skip to main content


Title: Multi-Objective Process Optimization of Additive Manufacturing: A Case Study on Geometry Accuracy Optimization
Despite recent research efforts improving Additive Manufacturing (AM) systems, quality and reliability of AM built products remains as a challenge. There is a critical need to achieve process parameters optimizing multiple mechanical properties or geometry accuracy measures simultaneously. The challenge is that the optimal value of various objectives may not be achieved concurrently. Most of the existing studies aimed to obtain the optimal process parameters for each objective individually, resulting in duplicate experiments and high costs. In this study we investigated multiple geometry accuracy measures of parts fabricated by Fused Filament Fabrication (FFF) system. An integrated framework for systematically designing experiments is proposed to achieve multiple sets of FFF process parameters resulting in optimal geometry integrity. The proposed method is validated using a real world case study. The results show that optimal properties are achieved in a more efficient manner compared with existing methods.  more » « less
Award ID(s):
1657195
NSF-PAR ID:
10023972
Author(s) / Creator(s):
; ; ; ;
Date Published:
Journal Name:
Annual International Solid Freeform Fabrication Symposium
Page Range / eLocation ID:
656-669
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Despite recent research efforts improving Additive Manufacturing (AM) systems, quality and reliability of AM built products remains as a challenge. There is a critical need to achieve process parameters optimizing multiple mechanical properties or geometry accuracy measures simultaneously. The challenge is that the optimal value of various objectives may not be achieved concurrently. Most of the existing studies aimed to obtain the optimal process parameters for each objective individually, resulting in duplicate experiments and high costs. In this study we investigated multiple geometry accuracy measures of parts fabricated by Fused Filament Fabrication (FFF) system. An integrated framework for systematically designing experiments is proposed to achieve multiple sets of FFF process parameters resulting in optimal geometry integrity. The proposed method is validated using a real world case study. The results show that optimal properties are achieved in a more efficient manner compared with existing methods. 
    more » « less
  2. In Machine learning (ML) and deep learning (DL), hyperparameter tuning is the process of selecting the combination of optimal hyperparameters that give the best performance. Thus, the behavior of some machine learning (ML) and deep learning (DL) algorithms largely depend on their hyperparameters. While there has been a rapid growth in the application of machine learning (ML) and deep learning (DL) algorithms to Additive manufacturing (AM) techniques, little to no attention has been paid to carefully selecting and optimizing the hyperparameters of these algorithms in order to investigate their influence and achieve the best possible model performance. In this work, we demonstrate the effect of a grid search hyperparameter tuning technique on a Multilayer perceptron (MLP) model using datasets obtained from a Fused Filament Fabrication (FFF) AM process. The FFF dataset was extracted from the MakerBot MethodX 3D printer using internet of things (IoT) sensors. Three (3) hyperparameters were considered – the number of neurons in the hidden layer, learning rate, and the number of epochs. In addition, two different train-to-test ratios were considered to investigate their effects on the AM process data. The dataset consisted of five (5) dominant input parameters which include layer thickness, build orientation, extrusion temperature, building temperature, and print speed and three (3) output parameters: dimension accuracy, porosity, and tensile strength. RMSE, and the computational time, CT, were both selected as the hyperparameter performance metrics. The experimental results reveal the optimal configuration of hyperparameters that contributed to the best performance of the MLP model. 
    more » « less
  3. Low precision deep neural network (DNN) training is one of the most effective techniques for boosting DNNs’ training efficiency, as it trims down the training cost from the finest bit level. While existing works mostly fix the model precision during the whole training process, a few pioneering works have shown that dynamic precision schedules help NNs converge to a better accuracy while leading to a lower training cost than their static precision training counterparts. However, existing dynamic low precision training methods rely on manually designed precision schedules to achieve advantageous efficiency and accuracy trade-offs, limiting their more comprehensive practical applications and achievable performance. To this end, we propose LDP, a Learnable Dynamic Precision DNN training framework that can automatically learn a temporally and spatially dynamic precision schedule during training towards optimal accuracy and efficiency trade-offs. It is worth noting that LDP-trained DNNs are by nature efficient during inference. Further more, we visualize the resulting temporal and spatial precision schedule and distribution of LDP trained DNNs on different tasks to better understand the corresponding DNNs’ characteristics at different training stages and DNN layers both during and after training, drawing insights for promoting further innovations. Extensive experiments and ablation studies (seven networks, five datasets, and three tasks) show that the proposed LDP consistently outperforms state-of-the-art (SOTA) low precision DNN training techniques in terms of training efficiency and achieved accuracy trade-offs. For example, in addition to having the advantage of being automated, our LDP achieves a 0.31% higher accuracy with a 39.1% lower computational cost when training ResNet-20 on CIFAR-10 as compared with the best SOTA method. 
    more » « less
  4. Abstract

    Additive manufacturing (AM) has emerged as a promising approach to achieve energetic materials (EMs) with intricate geometries and controlled microstructures, which are crucial for safety and performance optimization. However, current AM methods still face limitations such as limited densities and inadequate solids loading. To overcome these limitations, we have developed a pressure‐assisted binder jet (PBJ) process that has the potential to allow for the fabrication of intricate EMs while preserving their desired properties. This study aims to investigate the effects of printing parameters on the microstructures and properties of EMs, including density, solids loading, mechanical properties, and heterogeneity. Our results demonstrate that the PBJ process achieves exceptional properties in EMs, including densities up to 83.4 % and solids loading up to 95.4 %, surpassing those achieved by existing AM processes. Furthermore, the mechanical properties of the fabricated EMs are comparable to those achieved using conventional fabrication techniques, including a compressive strength of 3.32 MPa, a Young's modulus of 16.68 MPa, a Poisson's ratio of 0.45, a shear modulus of 5.73 MPa, and a bulk modulus of 21.01 GPa. Various test cases were printed to showcase the ability of the PBJ process to create EMs with complex structures and exceptional properties. Micro‐computed tomography was employed to analyze the influence of printing parameters on the internal composition and microstructures of the printed specimens.

     
    more » « less
  5. null (Ed.)
    Purpose Mechanical anisotropy associated with material extrusion additive manufacturing (AM) complicates the design of complex structures. This study aims to focus on investigating the effects of design choices offered by material extrusion AM – namely, the choice of infill pattern – on the structural performance and optimality of a given optimized topology. Elucidation of these effects provides evidence that using design tools that incorporate anisotropic behavior is necessary for designing truly optimal structures for manufacturing via AM. Design/methodology/approach A benchmark topology optimization (TO) problem was solved for compliance minimization of a thick beam in three-point bending and the resulting geometry was printed using fused filament fabrication. The optimized geometry was printed using a variety of infill patterns and the strength, stiffness and failure behavior were analyzed and compared. The bending tests were accompanied by corresponding elastic finite element analyzes (FEA) in ABAQUS. The FEA used the material properties obtained during tensile and shear testing to define orthotropic composite plies and simulate individual printed layers in the physical specimens. Findings Experiments showed that stiffness varied by as much as 22% and failure load varied by as much as 426% between structures printed with different infill patterns. The observed failure modes were also highly dependent on infill patterns with failure propagating along with printed interfaces for all infill patterns that were consistent between layers. Elastic FEA using orthotropic composite plies was found to accurately predict the stiffness of printed structures, but a simple maximum stress failure criterion was not sufficient to predict strength. Despite this, FE stress contours proved beneficial in identifying the locations of failure in printed structures. Originality/value This study quantifies the effects of infill patterns in printed structures using a classic TO geometry. The results presented to establish a benchmark that can be used to guide the development of emerging manufacturing-oriented TO protocols that incorporate directionally-dependent, process-specific material properties. 
    more » « less