Bridging the Gap Between Digital and Physical: Predicting Sintering Shrinkage in CMF Additive Manufacturing

ColdMetalFusion (CMF) is unlocking new possibilities in additive manufacturing by combining the geometric freedom of 3D printing with the proven robustness of powder metallurgy. However, the sintering phase introduces a critical hurdle: part shrinkage and distortion. As the green part densifies in the furnace, complex geometries warp, causing out-of-tolerance failures.

To demonstrate how to address this challenge, we virtually predicted the behavior during and after the sinter process of an impeller using the Rheona™ simulation platform, proving that digital twin simulation is not just an approximation, but a highly accurate predictive tool.

Predicted shrinkage for an an IN635 impeller model via the Rheona™ simulation platform.

The Challenge: Sintering Complex Impeller Geometries

Determining if a complex geometry like an impeller is fit for CMF can be notoriously tricky without simulation. The combination of thin, swept blades and a massive central hub creates varied thermal masses; simulation allows engineers to predict upfront whether the design will successfully densify, or if it will succumb to non-uniform cooling, differential shrinkage, and severe distortion.

For this simulation, an IN625 impeller model was printed using Headmade's CMF material and subjected to a 19-hour sintering cycle. The furnace profile featured strategic between 300°C and 600°C to manage binder burnout, followed by a peak temperature of 1280°C to drive the final densification.

Furnace profile for sintering the IN625 impeller.

Our goal at NODiVEC was simple: use Rheona to predict the final dimensions of the sintered part accurately.

The NODiVEC Simulation Setup

Setting up the digital twin requires translating the physical reality into robust computational mechanics. Here is how we configured the simulation within the Rheona environment:

To accurately predict the behavior of an Inconel 625 (IN625) impeller during a ColdMetalFusion (CMF) sintering cycle, the NODiVEC Rheona solver utilizes a fully coupled thermomechanical approach. Here is the technical breakdown of the simulation parameters and why they matter:

1. Spatial Discretization (The Mesh)

• The Setup: The continuous CAD geometry was discretized into a finite element mesh consisting of 34,211 elements and 48,420 nodes.

• The Engineering Impact: This specific nodal density strikes an optimal balance. It provides spatial resolution to capture localized distortions on the thin impeller blades, while maintaining a small memory footprint and enable rapid solve times (under 7 minutes).

  
  

2. Volumetric Baseline (Material Calibration)

• The Setup: The initial relative density of the green part was defined as 64.6% and the thermo-mechanical properties were gathered from literature.

• The Engineering Impact: In solid-state sintering, densification is essentially the elimination of porosity. By establishing this exact baseline, the solver maps the maximum allowable volumetric strain before theoretical full density is reached. This acts as the fundamental boundary condition for the conservation of mass equations throughout the thermal cycle.

  
  

3. Kinematic Constraints (Boundary Conditions)

• The Setup: A localized Dirichlet boundary condition was applied. Specifically, an absolute zero-penetration fixation was mapped to the nodes interfacing with the setter plate.

• The Engineering Impact: Sintering shrinkage is mathematically isotropic (uniform in all directions) unless restricted. By constraining these bottom nodes, the solver accurately mimics the frictional drag between the IN625 part and the ceramic setter. As the part attempts to shrink, this boundary constraint induces localized shear stresses, which propagate upward and manifest as macroscopic warpage or "cupping" at the base of the part.

  
  

4. The Viscous Sintering Model (Thermomechanical Coupling)

• The core of the simulation relies on modeling the IN625 powder compact not as a solid, but as a highly viscous, temperature-dependent, non-Newtonian fluid.

• The Driving Force (Capillary Stress): Densification is driven by the thermodynamic desire to minimize surface area. The solver calculates this driving capillary pressure using the specific surface energy of IN625 and the physical particle radius.

• Rheology and Arrhenius Kinetics: As the temperature rises, the material's viscosity drops, allowing flow. However, IN625 undergoes complex microstructural changes (such as the dissolution of precipitates) as it heats. To model this, the solver uses a piecewise Arrhenius formulation. This allows the activation energy and viscosity curves to mathematically shift at specific thermal boundaries (up to 1310°C), accurately tracking the alloy's changing resistance to deformation.

• Microstructural Coarsening (Grain Growth): If the driving capillary force remained constant, the part would mathematically shrink to 100% density. In reality, powder particles coalesce and grow at peak temperatures, reducing the specific surface area and throttling the driving force. By enabling a dynamic grain growth model, the solver accurately chokes the densification rate near the end of the cycle, predicting a realistic final density of <100%.

Under the hood, Rheona utilizes a highly optimized backend, utilizing an Intel Pardiso solver

Real-World Accuracy: Simulation vs. Physical Reality

The true test of any simulation software is how well it aligns with calipers on the workbench. After the 19-hour physical sintering cycle completed, we measured impeller and compared it directly to the exported Rheona results.

Final part measuremetns if the manufactured impellerv (left) vs the predicted results (right). The impeller geometry was provided by CADMore Metal.

Beyond Dimensional Shrinkage: Predicting Sintering Defects

• Predicting final dimensions is only half the battle. The true value of a digital twin lies in its ability to foresee catastrophic failures before a physical part is ever printed. Rheona’s predictive capabilities extend far beyond basic scaling factors to capture complex mechanical interactions:

• Geometrical Inaccuracies and GD&T Failures: Simple scaling factors cannot account for anisotropic (non-uniform) shrinkage. Rheona predicts specific geometric deformations, such as out-of-roundness on critical rotating components like the impeller's hub, or loss of flatness across mating surfaces. This allows engineers to apply digital compensations to the green part geometry before printing.

• Gravity-Induced Drooping (Slumping): During the high-temperature viscous flow phase, unsupported overhangs and thin features become highly susceptible to gravity. To validate our solver's accuracy, we used Rheona to simulate complex T-shaped pipe connector geometries from a recent academic study (Torresani, Elisa, et al., 2024). In the physical experiments, the concentrated weight of the vertical components caused the unsupported horizontal tubes to severely slump and droop during sintering. Rheona successfully simulated this gravity-induced viscoelastic deformation, accurately predicting the final drooping behavior with absolute deviations of less than 1 mm compared to the real-world measurements. This capability allows engineers to visualize sag and design effective ceramic setters orsupports before committing to a print.
  

  

  See this blog post for more information.
  

• Thermal Stress and Cracking: Differential heating and cooling rates across a part—especially between thick hubs and thin blades—generate massive internal stresses. By continuously tracking stresses and localized strain throughout the thermal cycle, Rheona highlights high-risk zones where the green part is likely to fracture or tear.

• Friction and Base Warpage: As the part densifies and shrinks, it physically drags across the setter plate. If the friction coefficient is too high, the base of the part will warp, cup, or shear. Rheona mathematically models these boundary interactions, allowing you to optimize setter materials, release agents, and part orientation to mitigate drag-induced defects.

The Digital Gatekeeper: Determining CMF Suitability

Not every part is an ideal candidate for ColdMetalFusion right out of the box. Massive cross-sections adjacent to paper-thin walls can create unsolvable thermal gradients and tear the part apart from the inside during sintering.

Simulation acts as a rapid, zero-cost screening tool. Before committing to physical production, engineering teams can use Rheona to determine if a part's geometry is inherently suitable for the CMF process. If the simulation reveals unavoidable warping or cracking, the team immediately knows that Design for Additive Manufacturing (DfAM) modifications—such as lightweighting, adding support ribs, or hollowing out dense hubs—are required.

Unprecedented Speed

What is arguably as impressive as the dimensional and structural accuracy is the speed at which it was achieved. To simulate the entire 19-hour adaptive time-stepping cycle, the Rheona solver took exactly 405.9 seconds (6.7 minutes) of wall time.

Engineers no longer have to wait overnight for a solver to finish. You can iterate designs, adjust your setter strategies, assess CMF suitability, and validate the thermal outcomes in minutes.

The Bottom Line

Sintering distortion and defects no longer need to be solved through expensive, iterative trial-and-error prints. By pairing high-quality CMF processes with the robust predictive power of NODiVEC’s Rheona solver, engineers can achieve first-time-right manufacturing.

When your simulations predict reality this closely, you can design with absolute confidence.