Date of Award
5-2024
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Mechanical Engineering
Committee Chair/Advisor
Garrett J Pataky
Committee Member
Marian Kennedy
Committee Member
Enrique Martinez Saez
Committee Member
Huijuan Zhao
Abstract
The expansion of the design space due to additive manufacturing (AM) has been a large motivator for the success of this family of processes. Despite the complexity of the physics in metal laser powder bed fusion AM causing significant stresses and strains in finished parts, the design advantages and subsequent performance improvements continue to drive the expansion of AM. Because the trial-and-error approach to AM part development is cost prohibitive, simulation of prints has become crucial. However, full thermo-mechanical simulation is susceptible to the same pitfall of time and computational cost in order to attain part-scale results. The development of lightweight AM print simulations that balance computational cost while retaining an acceptable level of accuracy is critical to the continued implementation and improvement of the AM process. In order to develop faster simulations, simplifications to and assumptions of the system must be made. In the context of simulating the mechanical behavior of printed AM parts, the work presented in this dissertation found that the part response can be captured in a pair of vectors that discretize the core and shell of a part and the in-plane and build-direction environments. This is implemented in the inherent strain method, where an initial small thermo-mechanical print simulation is performed on a millimeter-scale representative volume. From this, the vectors of inelastic strain are extracted using a novel method and subsequently applied to a part-scale mechanical simulation to yield the part’s critical post-print strain distribution. The parts were simulated as Inconel 625 on a 304 stainless steel iii build plate and validated against experimental measurements provided by the National Institute of Standards and Technology. Compared to the traditional approach of using only a single inherent strain vector to capture the full thermo-mechanical behavior, the piecewise vector set developed in this work significantly improved the median error from 82% down to 17%, with individual measurement location errors as low as 1%. The inclusion of the separate experimentally shown compressive core and tensile shell due to uneven cooling during the AM process, as well as the recognition of the different mechanical contexts of the in-plane and out-of-plane directions more accurately captured print behavior. This dissertation defines the inherent strain extraction technique and application, while improving the part-scale AM simulation cost-accuracy tradeoff balance.
Recommended Citation
Morand, Lucas M., "Novel Vector Assignment Approach for Inherent Strain Modeling of Laser Powder Bed Fusion Manufacturing" (2024). All Dissertations. 3571.
https://tigerprints.clemson.edu/all_dissertations/3571
Author ORCID Identifier
0000-0002-3330-5500