In Direct Energy Deposition, manufacturers require a simple and economical technique to simulate a given powder flux. This study proposes a novel analytical method based on a straightforward measurement technique to reverse engineer the powder flow distribution of a coaxial annular nozzle up to its focus position. Different axial profiles of powder concentration were obtained by blowing powder over a coated plate at different working
distances. The observed marks were then discretised into a mixture of Gaussian density of probabilities. A predictive model correlating information between the powder flow profiles to reproduce the powder flow as a sum of continuous statistical functions, each associated to an individual trajectory was developed. This method is assessed on a random nozzle to prove its reliability. The model was trained using data from seven low working distance profiles. It was then tested on two horizontal planes closer to the powder focus and one tilted plane taken at a low working distance. The model shows good predictability with an average R2 of 0.874 for the profiles of the plane nearest to the powder focus, which is very similar to the training phase R2 Regarding the tilted plane, the model was able to predict expected variations on the dispersion and the centroids’ positions, which demonstrates the validity of an application to complex surfaces.