For corrosion evaluation of steel products in atmospheric corrosion environment, exposure test of test specimens has generally been conducted. However, it is not possible to evaluate the detailed changes of corrosion loss due to environmental variance.

For this problem, the electric resistance technique which converts the corrosion loss from continuously measuring the increase of electric resistance seems to be more suitable. In order to measure the corrosion loss with high accuracy, it was found that the structure of the sensor which can minimize the effect of the localized corrosion which controls the whole resistance by increasing the local electric resistance and make the dry and wet cycle of the sensor equal to the evaluation target is important.

The corrosion inhibition performance of imidazoline derivatives, which have different dispersion quality on carbon steel was investigated with electrochemical tests and X-ray photoelectron spectroscopy in model oil and gas environments. From the potentiodynamic polarization results, used imidazoline derivatives effectively inhibited anodic and cathodic reactions. X-ray photoelectron spectroscopy analysis results suggest that used imidazoline derivatives adsorbed on the carbon steels to inhibit corrosion reaction of the steels. The possible corrosion inhibition mechanism was proposed.

In case of RBM（Risk Based Maintenace） assets risk evaluation, metal corrosion mechanism has been decided by the experts. But, the perfect estimation of corrosion mechanism to too many corrosion environments is difficult and the population of such experts will decrease in near future. As the countermeasure, AI system is proposed here and continues to reach more reliable RBM. This AI system has 2 Python program codes. One is RBS（Rule Based System）. The other is Decision Tree Analysis. RBS has 172 metal damage mechanisms classified to fatigue, Creep, Wet/Dry Corrosion, Metal degradation and others. RBS and Decision Tree Analysis can predict effectively each asset corrosion mechanism using small amount of data such as kind of factory, kind of equipment, material, chemical environment, and precisely predict the risk for RBM using a lot of data such as chemical concentration. The issue to resolve is that there are too many natural language characters in data reducing the reliability of estimation of corrosion mechanism.

In order to investigate the influence of salts including biomass fuels on high-temperature corrosion process, we have tested in sulfur dioxide gas atmosphere where chlorine gas concentration and salt ratio in ash is changed. The corrosion mass loss increased rapidly when the salt ratio in ash increased more than 50 wt.％ with chlorine gas in sulfur dioxide gas atmosphere. These results indicate that the highly corrosive iron chloride is generated by dissolved chlorine gas from the gas phase and iron ion from steel in molten salt.