Application of Extreme Value Statistics to Corrosion

Area of Science:

• Materials Science
• Statistical Analysis
• Corrosion Engineering

Background:

• Corrosion failure prediction requires quantitative evaluation, particularly for localized corrosion.
• Extreme value statistics offers a robust framework for this assessment.
• Understanding corrosion probability is key to accurate failure prediction.

Purpose of the Study:

• To review the application of extreme value statistics in corrosion analysis.
• To highlight the utility of extreme value statistics for predicting corrosion failure.
• To propose a standardized procedure for estimating maximum pit depth.

Main Methods:

• Application of extreme value statistics, including Gumbel distribution.
• Utilizing the concept of return period for surface area-dependent pit depth evaluation.
• Analysis of reported corrosion failure cases, primarily from Japan.

Main Results:

• Extreme value statistics is effective for assessing maximum pit depth and minimum time for crack generation.
• Gumbel distribution can evaluate maximum pit depth based on surface area.
• A standardized procedure for maximum pit depth estimation is proposed.

Conclusions:

• Extreme value statistics is a valuable tool for quantitative corrosion failure assessment and prediction.
• Data classification based on corrosion mechanisms is essential prior to applying extreme value analysis.
• The proposed methods are particularly useful for localized corrosion scenarios.