Skip to main content

Celonis Product Documentation

Process Simulation

Overview

Process Simulation allows you to measure the impact of potential changes on your processes and the effect they would have on your operations. You can simulate any part of the process and, thereby, assess the impact of decisions beforehand.

Implementing process changes can be very costly and resource-intensive while organizations cannot know in advance whether all changes to a process will increase their execution capacity. For example, removing a bottleneck in one step of the process might create a bottleneck in another step of the process. Simulation is very powerful in general when it comes to assessing the combined impact of several changes in the process.

Process Simulation works in three steps:

  1. Digital Twin Extraction. A Digital Twin is extracted by the user based on the actual process data. It is used as a baseline model for comparing the scenarios (i.e. to-be processes of step 2). The Digital Twin is a graphical model of the as-is process. It contains a BPMN model extended with various statistical parameters about the process. Those parameters describe the incoming cases (e.g. how new cases arrive), the resources working on the process, the activities (their processing times, automation rates), and the connections between those activities (e.g. branching probabilities). The Digital Twin extraction is done inside the Studio. Once created, the Digital Twin needs to be published in Studio before we can proceed to step 2.

  2. Scenario Creation. The user creates one (or more) Scenario(s) that will be simulated. Each scenario uses the Digital Twin as a starting point. For example, the user can decide on automating a manual activity in the Digital Twin. The user can then run simulations for the Scenarios and Digital Twin on a given duration to visualize the impact on the KPIs. Simulation Scenarios are created from the Simulation Dashboard in Apps.

  3. Results Comparison View. This view compares the KPIs of the Scenarios to the Digital Twin. Reports can be generated from the comparison of results.

Technical Description: How it works

Digital Twin Extraction

  • A parametric simulation model is needed to run a simulation

  • To model the as-is process, we calculate the parameter distributions of the as-is data and fit parametric distributions to it

  • Some parameters can’t be extracted from the data and need to be provided manually (e.g. number of FTE)

  • This results in the Digital Twin model, containing Inter-arrival Times, Processing Times, Resource configuration and more

  • After extraction, the parameters can be changed to create what-if scenarios

Scenario Simulation

ProcessSimulation.png
  1. Cases are generated based on a clock and the Inter-arrival Times distribution for each day and start to move through the process

  2. A case gets processed by a resource of Resource Pool A until its Processing Time is over

  3. The Branching Probability of a case following the path between Activity A and Activity B is 20% and 80% for the path between Activity A and Activity C

  4. A case is passing between two Activities for the duration of its Enabling Time

  5. In Resource Pool B, two resources are available in total. One resource is currently available/idle to process a case of the assigned Activities (only Activity B in this case). It is idle because in Activity B no case is waiting in the queue. The second resource is currently processing a case in Activity B

  6. A case is waiting in the queue of Activity D for as long as no resource is available in Resource Pool A. The two FTE of the pool are available based on their working hours and schedule but are currently occupied

  7. Every time a case moves forward through the process, its simulated Case ID, Activity and Timestamp are recorded in an event log until the simulation duration (clock) finishes

Results Comparison

  • The result of a simulation is an eventlog (not stored) which is used to calculate a series of KPIs

    • Throughput Time distribution

    • Cost distribution (based on Processing Time & Resource Cost)

    • Workload (percentage of time a Resource is occupied)

    • Average Throughput Time per Activity

    • ... and more

  • Results can be compared side-by-side for easy comparison of process changes and their effect on the relevant Process KPIs

Tip

To learn more about Celonis Process Simulation, see the free online training from Celonis Process Simulation (1h).