Gustaf Östberg

Evaluation of a design for inconceivable event occurrence

Abstract The importance of inconceivable events increases with growing complexity of designs and increasing consequences of failures. Airplane crashes and accidents of nuclear power plants are spectacular examples but also more mundane cases warrant consideration. It is difficult to deal with inconceivable events because they usually depend on interference by some factors which are not readily included in the models and systems used for design of industrial products with respect to normal performance. To be able to diminish the risk of occurrence of inconveivable events one has to recognize that detection of such phenomena requires another state of mind than that typical in normal design work. There are organisational and managerial obstacles that may prevent measures to be taken in this direction. Hence success in dealing with inconceivable events means that one has to go beyond present practice in industrial design.

What goes on when a designer thinks

Design can be thought of as a model for such endeavours as are intended to result in industrially manufactured products by means of thinking and other intellectual activity. Familiarity with the thinking involved in the designing process is important, not only for those engaged in training designers, but for anyone desirous of systemizing the endeavour. One procedure for approaching an understanding of the way designers think is to describe it with the help of different metaphors. There are some metaphors for thinking about complicated and interrelated phenomena that should lend themselves to illustrating the thinking involved in design, albeit with reservations — e.g. the difficulty of allowing due consideration for what is unthinkable and indescribable in this context.

Statistical evaluation of elevated temperature minimum yield/proof stress values for design purposes

Abstract Evaluations of minimum elevated temperature yield/proof stress values have been performed for one ferritic and four austenitic heat resistant steels. Besides analyses with the methods recommended by the ISO, unbiassed statistical analyses have also been performed. After some modifications the ISO methods yielded results in good agreement with experimental data and with Swedish design stress values. The results of the statistical analyses were found to be sensitive to the distribution of experimental data which, for some of the steels tested, was rather irregular. Thus, the fact that the distribution was not normal but more rectangular resulted in too low values for the desired level of confidence. This effect could not easily be compensated for since the shape of the distribution varied between steels. Attempts to determine statistically the degree of confidence inherent in the ISO methods were not very successful. However, comparisons with experimental data show that the degree of confidence is 99 per cent or better.

A survey of expert opinions on the effectiveness of non-destructive examination of pressure vessels

Abstract There is an increasing awareness of the fact that our knowledge about the risk of not detecting defects in heavy section, welded pressure vessels is unsatisfactory. There is only limited understanding of what confidence can and should be given to non-destructive examination. In order to improve the basis for further development as well as for probabilistic assessments, an inquiry into the efficiency of non-destructive examination of pressure vessels was made. Six qualified and experienced specialists on non-destructive examination were interviewed in order to determine their opinion about the efficiency of defect detection. The observations showed that such specialists do not possess the knowledge of statistics that is desirable for a trustworthy assessment of probabilities. They proved to be very uncertain about the efficiency in question, often referring to the assumed high quality of the manufacture of the vessels as a guarantee of safety. In the light of these observations, the current views on the efficiency of defect detection therefore, appear less trustworthy than commonly believed.

On the trustworthiness of safety assessments for nuclear pressure vessels

Abstract The progress made in improving the safety against failure of nuclear pressure vessels has relied, among other things, on intangible factors like good workmanship, tacit knowledge, sense of responsibility, etc. This has long been recognized by those for whom this effort is made. The trust of the public and decision-makers is based on an appreciation of the awareness of the technological community also of the importance of non-technical, critical aspects. In order to maintain the trustworthiness of safety assessments in this respect it is therefore necessary to follow up previous research and development work with studies of problems of modelling for probabilistic calculations, collection of data on defect detection, intelligence concerning inconceivable events, competence and managerial aspects.

A review of present knowledge concerning the influence of segregates on the integrity of nuclear pressure vessels

Observations have been made in tests on steel for nuclear pressure vessels, indicating that segregates in the steel may reduce its resistance to failure due to propagation of cracks. In order to assess this risk, one would need information and data on the occurrence of segregates, in particular their structure and distribution, as well as on the fracture toughness of segregated steel. The available information in these respects is insufficient for a regular fracture mechanical probabilistic analysis. In view of the demand for a judgement of the risk in question, it would be worthwhile, nevertheless, to make an evaluation of various pieces of circumstantial evidence relating to segregates and their influence on the integrity of nuclear pressure vessel steel.

What is a materials data system

During the last two decades considerable efforts have been made to develop computerized data systems for engineering materials. The results have not come up to the expectations of systems that can be used by designers for selecting of materials. Some factors have been recognized as responsible for the slow progress. It has proved difficult, however, for those involved in this development to make use of such information about the systems in question. Conflicts have occurred between different parties involved. It now seems that too little attention has been paid to the historical background of knowledge and systems related to civiliziation and society. A recently published review of world history from the point of view of the emergence of information technology has provided a basis for some considerations about the nature and roles of data systems. It can be shown that certain activities or features characterize each period of civilization and the development of computers and computerization should be considered as part of this sequence. The same is true for the world history of thinking and knowledge. In the present paper these historical considerations are combined with the observation that in their thinking and perception of ideas people appear to carry with them traditions from the past in varying degrees. Hence differences of opinion about the nature and roles of materials data systems may be explained by differences in the imprint of historical developments on peoples ideas. In order to apply this lesson to the development of materials data systems, one has to learn also from the varying interpretations of world history that have been written through the ages. The best approach will probably be incremental and pragmatic rather than holistic and synoptic.

Some reflections on relationships between structure, properties, and performance of engineering materials

The main object of materials science and technology is relationships between structure and properties. The practical use of engineering materials also involves consideration of things other than current theories for such relationships. Uncertainties of various kinds prompt reflections on the validity and applications of common practice based on experience. Among other things, this is true for decisions relating to reliability and safety. Hence, fundamental issues like determinism vs. probabilistics may become interesting to anyone who wants to deal efficiently with problems related to the use and performance of engineering materials.

The influence of technical testing methods on perceptions of constructions

In principle, methods of testing the reliability and safety of constructions are designed with the aim of representing critical conditions for their use. Because of variations within technical systems, including the one for which a certain test was once developed, the validity of the method also varies. The adoption of a standardized test for different applications had to be based on the assumption that the test deals adequately with a real situation. As a consequence of this acceptance of a trustworthy relationship between test and reality, an imprint of the former on the latter occurs. This effect may even cause the perception of a reality to be defined by the corresponding test. Examples of such an influence may be found in the field of materials testing. Related phenomena can be dealt with by reference to current concepts of cognitive science and psychology of the perception of images.

On the Management of Incompatibilities in Knowledge Systems

In our traditional school education we are induced to think that we can apply systems concepts when dealing with problems not only in mathematics, physics and chemistry but in other fields also. The success of the systems approach in these sciences makes us believe that the same methods can also be satisfactorily applied to live organisms. The failure of systems analysis to account for the behavior of such organisms is often blamed on deficiencies in the logic of the systems. It appears, however, that the main reason for the lack of success of attempts to develop, for instance, computerized data bases for materials to be used by designers is that they have neglected other factors than the hard, quantifiable ones. This is particularly true for tacit or personal knowledge and unknowable factors. In order to cope with these deficiencies and the incoherence in so-called natural systems one has to consider them in their own context. Even if most incompatibilities can then be overcome by means of management, there may remain value conflicts that cannot be settled except by the use of power.