Reinforcement learning for real-time control of Resin Transfer Molding: Bridging simulation and experiments

A comparative study on mechanical properties of sisal-leaf fibre-reinforced polyester composites prepared by resin transfer and compression moulding techniques

This paper presents results from an experimental investigation into real-time sensing and control of resin flow in an resin transfer molding (RTM) process. The objective of the research was to develop intelligent process control methodologies using in-situ sensors and process models, for real-time control of the RTM process. The real time control of the RTM process enables an increase in throughput, high yields, low defects, and consistent repeatability of quality between the parts. We concentrated on controlling the resin flow using multiple injection ports and vent locations. The results from the preliminary investigations indicate the feasibility of implementing real-time control on the RTM production floor.

Vacuum Assisted Resin Transfer Molding (VARTM) and Resin Transfer Molding (RTM) are among the most significant and widely used Liquid Composite manufacturing processes. In RTM preformed-reinforcement materials are placed in a mold cavity, which is subsequently closed and infused with resin. RTM numerical simulations have been developed and used for a number of years for gate assessment and optimization purposes. Available simulation packages are capable of describing/predicting flow patterns and fill times in geometrically complex parts manufactured by the resin transfer molding process. Unlike RTM, the VARTM process uses only one sided molds (tool surfaces) where performs are placed and enclosed by a sealed vacuum bag. To improve the delivery of the resin, a distribution media is sometimes used to cover the preform during the injection process. Attempts to extend the usability of the existing RTM algorithms and software packages to the VARTM domain have been made but there are some fundamental differences between the two processes. Most significant of these are 1) the thickness variations in VARTM due to changes in compaction force during resin flow 2) fiber tow saturation, which may be significant in the VARTM process. This paper presents examples on how existing RTM filling simulation codes can be adapted and used to predict flow, thickness of the preform during the filling stage and permeability changes during the VARTM filling process. The results are compared with results obtained from an analytic model as well as with limited experimental results. The similarities and differences between the modeling of RTM and VARTM process are highlighted.

The purpose of this study was to develop high-performance wood strand panels for automotive application using resin transfer and compression resin transfer molding technologies. Wood strand preforms bonded with 1%, 5%, and 20% by weight low-density polyethylene and 1% by weight high-density polyethylene were impregnated with resin during the molding process and compared. The results showed that 1% low-density polyethylene is sufficient to bind wood strands into a stable preform for handling and processing. Permeability measurements using a volumetric interpretation of Darcy’s law accounted for resin flow through the thickness of the preform. When compared with previously published data on other natural fiber composites, resin transfer molded wood strand composites generally exhibited superior mechanical properties and significantly greater dimensional stability when exposed to moisture.

Two phenylethynyl terminated oligomers designated PETI-298 and PETI-330 were developed at the NASA Langley Research Center and have emerged as leading candidates for composite applications requiring high temperature performance (i.e. ≥ 288 °C for 1000 hours) combined with the ability to be readily processed into composites without the use of an autoclave or complex or lengthy cure or postcure cycles. These high performance/high temperature composites are potentially useful on advanced aerospace vehicles in structural applications and as aircraft engine components such as inlet frames and compressor vanes. The number designation (i.e. 298, 330) refers to the glass transition temperature in degrees Centigrade as determined on neat resin cured for 1 hour at 371 °C. The resins are processable by non-autoclave techniques such as resin transfer molding (RTM), vacuum assisted RTM (VARTM) and resin infusion (RI). Both resins exhibit low complex melt viscosities (0.1-10 poise) at 280 °C and are stable for ≥ 2 hours at this temperature. Typically, the resins are melted, de-gassed and infused or injected at 280 °C and subsequently cured at 371 °C for 1-2 hours. Virtually no volatiles are evolved during the cure process. The resin synthesis is straightforward and has been scaled-up to 25 kg batches. The chemistry of PETI-298 and PETI-330 and the RTM AS-4 and T-650 carbon fabric laminate properties, and those of BMI-5270 for comparison, are presented.

The present study provides the performance evaluation of 2x2 twill woven composite (S2-Glass and C-50 resin system) material for Integral Armor applications. The laminates were fabricated by two low cost processes: RTM (Resin Transfer Molding) and VARIM or RI (Vacuum Assisted Resin Infusion Molding). These components are expected to be under fatigue loading. Fatigue behavior of the twill woven laminate is presented. Tension-Compression (R = -1) fatigue experiments were performed for both RI and RTM panels. All the fatigue tests were performed at 1 Hz frequency. S-N diagram and stiffness degradation over the fatigue life of the specimen was obtained. The fatigue tests indicated that with the same volume of fibers and with more resin, RTM process improved the fatigue performance over RI process. The ratios of logarithm of fatigue life cycles of RTM and RI is almost constant. Therefore on the absolute scale the ratio keeps increasing at higher fatigue life or at lower fatigue loads. However, with the same volume of fibers and resin, the fatigue performance of RTM and RI processed panels were comparable. It was also observed that both RTM and RI exhibited classical three-stage degradation in stiffness.

High temperature phenylethynyl-terminated imide oligomers derived from asymmetric diphenyl ether diamines for resin transfer molding

In both the aerospace and automotive industries, polymer-based composite materials are increasingly being used to replace metallic structures. To reduce the cost of fabricating complexly contoured polymer composite structures, manufacturers are actively developing low-cost alternatives to hand lay-up/autoclave curing and other labor and capital-intensive processes. One technique, that has proven to be an economical method for fabricating polymer composite structures containing complex contours, is resin-transfer molding (RTM). In the RTM process a dry fiber preform is placed between two faces of a mold. Resin is injected into the mold at low pressure and fills the mold cavity permeating the fiber reinforcement. After the part is fully cured it is removed from the mold. The key to efficient and cost-effective resin-transfer molding is to have an automated process in which the infusion of resin is controlled so that the flow front rapidly and evenly permeates the fiber preform. Several problems can occur during the transfer of the resin into the mold such as lack of penetration into all regions of the mold or incomplete infiltration of the fiber preform. Thus, in situ sensors are required to monitor the RTM process and ensure that pumping of resin is not stopped prematurely. Laser-based ultrasound (LBU) is well suited for this monitoring application since it can work with heated molds having nonplanar surfaces and can acquire ultrasonic C-scan images in times that are short compared to the total transfer time of the resin. These ultrasonic images can be used to enable rapid assessment of degree-of-completion of resin transfer, and indicate problems associated with the resin transfer or with defects in the mold. In this paper, results are presented which demonstrate the capability of laser-based ultrasound to monitor the resin flow front along the inner surface of the mold during the RTM process.

Resin Transfer Molding (RTM) is a process in which a thermoset polymer is injected into a closed mold filled with dry fabric reinforcement. Achieving net-shaped, void-free composite parts in RTM depends on using well-designed molds that promote uniform resin flow throughout the fiber preform. Traditionally, the design and fabrication of RTM molds, including the placement of injection gates and vent channels, involves an expensive and iterative process. This study investigates the integration of additive manufacturing (AM) techniques for the rapid prototyping of RTM molds with embedded sensing capabilities for real-time monitoring of the resin infusion process. In this work, carbon nanotube (CNT)-coated veils are incorporated as in-situ sensors within the AM-fabricated molds. The electrical conductivity of veils changes in response to resin infiltration, allowing for the detection of resin arrival and saturation at predefined locations within the mold. Experimental sensor data acquired during RTM trials are systematically compared with numerical predictions from resin flow simulations. Results show that the sensor signals provide local measurements of resin arrival times, supporting the feasibility of AM-enabled, sensor-integrated molds for RTM process monitoring and potential real-time control.

Resin Transfer Molding (RTM) is one of the most widely known composite manufacturing techniques of the liquid molding family, being extensively studied and used to obtain advanced composite materials comprised of fibers embedded in a thermoset polymer matrix. Nowadays, RTM is used by many industrial sectors such as automotive, aerospace, civil and sporting equipment. Therefore, the objective of this study is to verify the effect of calcium carbonate mixed in resin in the RTM process. Several rectilinear infiltration experiments were conducted using glass fiber mat molded in a RTM system with cavity dimensions of 320 x 150 x 3.6 mm, room temperature, maximum injection pressure 0.202 bar and different content of CaCO3 (10 and 40%) with particle size of 75μm. The results show that the use of filled resin with CaCO3 influences the preform impregnation during the RTM molding, changing the filling time and flow from position, however it is possible to make the composite with a good quality and low cost.

Silica fibre/phenolics composites produced by resin transfer moulding (RTM) solution impregnation technique are finding applications as aerospace structures. Typical phenolics contain solvent to facilitate injection and mould filling. It has been found that volatilisations of solvent and condensation polymerisation byproduct would cause void in cured composites, which is believed to affect the properties of the final product. In the present study, the role of solvent in the RTM impregnation process of silica fibre/phenolics composites was investigated. Experiments on wettability and flow were performed to evaluate the thermodynamical interaction on microscopic level occurring in fibre and resin system. Void content and interfacial strength were also examined. For the first time it has shown that the RTM process of silica fibre/phenolics composites is highly solvent dependent. Phenolics form distribution gradient in RTM mould with respect to isomeric composition under the effect of solvent, which results, to a varying extent, in the inhomogeneity of void content and thus interlaminar shear strength of the resulting product. The present work gives an insight into the role of organic solvent in RTM processing and this will help in choosing the best possible solvent for the solution impregnation system.

In the process modeling and manufacturing of large geometrically complex lightweight structural components comprising of fiber-reinforced composite materials with complex microstructures by Resin Transfer Molding (RTM), a polymer resin is injected into a mold cavity filled with porous fibrous preforms. The over-all success of the manufacturing process depends on the complete impregnation of the fiber preform by the polymer resin, prevention of polymer gelation during filling, and subsequent avoidance of dry spots. Since the RTM process involves the injection of a cold resin into a heated mold, the associated multi-physics encompasses a moving boundary value problem in conjunction with the multi-disciplinary and multi-scale study of flow/thermal/cure and the subsequent prediction of residual stresses in side the mold cavity. Although experimental validations are indispensable, routine manufacture of large complex structural geometries can only be enhanced via computational simulations; thus, eliminating costly trial runs and helping designers in the set-up of the manufacturing process. This manuscript describes an in-depth study of the mathematical and computational developments towards formulating an effective simulation-based design methodology using the finite element method. The present methodology is well suited for applications to practical engineering structural components encountered in the manufacture of complex RTM type lightweight composites, and encompasses both thick and thin shell-type composites with the following distinguishing features: (i) an implicit pure finite element computational methodology to track the fluid flow fronts with illustrations first to isothermal situations to overcome the deficiencies of traditional explicit type methods while permitting standard mesh generators to be employed in a straightforward manner: (ii) a methodology for predicting the effective constitutive model thermophysical properties, namely, the permeability tensor of the fiber preform microstructures in both virgin and manufactured states, the conductivity tensor, and the elasticity tensor; (iii) extension of the implicit pure finite element methodology to non-isothermal situations with and without influence of thermal dispersion to accurately capture the physics of the RTM process; (iv) stabilizing features to reduce oscillatory solution behavior typically encountered in the numerical analysis of these classes of problems: and (v) as a first step, preliminary investigations towards the prediction of residual stresses induced in the manufacturing process during post-cure cool-down. The underlying theory and formulations detailing the relevant volume averaging and homogenization techniques are first outlined for the multi-scale problem. Then the implicit pure finite element methodology, followed by the models for permeability prediction, is presented and compared for the case of isothermal mold filling. Applications of the pure finite element method is next extended to non-isothermal situations to accurately capture the flow/thermal/cure effects and the physics of the RTM process. Subsequently, a preliminary attempt is made to integrate the developments with the problem of thermoelasticity for residual stress prediction during post-cure cool-down. Where applicable, extensive validations of numerical results are made with analytical solutions and/or available experimental data. From these comparisons, relevant conclusions are drawn about the effectiveness of the present developments and their subsequent application to large-scale practical analysis of fiber-reinforced composite structures. Finally, some future directions relevant to the present study encompassing the multi-physics and multi-scale aspects of fibrous preforms with complex microstructures for use in lightweight composites are outlined.

Resin Transfer Molding (RTM) is a manufacturing process in which a liquid resin is injected into a closed mold pre-loaded with a porous fibrous preform, producing complex composite parts with good surface finishing. Resin flow is a critical step in the process. In this work, the numerical study of the resin flow in RTM applications was performed employing a general Computational Fluid Dynamics software which does not have a specific RTM module, making it necessary to use the Volume of Fluid method for the filling problem solution. Examples were presented and compared with analytical, experimental and numerical results showing the validity and effectiveness of the present study, with maximum difference among these solutions of around 8%. Besides, based on the computational model for the RTM process, a new computational methodology was developed to simulate Light Resin Transfer Molding (LRTM). In this process, resin is injected into the mold through an empty injection channel (without porous medium) which runs all around the perimeter of the mold. The ability of FLUENT® package to simulate geometries which combine porous media regions with open (empty) regions was used. Two specific cases were simulated, showing the differences in time and behavior between RTM and LRTM processes.

Application of ultrasonics for the process control of Resin Transfer Moulding (RTM)

Tool articulation is proposed as a means to improve mold fill time and to reduce void generation in resin transfer molding (RTM). Following a brief review of conventional RTM and a discussion of the limitations on the factors that control fluid flow through porous media, the articulated tool concept is described. The sequence of motion of an articulated segmented mold necessary for consolidation, void removal and accelerated fluid flow through a fibrous preform is described. An analysis of the process using a fiber preform with orthotropic permeability is outlined from which mold fill time is obtained. This is compared with conventional RTM mold fill times using typical resin properties and fiber volume fractions. In addition, void reduction in the articulated process relative to conventional RTM is investigated. For the conservative assumptions used in the analysis, the articulated process demonstrates an improvement by a factor of at least five in mold fill time compared with conventional RTM, accompanied by reduced void generation.