The completion and validation of its Physical Test Bed for large composite liquid resin infusion processes confirms pre-industrial integration readiness and opens the door to full industrial demonstration – accelerating the project’s mission to enable the sustainable, zero-defect manufacturing of composite parts. 

FLASH-COMP, an EU project Insight Media has had the pleasure of being a part of, was launched three years ago with a clear objective: to enable flawless and sustainable production of composite parts through a human-centred digital approach, integrating advanced sensing, simulation and decision-support technologies into liquid resin infusion (LRI) manufacturing processes.  

Working towards achieving this vision has involved more than developing the digital tools alone. It has also required the building of a validated physical environment where sensing architectures, data pipelines and monitoring logic can be tested under realistic conditions before the FLASH-COMP solution can then be put through its real-world paces under full industrial production constraints. 

Now, with the completion of the project’s Physical Test Bed for LRI of large composite parts, the project has reached a defining milestone in that journey. With consolidated pre-industrial validation successfully achieved, the FLASH-COMP team now has the validated pre-industrial baseline required to move confidently into full industrial-scale deployment in the coming months. 

This key deliverable has now been published, fully documenting the design, implementation and validation of this modular and reconfigurable test bed environment, which was developed at INEGI in Portugal. The paper outlines the tests carried out and how their success under controlled, yet representative, manufacturing conditions, shows that FLASH-COMP is ready to move on to full industrial implementation and demonstration. 

How the test bed works 

The Physical Test Bed provides a modular and reconfigurable monitoring system integrating sensing, data acquisition and local processing that is designed to support the controlled integration and maturation of heterogeneous sensing, data acquisition and decision-support technologies applied to LRI processes. The objective was to validate functional integration, data coherence and system responsiveness under controlled yet representative manufacturing conditions, while remaining explicitly positioned below full industrial deployment. 

The initial implementation relied on a robot-assisted configuration, enabling flexible sensor positioning and controlled scanning trajectories. This high degree of configurational flexibility allowed rapid reconfiguration of sensor layouts and repeatable measurement scenarios, supporting early-stage validation of sensing architectures, data acquisition pipelines and interaction with the project’s digital components. 

As validation activities progressed, the architecture evolved towards a simplified and mechanically stable monitoring rig. By progressively reducing degrees of freedom while preserving validated sensing logic and data-flow principles, this transition created a practical transition from flexible laboratory setups to fixed and repeatable industrial configurations. The rig enforces stable sensor positioning, controlled scanning paths and standardised mounting interfaces, allowing monitoring performance to be assessed under reduced configurational flexibility and conditions closer to real production environments. 

Integrating sensing technologies 

A central achievement of this milestone has been the successful integration of heterogeneous sensing subsystems, which include optical, thermal and process-related sensors, within a synchronised data acquisition architecture. A dedicated local DAQ system ensures time-aligned and coherent collection of multi-source data streams, preserving representativeness of industrial manufacturing conditions in terms of signal timing, data quality and interoperability. 

Preprocessing, simulation coupling and data interpretation tools are executed locally, in close proximity to the physical process. This architectural choice enables timely interpretation of sensor data, reduces dependency on external connectivity and validates data pipelines prior to their full interaction with the FLASH-COMP data space and decision-support services. The result is a validated sensing-to-simulation-to-DSS data flow, from physical measurement through preprocessing and simulation, to structured digital interaction. 

Interoperability 

The test bed’s monitoring environment operates within a clearly defined physical/digital interoperability framework. Sensor data are transmitted via a structured communication layer, stored in a central process database and selectively coupled to real-time simulation components. Outputs are made available to downstream analytical tools, including FLASH-COMP’s Defect Severity Estimation Tool (DSET) and the Decision Support System (DSS), while maintaining traceability to the process map and failure modes and effective analysis (FMEA). 

Within this structure, the test bed acts as the calibrated interface between the manufacturing process and the digital decision-support framework. It ensures that sensor positioning, geometric referencing, synchronisation and contextualisation are coherent and traceable. Interpretation of deviations and the generation of corrective actions remain within the digital layers, preserving a clear separation between physical data generation and digital decision logic. 

This end-to-end flow, from sensing through local processing and simulation to the data space and DSS, was exercised and validated as a fully-integrated, physical-to-digital monitoring framework 

Validation 

Pre-industrial validation included structured vacuum infusion trials on representative composite parts. Qualitative benchmarks compared visual and photographic observations of resin flow progression and surface phenomena with corresponding digital outputs generated by monitoring and simulation tools. This verification confirmed consistency between physical process behaviour and its digital representation at pre-industrial scale. 

Robustness was deliberately assessed through controlled variability. Different part geometries, infusion layouts, sensor configurations and environmental conditions were introduced to evaluate monitoring stability and data quality. Operational challenges, such as surface irregularities, environmental influences like movement and vibration, sensor placement constraints and signal instability, were characterised, and mitigation strategies were defined to reduce technical risk ahead of industrial validation. 

A path to industrial demonstration 

This milestone for FLASH-COMP confirms that sensing, data acquisition and local processing operate together reliably at pre-industrial scale and reduces technical uncertainty ahead of industrial demonstration under real production constraints, which is the work now being carried out at the project’s use case at yacht builder Azimut Benetti, based in Turin, Italy. 

The dedicated monitoring rig now serves as the transitional platform for stress-testing under increased variability, reduced configurational flexibility and production-rate considerations. 

By consolidating sensing integration, synchronised data acquisition and geometry-aware physical/digital interfacing within a controlled but representative LRI context, FLASH-COMP has completed a critical step towards achieving its aims. The monitoring architecture has moved from concept and laboratory flexibility to demonstrated integration readiness, enabling the system to be deployed and stress-tested under real production conditions.