FTA Studio vs OpenFTA
OpenFTA is the Java desktop fault tree tool almost everyone tries first, because it is the one that comes up when you search for a free one. It is GPL-licensed, it runs on Windows, macOS and Linux, and it is genuinely, permanently yours. FTA Studio has a free tier too — but the two words mean different things, and the difference matters more than any feature table. Here is the honest version.
Two kinds of free, stated plainly
OpenFTA is free as in GPL. Download it, run it offline forever, no account, no telemetry, no cap on anything, no vendor who can change the terms. If the project disappears tomorrow your copy keeps working, and you have the source. For work that cannot touch a network, or an institution that cannot sign up for things, that is not a lesser kind of free — it is the stronger one, and no SaaS tier matches it.
FTA Studio Community is free as in tier. It costs nothing and does not expire, but it wants an account, it runs in the browser, and it caps you at 3 projects, 10 trees per project and 3 members. Above that it is $99 per seat per year. We think that is a fair deal; it is not the same deal.
At a glance
| FTA Studio | OpenFTA | |
|---|---|---|
| Licence | Proprietary; free Community tier | GPL — yours, with source |
| Runs offline | No — browser, needs the service | Yes, entirely local |
| Account required | Yes | No |
| Install | None — open a URL | Java desktop app |
| Caps | 3 projects / 10 trees each / 3 members on Community | None |
| Cut-set algorithm | MOCUS | MOCUS |
| Common-cause failure | Beta-factor CCF | Not modelled |
| Monte Carlo uncertainty | Yes | No |
| Actively developed | Yes | Essentially dormant; UI unchanged since the early 2010s |
| Sharing with a reviewer | A URL, or JSON | Send them the file and the installer |
The two gaps that decide it for safety work
Everything above is a preference until you reach these two, which are not preferences.
No common-cause failure modelling
Redundancy is the whole reason safety architectures exist, and CCF is the reason redundancy delivers less than the arithmetic suggests. Our worked 2-of-3 example puts a number on it: at β = 0.05, the common-cause term (5.0e-4) is larger than the independent path (3.0e-4). A tool that cannot express β does not just lose precision on a redundant architecture — it reports the wrong answer, in the optimistic direction, on exactly the systems people build fault trees for.
No Monte Carlo
Failure rate data comes with uncertainty, often an order of magnitude of it. A single point estimate carried through to a top-event probability implies a confidence nobody has. Propagating the distribution is standard practice in PRA and increasingly expected elsewhere; without it you are submitting a number with no interval attached.
Our own market survey reaches the same conclusion and has for a while: OpenFTA is sound for education and introductory use, and the missing CCF and Monte Carlo rule it out for production safety-case work. That was written before this page and we are not going to soften it now that we have a comparison to make.
When to pick FTA Studio
- The tree has redundancy. Voting architectures, dual channels, anything where CCF is the dominant term.
- You need an interval, not a point. Monte Carlo over lognormal inputs, because the data has spread.
- Reviewers must open it. A URL beats an installer and a file, especially when the reviewer is a customer.
- Mixed machines, no admin rights. Nothing to install is sometimes the entire argument.
- You want it maintained. Someone answers support, and the tool changes when standards do.
When to pick OpenFTA
- Air-gapped or offline work. Classified, ITAR, CUI, or simply a lab with no network. This is decisive and we cannot match it.
- You need the source. GPL, auditable, modifiable, no vendor between you and the algorithm.
- No account, on principle or by policy. Some institutions genuinely cannot.
- Teaching FTA. Free for a whole cohort with no seat management, and a simple tree is exactly what a course needs.
- Trees without redundancy. If there is no CCF to model, the biggest gap does not apply to you.
If you are moving from OpenFTA
There is no importer, and we will not pretend one is coming: OpenFTA's format is its own, and the population moving is small enough that hand-rebuilding is honestly the faster path for most trees. What usually helps is starting from a worked template in the same shape as your system and editing it, rather than rebuilding gate by gate from a blank canvas.
The thing worth doing while you rebuild is adding what OpenFTA could not express — the β on your redundant groups, and a distribution rather than a point on the rates you were least sure about. That is the part that changes the answer, and it is the reason to move at all.
OpenFTA originated with Formal Software Construction and is GPL-licensed. We are not affiliated with it or its maintainers, and it deserves credit: it introduced a great many engineers to fault tree analysis and asked nothing in return. Details here reflect publicly available information as of September 2026 and the project's dormant state; if it is revived, or if something below is wrong, email support@ftastudio.app and we will correct it.