Root Cause Analysis levels

One additional benefit of root cause analysis is that once the fundamental problem has been identified and addressed many other problems will be eliminated with just one action. It may be found that the high-level problem in T-101 was just one of the problems that will be solved. In particular, the root cause analysis may identify common cause effects. These are problems that affect many systems, and that, if successfully addressed, can eliminate many problems at once. 

The level of root cause analysis that is typically carried out for each of these levels is discussed below based on an example of a seal failure of Pump P-101 A in Example 2 in Chapter 1. 

At this level, the root cause analysis needs to address the supervisors primary responsibilities, which are to make sure that the immediate situation is brought safely under control, and to make short-term changes that prevent the immediate recurrence of this or similar events. 

A key word in the above definition is correctable. Managers at all levels need the root cause analysis to provide them with practical guidance—something that will help them correct their problems, and thereby improve their culture. The need for defining root causes that can help lead to solutions is demonstrated in the following paraphrased definition, adapted from Mark Paradies,... 

One of the difficulties associated with root cause analysis lies in determining the level at which to stop. After all, if the causes of an incident are pursued for long enough the team will eventually be discussing the philosophical, moral, and theological issues to do with human nature. This is obviously absurd a sensible stopping point is required. 

When we look at the factors that are identified in the interviews, we can see that the different factors, and hence, the related explicit uncertainty that is caused by them, can all be related to the different stages of the product lifecycle as, for example, defined by Bedford, Quigley and Walls (Bedford et al. 2006), i.e. concept and definition, design and development, manufacturing and installation, operation and maintenance. We see for example that the concept root-cause analysis (here, uncertainty is caused by this concept because we do not know the extent to which we are able to find the root-cause) can typically be related to the operation and maintenance phase. On the other hand, the concept supplier management (here, uncertainty is caused by the unknown level of quahty with which supplier management is applied) typically relates to the design and development phase. 

For Statistical Forecast, as the company already has a good forecast level with demand planning tools and processes in place, it is recommended to implement a root cause analysis to map and understand the reasons of low forecast accuracy by SKU, and then, implement an effectively action plan to fix the problems. Another opportunity is to implement medium to long term forecast tool to support the generation of forecasts for a longer horizon (e.g., 1-5 years). 

Real-time process supervision is performed by means of a set of tools and methods, which ensure safe process operation in normal situation as well as in the presence of failures or undesired disturbances. Process supervision is conducted by integrating several tools, each specifically designed for a specific activity. These activities are fault detection, fault isolation, diagnosis (root cause analysis), fault quantification (determination of the severity of the fault), and the decision making to accommodate the fault. The presence of a fault is detected at the monitoring level, which determines whether the process is in normal operation or not. Other tools for fault isolation, diagnosis, etc., are executed when an abnormal process state is detected by the process monitoring tool. 

As a result of the potential significance of this occurrence, a formal, detailed root cause analysis was performed. A high level of effort was expended but the effort was justified due to the consequences of a repeat occurrence. 

Frank Bird introduced his model of root cause analysis in his book, Practical Loss Control Leadership (Bird et al., 2003). The model identifies three levels of causation that must be addressed for proper countermeasure formation. Understanding and identifying factors of the incident nnder all three levels is necessary to address the incident at a root level and prevent fntnre occnrrence. 

There are multiple tools for the analysis of root causes of events. The analysis of root causes of hardware failure is well defined, the analysis of the root causes of human performance is well developed in general, but there are still some limitations in the tools to analyse management and organizational factors. In addition, sometimes the root cause analysis is not performed to the level of detail required to identify management and organizational problems. 

One of the simplest root cause analysis techniques is to determine the causes of accidents/ind-dents at different levels. During any hazard analysis we are always trying to determine the root cause of any accident or incident. Experts who study acddents often do a breakdown or analysis of the causes. They analyze them at three different levels ... 

The shift from reactive corrective maintenance toward proactive predictive maintenance represents a significant move toward enhanced reliability. However, efforts designed to identify problems before failure are not sufficient to optimize reliability levels. Ultimately, for enhanced reliability, the root causes of maintenance problems have to be determined, in order to eliminate them. The high-est-priority use of root cause analysis (RCA) should be for chronic, recurring problems (often in the form of small events), since these usually consume the majority of maintenance resources. Isolated problems can also be analyzed by RCA. 

For this simple example, a similar analysis could be based on a quantitative approach, namely, a dynamic model based on an unsteady-state mass balance. The level could be changing as a result of process changes or a sensor failure. However, for more complicated processes, a reasonably accurate dynamic model may not be available, and thus a qualitative approach can be used to good advantage. The diagnosis of the abnormal event could lead to a subsequent root cause analysis, where the source of the abnormality is identified and appropriate corrective actions are taken. 

The theoretical basis of the different causal models becomes more obvious at the upper management level. MORT was the first comprehensive model to include organisational and individual factors at the top management level. At this level, it draws from quality assurance management principles. The SMORT and ILCI models have been influenced by this pioneering work and represent variations on the same theme. The concept of root causes originates from the MORT model. The checklist above shows the different items of a root-cause analysis. 

Previously, it was mentioned that root cause analysis of craters usually is very difficult. That is true, but certain causes turn up again and again. One example is oil and water in compressed air. Water usually does not cause craters, but it is an excellent carrier for oil and droplets of oily water do give craters. Compressed air should be tested periodically for water and oil content using Drager tubes or some other technique. At the point of paint application, there should be no oil detectable and the water level should be less than 150 mg/m ... 

Performance information for the incumbent resin was missing from the early parts of the decision-making process. The decision that the technical problem was the performance of the new resin was based on anecdotal information from plant personnel on the performance of the incumbent resin. That is, the plant personnel believed that the reject level for parts made from the incumbent resin was less than 5 %. A statistical analysis of the part defect rates was not performed. This lack of information early in the process allowed the plant manager to propose a poor technical solution without understanding the root cause for the defect. Later in the troubleshooting process, a statistical analysis of the defect rate indicated that the incumbent resin had a defect rate that was statistically equivalent to the new resin. 

If the root cause is assignable, corrective action needs to be taken and documented. If three consecutive runs over action levels occur, a problem analysis corrective action report (PACAR) must be issued. 

---