CAHR

(more see downloads)

CAHR - A Human Reliability Assessment approach

CAHR means "Connectionism Assessment of Human Reliability". The Database-System CAHR is a tool for analyzing operational disturbances, which are caused by inadequate Human actions or organizational factors. It was implemented using MICROSOFT ACCESS. CAHR contains a generic knowledge base for the event analysis  that is extendable by the description of further events. The knowledge-base contains information about the system-state and the tasks as well as for error opportunities and influencing factors (PSFs).
 

The method applies a new way in the analysis and assessment of the human role in technical systems. It's philosophy is:

 

Areas where CAHR was applied inculde

The model allows to find causes and improvements for human errors in a single event as well as an assessment of error possibilities and measures in future settings.

In contrast to past methods for event analysis, the process developed in this study is an analytical and not a classifying method. In the method, an event is systematically broken down, analyzed, and described for the purpose of a qualitative error and cause determination. The analysis method proceeds from general questions to substantive information that was observable during the event, and, via actions it moves on to error indications and performance-shaping factors. This object oriented procedure facilitates the analysis especially when initially only little information is available about the event. Although the method developed here also employs known taxonomies from literature for support during analysis, it is not tied to any fixed, predetermined descriptors; it is, in other words, an open procedure. In that way, the approach makes it possible to preserve also the original information of the events and makes the analysis of an event replicable in general. Besides, there is no compulsion to categorize an event in a certain classification scheme that might possibly not reproduce actually observed aspects correctly and that could thus lead to misinterpretations.
 
 

Figure 1-1 Structure of the System

 

 

Structure of the system

CAHR concsists of the following Modules:

 


Here we put in all relevant information on the event regarding general characteristics of the plant and of the system state (Figure 1-2). By simply clicking on the arrows, one can quickly enter all necessary characteristics of the procedure. The contents of characteristic input can also be expanded.

Figure 1-2 Put in Characteristics into the Databank


 


The event descriptions are filed in a text field and a commentary field (Figure 1-3). The fields can be searched for certain concepts by means of full text search.

 

Figure 1-3 Event Description in Databank

 

Input of Relevant Event Information


All relevant information concerning human actions is put in interactively. Figure 1-4 by way of example shows the input into the databank for the sub-taxonomy object-person. In event analysis, one builds up the input of general questions on information items that were observable during the event (object, action) into error data and performance-shaping factors. This is done in an implicative way, that is to say, for example, one may enter only one error in the "indication" column if an object and an action have already been stated. The same applies to the performance-shaping factors in the column "property." In that way one can make sure that the effectiveness relationship between actions, errors, and performance-shaping factors will be illustrated so that it is determined what the performance-shaping factor acted on.

 

Figure 1-4 Input of Human Factor Information Into Databank

  

This object oriented procedure facilitates the analysis of events because one starts with information that was observable on the technical system. In addition, it avoids allocation of culpability because the event is analyzed by starting with the error situation and not by starting with the persons who are involved. By offering decision making aids during concept selection to the person who wants to describe an event with the help of this procedure, the concepts are sorted according to frequency of use.
 

Qualitative und quantitative Assessment

 

The connectionism approach allows qualitative and quantitative analyses of data in a uniform model. In that way it becomes possible to supply both information for the evaluation of human reliability and for the optimization of the technical system in a uniform database. First of all, one can interrogate frequencies of interrelationships of any concepts within the data structure (for instance, relationships between errors and performance-shaping factors). Permissible logic tie-ins between concepts here are AND, OR, AND NOT, related to the various classes of the man-machine system and the various description stages (object, action, indication, property). With the help of the connectionism approach, one can implement all of these analysis possibilities on any degree of detailing of the event description (for individual concepts or general classes).

 

Figure 1-5 Illustration of a Quantitative Inquiry in the Databank System

 

For each interrelationship, that was observed with a frequency of more than 0, one can determine qualitative interrelationships on the basis of all cases of available information observed so far. A typical inquiry to the collective data is for instance: "How many errors of confusion were observed, what performance-shaping factors were observed, and what precautions were initiated against repetition?" Figure 1-6 and Figure 1-7 show this inquiry within the databank. It is answered by the databank system with the help of various lists; information items gained can be traced back all the way to the individual cases.

 

Figure 1-6 Illustration of a Qualitative Inquiry in the Databank System

 

New generic concept and sub-concept relationships for any concepts can be established by means of a class editor and can be used in the event analysis as if they had been mentioned themselves (Figure 1-7).

 

Figure 1-7 Illustration of Class Formation

Tables for Diagram Preparation

Data can be processed in another way by inquiries and cross reference tables to illustrate simple interrelationships or reciprocal relations in the form or tables or diagrams. Inquiries are styled interactively and are stored. The data can be processed graphically in various ways. All possibility from the customary WINDOWS PRODUCTS are available to vary the graphic processing (for example, copy the graphs and tables in Text Processing or Table calculations).

 

Figure 1-8 Analysis of Characteristics in the Databank System