Authors: Treesa Rose Joseph, Software Developers, SINTEF Ocean; Sridevi Krishnamurthi, Research Scientist, SINTEF Industry; Sylvain Gouttebroze, Senior Research Scientist, SINTEF Industry

Battery pilot lines generate large volumes of machine data every second. However, these values gain meaning only when combined with material information, such as recipes, batch identifiers, and preparation steps. When these datasets are connected, the complete process chain becomes transparent. This include information on what was produced, how was the process, and which factors contributed to the battery performance.

Material Preparation and Pilot Line Processing – how Data is created

In practice, battery manufacturing spans both manual laboratory work and automated pilot line processing. In the early stages, materials are weighted, mixed, and characterised through workflows that often involve manual steps and offline measurements.

After preparation, materials are transferred to the pilot line, where processing happens through automated equipment such as coaters, dryers, and calendaring machines. These systems operate with fixed setpoints and dense sensor systems. Subsequently, these systems are continuously recording parameters such as temperature, pressure, speed, and flow rate. Data exchange typically occurs through OPC UA based networks, and the resulting time-series data is stored in systems optimised for equipment monitoring and process control. As a result, pilot line trials generate large volumes of detailed process data.

Figure1: Dataspace components

Limitations of Machine-Only Data

Machine data captures detailed process conditions, but it does not fully explain variations in trial outcomes. As a result, two runs may appear similar from a process parameter perspective, while producing different results due to differences introduced earlier during material preparation. Without direct access to material related information, interpreting these discrepancies becomes difficult, and the root causes often remain unclear.

Importance of Material Context

Reliable interpretation of pilot line results requires complete material provenance. This includes where the material originated, how it was formulated, which preparation steps were applied, and which recipe was used. Without this information, trial results are difficult to compare, and root cause analysis becomes unreliable.

In many pilot line setups, material preparation data and machine data are stored in separate systems. Laboratory workflows focus on flexibility and experimentation, while pilot line systems focus on automation and data collection. As a result, the connection between a trial and the exact material that was processed is often weak/missing or not machine-readable/accessible.

BATMACHINE and the Role of Kadi4Mat

Figure2: Overview of the BATMACHINE Dashboard showing the header tabs for KADI (material management)m Trials (trial and equipment data), and Dashboards (link to Grafana for Real-Time sensor visualisation).

BATMACHINE addresses this gap by developing an integrated Data Architecture for battery pilot lines. As described in a previous article, BATMACHINE provides a shared dataspace (dashboard shown in Figure 2) that aggregates different data. Specifically, it analyses time-series machine data, semantic knowledge graphs, visualisation tools and process control interfaces. This, in turn, enables consistent access to data across the pilot line.

Figure3: Kadi4Mat Interface displaying recipe and material quantity information

Within this architecture shown in Figure1, Kadi4Mat (Figure 3) is used as the system responsible for material management. It is a research data management tool used to track material provenance.  This way, it is possible to provide the material metadata link that was missing earlier. In BATMACHINE, each pilot line trial connects to a KADI4MAT recipe and material batch, ensuring that machine data is directly traceable to the underlying material preparation and process context. This integration ensures that analyses can account for both upstream material variations and downstream process behaviour, providing a coherent view of the full process.

The Importance of Data Integration

Linking pilot line trials to their corresponding recipes and material batches enables accurate comparison of experiments based on real material variations. In fact, it clarifies how preparation steps influence both the process and final cell performance. Supporting faster optimisation, better reproducibility, and a smoother transition from pilot scale to production using a data-driven approach, battery manufacturing can improve its sustainability, material usage and energy consumption.