What is DES?
Discrete Event Simulation (DES) is a modeling approach that focuses on the occurrence of actual events in a simulation.
Discrete Event Simulation differs from Continuous Simulation (CS) in that the state of change is not measured through the passage of time, but is event-driven. This type of modeling focuses on ‘discrete’ – which is to say separate/instantaneous) events – rather than ‘continuous’ (gradual) events. Time is abstracted away and individual relevant ‘events’ that occur within a system are brought to the fore.
No changes occur between events, so discrete event models can jump straight to the occurrence time of the next event in a simulation, without consequence, and long time intervals can be vastly reduced when simulated.
The concept revolves around the occurrence of ‘state changes’ or phase transitions, rather than the passing of time or continuous movements. This is known as ‘next-event time progression’, as opposed to the ‘fixed-increment time progression’ used in CS.
Further, the developer does not need to simulate unnecessary aspects of the model — only things that are simulated are classed as events, thereby reducing complication.
Discrete event: these occur instantaneously, as opposed to gradually. Think of the abrupt (‘discrete’) state change of a water balloon popping, as opposed to the slower (‘continuous’) deflation of a helium balloon.
Event-centric: a focus on the occurrence of actual events in a sequence (where no changes happen in the interval between events), and time is not considered an important factor.
Nondeterministic: models which contain stochastic (random) elements may output varying results from run-to-run, even if the same set of starting parameters/instructions are provided. DESs can either be deterministic or nondeterministic.
DES can be considered more of a flowchart than a timetable of events, and can help users separate the signal from the noise — by reducing large temporal transitions into an events-oriented ‘schedule’, we can abstract aspects of the model and simplify complex scenarios for the user.
The concept of time is not tied as tightly to the simulation logic — the user can add a time element, but this is mostly cosmetic/auxiliary.
DES, like other simulation approaches, can include deterministic or stochastic elements. For instance, if you want to model a range of scenarios, you could draw initial parameters from a probability distribution, simulating potential conditions for the simulation. This can help approximate aspects of the model that we might be uncertain about.
How is DES used?
A common use of DES is in inventory management for FMCG (Fast-Moving Consumer Goods). No matter the amount of time lapsed between outgoing orders, stock must be replenished for the warehouse to maintain its inventory obligations (even as consumer demand might episodically ebb and flow).
DES can demonstrate the logistics of these reorder liabilities, and allow managers to review current processes in light of the simulations’ results. They can simulate potential stock picking methodologies, such as one order, batch, or zone picking, and then find the most efficient way of fulfilling multi-SKU (Stock-Keeping Unit) obligations to reduce lead time and replenish continually-diminishing inventory.