DEVELOPMENT OF THE RAMBERG-OSGOOD MECHANICAL STRESS-STRAIN CURVE USING THE ARTIFICIAL NEURAL NETWORK METHOD TO EVALUATE MECHANICAL BEHAVIOUR OF 316L STAINLESS STEEL IN THE LIQUID LEAD

Abstract

Romania, through RATEN ICN, is involved in the construction of the ALFRED demonstrator (Advanced Lead Fast Reactor European Demonstrator), in which the core of the reactor uses cooling in the liquid lead environment. This constitutes one of the arguments for the development of studies on innovative materials of generation IV, having the stated objective of the problem of the contact between the liquid lead and the structural materials specific to this type of reactor. The purpose of this paper is to highlight the changes in the thermomechanical behaviour induced by the contact between the 316L austenitic steel and the liquid lead, as well as its modelling through the equation of the Ramberg-Osgood-type mechanical stress-strain curve. The tensile tests in air and liquid lead were carried out at strain rates of the specimens in the range 10-3 s-1 ~ 10-5 s-1 and in a range of temperatures 350oC - 400 oC. To highlight the changes induced by the contact with the liquid lead on the thermomechanical behaviour of the 316L steel, the artificial neural network method, called the "Multilayer Feedforward Neural Network", was used in the processing of the experimental database. The obtained Ramberg-Osgood-type mechanical stress-strain curve is applied for both the air and the liquid lead environment at a temperature of 375oC and includes the following parameters as input: temperature, strain rate, yield stress, and maximum stress at necking. The two equations obtained for the air environment and the liquid lead environment at a temperature of 375oC were verified to the experimental data and a very good prediction agreement was obtained.