BPMN simulation

BPMN describes the flow. Simulation needs four more inputs

A BPMN diagram says which steps exist, who performs them and where the flow branches. It cannot be computed: arrival rate, spread, capacity and calendar are not part of the notation. This page covers which BPMN tools actually simulate, what their simulation does — and when you do not need the notation for the number at all.

NotationBPMN 2.0 (OMG)
Simulation dataBPSim (WfMC)
FlowVisualSix blocks, no BPMN
In short

BPMN 2.0 is a notation for the flow, not a computational model. A simulation needs four inputs that appear in no BPMN diagram: how many cases arrive, how much processing durations vary, how much capacity each role has, and which calendar applies. Tools such as Bizagi Modeler, Signavio, ARIS, Trisotech and Cardanit add these in their own dialogues — partly through the WfMC extension BPSim, partly proprietary, depending on product and edition. If you need standards-compliant documentation, model in BPMN. If you only need the number, skipping the notation is faster.

01Foundation

What BPMN contains — and what it does not

The distinction is not pedantry. It explains why an imported diagram cannot be computed at the press of a button, whichever tool opens it.

What BPMN 2.0 describes: activities, events, gateways, sequence flows, pools and lanes, messages, sub-processes. In other words the structure of the flow — completely, in a standardised form, exchangeable between tools.

What BPMN 2.0 does not describe: behaviour under load. Four inputs are missing:

InputExampleWhy it decides the result
Arrival process120 cases/day, peak day 180Without volume there is no utilisation and therefore no constraint
Spread of durationsprocessing 5–15 min instead of "10 min"Queues come from variation, not from averages
Capacity per role3 people × 7 productive hoursDetermines whether a step runs above or below 100 %
Calendar and costMon–Fri 08:00–17:00, €53/h fully loadedTurns waiting time into a number in money

The WfMC extension BPSim (Business Process Simulation Interchange Standard) exists precisely for this: it attaches those parameters — arrivals, distributions, resources, costs, scenarios — to a BPMN file as a separate section. In practice its adoption is limited; most tools store simulation parameters in their own fields, which is why switching tools rarely saves you from re-entering the numbers.

The consequence for your project: importing a BPMN diagram saves you the structure, not the data. And the data is the work.

02Tools

Which BPMN tools actually compute

"Simulation ✓" on a comparison page covers everything from "colours in boxes" to "discrete-event simulator". This ranking is by computational depth, not vendor size — and it is a starting point for your own check, not a replacement for it.

ToolKind of simulationNote
Bizagi ModelerStaged: process validation, time, resource and calendar analysisThe best-known entry point; modelling is freely usable, simulation scope depends on edition
SAP SignavioSimulation module inside the suiteIts strength is the process archive and governance; simulation is one component
Software AG ARISSimulation module, historically deepSensible when ARIS is already in operation
Trisotech, CardanitBrowser-based process simulation, close to BPMNSmaller vendors, often more direct to operate
CamundaNo simulationAn execution engine: it runs real cases through the process, not simulated ones. Frequently confused
bpmn.io / bpmn-jsNo simulationA library for drawing and embedding, no engine
Simulation labs (Arena, Simul8, AnyLogic)Very strong, BPMN usually only as an importCompute deeper than any BPM suite, but the notation is not the model there

Three questions to ask any vendor before committing:

  1. Is simulation included in this edition, or in a more expensive one?
  2. Does it represent queues in front of busy steps — or does it merely add processing times along the paths?
  3. Do I get a result as a range (P10–P90) or as a single number?

Question 2 is where the category splits. A tool that sums path durations calls its result "lead time" but means processing time. In administrative processes the difference is rarely smaller than a factor of three.

03Practice

The sequence that works in practice

Not either/or. The two artefacts have different readers and different lifespans.

The BPMN diagram is for people and systems that must know the flow — quality management, audit, training, software development. It lives for years and is versioned.

The computational model is for one decision — is this automation worth it, does the process survive double the volume, where do we add half a position. It lives for weeks and is thrown away once the decision is made.

The usual sequence:

  1. Document the flow in BPMN where documentation is genuinely required — structure, responsibilities, exceptions.
  2. For the decision, set up a coarse computational model: five to ten steps, not thirty. The constraint almost never sits in the detail.
  3. Enter volumes, ranges, capacities and rates — the work BPMN does not save you.
  4. Compute, save the baseline, change one lever, compare.
  5. Justify the decision with the result and keep the diagram for documentation.

Doing step 2 inside the same tool as step 1 saves a tool switch and buys you pressure to add detail. A documentation-grade diagram has thirty activities; a computable model needs ten — and every additional activity costs you five input fields.

04Position

Why FlowVisual is deliberately not a BPMN tool

Six building blocks instead of a notation. That is a restriction, and it is intentional.

FlowVisual knows start, step, decision, merge, abort and end. No pool, no lane, no message event, no sub-process marker, no notation validation.

What that costs: you will not get standards-compliant process documentation here. If your quality management requires a BPMN 2.0 diagram, you need a BPMN tool — there is no way around it.

What that buys: in the workshop the department argues about the process instead of about the notation. A model stands in ten minutes rather than two hours. And every input field that exists is one that changes the result: volume, range, capacity, role, system, rate.

The honest rule of thumb:

  • You need a document that satisfies a standard or an audit → BPMN tool.
  • You need a number a decision maker will sign → computational model, notation secondary.
  • You need both → both, in sequence and not in the same artefact.

The longer version of this comparison is in the article BPMN tool with simulation.

At a glance
Notation
BPMN 2.0 deliberately unsupported — six building blocks instead of a notation
Import
No BPMN import. The structure is redrawn in ten minutes; the data would have to be entered either way
Engine
Discrete-event simulation with queues in front of busy steps, Monte Carlo repetition
Result
Throughput, lead time, utilisation and cost as P10–P50–P90
Recommendation
BPMN tool for documentation, computational model for the decision

Frequently asked

Can a BPMN diagram be simulated directly?

Not without additional inputs. BPMN describes the structure of the flow, not behaviour under load. Before any run you must supply arrival rate, spread of processing durations, capacity per role and calendar. Tools with a simulation module ask for these in their own dialogues; the standard extension for exchanging them is BPSim, which is not implemented everywhere.

What is BPSim?

The Business Process Simulation Interchange Standard from the WfMC. It attaches simulation parameters — arrivals, distributions, resources, costs, scenarios — to a BPMN file as a separate section so they become exchangeable between tools. The idea is right; adoption is limited, which means switching tools in practice usually means re-entering the numbers.

Is Camunda a simulation tool?

No, and this is one of the most common confusions. Camunda is an execution engine: it runs real cases through a BPMN process and measures what happens. Simulation computes a process that does not exist yet — for example after a planned automation. Both are useful, but they answer different questions.

How many activities should a simulation model have?

Five to ten. A documentation-grade BPMN diagram often has thirty or more, which is harmful for computation because every extra activity demands input fields without changing the result. The constraint is the step with the highest utilisation, and that is just as visible in a coarse model as in a fine one.

Does FlowVisual support BPMN?

No, deliberately not. There are six building blocks and no notation validation. Anyone who needs a standards-compliant BPMN 2.0 diagram for quality management or audit needs a BPMN tool in addition. The benefit of doing without is speed: a computable model stands in ten minutes, and in the workshop people argue about the process rather than the notation.

FlowVisual

The number first, the diagram afterwards

Build the flow coarsely from six blocks, enter volumes and ranges, and let it run. Once the decision is made you will also know which part of the process is worth documenting.

Guide: seven steps to the number