System Modelling and Continuous vs Discrete-Event Dynamics
System modelling keyword is the process of building an interpretable representation of a system so you can analyze, predict, or design it. In engineering practice, a model is typically defined by: (i) inputs (what you control), (ii) state (what must be remembered to predict the future), (iii) dynamics (how the state changes), and (iv) outputs/observations (what you measure). Modelling supports tasks such as simulation, control design, performance analysis, and decision-making.
A useful way to view system modelling is as a mapping:
- Real system model (abstraction)
- Model dynamics predictions about the real system
Common modelling choices differ mainly in the form of time and the mechanism of state change. Two especially important contrasts are continuous-time evolution versus discrete-event evolution.
Key terms: System model, State, Abstraction, Dynamics.
Discrete-Event Simulation (DES) Fundamentals
System modelling: what you specify
In most modelling workflows you decide what to keep and how to represent time and events. A minimal generic structure is:
- Variables: determine which quantities are state vs inputs vs outputs.
- Time representation: continuous time, discrete time, or event-driven time.
- Update rule: differential/recurrence relations, or event-triggered transitions.
- Constraints/assumptions: what is ignored, what approximations are acceptable.
This is why two models of the same real system can yield different forms—because the modelling objective (e.g., control vs throughput analysis) changes the level of abstraction.
Key terms: [Input]{def="Signals/decisions that drive the model}, Output, Transition rule, Assumption.
Pro Tip: If your model needs “everything between two observations,” choose continuous or discrete-time dynamics. If your model only needs “what happens when something occurs,” choose discrete-event modelling.
Differences between continuous and discrete-event systems
Below are core conceptual differences. “Continuous” here refers to continuous-time evolution (state changes as time flows). “Discrete-event” refers to evolution only when events occur (state changes occur at isolated time instants).
Side-by-side comparison
| Aspect | Continuous-time system | Discrete-event system |
|---|---|---|
| Time basis | State evolves for all | State changes only at event times |
| When updates happen | Every moment (via differential dynamics) | Only when an event occurs |
| Typical math form | Differential equations, integrators | Event scheduling + state transitions |
| State change mechanism | Usually smooth/continuous (unless modeled with impulses) | Instantaneous jumps at event times |
| Sources of discontinuity | Modelled explicitly (e.g., impacts, switch logic) | Naturally produced by events (arrivals, service completions) |
| Example | Mass-spring-damper, RC circuit | Queueing network, manufacturing processes with jobs flowing through stations |
Key terms: Event time, Event, Continuous dynamics, Jump discontinuity.
How to decide whether your model should be continuous-time or discrete-event
- 1Step 1
Ask whether the system’s key changes happen continuously (physics, flows) or at distinct occurrences (arrivals, completions).
- 2Step 2
If you need behavior at arbitrary intermediate times, continuous-time is natural. If you only care about ordering/throughput at event instants, discrete-event is natural.
- 3Step 3
If state changes only when conditions like 'job arrives' or 'server becomes free' occur, treat those as events.
- 4Step 4
Continuous: integrate dynamics in time. Discrete-event: maintain a future event list and advance time to the next event.
- 5Step 5
Ensure the abstraction preserves the metrics you care about (e.g., stability, accuracy over time vs queue length, wait time distribution).
A practical mental model
Continuous-time systems are like watching the system evolve on a stopwatch tick-by-tick (resolution in time). Discrete-event systems are like watching a ledger of occurrences—time jumps from one occurrence to the next.
Common modelling pitfall
Don’t call something 'discrete-event' just because it is simulated with computer time steps. If the state can change only when events occur, DES uses an event calendar; otherwise you are effectively doing discrete-time updates.
Time evolution in continuous vs discrete-event modelling
All moments
Continuous modelState evolves continuously according to continuous dynamics."
Between events (if any)
Instant viewIn DES, if no event happens, the state remains unchanged."
Discrete-event model
Event momentsState jumps at event times; the simulator advances to the next event."
Typical examples (to anchor intuition)
- Continuous-time examples: electrical circuits governed by differential equations; mechanical systems approximated by Newton’s laws; chemical concentration evolution modeled continuously over time.
- Discrete-event examples: queueing systems (arrivals/services), network packet switching (packet arrivals/departures), batch manufacturing workflows (start/end of processing steps).
In DES, the simulator usually tracks a set of scheduled future events and repeatedly:
- selects the next imminent event,
- advances the model clock to that event time,
- updates state according to the event,
- schedules any resulting future events.
Key terms: Event calendar, Simulation clock, Event scheduling.
Conceptual differences: continuous vs discrete-event
Relative emphasis for learning/analysis tasks (qualitative scale).
Quick FAQs
Core Terms for System Modelling
Knowledge Check
In a discrete-event model, the system state changes primarily when: