Optimization of the Infusion Process through Numerical Permeability Analysis and Simulation
Topic(s) : Manufacturing
Co-authors :
Nihad SIDDIG (FRANCE), Paris Dilip MULYE (FRANCE), Yves LE GUENNEC (FRANCE), Philippe LE BOT (FRANCE), Damien LECOINTE , Yvan DENIS (FRANCE), Elena SYERKO (FRANCE), Olivier FOUCHÉ (FRANCE), Christophe BINETRUY (FRANCE)Abstract :
The liquid resin infusion (LRI) process is continuously used to manufacture large-sized industrial parts for a wide range of applications. Although simple, this process yields several variabilities and challenges. Online process monitoring has been used to identify process disturbances allowing the detection of dry spots and reducing the number of parts rejected as scrap [1]. The research project MONOCLE, led by the Jules Verne French Institute in collaboration with a consortium of key industrial partners (Naval Group, SICOMIN, PREDICT, BUREAU VERITAS, and PCMI) aims to develop a digital approach for monitoring the infusion process for complex parts. The ultimate goal is to provide operators with an effective decision-making tool during the process.
The optimization of the infusion process requires the use of simulations to save the considerable time and effort associated with experiments. To be trustworthy, these numerical methods must be validated to ensure accurate predictions of potential defects, especially in thick and complex parts [2]. Material characterization is required for process modeling and simulation. Permeability characterization is vital for modeling the flow front and the pressure evolution in fibrous media, facilitating the prediction of residual defects. The main challenge lies in the prediction of the flow behavior under variable process conditions. This work aims to establish a physics-based model for the infusion of thick parts, apply a numerical approach for permeability characterization as input for the simulations, validate the numerical results with experiments through the monitoring system and then use the model to predict the process behavior for different scenarios to assist the decision-making process.
In this study, infusion experiments were carried out on thick components of 25 mm using a test bench equipped with a robust monitoring system. Sensors were placed throughout the part to monitor and control various parameters during the infusion. All preform permeability components are obtained virtually using the image-based flow and permeability software suite PoroS [3], which quantifies the ability of porous media to transmit fluids at micro- and/or meso-scale by solving single and/or dual-scale flow problems (Fig. 1). This cloud-based software allowed to predict the full permeability tensor based on a 3D segmented image of the thick stack of fabrics obtained from µCT scans.
The permeability values were used as input for the process simulations, which were carried out with the commercial software PAM-RTM® from ESI group. The results were compared and validated against the experiment (Fig. 2). Subsequently, the validated model provided a basis for the numerical prediction of multiple scenarios with variable material and process parameters. These simulations are input for Reduced-order models (ROM), which are used as a decision-making tool for the monitoring of composite parts produced by the infusion process.
The optimization of the infusion process requires the use of simulations to save the considerable time and effort associated with experiments. To be trustworthy, these numerical methods must be validated to ensure accurate predictions of potential defects, especially in thick and complex parts [2]. Material characterization is required for process modeling and simulation. Permeability characterization is vital for modeling the flow front and the pressure evolution in fibrous media, facilitating the prediction of residual defects. The main challenge lies in the prediction of the flow behavior under variable process conditions. This work aims to establish a physics-based model for the infusion of thick parts, apply a numerical approach for permeability characterization as input for the simulations, validate the numerical results with experiments through the monitoring system and then use the model to predict the process behavior for different scenarios to assist the decision-making process.
In this study, infusion experiments were carried out on thick components of 25 mm using a test bench equipped with a robust monitoring system. Sensors were placed throughout the part to monitor and control various parameters during the infusion. All preform permeability components are obtained virtually using the image-based flow and permeability software suite PoroS [3], which quantifies the ability of porous media to transmit fluids at micro- and/or meso-scale by solving single and/or dual-scale flow problems (Fig. 1). This cloud-based software allowed to predict the full permeability tensor based on a 3D segmented image of the thick stack of fabrics obtained from µCT scans.
The permeability values were used as input for the process simulations, which were carried out with the commercial software PAM-RTM® from ESI group. The results were compared and validated against the experiment (Fig. 2). Subsequently, the validated model provided a basis for the numerical prediction of multiple scenarios with variable material and process parameters. These simulations are input for Reduced-order models (ROM), which are used as a decision-making tool for the monitoring of composite parts produced by the infusion process.