Validating porosity results faces inherent challenges given the cost and effort required to obtain accurate and statistically representative experimental datasets. As a result, some deviation between simulated and measured data is to be expected. This is the case across all ranges of porosity, but particularly so at relatively low porosity cases (≦ ∼1%), given both experimental error as well as practical challenges of achieving sufficient simulation resolution without becoming computationally prohibitive. These challenges may be reduced from an experimental side by gathering a sufficient number of high-quality measurements to boost confidence in the magnitudes and trends expected.
It should be noted that there are various contributing factors to almost never achieving fully dense parts experimentally. These include the following:
Entrapped gases in the powder themselves can become trapped in the part as it is being built. This is powder-supplier-dependent and puts a lower bound on achievable porosity with their powder.
Outgassing by contaminants may occur during melt pool formation. This is dependent on end-user powder handling protocols and puts a lower bound on achievable porosity.
Vaporization of the powdered material will occur during scanning. There will always be at least a small amount of this gas vapor that becomes trapped in the melt pool during solidification, which appears as porosity. This is alloy dependent and puts a lower bound on achievable porosity.
Large plasma plumes at high energy density result in keyhole formation. The shape of keyholes make trapped vapor even more likely. As energy density increases, the amount of keyhole porosity will increase.
We do not simulate any of these phenomena. When our lack-of-fusion predicted % porosity levels become very low, these issues that we do not predict start to dominate and keep the actual part from becoming 100% dense even though the simulation indicates 100% density.
Finally, the validity of a Porosity simulation diverges more significantly from empirical data when porosity is greater than 30%. This can be caused by high variance in empirical data at high levels of porosity or from simulation resolution of melt pools in low energy density scenarios. We do not recommend you simulate with settings that will generate porosity greater than 30%.