6th World Congress on Integrated Computational Materials Engineering (ICME 2022): Tuesday Plenary
Program Organizers: William Joost; Kester Clarke, Los Alamos National Laboratory; Danielle Cote, Worcester Polytechnic Institute; Javier Llorca, IMDEA Materials Institute & Technical University of Madrid; Heather Murdoch, U.S. Army Research Laboratory; Satyam Sahay, John Deere; Michael Sangid, Purdue University

Tuesday 12:00 PM
April 26, 2022
Room: Regency Ballroom AB
Location: Hyatt Regency Lake Tahoe

Session Chair: Kester Clarke, Los Alamos National Laboratory


12:00 PM  Plenary
Virtual Testing of Structural Composites: A Multiscale Perspective: Carlos Gonzalez1; 1Imdea Materials Institute
    The increasing incorporation of lightweight, high-performance composites into modern structural designs represents a natural step in the optimization roadmap defined to satisfy the demanding needs and regulations of the transport industry nowadays. The outstanding specific mechanical properties of composites make possible stronger, faster, lighter, safer and greener vehicles. But the full, efficient exploitation of the advantages of composites still tackles challenging tasks, such as the monitoring of structural integrity or the reduction of development costs. In particular, a pursued aim of the engineering community is the eventual replacement of expensive and time-consuming test campaigns by accurate computational predictions of the mechanical behaviour and failure of composite materials. In this context, virtual testing constitutes a framework of advanced analysis and simulation techniques capable of bridging different length scales by transferring, from one level to the others, relevant information of the properties and failure mechanisms of and between the constituent materials.

12:30 PM Lunch

1:30 PM  Plenary
Hybrid Twin – Combining Physics- and Data-Based Models in a Consistent Digital Thread Spanning the AM Process Chain: Monica Salgueiro1; Marcos Diez1; Camilo Prieto1; Bernardo Freire2; Mihail Babcinschi2; Mustafa Megahed3; 1AIMEN; 2University of Coimbra; 3ESI Group
    Despite the generally high computational effort, physics-based models represent efficient tools enabling engineers to forecast material and product behavior. Integrated physics-based models (virtual twin) do however require assumptions to be made when the physics is too cumbersome to represent, is not understood or to account for unknown factors. Data-based models (digital twin) deduced from online monitors and material characterization on the other hand are very quick, capture unknowns and process deviations. The digital twin is however limited to the parameter range for which data is available. Combining the virtual and the digital twins provides researchers and engineers with a hybrid twin capturing the physics as well as the unknowns. This presentation discusses the implications and requirements for the creation of hybrid twins and demonstrates several aspects based on direct energy deposition additive manufacturing.

2:10 PM  Plenary
Computational Design of Lithium-ion Batteries Using Multi-Scale Models and Machine Learning: Kandler Smith1; 1National Renewable Energy Laboratory
    Battery design and discovery has traditionally followed a trial and error process. To accelerate this process, engineers and scientists have formalized physical descriptions of batteries into computational models. The U.S. Department of Energy’s Computer Aided Engineering of Batteries program, for example, supported software companies to scale well-known transport/reaction models at the Li-ion battery electrode length scale to solve large 3D battery design problems around packaging, heat and electron transport, performance, safety and crashworthiness of electric vehicles. Present research looks to deepen our knowledge at the sub-electrode or microstructure length scale to better understand chemo/mechanical interactions that govern degradation of today’s graphite/nickelate chemistries and operating performance of future Li/silicon/sulfur chemistries. Multi-scale high performance computing models and machine learning support this research. Machine learning algorithms interpret microscopy and electrochemical data to provide better descriptions of material mechanical evolution. Machine learning also identifies chemical side-reaction sequences from atomistic and molecular calculations.

2:50 PM Break

3:10 PM  Plenary
Accelerating the Broad Implementation of Verification and Validation in Computational Models of the Mechanics of Materials and Structures: A Science and Technology Accelerator Study: Michael Tonks1; George Spanos; 1TMS
    Computational models associated with the mechanics of materials and structures are increasingly utilized to guide engineering decision-making. However, despite the increasing sophistication of such models, they are rarely sufficiently tested to yield suitably accurate, quantitative results for which the level of uncertainty has been adequately quantified. In this regard, inadequate verification and validation (V&V) and uncertainty quantification (UQ) practices can mask large errors and/or lead to misinterpretation of model limits, which can in turn lead to production delays, costly redesign, catastrophic failure, and even loss of life. Thus, on behalf of the National Science Foundation, The Minerals, Metals & Materials Society has led a science and technology accelerator study on Accelerating the Broad Implementation of Verification & Validation in Computational Models of the Mechanics of Materials and Structures. This presentation will provide some key highlights from that study, with a particular focus on the section on recommended V&V/UQ practices.