https://dr.ntu.edu.sg/bitstream/10356/152944/2/qwaa094.pdf

Data-driven design strategy in fused filament fabrication : status and opportunities

A hydrological system involves a significant number of parameters to describe the real-world phenomena which ultimately reflects into the computational burden during model fitting and simulation. In this study, the proposed Adaptive Emulator Modelling based Optimization (AEMO) framework is presented which minimizes the complexity and computational burden, and enhances the overall efficiency of the representation of a physical watershed model. Here, AEMO consists of machine learning, sensitivity analysis, adaptive modelling, and sampling. The efficacy of the proposed AEMO framework is carried out through the Soil and Water Assessment Tool (SWAT) hydrological model for Peachtree Creek Watershed, Atlanta, USA. This watershed is subjected to flash floods in case of heavy rains due to a number of narrow streams situated in urban areas, and the water levels can rise quickly within a few hours or minutes of a rainfall event. Hence, this type of watershed is more complicated during the simulation and it is necessary to analyze the sensitivity during calibration of each iteration to provide information for the next iteration. The results concluded that the AEMO achieved exceptional results in both performance and computational burden when compared to the existing method. The NSE performance metric for the default SWAT model and hybrid model (SWAT + AEMO) showed the values 0.49 and 0.81, respectively. Hence, it displayed that AEMO enhanced the SWAT conceptual model prediction. The proposed framework incorporating future climate data can provide accurate information on water and disaster management.

Integration of Adaptive Emulators and Sensitivity Analysis for Enhancement of Complex Hydrological Models

Metal fused filament fabrication (MF3) combines fused filament fabrication and sintering processes to fabricate complex metal components. To design for MF3, an understanding of part geometry, printing parameters and material properties' effects on processability, part quality and ensuing properties is required. However, such investigation is a complex problem having several linked geometry, process and material variables to be considered that influence the process outcome. Moreover, such investigations through the experimental trial-and-error approach are costly and time-consuming, and sometimes not even feasible due to so many input variables involved. This study investigated the sensitivity of key output parameters toward each of the input parameters in MF3. FEA-based thermomechanical process simulations were used to estimate the process outcome in response to variable inputs, and a systematic procedure for sensitivity analysis has been successfully developed for the printing phase of the MF3 process. Dimensionless sensitivity values for all output parameters were calculated in the response of each input parameter, which allows parameters with different units to be compared quantitatively with a single yardstick. Moreover, three different part geometries were studied to identify how the process sensitivity varies with part geometry. For each output parameter, the most influential input parameters were identified from the whole set of input parameters and their influence trends were evaluated for different part geometries. The present sensitivity analysis procedure is expected to be an invaluable tool not only for process parameters optimization but also for the development of material and part geometry for MF3, hence enabling design for MF3 (DfMF3).

https://link.springer.com/content/pdf/10.1007/s11665-021-05666-8.pdf

Process Sensitivity and Significant Parameters Investigation in Metal Fused Filament Fabrication of Ti-6Al-4V

Designing components is a complex task, which depends on the component function, the raw material, and the production technology. In the case of rotating parts with higher RPM, the creep and orientation are essential material properties. The PLA components made with the material extrusion process are more resistant than VeroWhite (material jetting) and behave similarly to weakly cross-linked elastomers. Also, based on the tensile tests, Young’s modulus shows minimal anisotropy. Multilinear isotropic hardening and modified time hardening models are used to create the finite element model. Based on the measurements, the finite element method simulation was identified. The deformation in the compressor wheel during rotation became definable. It was concluded that the strain of the compressor wheel manufactured with material extrusion technology is not significant.

/pdf/finite-element-method-modeling-of-additive-manufactured-2zup8nyl71.pdf

Finite Element Method modeling of Additive Manufactured Compressor Wheel

A spiral casing is an important component of Francis hydro turbine for the even distribution of kinetic energy along the stay and guide vanes. The fluid flow around the runner is dependent on the flow condition of a spiral casing. The shape of the casing plays an important role in proper flow distribution. In this study, the optimization of the shape of a spiral casing is based on a steady-state flow analysis. Numerical optimization is performed using response surface methodology (RSM) and multiobjective genetic algorithm (MOGA). The flow uniformity and head loss in the spiral casing are selected as objectives for the optimal design of the spiral casing. The optimal design is selected from the solution acquired by RSM and MOGA. Moreover, the flow characteristics in the initial and optimal designs of the spiral casing are compared. The flow conditions in the optimal design improve significantly with the optimal design of the spiral casing. Thus, the inlet conditions for the stay vane are improved with the optimal design of the spiral casing.

Improvement of flow behavior in the spiral casing of Francis hydro turbine model by shape optimization

Surrogate Based Sensitivity Analysis of Part Strength due to Process Parameters in Fused Deposition Modelling

This paper aims at creating a computational finite element model, which will be used for prediction of deformation and failure of specimen under static loads for specimens prepared using FDM based 3D printers with different raster fill patterns and density. The work is divided as follows: a) understanding heterogeneity in specimen printed using FDM; b) conducting strength testing experiments to estimate stiffness and failure corresponding to particular infill configuration; c) creating and validating computational finite element model of FDM printed specimen with various raster fill pattern and density. In this work, the computational model of FDM parts is created as a multi-layered composite model. The computational model generated in this work can be used to estimate the deformation of a specimen printed using FDM process under a specific load condition. This model can further be used within an optimization framework to maximize part strength for a part with any geometry by suitably selecting part orientation and slicing parameters for a given service load condition.Copyright © 2016 by ASME

A Finite Element Model to Predict the Part Strength of Fused Deposition Modeling Printed Parts

Analysis of Total Head Loss in Various Configurations of Spiral Casing: A Numerical Study

Evolutionary Algorithms for Programming Pneumatic Sequential Circuit Controllers

Metastasis in the vertebral body is a widely observed malignant disease. This study used Computer Tomography (CT) images of the lumbar region to detect and classify the metastases from the normal vertebral bodies. We also classify the two types of metastases; Sclerotic and Lytic. Segmentation of the vertebral body is done using a deep neural network. We compared the performance of our model with existing state-of-art models. Multi-resolution blocks in the proposed model help segment the vertebral body with lytic lesions in the margin of the region of interest. Through Wavelet image transformation, coefficients are derived from the vertebral region. Based on the extracted wavelet scattering features, vertebral bodies are classified into three classes; normal, sclerotic and lytic, using Principal Component Analysis (PCA).

Saravana Kumar Gurunathan

Papers

Surrogate Based Sensitivity Analysis of Part Strength due to Process Parameters in Fused Deposition Modelling

A Finite Element Model to Predict the Part Strength of Fused Deposition Modeling Printed Parts

Analysis of Total Head Loss in Various Configurations of Spiral Casing: A Numerical Study

Evolutionary Algorithms for Programming Pneumatic Sequential Circuit Controllers

Deep Learning Based Lumbar Metastases Detection and Classification from Computer Tomography Images