Reconstructing Synthetic Surface Topography Maps from an Experimental Measurement Using a Markov Random Field Graphical Network

Studying contact between engineering surfaces requires measuring their surface topography, which is time-consuming and requires the use of sophisticated equipment. Thus, to reduce the need for intricate measurements, researchers have implemented algorithms to numerically generate surface topography that, e.g., imitates the texture resulting from different manufacturing processes. However, such algorithms do not consider experimental parameters and, instead, create surface topography with specific, theoretical properties, which limits their applicability. In contrast, we implement a method based on a Markov Random Field graphical network to reconstruct synthetic surface topography that emulates any experimental surface topography measurement. We validate the method both statistically and physically using Inconel 718 specimens manufactured with laser-powder bed fusion, and we use several parameters to determine that the synthetic and experimental surface topography are similar. Furthermore, we investigate the effect of scaling and sampling on the accuracy of the synthetic surface topography. The knowledge resulting from this work enables reducing the experimental burden of performing surface topography measurements by replacing them with statistically and physically equivalent synthetic surface topography. Hence, this work finds application in studying, e.g., contacting surfaces, fatigue fracture, surface topography digital twins, and the relationship between manufacturing process parameters and surface topography.