Discrete Event Simulation (DES) models a warehouse as a sequence of timed, stochastic events - a pallet arrives, a conveyor segment transfers it, an AGV picks it up - and runs the system forward in simulated time to measure throughput, utilization, and queue behavior under realistic variability.
DES is the gold standard for validating complex automated system designs before capital commitment. It catches bottlenecks and undersizing that analytical models miss because it models variability, not just averages.
When to Use DES
DES is warranted when:
- Project value exceeds $5M (model cost is 0.5–1% of project cost)
- System has multiple interacting automated subsystems (AS/RS + conveyor + AMR)
- Order profile has high variability (CV > 0.3 on hourly arrival rate)
- Sizing results from analytical models are borderline (within 10% of capacity limit)
- Contract requires performance guarantees at FAT/SAT (model provides defensible basis)
DES is usually not warranted for simple manual operations, single-system installations below $3M, or early-stage concept screening (use analytical models for that - see Throughput Analysis).
Key Inputs
| Input Category | Specific Data Required |
|---|---|
| Order profile | Hourly order/line/unit arrivals, SKU velocity distribution, peak-to-average ratio |
| Product characteristics | Case dimensions, weights, fragility (affects conveyor speed, jam rate) |
| Equipment specs | Speeds, accelerations, cycle times, throughput rates per vendor datasheet |
| Layout | Travel distances, aisle lengths, conveyor path lengths, buffer locations and sizes |
| Failure / maintenance | MTBF and MTTR by equipment type (see Reliability and Design Safety Factors) |
| Staffing | Number of operators, break schedules, task times for manual touches |
Input quality drives output quality. A model built on ±20% equipment specs produces ±20% throughput predictions. Nail down vendor-confirmed cycle times before building.
Key Outputs
| Output | Use |
|---|---|
| Peak throughput (orders/hr, units/hr) | Validates capacity against design-day requirement |
| Equipment utilization % | Identifies bottlenecks (>85% utilization = constraint) |
| Queue lengths and dwell times | Sizes accumulation conveyors and buffer lanes |
| System throughput under failure | Quantifies redundancy value; validates bypass paths |
| Time-to-target analysis | Confirms system reaches steady state within shift start window |
Tool Landscape
| Tool | Strength | Typical User |
|---|---|---|
| FlexSim | Intuitive 3D, strong conveyor libraries | Integrators, consultants |
| Simio | Object-oriented, strong scheduling integration | Academic, advanced users |
| AnyLogic | Most flexible (Java-based), pedestrian/agent models | Research, complex systems |
| DELMIA (Dassault) | Deep manufacturing integration, digital twin path | Automotive, large manufacturers |
| AutoMod | Legacy tool, still used for high-fidelity conveyor modeling | Established integrators |
Most North American integrators use FlexSim or a proprietary DES tool built on one of these platforms.
Model Build Process
| Phase | Duration | Output |
|---|---|---|
| Data collection & validation | 1–2 weeks | Verified input dataset |
| Model construction | 2–3 weeks | Calibrated base model |
| Verification & validation | 1 week | Model vs. hand-calc comparison |
| Scenario runs | 1 week | Throughput curves, sensitivity results |
| Total | 5–7 weeks | Final report + model file |
Budget 6–8 weeks for a full DC model. Scope creep (adding subsystems mid-build) is the primary schedule risk.
Warm-Up and Run Length
A DES model starts empty. Results during the “warm-up period” - while the system fills to steady state - are discarded. For a DC model, warm-up is typically 1–2 simulated shifts. Run length for statistically valid results: 10–20 replications of a full operating week, or one sufficiently long run using the method of batch means.
Sensitivity Analysis
Standard scenarios to run:
| Scenario | Purpose |
|---|---|
| +20% / −20% volume | Capacity headroom and low-volume behavior |
| Single-point equipment failure (MTBF = actual) | Validates bypass routing and buffer sizing |
| Peak surge (2× design-day for 2 hours) | Confirms no permanent queue buildup |
| Reduced staffing (−1 operator) | Labor sensitivity for automated-manual hybrid zones |
DES vs. Analytical Models
| Factor | Analytical | DES |
|---|---|---|
| Build time | Hours | Weeks |
| Variability modeled | No (uses averages) | Yes (stochastic) |
| Bottleneck identification | Approximate | Precise |
| Cost | Minimal | $50K–$150K |
| Use case | Concept, early sizing | Final design validation |
Use analytical models (see AS-RS Sizing Methods, AMR Fleet Sizing) to establish initial sizing. Use DES to validate before contract signing.