Predefined trees

This chapter addresses methods and tools used successfully to identify multiple root causes. Process safety incidents are usually the result of more than one root cause. This chapter provides a structured approach for determining root causes. It details some powerful, widely used tools and techniques available to incident investigation teams including timelines, logic trees, predefined trees, checklists, and fact/hypothesis. Examples are included to demonstrate how they apply to the types of incidents readers are likely to encounter. 

This useful companion disk contains root cause analysis examples, predefined tree examples, practical checklists that can be customized, and incident evidence photograph examples. It includes a quick checklist for investigators traveling to an incident, examples of methodologies that may be usefiil in training the onsite team, and checklists and samples from the text that can be printed out at the incident site to help organize the team s work. 

In general, the companies surveyed use one of two main methodologies to determine root causes. The first involves timeline construction followed by logic tree development. The second involves timeline construction, identification of causal factors, followed by the use of predefined trees or checklists. These two approaches are discussed in detail in Chapter 9. 

Once the causal factors have been identified, the factors are analyzed using a root cause analysis tool, such as 5-AVhys or predefined trees. See Chapter 9 for a more detailed discussion of Barrier Analysis (sometimes called hazard-barrier-target analysis or HBTA) and Change Analysis (also referred to as Change Evaluation/Analysis or CE/A). In essence, these tools act as a filter to limit the number of factors, which are subjected to further analysis to determine root causes. 

Like checklists, the comprehensiveness of the various predefined trees varies. Some are very detailed with numerous categories and subcategories, whereas others may not fully reach root causes. This is hardly surprising, as the predefined trees are essentially a graphical representation of numerous checklists, organized by subject matter, such as human error, equipment failure, or other topics. The more comprehensive techniques were developed from many years of incident experience and management system experience across the chemical and allied industries. 

The use of predefined trees is also discussed in detail in Chapter 9. See the accompanying CD-ROM for examples of predefined trees, including Comprehensive List of Causes (CLC) and SOURCE/Root Cause Map. 

Timeline construction, identification of causal factors, and predefined trees or checklists. 

Figure 9-1, the two flowcharts describing root cause determinations using Methods A and B, presents general frameworks for root cause determination. Method A focuses on the logic tree method using a simplified fault tree approach. Method B focuses on the predefined tree method. 

The diagram can assist the overall methodology to identify multiple causes, when used in conjunction with logic trees or predefined trees. 

Root Cause Determination Using Predefined Trees— Method B... 

FIGURE 9-27. Flowchart for root cause determination— predefined tree/checklist. 

The causal factors need to be examined further to determine why those factors existed. The investigation team may use a predefined tree to examine each causal factor individually. The first causal factor is analyzed starting at the top of the tree, and then working down all of the branches as far as the facts permit. When an appropriate subcategory on one of the branches is identified, it is recorded as a root cause. The remaining branches are checked as one causal factor may have multiple root causes. The procedure is then repeated for each causal factor in turn. 

Several quality assurance tests should be applied when using predefined trees. Firstly, predefined trees are designed to capture most root causes, but may not be comprehensive. A completeness check should be conducted on each branch of the tree to see if there are other root causes associated with the category of that branch that are not listed on the tree. 

Some predefined trees do fully reach down to the root cause level. A system test should be applied to each identified root cause to ensure that it relates to a management system failure. By applying the 5-AAdiys tool to each cause identified at the end of the relevant branches of the tree, the investigator can determine if another underlying cause can be identified. 

After the predefined tree has been used, a final generic cause test should be applied. The plant operating history, especially previous incidents, is considered to indicate if other generic management system problems exist. For example, repetitive failures may indicate generic causes that would not be apparent by only investigating the current incident. It is also an opportunity for a final overall review of the investigation to focus on the big picture, not just individual facts or causal factors. The team should ask, Are there any other causes that anyone has in mind that have not been included ... 

Once the actual incident scenario is understood and its multiple causal factors identified, this information may be used to determine the incident s root causes. One means of performing root cause analysis involves the use of ready-made, predefined trees. A predefined tree provides a systematic approach for analyzing and selecting the relevant elements of the incident scenario. It is a deductive approach, looking backward in time to examine preceding events necessary to produce the specified incident. 

Unlike the procedure followed in developing logic trees, the investigation team does not construct the tree. Rather they apply each causal factor to each branch of the predefined tree in turn, and those branches that are not relevant to the incident are discarded. This prescriptive approach offers consistency and repeatability by presenting different investigators with the same standard set of possible root causes for each incident. 

While the use of predefined trees does not directly challenge the investigation team to think laterally of other possible causes, many predefined trees present a wide range of causal factors, some of which the team may not have otherwise considered. It is therefore possible, but unlikely, that the incident could involve a novel root cause that was not previously experienced by those who developed the predefined tree. The addition of a final test based on another tool, such as brainstorming, can overcome this apparent weakness. 

Application of the MORT technique is based on a predefined tree structure laid out vertically (top down) in eight interconnecting trees. The structure is quite complex, but contains 98 generic problem areas and up to 1500 possible causes. A user s manual 02) provides detailed instructions on how to use the tree. 

The MORT technique has received domestic and international recognition, and has been applied to a wide range of projects from investigation of occupational incidents to hazards identification. It is supported by detailed documentation, and has been subjected to continued development efforts since it was originally introduced. Today there are several predefined trees available from public and proprietary sources that are based, at least in part, upon tbe MORT tool. 

Although there are differences between various predefined trees, the basic method to perform a root cause analysis using the trees is similar whichever tree is used. The following basic steps apply ... 

The first causal factor is then analyzed starting at the top of the predefined tree and working down the branches as far as the facts... 

Predefined trees are relatively easy to use and generally require less training and effort to conduct root cause analysis tban logic trees. 

The following is an example of the use of a predefined tree to analyze an environmental incident. While the structure (numher of branches and levels) and terminology of predefined trees vary, this example demonstrates the basic method. 

Each of the causal factors can now he analyzed for its specific root causes using a predefined tree, as shown in Figure 9-3 f. 

The following is an analysis of one of these causal factors contractor operator (CO) falls asleep. The basic technique works with any of the predefined trees commonly used within the process industry. However, for the purposes of this example, a proprietary tool C) has been selected, and therefore the structure of the tree and the terminology used is specific to that tree. 

In this case, the causal factor (contract operator falls asleep) is identified as a Human Performance Difficulty (one of the four major problem categories at the top of the tree), and the other three categories are discarded. (Different predefined trees use different terminology and structure, but generally cover similar choices.)... 

The investigator then follows the Human Performance Difficulty category through a series of questions (or subcategories). These questions help the investigator identify which of several human performance related branches (sometimes known as basic causes) to investigate further. (Some predefined trees use statements rather than questions, but the selection process is similar.)... 

A different predefined tree may express this question as one or more simple statements, such as ... 

There are a number of quality assurance checks that should be considered when conducting an incident investigation using predefined trees. Most of these checks have already been discussed, but will now be summarized in relation to the corresponding phase of the investigation. 

After the root causes have been identified from the predefined tree, a generic cause test should be applied. By considering the plant operating history, especially other incidents that may indicate repetitive failures, the investigator may identify other generic management system problems. 

---