Underground parallel pipelines domino effect: An analysis based on pipeline crater models and historical accidents

This paper focuses on the analysis of the possibility of domino effect in underground parallel pipelines relying on historical accident data and pipeline crater models. An underground pipeline can be considered as safe following an accident with an adjacent gas or liquefied pipeline when it remains outside the ground crater generated. In order to prevent the domino effect in these cases, the design of parallel pipelines has to consider adequate pipeline separations based on the crater width, which is one of the widely used methods in engineering applications. The objective of this work is the analysis of underground petroleum product pipelines ruptures with the formation of a ground crater as well as the evaluation of possible domino effects in these cases. A detailed literature survey has been carried out to review existing crater models along with a historical analysis of past accidents. A FORTRAN code has been implemented to assess the performance of the Gasunie, the Batelle and the Advantica crater models. In addition to this, a novel Accident-Based crater model has been presented, which allows the prediction of the crater width as a function of the relevant design pipeline parameters as well as the soil density. Modifications have also been made to the Batelle and Accident-Based models in order to overcome the underestimation of the crater width. The calculated crater widths have been compared with real accident data and the performance evaluation showed that the proposed Accident-Based model has a better performance compared to other models studied in this work. The analysis of forty-eight past accidents indicated a major potential of underground parallel pipelines domino effect which is proven by two real cases taken from the literature. Relying on the investigated accidents, the crater width was smaller than or equal to 20 m in most cases indicating that the definition of underground pipeline separations at around 10 m would be sufficient to ensure a small probability of the domino effect.     

An optimization of artificial neural network modeling methodology for the reliability assessment of corroding natural gas pipelines

A fast calculation of the reliability is meaningful to the in-line inspection of corroding natural gas pipelines. However, the traditional Monte Carlo simulation(MCS) method is time consuming for the low possibilities of the pipeline failure. The artificial neural network(ANN) is preferable for the complex nonlinear situation. An optimization of artificial neural network modeling methodology for the reliability assessment of corroding natural gas pipelines is proposed in this paper. To reduce the influence of training sets random behaviors on the calculation results, some algorithms are used to optimize the sequence of the training samples and the initial parameters of ANN models. The optimized model is applied to the reliability assessment of a corroded pipe with two successive inline inspections. According to the physical parameters of the pipeline, the trend of corroding pipeline reliability in time is predicted. The comparison between the trained ANN model, the MCS method and non-optimized ANN model shows the advantages the proposed modeling process. The methodology given in this paper is general and it can be applied to evaluate the reliability of other kind of structure safeties in practical systems.     

支持向量机对多环芳烃毒性的定量构效预测

为了实现多环芳烃(PAHs)毒性的有效预测,提出应用定量构效技术对多环芳烃的空气-正辛醇分配系数(KOA)和致癌性进行预测。应用分子描述符和试验值确立构效关系,采用支持向量机算法(SVM)和人工神经网络算法(ANN)分别建立了PAHs的KOA回归预测模型和致癌性分类预测模型。利用网格划分(GS)、遗传算法(GA)、粒子群算法(PSO)对SVM进行参数寻优。应用均方误差(MSE)、拟合决定系数R2和分类准确率(Accuracy)分别对模型进行了验证与评价。结果表明,最佳回归预测模型GS-SVR的MSE为0.059 7,R2为0.913 0;最佳分类预测模型GA-SVC的Accuracy为95%。研究表明:应用SVM所建两种模型的稳定性和预测能力都优于应用ANN建立的模型;参数优化后模型的稳定性和预测能力得到了提高。     

Development of HGAPSO-SVR corrosion prediction approach for offshore oil and gas pipelines

Managing the oil and gas pipelines against corrosion is one of the major challenges of the oil and gas sector because of the complexities associated with the initiation, stabilization, and growth of the corrosion defects. The present research attempts to develop a model for predicting the maximum depth of pitting corrosion in oil and gas pipelines using SVM algorithm. In order to improve the SVM performance, Hybrid PSO and GA was utilized. Monte Carlo simulation was used to determine the time lapse for the pit depth growth. In order to implement the above modeling approaches and to prove their efficiency and accuracy against a large database, a total of 340 data samples for corrosion depth and rate are retrieved from the Iranian Oilfields. The performance of the new algorithm shows that it has higher stability and accuracy. In addition, the forecasting results of the new algorithm are compared with the 11 intelligent optimization algorithms, it shows that the novel hybrid algorithm has higher accuracy, better generalization ability, and stronger robustness. The coefficient of determination (R²) value in the testing phase for SVM-HGAPSO was estimated by 0.99. Proposed hybrid model and Monte-Carlo simulations pitting corrosion based on Poisson square wave process have been used to predict the time evolution of the mean value of the pit depth distribution for different categories of maximum pitting rates (low, moderate, high and sever). The models was validated with 4 field data for each of the pitting corrosion categories and the results agreed well. The pipelines under severe pitting corrosion rate were, more conservatively predicted by HGAPSO-SVR than those under low, moderate and high pitting corrosion rates. The results obtained demonstrate the potentials of this technique for the integrity management of corroded aged pipelines.     

基于ISOA-RBPNN的埋地管道剩余强度预测*

为提高腐蚀管道剩余强度的预测精度，提出引入弹性梯度下降法改进BP神经网络，并融合改进海鸥优化算法(ISOA)，构建腐蚀管道剩余强度预测模型。关于改进BP神经网络模型的参数寻优，首先采用Cat混沌映射初始化改进海鸥优化算法(SOA)初始种群的分布，提升寻优能力，优化SOA的搜索方向和攻击形式，增强其全局搜索能力并提高收敛速度，然后用ISOA对弹性BP神经网络（RBPNN）模型中的权值和阈值进行寻优，最后构建ISOA-RBPNN预测模型。以管道爆破数据为例，利用MATLAB进行仿真模拟，并与PSO-BPNN模型和IFA-BPNN模型预测结果进行对比分析。研究结果表明:ISOA-RBPNN模型的各项评价指标均优于其他2个模型，预测结果较实际值误差更小，在预测腐蚀管道剩余强度领域具有更好的性能，可为后续研究腐蚀管道剩余寿命和制定维修策略提供参考依据。     

Prediction of gas explosion pressures: A machine learning algorithm based on KPCA and an optimized LSSVM

A gas explosion, as a common accident in public life and industry, poses a great threat to the safety of life and property. The determination and prediction of gas explosion pressures are greatly important for safety issues and emergency rescue after an accident occurs. Compared with traditional empirical and numerical models, machine learning models are definitely a superior approach. However, the application of machine learning in gas explosion pressure prediction has not reached its full potential. In this study, a hybrid gas explosion pressure prediction model based on kernel principal component analysis (KPCA), a least square support vector machine (LSSVM), and a gray wolf optimization (GWO) algorithm is proposed. A dataset consisting of 12 influencing factors of gas explosion pressures and 317 groups of data is constructed for developing and evaluating the KPCA-GWO-LSSVM model. The results show that the correlations among the 12 influencing factors are eliminated and dimensioned down by the KPCA method, and 5 composite indicators are obtained. The proposed KPCA-GWO-LSSVM hybrid model performs well in predicting gas explosion pressures, with coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE) values of 0.928, 26.234, and 12.494, respectively, for the training set; and 0.826, 25.951, and 13.964, respectively, for the test set. The proposed model outperforms the LSSVM, GWO-LSSVM, KPCA-LSSVM, beetle antennae search improved BP neural network (BAS-BPNN) models and reported empirical models. In addition, the sensitivity of influencing factors to the model is evaluated based on the constructed database, and the geometric parameters X1 and X2 of the confined structure are the most critical variables for gas explosion pressure prediction. The findings of this study can help expand the application of machine learning in gas explosion prediction and can truly benefit the treatment of gas explosion accidents.     

基于遗传-神经网络算法的含均匀腐蚀缺陷油气管线爆破压力预测研究*

为提高含均匀腐蚀缺陷油气管线爆破压力的预测精度，保障长输油气管线的安全运行，将遗传算法和BP神经网络相结合，建立含均匀腐蚀缺陷油气管线爆破压力预测的遗传-BP神经网络（GA-BPNNs）模型。采用已有文献实验数据，分析对比该模型与AGA NG-18，ASME B31G，修正B31G，PCORRC，DNV RP-F101和SHELL 92等方法用于X46，X52，X60，X65，X80等材质油气管线含均匀腐蚀缺陷时爆破压力的计算误差。结果表明:GA-BPNNs模型用于含均匀腐蚀缺陷油气管线爆破压力预测时，误差在-7.78%～6.06%之间，预测精度明显高于目前国内外通用规范的计算结果；该模型操作简单，适用范围广，工程实用性好，为含缺陷压力管道爆破压力的预测提供更好的思路和方案。     

基于MPSO-WLS-SVM的矿井瓦斯涌出量预测模型研究

为有效预防瓦斯灾害,以预测矿井瓦斯涌出量为研究目的,提出经改进的粒子群算法(MPSO)优化的加权最小二乘支持向量机(WLS-SVM),并用其预测非线性动态瓦斯涌出量。算法通过对WLS-SVM的正则化参数C和高斯核参数σ寻优,建立基于MPSO优化的WLS-SVM的瓦斯涌出量预测模型,并利用某矿井监测到的各项历史数据进行实例分析。试验结果表明:该预测模型预测的最大相对误差为5.99%,最小相对误差为0.43%,平均相对误差为2.95%,较其他预测模型有更强的泛化能力和更高的预测精度。     

Analysis of crater formation in buried NG pipelines: A survey based on past accidents and evaluation of domino effect

The formation of a crater by the abrupt and catastrophic rupture of a high-pressure pipeline can be highly relevant, especially when the crater uncovers other pipelines, which could undergo a domino effect with a significant increase of the consequences on people or on the environment. However, this scenario has been only partially studied in the literature. To assess the influence of the pipeline parameters on the dimensions of the resulting crater, a statistical analysis of accidental ruptures of buried natural gas pipelines that have involved the formation of a crater was carried out. Mathematical expressions are proposed to describe the proportionality relationships found, which can be very useful to support adequate separation distances in the design and construction of parallel corridors of pipelines after appropriate escalating effects are considered. Finally, detailed event trees were developed to calculate the probability of occurrence of the final outcomes, as well as the identified domino sequences, based on a qualitative and quantitative analysis of the data. The study of these accident scenarios, based on actual cases, represents a useful and needed advance in risk analysis of natural gas transportation through pipelines.     

基于随机搜索优化XGBoost的瓦斯涌出量预测模型*

为防治瓦斯灾害,解决井下瓦斯涌出量在预测过程中因影响因素繁多带来的精度较低问题,提出1种基于套索(Lasso)回归与随机搜索优化极限梯度提升(XGBoost)的模型进行瓦斯涌出量预测。以沈阳某煤矿综采面瓦斯涌出量历史数据为例,综合考虑影响瓦斯涌出量的影响因素。首先利用Lasso回归提取对瓦斯涌出量有重要影响的特征数据,作为预测输入;采用随机搜索算法对XGBoost模型4种主要参数进行寻优,选取最优参建立预测模型获得预测指标并分析比较其他模型。研究结果表明:Lasso回归筛选出的影响因素结合随机搜索获得的最优参数组合优化XGBoost比其他模型预测精度更高,平均相对误差为1.53%,均方根误差为0.140 3 m3/min,希尔不等系数为0.013 2,研究结果可为现场瓦斯管理提供参考依据。     

基于改进PSO-BPNN的输油管道内腐蚀速率研究

为解决输油管道易腐蚀,且腐蚀程度难以测量的问题,提出使用改进的粒子群算法（PSO）优化误差反向传播神经网络（BPNN）对输油管道内腐蚀速率进行预测。改进的PSO算法提升了自身搜索到全局最优的能力,可为BPNN提供最优初始权值和阈值,从而有效避免BPNN易陷入局部最优的问题发生。以某条输油管线为例,分别运用标准的BPNN模型、PSO-BPNN以及改进的PSO-BPNN对该管线内腐蚀速率进行预测。结果表明:基于改进的PSO-BPNN的预测结果平均相对误差为5.57%,预测精度较BPNN和PSO-BPNN有明显提升。使用改进的PSO-BPNN预测输油管道的腐蚀速率可为管道的检测维修提供可靠的理论和技术支撑。     

Research on flow assurance of deepwater submarine natural gas pipelines: Hydrate prediction and prevention

The formation of hydrate will lead to serious flow assurance problems in deepwater submarine natural gas transmission pipelines. However, the accurate evaluation model of the hydrate blocking risk for submarine natural gas transportation is still lacking. In this work, a novel model is established for evaluating the hydrate risk in deepwater submarine gas pipelines. Based on hydrate growth-deposition mechanism, the mathematical model mainly consists of mass, momentum and energy conservation equations. Meantime, the model results are obtained by finite difference method and iterative technique. Finally, the model has been applied in the production of deepwater gas field (L Gas Field) in China, and the sensitivity analysis of relevant parameters has been carried out. The results show that: (a). The mathematical model can well predict the hydrate blockage risk in deepwater natural gas pipelines after verification. (b). Hydrate is easily formed at the intersection of horizontal pipeline and vertical riser, and the maximum blocking position often occurs in middle of the riser. (c). The hydrate blockage degree and length of hydrate formation region (HFR) decrease with the increase of gas transport rate. (d). The hydrate blockage degree and length of HFR decrease with the increase of gas transport temperature. (e). The hydrate blockage degree and length of HFR increase with the extension of horizontal pipeline. (f). Injecting inhibitors can effectively inhibit hydrate formation and blockage, but the improvement of transmission measures can significantly reduce the dosage of inhibitor. It is concluded that measures such as increasing gas transportation rate and temperature, shortening horizontal pipeline length, optimizing inhibitor injection point and injection rate can play a safe, economic and efficient role in hydrate preventing and controlling.     

IAFSA-GRNN在油田集输管道CO2腐蚀速率预测中的应用

为提高油田集输管道CO₂腐蚀速率预测的准确性,针对原始广义回归神经网络(GRNN)预测精度低的问题,提出改进的群智能算法优化原始GRNN的预测模型;分别使用GRNN模型、人工鱼群算法(AFSA)优化的GRNN(AFSA-GRNN)模型和自适应改进的AFSA-GRNN(IAFSA-GRNN)模型预测X65管线钢的CO₂腐蚀速率。结果表明:采用AFSA和IAFSA优化光滑因子S后,能大大提高GRNN模型的预测精度,预测结果的平均相对误差由36.09%分别减小至7.20%和6.90%;与AFSA相比,IAFSA优化的GRNN不仅具有更高的预测精度,还具有更快的收敛速度。AFSA-GRNN在第164次迭代计算时收敛,而IAFSA-GRNN在第109次迭代计算时收敛,说明AFSA经自适应优化能提高优化过程的收敛速度和GRNN的预测精度。     

煤与瓦斯突出预测的QGA-LSSVM模型

为快速、有效地对煤与瓦斯突出类型作出预测,运用灰色关联和因子分析模型对所选主要的判别指标进行分析提取,利用量子遗传算法(QGA)对最小二乘支持向量机(LSSVM)的参数作寻优处理,最终建立QGA-LSSVM煤与瓦斯突出预测模型。选取从砚石台矿区历史实测的数据,以96∶20的比例对该模型进行训练与测试,并将预测结果与其他预测模型的预测效果进行了比较。研究结果表明:对判别指标进行灰色关联分析可以有效去除对煤与瓦斯突出影响作用小的指标;用因子分析进行公共因子提取,可以有效减少数据信息冗余;利用QGA优化的LSSVM模型能使结果避免陷入局部最优解,用该模型可以有效预测煤与瓦斯突出类型,误判率为0。     

基于链式多重插补的WOA-ELM煤与瓦斯突出预测模型*

为了提高缺失数据下煤与瓦斯突出预测准确率,提出1种基于链式支持向量机多重插补（MICE_SVM）的鲸鱼优化算法（WOA）-极限学习机（ELM）预测模型,以淮南朱集矿区为例,选取5个煤与瓦斯突出影响指标作为模型特征,采用提出的MICE_SVM算法插补突出事故数据中缺失值,利用WOA优选ELM输入层权值及隐含层阈值,构建煤与瓦斯突出预测模型,将插补后数据用于WOA-ELM模型的训练与测试,并与其他模型的预测效果对比。研究结果表明:MICE_SVM插补前、后的有突出数据预测准确率分别为83.02%,90.41%,MICE_SVM显著提高了有突出预测准确率,对无突出和整体的预测准确率提高不明显;数据插补后WOA优化ELM对无突出、有突出和整体的预测准确率分别为97.94%,96.25%,96.48%,较优化前分别提高了5.79%,5.84%,5.55%,数据插补后WOA-ELM为最佳预测模型。     

A new pre-assessment model for failure-probability-based-planning by neural network

At present, the prediction of failure probability is based on the operation period for laid pipelines, and the method is complicated and time-consuming. If the failure probability can be predicted in the planning stage, the risk assessment system of gas pipeline will be greatly improved. In this paper, the pre-laying assessment model is established to minimize risk of leakage due to piping layout. Firstly, Fault Tree Analysis (FTA) modeling is carried out for urban natural gas pipeline network. According to expert evaluation, 84 failure factors, which can be determined in the planning stage, are selected as the input variables of the training network. Then the FTA model is used to calculate the theoretical failure probability value, and the failure probability prediction model is determined through repeated trial calculation based on BP (Back Propagation Neural Network) and RBF (Radial Basis Function), for obtaining the optimal network parameter combination. Finally, two prediction models are used to calculate the same example. By comparing our pre-assessment model with the theoretical prediction consequences of the fault tree, the results show that the error of RBF prediction model can be close to 3%, which proves the validity and correctness of the method.     

Integrated leakage detection and localization model for gas pipelines based on the acoustic wave method

With the development of natural gas transportation systems, major accidents can result from internal gas leaks in pipelines that transport high-pressure gases. Leaks in pipelines that carry natural gas result in enormous financial loss to the industry and affect public health. Hence, leak detection and localization is a major concern for researchers studying pipeline systems. To ensure the safety and improve the efficiency of pipeline emergency repair, a high-pressure and long-distance circular pipe leakage simulation platform is designed and established by similarity analysis with a field transmission pipeline, and an integrated leakage detection and localization model for gas pipelines is proposed. Given that the spread velocity of acoustic waves in pipelines is related to the properties of the medium, such as pressure, density, specific heat, and so on, this paper proposes a modified acoustic velocity and location formula. An improved wavelet double-threshold de-noising optimization method is also proposed to address the original acoustic wave signal collected by the test platform. Finally, the least squares support vector machine (LS-SVM) method is applied to determine the leakage degree and operation condition. Experimental results show that the integrated model can enhance the accuracy and precision of pipeline leakage detection and localization.     

Gas leakage consequence modeling for buried gas pipelines

One of conservation transfer methods for such widely-used gases as natural gas and hydrogen is buried pipelines. Safety of these pipelines is of great importance due to potential risks posed by inefficiencies of the pipelines. Therefore, an accurate understanding of release and movement characteristics of the leaked gas, i.e. distribution and speed within soil, the release to the ground surface, the movement of hydrogen gas through the ground, gas underground diffusion, gas dispersion in atmosphere, and following consequences, are very important in order to determine underground dispersion risks. In the present study, consequences of gas leakage within soil were evaluated in two sub-models, i.e. near-field and far-field, and a comprehensive model was proposed in order to ensure safety of buried gas supply pipelines. Near-field model which is related to soil and ground and its output is the gas released at different points and times from ground surface and it was adopted as input of far-field sub-model which is dispersion model in atmosphere or an open space under the surface. Validation of near-field sub-model was performed by the experimental data obtained by Okamoto et al. (2014) on full-scale hydrogen leakage and then, possible scenarios for far-field sub-model were determined.     

基于GASA-BP神经网络的煤层瓦斯含量预测方法研究*

为提高煤层瓦斯含量预测的精准度和效率,提出1种利用遗传算法（GA）和模拟退火算法(SA)混合初始化BP神经网络（BPNN）的瓦斯含量预测新模型（GASA-BPNN模型）。利用灰色关联分析法（GRA）筛选瓦斯含量主控因素并作为GASA-BPNN预测模型的输入。为解决BPNN收敛速度慢和易陷入局部极小陷阱的问题,将GA和具有时变概率突跳性的SA整合为GASA算法协同初始化BPNN的权值和阈值,有效地提高BPNN的参数学习能力。将该模型应用于煤炭生产现场,结果表明:BPNN模型、GA-BPNN模型和GASA-BPNN模型瓦斯含量预测总平均相对误差分别为15.79%,9.03%,5.56%。相比BPNN模型和GA-BPNN模型,GASA-BPNN模型对样本的泛化能力更强,参数训练速度最快并且预测精准度最高。