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 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.

methods/flow-analysis · methods/process-mining-basics · concepts/devils-quadrangle