Crude oil pretreatment system optimization design method based on digital twinning

Abstract

The invention discloses a crude oil pretreatment system optimization design method based on digital twinning, and the method comprises the steps: constructing a digital twinning model comprising a design module, an operation module and a maintenance module, and combining a fault tree analysis and sensitivity analysis method, and precisely recognizing a key fault event which has the greatest influence on the system performance. Meanwhile, a case knowledge base is constructed based on operation and maintenance data, and targeted guidance is provided for design optimization. According to the method, through a closed-loop feedback path from the maintenance stage to the design stage, the high-frequency fault rate is reduced, the system operation efficiency and the design adaptability are improved, the reliability of near infrared spectrum detection is improved while the design of the crude oil pretreatment system is optimized, and the method has important industrial application value.

Description

Technical Field

[0001] The present invention relates to the field of near-infrared rapid analysis of crude oil in refining enterprises, and specifically to an optimization design method for a crude oil pretreatment system based on digital twin. Background Technique

[0002] The crude oil pretreatment system is an important device for quickly analyzing and evaluating the properties of crude oil in refining enterprises. It is used to preliminarily process crude oil samples to meet the detection conditions of near-infrared (NIR) spectroscopy. The stable operation of the pretreatment system plays an important role in ensuring the accurate evaluation of crude oil properties and improving the efficiency of downstream processes. In refining enterprises, the design of the crude oil pretreatment system has always been mainly "forward design", and the system model is designed through theoretical analysis and engineering experience. There are certain limitations in this traditional design mode in actual operation: there may be deviations between the design and the actual operating conditions, resulting in frequent failures during the operation stage and increased maintenance costs; at the same time, due to the lack of feedback of operation and maintenance data to the design stage, it is difficult to optimize and iterate the system design according to actual needs.

[0003] In recent years, as an emerging intelligent technology, digital twin technology has been widely studied and preliminarily applied in the field of industrial processes, showing broad application prospects. However, the application of this technology in the petrochemical field is relatively less and the depth is insufficient, mainly concentrated in links such as operation status monitoring and fault maintenance, and it fails to fully utilize maintenance information to guide design optimization, resulting in the fragmentation of data and functions in different stages of the life cycle.

[0004] Therefore, applying digital twin technology to the crude oil pretreatment system, fully combining the data streams in multiple stages such as operation, maintenance, and design within the equipment life cycle, and realizing the feedback optimization of the design stage, can not only effectively improve the pertinence and rationality of system design, but also significantly improve the operation efficiency and economy of the entire system, and has important application value. Summary of the Invention

[0005] Aiming at the problems existing in the background technique, the present invention discloses an optimization design method for a crude oil pretreatment system based on digital twin. By using the digital twin model combined with fault tree analysis and sensitivity analysis methods, key fault events are located, a case knowledge base is constructed through operation and maintenance data, and targeted design optimization guidance suggestions are generated, and finally a cyclic feedback from fault data to optimized design is realized. The method has the following steps:

[0006] 1) Build a digital twin model of the crude oil pretreatment system. The model includes a design module, an operation module, and a maintenance module. The data of each module is integrated into a unified platform. The design module supports design optimization iteration and establishes a 3D visualization model, which can switch different views to display operation and maintenance data according to different components. The operation module collects real-time operation data through sensors and transmits it into the model, and fuses the data into the 3D model to achieve operation visualization. The maintenance module includes a fault detection and a fault diagnosis model, which collects maintenance data information as the basis for optimizing the design;

[0007] 2) Build a fault tree FT of the crude oil pretreatment system based on system mechanism and engineering experience to form a fault propagation path, specifically as follows:

[0008] a) Identify the fault sources of the crude oil pretreatment system during the operation stage as the basic events of the FT

[0009] n 1 is the number of fault sources;

[0010] b) Determine the top event TE, which is the system fault for the crude oil pretreatment system;

[0011] c) According to the hierarchical relationship, perform a tree decomposition layer by layer from the top event to the intermediate events to the basic events. n 2 is the number of intermediate events, and the intermediate events directly connected to the top event are fault events

[0012] n 3 is the number of fault events;

[0013] d) Analyze the logical relationship between events and connect them through logic gates to form the fault tree FT;

[0014] 3) Sensitivity analysis of basic events. Assign an initial probability of occurrence to the basic events according to engineering experience Calculate the event probability according to the logic gates in the FT from bottom to top. Adjust the initial probability of the basic events in turn with a probability increment of 0.10, and calculate the change in the probability of the fault event from bottom to top. Finally, obtain the fault event FE i For the basic event BE j The sensitivity index of:

[0015]

[0016] In the formula is the sensitivity of the fault event FE i to the basic event BE j , is the probability increment of BE j , and They are BE respectively j FE before and after probability adjustment i Probability, and then trace the root cause basic events of fault events based on sensitivity indicators;

[0017] 4) Establish a case knowledge base based on the optimization and operation and maintenance data of the crude oil pretreatment system, store the existing material and structure improvement cases of each component in the system, and the data comes from enterprise historical data and external data, including operation data, fault detection and diagnosis data, and design optimization records, and construct a design optimization guidance library for different fault scenarios;

[0018] 5) During the operation of the crude oil pretreatment system, record the information of each fault event that causes the near-infrared spectrum of the crude oil sample to be abnormal according to the operation and maintenance information of the digital twin model, and count the fault frequency;

[0019] 6) For the fault event FE now occurring during operation, update the corresponding fault frequency and judge whether it exceeds the fault frequency threshold e FE given based on historical maintenance data. If it does not exceed the threshold, start the maintenance process; otherwise, optimize the design:

[0020] a) Trace the root cause basic events through FT sensitivity analysis, and construct the feature vector C now of the current fault FE now =[f 1 ,f 2 ,…,f n , where f 1 ,f 2 ,…,f n are continuous fault characteristics and n is the number of characteristics;

[0021] b) Calculate the relevance Rel(FE case and FE now ) in the case knowledge base: now ,FE case ):

[0022] Rel(FE now ,FE case ) = Mch(FE now ,FE case ) + Sim(FE now ,FE case )

[0023] In the formula, the matching degree Mch(FE now ,FE case ) is obtained based on the rule: if the fault event and the basic event are both the same, the score is 4; if one of them is the same, the score is 2; if both are different, the score is 0. The similarity Sim(FE now, FE case ) Calculated based on the Euclidean distance:

[0024]

[0025] Where C case is the fault feature vector of FE case Sort the cases in the case knowledge base according to the relevance, and take the top m = 5 cases as the design optimization guidance suggestions;

[0026] c) Update the knowledge base with the current design optimization plan obtained with the assistance of the guidance suggestions as a new case.

[0027] Beneficial effects:

[0028] The present invention discloses an optimization design method for a crude oil pretreatment system based on digital twin. This method uses digital twin technology combined with fault tree and sensitivity analysis to locate key faults and establish a case knowledge base through operation and maintenance data, and generate targeted optimization design plans. This method can achieve a closed-loop feedback from fault data to design optimization, significantly improving the pertinence and rationality of system design; at the same time, combining the case knowledge base with optimization algorithms to provide intelligent design improvement plans for different fault scenarios, thereby effectively reducing the fault occurrence rate, improving the operation efficiency and economy of the crude oil pretreatment system, and having important industrial application value. Description of the Drawings

[0029] Figure 1 is the flow chart of the optimization design method for the crude oil pretreatment system of the present invention;

[0030] Figure 2 is the three-dimensional visualization model of the crude oil pretreatment system in the embodiment of the present invention;

[0031] Figure 3 is the fault tree of the crude oil pretreatment system in the embodiment of the present invention;

[0032] Figure 4 is the sensitivity analysis result of the fault tree of the crude oil pretreatment system in the embodiment of the present invention. Specific Embodiments

[0033] The following further describes the present invention in conjunction with the drawings and specific embodiments. The implementation effect of this method in the design optimization process of the crude oil pretreatment system is described by the specific operation process. This implementation case is implemented on the premise of the technical solution of the present invention, but the protection scope of the present invention is not limited to the following embodiments.

[0034] The present invention takes the crude oil pretreatment system of a refining and chemical enterprise as an example to illustrate the effectiveness and implementation process of the method. This system is used for the rapid near-infrared spectroscopy analysis of the crude oil samples fed into the atmospheric and vacuum distillation unit (CDU). It mainly includes: a fast loop that introduces the crude oil samples collected from the CDU feed pipeline into the pretreatment system; an electrically heated constant temperature oven to heat the samples; a water and impurity removal filter to filter out the moisture and impurities in the samples; a pressure controller to provide a stable pressure for pumping the samples to the flow-through cell; a flow-through cell and a near-infrared spectroscopy detector to detect the absorption spectra of the samples, etc. The flow chart of the method of the present invention is as shown in Figure 1 as follows, and the specific process is as follows

[0035] 1) Build a digital twin model of the crude oil pretreatment system. The model includes a design module, an operation module, and a maintenance module. The data of each module is integrated into a unified platform. Among them, the design module supports design optimization iteration and establishes a three-dimensional visualization model as shown in Figure 2 as follows. Different views are switched according to different components to display the operation and maintenance data. The operation module collects the real-time operation data through sensors and transmits it into the model, and fuses the data into the three-dimensional model to achieve operation visualization. The maintenance module includes a fault detection and a fault diagnosis model, which collects maintenance data information as the basis for optimizing the design. In this embodiment, the fault detection and diagnosis model is provided by the enterprise;

[0036] 2) Build a fault tree FT of the crude oil pretreatment system based on system mechanism and engineering experience to form a fault propagation path, specifically as follows:

[0037] a) Identify the fault sources of the crude oil pretreatment system during the operation stage as the basic events of the FT

[0038] n 1 is the number of fault sources;

[0039] b) Determine the top event TE. In this embodiment, the top event is the failure of the crude oil pretreatment system;

[0040] c) According to the hierarchical relationship, decompose the tree layer by layer from the top event to the intermediate events to the basic events. n 2 is the number of intermediate events. The intermediate events directly connected to the top event are fault events

[0041] n 3 is the number of fault events;

[0042] d) Analyze the logical relationship between events and connect them through logic gates to form a fault tree FT as shown in Figure 3 as follows. The event descriptions are shown in Table 1;

[0043] Table 1 Event descriptions of the fault tree of the crude oil pretreatment system

[0044]

[0045]

[0046] 3) Basic event sensitivity analysis, assign initial probability values to basic events according to engineering experience Calculate the event probability from bottom to top according to the logic gates in the FT, adjust the initial probability of the basic event in turn with a probability increment of 0.10, calculate the change of the fault event probability from bottom to top, and finally obtain the fault event FE i For the basic event BE j The sensitivity index of:

[0047]

[0048] In the formula Is the fault event FE i For the basic event BE j The sensitivity of, Is BE j Probability increment, And Are the FE of BE respectively j Probability before and after probability adjustment, the FT sensitivity analysis of this embodiment is as i Shown, and then trace back to the root cause basic event of the fault event based on the sensitivity index; Figure 4

[0049]

[0049] 4) Establish a case knowledge base based on the optimization and operation and maintenance data of the crude oil pretreatment system, store the existing material and structure improvement cases of each component in the system, and the data comes from enterprise historical data and external data, including operation data, fault detection and diagnosis data, and design optimization records, and construct a design optimization guidance library for different fault scenarios;

[0050] 5) During the operation of the crude oil pretreatment system, record the information of each fault event that causes the abnormal near-infrared spectrum of the crude oil sample according to the operation and maintenance information of the digital twin model, and count the fault frequency;

[0051] 6) In this embodiment, a fault event FE occurs during the operation of the crude oil pretreatment system now , after being diagnosed by the fault diagnosis model as the water content of the crude oil sample, it is found that the fault frequency exceeds the fault frequency threshold e after updating the corresponding fault frequency FE =4, and the design needs to be optimized:

[0052] a) Trace back to the root cause basic event through FT sensitivity analysis, and the basic event with the highest sensitivity to the water content fault of the sample is found to be improper selection of the filter membrane. Combine the operation data to construct the feature vector C of the current fault FE now Of now= [32.1, 43.5, 4.5. The continuous features are injection temperature, sample temperature in the connecting cell, sample flow rate, and static cup bypass flow rate respectively;

[0053] b) Calculate the case FE in the case knowledge base case with FE now the relevance Rel(FE now , FE case ):

[0054] Rel(FE now , FE case ) = Mch(FE now , FE case ) + Sim(FE now , FE case )

[0055] Wherein, the matching degree Mch(FE now , FE case ) is obtained based on the rule: if the failure type and the component are both the same, the score is 2; if one of them is the same, the score is 1; if both are different, the score is 0. The similarity Sim(FE now , FE case ) is calculated based on the Euclidean distance:

[0056]

[0057] Where C case is the failure feature vector of FE case . Sort the cases in the case knowledge base according to the relevance. The top m = 5 cases as the design optimization guidance suggestions are shown in Table 2;

[0058] Table 2 Design Optimization Guidance Cases after Sorting

[0059]

[0060] c) Refer to the given optimization guidance, try to replace the filter material in the water removal filter with a 4.5μm spiral hydrophobic filter membrane, and shorten the pipeline from the static cup to the flow cell. Observe the near-infrared spectrum detection results of the crude oil sample. The spectrum detection becomes stable within 5 minutes. Continuing to track the crude oil pretreatment effect after the improvement of the system components, it is found that the crude oil spectrum can remain stable. Update the knowledge base with the current design optimization plan as a new case.

[0061] As can be seen from the above analysis, the present invention can use the fault tree and sensitivity analysis to find the basic events of the root cause of the fault, and provide guidance for design optimization by establishing a case knowledge base for the crude oil pretreatment system. It can optimize the design specifically when high-frequency faults occur, realizing the closed-loop feedback from the maintenance stage to the design stage. This enables the design stage to achieve optimization iteration, effectively improving the pertinence and rationality of the crude oil pretreatment system design, thereby reducing the failure rate, facilitating subsequent process decision-making and process control, and enhancing production efficiency.

Claims

1. A crude oil pretreatment system optimization design method based on digital twin, characterized in that By building a digital twin model, combining fault tree analysis and sensitivity analysis, identifying key fault events, and establishing a case knowledge base based on operation and maintenance data, generating design optimization guidance suggestions, and realizing closed-loop feedback from fault data to optimized design, the following steps are involved: 1) Build a digital twin model of the crude oil pretreatment system; 2) Based on the system mechanism and engineering experience, the fault tree FT of the crude oil pretreatment system is constructed to form the fault propagation path, as follows: a) Identify the source of failure in the crude oil pretreatment system during the operation phase as the basic event of FT n1 is the number of fault sources; b) determining the top event TE, which is a system failure for the crude oil pretreatment system; c) According to the hierarchical relationship, from the top event to the middle event The basic events are decomposed layer by layer in a tree shape. n2 is the number of intermediate events. The intermediate events directly connected to the top event are fault events. n3 is the number of fault events; d) Analyze the logical relationship between events and connect them through logic gates to form a fault tree FT; 3) Basic event sensitivity analysis, assigning initial values ​​of occurrence probability to basic events According to the logic gates in FT, the event probability is calculated from bottom to top, the initial probability of the basic events is adjusted in sequence, and the change of the probability of the fault event is calculated from bottom to top, and finally the fault event FE is obtained. i For basic event BE j Sensitivity index: In the formula Fault event FE i For basic event BE j sensitivity, BE j Probability increment, and BE j FE before and after probability adjustment i Probability, and then trace the root cause of the failure event based on the sensitivity index; 4) Establish a case knowledge base based on the optimization and operation and maintenance data of the crude oil pretreatment system, store the existing material and structure improvement cases of each component in the system, and generate a design optimization guidance library for different failure scenarios; 5) During the operation of the crude oil pretreatment system, based on the operation and maintenance information of the digital twin model, record the information of each fault event that causes the abnormal near-infrared spectrum of the crude oil sample and count the fault frequency; 6) FE for fault events that occur during operation now , update the corresponding fault frequency and determine whether it exceeds the fault frequency threshold e FE If the threshold is not exceeded, the maintenance process is started, otherwise the design is optimized: a) Trace the root cause basic events through FT sensitivity analysis and construct the current fault FE based on the operation data now The eigenvector C now =[f1,f2,…,f n ],f1,f2,…,f n is the continuous fault feature, n is the number of features; b) Calculate the case FE in the case knowledge base case with FE now Correlation Rel(FE now ,FE case ), sort the cases in the case knowledge base according to their relevance, and take the top m cases as guidance suggestions for design optimization; c) Update the knowledge base with the current design optimization solution obtained with the assistance of guidance suggestions as a new case.

2. The crude oil pretreatment system optimization design method based on digital twin according to claim 1 is characterized in that The digital twin model includes design module, operation module, and maintenance module, and the data of each module are integrated into a unified platform.

3. The crude oil pretreatment system optimization design method based on digital twin according to claim 2 is characterized in that In the digital twin model, the design module supports design optimization iteration and establishes a three-dimensional visualization model, switches different views according to different components, and displays operation and maintenance data.

4. The method for optimizing the crude oil pretreatment system based on digital twin according to claim 2 is characterized in that In the digital twin model, the operation module collects real-time operation data through sensors and transmits it to the model, and fuses the data into the three-dimensional model to achieve operation visualization.

5. The crude oil pretreatment system optimization design method based on digital twin according to claim 2 is characterized in that In the digital twin model, the maintenance module includes fault detection and fault diagnosis models, and collects maintenance data information as the basis for optimized design.

6. The crude oil pretreatment system optimization design method based on digital twin according to claim 1 is characterized in that The knowledge base is constructed by combining enterprise historical data and external data, including operation data, fault detection and diagnosis data, and design optimization records.

7. The crude oil pretreatment system optimization design method based on digital twin according to claim 1 is characterized in that The initial probability of the basic event is given according to engineering experience, and the probability increment is set to 0.

10.

8. The crude oil pretreatment system optimization design method based on digital twin according to claim 1 is characterized in that Frequency threshold e FE Based on historical maintenance data.

9. The crude oil pretreatment system optimization design method based on digital twin according to claim 1 is characterized in that Calculate the correlation Rel(FE now ,FE case ): Rel(FE now ,FE case )=Mch(FE now ,FE case )+Sim(FE now ,FE case ) In the formula, the matching degree Mch(FE now ,FE case ) Based on the rules, the score of the fault event and the basic event is 4 if they are the same, 2 if one of them is the same, and 0 if they are not the same. The similarity Sim(FE now ,FE case ) is calculated based on the Euclidean distance: Where C case For FE case The feature vector of , k represents the kth continuous feature variable in the feature vector.

10. The method for optimizing the crude oil pretreatment system based on digital twin according to claim 1, characterized in that m is set to 5.

Images