Reliable reverse modeling of additive manufacturing by Tarasankar DebRoy - a keynote at RAAM2023

00:00 Introduction
02:49 Type of modeling
10:29 Reverse modeling
13:26 Improving reliability
17:59 Tailoring fusion zone geometry - a case study
23:01 Main contributions
Heat transfer and fluid flow models can provide valuable insights into additive manufacturing processes and materials such as the fusion zone geometry, temperature fields, and cooling rates, that cannot be easily obtained otherwise. However, their widespread adoption faces two main challenges. First, the computed results do not always agree with experimental results. This discrepancy is contributed by the uncertainty of some important parameters, such as energy absorption efficiency that cannot be precisely specified from theory. Second, the existing models typically operate in a one-way fashion where the models calculate temperature fields, cooling rates, and fusion zone geometry from the specified inputs of thermophysical properties and process variables. This one-way nature makes it impossible to calculate the additive manufacturing variables required to achieve a specified outcome of fusion zone geometry, temperatures, and cooling rates. Here we show that by combining a forward heat transfer and fluid flow model with an appropriate genetic algorithm, the uncertain model parameters can be estimated from a set of experimental data, improving the reliability of the models. Furthermore, the combination of a genetic algorithm and a heat transfer and fluid flow model makes the model bi-directional, thus improving the practical utility of the model. The bi-directional model can calculate verifiable multiple combinations of process parameters that can all provide a specified track geometry.