The Combination of Modeling Techniques to Predict the Service Behaviour of Heat Resistant Alloys

Abstract:

Article Preview

It is here demonstrated that combinations of various modeling techniques can be used to predict the service behaviour of heat resistant alloys, and sometimes to design “made-to-measure” alloys for specific high temperature applications. A first example is given, where an affordable creep-resistant nickel-base superalloy for operation around 750°C has been designed without any experiment. Based on the analysis of huge databases on existing alloys, Gaussian processes were used to predict its thermomechanical properties. Thermo-Calc allowed to design its fabrication process and to assess its weldability and its thermodynamical stability at service temperature. The alloy has then been tested and the validity of the modeling approach verified a posteriori, in particular a creep rupture life of 100 000 h at 750°C under 100 MPa. A second example is given, in the case of high-carbon Fe-Ni-Cr based alloys for reformer tube applications (HP steels). Their mechanical properties are predicted through the analysis of existing data with artificial neural networks. Parallelly, their thermodynamical stability in operating conditions is assessed using Thermo-Calc in combination with Dictra, to simulate the precipitation of carbides in the austenite matrix during service. It is therefore tried to understand microstructural evolution in service, damage mechanisms, and durability.

Info:

Periodical:

Materials Science Forum (Volumes 539-543)

Main Theme:

Edited by:

T. Chandra, K. Tsuzaki, M. Militzer , C. Ravindran

Pages:

3070-3075

DOI:

10.4028/www.scientific.net/MSF.539-543.3070

Citation:

F. Tancret and J. Laigo, "The Combination of Modeling Techniques to Predict the Service Behaviour of Heat Resistant Alloys", Materials Science Forum, Vols. 539-543, pp. 3070-3075, 2007

Online since:

March 2007

Export:

Price:

$35.00

In order to see related information, you need to Login.

In order to see related information, you need to Login.