---
title: Process Simulation
type: method
tags: [bpm, quantitative-analysis, simulation, queueing]
sources: ["[[sources/2018-dumas-fundamentals-of-bpm]]"]
created: 2026-04-13
updated: 2026-04-13
---

# Process Simulation

Discrete-event simulation of a process model to estimate performance under stochastic arrivals, probabilistic branching, and resource contention — when [[methods/flow-analysis]] is too restrictive ([[sources/2018-dumas-fundamentals-of-bpm]], §7.3).

## Inputs (§7.3.2)
- **Process model** with probabilities on XOR branches.
- **Inter-arrival time distribution** (e.g., exponential, empirical).
- **Activity duration distributions**.
- **Resources** — counts, skill-roles, calendars.
- **Warm-up + run length + replications**.

## Anatomy (§7.3.1)
- Event list (next-event time-advance).
- Entity/case generator.
- Resource pools with queues.
- Collection of statistics (cycle time, waiting time, utilisation).

## Queueing theory as a lightweight alternative (§7.2)
For analytically tractable cases:
- **M/M/1** — single server, exponential arrivals & service.
- **M/M/c** — c parallel servers, common queue.
- Little's Law applies.

## Tools (§7.3.3)
Bizagi Modeler, Signavio, BIMP (formerly BIMP.cloud), ARIS — plus research tools like ProM simulation extensions. For **PPM-evaluation use cases**, see also [[sources/2024-riess-synbps-simulation-framework|SynBPS]] (Riess 2024): a parametric Python framework based on Markov chains that prioritises *controlled-variable hypothesis testing* over calibration to real logs.

## Cautions (§7.3.4)
- Garbage-in-garbage-out: duration distributions must be grounded in data.
- Validate against reality before trusting.
- Don't simulate the wrong question.

## Related
[[methods/flow-analysis]] · [[methods/process-mining-basics]] · [[concepts/devils-quadrangle]]