporator: see Table 1; Overall exchange coefficient of the evaporator: 1000 W / m2 / °C.

[0104] The thermodynamic model developed in this invention is used under these operating conditions to iteratively predict using ProSimPlus the excess chemical potentials of the apparent species, after having predicted the excess chemical potentials of the true species present in the real mixture and the equilibrium properties of said mixture.

[0105] This is how the boiling temperature and the outlet flow rate of the mixture at the evaporator outlet are calculated (see Table 1). Table 1 1 2 3 4 5 6 7 8 Ladder Pilot Pilot Industrial Industrial Industrial Industrial Industrial Industrial Simulation input data Exchange surface m 2< 0,14 0,14 3,00 6,00 6,00 6,00 6,00 6,00 Sorbitol solution feed rate kg / h 25 30 915 1500 1500 1800 1800 1800 Power supply temperature (°C) 40 40 40 40 60 60 60 60 Solids of molten sorbitol % 99,25 99,25 99,25 99,25 99,25 99,25 95,00 99,50 Simulation results Heating steam temperature (°C) 139,5 146,7 166,9 154,4 154,4 167,8 121,1 173,5 Heating steam pressure (bar) 3,555 4,347 7,327 5,341 5,341 7,490 2,054 8,595 Temperature of the concentrated product (°C) 123,7 123,7 123,7 123,7 123,7 123,7 73,0 130 Heating steam flow rate Kg / h 10,0 12,8 390,0 644,9 644,9 732,9 530,8 755,7 Power exchanged kW 6,37 7,64 233,04 382,03 356,33 427,59 326,13 437,93 DTLM (mean logarithmic temperature difference) of the evaporator (°C) 45,5 54,6 77,7 63,7 59,4 71,3 54,4 73,0 Quantity of water evaporated kg / h 7,37 8,84 269,66 442,07 442,07 530,48 473,68 533,67 Melted sorbitol output flow rate kg / h 17,64 21,16 645,34 1057,93 1057,93 1269,52 1326,32 1266,33 Conclusion of the Example

[0106] Among the control parameters determined by simulation using the method of the invention, the boiling delay, i.e. the difference between the boiling temperature of a solution containing sorbitol and that of the pure solvent (water) is one of the most critical data.

[0107] In addition, the regulation of the heating steam (flow rate, pressure, temperature) is important: it determines the regularity and stability of the evaporator.

[0108] The use of the invention for dimensioning (designing and scaling up) is thus illustrated in Table 1 and calibrating an industrial installation using scraped film evaporation technology.

[0109] The simulations carried out in Table 1 thus make it possible to size and control a scraped film thin-film evaporator according to the needs and utilities (steam, feed product, etc.) by indicating the optimal parameters to approach the operating point as quickly as possible.

[0110] These simulations make it possible to limit the number of attempts scale-up and start-up of an industrial installation.

[0111] Unless otherwise specified, the word "or" is equivalent to and / or. Similarly, the word "a" is equivalent to "at least one" unless otherwise specified.

Claims

1. A method for the predictive determination of an excess chemical potential of an apparent species, the method being carried out by a computer and being characterized in that it involves the following steps: a. performing a typology of the chemical species which are present in a solution, in order to determine at least one true species, and generating, for each true species determined, a paving of the molecular surface of each chemical species through a combination of Platonic solids, so that each true species be associated with a form having an outer surface consisting of a combination of flat surface segments of the same area b. associating with each surface segment obtained for a true species a vector of physical magnitudes, c. determining, from the vectors of physical magnitudes obtained, a group of contributions relative to the considered true species and, from the calculated contributions for the true species present in the solution, determining an excess chemical potential of an apparent species.

2. The method for the predictive determination of an excess chemical potential as claimed in claim 1, wherein the solution is an aqueous solution of sugars and / or of polyols and / or of carbohydrates.

3. The method for the predictive determination of an excess chemical potential as claimed in claim 1, wherein the combination of Platonic solids is a combination of icosahedra.

4. The method for the predictive determination of an excess chemical potential as claimed in any of claims 1 to 3, wherein at least one of said physical magnitude vectors contains a physical magnitude chosen from the charge density, the tendency to form hydrogen bonding as a hydrogen atom donor, the tendency to form hydrogen bonding as a hydrogen atom acceptor, and the electrostatic potential.

5. The method for the predictive determination of an excess chemical potential as claimed in any of claims 1 to 4, wherein the typology of the chemical species present enables distinction between various complexation states of the same molecule.

6. The method for the predictive determination of an excess chemical potential as claimed in any of claims 1 to 5, wherein the typology of the chemical species present enables distinction between various solvation states, in particular hydration states, of the same molecule.

7. The method for the predictive determination of an excess chemical potential as claimed in any of claims 1 to 6, further including a step b' consisting in producing a molecular histogram of at least one of said physical magnitudes.

8. The method for the predictive determination of an excess chemical potential as claimed in any of claims 1 to 7, further including a step consisting in determining a magnitude deduced from the excess chemical potential, preferably chosen from an activity coefficient of a chemical species or of a molecule present, a boiling point of the solution, and a solubility of a chemical species or of a molecule present.

9. The method for the predictive determination of an excess chemical potential as claimed in any of claims 1 to 8, wherein the solution has a solvent mass content of less than 30% relative to the total mass of the solution, advantageously less than 20%.

10. The method for the predictive determination of an excess chemical potential as claimed in claim 1, wherein the Platonic solids have a characteristic size of between 0.0005 Å and 100 Å, preferably between 0.0005 Å and 6 Å.

11. The use of a method as claimed in any of claims 1 to 10, for the calibration of an industrial facility.

12. The use as claimed in claim 11, wherein the industrial facility is a facility for producing sugars and / or polyols and / or carbohydrates.

13. A computer system comprising a processor, said processor being configured to implement a method as claimed in any of claims 1 to 10 when it receives specific instructions.

14. A computer program product comprising a code configured to implement a method as claimed in any of claims 1 to 10 when it is run by a processor or an electronic control unit.