Process simulation software: four categories, four purposes

Asking “which process simulation tool should we use?” almost always produces a feature list and almost never a good decision. The reason: tools that appear on the same search results page solve fundamentally different problems. Buy from the wrong category and you find out three months into the training, or the first time someone in a steering committee asks a follow-up question.

So let us sort them by purpose rather than by feature count.

Category 1: Drawing tools

Visio, Lucidchart, draw.io, Miro, Whimsical

They produce a picture of the process. That is worth more than engineers like to admit: a shared picture ends half the misunderstandings in a workshop, and for documentation, training and audits it is exactly right.

What they do not do is compute. There is no concept of time, volume, capacity or queue. A box labelled “Review, 10 min” is text inside a rectangle, not a model. The question “what happens if volume rises by 30 %?” has no answer in a drawing tool, only an opinion.

Category 2: BPM suites

SAP Signavio, Software AG ARIS, Bizagi, Camunda, Appian, Pega

This category is about process management at scale: process landscapes spanning hundreds of flows, approval and version control, links to roles, systems and risks, and in some cases execution of the process as a running application.

Simulation here is one module among many — its depth and availability depend on product and edition. Camunda, for instance, is not a simulator at all but an execution engine: it runs real cases through the process rather than simulated ones.

The honest assessment: if your organisation already runs a BPM suite, model inside it. If it does not, entering through a suite to answer “where does this one process break, and what is a change worth?” is a very heavy path. You are buying governance where you need arithmetic.

Category 3: Simulation labs

Arena, Simul8, AnyLogic, Siemens Plant Simulation, FlexSim, Witness

These have the strongest engines. Discrete-event simulation, partly agent-based, partly system dynamics, with distributions, shift calendars, setup times, transport logic, experiment designs and statistical analysis. If you are designing a production line, a warehouse, an emergency department or a terminal, this is the right category — and the only one.

The price is twofold. First, the learning curve: these tools assume you know what a lognormal distribution is and why the warm-up period distorts your results. Two to four weeks to a defensible model is normal, not pessimistic. Second, the heritage: the interaction model comes from the factory floor. An administrative process with approvals, email queries and three departments can be represented, but it always feels slightly translated.

Category 4: Decision instruments

FlowVisual and similar approaches

This category answers a narrower question: where is the constraint in this process, and what is a specific change worth in money and time? It computes on the same principles as the labs — discrete events, Monte Carlo repetition, ranges instead of point values — but deliberately omits transport logic, setup matrices and experiment designs.

What it buys you: a model in an hour instead of three weeks, and a document at the end that somebody without a statistics background can read and defend.

The omission is the point of the category. If you are designing a production line, you are in the wrong place.

The categories side by side

Three questions before you buy

1. Do you need a picture or a number? If the answer is “a picture”, do not buy anything with an engine. You will not feed it, and an unfed model is worse than no model — it looks like somebody did the maths.

2. Is the question one-off or ongoing? A one-off question (“is this automation worth it?”) does not justify a suite. An ongoing obligation (“we manage 300 processes under approval control”) does not justify a single-purpose tool.

3. Who has to believe the result? This is the question most often missing. If a CFO has to sign the number, it needs a range, a traceable origin and an honest statement of what it does not cover. A tool that outputs a single number to two decimal places has not understood the question.

How to spot a useless comparison

Comparison tables online — including vendors’ own — share three recurring weaknesses:

• Checkmarks without depth. “Simulation ✓” covers everything from “colours the boxes” to “discrete-event engine with confidence intervals”.
• Prices without editions. Almost every product named above has editions in which the decisive capability is missing.
• Silence about onboarding. For simulation tools, training is routinely the larger cost block, not the licence.

So ask the same thing in every sales conversation: show me my model, with my numbers, all the way to a result — today. Whatever fails there will fail in the project too.

Where FlowVisual sits

FlowVisual is deliberately category 4. It did not start as a software product but inside consulting engagements, where the same question kept coming: and how do we know that? You assemble the process from building blocks, run it under load and evaluate it in money — with a P10–P90 band instead of false precision, locally on your machine, without an account.

Designing a factory? Take category 3. Managing 300 processes? Category 2. Need to prove next week what a change to one process is worth? That is what FlowVisual is built for.