Data-Driven Models for Forecasting Failure Modes in Oil and Gas Pipes

Elshaboury, Nehal; Al-Sakkaf, Abobakr; Alfalah, Ghasan; Abdelkader, Eslam Mohammed

doi:10.3390/pr10020400

Cited by 18 publications

(8 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2) The performance evaluation parameter is the mean square error 3) The training algorithm is the Bayesian Regularization algorithm It can be clearly seen that the proposed work attains much higher accuracy of 91.91% compared to 85% of previous work [1]. This can be attributed to the regression learning based BR trained ANN design which has a steep descent of error compared to the naïve Baye's classifier or the conventional Bayesian Regularization algorithm.…”

Section: Resultsmentioning

confidence: 93%

See 1 more Smart Citation

A Data Driven Machine Learning Model for Fault Detection in Industrial Pipelines

Vyas,

Bagde

2024

IJSREM

View full text Add to dashboard Cite

Industry 4.0 has marked a paradigm shift in the way production and manufacturing operates. Process monitoring and fault diagnosis are important for the safety and reliability of industrial processes especially for smart manufacturing. As a data-driven process monitoring methodology, multivariate statistical analysis techniques, such machine learning based approaches have become extremely critical for automation. Pipelines carrying oil and gas are essential for a nation's economic sustainability. In order to maximise their function and prevent product losses during the transportation of petroleum products, they must be carefully inspected. However, they are susceptible to failure, which could have negative effects on the environment, financial loss, and safety. Therefore, evaluating the pipe's state and quality would be crucial. Despite being time-consuming and expensive, a number of inspection procedures are used to assure the safety of pipelines. However, due to the time consumption and error prone nature of manual inspection, data driven models are being explored for forecasting failure in oil and gas pipelines. The proposed work presents a Bayesian Regularized classifier for forecasting failure and the results show that the proposed approach outperforms the existing baseline techniques in terms of forecasting accuracy. Keywords:- Industry 4.0, oil pipelines; failure forecasting, Bayesian Regularization, Confusion Matrix, Classification Accuracy.

show abstract

Section: Resultsmentioning

confidence: 93%

“…Petroleum products are transported by pipelines, the backbone of the oil and gas sector, in a range of locations (such as onshore or offshore) [1]. The first oil pipeline was built in Pennsylvania in 1879, and it had a diameter of 6 inches and a length of 109 miles [2].…”

Section: Failures In Oil and Gas Pipelinesmentioning

confidence: 99%

A Data Driven Machine Learning Model for Fault Detection in Industrial Pipelines

Vyas,

Bagde

2024

IJSREM

View full text Add to dashboard Cite

show abstract

“…here are over two million miles of pipelines in the world, with over two-thirds contained in the U.S. alone [1]. Unfortunately, there have also been about 10,000 oil and gas pipeline failures to date in the U.S. since 2002 [2], and over half of the pipelines in the U.S. are over 40 years old [3].…”

Section: Introductionmentioning

confidence: 99%

Machine-Learning-Assisted Leak Detection Using Distributed Temperature and Acoustic Sensors

Gemeinhardt,

Sharma

2024

IEEE Sensors J.

View full text Add to dashboard Cite

Small leaks, many of which often go undetected using conventional gauges, remain an urgent problem for an aging pipeline infrastructure. This paper proposes a method that allows for an automated and robust leak detection and localization system by fusing Distributed Acoustic Sensor (DAS) and Distributed Temperature Sensor (DTS) data for machine learning. Distributed fiberoptic sensing creates an advantage over conventional gauges by providing real-time continuous measurements along the entire length of the fiber-instrumented pipeline at a high spatiotemporal resolution and sensitivity that can detect small leaks, as well as the leak location. High sensitivity, however, can create noisy data. Thus, machine learning is applied for a robust method of distinguishing non-leak data from leak signatures for accurate leak detection and localization. The workflow is demonstrated on an experimental pipeline setup exposed to environmental noise. The results illustrate reliable detection of small leaks between 0.04 L/s to 0.30 L/s with F1 scores over 0.9, on a range of DAS frequency bands combined with DTS data. The fusion of two different types of distributed fiber-optic sensors not only increases the likelihood of small leaks being detected but decreases the number of false alarms by filtering noise that only appears in one sensor domain. Furthermore, the machine learning approach utilizes segmentation, allowing for precise localization and quick investigation, as well as AI explainability of the sensor data deemed a leak signature.

show abstract

“…The prediction results were compared and had less than 5% false detection [27]. A multilayer perceptron neural network was developed with an accuracy of 84% for historical oil and gas pipeline data [28]. A comprehensive review of machine-learning approaches for oil and gas pipeline failure predictions was also conducted [29].…”

Section: Introductionmentioning

confidence: 99%

Threat and Risk Analysis-Based Neural Network for a Chemical Explosion (TRANCE) Model to Predict Hazards in Petroleum Refinery

Gabhane

Rao

2023

Toxics

View full text Add to dashboard Cite

Risk analysis and prediction is a primary monitoring strategy to identify abnormal events occurring in chemical processes. The accidental release of toxic gases may result in severe problems for people and the environment. Risk analysis of hazardous chemicals using consequence modeling is essential to improve the process reliability and safety of the refineries. In petroleum refineries: toluene, hydrogen, isooctane, kerosene, methanol, and naphtha are key process plants with toxic and flammable chemicals. The major process plants considered for risk assessment in the refinery are the gasoline hydrotreatment unit, crude distillation, aromatic recovery, continuous catalytic reformer, methyl–tert–butyl–ether, and kerosene merox units. Additionally, we propose a threat and risk analysis neural network for the chemical explosion (TRANCE) model for refinery incident scenarios. Significantly, 160 attributes were collected for the modeling on the basis of the significance of failure and hazardous chemical leaks in the refinery. Hazard analysis shows that the leakages of hydrogen and gasoline at the gasoline hydrotreatment unit, kerosene at the kerosene merox plant, and crude oil at crude-distillation units were areas of profound concern. The developed TRANCE model predicted the chemical explosion distance with an R2 accuracy value of 0.9994 and MSE of 679.5343.

show abstract

Data-Driven Models for Forecasting Failure Modes in Oil and Gas Pipes

Cited by 18 publications

References 37 publications

A Data Driven Machine Learning Model for Fault Detection in Industrial Pipelines

A Data Driven Machine Learning Model for Fault Detection in Industrial Pipelines

Machine-Learning-Assisted Leak Detection Using Distributed Temperature and Acoustic Sensors

Threat and Risk Analysis-Based Neural Network for a Chemical Explosion (TRANCE) Model to Predict Hazards in Petroleum Refinery

Contact Info

Product

Resources

About