One of the biggest drawbacks of liquid-resin-based manufacturing processes is the difficulty of intervening in the infusion process once resin has started flowing inside the mould. Consequently, if fibre laminates are not properly placed, their quality is not optimal or resin does not flow as expected, defects appear.
These flaws can be reduced by introducing excess resin, but this produces a lot of waste and requires a tedious and expensive quality control loop. As the problem relates to the behaviour of the resin flow, it may be possible to reduce waste and quality control loop associated costs by establishing corrective actions over the resin infusion stage. But to do this, controlling both the resin infusion process as well as the pre-forming stage is critical.
Creating a zero-defect manufacturing system
FLASH-COMP will develop a decision support system (FLASH-DSS) based on a smart defect severity estimation tool (FLASH-DSET) that can identify defectiveness at an early stage by means of novel inspection and monitoring devices (FLASH-IM). This strategy will allow corrective actions to be adopted and consequently optimise the manufacturing process towards zero-defects and reducing the generation of waste.
The corrective strategies will be approached in two ways: FeedForward Control, which refers to the application of corrective actions prior to infusion process, and FeedBack Control, which focuses on correcting the process ‘on-the-fly’ once the infusion process has begun.
Predicting defects and defining solutions in real-time
The application of corrective measures to modify the behaviour of a liquid is not trivial, as it is ruled by complex fluid mechanic physics. FLASH-COMP will ground the FLASH-DSET in model order reduction simulation techniques, which will allow for similar accuracy as physics-based numerical simulations on high-performance computers, but almost in real time.
However, as the objective is to optimise the process in-line, it is necessary to consider potential discrepancies between the predictions of simulations and the actual measurements obtained by inspection and monitoring instruments. Therefore, “hybrid twin” technology will be used to build a deviation model that will describe such discrepancies, allowing input variables simulated off-line to be replaced by the measurements of the inspection and monitoring instruments. A hybrid-twin-based FLASH-DSET capable of predicting defects and determining if corrective actions should be adopted will be built.
Intelligent vision and sensor systems
For the inspection of the pre-forming stage, the systems proposed in the project are based on a combination of visible, spectral and geometrical linear cameras and profilers. Besides from obvious on-sight defects – easily identifiable with smart vision and point analysis techniques – some flaws related to raw material quality are only trackable by means of hyper- or multi-spectral cameras. But these instruments are expensive, bulky and hard to industrialise. FLASH-COMP therefore intends to integrate spectrally-sensitive features to affordable linear cameras.
For the monitoring of the infusion stage and to avoid current problems with 2D images, FLASH-COMP will create a novel 3D vision element –FLASH-LiDAR– capable of quantifying the resin volumetric rate in real-time. As it is not possible to properly visualise the flow front in some LRI processes, the use of mould and interlayer level embedded sensors will be tested as well, increasing the amount of relevant information captured and extrapolated from the resin flow behaviour.
The proposed inspection and monitoring techniques (FLASH-IM) will use AI models with relevant process data to train the defect severity estimation tool (FLASH-DSET). Consequently, FLASH-DSET will be able to suggest instant process modifications that will be materialised as automatic or manual control decisions through the decision support system.
One clear message from the FLASH-COMP project is that the solutions are not designed to replace people – they are designed to help people do their jobs more effectively.
FLASH-COMP is creating a decision support tool that will help workers to control the manufacturing process by either intervening directly or providing suggestions on how to proceed to guarantee the correct progress of the process. It will therefore be essential to put proper consideration into how this interaction between the decision support tool and the person involved works, to ensure that the FLASH-COMP solution is both accepted and valued by those who use it.
FLASH-COMP will integrate the needs and requirements of the users in the technical development of the FLASH-COMP solution, and use their suggestions about how to make the solutions more user-friendly and useful. It will also look to understand the impact of the new solution on the current work environment, identifying learning objectives to help current and future workers cope with the technological, process and cultural change that may come with the new solution.
Knowledge and methodologies from social sciences and humanities (SSH) will play a key role in this aspect of the FLASH-COMP project, with the aim of involving the end-users and workers from the beginning to shape the solution and to properly consider the learning competencies required to assure an effective integration of the FLASH-COMP solution.
Several interviews with end users will take place at the beginning of the project, as well as observations of the different tasks they perform daily. Through an iterative process, the user’s perception of the technologies will be assessed and considered as the development of the solution progresses. This will be done through focus groups that include participants from a wide range of staff positions, from plant operators to management.
The FLASH-COMP project is divided up into nine work packages. Find out more about what goes on in each work package below.