Impurity deep removal and purification device in ES fiber solvent recovery

Abstract

The invention provides an impurity deep removal and purification device in ES fiber solvent recovery. Relates to the technical field of electrostatic spinning process and solvent recovery, and comprises a pre-separation module used for removing fiber residues and solid particles larger than 5 microns in solvent-enriched gas; the multi-stage deep condensation module comprises two stages of condensers, and the two stages of condensers condense the solvent steam according to the gradually decreased temperature and intercept micro-droplet impurities; and the composite adsorption module is used for adsorbing residual organic trace impurities and removing water. The impurity deep removal and purification device in ES fiber solvent recovery has the advantages of being low in energy consumption, easy and convenient to maintain, stable in operation and capable of being continuously online for 72 h or above without shutdown, the solvent recovery rate is remarkably increased, the operation cost is reduced, and the potential risk of organic vapor emission to the environment and operation safety is effectively avoided.

Description

Technical Field

[0001] This invention relates to the fields of electrospinning process and solvent recovery technology, specifically to a device for deep removal and purification of impurities in ES fiber solvent recovery. Background Technology

[0002] In electrospinning (ES), polymer solutions are stretched into nano / submicron-sized fibers by a high-voltage electric field, which are widely used in filter materials, biomedical scaffolds, and functional composite materials. During the spinning process, organic solvents such as dimethylformamide, tetrahydrofuran, and ethyl acetate are released in large quantities along with the fibers. Common recovery methods include condensation recovery, adsorption-desorption, distillation separation, and membrane separation. Condensation recovery liquefies the vapors at low temperatures; adsorption-desorption uses activated carbon or molecular sieves to trap organic vapors; distillation separation purifies the solvent based on boiling point differences; and membrane separation utilizes the selective permeation of membranes into the solvent-impurity system for separation.

[0003] However, existing technologies have insufficient deep purification capabilities in practical applications: conventional condensation and adsorption methods are unable to completely remove trace amounts of polymer colloids, microdroplet impurities, and moisture, resulting in the purity of the recovered solvent being lower than the process requirements. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a device for deep removal and purification of impurities in ES fiber solvent recovery, solving the problem of how to achieve deep removal and purification of impurities in ES fiber solvent recovery through a combination of multi-stage separation and model predictive control.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a deep impurity removal and purification device for ES fiber solvent recovery, comprising:

[0006] The pre-separation module is used to remove fiber residues and solid particles larger than 5μm from the solvent-enriched gas.

[0007] A multi-stage deep condensation module includes two-stage condensers. The two-stage condensers condense solvent vapor and trap micro-droplet impurities by decreasing temperature at each stage. The two-stage condensers integrate a temperature gradient dynamic adjustment method based on a thermodynamic-mass transfer coupling numerical model to optimize the condensation efficiency of each stage in real time.

[0008] The composite adsorption module is used to adsorb residual organic trace impurities and remove moisture;

[0009] The membrane dehydration module is used to selectively retain water molecules and polymer colloids through a hydrophobic nanofiltration membrane, thereby achieving further removal of residual water and colloids from the solvent;

[0010] The model prediction control module is used to predict the removal efficiency of each purification module through real-time monitoring of multivariate data, and automatically generate the optimal process parameter adjustment scheme. The model prediction control module includes an online detection subunit, a model prediction calculation subunit, and an optimization instruction subunit.

[0011] Preferably, the method for dynamically adjusting the temperature gradient of the multi-stage deep condensation module is based on the following thermodynamic-mass transfer coupling model:

[0012] R i =k i A i (p i -p eq，i (T i ))

[0013] Among them, R i Let k be the steam condensation rate of the i-th stage condenser. i Let A be the overall mass transfer coefficient of the i-th stage condenser, encompassing both convection and conduction effects. i p is the effective heat transfer surface area of ​​the i-th stage condenser. i For the partial pressure of the solvent vapor entering the i-th stage condenser, p eq，i (T i T is the condensation temperature of the i-th stage. i The corresponding equilibrium saturated vapor partial pressure.

[0014] Preferably, the composite adsorption module includes an activated carbon layer, a molecular sieve layer, and a silica gel layer.

[0015] Preferably, the online regeneration strategy of the composite adsorption module is based on the Sips isothermal adsorption model, and the formula of the Sips isothermal adsorption model is:

[0016]

[0017] Where, q e To balance the adsorption capacity, i.e., the mass of impurities adsorbed per unit mass of adsorbent, q m The maximum adsorption capacity corresponds to the amount of adsorption under saturated adsorbent conditions, where K is the adsorption equilibrium constant and C is the maximum adsorption capacity. eq is the concentration of the impurity to be adsorbed in the solvent at equilibrium, and n is the heterogeneity index, used to characterize the non-uniformity of adsorption sites.

[0018] Preferably, the hydrophobic nanofiltration membrane used in the membrane dehydration module has a pore size range of 1 nm to 10 nm, a static contact angle ≥ 120°, and a membrane flux ≥ 20 L / (m²). 2 (·h), dehydration rate ≥95%, after cross-linking modification, it can withstand pH2~12 solutions.

[0019] Preferably, the online detection subunit is used to collect the moisture content, colloid concentration, and temperature, pressure, and flow rate of the recovered solvent in real time.

[0020] Preferably, the model prediction calculation subunit performs rolling time-domain optimization based on the following state-space model:

[0021]

[0022] Where, x k Let u be the state vector at time k, containing the moisture content and colloid concentration at the outlet of each module. k Let y be the control vector at time k, containing the condensation temperature gradient, adsorption layer flow rate, and transmembrane pressure difference. k Let w be the detection vector, matrix A be the system transition matrix describing the evolution between states, matrix B be the control matrix describing the influence of inputs on states, and matrix C be the output matrix describing the mapping from states to measured values. k v k The process noise and measurement noise are respectively, and both satisfy a zero-mean Gaussian distribution.

[0023] Preferably, the optimization instruction subunit is used to adaptively adjust the operating parameters of each module of the device based on the prediction results of the model prediction calculation subunit, so as to ensure that the purity of the recovered solvent is ≥99.5%.

[0024] This invention provides a device for deep removal and purification of impurities in ES fiber solvent recovery. It has the following beneficial effects:

[0025] This deep impurity removal and purification device for ES fiber solvent recovery employs a four-stage series purification process—pre-separation, multi-stage condensation, composite adsorption, and membrane dehydration—combined with a dynamic temperature gradient adjustment method based on a thermodynamic-mass transfer coupled numerical model. This process enables comprehensive and deep removal of microdroplets with a diameter <5μm, polymer colloids, and trace amounts of moisture from electrospinning tail gas, ensuring that the purity of the recovered solvent consistently reaches or exceeds 99.5%. This completely solves the problem of traditional condensation and adsorption methods being unable to remove ultrafine impurities.

[0026] Furthermore, the multivariate coupling adaptive adjustment based on model predictive control can predict the removal efficiency of the purification module in real time and automatically generate the optimal temperature, flow rate and differential pressure adjustment scheme, realizing closed-loop intelligent optimization of the process parameters of the device. The system has low energy consumption, simple maintenance, stable operation and can be continuously online for more than 72 hours without shutdown, which significantly improves the solvent recovery rate, reduces operating costs, and effectively avoids the potential risks of organic vapor emissions to the environment and operational safety. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the process of realizing the invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Example 1

[0030] like Figure 1 As shown, this embodiment of the invention provides a deep impurity removal and purification device for ES fiber solvent recovery, including a pre-separation module for removing fiber residue and solid particles larger than 5μm from the solvent-enriched gas.

[0031] The specific implementation method is as follows:

[0032] The pre-separation module consists of a cyclone separator and a hydrophobic microporous filter connected in series. It is responsible for the initial interception of fiber residue and solid particles larger than 5μm, ensuring the stable operation of the subsequent deep purification unit.

[0033] Cyclone separator:

[0034] Processing capacity: The designed processing flow rate for spinning tail gas is 1.8 m³ / s. 3 / h, inlet air velocity is about 15m / s.

[0035] Separation characteristics: Vortex rotation radial separation, with an efficiency of ≥92% for retaining particles with a diameter ≥5μm.

[0036] Fiber residue removal: Tests showed that the fiber residue content in the exhaust gas decreased from 20 mg / m³. 3 Reduced to ≤2mg / m 3 .

[0037] Pressure drop index: Typical pressure drop is about 200 Pa.

[0038] Slag discharge method: The bottom conical dust collection hopper automatically discharges the slag at set times, triggering a discharge cycle every 8 hours, without the need to stop the machine.

[0039] Hydrophobic microporous filter:

[0040] Filter media structure: It adopts a hydrophobic polytetrafluoroethylene membrane with a pore size of 2μm.

[0041] Retention efficiency: ≥98% for the collection efficiency of 1–5μm particles remaining after the cyclone separator; solid residue at the exhaust gas outlet ≤0.05mg / m³. 3 .

[0042] Dehydration function: The membrane surface has a hydrophobic contact angle of ≥120°, which can additionally remove water droplets carried by the air, reducing the water content from about 300ppm after cyclone separation to ≤100ppm.

[0043] Operating parameters: Filtration differential pressure is stable at 300–500 Pa, and the module temperature resistance is 80℃.

[0044] Online cleaning: Equipped with a reverse airflow pulse cleaning system, it automatically blows once every 12 hours, with a blowing time of 2 seconds, ensuring a filter life of ≥1000 hours.

[0045] Through the above two-stage pre-separation measures, fibers and solid particles larger than 5μm in the electrospinning tail gas can be efficiently retained, reducing the fiber residue content from the initial 20mg / m³. 3 Reduced to ≤0.05mg / m³ 3 This lays a reliable foundation for the efficient purification of subsequent multi-stage condensation, adsorption, and membrane dehydration modules.

[0046] The multi-stage deep condensation module includes two-stage condensers. The two-stage condensers condense solvent vapor and trap micro-droplet impurities by decreasing the temperature at each stage. The two-stage condensers integrate a dynamic temperature gradient adjustment method based on a thermodynamic-mass transfer coupling numerical model to optimize the condensation efficiency of each stage in real time.

[0047] The method for dynamically adjusting the temperature gradient of a multi-stage deep condensation module is based on the following thermodynamic-mass transfer coupling model:

[0048] R i =k i A i (p i -p eq，i (T i ))

[0049] Among them, R i Let k be the steam condensation rate of the i-th stage condenser. i Let A be the overall mass transfer coefficient of the i-th stage condenser, encompassing both convection and conduction effects. i p is the effective heat transfer surface area of ​​the i-th stage condenser. i For the partial pressure of the solvent vapor entering the i-th stage condenser, p eq，i (T i T is the condensation temperature of the i-th stage. i The corresponding equilibrium saturated vapor partial pressure.

[0050] The specific implementation method is as follows:

[0051] The multi-stage deep condensation module consists of two condensers connected in series, used to efficiently remove micron- and submicron-sized droplet impurities from electrospinning exhaust gas.

[0052] First, the initial operating temperature of the first condenser is set at 40℃, and the heat transfer surface area is approximately 2.0 m². 2 The overall mass transfer coefficient is taken as 1.0 × 10⁻⁶. -4 kg / (m 2 The solvent vapor partial pressure entering this stage of the condenser is approximately 2000 Pa, corresponding to an equilibrium saturated vapor pressure of approximately 1200 Pa. Measured by an online liquid flow meter and turbidity sensor, the actual condensation rate is approximately 0.016 kg / s. The second condenser is set at 20°C with a surface area of ​​approximately 2.5 m². 2 The overall mass transfer coefficient is 1.2 × 10⁻⁶. -4 kg / (m 2 The inlet vapor partial pressure is about 800 Pa, the equilibrium vapor pressure is about 200 Pa, and the condensation rate is about 0.18 kg / s.

[0053] During operation, when the liquid flow rate at the outlet of the first condenser falls below 10 g / s or the turbidity exceeds 1 mg / L, the control system automatically initiates dynamic temperature gradient adjustment: firstly, the temperature of the first condenser is reduced by 2°C to 38°C, and the equilibrium vapor pressure is correspondingly reduced to approximately 1100 Pa; after adjustment, the liquid flow rate recovers to 18 g / s, and the turbidity decreases significantly; if the target is still not met, the system can further reduce the temperature by 2°C each time until it reaches 30°C or meets the flow rate threshold. The second condenser also aims to maintain a condensation flow rate of at least 150 g / s. When the measured value falls below this threshold, the temperature can be automatically fine-tuned to 18°C ​​or lower until the condensation efficiency is restored.

[0054] Through the dynamic temperature regulation method driven by the above-mentioned thermodynamic-mass transfer coupling numerical model, this embodiment maintains the efficient synergistic operation of the two-stage condenser under spinning load or ambient temperature and humidity fluctuations. The rejection rate of microdroplets with a diameter of less than 5 μm can be stabilized at over 99.9%, significantly improving the matching degree of subsequent adsorption and membrane dehydration loads. The purity of the recovered solvent can be maintained at over 99.5% continuously without frequent manual adjustments, making it suitable for stable online operation for more than 72 hours.

[0055] The composite adsorption module is used to adsorb residual organic trace impurities and remove moisture. The composite adsorption module includes an activated carbon layer, a molecular sieve layer, and a silica gel layer.

[0056] The online regeneration strategy of the composite adsorption module is based on the Sips isothermal adsorption model, the formula of which is:

[0057]

[0058] Where, q e To balance the adsorption capacity, i.e., the mass of impurities adsorbed per unit mass of adsorbent, q mThe maximum adsorption capacity corresponds to the amount of adsorption under saturated adsorbent conditions, where K is the adsorption equilibrium constant and C is the maximum adsorption capacity. eq is the concentration of the impurity to be adsorbed in the solvent at equilibrium, and n is the heterogeneity index, used to characterize the non-uniformity of adsorption sites.

[0059] The specific implementation method is as follows:

[0060] The composite adsorption module consists of an activated carbon layer, a molecular sieve layer, and a silica gel layer, with a total filler mass of 10 kg, including 5 kg of activated carbon, 3 kg of molecular sieve, and 2 kg of silica gel. This module is used to further remove residual organic trace impurities and moisture after multi-stage condensation.

[0061] Feeding and performance indicators:

[0062] Concentration of organic impurities in the solvent after condensation: 100 mg / L.

[0063] Moisture content: 1000ppm.

[0064] The required impurity concentration after adsorption is ≤5mg / L, and the moisture content is ≤50ppm.

[0065] Sips isothermal adsorption model parameters:

[0066] Based on experimental calibration, the Sips model was selected to describe the adsorption equilibrium:

[0067] Maximum adsorption capacity: 120 mg / g.

[0068] Equilibrium constant K: 0.15 (L / mg).

[0069] Heterogeneity index n: 0.8.

[0070] Equilibrium adsorption capacity calculation: 93 mg / g.

[0071] This means that each gram of adsorbent can retain approximately 93 mg of organic impurities, ensuring that the outlet concentration is below 5 mg / L.

[0072] Online operation and regeneration triggering:

[0073] The online mass spectrometer and moisture sensor continuously monitor the module outlet. When the concentration of organic impurities rises to 50 mg / L or the moisture content increases to 150 ppm, the adsorbent is considered to be approaching saturation.

[0074] The control system automatically switches to regeneration mode without shutting down the machine, raising the module temperature to 90°C and introducing dry nitrogen gas (flow rate 0.6 m³ / h). 3 ( / h), keep for 8 minutes.

[0075] Regeneration efficiency and cycle stability:

[0076] After online regeneration, the adsorbent regeneration rate reaches over 95%; in 10 consecutive adsorption-regeneration cycle tests, the module's retention rate of organic impurities remains above 95%, and its moisture removal efficiency remains stable above 98%, meeting the process requirements for continuous online operation.

[0077] The membrane dehydration module is used to selectively retain water molecules and polymer colloids using a hydrophobic nanofiltration membrane, thereby further removing residual water and colloids from the solvent. The hydrophobic nanofiltration membrane used in the membrane dehydration module has a pore size range of 1 nm to 10 nm, a static contact angle ≥120°, and a membrane flux ≥20 L / (m²). 2 (·h), dehydration rate ≥95%, after cross-linking modification, it can withstand pH2~12 solutions.

[0078] The model predictive control module is used to predict the removal efficiency of each purification module through real-time monitoring of multivariate data, and automatically generate the optimal process parameter adjustment scheme. The model predictive control module includes an online detection subunit, a model predictive calculation subunit, and an optimization instruction subunit.

[0079] The online detection subunit is used to collect data in real time on the moisture content, colloid concentration, temperature, pressure, and flow rate of the recovered solvent.

[0080] The model prediction computation subunit performs rolling time-domain optimization based on the following state-space model:

[0081]

[0082] Where, x k Let u be the state vector at time k, containing the moisture content and colloid concentration at the outlet of each module. k Let y be the control vector at time k, containing the condensation temperature gradient, adsorption layer flow rate, and transmembrane pressure difference. k Let w be the detection vector, matrix A be the system transition matrix describing the evolution between states, matrix B be the control matrix describing the influence of inputs on states, and matrix C be the output matrix describing the mapping from states to measured values. k v k The process noise and measurement noise are respectively, and both satisfy a zero-mean Gaussian distribution.

[0083] The optimization instruction subunit is used to adaptively adjust the operating parameters of each module of the device based on the prediction results of the model prediction calculation subunit, so as to ensure that the purity of the recovered solvent is ≥99.5%.

[0084] The specific implementation method is as follows:

[0085] Application scenario: Dynamic optimization under high temperature and high humidity conditions.

[0086] Under summer workshop conditions of 35℃ and 70% relative humidity, the moisture content and colloid concentration in the electrospinning exhaust gas increased significantly. The online detection subunit collected the following data in real time:

[0087] The moisture content of the recovered solvent is 1200 ppm.

[0088] The colloidal concentration is 15 mg / L.

[0089] The first-stage condenser has a temperature gradient of 40℃→20℃ and a flow rate of 1.0m. 3 / h.

[0090] The flow rate of the adsorption module bed is 0.8 m. 3 / h; transmembrane pressure difference 0.3 bar.

[0091] The model prediction computation unit substitutes the above data into the pre-trained state-space model and then makes a prediction:

[0092] If the current temperature gradient is maintained, the first-stage condensation rejection rate will drop to 92%, and the second-stage condensation rejection rate will be less than 85%.

[0093] The adsorption bed will saturate within 5 hours at the current flow rate, and the water removal rate cannot be guaranteed to be greater than 95%.

[0094] Based on the rolling time-domain optimization results, the optimized instruction subunit issues and executes the following adjustment scheme:

[0095] Condensation module: The inlet temperature of the first-stage condenser is reduced from 40℃ to 36℃, and the temperature of the second stage condenser is reduced from 20℃ to 16℃.

[0096] Adsorption module: Reduces bed flow rate from 0.8m 3 / h decreased to 0.6m 3 / h, extending the gas-solid contact time.

[0097] Membrane dehydration module: Increases the transmembrane pressure difference from 0.3 bar to 0.35 bar, accelerating water removal.

[0098] Within 10 minutes of the adjustment, online monitoring showed:

[0099] The moisture content dropped rapidly to 200 ppm.

[0100] The colloid concentration decreased to 2 mg / L.

[0101] The condensation rejection rate has rebounded to 99.5%.

[0102] The system operated stably for 48 hours, with the purity of the recovered solvent consistently exceeding 99.7%.

[0103] Example 2

[0104] Unlike Example 1, the application scenario of this example is adaptive response under conditions of sudden load increase.

[0105] When production switched to a high-concentration polymer solution (the DMF concentration in the solvent-enriched gas suddenly increased from 10 vol% to 18 vol%), the online detection subunit monitored in real time that:

[0106] The moisture content rose sharply to 900 ppm.

[0107] The colloid concentration increased from 5 mg / L to 12 mg / L.

[0108] The inlet flow rate of the condenser module is 1.0 m³ / s. 3 / h increased to 1.5m 3 / h.

[0109] The pressure drop of the adsorption module bed increased from 0.2 bar to 0.4 bar.

[0110] The model prediction calculation sub-unit makes predictions in the rolling time domain. If adjustments are not made in a timely manner:

[0111] The condensation rejection rate will drop to 90% within 2 hours.

[0112] The adsorption bed will become saturated ahead of schedule (6 hours), resulting in solvent and moisture backflow.

[0113] Optimize instruction sub-unit generation and execute the following process parameter updates:

[0114] Condensation module: Simultaneously increase the two-stage temperature gradient by 2℃, i.e., the first stage 42℃→22℃, and the second stage 22℃→12℃.

[0115] Adsorption module: Incremental regeneration strategy is activated, adsorption under main operating conditions is paused, and purging with 90℃ hot nitrogen gas for 6 minutes is performed immediately.

[0116] Membrane dehydration module: Temporarily increase the transmembrane pressure difference from 0.25 bar to 0.40 bar and switch to high-intensity vibration self-cleaning mode.

[0117] Within 15 minutes of the adjustment, the system indicators recovered and stabilized.

[0118] The moisture content should be controlled below 150 ppm.

[0119] The colloid concentration decreased to 3 mg / L.

[0120] The condensation rejection rate remained stable at 99.8%.

[0121] The adsorption bed can work continuously for 12 hours without regeneration.

[0122] Example 3

[0123] Unlike Example 1, the application scenario of this example is long-term stable operation in low-temperature winter.

[0124] Under winter workshop conditions with an ambient temperature of only 5°C and a relative humidity of 30%, the residual solvent vapor in the electrospinning exhaust gas is close to room temperature, leading to a decrease in the efficiency of the initial multi-stage deep condensation and adsorption module. The online detection subunit collected the following real-time data:

[0125] Moisture content of the recovered solvent: 850 ppm.

[0126] Colloidal concentration: 8 mg / L.

[0127] Condensation module inlet gas temperature: 18℃, flow rate: 1.2m³ / h 3 / h.

[0128] Adsorption module bed flow rate: 0.7m 3 / h, bed pressure drop 0.25 bar.

[0129] Transmembrane pressure differential during membrane dehydration: 0.20 bar.

[0130] Based on the model's predictive calculations and sub-unit analysis, if the current operating conditions are maintained:

[0131] The condensation rejection rate will decrease from 99.2% to 95%.

[0132] The adsorption bed becomes saturated within 8 hours, and the reflux moisture content rises to 400 ppm.

[0133] The dehydration rate of the membrane dehydration module may drop to 85%.

[0134] To achieve stable operation over a long period (>96 hours), the following adjustment scheme is optimized for automatic issuance and execution by the instruction subunit:

[0135] Preheating measures: An electric preheater is added between the pre-separation module and the first-stage condenser to raise the exhaust gas inlet temperature to 30°C, reduce the temperature difference with the condensation temperature, and improve the condensation driving force.

[0136] Condensation module adjustment: Increase the first-stage condensing temperature from the original setting of 22℃ to 24℃, and the second-stage temperature from 12℃ to 14℃, and evenly distribute the flow rates of the two stages to 1.0m³. 3 / h.

[0137] Adsorption module optimization: reducing bed flow rate to 0.5 m 3 / h, and start the semi-continuous micro-regeneration mode - purge with 85℃ hot nitrogen for 3min every 12h to release adsorption sites.

[0138] Membrane dehydration enhancement: The transmembrane pressure difference is temporarily increased to 0.40 bar, and pulse oscillation cleaning is started—lasting 1 minute every 24 hours—to prevent colloids from depositing on the membrane surface.

[0139] Within 15 minutes of the adjustment, online monitoring showed:

[0140] The moisture content dropped rapidly to 120 ppm.

[0141] The colloidal concentration decreased to 0.8 mg / L.

[0142] The condensation rejection rate was restored to 99.7%.

[0143] The adsorption module has a retention efficiency of >98%.

[0144] Membrane dehydration dehydration rate > 96%.

[0145] After 96 hours of continuous testing, the device did not need to be shut down in a low-temperature environment, and the purity of the recovered solvent remained above 99.8%. Energy consumption was reduced by about 12% compared with conventional operating conditions, which greatly extended the effective life of the adsorbent and membrane module, and achieved long-term, efficient, stable and low-maintenance operation in winter.

[0146] Example 4

[0147] Unlike Example 1, this example focuses on adaptive optimization for production switching and high-viscosity solution conditions.

[0148] When the production task was switched from conventional polyacrylonitrile solution to high-viscosity polylactic acid solution, the solvent system changed from a mixture of dimethylformamide and tetrahydrofuran (volume fractions of 60% and 40%) to pure THF. The THF concentration in the exhaust gas instantaneously increased to 25 vol%, while the solution viscosity increased from 800 mPa·s to 1200 mPa·s. The online detection subunit collected the following data in real time:

[0149] The moisture content of the recovered solvent is 2500 ppm.

[0150] Colloidal concentration 20 mg / L.

[0151] The condenser module inlet temperature is 16℃, and the gas flow rate is 1.0 m³ / s. 3 / h increased to 1.8m 3 / h.

[0152] The pressure drop of the adsorption module bed increased from 0.3 bar to 0.5 bar.

[0153] The transmembrane pressure differential remained at 0.25 bar during membrane dehydration, but the dehydration rate had dropped to 80%.

[0154] Based on the aforementioned multivariate states and combined with digital twin simulation prediction, the model prediction calculation subunit derives the following under the current operating conditions:

[0155] The condenser module retention rate will rapidly decline from 99.4% to 92%.

[0156] The adsorption module is expected to saturate within 4 hours and exhibit significant backflow.

[0157] The membrane dehydration module needs to be cleaned alternately; otherwise, colloidal contamination will reduce the flux by 30%.

[0158] The optimization instruction subunit immediately issues and implements the following comprehensive adjustment plan:

[0159] Preheating adjustment: Start the workshop hot air preheater to raise the temperature of the exhaust gas at the inlet of the condensing module to 28°C, reduce the temperature difference with the two-stage condensing temperature to enhance the condensation driving force.

[0160] Condensation module: The two-stage condensation temperatures were slightly adjusted from the original settings (18℃ for the first stage and 8℃ for the second stage) to (20℃ for the first stage and 10℃ for the second stage), and the flow distribution was optimized to 1.2m³ / s each. 3 / h.

[0161] Adsorption module: Switch to intermittent regeneration mode—purge the adsorption bed with 85℃ hot nitrogen for 5 minutes every 6 hours, and after purging, spray with 0.8m... 3 / h low flow rate to restore adsorption.

[0162] Membrane dehydration module: Increases the transmembrane pressure difference to 0.4 bar, and performs pressure difference reverse cleaning for 3 minutes every 12 hours to remove colloidal blockage on the membrane surface.

[0163] Digital twin calibration: The digital twin model is started simultaneously to calibrate the hot air preheating efficiency and membrane fouling rate in real time, ensuring that the simulation and field conditions have an error of ≤5%.

[0164] Within 10 minutes of implementing the above solution, the system indicators improved significantly:

[0165] The moisture content dropped to 180 ppm.

[0166] The colloidal concentration decreased to 1.5 mg / L.

[0167] The condenser module retention rate has recovered to 99.6%.

[0168] The adsorption module operates stably with a regeneration-adsorption cycle interval of more than 8 hours.

[0169] The membrane dehydration rate returned to over 96%, and the flux recovered to its original level.

[0170] After 120 hours of continuous operation, this embodiment has demonstrated that it can still respond quickly, adjust intelligently, and maintain high efficiency and stability under solvent system switching and high viscosity conditions, fully demonstrating the synergistic advantages of model predictive control and digital twin methods.

[0171] The four embodiments described above address various extreme operating conditions, including high temperature and humidity, sudden load increases, low temperature winters, and solvent system switching. They sequentially employ methods such as online monitoring, dynamic temperature gradient adjustment driven by a thermodynamic-mass transfer coupling model, an online adsorption regeneration strategy based on the Sips model, and collaborative optimization using model predictive control and digital twin simulation to adaptively adjust the multi-stage condensation, composite adsorption, and membrane dehydration modules. Practice has demonstrated that under various operating conditions, the device can rapidly reduce the moisture content of the recovered solvent from thousands of ppm to below hundreds of ppm within minutes, control the colloid concentration at 1–3 mg / L, maintain a stable condensation rejection rate of ≥99.5%, and maintain adsorption and dehydration efficiencies above 95%. Furthermore, it can operate continuously online for 72–120 hours without manual intervention, significantly improving solvent purity, recovery rate, and system stability.

[0172] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A device for deep removal and purification of impurities in ES fiber solvent recovery, characterized in that, include: The pre-separation module is used to remove fiber residues and solid particles larger than 5μm from the solvent-enriched gas. A multi-stage deep condensation module includes two-stage condensers. The two-stage condensers condense solvent vapor and trap micro-droplet impurities by decreasing temperature at each stage. The two-stage condensers integrate a temperature gradient dynamic adjustment method based on a thermodynamic-mass transfer coupling numerical model to optimize the condensation efficiency of each stage in real time. The composite adsorption module is used to adsorb residual organic trace impurities and remove moisture; Membrane dehydration module, used to selectively retain water molecules and polymer colloids through hydrophobic nanofiltration membrane; The model prediction control module is used to predict the removal efficiency of each purification module through real-time monitoring of multivariate data, and automatically generate the optimal process parameter adjustment scheme. The model prediction control module includes an online detection subunit, a model prediction calculation subunit, and an optimization instruction subunit.

2. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The temperature gradient dynamic adjustment method of the multi-stage deep condensation module is based on the following thermodynamic-mass transfer coupling model: R i =k i A i (p i -p eq，i (T i )) Among them, R i Let k be the steam condensation rate of the i-th stage condenser. i Let A be the overall mass transfer coefficient of the i-th stage condenser, encompassing both convection and conduction effects. i p is the effective heat transfer surface area of ​​the i-th stage condenser. i For the partial pressure of the solvent vapor entering the i-th stage condenser, p eq，i (T i T is the condensation temperature of the i-th stage. i The corresponding equilibrium saturated vapor partial pressure.

3. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The composite adsorption module includes an activated carbon layer, a molecular sieve layer, and a silica gel layer.

4. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The online regeneration strategy of the composite adsorption module is based on the Sips isothermal adsorption model, and the formula of the Sips isothermal adsorption model is as follows: Where, q e To balance the adsorption capacity, i.e., the mass of impurities adsorbed per unit mass of adsorbent, q m The maximum adsorption capacity corresponds to the amount of adsorption under saturated adsorbent conditions, where K is the adsorption equilibrium constant and C is the maximum adsorption capacity. eq is the concentration of the impurity to be adsorbed in the solvent at equilibrium, and n is the heterogeneity index.

5. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The hydrophobic nanofiltration membrane used in the membrane dehydration module has a pore size range of 1 nm to 10 nm, a static contact angle ≥120°, and a membrane flux ≥20 L / (m²). 2 (·h), dehydration rate ≥95%, after cross-linking modification, it can withstand pH2~12 solutions.

6. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The online detection subunit is used to collect data in real time on the moisture content, colloid concentration, temperature, pressure, and flow rate of the recovered solvent.

7. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The model prediction calculation subunit performs rolling time-domain optimization based on the following state-space model: Where, x k Let u be the state vector at time k, containing the moisture content and colloid concentration at the outlet of each module. k Let y be the control vector at time k, containing the condensation temperature gradient, adsorption layer flow rate, and transmembrane pressure difference. k The detection vector is defined by matrix A, which is the system transition matrix; matrix B is the control matrix, describing the influence of the input on the state; and matrix C is the output matrix. k v k The process noise and measurement noise are respectively, and both satisfy a zero-mean Gaussian distribution.

8. The deep impurity removal and purification device for ES fiber solvent recovery according to claim 1, characterized in that: The optimization instruction subunit is used to adaptively adjust the operating parameters of each module of the device based on the prediction results of the model prediction calculation subunit.

Images