3-5 November 2021
Wirtschaftskammer Österreich
Europe/Vienna timezone

Powder bed fusion additive manufacturing - comparison of different process simulation methods

3 Nov 2021, 15:30


Oral Presentation AM Process- and Quality Control AM PROCESS and QUALITY CONTROL


Mr Etienne Bonnaud (SWERIM)


Additive manufacturing is an attractive and now well-accepted technique. Nevertheless, to reach its full potential, the method needs, among other issues, to cope with distortion and residual stress. High, local and quickly moving heat input leads to high temperature gradients that give rise to residual stresses during solidification and cooling. Due to residual stresses, components can crack already during manufacturing, distortions can develop when components are cut off their build plate and fatigue life can be shortened as residual stresses add up to external service stresses.

Numerical simulations based on the finite element method are a well-suited tool to predict residual stresses and resulting distortions. Simulations techniques for additive manufacturing have been derived from simulation techniques previously developed for welding applications. Nevertheless, the very small size and the intensity of the heat source require small elements and small time steps which leads to heavy simulations and long run-times. Drastic simplifications must therefore be made which unfortunately alter the precision of the results. Several methods of different complexity levels are currently available as shown in the following non-exhaustive list:

  1. mechanical global/macro models with classical layer lumping and dumping,
  2. thermo-mechanical global/macro models with classical layer lumping and dumping,
  3. thermo-mechanical local/meso models from which inherent strains are extracted and subsequently mapped on the full-size component of interest.
  4. thermo-mechanical global/macro models with use of an advanced additive manufacturing built-in plug-in in the commercial code Simulia Abaqus featuring use of exact heat source path and partial element activation,

In this study, these four models are run for the exact same component and process (geometry, material properties, heat source parameters, boundary conditions), see Fig.1, and relevant sensitivity studies are carried out. Distortion results are compared to one another and to actual measurements. As expected, the level of simplification has a strong influence on the reliability of the results.

Figure 1: Actual printed component and simulation model

Figure 1: Actual printed component and simulation model.

Speaker Country Sweden

Primary author

Mr Etienne Bonnaud (SWERIM)

Presentation Materials