ETD

Archivio digitale delle tesi discusse presso l'Università di Pisa

Tesi etd-04202020-123347


Tipo di tesi
Tesi di laurea magistrale
Autore
BENASSI, MATTEO
URN
etd-04202020-123347
Titolo
Analisi numerico-sperimentale delle strategie di modellazione per la simulazione del processo SLM
Dipartimento
INGEGNERIA CIVILE E INDUSTRIALE
Corso di studi
INGEGNERIA MECCANICA
Relatori
relatore Prof. Monelli, Bernardo Disma
relatore Moda, Mattia
tutor Palladino, Marco
Parole chiave
  • additive manufacturing
  • selective laser melting
  • laser powder bed fusion
  • distortion
  • residual stress
  • finite element modelling
  • inherent strains methods
  • thermo-structural simulation
  • experimental validation
  • Ansys
Data inizio appello
06/05/2020
Consultabilità
Non consultabile
Data di rilascio
06/05/2090
Riassunto
Selective Laser Melting (SLM) is an Additive Manufacturing (AM) process that allows the production and optimization of functional parts with a cost substantially independent of their geometric complexity. Despite his flexibility, SLM is affected by several issues, such as cracks, porosity and stress-driven distortions. The development of new components is currently based on expensive and time-consuming trial-and-error procedures. Therefore, the adoption of dedicated simulation strategies, aimed at reducing experimental iterations, could greatly improve the reliability and convenience of the SLM technology. Due to the multi-scale nature of the involved physical phenomena, the process modeling is not a straightforward task. Meso-scale models simulate the scanning process, but their high computational costs prevent their application on significant build volumes. On the other hand, macro-scale models allow the simulation of the entire process by introducing substantial simplifications of the thermo-structural problem.
This work is aimed at benchmarking the available macro-scale models based on their assumptions, inputs, outputs, flexibility, computational efficiency, number of built-in options, and open source level.
The three analyzed approaches are the Transient-Thermal (TT) and Inherent Strain (IS) methods implemented in Ansys Workbench, and the Nonlinear Track Superposition (NTS) method developed by the University of Pisa. The measured residual stresses and distortions were compared with those predicted by the above methods on several samples designed to highlight their differences. Residual stresses were measured through the hole drilling method on four cylindrical samples (25 mm diameter × 10 mm length), whilst distortions were measured with contact (CMM) and non-contact (structured 3D light scanning) techniques on two sets of thin-walled cylinders (30 mm outer diameter, 28 mm inner diameter and three different heights: 10, 20, 30 mm; totaling 20 samples), five different types of cantilever-shaped specimens (21 samples) and six arch-shaped samples.
Residual stress profile was accurately estimated by NTS and IS approaches. By contrast, about 10% average overestimation was found with TT method. Regarding the distortions prediction, IS provided the worse estimations than those obtained using both TT and NTS methods. For cantilevers-shaped specimens, NTS successfully estimated the deflection with a maximum deviation of 5%. IS demonstrated to collect error of 10-20% on deflection prediction for geometries even when are really close to the ones used for the calibration. TT demonstrated to have an estimating error of about 5-30% depending on the parts geometry.
Ultimately, we found that each simulation strategy can be useful depending on the needs. NTS is the most accurate and stable method for the prediction of distortions and also the only one taking into account the actual scanning strategy. However, being at an early development stage, it lacks some capabilities and user-friendliness compared with the other methods. IS is both fast and easy to use, it could be useful for first-order approximations of residual stresses and distortions, but it lacks accuracy on uncalibrated geometries. TT is the only method assessing the macro-scale thermal history, albeit significantly approximated, and including its effects on the residual stress field. Unfortunately, it is also the most demanding in terms of computational resources. Overall, NTS seems to be the most promising method for the prediction of process-induced residual stresses and distortions, while a simulation tool evaluating the part-scale thermal history could be complementary depending on the investigated issues.
File