Stochastic Modelling and Simulation consists of modelling a real system in a virtual environment, mimicking its behaviours and interactions. As it is stochastic, it is possible to incorporate the randomness and uncertainty of the real world. Its advantages include flexibility and agility in modifying the modelled system, for example layout, number of resources, procedural decisions, storage policies, or planning rules — commonly referred to as scenarios. Simulation models are white boxes, meaning they allow users to observe, understand and diagnose their behaviours. The definition of metrics enables the comparison of different scenarios and supports informed decision-making.
Traditional analytical tools allow the system’s capacity to be assessed in a static manner. However, reality is dynamic. Simulation has the capability to mimic the dynamics of a real system, both in the intra-process flow and in inter-process interactions. It therefore enables a holistic analysis of the system. Taking this advantage into account, simulation has been used for decades as a technique for diagnosing, evaluating and improving processes. These processes may be industrial, hospital-based, administrative, among others.
Because it is stochastic, simulation is capable of incorporating randomness — an inherent feature of everything around us. This capability can be integrated into several aspects of the simulation model, namely in process decisions (for example process failure or priority allocation) or in the duration of an activity (for example random, dependent or optional time).
Additionally, it is possible to mimic the uncertainty associated with specific processes, enabling system risk analysis. This capability is particularly relevant in critical contexts, providing decision-makers with information about the risk of system unavailability or failure.
By modelling the system in a virtual environment, it becomes possible to modify it flexibly and quickly. Examples include changing the layout, varying the number of resources, altering procedural decisions, evaluating storage policies or testing planning rules. By performing such modifications in a virtual environment, real operations can continue without disruption, risks to worker safety are reduced, and the system can even be tested before implementation.
Each modification may constitute a scenario, enabling direct comparison between them. Scenarios can be compared based on predefined metrics, such as time in system, waiting time, throughput time or resource utilisation. Based on the comparison of these metrics for each scenario, it becomes possible to make an informed decision regarding the scenario best suited to the organisation’s needs.
These simulation models are white boxes, meaning it is possible to observe, understand and diagnose the behaviour of the simulation model. This capability enhances process diagnostic capacity, as it allows each process to be analysed step by step, as well as each inter-process interaction.
With the advent of digitalisation, the requirements for simulation have increased. Upstream, this includes the incorporation of historical and dynamic data (data-driven) and the growing need for flexible modelling (digital shadow). Downstream, it involves exporting the data generated from simulated scenarios to the organisation’s other information systems.
In the current era of the Digital Twin, two key needs are emerging: integration and autonomy. Integration refers to the need for simulation to be connected to the organisation’s digital infrastructure in order to replicate changes occurring in the real system — reducing the need for manual modelling and increasing resilience. Autonomy focuses on the ability of simulation models to autonomously and automatically run scenarios that: (a) support regular operational tasks, (b) search for appropriate alternatives, and (c) anticipate future constraints. Respectively, this means simulating, predicting and prescribing.
With the goal of achieving integration and autonomy, the Stochastic Modelling and Simulation team has been developing solutions based on the use of open-source software for simulation. The benefit of low acquisition costs is an excellent complementary advantage.
Opinion article by Marcelo Henriques, Researcher in Stochastic Modelling and Simulation at the CCG/ZGDV Institute.
