A Study on Stochastic Models for Localized Corrosion

对于不锈钢等耐腐蚀材料，已知超过一半的腐蚀失效是由局部腐蚀引起的。众所周知，局部腐蚀数据在空间或时间上表现出广泛的分散性，即统计和概率变化。为此，需要对局部腐蚀扩展的统计和概率性质进行更详细的研究，以便更可靠地评估各种结构材料的失效寿命。本研究旨在完善先前提出的点蚀出生和死亡随机模型，并在该模型的基础上用蒙特卡罗模拟方法模拟实际的点蚀过程。将模拟的点蚀电位和诱导时间分布与实验观测的点蚀电位和诱导时间分布进行了比较，以确定该模型的可信度。出生和死亡随机模型的基本假设为…1)坑的产生服从Posson随机过程，坑的感生时间服从指数概率分布，坑产生率对应的转移概率由势函数给出。2)坑核尺寸随时间增大，速率定律的比例常数表示为势的指数函数。3)在临界尺寸以下，再钝化会使坑核停止生长。坑核的死亡过程也服从泊松过程，因此到死亡的寿命分布规律也服从指数概率规律。基于上述模型和假设，蒙特卡罗模拟得到了如下结果：1)点蚀势分布服从正态概率分布。2)点蚀电位中值与电位扫描速度的平方根成正比。3)成坑诱导时间的概率分布服从扭曲分布。上述结果与实验中观察到的结果具有较高的可信度。基于该模型，分析了非晶态合金热处理后再结晶对非晶态合金异质性的影响。较少

For corrosion resistant materials such as stainless steels, more than half cases of corrosion failure had been known to be caused by localized corrosion. It is well known that data of localized corrosion show wide scatter in space or in time, namely statistical and probabilistic variations. For this reason, statistical and probabilistic nature of localized corrosion shckild be studied in more detail in order to evaluate failure life of materials for various structures with more confidence.This study aims to refine the birth and death stochastic model of pitting corrosion which had been proposed previously and to simulate the actual pit foreation process by using Monte Carlo similation based on the proposed model. Simulated distributions of pitting potential and induction time for pit generation were compared with the observed distributions by the experiments in order to establish the confidence of the proposed model. The basic assumptions for the birth and death stochastic model are as … More follows: 1) Pit generation obeys a Posson stochastic process, so that the distribution of pit senation time follows exponential probability distribution.Transition probability corresponding to the pit generation rate is given by a function of potential. 2) The size of pit nucleus increases with time and the proportional constant of the rate law is expressed as an exponential function of potential. 3) Below a critical size, the growth of pit nucleus will stop due to repassivation. Death process of pit nucleus obeys also Poisson process, so that the distribution law for life time until death follows also exponential probability law. 4) Large pit which grows over a critical size will not die and continue to grow.Based on the above model and assumptions, Monte Carlo simulation provided the following results: 1) The distribution of pitting potential obeys normal probability distribution as a first approximation. 2) Median of pitting potential is proportional to the square root of potential sweep velocity. 3) The probability distribution of induction time for pit generation obeys distorted distribution. The above results were found to simulate the results observed in the experiments with a high confidence. Increase in heterogenuity due to recrystalization of amorphous alloys by heat treatment was analyzed based on the model. Less