How to successfully build a business process simulation model with BPSim

The digital twin of an enterprise is a very popular topic and a not so far utopia. This post focuses on the main pillar of this practice: Business Process Simulation. Cardanit has a recipe to help you succeed in building a reliable simulation model thanks to science-backed algorithms and expert consultants.

Why do you need Business Process Simulation?

The traditional Business Process Management cycle suggests to design, map, execute, monitor and optimize business processes in an iterative way. It’s usually possible to incorporate several small improvements over time with relative ease. However, major changes generate friction and indeed Change Management is a separate and dignified discipline.

Before revolutionizing an established and working process, you may want to see proof of your return on investment. Trial and error experimentations are risky and expensive. Unfortunately, there is no silver bullet but we believe the technology is mature: all the ingredients to scientifically and successfully handle process optimization are ready to be connected and exploited.

Digital twins from engineering to BPM

As you may know, Cardanit is a product of ESTECO. Our company has more than 20 years of experience in managing the complexity of engineering digital twins. It’s a natural choice to follow the same approach with Cardanit and Business Process Management. Simulation and optimization are among our core competencies and we’re going to make the best out of them in the BPM world, too.

Since we developed some strategic partnerships, we’re not alone in this. We chose the Lanner L-SIM software as the basis for the process simulation capability. A solid scientific foundation is paramount to us and that's why we collaborate with prof. Massimiliano de Leoni from the University of Padua (Italy), a major expert in Process Mining and related fields. With the aim of bringing our competencies and researches into real-world scenarios, we also partnered with Business Consultants like the Italian agency QUIN.

Business Process Simulation with BPSim

The Workflow Management Coalition, with a great contribution from Lanner among others, developed a standard for this discipline: BPSim. With BPSim you can simulate different perspectives of a process and understand its important features:

• The control flow - Are there any deadlocks? Is it possible to reach all the tasks? What is the most probable end event?
• Time and cost analysis - What is the ideal processing time, if all tasks start as soon as possible? How much does the process cost on average?
• Resource analysis - If multiple instances of the process rely on the same resources, are there any bottlenecks? Where are the most serious ones?
• Calendar analysis - How do work shifts impact the process?

Business Process Simulation is not new in the BPM field, as the first release dates back to 2013. Today, simulation is a well-established technique and it’s an essential ingredient of the digital twins and of the Industry 4.0. However, its adoption seems slow. At ESTECO we experienced a similar scenario within the automotive industry twenty years ago. In the BPM market today, RPA (Robotic Process Automation) and Process Discovery seem to receive more attention, perhaps because they seem easier to adopt - but are they? There are no shortcuts when creating simulation models and you must have the right support to produce reliable ones.

Lanner's L-Sim simulation engine can handle many scenarios, huge models, and the complex interactions between elements. Cardanit is going to implement the best possible user experience to facilitate Business Analysts' work, without hiding the complexity of the task. A simulation model is made of two parts: the BPMN process flow and the BPSim parametrization (for example task durations, gateways probabilities, resource assignments, calendars). The goal is to make this model accurate and reliable. Sometimes you need to iterate between modeling and simulation in order to calibrate the overall model. The standard way to do this is by trial and error. But we’re working on a new approach.

Process Mining and Process Simulation together

We want to exploit Process Mining techniques to speed up the model calibration. We're researching this with Prof. de Leoni, and we strongly believe that it's possible to build a first tentative model in an automatic way and then to assist analysts with science-backed techniques.

Process Mining algorithms proved their ability to discover process models from event logs. We’re extending those algorithms to extract also simulation parameters like task duration, resource allocation and work shift calendars. And we don’t think that Process Mining can replace the work of business analysts. We believe that they complement each other. We found it out directly by experience when working with QUIN on a bank process for a loan request from both physical and online channels.

Both the work with prof. de Leoni’s research team and the one with QUIN have been presented at well-attended conferences last year because we want our approach to be scientifically validated. If you are interested, you can see the videos of the presentations in the proceedings of the 19th International Conference on Business Process Management (BPM 2021, Rome) and of the International CAE Conference (Vicenza, Italy). The former focuses on BPSimpy, a library to programmatically create a BPSim model. It’s the foundation of the communication between Process Mining algorithms with the L-Sim simulation engine. The latter shows that Process Mining and Process Simulation can open the door to mathematical optimization in the BPM field.

Data, algorithms and human knowledge

We’re going ahead with both research and development. Data, algorithms and human knowledge must interact and collaborate; we would like Cardanit to be the shared repository for this collaboration and the conductor of this orchestration. If you are interested in becoming a partner, or in developing a proof of concept, you can contact us.

Further reading

BPMN for ESTECO technology: Improving engineering design projects

Six ingredients for effective business process mapping