Corrosion Prediction Research Articles

Marine steel corrosion prediction and zonation using feature extraction and machine learning in the seas around China

To reduce the losses caused by the marine corrosion of steel, it is important to establish a prediction model to determine the corrosion rate of steel in depth-varying aggressive marine environments. The use of statistical feature extraction methods and machine learning modeling for marine steel corrosion prediction and zoning in the seas around China is investigated. In this study, 856 samples were collected. Mean and standard deviation were selected as environmental characteristics and corrosion loss time-varying relationships were log-transformed. Subsequently, four main supervised machine learning (ML) algorithms including Decision Tree (DT), Random Forest (RF), Gradient Boosting (GB), and XGBoost were explored for predicting corrosion loss in different depth-varying marine exposure zones. The GB model showed the best prediction accuracy and generalization ability with MSE, RMSE, MAE, and R2 values of 0.08, 0.43, 0.19, and 0.92, respectively. The spatial and temporal distribution of corrosion loss and zoning map in the seas around China were obtained. According to the corrosion zoning map of the splash zone, the South China Sea has a higher degree of corrosion, particularly in its northwestern region.

Spatial distribution of corrosion products from a bridge pier

AbstractThis paper studies the spatial distribution of corrosion by-products by a bridge pier within a conductive medium. An electrochemical impedance spectroscopy (EIS) technique was used to investigate an uncoated metallic bridge pier submerged in static distilled water. An equivalent circuit model, derived from EIS results, served as the foundation for the study. Further, the role of diffusion was analysed, considering its significance in characterising the transfer of particles from the pier into the surrounding water. This exploration revealed the complex interaction between the diffusion processes of various corrosion by-products as a function of distance. In addition, by evaluating the spatial distribution of iron (II) corrosion by-products and modelling nanoparticle diffusion, the research examined the impact of diffusion and concentration on corrosion particle transmission. The findings, analysed via Nyquist and Bode plots, demonstrate significant differences between theoretical and empirical diffusion coefficients. Results indicated that under natural corrosion conditions, the primary product of the corrosion reaction, iron (II), disperses into the medium when oxidation occurs. The elevated resistivity due to the presence of iron (II) underscores the diffusion effect, leading to corrosion product precipitation and reaching saturation level. Additionally, the results demonstrated ideal values for the diffusion coefficient, which are crucial for advanced corrosion modelling. The results emphasised the need for empirical data to improve corrosion prediction models and informed maintenance strategies for submerged structures.

Artificial intelligence enhanced fault prediction with industrial incomplete information

With the rapid advancement of sensor and information technology, fault prediction for industrial equipment has become increasingly feasible. However, accurate fault prediction is heavily dependent on the integrity of monitoring data. Unfortunately, limitations in data collection and storage technologies often result in incomplete and missing data, leading to decreased accuracy or even failure of predictive models. To address this challenge, this paper introduces a novel data-enhanced fault prediction method leveraging artificial intelligence, which stands out for its ability to handle incomplete information effectively. The innovation lies in the first-time division of incomplete information into incomplete variables and missing variables, each addressed with tailored AI-based solutions. Furthermore, the method integrates complete data with uncertain information through a dynamic Bayesian network, pioneering a new approach to compensating for data integrity issues and minimizing prediction errors. The enhanced data is then utilized for long-term performance predictions, setting a new standard in fault prediction accuracy. The developed model exhibits exceptional adaptability by integrating advanced techniques such as parameter uncertainty analysis, sensitivity analysis, and dynamic range analysis. These enhancements significantly boost the precision and comprehensiveness of fault predictions. The method’s feasibility and superiority are demonstrated through the prediction of CO2 corrosion in subsea pipelines, proving that the proposed approach significantly improves fault prediction accuracy even under incomplete information conditions.

Corrosion prediction model for long-distance pipelines based on NLFE-NGO-ELM

Abstract To improve the accuracy of buried pipeline corrosion rate prediction, we developed a Northern Goshawk Optimization-based Extreme Learning Machine (NLFE-NGO-ELM) model enhanced by Non-linear Feature Expansion. Feature selection was used to expand the small dataset and reduce the dimensionality of factors affecting pipeline corrosion. The Northern Goshawk algorithm was used, and 106 sets of pipeline external corrosion detection data were simulated in MATLAB as a case study. The ELM optimizer was analyzed using Grey Wolf Optimization (GWO-ELM), Northern Goshawk Optimization (NGO-ELM), and BP optimization as comparison models. Using NLFE-NGO-ELM significantly improved the model's prediction accuracy and speed. The NGO-ELM model was applied to predict the corrosion rates of 21 pipelines, achieving an absolute error of only 2.40% and an R value of 0.97255, surpassing the predictive performance of NGO-BP and GWO-ELM models. These results indicate that the current model exhibits superior learning performance and stronger fitting capabilities.

A probabilistic time-dependent corrosion wastage model for offshore platforms based on reverse diffusion model

Corrosion wastage poses significant challenges to offshore platforms, making accurate corrosion depth prediction crucial for structural integrity and extending service life. This study addresses the limitations in measurement processes by integrating a reverse diffusion model into the traditional probabilistic corrosion wastage model, enabling comprehensive analysis of corrosion data. The reverse diffusion model effectively manages data uncertainty and noise, leading to a proposed probabilistic model that incorporates measurement uncertainty. This model not only mitigates data noise and errors but also improves the reliability and robustness of corrosion depth predictions for offshore platforms. Initially, statistical analysis of corrosion data from offshore mobile platforms confirmed that the corrosion rate and depth distribution of local components follow a Weibull distribution. The effectiveness of the reverse diffusion model was validated by comparing the probability distributions of original data with those processed using the model. Finally, leveraging the optimized data and an empirical corrosion model for ships, a probabilistic model for time-dependent corrosion wastage in offshore platforms was developed, and its reliability was assessed. The proposed model enhances the accuracy of corrosion predictions, contributing to better management of corrosion risks in offshore platform structures.

Automated matching and visualisation of magnetic flux leakage data in shale gas pipeline based on ICP and DBSCAN algorithm

ABSTRACT Magnetic flux leakage (MFL) is a non-destructive testing method for shale gas pipeline safety monitoring. Despite many pipelines have completed two or more rounds of internal inspection and have accumulated substantial data, effectively analyzing and utilizing this MFL data remains a significant challenge. To that end, this study proposes a novel combined model that integrates Iterative Closest Point (ICP) and DBSCAN (Density-Based Spatial Clustering of Applications with Noise) fusion to match multiple MFL data. The model outputs the matching result and visualisation images of MFL data acquired in different years. Furthermore, RMSE (root mean square error) and overlap rate have been used to evaluate the model. The results indicate that the average distance error of matched defects between two datasets decreased by 72.5%, and the overlap rate increased by 20%. Additionally, the DBSCAN Fusion shows better computational efficiency with an increase in defect quantity within the pipe segments. Finally, the impact of different datasets on the model’s matching accuracy is discussed in this study. The findings show that small-scale datasets or higher false detection rates lead to decreased accuracy. The proposed model fully leverages MFL data to achieve rapid matching of defects, offering a dependable and effective technical solution for safety monitoring and corrosion prediction in shale gas pipelines.

Investigation of ash problems potential for a blend of high-sulfur coal and bamboo pellet biomass

Abstract Biomass bamboo can be utilized as co-firing fuel in power plants in order to provide more green energy. However, the characteristics of biomass bamboo with high potassium and chlorine that probably increase ash problem aspects need to be investigated before the implementation. This study aims to obtain an initial prediction of slagging, fouling, abrasion, and corrosion for bamboo pellets and high-sulfur coal blends using empirical indices. This study used seven slagging parameters, four fouling parameters, one abrasion parameter, and two corrosion parameters. The result shows that a composition of 10-30 wt.% bamboo pellets is more recommended than others. Despite the composition of 50 wt.% and above being predicted to reduce the slagging risk of high-sulfur coal, the composition of 40 wt.% and above is predicted to increase the fouling risk. At higher compositions, 80 wt.% and above are predicted to increase the corrosion risk. Further test using a drop tube furnace is necessary to obtain comprehensive results. Moreover, the calorific value aspect needs to be considered depending on the need of the power plant for the efficiency aspect.

A life prediction model for P110S steel in deep-well environments with H2S/CO2 coexistence based on multi-factor chemometric drive

A mechanistic-chemometrics model for life prediction of P110S steel in deep-well environments with H₂S/CO₂ coexistence was proposed. The model was developed by considering the interaction mechanism between uniform and pitting corrosion, then modified using a multi-factor chemometric drive incorporating temperature/pressure, flow velocity, stress. Finally, the pit-to-crack transition was modeled using finite element design and direct current potential drop measurements, completing the life prediction process. The model predicts a lifespan of 2.28–5.25 years at different well depths, and this result was validated with on-site data, indicating the model’s accuracy. The knowledge paradigm provided herein will assist in corrosion prediction.

Carbon Steel Corrosion Induced by Sulfate-Reducing Bacteria: A Review of Electrochemical Mechanisms and Pathways in Biofilms

Microbial metal corrosion has become an important topic in metal research, which is one of the main causes of equipment damage, energy loss, and economic loss. At present, the research on microbial metal corrosion focuses on the characteristics of corrosion products, the environmental conditions affecting corrosion, and the measures and means of corrosion prevention, etc. In contrast, the main microbial taxa involved in metal corrosion, their specific role in the corrosion process, and the electron transfer pathway research are relatively small. This paper summarizes the mechanism of microbial carbon steel corrosion caused by SRB, including the cathodic depolarization theory, acid metabolite corrosion theory, and the biocatalytic cathodic sulfate reduction mechanism. Based on the reversible nature of electron transfer in biofilms and the fact that electrons must pass through the extracellular polymers layer between the solid electrode and the cell, this paper focuses on three types of electrochemical mechanisms and electron transfer modes of extracellular electron transfer occurring in microbial fuel cells, including direct-contact electron transfer, electron transfer by conductive bacterial hair proteins or nanowires, and electron shuttling mediated by the use of soluble electron mediators. Finally, a more complete pathway of electron transfer in microbial carbon steel corrosion due to SRB is presented: an electron goes from the metal anode, through the extracellular polymer layer, the extracellular membrane, the periplasm, and the intracellular membrane, to reach the cytoplasm for sulfate allosteric reduction. This article also focuses on a variety of complex components in the extracellular polymer layer, such as extracellular DNA, quinoline humic acid, iron sulfide (FeSX), Fe3+, etc., which may act as an extracellular electron donor to provide electrons for the SRB intracellular electron transfer chain; the bioinduced mineralization that occurs in the SRB biofilm can inhibit metal corrosion, and it can be used for the development of green corrosion inhibitors. This provides theoretical guidance for the diagnosis, prediction, and prevention of microbial metal corrosion.

Corrosion degree detection of ferromagnetic materials using a hybrid machine learning approach and self-magnetic flux leakage technology

Corrosion of ferromagnetic materials can lead to performance degradation or even failure in engineering structures. Therefore, detecting the degree of corrosion in these materials is crucial. Traditional detection methods are primarily manual and lack precision, highlighting the need for developing automated, high-precision detection techniques. In this paper, we developed a novel self-magnetic flux leakage (SMFL) scanner. We enhanced the Marine Predator Algorithm (MPA) by integrating Chaos Opposition (CO) and Group Learning (GL), resulting in the COGL-MPA-XGB model. This model optimizes Extreme Gradient Boosting parameters for accurate corrosion prediction, achieving a 96% coefficient of determination (R2) and a mean error of 1.87. We tested the method on 170 rod-shaped steel samples with varying degrees of corrosion, evaluating its effectiveness through ablation experiments and comparisons with three other machine learning models. The results show that the combined SMFL and hybrid model strategy reduces the maximum detection error by 73.7% and offers a new perspective for quantitative corrosion detection.

Development of a Predictive Model for Carbon Dioxide Corrosion Rate and Severity Based on Machine Learning Algorithms.

Carbon dioxide corrosion is a pervasive issue in pipelines and the petroleum industry, posing substantial risks to equipment safety and longevity. Accurate prediction of corrosion rates and severity is essential for effective material selection and equipment maintenance. This paper begins by addressing the limitations of traditional corrosion prediction methods and explores the application of machine learning algorithms in CO2 corrosion prediction. Conventional models often fail to capture the complex interactions among multiple factors, resulting in suboptimal prediction accuracy, limited adaptability, and poor generalization. To overcome these limitations, this study systematically organized and analyzed the data, performed a correlation analysis of the data features, and examined the factors influencing corrosion. Subsequently, prediction models were developed using six algorithms: Random Forest (RF), K-Nearest Neighbors (KNN), Gradient Boosting Decision Tree (GBDT), Support Vector Machine (SVM), XGBoost, and LightGBM. The results revealed that SVM exhibited the lowest performance on both training and test sets, while RF achieved the best results with R2 values of 0.92 for the training set and 0.88 for the test set. In the classification of corrosion severity, RF, LightGBM, SVM, and KNN were utilized, with RF demonstrating superior performance, achieving an accuracy of 99% and an F1-score of 0.99. This study highlights that machine learning algorithms, particularly Random Forest, offer substantial potential for predicting and classifying CO2 corrosion. These algorithms provide innovative approaches and valuable insights for practical applications, enhancing predictive accuracy and operational efficiency in corrosion management.

Prediction model for the bond behaviour of low-corrosion reinforced concrete considering corrosion time variability

Corrosion greatly affects the bond behaviour of reinforced concrete. In the case of reinforcing low-corrosion reinforced concrete, the bond strength of steel bars and concrete is at an enhanced stage. Accurate prediction of bond behaviour at this stage is crucial for the rational reinforcement. This study designed ten groups, totalling 30 pull-out specimens, were accelerated corroded with various current densities and electrification times. Through 3D scanning and image analysis technology, we observed that increasing the current density yielded less homogeneous corrosion, creating pit-like rust marks. Meanwhile, we developed a model for predicting the corrosion level. The model is applicable to the corrosion level design of accelerated corrosion. Pull-out tests showed that as corrosion increased, the specimen failure mode changed from pull-out to splitting and back to pull-out, with the bond strength first increasing and then decreasing. A bond-slip prediction model for corroded reinforced concrete was also established, considering the effects of corrosion on bond strength in different stages, especially the enhancement stage. The model performs well in predicting the bond behaviour of low-corrosion reinforced concrete and can be used in conjunction with corrosion prediction models and provides guidance for reinforcement.

Machine learning-based corrosion rate prediction of steel embedded in soil

Predicting the corrosion rate for soil-buried steel is significant for assessing the service-life performance of structures in soil environments. However, due to the large amount of variables involved, existing corrosion prediction models have limited accuracy for complex soil environment. The present study employs three machine learning (ML) algorithms, i.e., random forest, support vector regression, and multilayer perception, to predict the corrosion current density of soil-buried steel. Steel specimens were embedded in soil samples collected from different regions of the Wisconsin state. Variables including exposure time, moisture content, pH, electrical resistivity, chloride, sulfate content, and mean total organic carbon were measured through laboratory tests and were used as input variables for the model. The current density of steel was measured through polarization technique, and was employed as the output of the model. Of the various ML algorithms, the random forest (RF) model demonstrates the highest predictability (with an RMSE value of 0.01095 A/m2 and an R2 value of 0.987). In light of the feature selection method, the electrical resistivity is identified as the most significant feature. The combination of three features (resistivity, exposure time, and mean total organic carbon) is the optimal scenario for predicting the corrosion current density of soil-buried steel.

Towards proactive corrosion management: A predictive modeling approach in pipeline industrial applications

Corrosion is the most significant factor affecting pipeline integrity management. It may lead to facility damage, production interruption, environmental pollution and even explosion. Corrosion management developed in recent years is a critical means to ensure the safety and reliability of pipeline systems. However, developing proactive corrosion management is difficult due to limited availability of data on target pipelines and corrosion factors. The motivation of this study is to develop a proactive corrosion management approach based on a predictive framework. The proposed framework integrates multiple corrosion factors, diverse working conditions, multi-phase degradation, and probabilistic analysis to offer an efficient analysis tool. A numerical analysis of corrosion prediction is conducted using the corrosion data of the training pipeline and the target pipeline, and the corrosion rate, corrosion depth, and remaining useful life (RUL) of the target pipeline are evaluated. Based on the RUL information, the maintenance time of the pipeline is further analyzed to maximize the benefits of maintenance. This study offers a comprehensive approach to pipeline corrosion prediction, facilitating proactive maintenance strategies and improving asset management in the pipeline industry.

Managing Green and Sustainable Technologies: Climate-Informed Corrosion Prediction for Steel Structures

The unpredictability of atmospheric circumstances is one of the major elements that contribute to the capability to anticipate the corrosion growth in metal structures over time accurately. Climate shifts can potentially modify the long-term attributes of these factors throughout the operational life of metal structures, both those currently in existence and those newly developed. The impact of climate irregularity on the probabilistic nature of atmospheric variables, which significantly impact corrosion situations, can add intricacy to corrosion predictions in these constructions. This project presents an incorporated framework to quantify the impact of climate alteration on the corrosion rates of steel structures in Jordan. It considers the changes in environmental conditions, specifically temperature, relative humidity, and wind speed, and their impacts on atmospheric corrosion. Global Climate Models are employed to assess the long-term effects of climate transformation on these environmental circumstances. An analytical model for anticipating corrosion rate is integrated with climate transformation models to predict modifications in the corrosion rates of steel parts relative to historical situations. This project also examines the impact of climate transformation on the fluctuations of these climatic parameters and offers a contrast between historical data and projected conditions across the country. The findings indicate a significant increase in corrosion rates across Jordan, which calls for localized green building codes and standards to ensure that future infrastructure is sustainable and capable of withstanding the new climatic norms. This approach addresses the immediate challenges posed by climate change and contributes to the broader goals of sustainable urban development and managing green technology adoption in Jordan. Doi: 10.28991/CEJ-2024-010-08-016 Full Text: PDF

Advanced multi-image segmentation-based machine learning modeling strategy for corrosion prediction and rust layer performance evaluation of weathering steel

Corrosion images are the most accessible type of data to collect. To utilize these data to monitor and evaluate the corrosion of weathering steel (WS) efficiently. This study established a dataset of corrosion images of WS by accelerated corrosion experiments with dry and wet cycles. Image segmentation algorithms and machine learning were used for training, modeling, and data mining of the image data. The results showed that 22 features in the corrosion images were highly correlated with the corrosion thickness loss, with the highest correlation coefficient exceeding 0.7. The rust layer texture and color features showed a synergistic evolution during the corrosion process. On this basis, a classification model of the rust layer was developed to quantify the evolution of the protective properties of the rust layer during the corrosion process. In addition, the established integrated learning model achieves accurate prediction (R2 > 0.94) for WS corrosion thickness loss.

Interactive effects of moisture, chloride, and carbonation on rebar corrosion in mortar

This study investigates the synergistic effects of chloride ions, carbonation, and relative humidity on rebar corrosion in mortar, elucidating the complex interactions at the rebar-mortar interface. Findings reveal that chloride ions and carbonation significantly alter the volumetric water content and resistivity of mortar, leading to non-binary corrosion behaviour through area effect, ohmic control, and anode-to-cathode ratio. In chloride-free, non-carbonated mortar, the area effect dominates, whereas the presence of chloride introduces ohmic control and a saturation-dependent anode-to-cathode ratio. In carbonated mortar, a distinctive ohmic control mechanism emerges, governed by local ionic conductivity. This study highlights the critical need to consider these combined effects for accurate prediction and mitigation of rebar corrosion in concrete structures. By advancing the understanding of these mechanisms, the research provides valuable insights for improving the durability of reinforced concrete structures exposed to harsh environmental conditions.

Transfer learning applications for corrosion prediction for electrodissoluted specimens

Corrosion monitoring is a critical aspect of Structural Health Monitoring (SHM), as it presents inherent challenges in tracking its progression and requires strategically placed sensors for effective surveillance. Moreover, the intricate interplay between environmental factors and the material composition of corroding components further complicates this phenomenon. In this study, a novel neural network framework is designed to predict corrosion evolution based on environmental data, focusing on three distinct aluminium alloys: Al7475, Al2024, and F357. A dedicated neural network model is developed for Al7475, trained on electrodissolution data, including process parameters and profilometry measurements. This model predicts corrosion progression in terms of volume loss, corroded depth, and area based on charge measurements. To address the challenge of material-environment coupling, transfer learning is employed to adapt the neural network to Al2024 and F357. The oxidation charge passed during the electrodissolution mimics in an artificial way the corrosion process due to environmental conditions in the attack site. Consequently, the implemented neural network establishes a robust connection between corrosion process measurements and profilometry data, enabling accurate corrosion progression prediction based on oxidation charge. This framework allows the prediction of a component's residual useful life based on the assessment of the corrosion state, with the knowledge of volume loss further supporting mechanical assessment by evaluating resistant section or pit dimensions, ultimately aiding in the pit-crack transition analysis.

Corrosion Prediction in the Oil Industry Using Deep Learning Techniques

Corrosion presents a significant challenge in the oil industry, causing both immediate and long-term damage. Effective early prediction and monitoring of corrosion are crucial to mitigating economic losses and environmental impacts. However, traditional methods for predicting and detecting corrosion are often time-consuming and inefficient. This study leverages convolutional neural networks (CNNs) within a deep learning framework to develop two automated detection models for internal and external corrosion. These models can extract hierarchical features directly from raw pixel data, enhancing prediction accuracy and efficiency. Our dataset, provided by the Iraqi Oil Company, includes drone-captured images (162 photos: 91 depicting corrosion and 71 showing no signs of corrosion) and ultrasonic sensor readings (250 rows of oil pipeline thickness measurements). We assess the performance of our CNN models using metrics such as accuracy, precision, recall, and F-score, and we perform regression analysis to evaluate prediction errors. This research introduces two innovative systems: a 2D CNN for classifying the presence or absence of external corrosion, and a 1D CNN for assessing internal corrosion levels, identifying areas with the highest corrosion rates, and estimating the remaining operational lifespan based on these rates. Additionally, we develop a user-friendly interface for these systems. Comparative analysis demonstrates the superior efficiency of our proposed approach over traditional and alternative methods. Our findings advance the understanding of artificial intelligence applications in corrosion prediction, offering robust models to prevent unexpected corrosion failures. Future work will explore the integration of additional factors, such as humidity and temperature sensors, to further enhance the system's accuracy and reliability.

Data-driven atmospheric corrosion prediction model for alloys based on a two-stage machine learning approach

Alloy material corrosion is a comprehensive problem affected by multi-dimensional factors, including the time, environment, and material element. Its corrosion prediction has been facing the challenges of complex mechanism and small sample size. However, parametric methods are limited in terms of modeling accuracy and generalization performance for high-dimensional feature problems. Non-parametric machine learning algorithms, while having excellent modeling capabilities, need to be optimized in conjunction with the structure of the corrosion problem. Therefore, we develop a two-stage hybrid intelligent machine learning model for improving alloy corrosion prediction accuracy. The model takes the time, environment, and material information of the alloy corrosion process as inputs and the corrosion rate as output. The first stage of the model utilizes tree ensemble learning (TEL) models for initial prediction modeling and feature importance knowledge mining. In the second stage, the neural network (NN) is used to enhance the fusion of the multi-model prediction outputs, and it is optimized using feature importance modification (FIM). The proposed TEL-FIMNN model shows considerable performance improvement over the TEL and ANN models on real low-alloy steel corrosion dataset. Compared to other advanced multi-model fusion methods such as linear regression (LR), support vector regression (SVR), random forest (RF), and Bayesian model averaging (BMA), the proposed model also shows superior results. The TEL-FIMNN model provide a new structure from the perspectives of knowledge mining and fusion enhancement, which has good potential for multi-dimensional corrosion prediction.