WO2020155865A1 - 一种延迟焦化模型集成方法 - Google Patents

一种延迟焦化模型集成方法 Download PDF

Info

Publication number
WO2020155865A1
WO2020155865A1 PCT/CN2019/124316 CN2019124316W WO2020155865A1 WO 2020155865 A1 WO2020155865 A1 WO 2020155865A1 CN 2019124316 W CN2019124316 W CN 2019124316W WO 2020155865 A1 WO2020155865 A1 WO 2020155865A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
model
delayed coking
coking
fingerprint data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2019/124316
Other languages
English (en)
French (fr)
Inventor
钱锋
杨明磊
钟伟民
杜文莉
李智
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
East China University of Science and Technology
Original Assignee
East China University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by East China University of Science and Technology filed Critical East China University of Science and Technology
Publication of WO2020155865A1 publication Critical patent/WO2020155865A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C10/00—Computational theoretical chemistry, i.e. ICT specially adapted for theoretical aspects of quantum chemistry, molecular mechanics, molecular dynamics or the like
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/10—Analysis or design of chemical reactions, syntheses or processes
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Definitions

  • the invention relates to the industrialization optimization of a refinery and chemical process, in particular to a delayed coking model integration method based on fingerprint data.
  • Delayed coking is the main device for processing residual oil, asphalt, slop oil and other heavy oil products in the refining and chemical process.
  • the common point of delayed coking and other forms of coking is the process of using thermal cracking to deeply react the residual oil into gas, gasoline, diesel, wax oil and solid product coke.
  • the difference between delayed coking and other coking methods is that the residual oil flows through the furnace tube of the heating furnace at a high flow rate, is heated to the temperature required for the reaction of 490-510°C, and then enters the coke drum, where it is carried by itself. Heat, cracking, condensation and other reactions.
  • the residence time in the furnace tube is very short, which delays the cracking, condensation, and decomposition reactions to the coke tower, avoiding (reducing) the coking of the furnace tube.
  • the coke produced by the reaction is focused in the tower, and the high-temperature oil vapor generated from the vapor volatilization line enters the fractionation tower and exchanges heat with the raw materials.
  • the heavy oil enters the heating furnace with the raw materials. After the components of the light oil are separated, gas is obtained. , Gasoline, diesel, wax oil and other products. What is commonly used in the technological process is one heating furnace with two coke towers, as shown in Figure 1.
  • the hot residual oil enters one of the coke towers, and when the generated coke accumulates to a certain height, it is decoked. At the same time, the hot residual oil is switched to another coke tower to ensure the continuous operation of the coke tower and the subsequent fractionation process.
  • the key to delayed coking process simulation is to simulate its reaction process and fractionation process.
  • the reaction kinetics of lumping raw materials and products is generally used to describe the reaction kinetics, and there are 6 lumps, 10 lumps, 12 lumps, and so on.
  • This kind of kinetic model requires more detailed analysis of raw materials, and accurate four-component information in addition to macroscopic properties.
  • there are often only a few properties analysis data for residual oil and the key four-component analysis is not listed in the routine analysis items, and the frequency is often half a month to once a month.
  • the present invention aims to provide a delayed coking model integration.
  • a delayed coking model integration method including the steps:
  • step (3) According to the delayed coking product data and the yield obtained in step (2), the fingerprint data of the full fraction delayed coking product is formed by fitting;
  • the integrated fractionation model realizes the separation of gas, liquefied gas, gasoline, diesel and wax oil.
  • the macroscopic properties of the raw material in step (1) include distillation range, density, sulfur content, residual carbon and nitrogen content; the components in the raw material are four components, which are saturated and aromatic. Points, gum and asphaltene.
  • step (2) is to use the kinetic model to calculate the yield of the coking product according to the macroscopic properties of the raw material and the operating conditions; a 10 lumped kinetic model is more preferred.
  • the operating conditions include reaction temperature and reaction pressure.
  • the coking product property data in step (3) includes gas composition, liquefied gas composition, gasoline real boiling point distillation data, diesel real boiling point distillation data, and wax oil real boiling point distillation data; the fitting uses ASPEN Crude oil characterization tool.
  • step (4) is to use the full fraction delayed coking product generated in step (3) as a raw material and input it to the rectification tower model to realize the separation of gas, liquefied gas, gasoline, diesel and wax oil.
  • the method further includes the step of correcting the fingerprint data described in step (1).
  • the macroscopic properties of the raw materials and the four-component data are combined with the differential evolution algorithm with triangular mutation to correct the fingerprint data.
  • the simulation and/or optimization includes delayed coking process model development, device size optimization, and production plan optimization model verification.
  • the present invention provides a reliable delayed coking device model.
  • Figure 1 is a simplified flow chart of the delayed coking process.
  • Figure 2 shows the correlation between the macroscopic properties and the four components.
  • Figure 3 is a simplified flow chart of fingerprint database calibration.
  • the inventor provides a delayed coking model integration method based on fingerprint data.
  • the method is based on 10 lumped delayed coking reaction kinetic models, actual industrial operating data (macro properties, four components, operating conditions, etc.), and distillation tower models.
  • the fingerprint database method is used to correlate the macro properties with four component analysis. , To form a practical kinetic model for industrial sites, and to delump the products at the same time, so that the delayed coking reaction can be lumped up to the full fraction data, and it can be used as a fractionation tower to simulate operation and realize the integration of delayed coking device models. Level optimization applications provide reliable model support.
  • the method for integrating delayed coking models based on fingerprint data includes the following steps:
  • Raw material fingerprint data Establish the fingerprint data association between the macroscopic properties of the raw materials in the delayed coking unit of the refinery and the content of the four components in the raw materials. Input the macroscopic properties of the raw materials to obtain the four-component information of the raw materials;
  • step 2 On the basis of the correlation obtained in step 1, use the ten-lumped kinetic model to calculate the yield of coking products according to the properties of raw materials and operating conditions;
  • Product fingerprint data According to the coking product data collected on site, it mainly includes gas composition, liquefied gas composition, gasoline real boiling point distillation data, diesel real boiling point distillation data, and wax oil real boiling point distillation data, using the ASPEN crude oil characterization tool to perform fitting, combined with the second step to get The yield of the coking reaction is formed to form fingerprint data from the total product to the whole fraction of the coking reaction;
  • step 4 Integrated fractionation model.
  • the full fraction delayed coking product generated in step 3 is used as a raw material and input to the rectification tower model to realize the separation simulation of gas, liquefied gas, gasoline, diesel and wax oil.
  • the method further includes the steps:
  • Delayed coking raw materials are residual oil, asphalt and oil slurry, etc.
  • the reaction is mostly based on cracking, which converts heavy, long-chain hydrocarbons into gas, liquefied gas, gasoline, diesel, wax oil and coke. Since most of the raw materials are above C40, the cracking reaction mechanism is extremely complicated.
  • a lumped method is used to divide the raw materials into saturated components, aromatic components, gums and asphaltenes. Its response characteristics.
  • the actual industrial process generally only has macro characteristics analysis, which mainly includes distillation range, density, sulfur content, residual carbon, and nitrogen content. A large number of studies have shown that there is a quantitative relationship between the macroscopic properties of the material and its four components.
  • the main ones are: (1) The higher the boiling point of the raw material, the higher the proportion of gum and asphaltene. On the contrary, the higher the proportion of saturation. (2) Residual carbon mainly exists in asphaltenes and gums; (3) Density reflects the heaviness of raw materials, and the relationship with the four components is similar to the boiling point; (4) Sulfur content is more heavily distributed in asphaltenes; (5) The nitrogen content is similar to the sulfur content.
  • the invention adopts a fingerprint data method to establish a quantitative correlation between distillation range, density, sulfur content, residual carbon, nitrogen content and the four components.
  • the fingerprint data of the present invention is a vector processing method, which expresses the relationship between the output in a complex system and one or more variables in a linear array, as shown in formula (1).
  • y is the four-component value
  • y 0 is the four-component reference value
  • k is the rate of change
  • ⁇ x is the macroscopic property change value
  • the general formula (1) is the general formula for calculating the four-component based on the macroscopic property.
  • Figure 2 shows the correlation between the macroscopic properties and the four components in the fingerprint data.
  • the quantitative correlation formula of the four components in the fingerprint data is as follows:
  • Sat, Aro, Res, and Asp represent saturated components, aromatic components, gums and asphaltenes, respectively.
  • the subscript 0 represents the reference value
  • ⁇ represents the deviation value from the reference property.
  • the main purpose of this step is to use the fingerprint data established in the previous step to realize the integration of industrial field macro data and lumped dynamics models.
  • the main contents include: field data collection, data reconciliation, model interface development and model calculation.
  • On-site data collection In the actual production process, most factories will use laboratory analysis data to record material property data, and provide the corresponding data points to collect data.
  • the invention uses an industrial field database system to collect raw material property data and store it in a local database.
  • the data to be collected mainly include the distillation range, density, nitrogen content, sulfur content and residual carbon data of the delayed coking mixed raw material.
  • Model interface development After sorting out the sample data, it is necessary to develop a corresponding interface to send the raw material property data to the kinetic model to realize automatic data transmission.
  • the ten ensemble delayed coking kinetic model of the present invention is developed using the aspen platform, and the interface program is developed using vb.net.
  • Each data is transmitted by compiling specific fields and using the data table in the aspen software. The data fields of various properties are shown in Table 1. .
  • n 10 ⁇ 5; preferably 10 ⁇ 3
  • the product fingerprint data component table the data of gas and liquefied gas can be directly obtained from the average of the composition; gasoline, diesel and wax oil are fitted to the distillation range curve through polynomials to obtain the proportions of the components in different temperature ranges, and then the corresponding Fingerprint data.
  • T 30% (T 10% + T 50% )/2 (12)
  • T 70% (T 50% + T 90% )/2 (13)
  • Y is the cumulative volume yield
  • A, B, C, D, E, F are polynomial parameters (no specific meaning)
  • x is the temperature.
  • the main purpose of this step is to use the fingerprint data established in the previous step to realize the integration of industrial field macro data with the main fractionation tower model.
  • the main contents include: field data collection, data reconciliation, model interface development and model calculation.
  • Field data collection use industrial field database system to collect real-time data and laboratory analysis data.
  • Real-time data includes fractionation tower operating conditions, such as reflux ratio, tower top pressure, sensitive plate temperature, etc.
  • laboratory analysis data includes gas and liquefied gas molecular composition data, gasoline, diesel and wax oil distillation data;
  • Model interface development After finishing the sample data, it is necessary to develop a corresponding interface to send the raw material property data to the distillation tower model to realize automatic data transmission. Refer to the delayed coking kinetic model using the interface program, and use the data table in the aspen software for transmission. Each component field is the component name.
  • Model calculation On the basis of the prepared real-time data and analysis data, it is automatically transmitted to the distillation tower model, and run in the aspen software to obtain the composition and flow of gas and liquefied gas; detailed fractions of gasoline, diesel and wax oil Composition of data and traffic.
  • it further includes:
  • Fingerprint data is the core of raw material characterization. After the oil refinery is switched to crude oil, the material properties are prone to large fluctuations. The fingerprint data needs to be re-calibrated to ensure the accuracy of the property correlation.
  • Fingerprint data correction is actually an optimization problem. Select the four-component prediction value and the minimum variance of the collected data from the industrial field as the goal, and convert the fingerprint data determination process into a function optimization problem to solve, namely:
  • the decision variable x includes the four-component correlation coefficients of various properties, and x actual and x calculate respectively represent the four-component data of raw materials calculated through actual industrial analysis and fingerprint database. Aiming at this type of optimization goal, the present invention uses a differential evolution algorithm with triangle mutation to solve the problem.
  • Differential evolution algorithm (differential evolution, DE) is a population-based random search algorithm, which has the characteristics of simple structure, fast convergence speed, and high robustness.
  • the mutation mechanism of the algorithm that is, the method of generating offspring is:
  • r' is the newly generated offspring individual
  • r 1 , r 2 , and r 3 are three different parent individuals randomly selected in the population
  • F is the differential evolution operator, which is generally a constant.
  • the present invention selects an improved differential evolution algorithm with triangular mutation. This method is proven to have significant effects in improving the convergence speed of the algorithm.
  • the improved mutation strategy can be expressed as:
  • r′ (r 1 +r 2 +r 3 )/3+(p 2 -p 1 )(r 1 -r 2 )+(p 3 -p 2 )(r 2 -r 3 )+(p 1- p 3 )(r 3 -r 1 ) (17)
  • the present invention provides an effective and reliable idea for the simulation and integration of the entire delayed coking process, and at the same time significantly improves the applicability of the model in industry.
  • the distillation range, density, sulfur content, residual carbon, nitrogen content and the four components are quantitatively correlated:
  • y 0 is the benchmark
  • k is the rate of change
  • ⁇ x is the amount of parameter change.
  • CCR Carbon Residue
  • Sat, Aro, Res, and Asp represent saturated components, aromatic components, gums and asphaltenes, respectively.
  • the subscript 0 represents the reference value
  • the reference values of Sat, Aro, Res and Asp are 0.5051, 0.1603, 0.1079 and 0.2266 respectively.
  • ⁇ represents the deviation value from the reference property (Table 6).
  • the gas mainly contains hydrogen sulfide, hydrogen, C1, C2 and C3, and the liquefied gas mainly contains C3 and C4.
  • the distillation range of gasoline is C5-230°C
  • the distillation range of diesel is 180°C-380°C
  • the distillation range of wax oil is 350°C- Final boiling point.
  • T 30% (T 10% + T 50% )/2 (12)
  • T 70% (T 50% + T 90% )/2 (13)
  • Y is the cumulative volume yield
  • A, B, C, D, E, F are polynomial parameters (no specific meaning)
  • x is the temperature.
  • the actual average data of raw materials for a certain week in the delayed coking unit of a refinery The macroscopic properties of the raw materials are: distillation range (IBP: 555; 5%: 559; 10%: 562; 30%: 584; 50%: 606; 70) %: 637; 90%: 735;), density (20° C.) is 0.9748, residual carbon is 20%, sulfur content is 1.9%, nitrogen content is 7920.1 ppm.
  • the corrected four-component output data obtained by the algorithm is shown in Table 5, and the corrected fingerprint data is shown in Table 6.
  • the above method is based on fingerprint data to establish the macroscopic properties of raw materials, including real boiling point distillation data, density, sulfur content and residual carbon, and quantitative correlation with saturated content, aromatic content, gum and asphaltene content, realizing industrial field data and delay
  • fingerprint data to establish the macroscopic properties of raw materials, including real boiling point distillation data, density, sulfur content and residual carbon, and quantitative correlation with saturated content, aromatic content, gum and asphaltene content, realizing industrial field data and delay
  • the integration of ten lumped coking models at the same time, combined with the analysis data of the coking product laboratory, the fingerprint data association between the reaction products in the lumped kinetics and the detailed composition of the actual plant products is established to realize the relationship between the kinetic model and the distillation model integrated.
  • the collected industrial field data needs to be processed by data reconciliation technology, combined with the improved differential evolution algorithm, to correct the raw material fingerprint data, so that the fingerprint data can accurately predict the four-component information on the basis of macroscopic properties.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Chemical & Material Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Analytical Chemistry (AREA)
  • Evolutionary Computation (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Production Of Liquid Hydrocarbon Mixture For Refining Petroleum (AREA)

Abstract

一种延迟焦化模型集成方法。所述方法包括步骤：(1)使延迟焦化装置原料宏观性质与原料中的组分含量之间的指纹数据关联；(2)集成焦化反应模型得到焦化产品收率；(3)根据焦化产品数据与步骤(2)的得到的收率拟合形成全馏分延迟焦化产物的指纹数据；和(4)集成分馏模型实现气体、液化气、汽油、柴油和蜡油的分离。

Description

一种延迟焦化模型集成方法 技术领域
本发明涉及炼油化工过程的工业化优化，尤其涉及一种基于指纹数据的延迟焦化模型集成方法。
背景技术
延迟焦化是炼油化工过程中处理渣油、沥青、污油以及其他重质油品的主要装置。延迟焦化和其他形式的焦化共同之点是采用加热裂解，使渣油深度反应转化为气体、汽油、柴油、蜡油和固体产品焦炭的过程。延迟焦化与其他焦化方法的不同点是渣油以高的流速流过加热炉的炉管，加热到反应所需的温度490-510℃，然后进入焦炭塔，在焦炭塔里靠自身带入的热量，进行裂化，缩合等反应。
由于渣油流速很快，在炉管内的停留时间很短，因而使裂化、缩合、分解反应延迟到焦炭塔内进行，避免(减轻)炉管结焦。反应生成的焦炭聚焦在塔内，生成的高温油汽从油汽挥发线进入分馏塔内，与原料换热，重质油随原料进入加热炉，轻质油的组份经分离后，得到瓦斯、汽油、柴油、蜡油等产物。工艺流程上普遍采用的是一个加热炉配两个焦炭塔，如图1所示。热渣油进入其中一个焦炭塔，当生成的焦炭堆积到一定高度后进行除焦，同时将热渣油切换到另一个焦炭塔，保证焦炭塔与后续分馏过程连续操作。
延迟焦化过程模拟关键在于模拟其反应过程和分馏过程。针对延迟焦化反应过程，目前普遍采用将原料和产品进行集总(lump)的反应动力学来描述，存在6集总、10集总、12集总等。这类动力学模型需要原料有较详细的分析，除宏观性质之外还需要有准确的四组分信息。然而，实际工业现场针对渣油往往只存在少数性质分析数据，而关键的四组分分析不列在常规分析项，频次往往是半个月到一个月一次。在这种情况下，如何在有限的原料信息下利用现有的集总动力学模型对反应过程模拟是提高模型实用性的关键；另一方面，针对分馏过程，由于集总反应产物为虚拟组分，无法直接进入精馏塔进行切割和蒸馏模拟。
因此，本领域迫切需要提供一种可靠的延迟焦化装置模型。
发明内容
本发明旨在提供一种延迟焦化模型集成。
在本发明的第一方面，提供一种延迟焦化模型集成方法，所述方法包括步骤：
(1)使延迟焦化装置原料宏观性质与原料中的组分含量之间的指纹数据关联；
(2)集成延迟焦化反应动力学模型得到焦化产品收率；
(3)根据延迟焦化产品数据与步骤(2)的得到的收率拟合形成全馏分延迟焦化产物的指纹数据；和
(4)集成分馏模型实现气体、液化气、汽油、柴油和蜡油的分离。
在另一优选例中，步骤(1)中所述的原料宏观性质包括馏程，密度、硫含量、残碳和氮含量；所述原料中的组分是四组分，为饱和分、芳香分、胶质和沥青质。
在另一优选例中，步骤(2)是利用动力学模型，根据原料宏观性质和操作条件计算焦化产品收率；更优选10集总动力学模型。
在另一优选例中，所述操作条件包括反应温度和反应压力。
在另一优选例中，步骤(3)所述焦化产品性质数据包括气体组成，液化气组成，汽油实沸点蒸馏数据、柴油实沸点蒸馏数据和蜡油实沸点蒸馏数据；所述拟合采用ASPEN原油表征工具进行。
在另一优选例中，步骤(4)是利用步骤(3)生成的全馏分延迟焦化产物作为原料，输入至精馏塔模型，实现气体、液化气、汽油、柴油和蜡油的分离。
在另一优选例中，所述方法还包括步骤：校正步骤(1)中所述的指纹数据。
在另一优选例中，利用原料宏观性质以及四组分数据，结合带有三角变异的差分进化算法对指纹数据进行校正。
在本发明的第二方面，提供一种如上所述的本发明提供的方法在延迟焦化模拟和/或优化方面的应用。
在另一优选例中，所述模拟和/或优化包括延迟焦化过程模型开发、装置尺度优化、和生产计划优化模型校核。
据此，本发明提供了一种可靠的延迟焦化装置模型。
附图说明
图1为延迟焦化过程简化流程图。
图2显示了宏观性质与四组分之间的关联。
图3为指纹数据库校正简化流程图。
具体实施方式
发明人经过广泛而深入的研究，提供一种基于指纹数据的延迟焦化模型集成方法。方法基于10集总的延迟焦化反应动力学模型、实际工业运行数据(宏观性质、四组分、操作条件等)以及精馏塔模型，应用指纹数据库方法，将宏观性质与四组分分析进行关联，形成工业现场实用的动力学模型，同时对产物进行解集总(delump)，使延迟焦化反应集总扩展至全馏分数据，作为分馏塔进行模拟操作，实现延迟焦化装置模型的集成，为工业级优化应用提供可靠的模型支撑。
具体地，本发明提供的一种基于指纹数据的延迟焦化模型集成方法包括以下步骤：
一、原料指纹数据。建立炼油厂延迟焦化装置原料宏观性质与原料中的四组分含量之间的指纹数据关联，输入原料宏观性质即得到原料四组分信息；
二、集成反应模型。在步骤1得到的关联关系基础上，利用十集总动力学模型，根据原料性质和操作条件计算焦化产品收率；
三、产物指纹数据。根据现场采集的焦化产品数据，主要包括气体组成，液化气组成，汽油实沸点蒸馏数据、柴油实沸点蒸馏数据和蜡油实沸点蒸馏数据，利用ASPEN原油表征工具进行拟合，结合第二步得到的收率，形成焦化反应集总产物到全馏分的指纹数据；
四、集成分馏模型。利用步骤3中生成的全馏分延迟焦化产物作为原料，输入至精馏塔模型，实现气体、液化气、汽油、柴油和蜡油的分离模拟。
在本发明的一种实施方式中，所述方法还包括步骤：
五、校正原料指纹数据。利用工业装置的原料宏观性质以及四组分数据，结合带有三角变异的差分进化算法对指纹数据进行修正，提高指纹数据对四组分预测的准确性。
分述每一个步骤如下：
1.原料指纹数据
延迟焦化原料为渣油、沥青和油浆等，反应多以裂解为主，由重质、长链的烃类转化为气体、液化气、汽油、柴油、蜡油和焦炭。由于原料大部分在C40以上，裂解反应机理极其复杂，目前普遍根据重质油品分子特性，采用集总方式，将原料分为饱和分、芳香分、胶质和沥青质四个集总分别考虑其反应特性。然而实际工业过程普遍只有宏观特性分析，主要包括馏程、密度、硫含量、残碳、氮含量。大量的研究表明，物料宏观特性与其四组分之间存在定量关联，主要有：(1)沸点越高的原料，胶质和沥青质的占比会越高，反之，饱和分占比越高；(2)残碳主要存在于沥青质和胶质中；(3)密度反映原料重质程度，与四组分的关系类似于沸点；(4)硫含量在沥青质中分布权重更大；(5)氮含量和硫含量类似。
根据工业现场的实际情况，构建延迟焦化原料宏观特性与四组分之间的关联是提高集总动力学模型准确性和应用价值的关键。本发明采用指纹数据方法建立馏程、密度、硫含量、残碳、氮含量与四组分之间的定量关联。本发明所述指纹数据是一种矢量处理方法，将复杂体系中的输出与一个或多个变量之间的关系通过线性数组的方式表达，如式(1)。
y＝y 0+k*Δx      (1)
式中y是四组分值，y 0为四组分基准值，k为变化率，Δx为宏观性质变化值，通式(1)即为基于宏观性质计算四组分的通式。
图2显示了指纹数据中宏观性质与四组分之间的关联。对于各宏观性质，其指纹数据中对四组分的定量关联式如下：
1)0％点
Figure PCTCN2019124316-appb-000001
2)5％点
Figure PCTCN2019124316-appb-000002
3)10％点
Figure PCTCN2019124316-appb-000003
4)30％点
Figure PCTCN2019124316-appb-000004
5)50％点
Figure PCTCN2019124316-appb-000005
6)70％点
Figure PCTCN2019124316-appb-000006
7)90％点
Figure PCTCN2019124316-appb-000007
8)密度(ρ，20℃)
Figure PCTCN2019124316-appb-000008
9)残碳(CCR)
Figure PCTCN2019124316-appb-000009
10)硫含量(S)
Figure PCTCN2019124316-appb-000010
11)氮含量(N)
Figure PCTCN2019124316-appb-000011
式1-12中，Sat，Aro，Res，Asp分别表示饱和分、芳香分、胶质和沥青质。下标0表示基准值，Δ表示与基准性质的偏差值。
2.集成反应模型
本步骤的主要目的在于利用上一步建立的指纹数据，实现工业现场宏观数据与集总动力学模型的集成。主要内容包括：现场数据采集、数据调和、模型接口开发和模型计算。
(1)现场数据采集：在实际生产过程中，大多数工厂都会使用实验室分析数据来记录物料性质数据，并提供相应数据点的位号以便采集数据。本发明利用工业现场数据库系统收集原料性质数据，并储存到本地数据库中。需要采集的数据主要包括延迟焦化混合原料馏程、密度、氮含量、硫含量和残碳数据。
(2)数据调和处理：受现场检测仪表可靠性的局限，直接从实验室分析系统上获取到的数据可能存在物料不平衡、性质误录入等问题，因此不能直接用于装置模型集成测算。为了确保模型样本数据的准确性，有必要对实时采集的数据建立调和标准，具体使用以下几种方法：1)采用周平均值来校正模型；2)根据统计数据和生产经验确定数据的值域，依此判断数据的准确性，将错误数据从本地数据库中删除；3)对于在特定期间内无法采集的数据，建立冗余的计算公式，通过采集其他数据来推导出该数据的值。
(3)模型接口开发：整理好样本数据后需要开发相应的接口将原料性质数据送至动力学模型中，实现数据自动传输。目前本发明中十集总延迟焦化动力学模型使用aspen平台开发，接口程序采用vb.net开发，各数据通过编制特定字段利用aspen软件中的data table进行传输，各性质数据字段如表1所示。
表1性质字段
Figure PCTCN2019124316-appb-000012
(4)模型计算：在原料条件准备好的基础上，输入相应的反应温度和压力，通过十集总动力学模型，在aspen软件中运行即可计算得到气体、液化气、汽油、柴油、蜡油和焦炭收率。
3.产物指纹数据库
通过集总动力学模型可以获得准确的气体、液化气、汽油、柴油、蜡油和焦炭收率，但各个产品都是单一的集总，无法直接进行精馏模拟，需将除焦炭以外的其他产品进行扩充，形成全馏分产物。建立产物指纹数据库可以将产品收率和详细的产品组成进行关联，形成精馏塔可以处理的物料，进而进行分馏模拟。本步骤结合实际案例介绍产物指纹数据库构建过程。
首先，建立焦化产物完整的组分列表，覆盖气体、液化气、汽油、柴油和蜡油。
其次，建立各产物的指纹数据组分表。
最后，结合气体、液化气、汽油、柴油和蜡油的实验室化验分析数据，统计n(n为10±5；优选10±3)次分析数据，以平均数的方式，计算气体和液化气的分子组成，汽油、柴油和蜡油的实沸点馏程数据，其中针对企业缺少30％温度点和70％温度点的情况，采用前后温度的算术平均进行预测。产物指纹数据组分表中，气体、液化气的数据直接从组成平均数可得到；汽油、柴油和蜡油则通过多项式拟合馏程曲线，得到不同温度区间组分的占比，进而获得相应指纹数据。
30％温度点(T 30％)计算公式：
T 30％＝(T 10％+T 50％)/2     (12)
70％温度点(T 70％)计算公式：
T 70％＝(T 50％+T 90％)/2     (13)
收率与温度拟合多项式：
y＝Ax 5+Bx 4+Cx 3+Dx 2+Ex+F    (14)
Y为累计体积收率，A、B、C、D、E、F为多项式参数(无具体意义)，x为温度。
4.集成分馏模型
本步骤的主要目的在于利用上一步建立的指纹数据，实现工业现场宏观数据与主分馏塔模型进行集成。主要内容包括：现场数据采集、数据调和、模型接口开发和模型计算。
(1)现场数据采集：利用工业现场数据库系统收集实时数据和实验室分析数据。实时数据包括分馏塔操作条件，如回流比、塔顶压力、灵敏板温度等；实验室分析数据包括，气体和液化气分子组成数据，汽油、柴油和蜡油蒸馏数据；
(2)数据调和处理：根据统计数据和生产经验确定数据的值域，依此判断实时数据和实验室分析数据的准确性，将错误数据从本地数据库中删除；
(3)模型接口开发：整理好样本数据后需要开发相应的接口将原料性质数据送至精馏塔模型中，实现数据自动传输。参考延迟焦化动力学模型使用接口程序，利用aspen软件中的data table进行传输，各组分字段即组分名字。
(4)模型计算：在实时数据和分析数据准备好的基础上，自动传输至精馏塔模型，在aspen软件中运行得到气体、液化气的组成与流量；汽油、柴油和蜡油的详细馏分组成数据和流量。
在本发明的一种实施方式中，还包括：
5.原料指纹数据库校正
指纹数据是原料表征的核心，当炼油装置切换原油后，物料性质易发生较大波动，需对指纹数据进行重新校正，确保性质关联的准确性。
指纹数据校正实际上是一个优化问题。选取四组分的预测值与工业现场采集数据的方差和最小为目标，将指纹数据的确定过程转化为函数的优化问题进行求解，即：
Figure PCTCN2019124316-appb-000013
其中，决策变量x包括各个性质的四组分关联系数，x actual和x calculate分别表示通过实际工业上分析和指纹数据库计算得到的原料四组分数据。针对这种类型的优化目标，本发明使用带有三角变异的差分进化算法对问题进行求解。
差分进化算法(differential evolution，DE)是一种基于种群的随机搜索算法，它具有结构简单、收敛速度快、鲁棒性高等特点。算法的变异机制，即生成子代的方法为：
r′＝r 1+F*(r 2-r 3)     (16)
其中，r′是新生成的子代个体，r 1,r 2,r 3是种群中随机选取的三个不同的父代个体，F为差分进化算子，一般为一个常数。
由于该目标决策变量数目众多，导致算法求解时计算量很大，因此需要对算法进行改进，加快其收敛速度。本发明选择了带有三角变异的改进差分进化算法，该方法被证明在提高算法收敛速度方面具有显著成效，其改进的变异策略可以表示为：
r′＝(r 1+r 2+r 3)/3+(p 2-p 1)(r 1-r 2)+(p 3-p 2)(r 2-r 3)+(p 1-p 3)(r 3-r 1)   (17)
其中
Figure PCTCN2019124316-appb-000014
p′＝|f(r 1)|+|f(r 2)|+|f(r 3)|    (19)
指纹数据库实时校正的简化流程图如图3所示。
本发明提到的上述特征，或实施例提到的特征可以任意组合。本案说明书所揭示的所有特征可与任何组合物形式并用，说明书中所揭示的各个特征，可以任何可提供相同、均等或相似目的的替代性特征取代。因此除有特别说明，所揭示的特征仅为均等或相似特征的一般性例子。
本发明的主要优点在于：
1、综合考虑当前10集总模型情况、原料工业分析现状以及精馏塔要求，建立原料指纹数据库和焦化反应产物解集总方法，提高集总模型在工业上的实用性，实现反应模型与分馏模型的集成。
2、本发明为延迟焦化全过程模拟集成提供有效、可靠的思路，同时显著提高模型在工业上的实用性。
下面结合具体实施例，进一步阐述本发明。应理解，这些实施例仅用于说明本发明而不用于限制本发明的范围。下列实施例中未注明具体条件的实验方法，通常按照常规条件或按照制造厂商所建议的条件。除非另外说明，否则所有的百分数、比率、比例、或份数按重量计。本发明中的重量体积百分比中的单位是本领域技术人员所熟知的，例如是指在100毫升的溶液中溶质的重量。除非另行定义，文中所使用的所有专业与科学用语与本领域熟练人员所熟悉的意义相同。此外，任何与所记载内容相似或均等的方法及材料皆可应用于本发明方法中。文中所述的较佳实施方法与材料仅作示范之用。
实施例
1.原料指纹数据
根据式(1)将馏程、密度、硫含量、残碳、氮含量与四组分之间定量关联：
y＝y 0+k*Δx    (1)
式中y 0为基准，k为变化率，Δx为参数变化量。
指纹数据中对四组分的定量关联式如下：
0％点
Figure PCTCN2019124316-appb-000015
5％点
Figure PCTCN2019124316-appb-000016
10％点
Figure PCTCN2019124316-appb-000017
30％点
Figure PCTCN2019124316-appb-000018
50％点
Figure PCTCN2019124316-appb-000019
70％点
Figure PCTCN2019124316-appb-000020
90％点
Figure PCTCN2019124316-appb-000021
密度(ρ，20℃)
Figure PCTCN2019124316-appb-000022
残碳(CCR)
Figure PCTCN2019124316-appb-000023
硫含量(S)
Figure PCTCN2019124316-appb-000024
氮含量(N)
Figure PCTCN2019124316-appb-000025
式1-12中，Sat，Aro，Res，Asp分别表示饱和分、芳香分、胶质和沥青质。下标0表示基准值，Sat，Aro，Res，Asp的基准值分别为0.5051、0.1603、0.1079和0.2266。Δ表示与基准性质(表6)的偏差值。
表6原料指纹数据
Figure PCTCN2019124316-appb-000026
2.产物指纹数据库
(1)焦化产物全组分列表如表2所示；
表2焦化产物全组分列表
Figure PCTCN2019124316-appb-000027
(2)形成各产物的指纹数据组分表，如表3所示。气体主要包含硫化氢、 氢气、C1、C2和C3，液化气则主要包含C3和C4，汽油馏程为C5-230℃，柴油馏程为180℃-380℃，蜡油馏程为350℃-终馏点。
表3焦化产物指纹数据组分表
Figure PCTCN2019124316-appb-000028
(3)选取某工厂延迟焦化装置某个月的实际数据，当月装置的操作工况为：进料负荷235吨/h、反应温度499.3℃、反应压力0.36MPag、产品包含气体、液化气、汽油、柴油、蜡油和焦炭。根据这些条件和实验室分析结果，结合以下拟合公式，计算后的产物指纹数据如表4所示。
30％温度点(T 30％)计算公式：
T 30％＝(T 10％+T 50％)/2     (12)
70％温度点(T 70％)计算公式：
T 70％＝(T 50％+T 90％)/2     (13)
收率与温度拟合多项式：
y＝Ax 5+Bx 4+Cx 3+Dx 2+Ex+F     (14)
Y为累计体积收率，A、B、C、D、E、F为多项式参数(无具体意义)，x为温度。
表4焦化产物指纹数据
Figure PCTCN2019124316-appb-000029
3.原料指纹数据校正
某炼油厂渣油延迟焦化装置某周的原料实际平均数据，原料的宏观性质为：馏程(IBP:555；5％:559；10％:562；30％:584；50％:606；70％:637；90％:735；)、密度(20℃)为0.9748、残碳为20％、硫含量为1.9％、氮含量为7920.1ppm。 根据上述条件，由算法求解得到校正后的四组分输出数据如表5所示，同时得到校正的指纹数据如表6所示。
表5校正后四组分预测值与实际值对比
Figure PCTCN2019124316-appb-000030
上述方法以指纹数据为基础，建立原料宏观性质，包括实沸点蒸馏数据，密度、硫含量和残碳，与饱和分、芳香分、胶质和沥青质含量的定量关联，实现工业现场数据与延迟焦化十集总模型的集成；同时，结合焦化产物实验室分析数据，建立集总动力学中反应产物与实际装置产物详细组成之间的指纹数据关联，实现动力学模型与精馏模型之间的集成。集成过程中，对采集得到的工业现场数据需采用数据调和技术进行处理，结合改进后的差分进化算法，对原料指纹数据进行校正，使指纹数据能够在宏观性质基础上准确预测四组分信息。
以上所述仅为本发明的较佳实施例而已，并非用以限定本发明的实质技术内容范围，本发明的实质技术内容是广义地定义于申请的权利要求范围中，任何他人完成的技术实体或方法，若是与申请的权利要求范围所定义的完全相同，也或是一种等效的变更，均将被视为涵盖于该权利要求范围之中。

Claims (10)

  1. 一种延迟焦化模型集成方法，其特征在于，所述方法包括步骤：
    (1)使延迟焦化装置原料宏观性质与原料中的组分含量之间的指纹数据关联；
    (2)集成延迟焦化反应动力学模型得到焦化产品收率；
    (3)根据延迟焦化产品数据与步骤(2)的得到的收率拟合形成全馏分延迟焦化产物的指纹数据；
    (4)集成分馏模型实现气体、液化气、汽油、柴油和蜡油的分离。
  2. 如权利要求1所述的方法，其特征在于，步骤(1)中所述的原料宏观性质包括馏程，密度、硫含量、残碳和氮含量；所述原料中的组分是四组分，为饱和分、芳香分、胶质和沥青质。
  3. 如权利要求1所述的方法，其特征在于，步骤(2)是利用动力学模型，根据原料宏观性质和操作条件计算焦化产品收率；优选10集总动力学模型。
  4. 如权利要求3所述的方法，其特征在于，所述操作条件包括反应温度和反应压力。
  5. 如权利要求1所述的方法，其特征在于，步骤(3)所述焦化产品性质数据包括气体组成，液化气组成，汽油实沸点蒸馏数据、柴油实沸点蒸馏数据和蜡油实沸点蒸馏数据；所述拟合采用ASPEN原油表征工具进行。
  6. 如权利要求1所述的方法，其特征在于，步骤(4)是利用步骤(3)生成的全馏分延迟焦化产物作为原料，输入至精馏塔模型，实现气体、液化气、汽油、柴油和蜡油的分离。
  7. 如权利要求1-6任一项所述的方法，其特征在于，所述方法还包括步骤：校正步骤(1)中所述的指纹数据。
  8. 如权利要求7所述的方法，其特征在于，利用原料宏观性质以及四组分数据，结合带有三角变异的差分进化算法对指纹数据进行校正。
  9. 一种如权利要求1-8任一项所述的方法在延迟焦化模拟和/或优化方面的应用。
  10. 如权利要求9所述的应用，其特征在于，所述模拟和/或优化包括延迟焦化过程模型开发、装置尺度优化、和生产计划优化模型校核。
PCT/CN2019/124316 2019-02-01 2019-12-10 一种延迟焦化模型集成方法 Ceased WO2020155865A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910104614.7 2019-02-01
CN201910104614.7A CN109817287B (zh) 2019-02-01 2019-02-01 一种延迟焦化模型集成方法

Publications (1)

Publication Number Publication Date
WO2020155865A1 true WO2020155865A1 (zh) 2020-08-06

Family

ID=66606354

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/124316 Ceased WO2020155865A1 (zh) 2019-02-01 2019-12-10 一种延迟焦化模型集成方法

Country Status (2)

Country Link
CN (1) CN109817287B (zh)
WO (1) WO2020155865A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109817287B (zh) * 2019-02-01 2023-08-11 华东理工大学 一种延迟焦化模型集成方法
CN115831250B (zh) * 2023-02-20 2023-06-06 新疆独山子石油化工有限公司 一种延迟焦化反应模型构建方法及装置、存储介质及设备
CN115862759B (zh) * 2023-02-20 2023-06-06 新疆独山子石油化工有限公司 一种延迟焦化反应优化方法及装置、存储介质及设备
CN116543851B (zh) * 2023-05-05 2025-05-27 西南石油大学 一种基于四组分模型的稠油裂解转化生焦预测方法
CN116609161A (zh) * 2023-05-24 2023-08-18 山东联化新材料有限责任公司 一种研究焦化不完全焦的分析方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007124002A2 (en) * 2006-04-21 2007-11-01 Exxonmobil Research And Engineering Company Application of abnormal event detection technology to delayed coking unit
CN104765346A (zh) * 2015-03-26 2015-07-08 华东理工大学 一种炼油过程全流程建模方法
CN104765347A (zh) * 2015-03-26 2015-07-08 华东理工大学 一种渣油延迟焦化过程中收率实时预测方法
CN105975685A (zh) * 2016-05-03 2016-09-28 华东理工大学 一种用于渣油延迟焦化过程的建模和优化方法
CN109817287A (zh) * 2019-02-01 2019-05-28 华东理工大学 一种延迟焦化模型集成方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE3325482A1 (de) * 1983-07-14 1985-01-24 Diamond West Energy Corp., Salt Lake City, Utah Entschwefelung von durch verzoegertes verkoken hergestelltem petrolkoks
US9665693B2 (en) * 2012-05-30 2017-05-30 Exxonmobil Research And Engineering Company System and method to generate molecular formula distributions beyond a predetermined threshold for a petroleum stream
CN104789256B (zh) * 2015-03-26 2016-07-06 华东理工大学 一种催化裂化装置的收率实时预测方法
CN106773674A (zh) * 2016-11-24 2017-05-31 武汉理工大学 一种延迟焦化过程多目标实时优化方法及系统
CN107942666B (zh) * 2017-11-27 2021-02-09 中国石油化工股份有限公司 基于面向方程法的延迟焦化吸收稳定优化系统
CN108520169B (zh) * 2018-04-09 2022-10-14 山西汾渭能源开发咨询有限公司 一种通过灰成分与挥发分预测焦炭硫转化方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007124002A2 (en) * 2006-04-21 2007-11-01 Exxonmobil Research And Engineering Company Application of abnormal event detection technology to delayed coking unit
CN104765346A (zh) * 2015-03-26 2015-07-08 华东理工大学 一种炼油过程全流程建模方法
CN104765347A (zh) * 2015-03-26 2015-07-08 华东理工大学 一种渣油延迟焦化过程中收率实时预测方法
CN105975685A (zh) * 2016-05-03 2016-09-28 华东理工大学 一种用于渣油延迟焦化过程的建模和优化方法
CN109817287A (zh) * 2019-02-01 2019-05-28 华东理工大学 一种延迟焦化模型集成方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
TIAN LIDA, SHEN BENXIAN, LIU JICHANG: "Study on Application of Structure-Oriented Lumping Model for Delayed Coking", PETROLEUM REFINERY ENGINEERING, vol. 42, no. 5, 31 May 2012 (2012-05-31), CN, pages 55 - 60, XP009522466, ISSN: 1002-106X *

Also Published As

Publication number Publication date
CN109817287A (zh) 2019-05-28
CN109817287B (zh) 2023-08-11

Similar Documents

Publication Publication Date Title
WO2020155865A1 (zh) 一种延迟焦化模型集成方法
CN105229627B (zh) 用于确定产品流或处理流的属性的系统、方法和设备
JP5116669B2 (ja) 石油精製用原料のコークスおよび瀝青の品質の評価方法
US7257451B2 (en) Method for creating a linear programming model of an industrial process facility
JP5460319B2 (ja) ほぼリアルタイムでのストリームの組成および特性の予測
Baroni et al. The applicability of isoconversional models in estimating the kinetic parameters of biomass pyrolysis
CN102187202A (zh) 使用近红外光谱预测烃工艺流稳定性的方法
CN104765346A (zh) 一种炼油过程全流程建模方法
Fu et al. Hybrid model for optimization of crude oil distillation units
CN113627037A (zh) 一种基于原油切割的直馏沥青性质预测方法
RS53390B (sr) Postupak predviđanja kvaliteta i prinosa sirove nafte
CN104765347B (zh) 一种渣油延迟焦化过程中收率实时预测方法
Chelgani et al. Estimation of free-swelling index based on coal analysis using multivariable regression and artificial neural network
Khalaf Steady state simulation of Basrah crude oil refinery distillation unit using aspen hysys
CN109507352A (zh) 一种预测石化生产中任意流股分子组成的方法
Agarwal et al. Prediction of coke CSR using time series model in Coke Plant
Reiter et al. Characterization of crude oil by real component surrogates
Saud et al. Study and optimization of the factors affecting the crude oil distillation process using ASPEN HYSYS
Ramli et al. Online composition prediction of a debutanizer column using artificial neural network
WO2017062416A1 (en) Hybrid analyzer for fluid processing processes
CA2740269A1 (en) Hybrid models of multi-component vapor liquid separation equipment
Maheshwari et al. Correlations for Pour Point and Cloud Point of middle and heavy distillates using density and distillation temperatures
CN121014082A (zh) 确定用于热解过程的反应模型的反应参数的值
CN116463142B (zh) 一种石油催化裂化方案获取方法及装置
Idris et al. Simulation studies on vacuum distillation column (VDU) using the application of Aspen Hysys v8. 4

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19914113

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19914113

Country of ref document: EP

Kind code of ref document: A1