Efficient recycling method for methanol synthesis tail gas
Through a method combining process simulation and machine learning, the operating parameters of methanol synthesis exhaust gas separation are optimized, and the effective gas is recovered using pressure swing adsorption-film combination separation technology, which solves the problems of fluctuations in the import flow rate and insufficient exhaust gas recovery during methanol synthesis, and improves the problem of fluctuations in the import flow rate and insufficient exhaust gas recovery. Methanol synthesis efficiency and product quality are reduced, and environmental pollution is reduced.
Patent Information
- Application Number
- CN202411214447.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-31
- Publication Date
- 2024-11-12
AI Technical Summary
In the prior art, the fluctuation of the import flow composition of methanol synthesis process and insufficient exhaust gas recovery in the methanol synthesis process leads to low methanol synthesis efficiency and unstable product quality.
Using process simulation and machine learning methods, the initial data set is established through Latin hypercube sampling, and the proxy model of methanol synthesis and exhaust gas separation processes are established, the exhaust gas separation operation parameters are optimized to improve methanol production, and combined with pressure swing adsorption-film combination separation Technology to recover H2 and CO effective gas.
The effective gas utilization rate and methanol production of the methanol synthesis process are improved, the methanol synthesis reaction process is stabilized, the environmental pollution caused by exhaust gas emissions is reduced, and resource conservation and cost reduction and efficiency are achieved.
Smart Images

Figure CN118942565A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of tail gas recovery, and in particular to a method for efficiently recovering and utilizing tail gas from methanol synthesis based on process simulation and machine learning. Background Art
[0002] Methanol is an important chemical raw material, a basic product and raw material for many carbon chemical industries, easy to store and transport, and an important energy carrier. Methanol synthesis is an important field in the energy industry and chemical industry. In view of the resource distribution pattern of my country's large coal reserves, lack of oil and gas, the coal-based synthetic methanol industry plays a vital role.
[0003] CO, CO for coal-based methanol synthesis 2 Add H 2 Methanol synthesis is a reversible reaction with many side reactions. Affected by many factors such as reaction temperature, pressure, components, and catalyst selectivity, the synthesis reaction will produce many by-products such as methane, ethane and other higher alkanes, ethanol, dimethyl ether, etc. while producing methanol. Specifically, the purified synthesis gas from the acid gas removal unit is mixed with hydrogen recovered from hydrogen, and then pressurized by the synthesis gas compressor and sent to the sulfur protector to remove sulfides. The circulating gas from the circulating gas compressor and the gas at the outlet of the sulfur protector are mixed and heated to carry out the methanol synthesis reaction. After the gas coming out of the reactor is cooled and condensed, it enters the high-pressure separator for gas-liquid separation. The separated crude methanol is sent to the low-pressure separator to enter the methanol distillation unit for refining. In order to prevent the accumulation of inert gas in the synthesis loop, part of the separated gas is returned as circulating gas to participate in the methanol synthesis reaction, and part is sent to the membrane separation unit as tail gas to recover hydrogen.
[0004] The exhaust gas contains H 2 , CO and other effective gases and a small amount of inert gas. The common tail gas recovery device uses membrane separation to recover hydrogen, and CO as an effective gas cannot be fully utilized. As the feed supplementary gas for the methanol synthesis unit, the tail gas recovery gas affects the import flow, hydrogen-carbon ratio and the final methanol product quality. When the coal sample is replaced upstream of the plant, the hydrogen-carbon ratio and flow of the synthesis gas imported for methanol synthesis will change, which will not meet the requirements of methanol synthesis. At this time, it can be adjusted by recycling gas. In order to further recover the effective gas, a pressure swing adsorption-membrane combined separation method is proposed. Pressure swing adsorption is an important industrial technology widely used in the separation of gas mixtures. It has a wide range of applications in gas purification and air separation. The common two-tower pressure swing adsorption process is mainly divided into six steps: adsorption, pressure reduction and equalization, exhaust, purge, pressure increase and equalization, and pressurization. It can be used for CO separation. Gas membrane separation technology is popular for its high operational flexibility, small footprint, stable operation, low energy consumption and quick results. The basic principle is to use a polymer film to select and "filter" the feed gas to achieve the purpose of separation. It can be used for H2 The tail gas is separated reasonably by a combined separation method, and the recycled effective gas is used to adjust the methanol synthesis inlet flow and composition to achieve a suitable hydrogen-carbon ratio, avoid synthesis gas fluctuations caused by changes in upstream coal samples, and realize dynamic control of methanol synthesis feed when upstream coal samples change.
[0005] The simulation process of methanol synthesis and tail gas separation is relatively complex, and it is difficult to express the precise relationship between the feed, tail gas and crude methanol product in the synthesis process. The relationship between these is affected by many factors and is difficult to display through simple mathematical formulas. In this case, the establishment of a proxy model through machine learning can not only avoid complex mechanism modeling, but also establish a "black box model" between feed and product indicators from the perspective of data. It can also be applied to the operation optimization of the methanol synthesis process, accelerate the convergence speed of the optimization process, and reduce the amount of calculations brought by a large number of process simulations. Therefore, applying machine learning technology to the chemical industry, by establishing a proxy model, predicting product conditions and optimal tail gas recovery conditions at a relatively low computational cost, can meet the development requirements of modern industry for intelligence, informatization and rapidity, and has become a major research focus.
[0006] The surrogate model is a mathematical model built based on data-driven, with a simple structure and fast calculation speed. Therefore, it is suitable for problems with complex mechanisms and difficult models to establish, and is also suitable for replacing complex mechanism models to evaluate some functions, reducing the number of calls or experiments of high-precision models, and providing a new solution to overcome the time-consuming problem of industrial simulation optimization. Among them, Kriging model Kriging (KRG), radial basis function (RBF), polynomial regression response surface (PRS), artificial neural network (ANN), etc. are all commonly used surrogate models. Current research shows that this technology plays a very important role in the optimization research of methanol synthesis process.
[0007] However, the existing technical solutions have the defects of fluctuating inlet flow composition and insufficient tail gas recovery in the methanol synthesis process. Summary of the invention
[0008] The technical problem to be solved by the present invention is to provide a method for efficiently recovering and utilizing methanol synthesis tail gas based on the combination of process simulation and machine learning in view of the above-mentioned defects in the prior art.
[0009] According to the present invention, a method for efficiently recovering and utilizing methanol synthesis tail gas is provided, comprising: The first step: set the initial data set according to Latin hypercube sampling; Step 2: Use the initial data set to establish a process simulation of methanol synthesis and tail gas separation; The third step: based on the data obtained from the process simulation, establish proxy models for the import and export data of methanol synthesis and tail gas separation respectively; Step 4: Using tail gas as imported supplementary gas for methanol synthesis, taking the optimal methanol yield as the goal, optimizing the tail gas separation process through iteration and using the proxy model, and obtaining the optimal tail gas separation operating parameters and the optimal methanol yield under the optimal tail gas separation operating parameters; Step 5: Obtaining a data set of optimal tail gas separation operating parameters and optimal methanol production that are optimized in sequence when the import data is changed; Step 6: Using the acquired data, a mathematical model of import data - optimal tail gas separation operating parameters - optimal methanol production is obtained.
[0010] Preferably, the method for efficiently recovering and utilizing methanol synthesis tail gas is a combined separation of pressure swing adsorption and membrane, which is used to simultaneously recover H 2 and CO effective gas combination, and the method also includes: in the process of efficient recovery and utilization of methanol synthesis tail gas, when the methanol synthesis import data changes, the tail gas separation operating parameters are adjusted based on the mathematical model of import data-optimal tail gas separation operating parameters-optimal methanol production to obtain suitable recycled gas to adjust the import data, thereby obtaining the optimal methanol production.
[0011] Preferably, the first step comprises: The inlet gas flow rate and composition of methanol synthesis are selected as the first feature, the inlet data of synthesis gas under different coal samples are monitored, and the fluctuation range of the inlet flow rate and composition is determined according to the monitoring results within a predetermined time period; the key operating parameters of tail gas separation are selected as the second feature, and the variation range of the key operating parameters is set; thereby, a sampling space is established, and in the sampling space, initial sample points that are evenly distributed in each dimension of the sampling space are selected through Latin hypercube sampling to establish an initial sample set.
[0012] Preferably, the decision variables affecting the establishment of the proxy model include inlet gas flow and composition, pressure, adsorption time, purge ratio and membrane area. Preferably, in the second step, the methanol synthesis process is simulated by using Aspen Plus software, the pressure swing adsorption process is simulated by using Aspen Adsorption software, and the membrane separation process is simulated by using Aspen Custom Modeler, and the process simulation of the corresponding process is established using the initial data set; Pressure swing adsorption is a dynamic adsorption process with a complex mechanism, mainly including: The component mass conservation and total mass conservation, for each component in the mixed gas, the mass convection term, the load accumulation in the gas phase and the accumulation in the adsorbent are all 0, and the total mass conservation during the adsorption process: , , in, v g represents the flow rate of the gas, c i represents the molar density of component i, Yes B represents the porosity of the bed, w i represents the load of component k adsorbed on the bed, ρ s represents the bulk density of the adsorbent, c g It represents the molar density of the mixed gas; The mass transfer rate model of the adsorption process adopts a linear model: , in, MTC i Indicates the components in the adsorption process i The mass transfer coefficient is a constant. w i * Indicates components i Balanced load; The adsorption isotherm describes the tendency of the components to adsorb onto the solid, that is, the amount of each component adsorbed on the solid at thermodynamic equilibrium. For mixed components, the extended Langmuir model is used to predict the adsorption isotherm: , in ,P represents the partial pressure of the component, IP 1 , IP 2 represents two adsorption isotherm parameters. The subscripts i and k represent two different components; The total momentum conservation equation uses the Ergun equation to deal with the relationship between pressure and flow rate in the bed: , in, P Indicates pressure, z represents the axial coordinate, Yes iv represents the pores between particles, μ is the dynamic viscosity, r p represents the particle radius, ρg is the gas phase molar density, ψ represents the shape factor of the particle; Energy conservation equation: Temperature is an important factor affecting the separation rate and separation effect of the pressure swing adsorption process, so the energy conservation of the system needs to be considered. Among them, the energy conservation equations of the gas phase and the solid phase are: , , in C vg represents the relative heat of gas at constant volume, HTC represents the gas-solid phase heat transfer coefficient, a p represents the specific surface area per unit volume of the bed, T g represents the gas phase temperature, T s represents the solid phase temperature, T amb Indicates the ambient temperature, H w represents the gas phase and wall heat transfer coefficient, D B represents the bed diameter; C ps represents the specific heat capacity of the adsorbent, C pa,i represents the specific heat of the adsorbed phase, ∆H i Indicates components i The adsorption heat of Membrane separation is mainly based on a gas diffusion model driven by the pressure difference between the feed side and the permeate side: , in, F represents the molar flow rate, y represents the mole fraction of a component, A represents the membrane area, D is the penetration rate, P Indicates pressure, corner mark P and R represent the permeate side and the retentate side, respectively.
[0013] Preferably, in the second step, in the process simulation of methanol synthesis, the gas passes through the sulfur protector, methanol synthesis reactor, compressor, heat exchanger and separator in sequence; in the process simulation of tail gas separation, the gas passes through double-tower pressure swing adsorption and membrane in sequence.
[0014] Preferably, the third step comprises: Each sample in the initial sample set is input into the simulation software in turn, and the process simulation is completed to obtain the simulation results; After the simulation is completed, the simulation results of the initial samples are used as training samples, and the initial proxy model of import data X-methanol production Y and tail gas data W, tail gas data W and tail gas separation key parameters OP-tail gas recovery Re is constructed through the machine learning method of artificial neural network: , , , Calculate the mean square error MSE and R of the initial proxy model on the test set 2 : , , Improving the R of the surrogate model by adding sample points 2 , in order to improve the model's fit and thus obtain the best surrogate model: , , .
[0015] Preferably, the fourth step comprises: Under the condition of fixed initial inlet flow rate composition, the key operating parameters of methanol synthesis tail gas separation are optimized to obtain the required tail gas separation requirements, the effective gas separated from the tail gas is used as the supplementary gas for the methanol synthesis inlet, the inlet flow rate composition is readjusted, and iterative optimization is performed repeatedly until the recycled gas flow rate and composition meet the convergence requirements, thereby obtaining the optimal operating parameters for tail gas separation and the optimal methanol production; , , in, i Indicates the number of iterative optimizations; The constraints include the upper and lower limits of the tail gas separation operating parameters, as well as the tail gas recovery constraints, namely: , , , , Preferably, the fourth step uses a genetic algorithm as an optimization algorithm, and finally obtains the optimal tail gas separation operating parameters and the optimal methanol production under the optimal tail gas separation operating parameters according to the sequence of initializing population - fitness evaluation - selection - crossover - mutation - updating population - iterative calculation.
[0016] Preferably, the sixth step comprises: Based on the samples collected in step 5, an initial proxy model of imported data-optimal tail gas separation operating parameters and an initial proxy model of imported data-optimal methanol production are constructed by using a machine learning method of an artificial neural network; On the basis of the initial proxy model, Latin hypercube sampling is used to increase the raw gas flow sample points. Through repeated iterations, the final import data-optimal tail gas separation operating parameter proxy model and import data-optimal methanol production proxy model are obtained: , . BRIEF DESCRIPTION OF THE DRAWINGS
[0017] A more complete understanding of the present invention and its attendant advantages and features will be more readily appreciated by reference to the following detailed description taken in conjunction with the accompanying drawings, in which:
[0018] Figure 1 A flow chart of a method for efficient recovery and utilization of methanol synthesis tail gas based on process simulation and machine learning according to a preferred embodiment of the present invention is schematically shown.
[0019] Figure 2 The data processing circuit diagram of the method for efficient recovery and utilization of methanol synthesis tail gas based on process simulation and machine learning according to a preferred embodiment of the present invention is schematically shown.
[0020] Figure 3 It is a schematic diagram of the methanol synthesis and pressure swing adsorption-membrane combined separation process according to a preferred embodiment of the present invention.
[0021] It should be noted that the drawings are used to illustrate the present invention, rather than to limit the present invention. Note that the drawings showing the structures may not be drawn to scale. In addition, in the drawings, the same or similar elements are marked with the same or similar reference numerals. DETAILED DESCRIPTION
[0022] In order to make the contents of the present invention clearer and easier to understand, the contents of the present invention are described in detail below in conjunction with specific embodiments and drawings.
[0023] In the methanol synthesis process, the composition and flow of the imported synthesis gas will fluctuate with the changes in the upstream coal sample, which will inevitably deviate from the optimal operating point and fail to convert as much raw gas into methanol products as possible, resulting in a waste of resources. In this case, if the best tail gas recovery and supplement situation can be selected in a targeted manner, the highest instantaneous methanol production will inevitably be obtained at every moment, and the total methanol production will inevitably be improved to a certain extent. In theory, this waste of resources can be completely achieved by appropriately adjusting the tail gas recovery situation in the methanol synthesis process to adjust the change in the hydrogen-carbon ratio of methanol import.
[0024] Therefore, it is necessary to re-optimize according to the changes in import data to select the tail gas recovery situation with the highest methanol production. This requires that the key parameters that affect the separation and recovery of tail gas in the methanol synthesis process must first be mastered. For different feed synthesis gas conditions, the optimal tail gas separation operating parameters with the highest methanol production under the corresponding import flow composition are found through appropriate optimization algorithms, thereby stabilizing the methanol synthesis process. Therefore, the present invention proposes a method for efficient recovery and utilization of methanol synthesis tail gas by combining process simulation with machine learning. The import and export samples of the methanol synthesis process are obtained through process simulation technology. According to the samples obtained by process simulation, a proxy model is established and screened through a machine learning method to select the best proxy model of the methanol synthesis device. Based on the proxy model, the problem of insufficient tail gas recovery in the existing device is taken into account, and a method for simultaneously recovering H 2 The combined separation device of tail gas and CO effective gas obtains the optimal tail gas separation operating parameters under different feed conditions through intelligent optimization algorithm, thereby improving the effective gas utilization and methanol production in the methanol synthesis process, avoiding the fluctuation of synthesis gas flow composition caused by changes in upstream coal samples, and realizing dynamic control of methanol synthesis feed under the condition of changes in upstream coal samples.
[0025] The method of the present invention focuses on the impact of efficient recovery and utilization of tail gas in the methanol synthesis process on the overall production process, combines artificial intelligence technology with the traditional chemical industry, is conducive to the refined and intelligent development of the methanol synthesis industry, and improves the mass and energy utilization efficiency of effective gas resources in the methanol synthesis process. It is of great significance to increase methanol production, reduce environmental pollution caused by tail gas emissions, and reduce costs and increase efficiency.
[0026] Figure 1 A flow chart of a method for efficient recovery and utilization of methanol synthesis tail gas based on process simulation and machine learning according to a preferred embodiment of the present invention is schematically shown. Figure 2 The data processing circuit diagram of the method for efficient recovery and utilization of methanol synthesis tail gas based on process simulation and machine learning according to a preferred embodiment of the present invention is schematically shown.
[0027] For example, the method for efficiently recovering and utilizing methanol synthesis tail gas based on process simulation and machine learning can be used to simultaneously recover H 2 Combination of effective gas and CO.
[0028] like Figure 1 and Figure 2 As shown, the method for efficiently recovering and utilizing methanol synthesis tail gas based on process simulation and machine learning according to a preferred embodiment of the present invention includes: The first step S1: According to Latin hypercube sampling (LHS), an initial data set is set.
[0029] The inlet gas flow rate and composition of methanol synthesis are selected as the feature X, the inlet data of synthesis gas under different coal samples are monitored, and the fluctuation range of the inlet flow rate and composition is determined according to the monitoring results within a predetermined time period. The key operating parameters of methanol synthesis tail gas separation are selected as the second feature OP, for example, they may include the inlet gas flow rate and composition, pressure (P), adsorption time (t), purge ratio (PFR) and membrane area (A). The key operating parameter variation range is set on an industrial basis, for example. Thus, a sampling space is established.
[0030] In the sampling space, by using Latin hypercube sampling, initial sample points that are evenly distributed in each dimension in the sampling space are selected to establish an initial sample set U0.
[0031] Through Latin hypercube sampling, the selected samples can satisfy the requirement of uniform distribution in all dimensions in space. This method can make the distribution of samples more uniform and reduce the number of sampling points, thereby speeding up the calculation.
[0032] The second step S2: Use Aspen Plus software to simulate the methanol synthesis process, use Aspen Adsorption software to simulate the pressure swing adsorption process, use Aspen Custom Modeler to simulate the membrane separation process, and use the initial data set to establish the process simulation of the corresponding process. The schematic diagram is as follows Figure 3 shown.
[0033] Among them, the physical property method can select the Peng-Rob method, which calculates all thermodynamic properties through the Peng Robinson cubic state equation.
[0034] For example, in the process simulation of methanol synthesis and tail gas separation, the gas passes through the sulfur protector, methanol synthesis reactor, compressor, heat exchanger and separator in sequence. 2S and COS are removed; the methanol synthesis reactor uses the plug flow reactor Rplug module for simulation and is calculated based on kinetic data. The final product mainly includes crude methanol and some by-products such as alkanes, alcohols, and ethers. The tail gas mainly includes H 2 ,CO,N 2 wait.
[0035] In the process simulation of tail gas separation, the gas passes through the double tower pressure swing adsorption and membrane in turn, and the tail gas under each set of inlet data is separated by pressure swing adsorption-membrane combination to obtain effective gas H 2 Select appropriate adsorbents and membranes, first separate CO by pressure swing adsorption, the simulation software used is Aspen Adsorption, and then separate H 2 , which can be simulated in Aspencustom modeler, so that the effective gas can be efficiently recovered.
[0036] Pressure swing adsorption is a dynamic adsorption process with a complicated mechanism, mainly including: The component mass conservation and total mass conservation, for each component in the mixed gas, the mass convection term, the load accumulation in the gas phase and the accumulation in the adsorbent are summed to 0, and the total mass conservation during the adsorption process, as shown in formulas (1) and (2): (1), (2), in, v g represents the flow rate of the gas, c i represents the molar density of component i, Yes B represents the porosity of the bed, w i represents the load of component k adsorbed on the bed, ρ s represents the bulk density of the adsorbent, c g It represents the molar density of the mixed gas; The mass transfer rate model of the adsorption process adopts a linear model, as shown in formula (3): (3), in, MTC i Indicates the components in the adsorption process i The mass transfer coefficient is a constant. w i * Indicates components i Balanced load; The adsorption isotherm describes the tendency of components to adsorb onto solids, that is, the amount of each component adsorbed on the solid at thermodynamic equilibrium. For mixed components, the extended Langmuir model is used to predict the adsorption isotherm, as shown in formula (4): (4), in ,P represents the partial pressure of the component, IP 1 , IP 2 represents two adsorption isotherm parameters. The subscripts i and k represent two different components; The total momentum conservation equation uses the Ergun equation to deal with the relationship between pressure and flow rate in the bed as shown in formula (5): (5), in, P Indicates pressure, z represents the axial coordinate, Yes iv represents the pores between particles, μ is the dynamic viscosity, r p represents the particle radius, ρ g is the gas phase molar density, ψ represents the shape factor of the particle; Energy conservation equation: Temperature is an important factor affecting the separation rate and separation effect of the pressure swing adsorption process, so the energy conservation of the system needs to be considered. The energy conservation equations of the gas phase and the solid phase are shown in formulas (6) and (7): (6), (7), in C vg represents the relative heat of gas at constant volume, HTC represents the gas-solid phase heat transfer coefficient, a p represents the specific surface area per unit volume of the bed, T g represents the gas phase temperature, T s represents the solid phase temperature, T amb Indicates the ambient temperature, H w represents the gas phase and wall heat transfer coefficient, D B represents the bed diameter; C ps represents the specific heat capacity of the adsorbent, C pa,irepresents the specific heat of the adsorbed phase, ∆H i Indicates components i The adsorption heat of Membrane separation is mainly based on the gas diffusion model driven by the pressure difference between the feed side and the permeate side, as shown in formula (8): (8), in, F represents the molar flow rate, y represents the mole fraction of a component, A represents the membrane area, D is the penetration rate, P Indicates pressure, corner mark P and R represent the permeate side and the retentate side, respectively.
[0037] The third step S3: Based on the data obtained from the process simulation, a proxy model of import and export data is established.
[0038] Preferably, first, each sample in the initial sample set U0 is sequentially input into the simulation software to complete the process simulation of methanol synthesis and tail gas separation to obtain the simulation results. For example, through Matlab and Aspen Plus communication, the operating points in the sample space are automatically changed sequentially, and the corresponding simulation results are simulated, calculated, output and recorded, including methanol production Y, tail gas parameters W and tail gas recovery Re.
[0039] After completing the simulation, the simulation results of the initial samples are used as training samples, and the initial proxy models of import data X-methanol production Y and tail gas data W, tail gas data W and tail gas separation key parameters OP-tail gas recovery conditions Re are constructed through the machine learning method of artificial neural network (ANN).
[0040] For example, the form of the proxy model is shown in formulas (9)-(11), where subscript 0 represents the initial proxy model: (9), (10), (11).
[0041] For the above proxy models, calculate the mean square error MSE and R of the model on the test set 2 , the calculation formula is shown in formula (12) and (13): (12), (13).
[0042] The calculated mean square error (MSE) represents the deviation between the estimated value and the actual value of the test sample, while R 2 It can be used to evaluate the goodness of fit of the surrogate model. On this basis, the surrogate model can be further improved by adding sample points to improve the fit of the model. 2 , and finally the best proxy model is selected, as shown in formulas (14)-(16): (14), (15), (16).
[0043] The fourth step S4: using the tail gas as the inlet supplementary gas for methanol synthesis, taking the optimal methanol production as the goal, and optimizing the tail gas separation process through iteration using the proxy model.
[0044] For example, under the condition of fixed initial inlet flow composition, the key operating parameter OP of methanol synthesis tail gas separation is optimized to obtain the required tail gas separation requirements, the recycled gas separated from the tail gas is used as the supplementary gas for the methanol synthesis inlet, the inlet flow composition is readjusted, and iterative optimization is performed repeatedly until the recycled gas flow and composition meet the convergence requirements, thereby obtaining the optimal operating parameters for tail gas separation and the optimal methanol production; (17), (18), in, i Indicates the number of iterative optimizations; The constraints include the upper and lower limits of the tail gas separation operating parameters, as well as the tail gas recovery constraints, namely: , , , , For the proxy model, its gradient information is difficult to calculate. Therefore, it is difficult for traditional optimization algorithms, such as the steepest descent method, quasi-Newton method, conjugate gradient method, etc., to obtain the global optimal solution. Evolutionary algorithms that draw on the idea of biological evolution, such as genetic algorithms and particle swarm optimization algorithms, have good robustness and wide applicability due to their self-organization, self-adaptation, self-learning, parallel search and other characteristics. Therefore, the present invention uses genetic algorithms as optimization algorithms to solve the optimization problem. According to the sequence of initializing population-fitness evaluation-selection-crossover-mutation-updating population-iterative calculation, the optimal exhaust separation operating parameter OP is finally found. optimalAnd the optimal methanol yield Y under this operating parameter optimal .
[0045] The fifth step S5: obtaining a data set of optimal tail gas separation operating parameters and optimal methanol production that are optimized in sequence when the import data is changed.
[0046] For example, the imported raw gas is supplemented according to the optimal tail gas recovery obtained in the fourth step, and then the new import flow composition sample set X after tail gas supplementation is constructed through the LHS sampling method, which is input into the proxy model in turn and optimized in the fourth step. Matlab is used to automatically change the imported synthesis gas flow composition, so as to obtain the sample data of the optimal methanol production and the optimal tail gas separation operating parameters under the import flow composition sample set after tail gas supplementation.
[0047] Step 6 S6: Using the data set obtained in step 5, a mathematical model of import data-optimal tail gas separation operating parameters-optimal methanol production is obtained. That is, the mathematical model reflects the relationship between import data and the optimal tail gas separation operating parameters and the optimal methanol production.
[0048] According to the samples collected in the data set obtained in the fifth step, the initial proxy model of the imported data-optimal tail gas separation operating parameters and the initial proxy model of the imported data-optimal product yield are constructed through the machine learning method of artificial neural network (ANN). On the basis of the initial proxy model, the sampling method mentioned above is used to increase the raw gas flow sample points, and iterate repeatedly to finally obtain the proxy model of the imported data-optimal tail gas separation operating parameters and the proxy model of the imported data-optimal methanol production. As shown in formulas (19) and (20): (19), (20).
[0049] Therefore, in the process of efficient recovery and utilization of methanol synthesis tail gas, when the methanol synthesis import data changes, the tail gas separation operating parameters are adjusted based on the mathematical model of import data-optimal tail gas separation operating parameters-optimal methanol production to adjust the import data, thereby obtaining the optimal methanol production.
[0050] Specifically, in the actual production process, if the upstream coal sample changes, that is, the methanol synthesis import data changes, the combined separation method combined with machine learning can adjust the tail gas separation operating parameters in a targeted manner and select the best tail gas recovery and supplement situation. Then, the highest instantaneous methanol production will be obtained at every moment. Therefore, the total methanol production will inevitably be improved to a certain extent.
[0051] The present invention proposes a pressure swing adsorption-membrane combined separation method to efficiently recover methanol synthesis tail gas, combines process simulation with machine learning, and is applied to tail gas recovery in the methanol synthesis process, which helps to adjust the import composition, stabilize the methanol synthesis reaction process, and can achieve the improvement of effective gas utilization and methanol production in the methanol synthesis process. This method is conducive to the refined and intelligent development of the methanol chemical industry, improves the mass and energy utilization efficiency of effective gas resources in the methanol synthesis process, and is of great significance to reducing environmental pollution caused by tail gas emissions, achieving resource conservation, and reducing costs and increasing efficiency.
[0052] Compared with the existing technologies and methods, the present invention has at least the following innovative effects:
[0053] 1. The method for efficiently recovering and utilizing methanol synthesis tail gas proposed in the present invention takes into account the problem of insufficient tail gas recovery in existing devices and designs a method for simultaneously recovering H 2 The pressure swing adsorption-membrane combined separation device for the effective gas of CO can improve the effective gas utilization rate and methanol output in the methanol synthesis process, reduce the environmental pollution caused by tail gas emissions, achieve resource conservation, reduce costs and increase efficiency.
[0054] 2. The method for efficiently recovering and utilizing methanol synthesis tail gas proposed in the present invention takes into account the changes in the methanol synthesis import flow and composition caused by changes in upstream coal samples, as well as the impact on the quality of methanol products, and uses the effective gas recovered from the tail gas to supplement the imported synthesis gas and adjust the hydrogen-carbon ratio to obtain the optimal methanol yield. Finally, a mathematical model of import data-optimal tail gas separation operating parameters-optimal methanol yield in the methanol synthesis process is obtained, which plays a very important role in the intelligent control of the methanol synthesis process, helps to realize the digital twin of the methanol synthesis process, and realizes the digital and intelligent transformation of methanol chemical enterprises.
[0055] 3. The present invention combines combined separation, process simulation and machine learning and applies them to tail gas recovery in the methanol synthesis process, which helps to adjust the inlet composition and stabilize the methanol synthesis reaction process. This is a new method for efficient recovery and utilization of methanol synthesis tail gas.
[0056] It should be noted that, unless otherwise specified, the terms "first", "second", "third", etc. in the specification are only used to distinguish the various components, elements, steps, etc. in the specification, and are not used to indicate the logical relationship or sequential relationship between the various components, elements, steps, etc.
[0057] It is to be understood that, although the present invention has been disclosed as a preferred embodiment, the above embodiment is not intended to limit the present invention. For any person skilled in the art, without departing from the scope of the technical solution of the present invention, the technical content disclosed above can be used to make many possible changes and modifications to the technical solution of the present invention, or modified into equivalent embodiments of equivalent changes. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of protection of the technical solution of the present invention.
Claims
1. A method for efficiently recovering and utilizing methanol synthesis tail gas, Features include: The first step: according to Latin hypercube sampling, an initial data set is set; wherein, the methanol synthesis inlet gas flow rate and composition are selected as the first feature, the synthesis gas inlet data under different coal samples are monitored, and the inlet flow rate and composition fluctuation range are determined according to the monitoring results within a predetermined time period; the key operating parameters of tail gas separation are selected as the second feature, and the key operating parameter variation range is set; thereby, a sampling space is established, and in the sampling space, initial sample points that are evenly distributed in each dimension in the sampling space are selected by Latin hypercube sampling to establish an initial sample set; The second step: using the initial data set to establish the process simulation of methanol synthesis and tail gas separation respectively; in the process simulation of methanol synthesis, the gas passes through the sulfur protector, methanol synthesis reactor, compressor, heat exchanger and separator in sequence; in the process simulation of tail gas separation, the gas passes through the double tower pressure swing adsorption and membrane in sequence; The third step: based on the data obtained from the process simulation, establish proxy models for the import and export data of methanol synthesis and tail gas separation respectively; the decision variables affecting the establishment of the methanol synthesis proxy model include the inlet gas flow rate and composition; the decision variables affecting the tail gas separation proxy model include the tail gas composition, pressure, adsorption time, purge ratio and membrane area; Step 4: Using tail gas as imported supplementary gas for methanol synthesis, taking the optimal methanol yield as the goal, optimizing the tail gas separation process through iteration and using the proxy model, and obtaining the optimal tail gas separation operating parameters and the optimal methanol yield under the optimal tail gas separation operating parameters; Step 5: Obtaining a data set of optimal tail gas separation operating parameters and optimal methanol production that are optimized in sequence when the import data is changed; Step 6: Using the acquired data set, a mathematical model of import data - optimal tail gas separation operating parameters - optimal methanol production is obtained; The method for efficiently recovering and utilizing methanol synthesis tail gas is a combined separation of pressure swing adsorption and membrane for simultaneously recovering H 2 and CO effective gas combination, and the method also includes: in the process of efficient recovery and utilization of methanol synthesis tail gas, when the methanol synthesis import data changes, the tail gas separation operating parameters are adjusted based on the mathematical model of import data-optimal tail gas separation operating parameters-optimal methanol production to obtain suitable recycled gas to adjust the import data, thereby obtaining the optimal methanol production.
2. The method for efficiently recovering and utilizing methanol synthesis tail gas according to claim 1, It is characterized in that In the second step, the methanol synthesis process was simulated using Aspen Plus software, the pressure swing adsorption process was simulated using Aspen Adsorption software, and the membrane separation process was simulated using Aspen Custom Modeler, and the process simulation of the corresponding process was established using the initial data set; Pressure swing adsorption is a dynamic adsorption process with a complex mechanism, mainly including: The component mass conservation and total mass conservation, for each component in the mixed gas, the mass convection term, the load accumulation in the gas phase and the accumulation in the adsorbent are all 0, and the total mass conservation during the adsorption process: , , in, v g represents the flow rate of the gas, c i represents the molar density of component i, Ɛ B represents the porosity of the bed, w i represents the load of component k adsorbed on the bed, ρ s represents the bulk density of the adsorbent, c g It represents the molar density of the mixed gas; The mass transfer rate model of the adsorption process adopts a linear model: , in, MTC i Indicates the components in the adsorption process i The mass transfer coefficient is a constant, w i * Indicates components i Balanced load; The adsorption isotherm describes the tendency of the components to adsorb onto the solid, that is, the amount of each component adsorbed on the solid at thermodynamic equilibrium; for mixed components, the extended Langmuir model is used to predict the adsorption isotherm: , in ,P represents the partial pressure of the component, IP 1 , IP 2 represents two adsorption isotherm parameters, and the subscripts i and k represent two different components; The total momentum conservation equation uses the Ergun equation to deal with the relationship between pressure and flow rate in the bed: , in, P Indicates pressure, z represents the axial coordinate, Ɛ iv represents the pores between particles, μ is the dynamic viscosity, r p represents the particle radius, ρ g is the gas phase molar density, ψ represents the shape factor of the particle; Energy conservation equation: Temperature is an important factor affecting the separation rate and separation effect of the pressure swing adsorption process, so the energy conservation of the system needs to be considered; the energy conservation equations of the gas phase and the solid phase are: , , in C vg represents the relative heat of gas at constant volume, HTC represents the gas-solid phase heat transfer coefficient, a p represents the specific surface area per unit volume of the bed, T g represents the gas phase temperature, T s represents the solid phase temperature, T amb Indicates the ambient temperature, H w represents the heat transfer coefficient between the gas phase and the wall, D B represents the bed diameter; C ps represents the specific heat capacity of the adsorbent, C pa,i represents the specific heat of the adsorbed phase, ∆H i Indicates components i The adsorption heat of Membrane separation is mainly based on a gas diffusion model driven by the pressure difference between the feed side and the permeate side: , in, F represents the molar flow rate, y represents the mole fraction of a component, A represents the membrane area, D is the penetration rate, P Indicates pressure, corner mark P and R represent the permeate side and the retentate side, respectively.
3. The method for efficiently recovering and utilizing methanol synthesis tail gas according to claim 1 or 2, It is characterized in that The third step includes: Each sample in the initial sample set is input into the simulation software in turn, and the process simulation is completed to obtain the simulation results; After the simulation is completed, the simulation results of the initial samples are used as training samples, and the initial proxy model of import data X-methanol production Y and tail gas data W, tail gas data W and tail gas separation key parameters OP-tail gas recovery situation Re is constructed through the machine learning method of artificial neural network: , , Improving the R of the surrogate model by adding sample points 2 , in order to improve the model's fit and thus obtain the best surrogate model: , , 。 4. The method for efficiently recovering and utilizing methanol synthesis tail gas according to claim 1 or 2, It is characterized in that The fourth step includes: Under the condition of fixed initial inlet flow rate composition, the key operating parameters of methanol synthesis tail gas separation are optimized to obtain the required tail gas separation requirements, the recycled gas obtained from tail gas separation is used as the supplementary gas for methanol synthesis inlet, the inlet flow rate composition is readjusted, and iterative optimization is performed repeatedly until the recycled gas flow rate and composition meet the convergence requirements, thereby obtaining the optimal operating parameters for tail gas separation and the optimal methanol production; , , in, i Indicates the number of iterative optimizations; The constraints include the upper and lower limits of the tail gas separation operating parameters, as well as the tail gas recovery constraints, namely: , , , 。 5. The method for efficiently recovering and utilizing methanol synthesis tail gas according to claim 1 or 2, It is characterized in that The fourth step uses a genetic algorithm as an optimization algorithm, and finally obtains the optimal tail gas separation operating parameters and the optimal methanol production under the optimal tail gas separation operating parameters according to the sequence of initializing the population - fitness evaluation - selection - crossover - mutation - updating the population - iterative calculation.
6. The method for efficiently recovering and utilizing methanol synthesis tail gas according to claim 1 or 2, It is characterized in that The sixth step includes: Based on the samples collected in step 5, an initial proxy model of imported data-optimal tail gas separation operating parameters and an initial proxy model of imported data-optimal methanol production are constructed by using a machine learning method of an artificial neural network; On the basis of the initial proxy model, Latin hypercube sampling is used to increase the raw gas flow sample points. Through repeated iterations, the final import data-optimal tail gas separation operation parameter proxy model and import data-optimal methanol production proxy model are obtained: , 。
