A typical failure analysis provides the failure mechanism, which can then be used to determine the definitive cause(s) of the failure so that future occurrences can be prevented. This latter phase is referred to as the “root cause analysis” (RCA).
Such a determination is most critical for corrosion-related failures. Upgrading the metallurgy is often the most logical solution to a corrosion failure. However, this approach may not be adequate if the exact or underlying cause(s) is not properly identified since process conditions may not be as originally designed. Thus, the selection of the upgraded alloy may be insufficient or overly conservative. An RCA will also greatly assist an investigation of the release of a hazardous chemical.
A root cause analysis is designed to identify the interrelating causes of the failure and includes a complete understanding of how to fix the problem and potential underlying issues. Recommendations may include not just a change of metallurgy but also modifying the process operation and conditions. Although an RCA starts with a thorough failure analysis, other major components include a cross-functional team and an organized procedure with assigned tasks.
Root Cause Analysis Functions
The RCA starts with a failure analysis that identifies the most likely failure mode but generally lacks examination of the pertinent external factors that were the actual cause of the failure. For example, failure mode might be concluded as a pitting chloride attack; however, the reason for the corrosion to occur at the failure site must be determined to prevent future attacks at that location or somewhere downstream. Note that without solving the initiating problem, just upgrading the metallurgy might (in the process) result in moving the corrosion problem to another location.
The next step is to assign a cross-functional team to assess the ultimate cause of the failure; identify the underlying conditions that led to the failure; and recommend corrective actions that may include process or design changes, additional maintenance requirements, improved training, or metallurgy upgrades at various locations. This team should include operational, maintenance, metallurgists / corrosion engineers, quality control of product specialists, engineering/process design personnel, and management.
Information Gathering
A critical part of the root cause analysis is the team’s collection and evaluation of process data, logs, and operational procedures. Such information can include:
- Service life history
- Maintenance and inspection records
- Operational excursions with details on temperature and pressure excursions
- Startup and shutdown practices and responses, including emergency shutdown procedures as well as the number of these
- Design and repair changes
The data analysis is an iterative and integrative process that examines the sequence of events with the goal of determining the underlying factors that led to the observed failure. Often, non-contributory factors might be noted that were unrelated to the failure being investigated but had the potential of causing a different failure or extensive process disruption. These must be addressed and should be compiled in a future actions list.
Once the process data, logs, and operational procedures have been summarized, the team can use that information to identify the causal factors. Questions that help to identify these factors include:
- What sequence of events led to the problem?
- What conditions allowed the problem to occur?
- What problems co-exist with the central problem and might have contributed to it?
The 5 Whys
As many causal factors as possible should be recorded. The team starts with the problem and brainstorms the causal factors. The “5 Whys” technique is an excellent tool for troubleshooting and identifying the potential causal factors. In practice, it may be necessary to ask “why” more than five times or fewer than five times. With this approach, the team continually keeps asking the question “why” until the underlying source of the potential factor is identified and an appropriate countermeasure is apparent. For example:
- Why was pitting corrosion observed at the failed location?
- Why did liquid droplets of water form?
- Why did the temperature decrease below the dew point?
- Why does the pump fail so frequently?
- Why is there no operational procedure for such disruptions or scheduled maintenance?
Fault Tree Analysis and Other Approaches
A more sophisticated approach is to use the 5 Whys as part of a logic/issue tree or Fault Tree Analysis (FTA). The FTA methodology is a systematic, backward-stepping process in which the FTA is constructed as a logical illustration of the events and their interrelationships that can produce the observed failure. The FTA is a graphical approach to display the possible combinations of sequences that might lead to the observed failed occurrence, which is noted at the top of the fault tree. Moving through each gate of the fault tree requires the event or outcome to have occurred or be required for passage to the next level. Some events may be unrelated, a correlated factor, a contributing factor, or a possible root cause. The FTA should include an analysis of potential unwanted interactions, adverse secondary faults, and the impact of human interaction on the system.
The FTA is often limited to major components, while Cause and Effect Chain charts—a type of Fishbone Diagram and also referred to as Cause Mapping®—provide a deeper analysis of causes and effects that is designed to uncover more hidden causes. One advantage of this technique is the focus on evidence-based causes. Another is root causes that cannot be eliminated because they represent fundamental requirements of the project, which can lead to process modifications that may minimize the impact of the cause leading to the observed failure.
Final Steps of the Root Cause Analysis
At the conclusion of the RCA, the team must summarize the underlying causes and their relative contributions, and then identify system problems that must be addressed for the path forward. Such system problems might include process, component, and human factors, as well as a combination of these. Ownership of the assignments prescribed and due dates for completion are necessary. In some instances, more than one failure path will be concluded. Any secondary mode(s) must also be addressed, as these could become a future failure mode(s).
For corrosion types of failures, corrosion testing may be required. Such testing requires that the process environment be as closely mimicked as possible; unfortunately, incorporating the necessary aggressive conditions may not be feasible. After the action items for correction have been implemented, regular inspections and evaluations are required to determine whether the actual problem has been solved and whether additional issues have not been created.
***
Note: Although this article has emphasized the application of root cause analysis to corrosion-related failures, RCA is widely used in many industries where the underlying cause of the failure was/is not obvious.