Solver evidence

A reproducible 1000 N compression benchmark

Use a simple block to check structural load signs, units, reactions and mesh refinement.

A small compression block is a useful place to test a structural workflow because its expected response can be calculated independently. The point is to compare a numerical result with a known solution, then inspect the evidence behind that result. A colored deformation plot by itself cannot establish that the load, material and supports mean what the user intended.

Define the problem before solving it

The Intact 0.16 verification case uses one native solid measuring 10 × 10 × 20 mm. Its cross-sectional area is 100 mm² and its height is 20 mm. The material has Young’s modulus E = 10 GPa and Poisson’s ratio ν = 0. Fix the bottom face and apply a total downward force of 1000 N to the top face, along native CAD −Z.

“Total force” is important. It is not 1000 N per node and not a pressure of 1000 in an unspecified unit system. The fixed bottom prevents displacement there; the load acts on the opposing top face. Record the actual face selections so an apparently correct setup cannot conceal a force applied to the wrong surface.

Compute the reference answer

For uniform axial compression, displacement magnitude is u = FL/(EA). Convert the modulus to compatible units: 10 GPa = 10,000 N/mm². Substitution gives 1000 × 20 / (10,000 × 100) = 0.020 mm. The loaded top moves in negative Z; 0.020 mm is the magnitude of its compression displacement.

The axial stress magnitude is F/A = 1000/100 = 10 N/mm², or 10 MPa. With a uniaxial stress state, the von Mises equivalent stress has the same magnitude. The fixed support should supply a positive Z reaction of 1000 N to balance the negative Z load. Poisson’s ratio zero removes lateral contraction coupling from this simple reference case.

Compare with actual solver evidence

The 0.16 release verification reports a live local AI-to-solver-to-viewer run with displacement 0.020000000000000122 mm, peak element-average von Mises stress 10 MPa and Z reaction 999.9999999999994 N. The final mesh had 208 nodes and 588 tetrahedra. Those small differences from the hand calculation are consistent with numerical arithmetic.

The same report records a residual of approximately 1.02 × 10⁻¹⁴ and a force-balance error of approximately 6.47 × 10⁻¹⁶. These are reported numerical diagnostics, not physical uncertainties or manufacturing tolerances. Read their definitions in the solver context before comparing values from another solver or another problem.

Refinement is part of the check

A single mesh can agree accidentally, particularly on a simple geometry. The reported benchmark passed displacement and stress refinement checks. When reproducing it, retain mesh counts, quality information, convergence status and results at successive resolutions. Look for stable quantities within the chosen acceptance target rather than merely selecting the smallest element size available.

Resource limits also matter. A refinement that stops at a budget without meeting its target should remain unverified. A failed or nonconverged run needs review; preparing a study or displaying an answer is not equivalent to completing a successful numerical analysis.

A practical reproduction brief

Ask the desktop workflow to create or inspect the block, confirm millimeter dimensions, select the bottom support and top load, and report the stated material properties. Then request the actual static structural solution, displacement direction and magnitude, stress, summed reaction, residual, mesh information and refinement evidence. Preserve the native geometry and revision during the study.

Keep the analytical calculation beside the result. If displacement differs by a factor of 1000, investigate unit conversion before interpreting a plot. If the reaction has the wrong sign or magnitude, inspect axes, support selection and load interpretation. These simple checks are often more informative than adding visual detail.

Know what the benchmark establishes

This is a single-solid compression test. It does not predict the bending of a vehicle, validate an entire assembly or exercise contact between parts. Intact’s supported structural scope does not include whole-assembly contact, nonlinear mechanics or production certification. A separate softer 1 GPa case reached the small-strain acceptance boundary and correctly required review.

The browser demo illustrates a prebuilt interaction; it does not execute this solver. Follow local setup for native Windows x86-64 computation and read how the workflow operates before interpreting results.

Use this guide in ChatGPT

Enable the installed Intact Simulation plugin in a new conversation. State the design goal and the checks you expect, then ask the assistant to report the native measurements and failed or unverified criteria. Use Review for a measured overview and ask for a parameter preview before committing an alternative. Keep the accepted revision available while you compare results.

Follow How to Use for the practical controls and Architecture for the audit and verification chain. Plugin setup explains the local runtime and your own ChatGPT login.

Install once.
Design through conversation.

Download Intact v0.20.0 for Windows. Install privately in Codex with your own account; CAD runs on your computer.

Follow the installation procedure →
Download Windows ZIP ↓