Additive Manufacturing (AM) is an advanced manufacturing technology used to manufacture custom and geometrically complex parts using a layer-by-layer material addition process. Variations in the AM process lead to ‘deviations’ in the manufactured part, resulting in geometric non-conformance. To date, the geometric conformance or tolerance quantification for AM relies on two major methods:​ experimental methods based on geometric benchmark test artifacts (GBTA) and predictive methods such as finite element analysis (FEA). A common limitation in these methods is non-compliance with ISO 1101 standard [ISO 1101, 2017], i.e., usage of GD&T (geometric dimensioning and tolerancing). GD&T enables complete geometric quantification of form, orientation, and location of any mechanical part and should be used for AM parts to ensure geometric conformance. Experimental methods lack GD&T quantification due to GBTA design issues. There are limitations in design guidelines in terms of geometric conformance and linkage of features to GD&T is missing. Designing GBTAs without considering these factors leads to unnecessary experimentation and partial GD&T characterization. On the other hand, predictive methods lead only to geometric deviation regions and/or residual stresses on the part but do not estimate the GD&T characteristics. Due to these limitations, there are process specific gaps which need to be addressed. Such as, the need for parametric optimization for GD&T and assemblability for fused filament fabrication (FFF) process;​ along with the consideration of bead geometry in modeling methodologies. In laser powder bed fusion (LPBF), there is a need for GD&T based data-sets considering the effect of removal of base plate. Further, there is no methodology present in literature that leads to GD&T predictions and assemblability information.

Based on these research gaps, this thesis aims at developing a framework to quantify and predict the geometric tolerances and assemblability of AM parts based on GD&T standards. A systematic GBTA design methodology is first proposed that links the features to GD&T and helps in assemblability investigation. Based on the new design methodology, new GBTAs are designed for conducting geometric quantification and prediction for different AM processes. For FFF, a new GBTA is designed to understand the effect of process parameters on GD&T and assemblability. The results suggest a direct dependence of GD&T and assemblability on the process parameters which were not studied before such as motor micro-stepping, component size, and material type. Further, a virtual geometric conformance investigation methodology was developed by converting the sliced file into a solid model. The resultant model called ‘the reverse CAD model’ is capable of performing accurate virtual geometric conformance investigation. This not only helps to reduce part rejection, but also helps in virtual design/parametric changes before manufacturing the part. For LPBF, a normative GBTA is designed for generic tri-planar GD&T quantification before and after removing the GBTA from the baseplate. The experimental GD&T results are compared with simulation results to understand the reliability of numerical simulations for GD&T prediction. The results provide a complete GD&T data-set for LPBF process, show a wide variation in GD&T results proving the need for GD&T based quantification, and provide quantified data about the usability of the simulation-based GD&T. Further, skin model shapes methodology is implemented for the first-time for LPBF to predict deviations leading to GD&T and assemblability estimation. To summarize, this thesis presents a framework for GD&T based assemblability investigation for AM using experimental and predictive methodologies.