A method and device for optimizing the structure of a cracking furnace burner

Abstract

The present application relates to the technical field of ethylene production, and more particularly to a method and device for optimizing the structure of a burner of a cracking furnace.The method for optimizing the structure of the burner of the cracking furnace comprises the following steps: S1, collecting parameter information to establish a cracking furnace model and determine the value range of each parameter of the burner structure; S2, selecting a plurality of parameters of the burner structure to perform Latin hypercube sampling, and establishing a sample set through simulation; S3, establishing a proxy model of the NO content at the outlet of the cracking furnace with respect to the parameters of the burner structure; and S4, calling the proxy model by using an optimization algorithm, optimizing the parameters of the burner structure of the cracking furnace in a certain three-dimensional decision space with a preset objective function, and obtaining the optimal parameters of the burner structure.The present application replaces the time-consuming CFD simulation process by using a proxy model, and further reduces the emission of nitrogen oxides by using a particle swarm optimization algorithm to find the optimal parameters of the burner structure.

Description

Technical Field

[0001] This invention relates to the field of ethylene production technology, and more specifically, to a method and apparatus for optimizing the burner structure of an ethylene cracking furnace burner to reduce nitrogen oxide emissions. Background Technology

[0002] The ethylene industry is the leading sector in the petrochemical industry, and ethylene production is a key indicator of a country's petrochemical development level. The cracking furnace plays a crucial role in ethylene plants; the burners at the bottom of the furnace release heat to provide the necessary heat for the cracking reaction in the furnace tubes at the top.

[0003] However, the gases produced by combustion are directly released into the air through the outlet of the pyrolysis furnace, often causing great pollution to the environment. Therefore, reducing its nitrogen oxide (NOx) emissions is of great significance for environmental governance.

[0004] Currently, there are two main optimization methods for NOx emission reduction at the cracking furnace outlet:

[0005] One approach is to add a tail gas treatment device to the pyrolysis furnace to discharge the treated gas into the air. However, this method requires an additional tail gas treatment device, which increases the cost.

[0006] Another approach is to reduce NOx formation at its source by designing and optimizing the structure of the pyrolysis furnace, such as the burner nozzles. Compared to the first method, the second method is more economical and reduces additional costs.

[0007] Low-NOx burners typically utilize staged combustion technology, primarily divided into air-staged and fuel-staged combustion. Air-staged combustion involves diverting a portion of the air from the first combustion zone into the second combustion zone, creating a low-oxygen combustion environment. This low-oxygen environment and lower temperature help reduce NOx formation. Fuel-staged combustion, on the other hand, introduces all the combustion air and primary fuel into the first combustion zone. The combustion of the primary fuel is completed in a large amount of excess air. This excess air dilutes the fuel and cools the flame temperature, thereby reducing NOx formation. The excess air is then introduced into the second combustion zone for combustion.

[0008] Since the parameters of the burner nozzle and other structural components have a significant impact on the combustion process, there is an urgent need for a method to optimize the burner structure in order to further reduce NOx emissions. Summary of the Invention

[0009] The purpose of this invention is to provide a method and apparatus for optimizing the structure of a pyrolysis furnace burner, thereby solving the problem that the structure of the pyrolysis furnace burner in the prior art has not been optimized for NOx emission reduction.

[0010] To achieve the above objectives, the present invention provides a method for optimizing the structure of a pyrolysis furnace burner, comprising the following steps:

[0011] Step S1: Collect parameter information to establish a pyrolysis furnace model and determine the value range of each parameter of the burner nozzle structure;

[0012] Step S2: Select several parameters of the burner nozzle structure and perform Latin hypercube sampling to establish a sample set through simulation;

[0013] Step S3: Establish a proxy model for the NO content at the pyrolysis furnace outlet with respect to the burner nozzle structural parameters;

[0014] Step S4: Use the optimization algorithm to call the proxy model and optimize the burner structure parameters of the pyrolysis furnace within a certain three-dimensional decision space with a preset objective function to obtain the optimal parameters of the burner structure.

[0015] In one embodiment, the parameter information collected in step S1 includes:

[0016] Information on the operating conditions, inlet fuel, oxidant content, burner and burner nozzle dimensions of industrial ethylene cracking furnaces.

[0017] In one embodiment, step S1 further includes:

[0018] The results of a single CFD simulation experiment were compared with the parameters collected under actual working conditions, and the best combustion model, turbulence model, radiation model, and NOx model were selected from the pyrolysis furnace model.

[0019] In one embodiment, the combustion model selected in step S1 is a non-premixed combustion model, the turbulence model is a standard k-ε model, the radiation model is a discrete ordinate model, and the NOx model is a post-processing model, mainly considering the mechanism of thermal NOx.

[0020] In one embodiment, several parameters of the burner nozzle structure in step S2 further include: the height difference between the primary and secondary burners, the radius of the primary burner, and the radius of the secondary burner.

[0021] In one embodiment, step S2 further includes:

[0022] A sample set was formed by Latin hypercube sampling of several parameters of the burner nozzle structure.

[0023] For each sample in the sample set, geometric modeling, fluid domain creation, region division, mesh generation, boundary condition setting, and simulation calculation are performed. The NO content under each set of structural parameters is recorded to establish a real sample set.

[0024] In one embodiment, the surrogate model in step S3 includes a radial basis function neural network surrogate model.

[0025] In one embodiment, step S3 further includes:

[0026] The real sample set is divided into a training set and a test set;

[0027] The input parameters of the burner nozzle structure are normalized.

[0028] Use a surrogate model to train the samples;

[0029] The NO content under different burner nozzle structural parameters was predicted using the finally established surrogate model.

[0030] In one embodiment, the optimization algorithm in step S4 includes a particle swarm optimization algorithm;

[0031] In step S4, the preset objective function is the weighted mole fraction of NO at the outlet of the pyrolysis furnace.

[0032] In one embodiment, step S4 is followed by:

[0033] After obtaining the optimal parameters of the pyrolysis furnace burner structure, CFD simulation was used to obtain the true value of the NO weighted mole fraction of the pyrolysis furnace burner structure with the optimal parameters, thus verifying the feasibility of the surrogate model and optimization algorithm.

[0034] To achieve the above objectives, the present invention provides a device for optimizing the structure of a pyrolysis furnace burner, comprising:

[0035] Memory is used to store instructions that can be executed by the processor;

[0036] A processor for executing the instructions to implement the method as described in any of the preceding descriptions.

[0037] To achieve the above objectives, the present invention provides a computer-readable medium having computer instructions stored thereon, wherein when the computer instructions are executed by a processor, the method described in any of the preceding claims is performed.

[0038] This invention proposes a method and apparatus for optimizing the structure of a pyrolysis furnace burner. Utilizing CFD simulation technology, the method first performs a three-dimensional model of the ethylene pyrolysis furnace, furnace tubes, burner, and burner. Then, the fluid domain model of the pyrolysis furnace is meshed, and boundary conditions are calculated. Simulation calculations are performed in ANSYS Fluent to obtain samples for individual structural parameters. By changing the structural parameters and repeating the same steps, an initial sample set can be established. A surrogate model is then built using this initial sample set, and an optimization algorithm is combined to optimize the burner structural parameters within the search domain of the structural parameters, thereby achieving nitrogen oxide emission reduction. Attached Figure Description

[0039] The above-described and other features, properties, and advantages of the present invention will become more apparent from the following description taken in conjunction with the accompanying drawings and embodiments, in which the same reference numerals consistently denote the same features.

[0040] in:

[0041] Figure 1 A flowchart of a method for optimizing the structure of a pyrolysis furnace burner according to an embodiment of the present invention is disclosed;

[0042] Figure 2 A 1 / 8 scale schematic diagram of a pyrolysis furnace according to an embodiment of the present invention is shown;

[0043] Figure 3a A schematic diagram of a furnace tube according to an embodiment of the present invention is disclosed;

[0044] Figure 3b A schematic diagram of a burner according to an embodiment of the present invention is shown;

[0045] Figure 3c A schematic diagram of a main burner according to an embodiment of the present invention is disclosed;

[0046] Figure 3d A schematic diagram of a secondary burner according to an embodiment of the present invention is disclosed;

[0047] Figure 4 A schematic diagram of the cracking furnace area division and grid according to an embodiment of the present invention is disclosed;

[0048] Figure 5 A flowchart of a particle swarm optimization algorithm according to an embodiment of the present invention is disclosed;

[0049] Figure 6 A block diagram of a pyrolysis furnace burner structure optimization device according to an embodiment of the present invention is disclosed.

[0050] The meanings of the labels in the figures are as follows:

[0051] 1. Furnace tube;

[0052] 2. Burners;

[0053] 21 main burners;

[0054] 22 secondary burners;

[0055] 23. Air inlet. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0057] This invention relates to a method for optimizing the burner structure of a pyrolysis furnace to reduce nitrogen oxide emissions. The method utilizes CFD (Computational Fluid Dynamics) simulations to establish a sample set of burner structures for different burners. By employing surrogate models and optimization algorithms, the burner structural parameters are optimized.

[0058] Figure 1 A flowchart of a method for optimizing the structure of a pyrolysis furnace burner according to an embodiment of the present invention is disclosed, as follows: Figure 1 As shown, the method for optimizing the burner structure of a pyrolysis furnace proposed in this invention includes the following steps:

[0059] Step S1: Collect parameter information to establish a pyrolysis furnace model and determine the value range of each parameter of the burner nozzle structure;

[0060] Step S2: Select several parameters of the burner nozzle structure and perform Latin hypercube sampling to establish a sample set through simulation;

[0061] Step S3: Establish a proxy model for the NO content at the pyrolysis furnace outlet with respect to the burner nozzle structural parameters;

[0062] Step S4: Use the optimization algorithm to call the proxy model and optimize the burner structure parameters of the pyrolysis furnace within a certain three-dimensional decision space with a preset objective function to obtain the optimal parameters of the burner structure.

[0063] Based on fuel-staged low-NOx burners, this invention provides an optimization method for the burner structure of an ethylene cracking furnace burner to further reduce nitrogen oxide emissions. By using computational fluid dynamics, the burner structure of the low-NOx burner is further optimized. By establishing a surrogate model and optimization algorithm, the optimal burner structure parameters are obtained, further reducing nitrogen oxide emissions.

[0064] Specifically, a sample set was established by performing Latin hypercube sampling on three dimensional parameters of the burner nozzle structure (main burner diameter, secondary burner diameter, and height difference between the main and secondary burners) and using CFD simulation. A surrogate model was then built based on this sample set. Finally, an optimization algorithm was used, with the NO weighted mole fraction at the pyrolysis furnace outlet as the objective function, to optimize the burner structure within a certain three-dimensional decision space, finding the optimal parameters for the burner structure, thereby further reducing NOx emissions.

[0065] These steps will be described in detail below. It should be understood that, within the scope of this invention, the above-described technical features of this invention and the technical features specifically described below (such as in the embodiments) can be combined and related to each other to form preferred technical solutions.

[0066] Step S1: Collect parameter information to establish a pyrolysis furnace model and determine the value range of each parameter of the burner nozzle structure.

[0067] In this embodiment, the collected parameter information includes: operating conditions of the industrial ethylene cracking furnace, inlet fuel, oxidant content, burner and burner nozzle size, etc.

[0068] Parameter information was collected, and a three-dimensional model of the pyrolysis furnace was established with reference to the original size drawings of the pyrolysis furnace and the low-NOx burner for fuel grading. Mesh generation and simulation calculations were performed to obtain simulation results under the original size and analyze them. Parameters including combustion model, turbulence model, radiation model, NOx model and boundary conditions were set, and the final simulation calculation was completed. By comparing the operating parameters, the rationality of the selection of combustion model, turbulence model, radiation model and NOx model in the simulation was verified, ensuring the accuracy and reliability of the simulation results.

[0069] Figure 2 A 1 / 8 scale schematic diagram of a pyrolysis furnace according to an embodiment of the present invention is shown, as follows: Figure 2 The diagram shows a 1 / 8 structure of a pyrolysis furnace. The exhaust gas outlet is located at the top of the pyrolysis furnace. The furnace tubes 1 are evenly arranged above the furnace chamber, where the ethylene pyrolysis reaction occurs. The burners 2 are located below the pyrolysis furnace and close to the furnace chamber. The 1 / 8 structure of the pyrolysis furnace contains 8 burners 2.

[0070] The specific structure of furnace tube 1 is as follows: Figure 3a As shown, the specific structure of burner 2 is as follows: Figure 3b As shown, the burner 2 includes the burner body, the main burner nozzle 21, and the secondary burner nozzle 22.

[0071] The specific structure of the main burner 21 is as follows: Figure 3c As shown, the specific structure of the secondary burner 22 is as follows: Figure 3d As shown.

[0072] Methane gas is introduced into the main burner 21 and the secondary burner 22, and air is introduced into the air inlet 23 of the burner body as an oxidant. The methane reacts with the air to release a large amount of heat, which is used to provide heat for the ethylene cracking reaction in the furnace tube.

[0073] The structure of the pyrolysis furnace, including the furnace chamber, furnace tubes, burners, and burners, is designed and drawn in three dimensions. The burner assembly is used as a part and assembled with the furnace chamber and furnace tubes to form the overall pyrolysis furnace assembly.

[0074] Figure 4 A schematic diagram of the cracking furnace area division and grid according to an embodiment of the present invention is disclosed, such as... Figure 4 As shown, the fluid domain model is imported into the mesh for mesh generation, using either hexahedral or tetrahedral meshes.

[0075] Furthermore, to further verify the rationality of the model selection, the results of a single CFD simulation experiment were compared with the parameters collected under actual working conditions. The optimal combustion model, turbulence model, radiation model, and NOx model in the pyrolysis furnace model were selected to verify the rationality of the selected combustion model, turbulence model, radiation model, and NOx model, as well as the accuracy of the boundary condition calculations.

[0076] Based on preliminary selection and comparison, in this embodiment, the combustion model in the pyrolysis furnace adopts a non-premixed combustion model, the turbulence model adopts the standard k-ε model, the radiation model adopts the discrete ordinate model (DOM), and the NOx model adopts a post-processing model, mainly considering the mechanism of thermal NOx.

[0077] Combustion models in Fluent include finite rate models, premixed combustion models, non-premixed combustion models, partially premixed combustion models, and probability density function (PDF) transport equation models.

[0078] Among them, the finite rate model can be further subdivided into laminar finite rate model, eddy dissipation model, and eddy dissipation concept (EDC) model;

[0079] The non-premixed combustion model refers to the process where fuel and oxidizer (air) enter the reaction zone without mixing, and instead enter the reaction zone directly for combustion.

[0080] Considering that methane and air enter the cracking furnace from different inlets, the combustion model selected in step S1 is the non-premixed combustion model.

[0081] To enhance the prediction of nitrogen oxide content in the cracking furnace, and considering both accuracy and low computational cost, a post-treatment model combining non-premixed combustion and a NOx model is used. The NOx post-treatment model primarily considers its thermal generation mechanism.

[0082] The standard k-ε model, as a turbulence model, has been experimentally verified to meet simulation accuracy requirements and save computation time, and is widely used in industrial production simulation.

[0083] Therefore, in the combustion simulation of the pyrolysis furnace, step S1 selects the standard k-ε model as the turbulence model to simulate the flow of gas in the furnace.

[0084] The radiation model was set as a discrete ordinate model (DOM), which is the recommended radiation model for pyrolysis furnace simulation and has been widely used.

[0085] Boundary conditions refer to parameters such as the type of inlet, mass flow rate of the material at the inlet, pressure and temperature at the inlet and outlet, and thermal radiation power of the wall in the pyrolysis furnace model.

[0086] Since the burner nozzle consists of a main burner and a secondary burner, the distribution calculation needs to be performed based on the ratio of the nozzle area of ​​the main burner to that of the secondary burner. At the same time, the hydraulic diameter of each inlet needs to be calculated based on the surface shape and area of ​​the inlet.

[0087] The hydraulic diameter is four times the ratio of the cross-sectional area to the perimeter of the flow path.

[0088] The Reynolds number Re is a dimensionless number used to characterize fluid flow, and its corresponding expression is:

[0089]

[0090] Where v, ρ, and μ are the fluid velocity, density, and viscosity coefficient, respectively, and d is the characteristic length;

[0091] For example, if fluid flows through a circular pipe, then d is the equivalent diameter of the pipe.

[0092] The Reynolds number can be used to distinguish whether a fluid flow is laminar or turbulent, and it can also be used to determine the resistance an object experiences when flowing in a fluid.

[0093] Since the flow rate q = ρvπr 2 Substituting into the formula for calculating the Reynolds number, we get... The Reynolds numbers for the air inlet, main burner, and secondary burner were calculated, and the magnitude of the turbulence intensity was determined as follows:

[0094] Since the model is a 1 / 8 scale structure of the real model, all symmetry planes must be set to symmetry. Heat dissipation from the furnace walls also needs to be considered; heat loss is calculated based on 1% of the total heat released during combustion, divided by the surface area of ​​the furnace walls.

[0095] The ethylene cracking reaction occurs inside the furnace tube itself, and the tube wall needs to absorb a large amount of heat. The temperature of the furnace tube surface will also affect the temperature distribution of the furnace. The temperature of the furnace tube is simplified as a function that varies with the height. Furthermore, the temperature functions of the two sections of the furnace tube divided according to the gas inlet and gas outlet are different. The surface temperature of the furnace tube is set using a user-defined function.

[0096] According to Fluent's simulation calculations, and by comparing with the data collected under operating conditions, the errors are all within a reasonable range. The relative error of the temperature at the pyrolysis furnace outlet is within 1%, the relative error of the excess oxygen mass fraction is within 5%, and the relative error of the furnace thermal efficiency is within 2%.

[0097] Through simulation comparison, the established pyrolysis furnace simulation model in this embodiment can simulate the combustion process in the pyrolysis furnace and predict the NO content at the outlet while ensuring accuracy.

[0098] Step S2: Select several parameters of the burner nozzle structure and perform Latin hypercube sampling to establish a sample set through simulation.

[0099] Assuming the boundary condition calculations are accurate and the model selections in the pyrolysis furnace simulation are reasonable, it can be concluded that the simulation calculations under these conditions can simulate the combustion conditions in a real pyrolysis furnace. Therefore, it can be used as a standard for simulation calculations of other samples.

[0100] Latin hypercube sampling was performed on three parameters of the burner nozzle structure (height difference between primary and secondary burners, radius of primary burner, and radius of secondary burner). Each parameter was uniformly sampled within a large range of values ​​to obtain several sets of samples, which were then combined to form a sample set.

[0101] For each set of samples, geometric modeling, fluid domain creation, region division, mesh generation, boundary condition setting, and simulation calculation are performed. The NO content under each set of structural parameters is recorded, that is, the NO weighted mole fraction at the pyrolysis furnace outlet, to establish a real sample set of NO content with respect to burner structural parameters.

[0102] A single CFD simulation takes more than 10 hours. Furthermore, since there are countless points to be determined in the three-dimensional decision space composed of the burner nozzle structural parameters, it is impossible to simulate by exhaustive methods. In addition, the effects of each structural parameter on NOx emissions are coupled, so it is not possible to analyze a single structural parameter in isolation.

[0103] Establishing a surrogate model can effectively solve this problem. A surrogate model refers to an approximate mathematical model that can be used for complex and time-consuming numerical analyses in optimization design. By using fewer CFD simulations to build a realistic sample set, and then building a surrogate model based on this realistic sample set, predictions can be made for any point in the decision space, thus avoiding the time-consuming CFD simulation process.

[0104] Latin hypercube sampling (LHS) is a statistical method used to generate sample points in a multidimensional space to better represent the entire parameter space. By generating sample sets using LHS, surrogate models can be built to predict the performance and other relevant indicators of cracking furnaces.

[0105] In this embodiment, the burner nozzle structural parameters are sampled using Latin hypercube sampling. Each structural parameter is sampled uniformly within its relatively large range of values ​​and combined to form a sample set. When the main burner is lower than the secondary burner, the height difference between the main and secondary burners is negative.

[0106] The simulation results for each set of structural parameters are obtained through geometric modeling and simulation calculations, and information such as temperature and NO weighted mole fraction at the outlet are recorded.

[0107] The pyrolysis furnace has a complex structure, being a complex assembly, making parametric modeling difficult. Furthermore, errors are prone to occur during mesh generation, preventing online data-driven implementation. To address this issue, an offline proxy model must be built using a real sample set.

[0108] Step S3: Establish a proxy model for the NO content at the pyrolysis furnace outlet with respect to the burner nozzle structural parameters.

[0109] Low-NOx burners achieve NOx emission reduction through fuel staging; however, the structural parameters of the burner nozzle affect the combustion process, thereby influencing the NOx emission levels at the pyrolysis furnace outlet. To optimize the burner nozzle's structural parameters, computational fluid dynamics (CFD) methods can be used to simulate the combustion process in the pyrolysis furnace. By comparing the NOx emissions under different burner nozzle structural parameters, the optimal burner nozzle parameters can be found to achieve the goal of lower NOx emissions.

[0110] However, due to the large scale of the pyrolysis furnace, the requirements for computational fluid dynamics simulation resources are high. Specifically, every time the control parameters of the burner are changed, such as the radius of the main burner and the radius of the secondary burner, the pyrolysis furnace model needs to be redrawn, meshed, and simulation calculations need to be performed. Changing the structural parameters one by one for CFD calculation is inefficient and costly.

[0111] This invention uses a small number of CFD simulation results as initial samples and establishes a surrogate model based on the initial samples. Then, it combines optimization algorithms to optimize the burner nozzle structure, thereby reducing the cost of CFD simulation calculations and finding the optimal parameters of the burner nozzle structure.

[0112] More specifically, using the sample set obtained in step S2, a proxy model of the NO content at the outlet with respect to the burner nozzle structural parameters is established to replace the computationally time-consuming CFD simulation process.

[0113] In step S3, the established real sample set is randomly sorted and divided into a training set and a test set. Simultaneously, the input parameters are normalized, and a surrogate model is used to train the samples. This surrogate model replaces the computationally intensive CFD simulation process to predict the NO content under different burner nozzle structural parameters.

[0114] In this embodiment, the radial basis function surrogate model (RBF model) is selected as the surrogate model.

[0115] The RBF model has no specific requirements for response characteristics and can fit any kind of function well, including functions with high nonlinearity. The model has strong robustness and adaptability, fast convergence speed, and low computational cost.

[0116] The basic form of radial basis functions is:

[0117]

[0118] In the formula, The coefficient matrix, Let ||·|| represent the output of the i-th neuron in the hidden layer, and let ||·|| represent the Euclidean distance from the input vector to the center point. A Gaussian function is chosen as the kernel function for the radial basis functions.

[0119] The burner's burner structural parameters (height difference between primary and secondary burners, radius of primary burner, and radius of secondary burner) are used as input parameters, and the NO weighted mole fraction at the pyrolysis furnace outlet under each set of burner structural parameters is used as the output of the surrogate model to establish a radial basis function surrogate model.

[0120] The RBF surrogate model trained on the training set was used to make predictions on the test set. The relative error between the predicted and true values ​​on the test set was within 10%, indicating that the established surrogate model has strong generalization ability and can predict other unknown samples.

[0121] Step S4: Use the optimization algorithm to call the proxy model and optimize the burner structure parameters of the pyrolysis furnace within a certain three-dimensional decision space with a preset objective function to obtain the optimal parameters of the burner structure.

[0122] Based on the fact that the accuracy of the surrogate model obtained in step S3 meets the requirements, step S4 uses the particle swarm optimization algorithm to optimize the input parameters. That is, within the range of the input parameters, the NO weighted mole fraction at the cracking furnace outlet is used as the optimization objective function to optimize the burner structure in a certain three-dimensional decision space, thereby obtaining the optimal structural size parameters in the search area and achieving further reduction of NOx emissions.

[0123] After constructing the surrogate model in step S3, it is still necessary to continue searching for the optimal solution that minimizes nitrogen oxide emissions within the decision space. To achieve this goal, optimization algorithms are needed for optimization.

[0124] This invention employs a particle swarm optimization (PSO) algorithm to optimize the structural parameters of a pyrolysis furnace burner. The PSO algorithm is inspired by studies of bird flock foraging behavior; it leverages collective information sharing within the flock to help the group find the optimal destination. PSO offers advantages such as fast convergence, few parameters, and simple implementation.

[0125] Using the NO weighted mole fraction at the cracking furnace outlet as the objective function, an initial particle swarm is generated by calling a radial basis function surrogate model, and the particle swarm optimization algorithm is used for optimization and iteration to find the global optimum.

[0126] After obtaining the optimal structural parameters, the true NO weighted mole fraction was obtained using CFD simulation to verify the feasibility of the surrogate model and optimization algorithm.

[0127] Figure 5 A flowchart of a particle swarm optimization algorithm according to an embodiment of the present invention is disclosed, such as... Figure 5 The particle swarm optimization algorithm shown is as follows:

[0128] The particle swarm optimization algorithm generates several initial particles. For each particle, its velocity and position are iteratively updated. Each particle updates its velocity and position by tracking its own historical best solution and the global historical best solution. The information of particle i can be represented by a D-dimensional vector, and its position is represented by X. i =(x i1 ,x i2 ,...,x iD ) T The speed is expressed as V i =(vi1,v i2 ,...,v iD ) T Similarly, for other vectors, in each iteration, the particle's velocity and position are updated according to the following formula:

[0129]

[0130]

[0131] in, It is the d-th dimension velocity of particle i in the k-th iteration;

[0132] In this embodiment, the parameters of the burner structure consist of three dimensions: the radius of the main burner, the radius of the secondary burner, and the height difference between the main and secondary burners. Therefore, the information of the particle is 3-dimensional, with dimension D=3.

[0133] c1 and c2 are acceleration coefficients, used to adjust the maximum step size of the particle's flight towards the global optimal solution and the individual optimal solution, respectively. If the value is too small, the particle may not be able to keep up with the changes in the target area. If the value is too large, the particle may fly over the target area and miss the optimal solution.

[0134] and It is a random number between 0 and 1. It represents the position of particle i at its individual extreme point in the d-th dimension. It is the location of the global extremum point of the entire population in the d-th dimension. It is the current position of particle i in the d-th dimension during the k-th iteration.

[0135] To improve the accuracy of the predicted samples, it is necessary to search within the broadest range of the sample set and use the particle swarm optimization algorithm for iterative optimization.

[0136] Finally, after multiple iterations, the global optimal solution within the search range was found. Based on the optimal structural parameters found, the model was rebuilt using SolidWorks and ANSYS Fluent and simulations were performed. Simulation calculations were conducted under the same conditions, and the NO weighted mole fraction at the outlet was recorded and compared with the predicted value of the surrogate model. The comparison showed that the simulation results under the optimal structural parameters found by the particle swarm optimization algorithm were basically the same as the minimum value in the real sample set, differing by only about 1%. This demonstrates that the method proposed in this invention is feasible and effective.

[0137] Meanwhile, the volume-weighted average temperature of the furnace tube region was analyzed for the optimized sample. The results showed that this parameter was not affected in any way, which can ensure that the reaction inside the furnace tube can proceed normally. This not only achieves the goal of reducing NOx emissions, but also avoids other negative impacts.

[0138] In summary, this invention proposes a method for optimizing the burner structure of a pyrolysis furnace. By combining a radial basis function surrogate model and a particle swarm optimization algorithm, NO emissions are further reduced by changing the structural parameters of the burner. This method utilizes a small amount of simulation data to establish a realistic sample set and constructs a radial basis function surrogate model through offline learning. Subsequently, the particle swarm optimization algorithm calls the surrogate model to optimize the sample points within the search area, thus achieving optimization of the pyrolysis furnace burner structure. The advantage of this invention lies in its portability, making it suitable for offline data-driven optimization scenarios where online sample updates are not possible.

[0139] Figure 6 A block diagram of a pyrolysis furnace burner structure optimization device according to an embodiment of the present invention is disclosed. The pyrolysis furnace burner structure optimization device may include an internal communication bus 601, a processor 602, a read-only memory (ROM) 603, a random access memory (RAM) 504, a communication port 605, and a hard disk 607. The internal communication bus 601 enables data communication between components of the pyrolysis furnace burner structure optimization device. The processor 602 can perform judgments and issue prompts. In some embodiments, the processor 602 may consist of one or more processors.

[0140] The communication port 605 enables data transmission and communication between the pyrolysis furnace burner structure optimization device and external input / output devices. In some embodiments, the pyrolysis furnace burner structure optimization device can send and receive information and data from a network via the communication port 605. In some embodiments, the pyrolysis furnace burner structure optimization device can transmit data and communicate with external input / output devices via the input / output terminal 606 in a wired manner.

[0141] The pyrolysis furnace burner structure optimization device may also include different types of program storage units and data storage units, such as hard disk 607, read-only memory (ROM) 603, and random access memory (RAM) 604, capable of storing various data files used for computer processing and / or communication, as well as possible program instructions executed by processor 602. Processor 602 executes these instructions to implement the main part of the method. The results processed by processor 602 are transmitted to an external output device via communication port 605 and displayed on the user interface of the output device.

[0142] For example, the implementation process document of the above-mentioned method for optimizing the structure of the pyrolysis furnace burner can be a computer program, stored in the hard disk 607, and can be loaded into the processor 602 for execution to implement the method of this application.

[0143] When the implementation process document of the pyrolysis furnace burner structure optimization method is a computer program, it can also be stored as an article of manufacture in a computer-readable storage medium. For example, computer-readable storage media can include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, magnetic stripes), optical discs (e.g., compact discs (CDs), digital multifunction discs (DVDs)), smart cards, and flash memory devices (e.g., electrically erasable programmable read-only memory (EPROM), cards, sticks, key drives). Furthermore, the various storage media described herein can represent one or more devices and / or other machine-readable media used for storing information. The term "machine-readable medium" can include, but is not limited to, wireless channels and various other media (and / or storage media) capable of storing, containing, and / or carrying code and / or instructions and / or data.

[0144] This invention proposes a method and apparatus for optimizing the burner structure of a pyrolysis furnace. It establishes an accurate simulation model of the pyrolysis furnace by modeling and simulating the furnace and comparing the results with data collected from real-world operating conditions. Based on this, simulations are performed on different burner structure models to create a real sample set. Through training and testing on this real sample set, a radial basis function neural network surrogate model is established and combined with a particle swarm optimization algorithm. Within the search region, the optimal parameters of the burner structure are searched using the NO content at the pyrolysis furnace outlet as the optimization objective. This further reduces NOx emissions based on the low-NOx burner design for fuel grading, thereby achieving the goal of reducing nitrogen oxide emissions at a lower cost. This invention can reduce nitrogen oxide emissions by using a smaller number of simulation calculations, replacing the computationally intensive CFD simulation process with a surrogate model, and finding the optimal parameters of the burner structure using a particle swarm optimization algorithm.

[0145] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0146] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0147] The above embodiments are provided for those skilled in the art to implement or use the present invention. Those skilled in the art can make various modifications or changes to the above embodiments without departing from the inventive concept of the present invention. Therefore, the protection scope of the present invention is not limited to the above embodiments, but should be the maximum scope that conforms to the innovative features mentioned in the claims.

Claims

1. A method for optimizing the structure of a pyrolysis furnace burner, characterized in that, Includes the following steps: Step S1: Collect parameter information to establish a pyrolysis furnace model and determine the value range of each parameter of the burner nozzle structure; Step S2: Select several parameters of the burner nozzle structure and perform Latin hypercube sampling to establish a sample set through simulation; Step S3: Establish a proxy model for the NO content at the pyrolysis furnace outlet with respect to the burner nozzle structural parameters; Step S4: Use the optimization algorithm to call the proxy model and optimize the burner structure parameters of the pyrolysis furnace in the three-dimensional decision space with a preset objective function to obtain the optimal parameters of the burner structure. Step S2 further includes: A sample set was formed by Latin hypercube sampling of several parameters of the burner nozzle structure. For each sample in the sample set, geometric modeling, fluid domain creation, region division, mesh generation, boundary condition setting, and simulation calculation are performed. The NO content under each set of structural parameters is recorded to establish a real sample set. The optimization algorithm in step S4 includes particle swarm optimization algorithm; In step S4, the preset objective function is the weighted mole fraction of NO at the outlet of the pyrolysis furnace. The process following step S4 also includes: After obtaining the optimal parameters of the pyrolysis furnace burner structure, CFD simulation was used to obtain the true value of the NO weighted mole fraction of the pyrolysis furnace burner structure with the optimal parameters, thus verifying the feasibility of the surrogate model and optimization algorithm.

2. The method for optimizing the burner structure of a pyrolysis furnace according to claim 1, characterized in that, The parameter information collected in step S1 includes: Information on the operating conditions, inlet fuel, oxidant content, burner and burner nozzle dimensions of industrial ethylene cracking furnaces.

3. The method for optimizing the burner structure of a pyrolysis furnace according to claim 1, characterized in that, Step S1 further includes: The results of a single CFD simulation experiment were compared with the parameters collected under actual working conditions, and the best combustion model, turbulence model, radiation model, and NOx model were selected from the pyrolysis furnace model.

4. The method for optimizing the burner structure of a pyrolysis furnace according to claim 3, characterized in that, The combustion model selected in step S1 is a non-premixed combustion model, and the turbulence model is a standard one. The radiation model is a discrete ordinate model, and the NOx model is a post-processed model.

5. The method for optimizing the burner structure of a pyrolysis furnace according to claim 2, characterized in that, The parameters of the burner nozzle structure in step S2 further include: the height difference between the primary and secondary burners, the radius of the primary burner, and the radius of the secondary burner.

6. The method for optimizing the burner structure of a pyrolysis furnace according to claim 1, characterized in that, The surrogate model in step S3 includes a radial basis function neural network surrogate model.

7. The method for optimizing the burner structure of a pyrolysis furnace according to claim 1, characterized in that, Step S3 further includes: The real sample set is divided into a training set and a test set; The input parameters of the burner nozzle structure are normalized. Use a surrogate model to train the samples; The NO content under different burner nozzle structural parameters was predicted using the finally established surrogate model.

8. A device for optimizing the structure of a pyrolysis furnace burner, comprising: Memory is used to store instructions that can be executed by the processor; A processor for executing the instructions to implement the method as described in any one of claims 1-7.

9. A computer-readable medium having stored thereon computer instructions, wherein when the computer instructions are executed by a processor, the method as described in any one of claims 1-7 is performed.

Images