Validating Rheona™ Distortion Predictions Against the NIST AM-Bench

For engineers in metal additive manufacturing, simulation software often feels like a black box. You feed it a mesh, wait hours (or days), and get a colorful contour plot. But when it comes time to commit to a costly Laser Powder Bed Fusion (LPBF) print, you need to know if you can actually trust those numbers to compensate for real-world distortion.

At NODiVEC, the development philosophy behind Rheona™ has always prioritized direct, transparent communication and accessible, production-focused tools over heavy, opaque academic research packages. A solver is only as good as its validation. To prove Rheona’s capabilities, we put it up against the gold standard in AM validation: the NIST AM-Bench 2022-01 (AMB2022-01) challenge.

Here is an honest, data-driven look at how Rheona handles complex macroscopic distortion.

The Challenge: AMB2022-01 Part Deflection

NIST’s AMB2022-01 benchmark involves an IN718 bridge structure built with 12 legs of varying geometries. To measure residual stress accumulation, NIST partially cuts the part from the 12.5 mm build plate using wire EDM, leaving one end clamped. As the internal stresses are released, the free end of the cantilever springs upward.

The physical measurement is taken at six specific ridges along the top of the bridge using a Coordinate Measurement Machine (CMM). The goal for any solver is to accurately predict this final Z-displacement profile.

Setting Up the Simulation in Rheona

Instead of bogging down the user with convoluted parameter trees, Rheona is designed to get straight to the physics.

We imported the NIST STL files, generated the mesh (252,8643elements), and assigned the IN718 material properties. Simulating the EDM release requires precise boundary conditions. Using Rheona’s node set management, we locked the NS_BPFIXATION region on the right side of the bridge—mirroring the exact clamped physical state of the NIST experiment.

The Results: Simulation vs. Reality

When running the inherent strain structural solver, we probed the exact node locations corresponding to NIST’s CMM measurements along the top ridges. Here is how Rheona’s Z-axis displacement predictions stack up against the rigorous physical measurements published by NIST:

What This Means for Production

Rheona captures the macroscopic springback profile with excellent accuracy at the critical free tip, predicting 1.78 mm of deflection compared to the physical 1.809 mm.

While there is some deviation in the mid-span (a known characteristic of fast inherent-strain approximations assuming uniform strain across complex geometries like the hollow Leg 10), capturing the peak distortion within ~<2% error at the tip is exactly what is needed for effective geometry compensation.

More importantly, Rheona delivers this level of accuracy without requiring a supercomputer. It provides a lightweight, highly responsive workflow that allows shop-floor engineers to predict distortion, iterate on part orientation, and generate counter-deformed geometries rapidly.

Simulation shouldn't be a research project. It should be a tool.