All articles
Comparison2 September 2026 · 4 min read

Free process simulation software: seven tools and their real price

In short

Free process simulation comes in four forms. Open source and unrestricted for commercial use: SimPy, Salabim, Ciw (all Python) and JaamSim (Java, with an interface) — you write the model as a program or operate a utilitarian interface. The AnyLogic Personal Learning Edition is free but explicitly not for commercial use. Bizagi Modeler allows free modelling; simulation scope depends on the edition. FlowVisual makes modelling and stress testing permanently free and charges for saving, comparing and exporting. The real price of open-source tools is statistical responsibility: warm-up period, replication count and distribution choice are yours.

Before the tools: "free" means four different things in this market, and the difference is legal, not cosmetic.

KindWhat you may doTypical trap
Open sourceEverything, including commercial useYou carry the statistical responsibility
Free tierA defined part, permanentlyThe deciding step sits beyond the line
Learning editionLearn and try, not commercialGets used in a client project, breaching the licence
Trial periodEverything, for 14 or 30 daysThe project outlasts the trial

The third row is the dangerous one. A model built in a learning edition does not belong in a proposal to a client, however good it is.

1. SimPy (Python, open source)

The most widely used free DES library. Process-based, with a very clear core: processes are generator functions, resources are objects with capacity, the clock jumps from event to event.

Strong when: your team writes Python anyway. Models live in version control, parameter sweeps are a loop, and analysis happens with the same tooling as the rest of your analytics.

Weak when: somebody outside the team has to review or operate the model. There is no interface, no animation and no statistics out of the box — you compute confidence intervals yourself.

2. Salabim (Python, open source)

Close to SimPy, but with animation and queue statistics built in. For those who want the Python route yet have to show something in a meeting.

Strong when: you want to make behaviour visible without buying a commercial tool.

Weak when: the audience does not want to see a script, however pretty the animation.

3. Ciw (Python, open source)

Built for queueing networks and rigorously documented. If you compute networks with several nodes, priorities and overflow routing, this has the most fitting vocabulary.

Strong when: the question is queue-theoretic and you want to check results against theory.

Weak when: you need costs, calendars and roles — you will have to build those alongside it.

4. JaamSim (Java, open source)

The one open-source tool with a real graphical interface: drag blocks, connect them, set parameters, run — including 2D animation.

Strong when: you need an interface and do not want to pay a licence.

Weak when: you expect commercial-lab ergonomics. The interface is functional, not friendly.

5. AnyLogic Personal Learning Edition

Free, with the full multi-method range — discrete, agent-based, system dynamics — but with clear conditions: not for commercial use, plus limits on model size and export.

Strong when: you want to learn, teach or find out whether multi-method modelling solves your problem.

Weak when: the result is destined for a proposal, a decision paper or a client presentation. Then you need the commercial edition, which is precisely what the learning edition exists to demonstrate.

6. Bizagi Modeler

BPMN modelling is free to use, and simulation runs in stages — process validation, time analysis, resource analysis, calendar analysis. How far you get for free depends on product state and edition; check that before planning, not after.

Strong when: you document in BPMN anyway and want the simulation in the same artefact.

Weak when: you need ranges instead of point values, or want to keep the model coarse: a documentation-grade diagram has thirty activities, a computable one needs ten.

7. FlowVisual (free tier)

Modelling and stress testing are permanently free — including the constraint display, utilisation on the peak day and lead time as a P10–P90 band. Saving, the before/after comparison and PDF output are paid.

Strong when: you want to know in half an hour where the constraint in an administrative process sits, with no account and no data leaving the device.

Weak when: you are designing manufacturing. Transport logic, setup matrices and experiment designs are deliberately absent.

What you actually pay by saving

Statistical responsibility. With the libraries, you decide on warm-up period, replication count and distribution choice. If you can do that, you save real money. If you cannot, you produce numbers with false confidence — which is worse than none.

Ability to involve others. The greatest value of a simulation arises in the workshop, when the department objects and corrects the model. A script cannot do that: whoever does not read what is in it cannot object to it either.

Maintenance. A script one person wrote becomes an artefact once that person leaves. A model in a tool with an interface is more likely to survive the handover.

The choice in three questions

  1. May the result be used commercially? If yes, learning editions are out — entirely, not "with a note".
  2. Does your team write Python? If yes, SimPy or Ciw is the cheapest serious option. If no, it is no option at all.
  3. Does somebody need to watch along? If yes, you need an interface — JaamSim, Bizagi or a tool with a free tier.

In summary

  • Four meanings of "free", and only two of them permit commercial use.
  • Open source is strong for coding teams, weak for participation.
  • Learning editions are not a route for client work.
  • Free tiers of commercial tools are the most pragmatic entry point, provided the line sits beyond what you need.

Frequently asked

Is there genuinely free process simulation for commercial use?

Yes: SimPy, Salabim, Ciw and JaamSim are released under free licences and may be used commercially. Not free for commercial use are learning editions of commercial products, such as the AnyLogic Personal Learning Edition. Free tiers of commercial tools sit in between — what the tier covers is what is permitted.

Is SimPy hard to learn?

The core is not: processes are generator functions, resources are objects with capacity, and a simple queueing model fits in fifty lines. What is hard is everything after that — warm-up period, replication count, distribution choice and confidence intervals. Commercial tools take some of those decisions off your hands; with a library they are entirely yours.

Can I simulate for free in Excel?

Only up to a point. Excel adds processing times but does not represent a queue in front of a busy step — and that is where lead time is created. Monte Carlo draws can be rebuilt; the coupling between steps cannot. A step behind the constraint looks relaxed in the spreadsheet and becomes the constraint itself once the first one is fixed.

Is a free tier enough for a business case?

For the analytical part, often yes — constraint, utilisation, lead time as a band. What free tiers typically omit is the output: saving, before/after comparison, PDF with the assumption list. Those are exactly the parts a steering committee reads, which is why they sit beyond the line.

What is the largest hidden cost?

Participation. Most of the value of a simulation arises in the workshop, when the department corrects the model. A tool without an interface excludes that part, and whatever is not corrected there surfaces later in the final presentation — as an objection to the result.

FlowVisual

Run the numbers on your own process

FlowVisual turns the figures in this article into a model that runs — with your volumes, your capacities, your range.

Guide: seven steps to the number