Summary
Key results
Digital twin of 3 production lines, covering 50+ stations and process variables
Ability to assess the impact of process changes and investments before implementation on the production floor
No consistent model for assessing changes in battery production
Battery production involved numerous interdependent elements – from assembly and intralogistics to workforce availability, machine failures, quality inspections, batch sizes, and buffer capacity constraints. A change to one parameter could affect the performance of subsequent production areas, so analyzing individual factors in isolation did not provide a complete view of the process.
The client lacked a consistent model or tool for structuring project data, mapping the entire process, and analyzing dependencies between its individual elements. The large volume of information made it difficult to identify bottlenecks and predict the impact of planned changes.
The organization needed a solution that would allow it to compare scenarios, evaluate investments, and adjust the process and associated costs to current needs without testing every change on a live production line. The client therefore entrusted Sii Poland with developing a digital twin of the production process.
Digital twin of the entire battery assembly process
A 2-person team of Sii Poland experts developed a digital twin in Siemens Plant Simulation within one quarter. The model replicated the entire battery assembly process across 3 production lines and more than 50 workstations, including parts supply, intralogistics, product collection, and dependencies between individual production areas.
Input data was loaded manually from Excel files generated by the manufacturing execution system. The baseline model reflected the current production process at the plant and was calibrated using historical data. Sii validated it step by step across successive production areas, first ensuring an accurate representation of the existing process and then enabling modifications to be tested without disrupting live production.
The scope of work included:
- Mapping cycle, changeover, and transportation times, batch sizes, work in progress, and buffer levels
- Incorporating intralogistics, warehouses, queues, workstation supply rules, and logistics constraints
- Accounting for workstation staffing, shift patterns, employee skills, and workforce movement
- Modeling machine failures based on frequencies derived from historical data as well as inspections affecting process performance
- Building an integrated analytics layer showing metrics including throughput, workstation utilization, lead times, work in progress, and queues
The model enabled the client to compare different scenarios for work organization, batch sizes, staffing, and buffer capacity. Sii supported the client in running approximately 20 initial simulations, then trained its team and prepared it to independently analyze further scenarios.
Better-informed investment decisions through process simulation
The digital twin gave the client a consistent view of the entire process and enabled changes to be assessed before implementation. The organization can compare scenarios involving staffing, work schedules, batch sizes, and buffer capacity, analyzing their combined impact on production performance.
The simulations identified workstations constraining throughput and logistics buffers with insufficient capacity. They also helped determine the optimal production batch size. As a result, the client can target optimization efforts and investments at the areas where changes are actually needed, rather than assessing their impact only after implementation on the production floor.
The reliability of the analyses was confirmed by comparing model results with actual production after the analyzed changes had been implemented. The deviation was approximately 1% compared with the number of units produced over a 24-hour period. After jointly analyzing around 20 initial scenarios and completing training, the client’s team was prepared to work with the model independently.
Key results
- Approximately 1% model deviation from actual production output over a 24-hour period
- Ability to assess process changes and investments before implementation
- Identification of throughput-constraining workstations and logistics buffers with insufficient capacity
- Determination of the optimal production batch size
- Client team prepared to independently analyze further scenarios