Analyzing water of simulated moving bed separation process by near infrared spectroscopy

Abstract

A simulated moving bed (SMB) separation method and apparatus suitable for determining the water content of a compound stream circulating in an SMB unit (e.g., comprising xylene) by recording a near infrared (NIR) spectrum of a set of samples selected such that the water content of the sample at least partially covers a water content interval of the compound stream; measuring the water content of the sample group by a reference measurement method; developing a multiple regression correlation model between the NIR spectrum of the sample set and the water content of the sample set; recording the NIR spectrum of the compound stream; and applying the multivariate regression correlation model to the NIR spectrum of the compound stream to determine the water content of the compound stream.

Description

Technical Field

[0001] This invention relates to methods for content measurement and to apparatus for controlling and regulating simulated moving bed (SMB) separation units (e.g., for separating aromatic compounds). More specifically, this invention relates to the measurement of water content in a feed stream of compounds (e.g., hydrocarbon compounds) circulating in various separation zones of an SMB unit. A particularly advantageous application of the method according to the invention is the separation of various xylenes in which the feed stream circulating in the unit contains isomers of C8 aromatics, a desorbent, and water, the content of which can vary depending on the measurement point in the SMB unit. The invention also relates to controlling and regulating an SMB unit based on the difference between one or more measured values ​​of water content and one or more set values. Existing technology

[0002] refer to Figure 1 The SMB unit 1 typically comprises a multistage column containing multiple adsorbent beds A1-A12 arranged in series along the flow direction of the fluid used in the column. The fluid that flows successively through the adsorbent beds is referred to as the main fluid to distinguish it from other secondary fluids that can be added to the main fluid via distribution and collection devices, also known as plates P1-P12, which are typically located between two successive beds (arranged perpendicular to the flow direction of the main fluid). The SMB unit also includes a recirculation loop 2, which is typically equipped with a recirculation pump 3 for recirculating the main fluid exiting the multistage column.

[0003] The plate includes at least one collection zone and a distribution and collection network (piping and valve system) that enables the collection of primary fluid and / or the injection of secondary fluids and the mixing of these secondary fluids with the primary fluid. The plate also includes at least one distribution zone that functions to distribute the fluid generated by the mixing of the primary and secondary fluids onto a particle bed located immediately downstream along the flow direction of the main fluid.

[0004] SMB separation processes typically include the following stages: feeding at least one feedstock F and a desorbent D into at least one multistage column, and removing at least one extract E and at least one raffinate R from the multistage column. The feed and discharge points in the plates of the multistage column shift over time during the displacement cycle to correspond to values ​​of the adsorbent bed, and identifying multiple separation zones of the column, particularly the following main zones: By definition, each operating zone is represented by a number: - Zone 1, used for desorption of the product to be separated (e.g., p-xylene), is located between the injection of the desorbent and the extraction of the extract; - Zone 2, used for desorption of impurities (such as isomers of the product to be separated), is located between the extraction of the extract and the injection of the feed. - Zone 3, used for adsorbing the products to be separated, is located between the injection of the feed and the removal of the raffinate; and - Zone 4 is located between the removal of the raffinate and the injection of the desorbent.

[0005] exist Figure 1 In this example, 12 adsorption beds are distributed in zones 1 to 4 according to a so-called 2 / 5 / 3 / 2 configuration, i.e., the bed distribution is as follows: 2 represents the number of beds in Zone 1; 5 represents the number of beds in Zone 2; 3 represents the number of beds in zone 3; and 2 represents the number of beds in zone 4.

[0006] SMB separation processes utilize zeolite adsorption technology, making it possible to replace difficult or even impossible distillation separations. This type of separation requires an optimal trade-off between the separating power of the zeolite and the ability of the product to transfer within it. This is because product separation occurs at the nanoscale within the zeolite, but for a continuous and efficient method, transfer within the zeolite-containing beads must be possible within both macroporous / mesoporous and microporous networks. Of course, these properties depend on the zeolite and beads themselves, as well as parameters such as the composition of the mixture being treated and the operating temperature.

[0007] A crucial parameter for improving separation remains the control of water content. This is because water acts as a "lubricant," allowing the substances to be separated to transfer well within the zeolite. Furthermore, excessive water will "clog" the zeolite and thus reduce productivity—the amount of product that can be separated per unit volume of zeolite. Therefore, depending on the system studied, the water content must be controllable.

[0008] The Karl Fischer method is a reference method for measuring the water content in a feed stream circulating in an SMB unit. Specifically, the Karl Fischer method measures water content via titration chemistry. The principle of the Karl Fischer method is based on the reaction of sulfur dioxide in a methanol solution in the presence of water being oxidized by iodine, according to Mathematical Equation (1), where imidazole (RN) shifts the equilibrium of Equation (1) to the right, resulting in a maximum reaction rate and a stable equilibrium point. Equilibrium is reached when the color changes from yellow to brown (iodine reduction). Titration can be performed by volumetric or coulometric methods, the latter being more suitable for measuring feed streams with low water content. The maximum permissible deviation of the Karl Fischer method is estimated to be ±10 ppm by weight of the target value of 105 ppm by weight of water in the standard.

[0009] Mathematics 1 While the Karl Fischer method is a reference method, it exhibits several drawbacks when used for monitoring SMB processes. Sampling analysis is dependent on the operator performing the analysis. Variation in operator skills can affect reproducibility, for example, depending on weighing accuracy and execution speed. Removal can introduce interference during the sampling process in this method, and sampling is sensitive to climatic conditions. Sampling time can vary depending on the flow rate of the fluid being removed, and the sample may no longer be representative. Real-time analysis is also not feasible due to the inability to achieve high-level control of the unit at the required analysis frequency, making the method's administration rather unresponsive. Therefore, the determination of water content in one or more streams circulating within the SMB unit could be improved. Invention Overview Against this background, the first objective of this specification is to overcome the problems of the prior art and to provide an SMB method and an SMB unit that enable better yields and productivity without affecting the purity of the separated products.

[0011] According to the first aspect, the above-mentioned objectives and other advantages are achieved by a simulated moving bed separation method (e.g., for separating xylene, particularly para-xylene), said method comprising a stage of measuring or determining the water content of the compound feed stream circulating in the simulated moving bed separation unit, said measurement stage comprising: a) Record the near-infrared spectra of a set of samples, selecting the samples such that the water content of the samples at least partially covers the water content range of the compound feed stream; b) Measure the water content of the sample group using a reference measurement method; c) Develop a multiple regression correlation model between the near-infrared spectra of the sample group and the water content of the sample group; d) Record the near-infrared spectrum of the compound stream; e) Apply the multiple regression correlation model from stage c) to the near-infrared spectrum of the compound stream to establish / determine the water content of the compound stream.

[0012] According to one or more embodiments, the method includes stage f): changing the water content in the compound stream in response to a deviation exceeding a threshold between the water content of the compound stream and a set water content in the compound stream.

[0013] According to one or more embodiments, the near-infrared spectrum of the compound stream is recorded by an online spectroscopic analysis device, and / or the water content of the sample group is measured by the Karl Fischer method.

[0014] According to one or more implementation schemes, the development of the multiple regression correlation model includes statistical analysis of the data.

[0015] According to one or more embodiments, the multivariate regression correlation model is developed by applying one or more analytical methods selected from partial least squares regression, principal component regression, multiple linear regression, neural network, direct classical least squares regression, indirect classical least squares regression, or inverse least squares regression to the near-infrared spectra of the sample group.

[0016] According to one or more embodiments, the multivariate regression correlation model is developed by applying one or more analytical methods selected from partial least squares regression, multiple linear regression, and artificial neural networks to the near-infrared spectra of the sample group.

[0017] According to one or more implementation schemes, the multivariate regression correlation model is developed by applying partial least squares regression to the near-infrared spectra of the sample group.

[0018] According to one or more embodiments, the recording phase a) and / or d) and / or the measurement phase b) are carried out under representative temperature and pressure conditions for the compound flow under consideration.

[0019] According to one or more embodiments, the method includes stage g): measuring the water content of the considered compound stream by a reference measurement method.

[0020] According to one or more embodiments, the near-infrared spectra of the sample group and / or the considered compound stream are recorded at at least one recording point selected from the following: - The area located between the injection of the desorbent and the removal of the extract; - The area located between the extraction of the extract and the injection of the raw material; - The area located between the injection of raw materials and the removal of raffinate; - The area located between the removal of the raffinate and the injection of the desorbent; - Main fluid recirculation loop; - Recirculation pump; - Raw material loop; - Raffinate circuit; and - Extraction circuit.

[0021] According to one or more embodiments, near-infrared spectra of the sample group and / or the considered compound stream are recorded at least in the extract loop or raffinate loop.

[0022] According to one or more embodiments, the sample group comprises at least 10 samples, preferably at least 20 samples, and most preferably at least 50 samples.

[0023] According to one or more embodiments, the sample set includes at least a portion of samples prepared in the laboratory or at least a portion of samples derived from at least one extraction point of the simulated moving bed separation unit.

[0024] According to one or more embodiments, the sample exhibits a water content between 10 wt ppm and 200 wt ppm, preferably between 30 wt ppm and 150 wt ppm, and most preferably between 40 wt ppm and 130 wt ppm.

[0025] According to the second aspect, the above-mentioned objectives and other advantages are achieved by a simulated moving bed separation unit (e.g., for separating xylene, particularly para-xylene), said simulated moving bed separation unit comprising means for measuring or determining the water content of the compound feed stream circulating in the simulated moving bed separation unit, said measuring means comprising: - At least one immersion probe or circulation pool is placed at the recording point of the considered compound flow in the simulated moving bed separation unit. - A system for analysis by near-infrared spectroscopy, which is adapted to generate near-infrared spectra of the compound stream at the recording point; - An operating system adapted to develop a multiple regression correlation model between the near-infrared spectra of a sample set and the water content of the sample set, and to apply the multiple regression correlation model to the near-infrared spectra of the compound stream to establish the water content of the compound stream; - A device for controlling and adjusting the water content in the compound stream in response to a deviation exceeding a threshold between the water content of the compound stream and a set water content; and - Optionally, the sampling system includes a retrieval device item adjacent to the immersion probe or the circulation pool; - Optionally, a reference measuring device for analyzing samples; - Optionally, an operating condition sensor adjacent to the immersion probe or the circulation pool.

[0026] The implementation of the methods and units according to the foregoing aspects, as well as other features and advantages, will become apparent upon reading the following description, which is given by way of example only and not as a limitation, and with reference to the following figures.

[0027] List of Attached Figures Figure 1 The diagram shows an SMB unit in which the main fluid circulates, and in which feedstock and desorbent are fed and extracted, and raffinate is removed.

[0028] Figure 2A schematic diagram of an apparatus according to the invention for controlling and regulating an SMB unit is shown. The apparatus specifically includes an immersion probe, a system for analysis by near-infrared spectroscopy, an operating system, and means for controlling and regulating water content.

[0029] Figure 3 This shows a linear correlation between the water content determined by the method according to the invention and the value obtained by the Karl Fischer method.

[0030] Description of the implementation plan Embodiments of the present invention will now be described in detail. Numerous specific details are set forth in the following detailed description to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention may be practiced without these specific details. In other instances, well-known features have not been described in detail to avoid unnecessarily complicating the description.

[0031] In this patent application, the term "comprising" is synonymous with "including" and "containing" (meaning the same) and is inclusive or open-ended, not excluding other unstated elements. It is to be understood that the term "comprising" includes the exclusive and closed term "consisting of." In this specification, the terms "substantially" or "materially" correspond to approximations of ±5%, preferably of ±1%, and very preferably ±0.5% (e.g., ±0.1%). For example, an effluent substantially comprising compound A or composed of compound A corresponds to an effluent containing at least 95% by weight of compound A. Furthermore, in this specification, the term "content" corresponds to the weight concentration of the entity under consideration.

[0032] In particular, the present invention relates to an SMB method comprising a stage of measuring or determining the water content of a compound (e.g., hydrocarbon compound) feed stream (main fluid or secondary fluid) circulating in various separation zones of an SMB unit. The water content is measured by a spectrum obtained by near-infrared spectroscopy (PIR) of the compound feed stream under consideration and by a method for processing said spectrum. Specifically, the measurement stage according to the invention includes: a) Record the NIR spectra of a set of samples, selecting the samples such that the water content of the samples at least partially covers the water content range of the compound stream under consideration, which is circulated in the SMB unit. b) Measure the water content of the sample group using a reference measurement method (e.g., the Karl Fischer method); c) Develop a multiple regression correlation model between the NIR spectra of the sample group and the water content of the sample group (e.g., through statistical analysis of the data). d) Record the NIR spectra of the considered compound feed stream circulating in the SMB unit; e) Apply the multiple regression correlation model from stage c) to the NIR spectrum of the compound stream under consideration to establish the water content of the compound stream under consideration; f) If the deviation between the water content of the compound stream and a set water content in the compound stream exceeds a threshold, change the water content of the compound stream; and g) Optionally, the water content of the compound stream under consideration is measured using a reference measurement method (optional validation phase).

[0033] Phases a), b), and c) are known as the model calibration phase; phases d), e), and f) are regular phases; and phase g) is a phase used to validate the quality of the model (e.g., the point stage).

[0034] Advantageously, NIR spectroscopy enables the study of phenomena arising from the interaction between matter and light, encompassing the electromagnetic spectrum from 0.8 µm to 2.5 µm. The resulting absorption spectra exhibit harmonic and combined vibrational bands. Chemical groups exhibiting absorption in the near-infrared region are primarily of the XH form, where X corresponds to a carbon, nitrogen, or oxygen atom, and H is a hydrogen atom. NIR spectroscopy can therefore detect the presence of water in a sample.

[0035] Furthermore, NIR spectroscopy is a rapid and non-destructive analytical method that can be installed online on the SMB unit (i.e., online analysis; remotely controllable) due to the use of fiber-compatible analytical systems (e.g., including circulating cells or immersion probes). Therefore, unlike point measurements requiring laboratory processing, this invention enables continuous acquisition of water content in the compound feed stream of the SMB unit. NIR spectroscopy can monitor changes over time and also allows for servo-controlled water injection flow rate.

[0036] According to one or more embodiments, the recording phases a) and / or d) and / or the measurement phases b) and / or g) are carried out under representative temperature and pressure conditions for the compound flow under consideration. Advantageously, this improves the accuracy of the development phase c) and the application phase e) of the multiple regression correlation model.

[0037] According to one or more embodiments, the sample set comprises at least 10 samples, preferably at least 20 samples, and very preferably at least 50 samples. According to one or more embodiments, the sample set includes at least a portion of samples prepared in a laboratory. According to one or more embodiments, the sample set includes at least a portion of samples from at least one extraction point of the SMB unit. According to one or more embodiments, the samples exhibit a water content variation, for example, between 10 wt ppm and 200 wt ppm, preferably between 30 wt ppm and 150 wt ppm, and very preferably between 40 wt ppm and 130 wt ppm. The samples are analyzed using a preferred online NIR spectroscopy technique and a reference measurement method in the laboratory.

[0038] Any NIR spectroscopy technique is acceptable for acquiring the spectrum of a sample. For example, an NIR spectroscopy device suitable for the method according to the invention is a "process" spectrometer, which does not require sample preparation. For online recording, acquisition can be performed, for example, using a transmittance / reflectance immersion probe or a circulating cell in the sampling loop. Acquisition can be performed at various optical path lengths, such as 5 mm and 10 mm, within the spectral range of 800 nm to 2200 nm.

[0039] According to one or more implementation schemes, the reference measurement method for water content is the Karl Fischer method.

[0040] Multiple regression correlation models enable the correlation between the NIR spectra of a sample group and its water content. Applying a multiple regression correlation model to the NIR spectra of a considered compound stream allows the establishment of the water content of that stream. Multiple regression applies at least one mathematical equation to the NIR spectra of the samples to output a representative value of water concentration for each sample and the considered compound stream. Specifically, the correlation between the NIR spectra of the sample group and a reference water content (e.g., obtained via Karl Fischer) is obtained through statistical analysis. Preferably, the spectra are pre-processed to eliminate interfering random variability. Common pre-processing methods include, for example, mean centering, normalization, and derivation of the data. The pre-processed data can then be utilized using one or more multivariate methods to reduce the dimensionality of the data and extract relevant information that can be correlated with the physicochemical properties of interest.

[0041] According to one or more embodiments, a multiple regression correlation model is developed by applying one or more analytical methods selected from partial least squares (PLS) regression, principal component regression (PCR), multiple linear regression (MLR), neural networks (ANN, CNN), direct classical least squares (DCLS) regression, indirect classical least squares (ICLS) regression, or inverse least squares (ILS) regression to the NIR spectra of a sample group. This list is not exhaustive. According to one or more embodiments, the one or more analytical methods are selected from partial least squares (PLS) regression, multiple linear regression (MLR), and artificial neural networks (ANN). According to one or more embodiments, the method is partial least squares (PLS) regression.

[0042] The correlation model is applied to the NIR spectrum of the considered compound stream to determine its water content. Once applied, the value given at the model's output corresponds to the property of interest, namely the water concentration of the considered compound stream.

[0043] According to one or more embodiments, the NIR spectra of the considered compound stream circulating in the SMB unit are recorded online. Preferably, the method according to the invention includes a method for continuously measuring / determining the water content present at one (or more) considered points in the SMB unit using an immersion probe, wherein, - Send the signal to the immersion probe. - Recover the NIR signal from the immersion probe and send it to the NIR spectrometer. - Recover the NIR spectrum of the considered NIR signal at the output of the NIR spectrometer. - A multiple regression correlation model is applied to the NIR spectrum under consideration to establish the water content of the compound feed at the considered point in the SMB unit.

[0044] refer to Figure 1 According to one or more embodiments, NIR spectra of the considered compound stream circulating in the sample group and / or SMB unit are recorded at at least one measurement point (e.g., 2, 3, or 4 measurement points), such as in zones 1, 2, 3, and / or 4 of a multistage column, recirculation loop 2 (e.g., recirculation pump 3), feed loop (e.g., feed pump), raffinate or extract loop (e.g., distillation column). Advantageously, online recording of NIR spectra enables data to be obtained under representative temperature and pressure conditions of the considered compound stream.

[0045] According to one or more embodiments, a feedstock containing C8 aromatics is fed into an SMB unit to produce raffinate and extract. According to one or more embodiments, the SMB unit is a hybrid unit utilizing an SMB stage and a crystallization stage.

[0046] According to one or more embodiments, for example, when the SMB unit is a unit for separating xylene in SMB, when there is a single measurement point on the unit, the point is preferably located on the extract loop or raffinate loop. When there are two measurement points on the unit, the first point is preferably located on the extract loop and the second point is preferably located on the raffinate loop. According to one or more embodiments, when there are two measurement points on the unit, the first point is preferably located on the extract loop or raffinate loop, and the second point is preferably located on the feed loop or desorbent loop, or on the recirculation loop (e.g., near the recirculation pump). Optionally, the measurement point may be located in the rectification zone of the raffinate distillation column. Furthermore, when the crystallization stage is performed, the measurement point may be located on the liquid feed stream at the outlet of the crystallization unit, i.e., on the production line where the mother liquor of the product of interest (e.g., para-xylene) to be separated is depleted.

[0047] Advantageously, the method according to the invention enables the control and regulation of the SMB unit. For example, when the deviation between the water content of the compound feed stream determined by the method and the setpoint water content curve present in the compound feed stream is greater than a threshold (e.g., 20 wt ppm, preferably 15 wt ppm, very preferably 10 wt ppm), a control and regulation phase is applied to one or more operating variables of the method (e.g., valve flow rate) to change the water content of the compound feed stream under consideration. According to one or more embodiments, a control and regulation phase is applied to at least one operating condition selected from: main fluid flow rate, feed flow rate, desorbent flow rate, extract flow rate, raffinate flow rate, water injection (e.g., injection into the feed or desorbent) flow rate, displacement time, and temperature in the SMB unit.

[0048] refer to Figure 2 The present invention also relates to a simulated moving bed separation comprising a measuring device for determining the water content of a compound stream (preferably containing xylene) circulating in the simulated moving bed separation unit, the measuring device comprising: I) At least one immersion probe (or circulation pool) 4, which is placed at the recording point of the considered compound flow 5 in the SMB unit. II) Systems for analysis via near-infrared spectroscopy; Optionally, the sampling system 7 includes a retrieval device item adjacent to the immersion probe or circulation cell; III) Optionally, a reference measuring device 8 for analyzing samples; IV) Operating System 9; V) Device 10 for controlling and regulating the water content in the considered compound feed stream; and VI) Optionally, the operating condition sensor 11 is located adjacent to the immersion probe or the circulation pool 4.

[0049] refer to Figure 2 An immersion probe 4 is immersed at an NIR spectral recording point in SMB unit 1, at which the water concentration in the considered compound feed stream 5 is determined. In response to a control signal, the NIR signal emitted at the recording point is collected by the immersion probe 4 and transmitted to a system 6 for analysis via NIR spectroscopy, which generates and records the NIR spectrum corresponding to that recording point. The NIR spectrum is then sent to an operating system 9 (e.g., a computer) for analysis.

[0050] Optionally, in the area near the recording point, the operating condition sensor 11 (e.g., temperature and / or pressure) is immersed in the SMB unit 1 so that the operating conditions of the recording point can be sent to the operating system 9.

[0051] Operating system 9 analyzes NIR spectra and optional operating conditions, and determines the water content of the considered compound stream 5 at the recording point by applying a multiple regression correlation model to the NIR spectra.

[0052] By comparing the water content of the considered compound stream thus obtained with one or more set water content values, the control and regulation device 10 acts on one or more operating variables of the method, such as the flow rate of a valve, to change the water content of the considered compound stream.

[0053] Optionally, in an area near the recording point, the sampling system 7 is positioned to retrieve one or more samples, which are analyzed by the reference measuring device 8. Advantageously, the sampling system 7 enables the provision of samples for the model calibration phase. Advantageously, the sampling system 7 enables the provision of samples for an optional validation phase. Example

[0054] A group of more than 50 samples was selected. These samples were primarily from two of the four effluents from the xylene separation process. Fifteen samples were prepared in the laboratory and acquired on the benchtop to increase sample variability; all other data were obtained from online NIR spectra acquired during dedicated testing. The same spectrometer was used for analysis. Spectra were acquired on a Metrohm-branded Process XDS spectrometer.

[0055] The modeled property is the water content of the effluent produced by the xylene separation process. For in-situ collections, the water content varied over time under stable pressure and temperature conditions during the test, while for laboratory collections, data were acquired at ambient temperature and atmospheric pressure. Each sample was taken and immediately analyzed on a Karl Fischer 756 coulometric meter equipped with generator electrodes featuring a diaphragm bearing the brand name Metrohm.

[0056] The establishment of a multiple regression model between the NIR spectra of the sample group and the water content of the sample group, as well as the preprocessing of the NIR spectra, were performed as follows: normalization of the total spectral area, extended multiplicative signal correction (EMSC), and mean centering. The selected model validation criterion was the root mean square error of cross-validation (RMSECV) (an example of PLS ​​regression). The lower this value, the better the prediction of the property of interest. For the model constructed to predict the water content of xylene mixtures, the RMSECV was 9.13, with 4 latent variables. Therefore, the mean error of the cross-validation model was less than 10 ppm compared to the error of the reference method. In this embodiment, statistical criteria were used to validate the model. The regression coefficient R was considered in the validation of this model. 2 The deviation and the RMSECV / RMSEC ratio (where RMSEC corresponds to the average calibration error; the closer the values ​​are to 1, the closer the ratio is to 1, which is the target). Figure 3 The figure shows the correlation line (solid line) obtained relative to the experimental data (shaded line) (horizontal axis: measured value, vertical axis: predicted value).

Claims

1. A simulated moving bed separation method, comprising determining the water content of a compound stream circulating in a simulated moving bed separation unit (1), wherein the water content determination Includes the following stages: - Record the near-infrared spectra of a set of samples, selecting the samples such that the water content of the samples at least partially covers the water content range of the compound feed stream; - The water content of the sample group was measured using a reference measurement method; - Develop a multiple regression correlation model between the near-infrared spectra of the sample group and the water content of the sample group; - Record the near-infrared spectrum of the compound feed stream; - The multivariate regression correlation model is applied to the near-infrared spectrum of the compound stream to determine the water content of the compound stream.

2. The method according to claim 1, comprising the following stages: The water content of the compound stream is changed in response to a deviation between the water content of the compound stream and a set water content in the compound stream, where the deviation exceeds a threshold.

3. The method according to claim 1 or claim 2, wherein the near-infrared spectrum of the compound stream is recorded by an online spectroscopic analysis device, and / or wherein the water content of the sample group is measured by the Karl Fischer method.

4. The method according to any one of the preceding claims, wherein the development of the multiple regression correlation model includes statistical analysis of the data.

5. The method according to any one of the preceding claims, wherein the multivariate regression correlation model is developed by applying one or more analytical methods selected from partial least squares regression, principal component regression, multiple linear regression, neural networks, direct classical least squares regression, indirect classical least squares regression, or inverse least squares regression to the near-infrared spectra of the sample group.

6. The method of claim 5, wherein the multivariate regression correlation model is developed by applying one or more analytical methods selected from partial least squares regression, multiple linear regression, and artificial neural networks to the near-infrared spectra of the sample group.

7. The method according to any one of the preceding claims, wherein the one or more recording stages and / or the measurement stages are performed under representative temperature and pressure conditions of the compound stream.

8. The method according to any one of the preceding claims, comprising the following stages: The water content of the compound feed stream is measured using the reference measurement method described above.

9. The method according to any one of the preceding claims, wherein the simulated moving bed separation unit comprises a multistage column, a main fluid recirculation loop (2), a recirculation pump (3), a feed loop, a raffinate loop, and an extract loop, and wherein near-infrared spectra of the sample group and / or the compound stream are recorded at at least one recording point selected from: - The zone of the multistage column located between the injection of the desorbent (D) and the extraction of the extract (E); - The zone of the multistage tower located between the extraction of the extract (E) and the injection of the feed (F); - The zone of the multistage column located between the feed injection (F) and the raffinate removal (R); - The zone of the multistage column located between the removal (R) of the raffinate and the injection (D) of the desorbent; - Main fluid recirculation loop (2); - Recirculation pump (3); - Raw material loop; - Residual raffinate circuit; - Extraction circuit.

10. The method of claim 9, wherein near-infrared spectra of the sample group and / or the considered compound stream are recorded at least in the extract circuit or the received raffinate circuit.

11. The method according to any one of the preceding claims, wherein the sample group comprises at least 10 samples, preferably at least 20 samples, and most preferably at least 50 samples.

12. The method according to any one of the preceding claims, wherein the sample set comprises at least a sample prepared in the laboratory or at least a sample derived from at least one extraction point of the simulated moving bed separation unit (1).

13. The method according to any one of the preceding claims, wherein the sample exhibits a water content between 10 wt ppm and 200 wt ppm, preferably between 30 wt ppm and 150 wt ppm, and most preferably between 40 wt ppm and 130 wt ppm.

14. The method according to any one of the preceding claims, used for separating xylene.

15. A simulated moving bed separation unit comprising a measuring device for determining the water content of a compound stream, preferably containing xylene, the compound stream circulating in the simulated moving bed separation unit (1), the measuring device comprising: - At least one immersion probe or circulation pool (4) is placed at the recording point of the compound flow (5) in the simulated moving bed separation unit (1); - A system (6) for analysis by near-infrared spectroscopy, which is adapted to generate and record the near-infrared spectrum of the compound stream at the recording point; - Operating system (9), which is adapted to develop a multiple regression correlation model between the near-infrared spectra of the sample group and the water content of the sample group, and apply the multiple regression correlation model to the near-infrared spectra of the compound stream to determine the water content of the compound stream; - A device (10) for controlling and adjusting the water content in the compound stream in response to a deviation between the water content of the compound stream and a set water content, wherein the deviation exceeds a threshold; - Optionally, the sampling system (7) includes a retrieval device item adjacent to the immersion probe or the circulation pool (4); - Optionally, a reference measuring device (8) for analyzing samples; - Optionally, an operating condition sensor (11) adjacent to the immersion probe or the circulation pool (4).

Images