Introduction
Efficient intralogistics is becoming increasingly important in high-mix, low-volume production. In battery pack assembly, the variety of products and small batch sizes make traditional material handling methods difficult to apply. Motivated by the European Green Deal’s push for electrification, the authors partner with VDL Nedcar to compare AGV vs. AMR strategies—Autonomous Mobile Robots (AMRs) for parts delivery and Automated Guided Vehicles (AGVs) for subassembly transport—to streamline assembly, supply, and transport in an Industry 4.0 framework.
Simulation model
To capture the nuanced interplay of AGV vs. AMR systems on a high-mix, low-volume production line, the authors developed a simulation in AnyLogic. They began by mapping the physical layout of the demo line—workstations, buffer areas, and charging stations—and then populating it with agents.
Battery packs move through the assembly line as AMRs retrieve components from storage and AGVs transport completed subassemblies between workstations. Each agent follows its own set of rules:
- Packs queue and wait their turn at stations.
- AMRs pick the best routes for pickups and drop-offs, stopping to recharge as needed.
- AGVs run nonstop, either waiting at stations for loads (Concept I) or heading back right away for the next load (Concept II).
Concept II: AGV leaving the workstation after unloading a product
In the research, three modeling approaches interlock seamlessly for the AGV vs. AMR comparison. Discrete‐event processes orchestrate task sequences and resource contention at workstations. Agent‐based logic endows each vehicle and trolley with autonomy, avoiding collisions and responding to delays. System dynamics equations quietly track battery state of charge, draining energy during travel and replenishing it during charging cycles.
AnyLogic’s native support for this multimethod approach made it the ideal choice: engineers could drag-and-drop flowcharts, script agents’ behaviors in Java, and embed stock-flow diagrams—all in a single environment. The built-in animation visualizes each AGV vs. AMR interaction, while the system automates hundreds of “what-if” trials, varying fleet sizes, recharge policies, and product mixes to pinpoint the optimal intralogistics strategy.
Results
The simulation evaluated both intralogistics concepts across single-product and mixed-product scenarios, varying the AGV vs. AMR fleet sizes:
Concept I (AGV waits at workstation):
- For high-mix, low-volume production, throughput climbs sharply once the AGV fleet exceeds three units.
- Optimal single product setup: 5 AGVs + 3 AMRs → ~15.8 parts/hour (AGV utilization 0.98; AMR utilization 0.50).
Concept II (AGV returns immediately):
- Delivers marginally higher throughput with very small fleets (< 4 AGVs) but suffers declining utilization as fleets grow.
- Recommended lean configuration: 4 AGVs + 3 AMRs for balanced performance.
Mixed products findings:
- Throughput drops to ~11.45 parts/hour with 5 AGVs + 2 AMRs, due to rework loops and station bottlenecks.
- Introducing vehicle-based or dedicated physical buffers and batching similar variants significantly mitigates congestion.
Overall, Concept I excels in larger high-mix, low-volume production setups by keeping AGVs continuously engaged, while Concept II may suit leaner operations where fleet costs must be minimized. The results underscore the importance of strategic buffer management and informed AGV vs. AMR fleet sizing to maximize throughput and utilization.