Human Error as a Factor of Corrosion Failures

The idea that a piece of equipment breaks down because of the error of a single mechanic in installation or a single design error is nice - but usually it doesn't hold up to scrutiny in the real world. The real roots are usually much deeper than a single metallurgical problem or one error on the part of the maintenance mechanic or even the fact that the machine is being run at twice the design load. When all the human errors and management system deficiencies are included, the typical machine breakdown is the result of somewhere between five and ten significant contributors.

The depth of the analysis into the roots of the failure is the key to accurately unearthing all of the failure sources. In an effort to better understand the general sources of plant failures a failure analysis company decided to look at the last three years' projects to determine the major failure contributors. There were a total of 131 analyses and we list the primary failure mechanisms below:

Note: In defining these five categories there is the possibility of confusion between corrosion fatigue and fatigue. The practice was to assign fatigue as the mechanism in those cases where the component would have eventually failed and corrosion was not needed to effect the failure. In those situations where the component would not have failed without the action of the corrosion, i.e., there was cyclic loading but it was not severe enough to cause cracking without corrosion, the cause was listed as corrosion fatigue. Also, note that only the primary failure mechanism has been listed here since many of these failures actually had multiple physical roots.

After analyzing the data for the mechanisms, the information was examined to determine the major failure causes. These were defined as those causes that were apparent from the component inspection, field analysis, and discussions with plant hourly and salaried, operating and maintenance personnel. Based on discussions with others and our experience, six error categories were selected, operational, design, maintenance practice, supervisory, manufacturing, and original equipment installation. The data was then sorted in these groupings.

According to a previous study reported by Congleton, the attribution of responsibility for corrosion failures investigated by a large US based chemical company was broken down into the following:

This data set indicates that only 8% of corrosion failures is unforeseeable. In other words 92% of the corrosion failures could be preventable!?! If this is true, the small % of corrosion damage attributed to a lack of human performance by the Hoar report, the Battelle studies and many others, i.e. roughly 30-40% of total corrosion damage, would be overly optimistic.

From "Understanding Why It Failed" by Neville W. Sachs, P.E. Sachs, Salvaterra & Associates, Syracuse, NY (back)