Assessment of the Scatter Ranges in the Long-term Characteristics of Heat-resistant Steels by Means of Data-mining-Method Application

Project Number 273

When exposed to higher temperatures steels have time-related expansion characteristics (creep). High-temperature components must therefore be designed on the basis of time-related characteristics.

The creep resistance has a scatter range of +/-20 % in direction of the expansion. This can result in an increase of the crack period by up to several 100 % (at +20 %) or a reduction by up to 80 % (at -20 %) related to the mean rate.

The scatter of the rates is due to the microstructure (chemical composition in context with heat treatment, manufacture, type of semifinished parts, and component geometry). Delimitation of this scatter by narrower definitions, i.e. relating to composition, cannot be achieved by economically reasonable measures. Both design and assessment of residual service life are calculated on the basis of the mean creep resistance rate.

Data-mining-tools look for data patterns and extract information from data. These intelligent technologies have a high innovation potential, could establish in several application areas and have contributed towards considerable structural changes and developments in the technical and economical field.

Three steps are envisaged:

By means of a neuronal network a prognosis model is elaborated based upon conventional data; it reliably classifies a melt into an existing scatter range.
In a second step, additional information on the microstructure is added to the conventional data base. Subsequently, the melts are classified into the scatter again while an improved approximation of the model is expected due to the additional evaluation of the microstructure.
In the third step operationally stressed components are assessed with the model regarding their degree of fatigue and/or residual service life.

The Stuttgart Materials Testing Institute (MPA) is realising compilation of the data basis (choice of data and characteristics) and data transfer via the ALIAS data bank. Modelling and analysis of the forecast quality are carried out at the Metal Shaping Institute (IMF) of the Bergakademie Freiberg Technical University.

The project will be carried out from January 1, 2005 to June 30, 2006, and headed by Dr. Karl Maile (MPA) and Dr. Heiner Gutte (IMF). Under the project No. 198 it will be mainly financed by means of the research association of the Arbeitsgemeinschaft der Eisen und Metall verarbeitenden Industrie (Association of iron- and steel processing industry, AVIF).The project is supported by the Working Group W12 (“Residual Service Life”) of the Association “Heat-resistant Steels”. From the part of VGB, the Technical Committee “Materials and Quality Supervision” is involved in the investigations.