February 2026 Volume 8

FORGING RESEARCH

Implementation and Integration The research implements a novel coupling strategy through the DynamiX tool, which integrates mean-field solvers with FORGE®'s process simulation capabilities. This architecture enables real-time visualization of microstructural evolution on selected workpiece cross-sections, sensor-based data extraction at specific material points, flexible solver selection between external mean-field codes and the integrated NHM solver, and micro macro linking between process and microstructural scales.

2. Full-Field Approach (Level-Set Method in DIGIMU®) The full-field model employs a Level-Set framework within a finite element context to explicitly track grain boundaries and their evolution. The methodology involves: • Explicit microstructure representation: Each grain is individually meshed with accurate representation of grain shapes and neighborhood relationships • Representative Volume Element (RVE) analysis: Boundary conditions reflect macroscopic thermomechanical loading experienced by material points • Physical mechanism simulation: Direct modeling of recrystallization, grain growth, and Smith-Zener pinning by second-phase particles • Heterogeneity capture: Preservation of topological features and local variations in stored energy and driving forces 3. Mean-Field Approach (Neighborhood Model – NHM) The NHM introduces an innovative statistical framework that advances beyond classical mean-field assumptions. The methodology encompasses: Microstructure representation: The model defines grain classes characterized by three state variables: grain radius (Ri) for class i, frequency ( η i) of grains in class i, and dislocation density ( ρ i). Neighborhood statistics: Unlike homogeneous medium assumptions, NHM considers statistical neighborhoods where each grain class is surrounded by grains from other classes with specific contact probabilities. Volumetric exchange mechanisms: Microstructural evolution occurs through grain boundary migration described by: dV(i,j) = dR(i,j) × Sc(i,j) = dR(i,j) × p(i,j) × SRi Shape evolution: The model allows grain shapes to evolve from spherical to ellipsoidal during deformation, better capturing changes in grain boundary surface area.

Figure 3: (a) User interface of Dynamix in FORGE®, (b) definition of the cutting plane (via normal vector data), and of the sensor grid density, (c) the initial microstructure is defined by a grain size distribution, and the material is characterized by a set of parameters. Experimental Validation Framework The validation methodology employs Inconel 718 as the model material system, with testing conditions spanning: temperature range of 940-1080°C (industrial forging window), strain rates of 0.001-1 s -1 , initial microstructure of 300-grain representative volume elements with 20 μ m average grain size, and second phase particle configurations of 0.6 μ m diameter particles at 0.5% volume fraction. Results Comparative Performance Analysis Computational Efficiency The three modeling approaches demonstrated dramatically different computational requirements: • JMAK model: Negligible additional computation time (integrated within FE solver) • NHM mean-field: 5-10 minute computation time per material point • DIGIMU® full-field: 2-8 hours per material point depending on microstructure complexity This represents a computational speed advantage of approximately 50-100x for the mean-field approach compared to full-field simulations, while maintaining substantially higher physical fidelity than phenomenological models.

Figure 2: 3D representation of the microstructure for the Maire model as illustrated in the work of Roth

FIA MAGAZINE | FEBRUARY 2026 67