Prediction model for corrosion rate of low-alloy steels under atmospheric conditions using machine learning algorithms

This work constructed a machine learning(ML)model to predict the atmospheric corrosion rate of low-alloy steels(LAS).The material properties of LAS,environmental factors,and exposure time were used as the input,while the corrosion rate as the output.6 dif-ferent ML algorithms were used to construct the proposed model.Through optimization and filtering,the eXtreme gradient boosting(XG-Boost)model exhibited good corrosion rate prediction accuracy.The features of material properties were then transformed into atomic and physical features using the proposed property transformation approach,and the dominant descriptors that affected the corrosion rate were filtered using the recursive feature elimination(RFE)as well as XGBoost methods.The established ML models exhibited better predic-tion performance and generalization ability via property transformation descriptors.In addition,the SHapley additive exPlanations(SHAP)method was applied to analyze the relationship between the descriptors and corrosion rate.The results showed that the property transformation model could effectively help with analyzing the corrosion behavior,thereby significantly improving the generalization ability of corrosion rate prediction models.

Corrosion behavior and corrosion products of a low-alloy weathering steel in Qingdao and Wanning

A newly developed low-alloy weathering steel has been exposed in two coastal sites （Qingdao in the north, Wanning in the south） in China for one year. The samples in Wanning corroded far more seriously than those in Qingdao. The rust layer formed on the steel was analyzed by X-ray diffraction （XRD）, scanning electron microscopy （SEM）, N2 adsorption approach, polarization curves, and electrochemical impedance spectroscopy （EIS）. The rust formed in Qingdao contains more X-ray amorphous compounds and is more compact than that formed in Wanning. Cr and Cu are enriched in the rust layer near the steel matrix, and the phenomenon is more obvious in Qingdao than in Wanning. The rust layer formed in Qingdao suppresses the anodic and cathodic reaction more remarkably than that formed in Wanning does. The rust layer formed in Qingdao possesses a higher ability to block the permeation of chloride ions than that formed in Wanning does.

INFLUENCE OF NaCl DEPOSITION ON ATMOSPHERIC CORROSION OF A3 STEEL IN THE PRESENCE OF SO_2

The regularities of atmospheric corrosion of A3 steel deposited with different amount of NaCl exposed to the air containing 1ppm SO2 at 80%RH and 25℃ were studied in laboratory. NaCl can accelerate the corrosion of A3 steel obviously under such condition. The relationship between the weight loss of A3 steel and the amount of NaCl deposition can be well described by using a quadratic function. Fourier transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), Scanning electron microscopy and electron dispersion X-ray analysis (SEM/EDAX) were used to characterize the corrosion products. In the absence of NaCl, FeSO4 · xH2O is the dominant corrosion products, while Fe3O4, FeSO4·H2O, β-FeOOH and γ-FeOOH dominate in the presence of NaCl.

Role of CO_2 in the initial stage of atmospheric corrosion of AZ91 magne-sium alloy in the presence of NaCl

The effect of CO2 and NaCl on the initial stage of atmospheric corrosion of AZ91 magnesium alloy was studied. The observation of surface morphology by optical microscopy （OM）, scanning electron microscopy （SEM） and the analysis of corrosion products by X-ray diffraction （XRD） were integrated to investigate corrosion evolution. The results showed that NaCl stimulated the corrosion by promoting the formation of thin electrolyte film, increasing the conductivity and breaking the protective film in the absence of CO2. The morphology of the corroded samples with deposited NaCl was more homogenous in the presence of CO2. It was suggested that NaCl-induced corrosion was inhibited in the presence of CO2 by the formation of slightly soluble corrosion products containing hydroxy carbonates and hydroxy chlorides that provided a partly protective layer on the surface of the magnesium alloy.

Comparison of the corrosion behavior of AM60 Mg alloy with and without self-healing coating in atmospheric environment

Coatings on the surface of Mg alloys are inevitably damaged during their practical applications,and corrosion can easily initiate from the damaged areas to accelerate the failure of Mg parts.A dual self-healing coating has already been developed to solve this problem in our previous work.Considering the practical application of this coating,it is necessary to investigate its service behavior in atmospheric environment.Thus,the corrosion behavior of AM60 Mg substrate with and without the self-healing coating was compared in Shenyang industrial atmospheric environment.The results show that the enrichment of sediments and rainwater on the scratch areas can accelerate the corrosion of the exposed Mg substrate.The inhibitors can be released from the damaged coating to inhibit corrosion.The dual self-healing coating shows better inhibition ability to narrow scratches due to the higher inhibitor concentration and less resumption.Also,the coating with wide scratches displays enough inhibition ability as well.The dual self-healing coating is a good alternative for Mg alloy parts in the practical applications.

Influence of NaCl Deposition on Atmospheric Corrosion of A3 Steel

Regularity of the initial atmospheric corrosion of A3 steel in the presence of NaCI was investigated. The results showed that NaCI can accelerate the corrosion of A3 steel. Dry mass gain of A3 steel in the presence of NaCI deposition increases with the exposure time, which can be correlated by using exponential decay function. The relationship between dry mass gain and amount of NaCI deposition at a certain exposure time follows a quadratic function. More amount of NaCI deposition will slow down the corrosion rate to some extent after exposure for a certain time. SEM/EDAX was used to characterize the corrosion surface and products. The surfaces of the NaCI treated A3 steel were obviously uneven, and some tiny crevices appear in the corrosion products that could help O2 transmit to the inner layer. The amount of oxygen in atomic percentage in the corrosion products increases with the amount of NaCI deposition.

GALVANIC CORROSION AND PROTECTION OF GECM/LY12CZ COUPLES UNDER DIFFERENT ATMOSPHERIC EXPOSURE CONDITIONS

Galvanic, compatibility between graphite epoxy composite materials (GECM) and LY12CZ aluminum alloy was evaluated in different atmospheric corrosion environments and by laboratory electrochemical measurements. Open circuit potential electrochemical measurements showed a relatively large potential difference about 1 volt between the GECM and LY12CZ aluminum alloy, and this difference provided the driving force for galvanic corrosion of the LY12CZ aluminum alloy as an anode. Having been exposed for 1, 3 or 5 years in Beijing, Tuandao and Wanning station, GECM/LY12CZ couples showed significant losses of strength and elongation. Protective coatings and non-conductive barriers breaking the galvanic corrosion circuit were evaluated under the same atmospheric corrosive conditions. Epoxy primer paint, glass cloth barriers and LY12CZ anodizing were effective in galvanic corrosion control for GECM/LY12CZ couples.

ATMOSPHERIC CORROSION OF 304AND 316 STAINLESS STEELS

Atmospheric corrosion of 304 and 316 stainless steels was studied by field exposure test, in which rain water was collected and analyzed by atomic absorption spectroscope (AAS). Emphasis was put on the dissolution of chromium and nickel from the stainless steels by atmospheric corrosion and rain runoff. AAS analyses shows that the amounts of yearly chromium and nickel dissolution were less than 150ng/cm 2 and 50ng/cm 2 respectively for both 304 and 316 stainless steels. XPS analysis reveals the marked Cr enrichment on the 316 stainless steel after one year field exposure. SEM shows the morphology of corrosion pits on the steel surfaces.

Atmospheric Corrosion of Steel A3 Deposited with Ammonium Sulfate and in the Presence of Sulfur Dioxide

A laboratory study of the atmospheric corrosion of carbon steel deposited with (NH4)2SO4 in the presence of SO2 is reported. The different levels of (NH4)2SO4 (0, 15, 30, 45, 60 μg·cm-2) were added on the surface of the samples before the exposure. The corrosion was investigated by a combination of gravimetry, Fourier transform infrared spectroscope and scanning electron microscopy. A detailed knowledge about the corrosion products was acquired, both quantitatively and qualitatively. The results show that the metal loss increased and the increasing tendency of corrosion rates slowed down with the increasing exposure time. The phase constituents of the corrosion products are mainly α-FeO(OH), γ-FeO(OH), and δ-FeO(OH).

Effect of CO_2 on Atmospheric Corrosion of UNS G10190 Steel under Thin Electrolyte Film

The atmospheric corrosion of UNS G10190 steel under a thin electrolyte film in the atmosphere polluted by CO_2, has been studied in the lab using an atmospheric corrosion monitor (ACM) in combination with XRD and SEM observations of the surface of steel. The ACM study indicated that the corrosion rate of the steel increased with increasing carbon dioxide concentration. The XRD and SEM observations showed that no carbonate was found in the corrosion product on the steel surface. The corrosion product consisted of two layers, i. e., inner and outer layer. From the experimental results, it was concluded that CO_2 played an enhancing role in the atmospheric corrosion of UNS G10190 steel. The film of the corrosion product showed slight protection.

Atmospheric corrosion behavior of Nb-and Sb-added weathering steels exposed to the South China Sea

The atmospheric corrosion behavior of new-type weathering steels(WSs)was comparatively studied,and the effects of Nb and Sb during corrosion were clarified in detail through field exposure and characterization.The results showed that the addition of Nb and Sb played positive roles in corrosion resistance,but there was a clear difference between these two elements.Nb addition slightly improved the rust property of conventional WS but could not inhibit the electrochemical process.In contrast,Sb addition significantly improved the corrosion resistance from the aspects of electrochemistry and rust layer.Compared with only 0.06wt%Nb,the combination of 0.05wt%Sb and 0.06wt%Nb could better optimize the rust structure,accelerate the formation of a high proportion of dense and protectiveα-FeOOH,repel the invasion of Cl^(−),and retard the localized acidification at the bottom of the pit.

Incubation and development of corrosion in microstructures of low alloy steels under a thin liquid film of NaCl aqueous solution

Electrochemical measurement, optical microscopy, and scanning electron microscopy were employed to investigate the corrosion behavior of some low alloy steels. The steels were held under a thin liquid film of 0.5wt% NaCl aqueous solution. It is found that the steels with the same chemical composition but different micmstructures exhibit obviously different corrosion behaviors. However, the corrosion behavior of the steels with different compositions but the same microstructnres may be similar in the present investigation. The corrosion rate of bainite is slower than that of ferrite and pearlite. The corrosion products of bainite are uniform and fine. The size of carbon-rich phases produces a great impact on the corrosion of the steels, whether in the initial stage or in the long tenn. It is easy to induce large pitting for carbon-rich phases with large size, which damages the compactness of the rust layer.

Atmospheric Corrosion Monitoring of Field-exposed Q235B and T91 Steels in Zhoushan Offshore Environment Using Electrochemical Probes

This work aimed to in-situ monitor the atmospheric corrosion of steels exposed to Zhoushan offshore environment by using electrochemical noise（EN） technique. A portable EN monitoring system was established and two electrochemical probes（named as Q235 B and T91） were designed. Experimental results indicated that the noise resistance of T91 steel was higher than that of Q235 B steel, revealing that the corrosion resistance of T91 was higher than that of Q235 B. A 60-day monitoring result indicated that the noise resistance was well correlated with the weight loss data. Wavelet analyses results of EN data indicated that Q235 B underwent uniform corrosion and T91 suffered from localized corrosion, which was further confirmed by the surface observation. It is concluded that EN can be used as a new method to identify the corrosion form and corrosion resistance in atmospheric conditions.

Influence of Al Content on the Atmospheric Corrosion Behaviour of Magnesium-Aluminum Alloys

The influence of AI content on the Mg-AI alloys corrosion performance during sodium chloride induced atmospheric corrosion has been studied. It was found that the corrosion rate of three Mg-AI alloys was accelerated with increasing AI content. The poor corrosion resistance was attributed to the galvanic coupling between the phase and eutectic phase or α phase and the formation of porous corrosion products.

INITIAL ATMOSPHERIC CORROSION OF ZINC IN THE PRESENCE OF NH_4Cl

Influence of NH_4 Cl on the initial atmospheric corrosion of zinc was investigated viaquartz crystal microbalance (QCM) in laboratory at 80%RH and 25℃. The resultsshow that NH_4 Cl can accelerate the initial corrosion of zinc. Mass gain increase withthe exposure time, but mass gain in the later doesn't change obviously due to theformation of the insoluble simonkolleite on zinc surface in the presence of NH_4 Cl.Fourier transform infrared spectroscopy (FTIR) and X-ray diffraction (XRD) wasused to characterize the corrosion products. Zn_5Cl_2(OH)_8·H_2O,(NH_4)_2ZnCl_4 andZnO are the corrosion products on zinc. Brief discussion on the mechanisms of at-mospheric corrosion of zinc in the presence of NH_4 Cl was introduced.

Mechanistic Studies of Atmospheric Corrosion Behavior of Al and Al-based Alloys in a Tropical Marine Environment

Atmospheric corrosion behavior of pure Al 1050 A, 5A02 and 6A02 aluminum alloys exposed to a tropical marine environment for 4 years was investigated. Synergetic effect of Cl-deposition rate and time of wetness resulted in an abnormal increase in weight loss and a significant fluctuation in corrosion rate. Pitting corrosion occurred on the three metals. Pits on 5A02 alloy were easy to initiate and inclined to propagate laterally to form higher corrosion area and shallower corrosion pits, while pits on 6A02 alloy presented the opposite appearances. This was further confirmed by the cyclic polarization experiments.

Initial corrosion behaviors of AZ91 magnesium alloy in the presence of SO_2

The effects of SO_2 on the initial atmospheric corrosion of AZ91D magnesiumalloy were investigated in laboratory. Met-allographic observation, SEM (Scanning ElectronMicroscopy), XRD (X-ray Diffraction) and XPS (X-ray Proton Spectrograph) were used to analyze anddiscuss the initial surface morphology of corrosion layers and corrosion products. The corrosionrate of the alloy increases with increasing the content of SO_2. The initial attack has thecharacteristics of localized corrosion and preferentially concentrates on a phase. MgO and Mg(OH)_2form at first, which provide a protective layer, then the existence of SO_2 decreases the pH of thethin solution on the alloy, accelerates dissolution process, and promotes the formation of MgSO_3centre dot 6H_2O and MgSO_4 centre dot 6H_2O, meanwhile cracks were found on the corrosion productswith corrosion continuation. These soluble corrosion products and the cracks provide the paths forfiltering oxygen and corrosion pollutants into the matrix, which results in severe localizedcorrosion and the loss of protective function of film.

Atmospheric corrosion of carbon steel in acid rain by accelerated weathering test

In this work, the accelerated weathering test of immersion-wet-dry combined cycles was carried ont. The artificial acid rain of NaHSO3-(NH4)2SO4-H2SO4 solution system was selected as immersion solution. Acidity of artificial acid rain affected corrossion weight loss of carbon steel, and pH 4.1 was found to be a critical point. Anodic reaction resistance of the samples during accelerated weathering test increased with exposed time by electrochemical measurement.

Influence of Formic Acid on Corrosion Behavior of Bronze under Thin Electrolyte Layer

The corrosion behavior of formic acid towards bronze under thin electrolyte layer(TEL)was investigated by means of electrochemical measurement.Bare bronze had smaller self-corrosion current density than Cu_(2)O patina bronze and CuCl patina bronze,while it had higher polarization resistance.The corrosion behavior of bronze materials had differences in various TELs and bulk solutions.The critical thickness of TEL for Bare Bronze was 200μm,which normly occurred in the transformation from anode control to cathode control.The thickness of TEL had a negligible effect on the corrosion rate of Cu_(2)O patina bronze when it was greater than 150μm.For CuCl patina bronze,the corrosion process accelerated with thinner TEL.SEM was used to analyze the morphology and composition of corrosion products.Cu_(2)O,Cu(OH)(HCOO)and Cu(HCOO)_(2)were formed on the surface of Bare Bronze and Cu_(2)O patina bronze,nevertheless,the main corrosion product of CuCl patina bronze was Cu_(2)Cl(OH)_(3).

An Early Stopping-Based Artificial Neural Network Model for Atmospheric Corrosion Prediction of Carbon Steel

The optimization of network topologies to retain the generalization ability by deciding when to stop overtraining an artificial neural network(ANN)is an existing vital challenge in ANN prediction works.The larger the dataset the ANN is trained with,the better generalization the prediction can give.In this paper,a large dataset of atmospheric corrosion data of carbon steel compiled from several resources is used to train and test a multilayer backpropagation ANN model as well as two conventional corrosion prediction models(linear and Klinesmith models).Unlike previous related works,a grid search-based hyperparameter tuning is performed to develop multiple hyperparameter combinations(network topologies)to train multiple ANNs with mini-batch stochastic gradient descent optimization algorithm to facilitate the training of a large dataset.After that,one selection strategy for the optimal hyperparameter combination is applied by an early stopping method to guarantee the generalization ability of the optimal network model.The correlation coefficients(R)of the ANN model can explain about 80%(more than 75%)of the variance of atmospheric corrosion of carbon steel,and the root mean square errors(RMSE)of three models show that the ANN model gives a better performance than the other two models with acceptable generalization.The influence of input parameters on the output is highlighted by using the fuzzy curve analysis method.The result reveals that TOW,Cl-and SO2 are the most important atmospheric chemical variables,which have a well-known nonlinear relationship with atmospheric corrosion.