Positioning
If you already use a commercial geotechnical finite element program, you know the vocabulary: Hardening Soil, Undrained (A), Rinter, ΣMstage, φ–c reduction. KATAI 2D speaks it. What it changes is everything around the analysis — the source is public, the formulation is written down, the benchmark inputs ship with the program, the whole thing runs from a script, and it costs nothing. This page says precisely where it stands, including where it does not yet stand at all.
Four things an open engine gives you
When a result is questioned in a design review, “the software says so” is not an answer. With KATAI 2D the return mapping, the convergence test and the boundary condition are in a public repository, and the equations are on the theory page.
A validation manual tells you a program was checked. A corpus of 26 input files and a suite of 207 tests lets you check it yourself, on your machine, on the version you are about to use.
A parametric study, a back-analysis against monitoring data, a probabilistic run of a thousand realisations: each is a Python loop or a batch of command lines. No seat is tied up, and nothing needs a mouse.
A project is a text file and the engine is a file you may keep for ever. In ten years the analysis can be rerun bit for bit, by you or by whoever inherits the structure — without a maintenance contract.
Every student gets the full program on their own laptop, with the source of every model they are being taught. No lab licences, no feature-limited edition.
Apache‑2.0 allows commercial use and proprietary extensions. A consultancy can wrap the engine in its own tools; a researcher can add a model and publish it.
By category
A comparison of categories, not of brands — features of individual commercial products change with every release, and their own documentation is the authority on them.
| KATAI 2D 0.9.0 | Commercial geotechnical FE (typical) | Limit-equilibrium programs | General-purpose FE codes | |
|---|---|---|---|---|
| Source code | Open, Apache-2.0 | Closed | Mostly closed | Varies |
| Cost | Free | Licence or subscription per seat | Licence per seat, usually lower | Free to very high |
| Deformations and structural forces | Yes | Yes | No — stability only | Yes |
| Geotechnical constitutive models | 7, incl. HS small, Soft Soil Creep, Hoek–Brown | Broad libraries, incl. advanced and user-defined models | Strength models only | Often few; user subroutines |
| Staged construction, consolidation, seepage, dynamics | Yes, as phases of one model | Yes | Limited | Possible, by hand |
| Three-dimensional analysis | No | In separate 3D products | In some | Yes |
| Liquefaction models | No | In several | Empirical, in some | User-supplied |
| Graphical pre- and post-processing | Desktop application in development; not in the open-source release | Yes, mature | Yes | Varies |
| Scripting | Native: Python API and CLI | Available in many, as an add-on layer | Varies | Yes |
| Input format | Documented JSON with a schema | Proprietary project files | Proprietary | Text decks |
| Verification evidence | 69 cases, inputs and tests public | Published validation manuals | Published manuals | Patch tests; geotechnical cases rare |
| Vendor support, training, certification | Community; issue tracker | Yes | Yes | Varies |
Measured
The PLAXIS 2D Validation Manual publishes analytic references beside its own results for a set of classical problems. Four of them were rebuilt from the problem statements and solved in KATAI 2D from input files that are in the repository.
| Problem | Analytic reference | PLAXIS 2D, as published | KATAI 2D 0.9.0 |
|---|---|---|---|
| Smooth rigid strip footing on elastic soil — force at 10 mm settlement | Giroud (1972): 15.15 kN/m | 15.24 (+0.6%) | 15.32 (+1.1%) |
| Strip load on incompressible Gibson soil — settlement | Gibson (1967): 0.050 m half-space | 0.047 m (finite layer) | 0.0454 m (−3.4% vs 0.047) |
| Smooth rigid circular footing, axisymmetric — limit pressure | Cox (1962): 225.6 kPa | 220.0 (−2.5%) | 233.9 (+3.7%) |
| Strip footing on clay, strength increasing with depth — limit pressure | Davis & Booker (1973): 7.80 kPa | 7.86 (+0.8%) | 7.91 (+1.4%) |
Read this table for what it is. The PLAXIS values are the published ones, quoted unchanged; PLAXIS was not re-run, and the two programs do not share a mesh. What is identical is the problem definition and the reference. On that basis both programs sit within a few percent of the exact solutions — and on the circular footing they bracket it from opposite sides. That is the expected outcome for two correct implementations of the same mechanics, and it is the only claim made here.
The four cases are reproduced in full — problem statements, formulas, command-line transcripts, file addresses — on a dedicated page.
Most of what you know transfers directly.
What differs is how you drive it: a script or a text file instead of a sequence of dialogs, and units that are fixed (kN, m, day) rather than chosen.
An honest recommendation includes the cases where the answer is no.
For two-dimensional deformation, stability, consolidation, seepage and site-response analysis with the standard soil models — most of everyday geotechnical finite element work — it is a complete tool, and a verifiable one.
Adopting an open engine does not require abandoning anything. Engineers use KATAI 2D next to a commercial package for three things in particular:
PLAXIS is a trademark of Bentley Systems, Incorporated. KATAI 2D is an independent project and is not affiliated with, endorsed by or derived from any commercial product. Published values are quoted from the cited manual for the purpose of comparison. The comparison document in the repository is plaxis-2d-validation-comparison.md.
KATAI 2D is free under Apache‑2.0, installs in one line and ships the input files of the benchmarks it is verified against. The fastest way to judge it is to rerun one.