Rapeseed oil alkali refining process based on three-section type oil-water separation tank

Abstract

The invention relates to the technical field of rapeseed oil alkali refining, and discloses a rapeseed oil alkali refining process based on a three-section type oil-water separation tank. In order to solve the problem of insufficient synergism of an oil-soap separation process in the prior art, the temperature of an alkali refining key link is accurately locked at 85-90 DEG C through a plate type heater, the temperature interval can ensure that phosphoric acid and phospholipid as well as sodium hydroxide and free fatty acid fully react, the reaction efficiency is improved, the proper flowability of nigre can be kept, and the oil-soap separation process is simple and convenient to operate. The method creates optimal conditions for efficient separation of the soap-removing centrifugal machine, effectively solves the core problem of low separation efficiency caused by out-of-control temperature, and remarkably improves the oil-soap separation effect.

Description

Technical Field

[0001] This invention relates to the field of rapeseed oil alkali refining technology, and in particular to a rapeseed oil alkali refining process based on a three-stage oil-water separator. Background Technology

[0002] Rapeseed oil, one of the most produced edible vegetable oils in my country, is rich in unsaturated fatty acids, vitamin E, and other nutrients, and has wide applications in food processing and catering. After being processed by pressing or solvent extraction, crude rapeseed oil contains various harmful impurities such as free fatty acids (FFA), phospholipids, gums, and mechanical impurities. These impurities not only affect the flavor, color, and shelf life of the oil but may also pose potential health risks. Therefore, refining is the core process in rapeseed oil production, and the alkali refining process, as a key step in removing free fatty acids and phospholipids, directly determines the quality of the final product, production costs, and environmental benefits.

[0003] However, the current alkali refining process has the following three major problems: First, the synergy of the oil-soap separation process is insufficient. In the traditional process, the desanding centrifugation parameters and saponification reaction conditions are not effectively matched, resulting in a high amount of soap residue, which significantly increases the processing burden of the subsequent water washing process. Secondly, the temperature control accuracy is low, and the temperature fluctuation range of the plate heater is large. If the temperature is too high, it will easily cause rapeseed oil to oxidize and deteriorate. If the temperature is too low, the saponification reaction will be insufficient, which restricts the improvement of separation efficiency. Third, the wastewater treatment process is outdated and lacks an oil-water separation design tailored to the characteristics of rapeseed oil. The wastewater has a high oil content, which not only wastes rapeseed oil raw materials but also increases the environmental protection costs for enterprises. Summary of the Invention

[0004] The technical problem to be solved by this invention is that the existing technology has the disadvantage of insufficient synergy in oil-soap separation process. To this end, we propose a rapeseed oil alkali refining process based on a three-stage oil-water separator.

[0005] To achieve the above objectives, this application adopts the following technical solution: a rapeseed oil alkali refining process based on a three-stage oil-water separator, comprising the following steps: S1: Precise Temperature Control for Acidification-Saponification: Crude rapeseed oil, after being filtered and purified, is first preheated by an energy-saving heat exchanger and then enters an acid mixer to be fully mixed with a food-grade phosphoric acid solution. It is then transferred to an alkali mixer to react with a sodium hydroxide solution and subsequently sent to a delayed reaction tank. The reaction temperature is stably controlled at 85-90℃ by a plate heater to ensure that free fatty acids are fully saponified and phospholipids are completely converted, laying the foundation for subsequent separation. S2: Oil-Soap Co-separation: Based on the characteristics of the oil-soap mixture after saponification, the operating parameters of the desoap centrifuge are optimized to enhance the oil-soap separation effect and make the oil-soap separation more thorough. The separated soap residue is temporarily stored and then transported outside the workshop for unified processing. S3: Water Washing - Oil-Water Recovery: After desoaping, the rapeseed oil enters the water washing centrifuge. The residual soap residue is removed by precisely controlling the water washing conditions. The wastewater generated by the water washing is introduced into the oil-water separation tank. The wastewater first enters the buffer zone to achieve a stable rise in the liquid level. By adjusting the height difference between the oil separation regulating device and the baffle, the specific gravity difference between rapeseed oil and water is used to achieve static center of gravity separation. Water slowly enters the still water zone from the bottom of the baffle, and the floating oil on the water surface slowly flows into the oil storage zone through the oil separation regulating device. Then, the water pump and oil pump are automatically started and stopped by the liquid level sensor or tuning fork to complete the efficient oil-water recovery. S4: Drying and impurity removal: The washed rapeseed oil enters a vacuum dryer to remove moisture and volatile substances, ultimately obtaining alkali-refined oil with the required purity, providing high-quality raw materials for subsequent decolorization and dewaxing processes.

[0006] Preferably, in step S1, crude rapeseed oil is filtered to remove impurities using a plate and frame filter or a bag filter. The plate and frame filter or bag filter has a mesh size of 100-200 mesh, a filtration temperature of 60-70℃, a preheating temperature of 35-40℃ for the energy-saving heat exchanger, a food-grade phosphoric acid solution concentration of 85%, diluted to a 10% solution for use, and a phosphoric acid addition amount of 0.05%-0.2% of the crude oil weight. The reaction time between the crude oil and phosphoric acid is 5-15 minutes, and the reaction temperature is 35-40℃.

[0007] Preferably, the sodium hydroxide solution needs to be determined based on the acid value of crude oleic acid, as shown in the following formula: ,in The coefficient is the amount of sodium hydroxide required to neutralize each unit of acid value. The excess phosphoric acid is the additional phosphoric acid used for neutralization, ranging from 0.05% to 0.1%.

[0008] Preferably, the reaction time in the delayed reaction vessel is 30 minutes, allowing sodium hydroxide to fully neutralize with free fatty acids to form soap particles, which then continue to flocculate and grow.

[0009] Preferably, in step S2, the oil-soap mixture is fed into a desoap centrifuge with a drum speed of 6000-7500 rpm, an inlet pressure of 0.2-0.5 MPa, and a back pressure control of 0.1-0.3 MPa.

[0010] Preferably, in step S3, soft water is added to the washing centrifuge. Specifically, 10%-15% of the oil weight of soft water is added, with a hardness of <50 ppm, and the water temperature is 5-10°C higher than the oil temperature to prevent emulsification.

[0011] Preferably, the oil-water separator is divided into three areas: a buffer zone on the far left, a still water zone and an oil storage zone on the right. The oil storage zone is located above the still water zone. An energy dissipator is installed at the water inlet of the buffer zone to ensure a smooth water flow.

[0012] Preferably, in S4, the temperature of the vacuum dryer is set to 90-105°C and the absolute pressure is ≤50 mbar.

[0013] Preferably, in S1 and S3, an online near-infrared spectrometer is integrated to monitor the acid value, peroxide value, phospholipid content, and moisture content of crude oil in real time, and the monitored data is sent to a central multivariate process control model. This model aims to minimize the chemical consumption and the variance of the key indicators of the final product, and dynamically solves a set of optimal process parameters. Then, an intelligent decision engine based on reinforcement learning is introduced. This intelligent decision engine uses historical production data as a training set to continuously learn and optimize the set point of the central multivariate process control model.

[0014] Preferably, the online near-infrared spectrometer is equipped with a high-temperature and pressure-resistant flow cell, with a detection band of 800-2500nm. The optical quantitative model established by the partial least squares algorithm can output the predicted values ​​of acid value, peroxide value, and phospholipid content in real time, with a prediction relative deviation greater than 3.0. The sampling interval of the central multivariate process control model is 30 seconds, and the control cycle is 2 minutes. Its internal model is established based on historical step response data through system identification methods. The intelligent decision engine is trained using a proximal strategy optimization algorithm. Its state space includes the current detection value of the online near-infrared spectrometer and the average value of key process parameters in the previous 3 hours. The action space is the adjustment suggestion of the set point of the central multivariate process control model. The reward function is the weighted sum of negative chemical consumption and product quality deviation.

[0015] The technical effects and advantages of this invention are as follows: In this invention, to address the problem of temperature fluctuations affecting separation efficiency in traditional processes, an innovative 85-90℃ constant temperature process system is constructed: the temperature of key alkali refining steps is precisely locked at 85-90℃ using a plate heater. This temperature range ensures that phosphoric acid and phospholipids, and sodium hydroxide and free fatty acids react fully, improving reaction efficiency, while also maintaining suitable fluidity of the soap residue, allowing it to form clear stratification with rapeseed oil. This creates optimal conditions for efficient separation in the desoap centrifuge, effectively solving the core problem of low separation efficiency caused by temperature runaway, and significantly improving the oil-soap separation effect. Attached Figure Description

[0016] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts: Figure 1This is a flowchart of the overall process for alkali refining of rapeseed oil based on a three-stage oil-water separator, according to the present invention. Detailed Implementation

[0017] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0018] Reference Figure 1 As shown, the present invention provides a technical solution: a rapeseed oil alkali refining process based on a three-stage oil-water separator, comprising the following steps: S1: Precise Temperature Control for Acidification-Saponification: Crude rapeseed oil, after being filtered and purified, is first preheated by an energy-saving heat exchanger and then enters an acid mixer to be fully mixed with a food-grade phosphoric acid solution. It is then transferred to an alkali mixer to react with a sodium hydroxide solution and subsequently sent to a delayed reaction tank. The reaction temperature is stably controlled at 85-90℃ by a plate heater to ensure that free fatty acids are fully saponified and phospholipids are completely converted, laying the foundation for subsequent separation. S2: Oil-Soap Co-separation: Based on the characteristics of the oil-soap mixture after saponification, the operating parameters of the desoap centrifuge are optimized to enhance the oil-soap separation effect and make the oil-soap separation more thorough. The separated soap residue is temporarily stored and then transported outside the workshop for unified processing. S3: Water Washing - Oil-Water Recovery: After desoaping, the rapeseed oil enters the water washing centrifuge. The residual soap residue is removed by precisely controlling the water washing conditions. The wastewater generated by the water washing is introduced into the oil-water separation tank. The wastewater first enters the buffer zone to achieve a stable rise in the liquid level. By adjusting the height difference between the oil separation regulating device and the baffle, the specific gravity difference between rapeseed oil and water is used to achieve static center of gravity separation. Water slowly enters the still water zone from the bottom of the baffle, and the floating oil on the water surface slowly flows into the oil storage zone through the oil separation regulating device. Then, the water pump and oil pump are automatically started and stopped by the liquid level sensor or tuning fork to complete the efficient oil-water recovery. S4: Drying and impurity removal: The washed rapeseed oil enters a vacuum dryer to remove moisture and volatile substances, ultimately obtaining alkali-refined oil with the required purity, providing high-quality raw materials for subsequent decolorization and dewaxing processes.

[0019] Reference Figure 1As shown in this implementation scheme: In step S1, crude rapeseed oil is filtered to remove impurities using a plate and frame filter or a bag filter. The plate and frame filter or bag filter has a precision of 100-200 mesh, and the filtration temperature is 60-70℃ (to reduce the viscosity of the oil and improve the filtration effect). The preheating temperature of the energy-saving heat exchanger is 35-40℃ (too high a temperature will cause excessive denaturation of phospholipids in the subsequent acid reaction, increasing the color of the degummed oil). The concentration of the food-grade phosphoric acid solution is 85%, diluted to a 10% solution for use. The amount of phosphoric acid added is 0.05%-0.2% of the weight of the crude oil. The reaction time between the crude oil and phosphoric acid is 5-15 minutes, and the reaction temperature is 35-40℃.

[0020] The sodium hydroxide solution needs to be determined based on the acid value of crude oleic acid, using the following formula: ,in The coefficient is the amount of sodium hydroxide required to neutralize each unit of acid value. The excess phosphoric acid is the additional phosphoric acid used for neutralization, ranging from 0.05% to 0.1%.

[0021] The reaction time in the extended reaction vessel is 30 minutes, allowing sodium hydroxide to fully neutralize with free fatty acids to form soap particles, which then continue to flocculate and grow.

[0022] In S2, the oil-soap mixture is fed into a desoap centrifuge with a drum speed of 6000-7500 rpm, an inlet pressure of 0.2-0.5 MPa, and a back pressure control of 0.1-0.3 MPa.

[0023] In S3, soft water is added to the washing centrifuge. Specifically, 10%-15% of the oil weight of soft water is added, with a hardness of <50ppm. The water temperature is 5-10℃ higher than the oil temperature to prevent emulsification.

[0024] The oil-water separator is divided into three areas: the leftmost buffer zone, the rightmost still water zone, and the oil storage zone. The oil storage zone is located above the still water zone. An energy dissipator is installed at the inlet of the buffer zone to ensure a smooth water flow.

[0025] In S4, the temperature of the vacuum dryer is set to 90-105℃ and the absolute pressure is ≤50 mbar.

[0026] In S1 and S3, an online near-infrared spectrometer is integrated to monitor the acid value, peroxide value, phospholipid content, and moisture content of crude oil in real time. The monitored data is then fed into a central multivariate process control model. This model aims to minimize the chemical consumption and the variance of key indicators of the final product, and dynamically solves for a set of optimal process parameters. A reinforcement learning-based intelligent decision engine is then introduced. This intelligent decision engine uses historical production data as a training set to continuously learn and optimize the setpoint of the central multivariate process control model.

[0027] Historical production data includes raw material properties, process parameters, and finished product quality. When the online near-infrared spectrometer detects an abnormality in the phospholipid composition of a batch of crude oil, the intelligent decision engine can surpass the real-time optimization of the central multivariate process control model and proactively adjust the phospholipid addition strategy and delayed reaction time in the acidification stage, fundamentally avoiding emulsification risks in subsequent processes. Specifically, a first online near-infrared spectrometer and a second online near-infrared spectrometer are integrated and installed on the acid mixer outlet pipe in step S1 and the water washing centrifuge outlet pipe in step S3, respectively. The first online near-infrared spectrometer collects data in real time. The first online near-infrared spectrometer collects spectral data from the acidified oil sample to predict phospholipid conversion rate and free fatty acid content. The second online near-infrared spectrometer collects spectral data from the washed oil sample in real time to predict residual soap particle concentration and moisture content. The first and second online near-infrared spectrometers transmit the predicted data to a central multivariate process control model via industrial Ethernet. This central multivariate process control model aims to minimize the total consumption of phosphoric acid, sodium hydroxide, and soft water over the next six control cycles (12 minutes), while simultaneously constraining the residual soap particle concentration of the washed oil to <50%. mg / kg, free fatty acid content constrained to <0.1%, phosphoric acid added in S1 using a phosphoric acid metering pump, sodium hydroxide added using a sodium hydroxide solution metering pump, and soft water added in S3 using a soft water addition metering pump. The central multivariate process control model outputs a set of dynamically optimized setpoint instructions every 2 minutes, including: the frequency of the phosphoric acid metering pump in step S1, the stroke of the sodium hydroxide solution metering pump, the temperature setpoint of the delayed reaction tank, and the flow rate of the soft water addition metering pump in step S3. The central multivariate process control model is based on a discretized linear state-space model for prediction, which is obtained from historical data through system identification. Its state equation is: ; Its output equation is: ; in, It is the state vector at time k. This includes estimating the saponification value, reaction temperature, and residual phosphoric acid concentration; It manipulates the variable vector. , The unit is , The flow rate of the sodium hydroxide solution is... This represents the flow rate of the phosphoric acid solution. It is a measurable perturbation vector. , The unit is , For the crude oil feed flow rate, The value of crude oleic acid; It is the output vector. , , Saponification value predicted by NIRS The reaction temperature; It is the system matrix of the corresponding dimension. and These are process noise and measurement noise, respectively, with the following covariance matrix: and ; At each sampling time The central multivariate process control model solves the following quadratic programming problem: Cost function (J): ; in, It is a prediction time domain (10 minutes). It controls the time domain (5 minutes); It is the control increment sequence to be optimized. ; Future based on k-time information Step output prediction; It is the future The step's setpoint vector comes from the upper-level RL engine; These are reference values ​​for the economically optimal manipulated variables provided by the upper-level RL engine; It is the output error weight matrix, which emphasizes the tracking of saponification values; It controls the incremental weight matrix and penalizes operations that are too drastic. It is a control weight matrix that drives the overall system towards the economically optimal operating point. ; The constraints are: ; The specific numerical constraints are as follows: , ; , ; , ; After solving the above quadratic programming problem, the optimal control increment sequence is obtained. It applies only to the first element of the whole: ; At the next sampling time k+1, the entire process is repeated, and the state is re-estimated and optimized based on the new measurement value to achieve rolling time-domain control.

[0028] The RL engine adjusts the setpoint and economic optimal value. It actively seeks the lowest reagent consumption range while ensuring quality, and its reward function directly aims to reduce unit consumption, thereby reducing the unit consumption of sodium hydroxide and phosphoric acid.

[0029] The online near-infrared spectrometer is equipped with a high-temperature and pressure-resistant flow cell, with a detection band of 800-2500nm. An optical quantitative model established using a partial least squares algorithm can output real-time predicted values ​​for acid value, peroxide value, and phospholipid content, with a relative deviation (RPD) greater than 3.0 (note that a larger RPD indicates better model prediction performance). The central multivariate process control model has a sampling interval of 30 seconds and a control cycle of 2 minutes. Its internal model is established based on historical step response data using a system identification method. The intelligent decision engine is trained using a near-end strategy optimization algorithm. Its state space includes the current online near-infrared spectrometer detection value and the average value of key process parameters over the past 3 hours. The action space consists of adjustment suggestions for the central multivariate process control model setpoint. The reward function is a negative weighted sum of chemical consumption and product quality deviation. The intelligent decision engine uses a near-end strategy optimization algorithm as its core. Its state space is a time-series data window of length 20, containing data including: phospholipid and free fatty acid content from 20 consecutive sampling points predicted by the first online near-infrared spectrometer, and corresponding historical values ​​of process parameters in stage S1. Its action space consists of instructions for adjusting the weight matrix in the objective function of the central multivariate process control model, specifically adjusting the penalty coefficients for phosphoric acid consumption, alkali consumption, and water consumption. The reward function is calculated as follows: ; in This is a unit consumption indicator. and for The absolute deviation between the measured value and the target value at any given time. These are the weighting coefficients. For phosphoric acid consumption, This is the unit consumption of sodium hydroxide. The values ​​represent the unit consumption of soft water, Soap (soap granule concentration), and FFA (free fatty acid content). When the first online near-infrared spectrometer detects that the predicted phospholipid content at five consecutive sampling points exceeds the historical average by ±3 standard deviations, it is determined to be an abnormal phospholipid composition. The intelligent decision engine immediately intervenes, temporarily lowering the penalty coefficient for phosphoric acid consumption in the central multivariate process control model and generating an instruction to extend the residence time setting of the delayed reaction tank in step S1 by 5-10 minutes. Real-time monitoring at the outlets of S1 and S3 replaces the traditional manual testing that lags for several hours, improving work efficiency. The central multivariate process control model can predict future trends and make adjustments based on small deviations detected in real time, improving work quality while reducing the consumption of phosphoric acid, sodium hydroxide, soft water, and auxiliary materials, thus reducing work costs.

[0030] With the goal of improving separation efficiency and ensuring oil quality, the alkali refining process is being restructured: 1. Precise temperature control process for acidification-saponification: Crude rapeseed oil, after being filtered and purified, is first preheated by an energy-saving heat exchanger and then enters an acid mixer to be fully mixed with a food-grade phosphoric acid solution. It is then transferred to an alkali mixer to react with a sodium hydroxide solution and subsequently sent to a delayed reaction tank. The reaction temperature is stably controlled at 85-90℃ by a plate heater to ensure that free fatty acids are fully saponified and phospholipids are completely converted, laying the foundation for subsequent efficient separation. 2. Oil-soap co-separation process: Based on the characteristics of the oil-soap mixture after saponification, the operating parameters of the desoap centrifuge are optimized to enhance the oil-soap separation effect and make the oil-soap separation more thorough. The separated soap residue is temporarily stored and then transported outside the workshop for unified processing. 3. Water washing-oil-water recovery linkage process: After desoaping, the rapeseed oil enters the water washing centrifuge, and residual soap residue is removed by precisely controlling the water washing conditions; the wastewater generated by water washing is introduced into a customized oil-water separation tank. The wastewater first enters the buffer zone to achieve a stable rise in liquid level. By adjusting the height difference between the oil separation regulating device and the baffle, the difference in specific gravity between rapeseed oil and water is used to achieve static center of gravity separation. Water slowly enters the still water zone from the bottom of the baffle, and the floating oil on the water surface slowly flows into the oil storage zone through the regulating device. Then, the water pump and oil pump are automatically started and stopped by the liquid level sensor or tuning fork to complete the efficient recovery of oil and water. 4. Drying and impurity removal process: The washed rapeseed oil enters a vacuum dryer to remove moisture and volatile substances, ultimately obtaining alkali-refined oil with the required purity, providing high-quality raw materials for subsequent decolorization and dewaxing processes.

[0031] The following effects can be achieved through this alkali refining process for rapeseed oil: To address the issue of temperature fluctuations affecting separation efficiency in traditional processes, an innovative 85-90℃ constant temperature process system was constructed. By using a plate heater, the temperature of key alkali refining processes is precisely locked at 85-90℃. This temperature range ensures that phosphoric acid and phospholipids, as well as sodium hydroxide and free fatty acids, react fully, improving reaction efficiency. It also maintains suitable fluidity of the soap residue, allowing it to form clear stratification with rapeseed oil. This creates optimal conditions for efficient separation in the desoap centrifuge, effectively solving the core problem of low separation efficiency caused by temperature runaway and significantly improving the oil-soap separation effect. Addressing the pain point of inefficient oil-water separation in washing wastewater, an innovative integrated process of buffer zone, stratification zone, and recovery zone is designed: wastewater first passes through a buffer zone to slow down the flow rate, preventing rapeseed oil emulsification caused by water flow impact and creating a stable environment for natural oil-water stratification; then, by adjusting the height difference between the oil separator and the baffle, the oil-water stratification interface is precisely matched to ensure thorough oil-water stratification; finally, automatic start-stop control enables separate recovery of oil and water, forming an integrated process of stable flow, stratification, and recovery, which significantly improves the oil-water separation rate, reduces the oil content of wastewater, and reduces the waste of rapeseed oil raw materials. To address the issues of parameter disconnect and poor coordination between stages in traditional processes, an innovative parameter linkage system for the entire process has been constructed: the addition ratio of phosphoric acid and sodium hydroxide is precisely controlled based on the phospholipid and free fatty acid content in crude oil; the residence time of materials in the extended reaction tank is optimized to ensure sufficient reaction while avoiding the secondary generation of impurities; the operating parameters of the desoap centrifuge are matched synchronously to reduce the load on the separation stage, and the entire process of reagent addition, reaction, and separation is coordinated and linked, significantly reducing the amount of soap residue and significantly improving the overall efficiency of alkali refining.

[0032] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A rapeseed oil alkali refining process based on a three-stage oil-water separator, characterized in that, Includes the following steps: S1: Precise Temperature Control for Acidification-Saponification: Crude rapeseed oil, after being filtered and purified, is first preheated by an energy-saving heat exchanger and then enters an acid mixer to be fully mixed with a food-grade phosphoric acid solution. It is then transferred to an alkali mixer to react with a sodium hydroxide solution and subsequently sent to a delayed reaction tank. The reaction temperature is stably controlled at 85-90℃ by a plate heater to ensure that free fatty acids are fully saponified and phospholipids are completely converted, laying the foundation for subsequent separation. S2: Oil-Soap Co-separation: Based on the characteristics of the oil-soap mixture after saponification, the operating parameters of the desoap centrifuge are optimized to enhance the oil-soap separation effect and make the oil-soap separation more thorough. The separated soap residue is temporarily stored and then transported outside the workshop for unified processing. S3: Water Washing - Oil-Water Recovery: After desoaping, the rapeseed oil enters the water washing centrifuge. The residual soap residue is removed by precisely controlling the water washing conditions. The wastewater generated by the water washing is introduced into the oil-water separation tank. The wastewater first enters the buffer zone to achieve a stable rise in the liquid level. By adjusting the height difference between the oil separation regulating device and the baffle, the specific gravity difference between rapeseed oil and water is used to achieve static center of gravity separation. Water slowly enters the still water zone from the bottom of the baffle, and the floating oil on the water surface slowly flows into the oil storage zone through the oil separation regulating device. Then, the water pump and oil pump are automatically started and stopped by the liquid level sensor or tuning fork to complete the efficient oil-water recovery. S4: Drying and impurity removal: The washed rapeseed oil enters a vacuum dryer to remove moisture and volatile substances, ultimately obtaining alkali-refined oil with the required purity, providing high-quality raw materials for subsequent decolorization and dewaxing processes.

2. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: In step S1, crude rapeseed oil is filtered to remove impurities using a plate and frame filter or a bag filter. The plate and frame filter or bag filter has a mesh size of 100-200 mesh, a filtration temperature of 60-70℃, and an energy-saving heat exchanger preheating temperature of 35-40℃. The food-grade phosphoric acid solution has a concentration of 85%, diluted to a 10% solution before use. The amount of phosphoric acid added is 0.05%-0.2% of the weight of the crude oil. The reaction time between the crude oil and phosphoric acid is 5-15 minutes, and the reaction temperature is 35-40℃.

3. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: The sodium hydroxide solution needs to be determined based on the acid value of crude oleic acid, as shown in the following formula: ,in The coefficient is the amount of sodium hydroxide required to neutralize each unit of acid value. The excess phosphoric acid is the additional phosphoric acid used for neutralization, ranging from 0.05% to 0.1%.

4. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: The reaction time in the extended reaction vessel is 30 minutes, allowing sodium hydroxide to fully neutralize with free fatty acids to form soap particles, which then continue to flocculate and grow.

5. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: In S2, the oil-soap mixture is fed into a desoap centrifuge with a drum speed of 6000-7500 rpm, an inlet pressure of 0.2-0.5 MPa, and a back pressure control of 0.1-0.3 MPa.

6. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: In S3, soft water is added to the washing centrifuge. Specifically, 10%-15% of the oil weight of soft water is added, with a hardness of <50 ppm. The water temperature is 5-10°C higher than the oil temperature to prevent emulsification.

7. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: The oil-water separator is divided into three areas: a buffer zone on the far left, a still water zone and an oil storage zone on the right. The oil storage zone is located above the still water zone. An energy dissipator is installed at the inlet of the buffer zone to ensure a smooth water flow.

8. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: In S4, the temperature of the vacuum dryer is set to 90-105℃ and the absolute pressure is ≤50 mbar.

9. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 1, characterized in that: In S1 and S3, an online near-infrared spectrometer is integrated to monitor the acid value, peroxide value, phospholipid content, and moisture content of crude oil in real time. The monitored data is then fed into a central multivariate process control model. This model aims to minimize the chemical consumption and the variance of key indicators of the final product, and dynamically solves for a set of optimal process parameters. A reinforcement learning-based intelligent decision engine is then introduced. This intelligent decision engine uses historical production data as a training set to continuously learn and optimize the setpoint of the central multivariate process control model.

10. The rapeseed oil alkali refining process based on a three-stage oil-water separator according to claim 9, characterized in that: The online near-infrared spectrometer is equipped with a high-temperature and pressure-resistant flow cell, with a detection band of 800-2500nm. An optical quantitative model established using a partial least squares algorithm can output real-time predicted values ​​for acid value, peroxide value, and phospholipid content, with a relative prediction deviation greater than 3.

0. The central multivariate process control model has a sampling interval of 30 seconds and a control cycle of 2 minutes. Its internal model is established based on historical step response data using a system identification method. The intelligent decision engine is trained using a near-end strategy optimization algorithm. Its state space includes the current online near-infrared spectrometer detection value and the average value of key process parameters over the past 3 hours. The action space consists of adjustment suggestions for the central multivariate process control model setpoint. The reward function is a negative weighted sum of chemical unit consumption and product quality deviation.

Images