How SuNing’s Digital Twin Boosts Warehouse Efficiency by 15%

SuNing’s AI R&D Center partnered with the SuNing Logistics Institute to build a digital twin platform that maps the physical logistics system to a virtual counterpart, automatically generates simulation models based on optimization needs, runs experiments, and feeds the optimized results back to the physical system, forming a closed‑loop.

The platform addresses the challenge of improving warehouse efficiency across 24 national logistics centers and 60 regional hubs by using system simulation to evaluate many strategy combinations without disrupting live operations.

Unlike traditional manual simulation, the digital twin automatically creates digital models from collected data, reflects both current and historical states, and reduces the labor‑intensive modeling and maintenance costs.

Overall Architecture

Each day a snapshot of the physical system is captured; when an optimization request arises, a strategy optimizer selects algorithms from a library, combines them with the snapshot, and a simulation model generator creates the corresponding model for the simulator to run experiments.

The optimizer iterates based on simulation outcomes until a satisfactory strategy is found, then the optimal policies are delivered to operators and the parameters are integrated into the production system.

Implementation in Large‑Item Warehousing

Mapping from the production system to the digital twin includes three aspects:

Space mapping : manual entry via a visual interface to map warehouse layout (see image).

Process mapping : abstraction of inbound, picking, and relocation processes into reusable components for the model generator.

Data mapping : synchronization of inventory status, tasks, and resources from the WMS and big‑data platform.

The platform integrates common warehouse strategies—product classification, slot allocation, order batching, replenishment, etc.—allowing users to manually select policies or let optimization algorithms automatically choose and tune parameters.

Performance metrics such as total operation time, travel distance, task‑specific durations, and equipment utilization are evaluated through simulation, enabling data‑driven optimization.

Visualization of the physical system (e.g., ABC classification of items per slot and aisle congestion) helps managers quickly identify bottlenecks.

Results and Insights

In the Nanjing large‑item logistics center, the recommended strategy increased overall operation efficiency by more than 15% while reducing resource consumption, demonstrating the platform’s ability to deliver cost‑effective improvements.

Key takeaways include the realization that a perfect real‑time digital twin is not always necessary; a daily snapshot can suffice, and IoT/5G are optional rather than mandatory for many optimization scenarios.

The digital twin lowers the barrier for simulation‑based optimization, making it accessible beyond specialist engineers.

Future development will see digital twins further integrated with AI, industrial IoT, and VR, expanding their impact across production and daily life.