Presentation on theme: "Welcome to the Demo of Event analysis and Assessment of Human Reliability with CAHR. Continue by pressing any key; ESC to stop. CAHR means "Connectionism."— Presentation transcript:
Welcome to the Demo of Event analysis and Assessment of Human Reliability with CAHR. Continue by pressing any key; ESC to stop. 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). Demo Version,
The method was first applied to the analysis of events in Nuclear Power Plants: First Study 1994 to 165 for events in Boiling Water Reactors (BWR) Second Study 1998 to 55 events in Pressurized Water Reactors (PWR) Third Study 2000 to events with communication problems The results of the method were used for: The assessment of cognitive aspects in the safety assessments of Nuclear Power Plants The further development of methods for the analysis and assessment of Human Reliability (HRA-methods), particular in the field of so-called Errors of Commission Currently we are also developing the method for: The prediction of cognitive errors in aircraft incidents The prediction of human behavior in decision problems (in process industry and transportation)
The method has an advanced view on the analysis and assessment of the Humans role in technical systems. General Procedure of the Method CAHR is structured into Event Analysis CAHR is structured into Event Analysis....and.... Event Evaluation Connecting link between both is the analysis structure of the Man-Machine System (MMS) because it considers... Organizational Aspects Ergonomical Aspects Cognitive Aspects Technical Aspects By combining several MMS an entire event can be described as a working-system. The Philosophy of the method is: Focus of analysis is the working system and not the human. Human blame or guilt is not an issue of analysis (or even distracts it). Human error result from the interrelation of several situational and causal factors of the working system The method uses a fixed structure but no fixed taxonomy (open method) and therefore is usable for any event or any technical area Strict differentiation between observable information (phenotypes) and causes (genotypes) in the event analysis and description
General informationPlant and plant typeDate and time of the eventCategorization of the importance of the eventState of description of the event The Case Description mages and describes general information about the event. Information for affected systems whether is was a human error or whether the human has coped with the event successfully
The Case Description also manages the event report. Textual description of the event Photos, further Information (via OLE-Objects)
The description of an event is implicative, i.e.: an error can only be described if there is some information already contained in the object and action column. The same holds for the PSFs. This ensures that the interrelations between observable information and errors and causes is represented correctly and hence it is clear where an error occurred and a PSF had been effective. Now we enter the most interesting part: The analysis regarding human errors. The input procedure works from general observable information (Object, Action) to errors and influencing factors (PSFs) related to the categories respectively classes of the MMS. The column Element enables the analyst to describe additional information that is useful and important for understanding the event A punctuation mark is used to distinguish statements and to assure a correct representation of the information regarding Object, Action, error and PSF. Maybe for instance that Valve A was opened too late and Valve C was closed too early.
The relevant information for human actions are described interactively. A Double-Click on the table provides us with the valid taxonomy for the class and aspect to describe, here for instance: What are the error-types of activities? Click...
The relevant information for human actions are described interactively. Another step of event description Events often consist of various sub-events: other persons involved other phases of the event different places For considering this we choose NEW
The relevant information for human actions are described interactively. For a new sub-event If the precaution measure should be described. This is of importance, if one wants to compare the realized measure and the real causes (what often is not equal). For analyzing the cognitive aspects of this sub-event.
The relevant information for human actions are described interactively. Here the new sub-event that has to be described using the taxonomy Here the precaution measure undertaken in this event
Now as we have analyzed and described such many events, what to do with them? Evaluating of course! CAHR performs this using: Simple queries regarding the events Qualitative analyses of the causes of human error Quantitative estimation for human error probabilities Event 1 : Event i : Event n
At first it is desirable to get an overview about frequencies, status of analysis etc. Let us see how many events of Boiling Water Reactors (BWR) were elaborated and which of them are errors and performances of Humans... Click and...
Errors Performance Quiet convenient to have Humans in the plant, isnt it? Would you blame an operator or speak about Human Failure considering such a portion of good performance? At first it is desirable to get an overview about frequencies, status of analysis etc.
Subsequently we can analyze the error deeper. Why do Humans make errors? We get a more precise picture, if we analyze the events regarding a question of analysis we have e.g.: Error of a certain group of personnel while operation a valve. The procedure of defining the query is analog to one of describing an event. Click and... We can combine different query items using logical relations like AND, OR, NOT
Subsequently we can analyze the error deeper. Why do Humans make errors?... we find 23 equal cases in 220 events or 25 in 400 sub-events...and of course uncertainties because an event is never described completely But with CAHR we are able to come up with an estimation for the Human Error Probability (HEP) that is - based on all performed studies - compatible to the estimations of other methods like THERP for instance.... with the extensive algorithm of the connectionism part of CAHR... Which cause leads to such an error? Click and...
Subsequently we can analyze the error deeper. Why do Humans make errors?... we find a series of causes in the areas of Organization Cognition Ergonomics Technique which led to the error in their interrelationships.... the frequencies... and the relative frequencies This enables a decision whether there are systematic errors and to find the optimal improvement measures. But, be aware of the interrelationships - the factors do interact with each other
No event is like the other! Quiet different terms are necessary to describe them sufficiently with all the individual relationships. Therefore, a taxonomy has to be proofed and revised anyway after each event. The variability of the events is covered by the fact that CAHR only prescribes the description structure and not the taxonomy. To deal with this openness in the event-evaluation, CAHR has some tools: A class-editor A taxonomy-editor
Using the Class-Editor different descriptors can be combined... For instance we can define a class for different persons In the evaluation of events we now are able to use a class in the same way as we can use the original descriptors.
Using the Taxonomy-Editor, descriptors can be changed or new descriptors can be implemented
Operational experience Module for HRA Probabilistic data Interface Event-InputModule for Diagnosis Factors and Causes HRA / Qualitative Analysis data Qualitative User-Output Event Possible factors and causes Connectionism Data-base Cause description Situational Pattern Question of Analysis User-Input EventsTaxonomies Class Editor Event-Database Query Summary of the method CAHR ACCESS
What can you do with CAHR? If you want to have a retrospective view into the past, you can......analyze and describe various events in a systematic manner......even from different technologies If you want to have a prospective view into the future, you can......can understand the relationships of factors leading to the error chain of an event...look for causes and situational conditions of importance for human errors...make safety assessments of human errors in risk analysis...find precaution measures for your system and assess the quality and effectiveness of them...improve and optimize precaution measures in your system
We hope you have enjoyed the tour. Sträter, O. (1997) Beurteilung der menschlichen Zuverlässigkeit auf der Basis von Betriebserfahrung. GRS-138. GRS. Köln/Germany. (ISBN ) Sträter, O. & Bubb, H. (1998) Assessment of Human Reliability based on Evaluation of Plant Experience: Requirements and their Implementation. Reliability Engineering and System Safety. Elsevier. Vol. 63, No. 2, p Reer, B., Sträter, O., Dang, V. & Hirschberg, S. (1999) A Comparative Evaluation of Emerging Methods for Errors of Commission Based on Applications to the Davis-Besse (1985) Event,. PSI. Schweiz. Nr (ISSN ) Sträter, O. & Reer, B. (1999) A Comparison of the Application of the CAHR method to the evaluation of PWR- and BWR-events and some implications for the methodological development of HRA. In: Modarres, M. (Ed). PSA 99 - Risk- Informed Performance-Based Regulation. American Nuclear Society. LaGrange Park, Illinois, USA. Do you desire to analyse and assess human errors? We would be pleased to assist you. You are interested in this data-base? We would be pleased to make an offer. Kontakt & Entwicklung: Prof. Dr. Oliver Straeter Universität Kassel Fachbereich Maschinenbau Arbeits- und Organisationspsychologie Heinrich-Plett-Strasse 40 D Kassel Tel: