CIMNE Seminar: "High Fidelity Digital Twins: Detecting Weaknesses in Structures", by Rainald Lohner

Digital Twins have appeared in nearly every field: civil, mechanical and aerospace engineering, medicine, social media, etc. The term `digital twin' can mean different things in different realms - even if we consider specialized field like civil engineering. The data stored in the digital twin can encompass vastly different information. In some instances a BIM enumeration of parts (e.g. number of buildings in a neighbourhood) is already considered by some a digital twin. This data can be enriched by accounting for the parts (number, type, dimensions, manufacturer, ... of: walls, doors, flooring, piping, HVAC systems, lighting, kitchen, etc.). And enriched further by the CAD data for visualization or product life cycle management. And enriched further by the detailed or approximate (ROM, POD, ML) computational models and sensors that monitor aspects of the built environment throughout its life. These `information rich' digital twins have emerged as a result of three megatrends:

a) The pervasive use of CAD systems,

b) The widespread availability and use of computational tools to `pre-compute, only then build' and `pre-compute, then operate'; and

c) The emergence of precise, rugged, connected and cheap cameras and sensors that may be used to monitor structures.

An adjoint-based procedure has been developed to determine weaknesses, or, more generally, the material properties of structures. The cost function is based on the difference between measured and computed results. The procedure assumes the existence of high-fidelity finite element models for the structure. This leads to an inverse problem of high dimensionality which is non-convex and non-smooth, requiring proper regularization. The adjoint-based procedure has been extended to consider the optimal placement of sensors, an optimal choice of standard loads, and techniques to 'zoom in' on weakened regions.

Several examples show the viability, accuracy and efficiency of the proposed methodology and its potential use for high fidelity digital twins.

Més posts de Formació