Method for establishing risk management model of hydrogen-doped natural gas pipeline

Abstract

The invention discloses a method for establishing a risk management model of a hydrogen-doped natural gas pipeline. Firstly, a Bow-Tie model system is used for analyzing a failure reason and an accident consequence development scene of the hydrogen-doped natural gas pipeline; then, the model is mapped into a Bayesian network by utilizing GeNIe software; and calculating the prior probability of the basic event through an expert evaluation method and a fuzzy set theory. Setting a conditional probability table of the Bayesian network by using a fault tree mapping method and a historical data calculation method; on the basis, simulating an equipment component failure process through a Markov model, introducing hydrogen corrosion and hydrogen-assisted fatigue crack propagation models to carry out hydrogen-related dynamic analysis, and finishing parameter setting of a dynamic Bayesian network; and finally, calculating the failure probability of the hydrogen-doped natural gas pipeline and a potential life loss result when an accident occurs in combination with a model output result. The method adopts the dynamic Bayesian network to predict the pipeline failure probability, is more in line with the actual operation condition, and can be widely applied to the risk management process of the hydrogen-doped natural gas pipeline.

Description

Technical Field

[0001] The present invention belongs to the field of theoretical research on hydrogen-blended natural gas pipelines and relates to a method for establishing a risk management model for hydrogen-blended natural gas pipelines. Background Art

[0002] As a clean energy carrier, hydrogen has attracted widespread attention for its advantages, including zero carbon emissions and high energy density, which can significantly reduce environmental pollution. However, due to its unique physical and chemical properties, such as its extremely low molecular weight, high diffusivity, and flammability and explosiveness, hydrogen faces high risks during storage and transportation. As critical infrastructure in the hydrogen energy supply chain, the safe operation of hydrogen pipelines will directly impact the sustainable development of the entire hydrogen energy industry.

[0003] Building new pipelines requires significant time and cost. Utilizing the existing natural gas pipeline network is the most cost-effective option. An increasing number of existing natural gas pipelines are being considered for conversion to hydrogen blending. This requires a comprehensive risk assessment of the converted pipelines to ensure their safety and reliability.

[0004] However, the existing risk models for hydrogen-blended natural gas pipelines are mostly static models, which differ greatly from actual operating conditions and lack a complete analysis of the consequences of pipeline failure. Summary of the Invention

[0005] To address the above technical issues, the present invention proposes a method for establishing a risk management model for hydrogen-blended natural gas pipelines. This method fully considers various factors contributing to pipeline failure. By coupling a Markov model with a hydrogen-related model, and employing a dynamic Bayesian network, it predicts pipeline failure probability, which is more consistent with actual operating conditions. It also analyzes the consequences of pipeline failure and constructs a risk management model, which has broader application prospects.

[0006] In order to achieve the above technical objectives, the present invention adopts the following technical means:

[0007] A method for establishing a risk management model for a hydrogen-blended natural gas pipeline comprises the following steps:

[0008] S1. Establish a fault tree model with hydrogen-blended natural gas pipeline failure as the top event and an event tree model for hydrogen-blended natural gas pipeline failure, and combine the two to establish a Bow-Tie model for hydrogen-blended natural gas pipeline failure;

[0009] S2. Use GeNIe software to map the Bow-Tie model into a Bayesian network;

[0010] S3. Using the expert evaluation method and fuzzy set theory to set the prior probability of basic events, the occurrence probability is defined for the parent node of the Bayesian network mapped in step S2. The strength relationship between the nodes connected by directed edges in the Bayesian network is expressed by conditional probability. The conditional probability is determined using the logic gates of the fault tree and the fault data provided by the PHMSA database;

[0011] S4: Dynamically make the Bayesian network obtained after step S3 dynamic. The change of the conditional probability of the dynamic Bayesian network depends on the dynamic analysis process, which specifically includes the following sub-steps:

[0012] S41. Use the Markov model to describe the change in the failure probability of equipment components. The variable state of the equipment component at the next moment is only related to the state variable at the current moment. The relationship between them is obtained through the transition probability matrix.

[0013] S42. The PCORRC model modified by HE is used to predict the burst pressure change of hydrogen-blended natural gas pipelines.

[0014] S43. Use the Paris formula and Irwin model to simulate the fatigue crack growth process of hydrogen-blended natural gas pipelines under the influence of hydrogen environment;

[0015] S44. Considering the errors in the measurement of initial defects during corrosion deepening and crack propagation, the errors are converted into the conditional probabilities of pipeline failure due to corrosion deepening or crack growth under different operating times. The conditional probabilities are input into the Bayesian network to obtain a dynamic model of hydrogen-blended natural gas pipeline failure.

[0016] S5. Use the hydrogen-blended natural gas pipeline failure model constructed in step S44 to output the failure probability of the hydrogen-blended natural gas pipeline in different years and the probability of occurrence of different accident consequences. Use the output results to calculate the potential loss of life, and combine the risk judgment matrix to perform risk management on the hydrogen-blended natural gas pipeline.

[0017] Beneficial Effects: The proposed hydrogen-blended natural gas pipeline risk management model, based on a Bayesian network, comprehensively considers various factors influencing pipeline failure. It utilizes expert evaluation and fuzzy set theory to calculate prior probabilities. Given the limited data available on hydrogen-blended natural gas pipeline failures, rigorous mathematical methods are employed to more accurately determine the probability of an event occurring. The Noisy-Max model is introduced to simplify the setting of conditional probabilities, integrating historical data with logic gate relationships to create a conditional probability table. Furthermore, the Markov model, hydrogen corrosion model, and hydrogen-assisted fatigue crack growth model are employed to study the dynamics of pipeline failure, providing a more realistic understanding.

[0018] In an optional embodiment, in step S1, by consulting relevant literature, examining the actual operating environment of the hydrogen-blended natural gas pipeline, and referring to the existing natural gas pipeline database, the causes of failure of the hydrogen-blended natural gas pipeline are systematically analyzed. Factors causing failure of the hydrogen-blended natural gas pipeline include: corrosion, natural forces, third parties, incorrect operation, pipeline design and construction problems, equipment failure, and hydrogen-induced damage. A fault tree model with failure of the hydrogen-blended natural gas pipeline as the top event is established. The consequence scenarios of failure of the hydrogen-blended natural gas pipeline are analyzed. Failure of the hydrogen-blended natural gas pipeline may cause flash fire, jet fire, and explosion. An event tree model of failure of the hydrogen-blended natural gas pipeline is established. The two factors are combined to establish a Bow-Tie model of failure of the hydrogen-blended natural gas pipeline.

[0019] Beneficial Effects: By comprehensively considering various factors that lead to the failure of hydrogen-blended natural gas pipelines, the Bow-Tie model achieves full risk control from "source cause" to "end impact". By analyzing the consequences of hydrogen-blended natural gas pipeline failures, the applicability of the model has been significantly improved.

[0020] In an optional embodiment, in step S3, the expert weights are calculated using the AHP method, the scoring results are fuzzified using a fuzzy function, the fuzzy numbers of each event are linearly integrated, and the fuzzification is performed using the left-right fuzzy ranking method to obtain the prior probability.

[0021] Beneficial effect: In the case of a lack of failure data on hydrogen-blended natural gas pipelines, the probability of event occurrence can be obtained more accurately using rigorous mathematical methods.

[0022] In an optional embodiment, in step S4, the device component is set to four states, namely, normal state, first degraded state, second degraded state and complete failure state. The four states of the device component cannot be transformed to the next higher state, and the state transition simulates the aging process of the device component through the transition probability matrix.

[0023] Beneficial effect: Using the Markov model to describe the aging process of equipment components is more consistent with actual operating conditions.

[0024] In an optional embodiment, in step S44, there is uncertainty in hydrogen corrosion and hydrogen-assisted fatigue crack propagation, that is, there is a measurement error in the initial defect. Assuming that there are 10,000 defects that meet the error distribution at the initial moment, the numerical simulation model is used to count the proportion of the number of failure defects at different moments, and the initial defect observation error is converted into a conditional probability parameter in the dynamic Bayesian network.

[0025] Beneficial effects: Using hydrogen corrosion and hydrogen crack propagation models to drive the change in failure probability of hydrogen-blended natural gas pipelines reflects the actual situation that pipeline cracks or corrosion defects become more threatening over time, so the constructed dynamic Bayesian network model is more accurate.

[0026] In summary, the hydrogen-blended natural gas pipeline risk management model established in the present invention can provide scientific guidance for the design, manufacturing and operation of hydrogen-blended natural gas pipelines, and reduce the loss of property and life caused by pipeline failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flow chart of the method for risk management of hydrogen-blended natural gas pipelines established for the present invention;

[0028] Figure 2 This is a Bayesian network diagram mapped using GeNIe software;

[0029] Figure 3 The hydrogen-blended natural gas pipeline model constructed for the present invention outputs pipeline failure diagrams under different hydrogen blending ratios. DETAILED DESCRIPTION

[0030] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Example 1: The purpose of the present invention is to establish a risk model for hydrogen-blended natural gas pipelines, such as Figure 1 As shown, the specific steps are:

[0032] (1) Bow-Tie modeling of hydrogen-blended natural gas pipeline failure: By reviewing relevant literature, examining the actual operating environment of the pipeline, and referring to the existing natural gas pipeline database, we systematically analyzed the causes of pipeline failure. Factors that cause pipeline failure include corrosion, third parties, incorrect operation, and hydrogen damage. A fault tree model with hydrogen-blended natural gas pipeline failure as the top event was established. The consequences of pipeline failure were analyzed. Pipeline failure may cause flash fire, jet fire, and explosion. An event tree model for hydrogen-blended natural gas pipeline failure was established. The two were combined to establish a Bow-Tie model for hydrogen-blended natural gas pipeline failure.

[0033] (2) Bayesian network mapping: Use GeNIe software to map the Bow-Tie model to a Bayesian network. When the fault tree is mapped to a Bayesian network, the basic events, intermediate events, and top events correspond to the parent nodes, intermediate nodes, and target nodes in the Bayesian network. The hydrogen-blended natural gas pipeline failure node and the accident development path node point to the accident consequence node. The different states defined in the accident consequence node represent different types of accident consequences, such as Figure 2 shown.

[0034] (3) Bayesian network parameter setting: The prior probability of basic events is set using expert evaluation method and fuzzy set theory. In response to the failure problem of hydrogen-blended natural gas pipelines, five experts in related fields were selected, and the seven-level language variable "very low, low, relatively low, general, relatively high, high, very high" was used as the semantic evaluation that the experts may give. The basic events in the Bayesian network were scored three times, with continuous feedback and correction, and the final round of scoring results were adopted. Then, the expert weights were calculated using the AHP method, the scoring results were fuzzified using fuzzy functions, the fuzzy numbers of each event were linearly integrated, and the fuzzification was performed using the left and right fuzzy ranking method to obtain the prior probability. In the Bayesian network model, the conditional probability is determined using the logic gate of the fault tree. The introduction of the Noisy-Max model can effectively reduce the difficulty of parameter setting, and the conditional probability of some nodes is calculated using the historical data provided by PHMSA.

[0035] (4) Dynamic Analysis Process: A Markov model is used to describe the change in the failure probability of equipment components. The equipment components related to the pipeline are set to four states. The variable state of the equipment component at the next moment is only related to the state variable at the current moment. The relationship between them is obtained through the transition probability matrix.

[0036] (5) The PCORRC model with hydrogen embrittlement correction is used to predict the changes in the bursting pressure of hydrogen-blended natural gas pipelines under the influence of HE.

[0037] d=d0+0.5t

[0038]

[0039] Where th is the pipe wall thickness, d is the corrosion depth, l is the corrosion length, d0 is the initial defect depth, t is the annual corrosion growth rate, P H is the hydrogen partial pressure, and SMTS is the specified minimum tensile strength of pipeline steel.

[0040] Paris formula and Irwin model are used to simulate the fatigue crack growth process of pipeline under the influence of hydrogen.

[0041]

[0042] The specific parameters C and m are obtained from relevant literature, a and c represent the depth and length of the crack, n represents the number of cycles, Q represents the shape parameter of the crack, Δσ represents the applied stress difference, ΔK represents the stress intensity factor, φ represents the parameter angle of the elliptical crack, σ y is the yield strength of the pipe, and φ represents the elliptic integral of the second kind.

[0043] Taking into account the uncertainty in the prediction process, that is, the errors in the initial defect observation, it is assumed that there are 10,000 initial defects that conform to the error distribution at the initial moment. Using the numerical simulation model, the proportion of the number of failure defects in each time segment to the total number of defects is counted and converted into conditional probability parameters in the dynamic Bayesian network.

[0044] (6) Model output result processing: The constructed hydrogen-blended natural gas pipeline failure model outputs the pipeline failure probability and the probability of occurrence of different accident consequences in different years. The output results are used to calculate the potential loss of life and, combined with the risk judgment matrix, to conduct risk management of the pipeline.

[0045] The constructed hydrogen-blended natural gas pipeline model outputs the pipeline failure probability under different hydrogen blending ratios as follows: Figure 3 As shown in the figure, compared with the traditional static model, this model can realize the dynamic change of pipeline failure probability over time, which is more in line with the actual operation of the pipeline.

Claims

1. A method for establishing a risk management model for hydrogen-blended natural gas pipelines, characterized in that: The following steps are involved: S1. Establish a fault tree model with hydrogen-blended natural gas pipeline failure as the top event and an event tree model for hydrogen-blended natural gas pipeline failure, and combine the two to establish a Bow-Tie model for hydrogen-blended natural gas pipeline failure; S2. Use GeNIe software to map the Bow-Tie model into a Bayesian network; S3. Using the expert evaluation method and fuzzy set theory to set the prior probability of basic events, the occurrence probability is defined for the parent node of the Bayesian network mapped in step S2. The strength relationship between the nodes connected by directed edges in the Bayesian network is expressed by conditional probability. The conditional probability is determined using the logic gates of the fault tree and the fault data provided by the PHMSA database; S4: Dynamically make the Bayesian network obtained after step S3 dynamic. The change of the conditional probability of the dynamic Bayesian network depends on the dynamic analysis process, which specifically includes the following sub-steps: S41. Use the Markov model to describe the change in the failure probability of equipment components. The variable state of the equipment component at the next moment is only related to the state variable at the current moment. The relationship between them is obtained through the transition probability matrix. S42. The PCORRC model modified by HE is used to predict the burst pressure change of hydrogen-blended natural gas pipelines. S43. Use the Paris formula and Irwin model to simulate the fatigue crack growth process of hydrogen-blended natural gas pipelines under the influence of hydrogen environment; S44. Considering the errors in the measurement of initial defects during corrosion deepening and crack propagation, the errors are converted into the conditional probabilities of pipeline failure due to corrosion deepening or crack growth under different operating times. The conditional probabilities are input into the Bayesian network to obtain a dynamic model of hydrogen-blended natural gas pipeline failure. S5. Use the hydrogen-blended natural gas pipeline failure model constructed in step S44 to output the failure probability of the hydrogen-blended natural gas pipeline in different years and the probability of occurrence of different accident consequences. Use the output results to calculate the potential loss of life, and combine the risk judgment matrix to perform risk management on the hydrogen-blended natural gas pipeline.

2. The method for establishing a risk management model for a hydrogen-blended natural gas pipeline according to claim 1, characterized in that: In step S1, by consulting relevant literature, examining the actual operating environment of the hydrogen-blended natural gas pipeline, and referring to the existing natural gas pipeline database, the causes of failure of the hydrogen-blended natural gas pipeline are systematically analyzed. Factors causing failure of the hydrogen-blended natural gas pipeline include: corrosion, natural forces, third parties, incorrect operation, pipeline design and construction problems, equipment failure, and hydrogen-induced damage. A fault tree model with failure of the hydrogen-blended natural gas pipeline as the top event is established. The consequence scenarios of failure of the hydrogen-blended natural gas pipeline are analyzed. Failure of the hydrogen-blended natural gas pipeline may cause flash fire, jet fire, and explosion. An event tree model of failure of the hydrogen-blended natural gas pipeline is established. The two factors are combined to establish a Bow-Tie model of failure of the hydrogen-blended natural gas pipeline.

3. The method for establishing a risk management model for hydrogen-blended natural gas pipelines according to claim 1, characterized in that: In step S3, the expert weights are calculated using the AHP method, the scoring results are fuzzified using a fuzzy function, the fuzzy numbers of each event are linearly integrated, and the fuzzification is performed using the left-right fuzzy ranking method to obtain the prior probability of the basic events.

4. The method for establishing a risk management model for a hydrogen-blended natural gas pipeline according to claim 1, characterized in that: In step S4, the device component is set to four states, namely normal state, first degraded state, second degraded state and complete failure state. The four states of the device component cannot be transformed to the next higher state. The state change process of the device component is simulated by the transition probability matrix.

5. The method for establishing a risk management model for a hydrogen-blended natural gas pipeline according to claim 1, characterized in that: In step S44, there is uncertainty in hydrogen corrosion and hydrogen-assisted fatigue crack propagation, that is, there is a measurement error in the initial defect. Assuming that there are 10,000 defects that meet the error distribution at the initial moment, the numerical simulation model is used to count the proportion of the number of failure defects at different moments, and the initial defect observation error is converted into a conditional probability parameter in the dynamic Bayesian network.

Images