Semi-Plenary Lecture

Julian Allwood (UK): (Bio)

AUTOMATED DESIGN OF FOLDING-SHEARING PROCESSES

The Folding-shearing process was invented to improve the material efficiency of deep drawing, by reducing the need for trimming, and is gaining commercial traction. The simplest application of the process converts a flat blank into a shrink corner, and if the product geometry is represented by flat planes connected straight line creases, this shrink corner is a single vertex. In recent studies, we have recognised that this vertex can be created by many different combinations of folding and shearing, for example with the first fold having different orientations, or with various geometries for the shearing stage. More complex part geometries can have several vertices, thus greatly increasing the number of process combinations that could convert an initial flat blank into the same final part. Here we describe an optimisation approach, in which all the options for process design are parameterised, and then explored using a dynamic programming algorithm. The algorithm can minimise total work done, or can target efficient blank shapes, to improve tessellation in blanking and hence material utilisation. The new approach is used to examine the options for forming a stylised B-pillar component.

Katia Mocellin (France): (Bio)

NUMERICAL ANALYSIS OF INCREMENTAL FORMING PROCESSES – FLOW FORMING OF TUBES AND PLATES

The production of metal products often requires the use of large, complex and costly tooling systems. These systems also rely on high-tonnage presses to achieve the desired plastic deformation.

Here, we will focus on less conventional processes that enable the precise and repeatable production of thin-walled parts (tubes or sheets) with complex geometries that cannot be achieved by deep drawing or without intermediate welding. These processes use relatively simple tooling and reduced force levels.

Incremental processes, such as flow forming, rely on the localized, progressive and incremental deformation of the material. This succession of small, localized plastic deformations allows for the eventual accumulation of large deformations in the material. The manufacturing process may also involve several passes and intermediate heat treatment to allow reach final geometries.

In this presentation, we will address the issue of modelling this type of process and understanding the material flow path. We will discuss which material characterization methods to use for the expected strain levels and strain rates. Additionally, we will analyse the impact of the manufacturing route and the formalism of the constitutive and damage law on the prediction of geometries and defects.

Dirk Mohr (Switzerland): (Bio)

MACHINE LEARNING MEETS HIGH-THROUGHPUT TESTING: ADVANCING SHEET METAL PLASTICITY AND FAILURE

After examining the limitations of conventional RNN architectures, this lecture introduces mechanics-informed recurrent neural networks specifically designed for constitutive modeling and discusses their implementation in explicit finite element solvers. The results demonstrate that mechanics-informed RNNs can serve as highly versatile material models, providing accurate surrogate representations of complex crystal-plasticity behavior. Compared with conventional crystal-plasticity fast Fourier transform simulations, the resulting CP-RNN models achieve substantial computational speed-ups, making the application of crystal plasticity to industrial-scale simulations increasingly practical.

Beyond neural-network architecture development, the lecture presents novel robot-assisted, high-throughput experimental methods for generating the large datasets required to identify machine-learning-based models of plasticity and failure. In particular, we demonstrate the first complete identification of a data-driven constitutive model using experimental data obtained from a random-walk multiaxial testing system. Together, these developments illustrate how mechanics-informed machine learning and automated experimentation can accelerate material characterization and enable more accurate and efficient predictions of sheet-metal deformation and failure.

Brad Kinsey (USA): (Bio)

POST-TRIMMING SPRINGBACK CONTROL IN DOUBLE-SIDED INCREMENTAL FORMING THROUGH TOOLPATH MANIPULATIONS

Double-sided incremental forming (DSIF) is a die-less sheet forming process in which a part is formed using two tools acting on opposite sides of the sheet along a toolpath. DSIF can be implemented using dedicated CNC-based machines or industrial robotic platforms, making it attractive for flexible, low-volume production. However, its geometric accuracy is often limited by springback, which occurs during tool removal, unclamping, and particularly trimming, due to the release of residual stresses. In this work, a geometry containing flat and convex walls is formed from an AA5052-H32 sheet and partially trimmed along its convex wall. Geometric deviations relative to the target geometry are experimentally measured at each springback stage, revealing post-trimming deviations of up to ~17 mm. A finite element model incorporating the Hill48 anisotropic model is developed to study stress evolution during forming and through-thickness residual stress distributions at different springback stages. The simulations show that opposing tensile and compressive residual stresses across the sheet thickness drive the observed geometric deviations. Additional experiments using symmetric geometries with flat, concave, and convex walls demonstrate that surface curvature strongly affects post-trimming deviation, with convex and concave walls producing opposite trim-edge movements. Based on these findings, a two-pass reverse-forming toolpath is used, in which an intermediate geometry is first formed to introduce compensating residual stresses between passes. This approach reduces the through-thickness stress gradients, leading to ~50% improvement in post-trimming geometric accuracy. The proposed methodology is applicable to both dedicated and robotic DSIF systems, as it relies solely on toolpath control.

Wolfram Volk (Germany): (Bio)

DEGRADATION AND RECOVERY OF ELASTIC MODULUS IN AISI 304 UNDER CYCLIC TENSILE LOADING AND PRE-STRAIN-INDUCED ANISOTROPY

The present study investigates the degradation of the elastic modulus of AISI 304 stainless steel under uniaxial loading at room temperature. The investigation focuses on the influence of plastic deformation and the anisotropy caused by the deformation. Cyclic uniaxial tensile tests were conducted in order to quantify the change in elastic stiffness as a function of accumulated plastic deformation. In order to further investigate the directional dependence, the specimens were uniaxially pre-stretched to a plastic deformation of 8% and then further examined by cyclic tensile tests along three different rolling directions (0°, 45°, and 90° relative to the rolling direction and pre-stretch direction). The findings demonstrate a substantial decrease in the elastic modulus in the pre-deformation direction (0°), while the 45° and 90° directions exhibit considerably smaller reductions, suggesting deformation-induced anisotropy of the elastic response. Furthermore, the impact of heat treatment on the elastic properties was examined. The findings indicate that appropriate heat treatment can result in a partial restoration of the elastic modulus. The elastic modulus was determined using two different approaches. Firstly, the conventional mechanical evaluation method was employed, based on the slope and R_p0.2. Secondly, a temperature-based approach was used, derived from the thermoelastic effect.

Stefania Bruschi (Italy): (Bio)

HOT DEFORMATION BEHAVIOR AND IN-SITU HEAT TREATMENT EFFECTS IN LASER POWDER BED FUSION TI6AL4V

This research investigates the hot deformation behaviour and microstructural evolution of Ti6Al4V samples produced via Laser Powder Bed Fusion (LPBF) and compressed at varying temperatures and strain rates. Two different LPBF Ti6Al4V conditions are considered, namely the as-built and heat-treated states. Utilizing an Arrhenius-type constitutive equation, peak flow stress was found to decrease significantly with temperature and increase with strain rate. However, as-built samples exhibit higher peak stress levels, particularly at lower testing temperatures, due to their initial metastable martensitic microstructure.

Microstructural analysis at the highest testing temperature showed that as-built samples undergo complete globularization and dynamic recrystallization, resulting in finer, equiaxed grains compared to heat-treated counterparts in the central regions of the samples characterized by higher levels of strain. Conversely, peripheral “dead metal zones” showed a microstructural convergence to a fine lamellar/Widmanstätten structure, indicating an “in-situ” heat treatment during the pre-heating stage.

Marion Merklein (Germany): (Bio)

TAILORED MATERIAL FOR SUSTAINABLE, ROBUST FORMING PROCESSES

Sustainability represents one of the key challenges for future manufacturing. The production of materials, including steels, aluminium alloys and copper, involves high-energy processes, while variations in material properties can adversely affect both component quality and forming process stability. Considering these aspects, manufacturing technologies must be assessed with regard to their potential to reduce energy consumption, improve material utilisation and increase process robustness. Promising approaches include lightweight design and the targeted tailoring of local material properties. Tailored materials can be produced by orbital forming, tailored heat treatment processes or local microstructural modification. In each case, the objective is to adapt the material response to the local requirements of both the forming operation and the final component application, thereby controlling material flow, reducing critical strain localization and extending the feasible process window. Particular attention is given to the technological implementation of these approaches and to the interactions between processes, materials, and energy consumption..

Hiroshi Utsunomiya (Japan): (Bio)

PLASTIC INSTABILITY PHENOMENON OF THREE-LAYER SANDWICH SHEET IN COLD ROLLING

Clad sheets are multi-materials having stacked structure of dissimilar metal layers. They are mostly produced by roll bonding process where stacked layers are reduced in thickness and simultaneously bonded. When an already bonded three-layer sheet is further rolled heavily, either of two periodic modes of plastic instability appears. One mode is the symmetric necking, while the other mode is the asymmetric necking, with respect to the rolling direction. They are often called ‘sausaging’ and ‘undulation’ modes according to the morphology of the inner layer. However, differences in formation mechanism and criteria between the two modes have not been understood. In this study, bifurcation conditions of the two modes were deduced from literature survey. Then, finite element analyses were conducted to reproduce the two modes. It was found that either mode occurs under conditions with large difference in flow stress, low work hardening rate, high reduction in thickness. It was also found that the mode depends on combination of stacked materials, sausaging mode appears with hard internal layer, while undulation mode appears with soft internal layer. A unified approach was proposed to predict the mode of plastic instablity in cold rolling.

Jianguo Lin (HK): (Bio)

GETTING BETTER MATERIAL DATA FOR HOT STAMPING APPLICATIONS

For industrial applications of innovative metal forming techniques, such as hot stamping of ultra-high strength steels and high-strength aluminium alloys, advanced experimental methods have been developed to characterize mechanical properties of materials under the corresponding forming conditions. This research focuses on the measurements of strain fields using DIC technologies and how to obtain more accurate material data using deep machine learning from previous data and errors for uniaxial and biaxial tensile tests. These testing methods have been improved significantly in recent years, including the development of AI-based material testing system and the results of application are introduced.

Kaan Inal (Canada): (Bio)

MACHINE LEARNING-BASED FRAMEWORKS FOR MICROSTRUCTURE ENGINEERING TO ENHANCE FORMABILITY

In modern manufacturing, components are designed primarily in the virtual space, allowing manufacturers to evaluate far more design cycles and accelerate the path from initial concept to production-ready components. In practice, however, this virtual design process is constrained by the substantial time and computational resources required to generate the large number of high-fidelity simulations that vehicle development demands. Metal forming simulations, such as those used to model sheet metal stamping, are particularly expensive because they must capture the complex, history-dependent material behavior that governs the final formed part. Recent advances in artificial intelligence (AI) offer a compelling opportunity to dramatically accelerate the generation of such simulation results. This work presents, for the first time, an advanced recurrent deep learning framework capable of predicting the entire simulated stamping response for previously unseen geometries, materials, and process parameters. By learning the underlying temporal evolution of the forming process, the framework reproduces the full stamping response rather than isolated outputs. The predictions are benchmarked against high-fidelity finite element models and demonstrate a significant reduction in prediction time while maintaining high accuracy in the key performance indicators of the stamping process. Next, the so-called U-PolyConformer, a spatiotemporal machine learning framework that combines U-Net convolutional neural networks with transformer layers, capable of capturing the full-field evolution of stress and strain under monotonic and random-walk loading conditions for aluminum alloys, will be presented. U-PolyConformer achieves a 7,900x speed-up over the ground-truth CPFEM simulations while producing high-fidelity results in both interpolative and extrapolative regimes. Comprehensive evaluations are presented for both models to demonstrate their capacity to generalize beyond the training distribution to novel microstructures, loading conditions, and strain-hardening behaviours. Finally, to highlight the potential of the proposed frameworks for microstructure optimization and to accelerate computational materials engineering workflows, a microstructure optimization framework based on static recrystallization is developed and used to improve forming limit strains for various strain paths.

Glenn S. Daehn (USA): (Bio)

THE AGILITY FORGE FOR POINT-OF-NEED PRODUCTION AND HISTORY-DEPENDENT MICROSTRUCTURAL CONTROL

The Agility Forge is an autonomous platform for incremental open-die forging, designed for the fabrication of high-mix structural metallic components. By utilizing numerically controlled press motion and induction heating, the system reduces design-to-product cycle times from months to a single week and has demonstrated large cost reductions compared to conventional rough machining of aerospace alloys. This technical capability enhances supply chain resilience by enabling the point-of-need production of high-value, low-volume components. Beyond manufacturing, the platform serves as a characterization tool to investigate history-dependent material behavior. It imposes controlled gradients in strain, strain rate, and temperature to enable the voxel-by-voxel mapping of flow stress and microstructural evolution. These experimental data inform an AI-integrated digital twin framework, which utilizes GPU-accelerated differentiable simulations and neural network surrogates for real-time process monitoring and path optimization. A primary challenge addressed is the assurance and optimization of component quality. Through an integrated “Design-Make-Measure-Learn” workflow, the platform synchronizes in-situ metrology with predictive materials models to facilitate Model-Based Qualification. This approach allows that autonomously forged parts achieve handbook-standard properties or allowing for the local engineering of microstructural states for performance optimization.

Lin Hua (China): (Bio)

INTELLIGENT ROLLING THEORY AND TECHNOLOGY FOR EXTREME RING COMPONENTS

Ring components are critical parts responsible for load-bearing and motion transmission, widely utilized in transportation and energy equipment such as automobiles, high-speed railways, aircraft, rockets, wind power generations, and nuclear power plants. Ultra-large ring components with diameters exceeding 2 meters not only possess extreme dimensions but also require ultra-fine microstructures; traditional split-welding manufacturing methods cannot satisfy the demands for such extreme microstructures and properties, creating an urgent need to develop integrated plastic forming technologies for ultra-large ring components with extreme dimensions and performance characteristics. Through in-depth theoretical analysis and experimental research, this study established a physical-mechanical model for the radial-axis rolling process of ring components, elucidated the mechanisms governing rolling motion stability and microstructure evolution, determined the conditions for progressive deformation and overall forming of ring components during rolling, and constructed a comprehensive theoretical framework for radial-axis rolling of ring components; it proposed methods for ensuring rolling motion stability, achieving deformation coordination, and regulating microstructure uniformity for ultra-large ring components, and invented a thermal mechanical motion coordinated shape-and-property control rolling technology; furthermore, it developed a digital twin-based intelligent process design, process measurement and control system for the radial-axis rolling of ultra-large ring components, manufactured rolling equipment and automated rolling production lines capable of processing 21-meter ultra-large ring components, and realized networked collaborative intelligent rolling production for these extreme-scale components. These research achievements have been extensively applied in the manufacturing of ultra-large ring components, including aero-engine casings, launch vehicle cabin sections, wind turbine bearings and towers, and nuclear power plant pressure vessels, thereby supporting the innovative development of major transportation and energy equipment systems.

Mingwang Fu (HK): (Bio)

DAMAGE AND FRACTURE IN THE DEFORMATION OF MATERIALS AND DEFORMATION-BASED MANUFACTURING

Deformation of materials and deformation-based manufacturing are important engineering practices and efficient manufacturing processes. The latter, in particular, is widely used to fabricate net-shape or near-net-shape parts via plastic deformation. In this process, the design of deformed parts, forming processes, tooling, defect prediction and avoidance, and product quality assurance and control must all account for damage and fracture in the working materials. Scientific insights into the formation and occurrence of damage and fracture, and into their mechanisms and behaviors, are crucial for deformation-based forming processes. In this talk, the focus will be on the mechanisms of void initiation, coalescence, and growth during material deformation; the formation and occurrence of damage and fracture; and their mechanisms, behaviors, and prediction and avoidance through experiments and simulations, from the perspective of the state of the art in damage and fracture research.

Junhe Lian (Germany): (Bio)

ADVANCING FORMING TECHNOLOGIES THROUGH MATERIAL-AWARE DIGITAL TWINS

Material and process are inseparable in forming. The state of materials (microstructure and properties) is not fixed during processes; it evolves with each forming step, and every forming operation is itself an act of material design, shaping not just geometry but also material state. For most of the history of our field, this second role has remained passive, with states inherited from the forming history rather than designed. This talk presents a pathway toward proactive engineering, along which material evolution is characterized by experiments, predicted by modeling, and finally designed and controlled through the process chain by artificial intelligence. Multiscale experiments generate the physical evidence, from macroscopic forming tests to in situ synchrotron diffraction. Physics-based modeling captures the underlying science through crystal plasticity and microstructure-informed models of anisotropy, damage and fracture, validated by blind predictions. Data-driven engineering accelerates these models with machine learning, turning extensive experimental and simulated data into predictive power. Digital twins then bring the material-aware surrogates, real-time monitoring, and closed-loop adaptation into the forming line to deliver the material design. Examples are showcased from sheet and bulk forming, including the local formability of advanced high-strength steels and closed-loop controlled open-die forging. The talk closes with the open ends of this pathway and outlines the potential to design materials via forming.

Alexander Brosius (Germany): (Bio)

MATERIAL CHARACTERISATION AND BENDING OF AA6082 AT CRYOGENIC TEMPERATURES

Cryogenic applications are found in numerous industries, such as aerospace, electrical engineering, and the liquefied natural gas industry. In sheet metal forming, cryogenic temperatures are used to improve the limited formability of aluminum alloys at room temperature without the disadvantages of hot forming, such as multiple process steps at high temperature and reduced strength due to recovery and recrystallization. Cryogenic forming prevents heat-induced microstructural changes and also leads to an increase in the strength and formability of aluminum alloys during forming, so that material failure occurs at higher strains. The challenge here is determining the material properties at very low temperatures, which is a requirement for numerical process design.

A nitrogen-based flow cryostat that enables material characterization at cryogenic temperatures was designed. Details and operation are presented in this contribution. With this setup, temperature-dependent flow curves and failure strains can be investigated at temperature levels down to 130 K.

In addition, this work shows a bending process at cryogenic temperatures as a possible strategy to overcome the limitation of forming aluminum alloys. Here, forming at cryogenic blank temperatures is used to increase the bending angle of aluminum blanks made of AA6082 compared to bending at room temperature with a high sheet thickness of 5 mm. For this purpose, three-point bending tests are carried out, whereby only the aluminum blank is cooled in liquid nitrogen, while the tools remain at room temperature.