Author(s)

M S A Oliveira & A C M Sousa

Abstract

Air-assisted water sprays or mists offer an excellent capability of controlling the rate of cooling. This is of particular relevance when dealing with heat treatment at very high temperatures, which is the case of nickel and titanium-based superalloys used in the aerospace and power industry. The paper provides a neural methodology for heat transfer analyses of data obtained experimentally. Its description will be given in detail, with particular emphasis on how a neural network can be trained to successfullypredict the resulting heat flux for specific input spray parameters. This particular knowledge can then be used to optimize theprocess, i.e. to establish the sprayconditions that would yieldthe coolingrate required to attain the specified mechanical properties, and to minimize the residual stresses. This methodology, aswill be advanced, can be integrated with deterministic multi-phase CFD models with the objective of reducing the number of time-consuming numerical computations to conduct, e.g. parametric analyses.

Keywords