Robust fatigue design by combining metrology and simulation

According to Industry 4.0 concepts, dimensions of every manufactured part are measured, and optical methods such as fringe projection and robotic 3D scanning with other optical sensors, offer promising solutions. A digital twin (DT) is formed from the measured geometrical data. The digital twin represents the actual physical part, created from optical scans and converted into an STL model.  

Using digital twin geometry in finite element analysis enables physical-instance-specific lifetime predictions, for example for weld misalignments that affect fatigue. For the fatigue life of welded structures, misalignments in the welds of individual parts are especially critical.

Traditional design and tolerancing methods often rely on nominal geometries and conservative assumptions, which can lead to over-engineering and increased costs. By using measured data and statistical models, engineers can evaluate the real structures and components produced in manufacturing. This makes it possible to quantify uncertainties and predict fatigue life, functional performance, and reliability more accurately.

One important benefit is the ability to define tolerances based on performance and reliability rather than on geometric limits from handbooks. Simulation models can link dimensional variation to the fatigue behaviour of welded structures, as well as to functional performance such as contact conditions and wear of machined surfaces. This allows tolerances to be optimised so that required reliability targets are met without unnecessary manufacturing effort. A key link to sustainability is the enabling of reliable lightweight structures. Lightweight design reduces raw material consumption, manufacturing energy use, and lifecycle emissions. 

VTT has excellent capabilities in creating a digital thread and in evaluating the reliability and quality of 3D data measured using optical instruments such as fringe projection and robotic laser scanning. The reliability of measured data is naturally critical to the success of the digital thread.

The workflow illustrated in the following figure connects design intent, measurement, simulation, and performance evaluation within a continuous digital thread. It begins with a CAD model containing defined critical tolerances. Measured dimensions from manufactured structures are characterized and used as inputs for parametric FEA, where key geometric features are varied. The resulting stress, fatigue-related performance metrics, and functional metrics, such as the stiffness of the structure between the attachment points, are collected and organized into a dataset that can be used to train AI-based surrogate models, enabling fast reliability assessment and supporting performance-based tolerancing and informed industrial decision-making.

The concepts described were developed in the research project “Trustworthy virtual experiments and digital twins” ViDiT. The project (22DIT01 ViDiT) has received funding from the European Partnership on Metrology, co-financed from the European Union’s Horizon Europe Research and Innovation Programme and by the Participating States.