A civil engineer writes a finite-element solver in his spare time. Not in a software lab. In an aero-engine company. More than twenty-five years later, that code is still solving stresses, meshes, and contact.
That is CalculiX: an open-source FEM package whose core was built by Guido Dhondt, a civil engineer by training (KU Leuven and a Princeton doctorate), while he worked at MTU Aero Engines in Germany. Klaus Wittig — also in the MTU environment — is the main figure behind GraphiX (CGX), the pre- and post-processor. Wittig is not presented here as a civil engineer: his documented role is CGX.
The official site says it without romance: a team of enthusiasts wrote the package in what little free time they had; MTU allowed it to be published. It did not start as a startup or a marketing product. It started next to real industrial problems.
Two names, two pieces
CalculiX is associated mainly with:
- Guido Dhondt — the finite-element solver (CrunchiX / CCX).
- Klaus Wittig — the graphical pre- and post-processor (GraphiX / CGX).
Dhondt studied civil engineering in Leuven and earned a doctorate in civil engineering at Princeton. In 1987 he joined MTU, working on fracture mechanics and finite elements. In autumn 1998 he began writing the core that would become CalculiX.
Wittig maintains CGX: geometry, meshing, model setup, and result visualization. The full package is CCX + CGX, not the solver alone.
The first goal was small
The original project was much more modest than today’s CalculiX. Dhondt wanted a simple finite-element program for linear-elastic static analysis with 20-node brick elements and reduced integration.
Then the project grew.
FEAP as an influence
Early development was strongly influenced by FEAP, associated with the work of Zienkiewicz and Taylor. New scientific tools rarely appear without ancestors. CalculiX inherited ideas, conventions, and decades of the finite-element method (FEM).
1998: the logbook already shows a serious solver
The historical logbook from November 1998 already mentions anisotropic materials, multiple steps, frequency calculations, prestress, buckling, and sparse-matrix solvers. This was not a weekend exercise.
Toward the end of 1999, geometric and material nonlinearity arrived. Later came cyclic symmetry, shells, beams, heat transfer, dynamics, and thermomechanical analysis. CalculiX was becoming a general-purpose solver.
CrunchiX and GraphiX
CalculiX CrunchiX (CCX)
This is the solver. It takes an input deck and runs the analysis:
model.inp → ccx → results.
CalculiX GraphiX (CGX)
This is the pre- and post-processor: create or edit geometry, generate meshes, prepare models, and visualize results.
C and FORTRAN
CalculiX combines C and FORTRAN. FORTRAN still dominates a great deal of mature numerical work; C supports other parts of the infrastructure. Instead of rewriting everything for fashion, the code grew on a pragmatic base.
A key decision: Abaqus-like syntax
Naming conventions and the input format are inspired by Abaqus. The authors note that this syntax was used with permission from HKS, the original company behind Abaqus. An input deck can contain concepts such as *NODE, *ELEMENT, *MATERIAL, *BOUNDARY, *STEP.
That lowers the barrier for people who already know Abaqus.
GPL and free software
CalculiX is distributed under the GNU General Public License v2. The code can be studied and modified. That is why it fits so well inside other open-source CAE projects.
What is CalculiX good for?
It is especially useful for 3D finite-element analysis.
Linear static analysis
Structure + loads + restraints → displacements, stresses, reactions.
Nonlinear analysis
Large displacements, plasticity, contact, nonlinear materials. These need more numerical care than the linear case.
Modal analysis
Natural frequencies and vibration modes: resonance, machinery, structural components, dynamics.
Buckling
Buckling of plates, shells, thin members, and parts under compression.
Heat transfer
Thermal problems and the path temperature → thermal expansion → stresses: thermomechanical analysis.
Where does it fit in civil engineering?
CalculiX does not try to compete directly with ETABS as a building-specific program. They are different tools.
ETABS thinks in stories, beams, columns, walls, and design codes.
CalculiX thinks much closer to nodes, elements, materials, contact, stress, and strain.
That is why CalculiX shines in local analysis: a steel connection, a base plate, anchors, a mechanical detail, a precast component, complex geometry, stress concentrations, special parts, thermal behaviour.
We are not inventing bridges or signed real-world projects run in CalculiX. What we do claim is local use in the office, plus the official examples and tutorials from the project itself.
The aero-engine example
The official site shows models of components from a small jet engine built around a turbocharger rotor: burst speed, low-cycle fatigue, creep, natural frequencies, resonance. That is the class of problem the program was born for: detailed component analysis, not the whole building.
CalculiX in modern workflows
It does not have to deliver the entire user experience by itself. Other programs use it as an engine. The ecosystem includes workflows with FreeCAD FEM, PrePoMax, Mecway, Coreform Cubit, and preCICE for coupling.
Typical architecture:
GUI / CAD / mesher → input deck → CalculiX → results.
The solver works as a specialized component.
OpenFOAM + CalculiX
A common open-source pairing:
OpenFOAM (CFD) → pressures / temperatures → CalculiX (FEM).
For example: simulate aerodynamic or hydraulic pressure in OpenFOAM, transfer those loads onto a structure, and compute stresses and deformations in CalculiX. For more advanced coupling there is preCICE. An open-source FEM engine can live inside larger multiphysics workflows.
What CalculiX is not
Its strength is the solver.
If the goal is to design a whole building quickly under codes, ETABS is usually more practical. If we need detailed three-dimensional stresses in a complex connection, CalculiX can make a lot of sense.
Why the mesh matters
Like any FEM software, CalculiX does not turn a model into a correct answer by magic. You still have to choose element type, mesh size, refinement, contacts, restraints, material, and convergence.
2004: the theory behind the solver becomes a book
In 2004 Guido Dhondt published The Finite Element Method for Three-Dimensional Thermomechanical Applications. The book describes an important part of the theory behind the solver: an explicit link between theory and implementation.
More than 25 years later
In 2026 CalculiX is still active. The official site shows CalculiX 2.23. At the 2026 World Congress for Computational Mechanics a paper was presented titled More than 25 Years of CalculiX: Sustaining an Industrial Open-Source FEM Code: CalculiX still matters in MTU Aero Engines simulation workflows, and it also has a community outside the company. That is remarkable for a project that began in spare time in 1998.
CalculiX + agents
The solver can be run from the CLI. A future workflow can look like this: CAD → mesher → an agent writes the input deck → ccx → the agent checks convergence → results → report. That makes it attractive when the main interface does not have to be the solver itself.
On this hub we treat that role as a deterministic engine, not the chat.
Short timeline
- 1987 — Dhondt joins MTU.
- Autumn 1998 — first lines of CalculiX (3D linear statics).
- Shortly after — frequencies and buckling.
- 1999 — nonlinearity; shells / beams; heat transfer; dynamics.
- CCX + CGX — solver and graphics under the GPL.
- Open-source CAE integrations.
- 2026 — more than 25 years; CalculiX 2.23.
Conclusion
CalculiX shows that scientific software does not need to start as a large commercial platform to matter. It began as a small FEM program. It grew because its authors were solving real problems. It used C and FORTRAN. It borrowed familiar conventions. It was published as free software. And it became a mature solver for both industrial and academic workflows.
For AEC, its greatest value is probably not replacing building-specific programs. It is offering an open FEM engine when we need to look at the structure at a much more local, detailed level — and remembering that behind the solver is a civil engineer who wrote the first lines.
References
- CalculiX. About the authors.
- CalculiX official website.
- CalculiX (Dhondt).
- Guido Dhondt. CalculiX source repository.
- Dhondt, G. The Finite Element Method for Three-Dimensional Thermomechanical Applications. Wiley, 2004.
- WCCM-ECCOMAS 2026. More than 25 Years of CalculiX: Sustaining an Industrial Open-Source FEM Code.
Evaluation and limitations
Treat the notes below as evaluation context, not as a verified calculation or code check.
Treat it as an FEM engine for local, detailed analysis. Mesh quality, contact, restraints, and convergence remain the modeller’s responsibility.