On the Stochastic Characterization and Prediction of Data-Driven Multi-Scale Materials Models for Composites

Abstract One of significant challenges in the metallic additive manufacturing (AM) is the presence of many sources of uncertainty that leads to variability in microstructure and properties of AM parts. Consequently, it is extremely challenging to repeat the manufacturing of a high-quality product in mass production. A trial-and-error approach usually needs to be employed to attain a product with high quality. To achieve a comprehensive uncertainty quantification (UQ) study of AM processes, we present a physics-informed data-driven modeling framework, in which multi-level data-driven surrogate models are constructed based on extensive computational data obtained by multi-scale multi-physical AM models. It starts with computationally inexpensive metamodels, followed by experimental calibration of as-built metamodels and then efficient UQ analysis of AM process. For illustration purpose, this study specifically uses the thermal level of AM process as an example, by choosing the temperature field and melt pool as quantity of interest. We have clearly showed the surrogate modeling in the presence of high-dimensional response (e.g. temperature field) during AM process, and illustrated the parameter calibration and model correction of an as-built surrogate model for reliable uncertainty quantification. The experimental calibration especially takes advantage of the high-quality AM benchmark data from National Institute of Standards and Technology (NIST). This study demonstrates the potential of the proposed data-driven UQ framework for efficiently investigating uncertainty propagation from process parameters to material microstructures, and then to macro-level mechanical properties through a combination of advanced AM multi-physics simulations, data-driven surrogate modeling and experimental calibration.

Download Full-text

On the use of the data- and physics-driven approaches for quasi-geostrophic double-gyre problem: application of Genetic Programming

10.5194/egusphere-egu21-8093 ◽

2021 ◽

Author(s):

Elnaz Naghibi ◽

Sergey Karabasov ◽

Vassili Toropov ◽

Vasily Gryazev

Keyword(s):

Genetic Programming ◽

Data Driven ◽

Time Signal ◽

Reduced Order ◽

Multi Scale ◽

Data Driven Approach ◽

Auto Correlation Function ◽

Gyre Circulation ◽

Proper Orthogonal ◽

Double Gyre

<p>In this study, we investigate Genetic Programming as a data-driven approach to reconstruct eddy-resolved simulations of the double-gyre problem. Stemming from Genetic Algorithms, Genetic Programming is a method of symbolic regression which can be used to extract temporal or spatial functionalities from simulation snapshots. &#160;The double-gyre circulation is simulated by a stratified quasi-geostrophic model which is solved using high-resolution CABARET scheme. The simulation results are compressed using proper orthogonal decomposition and the time variant coefficients of the reduced-order model are fed into a Genetic Programming code. Due to the multi-scale nature of double-gyre problem, we decompose the time signal into a meandering and a fluctuating component. We next explore the parameter space of objective functions in Genetic Programming to capture the two components separately. The data-driven predictions are cross-compared with original double-gyre signal in terms of statistical moments such as variance and auto-correlation function.</p><p>&#160;</p>

Download Full-text

Multi-scale proper orthogonal decomposition of complex fluid flows

Journal of Fluid Mechanics ◽

10.1017/jfm.2019.212 ◽

2019 ◽

Vol 870 ◽

pp. 988-1036 ◽

Cited By ~ 21

Author(s):

M. A. Mendez ◽

M. Balabane ◽

J.-M. Buchlin

Keyword(s):

Proper Orthogonal Decomposition ◽

Correlation Matrix ◽

Feature Detection ◽

Orthogonal Decomposition ◽

Data Driven ◽

Dynamic Mode ◽

Time Frequency ◽

Multi Scale ◽

Detection Capabilities ◽

Proper Orthogonal

Data-driven decompositions are becoming essential tools in fluid dynamics, allowing for tracking the evolution of coherent patterns in large datasets, and for constructing low-order models of complex phenomena. In this work, we analyse the main limits of two popular decompositions, namely the proper orthogonal decomposition (POD) and the dynamic mode decomposition (DMD), and we propose a novel decomposition which allows for enhanced feature detection capabilities. This novel decomposition is referred to as multi-scale proper orthogonal decomposition (mPOD) and combines multi-resolution analysis (MRA) with a standard POD. Using MRA, the mPOD splits the correlation matrix into the contribution of different scales, retaining non-overlapping portions of the correlation spectra; using the standard POD, the mPOD extracts the optimal basis from each scale. After introducing a matrix factorization framework for data-driven decompositions, the MRA is formulated via one- and two-dimensional filter banks for the dataset and the correlation matrix respectively. The validation of the mPOD, and a comparison with the discrete Fourier transform (DFT), DMD and POD are provided in three test cases. These include a synthetic test case, a numerical simulation of a nonlinear advection–diffusion problem and an experimental dataset obtained by the time-resolved particle image velocimetry (TR-PIV) of an impinging gas jet. For each of these examples, the decompositions are compared in terms of convergence, feature detection capabilities and time–frequency localization.

Download Full-text

Data-driven state detection algorithm for multi-scale systems and its application

Proceedings of the Institution of Mechanical Engineers Part C Journal of Mechanical Engineering Science ◽

10.1177/0954406213501590 ◽

2013 ◽

Vol 228 (7) ◽

pp. 1223-1234 ◽

Cited By ~ 1

Author(s):

Zeng Deliang ◽

Liu Jiwei ◽

Liu Jizhen

Keyword(s):

Failure Rate ◽

Detection Algorithm ◽

Data Driven ◽

Simulation Experiments ◽

Multi Scale ◽

State Detection ◽

Security And Reliability ◽

Roller Ring ◽

Technical Basis ◽

Over Time

To improve the security and reliability of equipment and reduce their failure rate, a data-driven state detection algorithm was proposed. The concepts of multi-scale system, multi-scale entropy and multi-scale exergy were defined. The algorithm is used for multi-scale systems whose state parameters change over time and have the characteristic of increasing monotonically on a dominant scale. An abrasion index for the middle speed roller ring mill was constructed, which was used to monitor the states of the instruments. Noise that affected the accuracy of the results was analyzed. The results of simulation experiments demonstrate the effectiveness of the algorithm, which can provide a technical basis for condition maintenance.

Download Full-text

Data-driven Urban Energy Simulation (DUE-S): A framework for integrating engineering simulation and machine learning methods in a multi-scale urban energy modeling workflow

Applied Energy ◽

10.1016/j.apenergy.2018.05.023 ◽

2018 ◽

Vol 225 ◽

pp. 1176-1189 ◽

Cited By ~ 34

Author(s):

Alex Nutkiewicz ◽

Zheng Yang ◽

Rishee K. Jain

Keyword(s):

Machine Learning ◽

Energy Simulation ◽

Energy Modeling ◽

Data Driven ◽

Learning Methods ◽

Multi Scale ◽

Machine Learning Methods ◽

Urban Energy

Download Full-text

Understanding COVID-19 nonlinear multi-scale dynamic spreading in Italy

Nonlinear Dynamics ◽

10.1007/s11071-020-05902-1 ◽

2020 ◽

Vol 101 (3) ◽

pp. 1583-1619

Author(s):

Giuseppe Quaranta ◽

Giovanni Formica ◽

J. Tenreiro Machado ◽

Walter Lacarbonara ◽

Sami F. Masri

Keyword(s):

Differential Evolution Algorithm ◽

Northern Region ◽

Multidimensional Analysis ◽

Data Driven ◽

Type Model ◽

Epidemic Spreading ◽

Epidemic Dynamics ◽

Multi Scale ◽

Evolution Algorithm ◽

Scale Data

Abstract The outbreak of COVID-19 in Italy took place in Lombardia, a densely populated and highly industrialized northern region, and spread across the northern and central part of Italy according to quite different temporal and spatial patterns. In this work, a multi-scale territorial analysis of the pandemic is carried out using various models and data-driven approaches. Specifically, a logistic regression is employed to capture the evolution of the total positive cases in each region and throughout Italy, and an enhanced version of a SIR-type model is tuned to fit the different territorial epidemic dynamics via a differential evolution algorithm. Hierarchical clustering and multidimensional analysis are further exploited to reveal the similarities/dissimilarities of the remarkably different geographical epidemic developments. The combination of parametric identifications and multi-scale data-driven analyses paves the way toward a closer understanding of the nonlinear, spatially nonuniform epidemic spreading in Italy.

Download Full-text