Quantifying Uncertainty in the Mechanical Performance of Additively Manufactured Parts Due to Material and Process Variation

Additive manufacturing is an extremely customizable process; however, variations in the chosen build parameters can lead to drastic differences in part performance. The performance variation due to process parameters is still not well understood, and propagating all uncertainties from the various sources has been a challenge. Sources of AM parts’ performance variability include uncertainties in feedstock, machine process parameters, material response during melting/solidification, process modeling, performance prediction, and post heat treatment.. Utilizing an ICME platform, MRL has developed a software to propagate uncertainty throughout various stages in the building cycle which was demonstrated and verified on quantifying uncertainty in distortion/residual stress and location-specific strength predictions in Ti6Al4V. The development of the software to include predictions of uncertainty in dynamic performance enhances its commercial use for DoD and non-DoD applications. . The flexibility of the software is developed in the ability to train the software on various machines and various materials. The deployment of the software using on-premises and cloud-based computational models drastically broadens the possible use of the software among various members of the AM supply chain.