Date of Award

Fall 2020

Degree Type


Degree Name

Master of Science (MS)


General Engineering

Committee Chair

Peter Lucon

First Advisor

Scott Coguill

Second Advisor

KV Sudhakar

Third Advisor

Jack Skinner


Metal additive manufacturing (AM) produces parts by addition as compared to subtraction of material. Selective laser melting (SLM) is an AM technique that prints objects layer-by-layer, selectively melting powders using a laser. The mechanical properties of SLM parts are affected by processing parameters and powder characteristics, both of which alter the molten melt pool flow field. Marangoni convection (M-flow) is a thermo-capillary mass transfer from a region of lower surface tension to a region of higher surface tension, referred to as a radially outward flow for SLM. However, in the presence of surface-active elements such as oxides and sulfides, the melt pool surface flow direction may shift from radially outward flow to radially inward flow (inverse M-flow). Balling and pores, the most common defects in SLM, have been correlated to the presence of inverse M-flow but the relation has yet to be quantified.

In order to quantify the surface flow using high-speed melt pool videos, numerical simulations and algorithms for melt pool flow were developed. The surface flow of a melt pool was simulated as binary images created using a MATLAB script. A particle tracking algorithm developed in MATLAB, using various functions from the MATLAB Image Processing Toolbox, was used to track the surface oxide particles in the simulated binary images. Various factors that may affect the particle tracking algorithm, such as 1) the melt pool size, 2) the image pixel size, 3) the size and the number of surface oxides, 4) the flow type, and 5) the particle velocity, were varied in the simulated binary images. Experiments were designed and results analyzed against the numerical error measured to generate a model using Design Expert 12, a statistical design analysis software package. Design Expert 12 was used to determine the significance of each factor within the model. The ANOVA results demonstrated that particle velocity and flow type have significant influence on the error of measured displacement. The error increased with decreasing displacement of the particles being tracked in the melt pool simulation. Melt pool algorithm error has been quantified and validated against simulated data, therefore it can be used to analyze actual melt pool data with known confidence.


A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in General Engineering