Despite the advantages in weight savings and parts consolidation of composites relative to steel and aluminum, their use for automotive structural components has traditionally been limited by the lack of reliable material and component modeling capabilities. Our team has developed models that allow us to better understand the multi-scale, multi-physics that govern the manufacturing process of composite materials so that we can simulate the manufacturing of continuous carbon fiberreinforced composite parts with much-improved precision. This paper details the design and manufacturing approach for a composite load bearing floor that is manufactured using high pressure resin transfer molding (HP-RTM). While resin transfer molding (RTM) is a relatively old technology that traditionally required cycle times that were far too long for most automotive applications, by injecting the resin at higher pressures, significant improvements are possible. Additionally, significantly faster-curing epoxy resin systems have become commercially available, enabling lower cycle times. Taken together, these developments have made HP-RTM a highly desirable molding process for volume manufacturing of complex parts. HP-RTM also easily allows for the use of continuous fibers, which are crucial for an assembly to meet rigorous crash performance requirements. The manufacturing process was modeled extensively to predict the optimal injection strategy and molding conditions. Feedback from these simulations was used to rationally develop the tooling design. This floor’s large size presented particular challenges for filling the part before the resin reached its gel point. The key variables

Fiber-reinforced composites are widely used for structural components in the aircraft, automotive, marine, and other industries due to their low density, high specific stiffness and strength, excellent durability, and design flexibility. During the fabrication of continuous fiber reinforced composite components, fiber direction changes, residual stresses, and out-of-the-mold deformations are unavoidable. As the 2-D fabric is deformed to a 3-D part geometry the fiber tows move, leading to fiber direction changes that result in relative angle changes between the fibers that make up the fabric. These changes have a significant effect on the behavior of the composite material system. The extent of fiber angle change in a non-crimp fabric system is largely dependent on the differences between the 2-D and 3-D geometries and the particular stitching parameters used to manufacture the fabric. In order to better understand the influence of fabrication induced fiber angle change on the performance of structural composite parts, detailed experiments and simulations were conducted. For the experiments, a reinforcement geometry was selected. Due to the complexity of the chosen geometry for draping, several slits were designed at strategic locations to allow the fabric to take the shape without wrinkling. During molding, two patterns were overlaid so that the slit locations after molding were staggered through the thickness to reduce their effect on the structural performance. Detailed draping analyses were performed taking into account the process steps, and the fiber angle changes were calculated using numerical models that were developed previously. Further, the fiber angles following draping were mapped onto the structural performance models used to simulate the crush tests. The predicted stiffness and strength results from the integrated fabrication and performance simulations were compared with the experimental measurements, and the correlations are presented in this paper.

The manufacturing of fiber composite materials involves a set of complex, interconnected processes that span across multiple physics and scales. The characterization of uncertainty in composite manufacturing predictions is a challenging task that involves high-dimensional, multiscale, multiphysics stochastic models. We demonstrate the use of a basis adaptation scheme within a polynomial chaos representation that permits the incorporation of a large number of stochastic variables in the analysis. We use the proposed PCE-based workflow to analyze the interplay of uncertainty through all the fiber composite manufacturing stages that comprise the Resin Transfer Molding (RTM) process. The proposed framework is centered on an integrated assessment of uncertainty in composite structures using probabilistic surrogate models for predefined QoI.

The need for an increase of production rate in aerospace industry implies a growing interest in composite manufacturing process simulation with a strong requirement on predictive accuracy. In this context, ESI enhanced its PAM-COMPOSITES software introducing fluid-solid coupled approach in PAM-RTM module in order to simulate more accurately Vacuum Assisted Resin Infusion (VARI) process. VARI consists in impregnating a dry preform laid on a rigid mold and placed under a distribution medium and a vacuum bag. During the impregnation resin flows preferentially into the distribution medium and then in the preform which may undergo deformations due to the flexibility of the vacuum bag. Usual 2.5D approaches, using shell elements with thickness depending on resin pressure, cannot account for resin flow through the thickness due to permeability and fiber fraction gradients implied by the material used and/or solid mechanics effects (such as compression in curvatures). ESI new approach, resulting from several years of collaboration with academics, consists in a 3D finite element modeling. It is based on the coupling of resin flow, governed by Darcy’s law, with the preform behavior, considered as porous medium undergoing deformations, through Terzaghi’s principle. Thus it results in more predictive filling time and properties (thicknesses, fiber volume contents, geometry) of the final product.

We apply manifold learning and sampling to the tasks of fabrication, manufacturing, and testing of composites. We specifically address the challenge associated with statistical inference on these tasks from a small size sample. Limitations on the sample size could emanate from constraints on computational resources as well as constraints on physical experiments. In either case, the analyst is typically presented with a short table that contains observations of environmental conditions and quantities of interest (QoI). In the case of numerical simulations, the QoIs can be at the discretion of the analyst while in a laboratory setting these are typically limited by access to sensing devices. We augment the statistical knowledge captured by the available dataset with knowledge of physics constraints (eg conservation laws) in order to enhance the predictive value of the dataset. Imposing these constraints typically requires additional experiments (either physical or numerical). We proceed differently as we discover, within the dataset, an intrinsic structure that is consistent with the manner in which the available data is interrelated. To that end, we rely on diffusion maps, a recent data-analytics procedure. This allows us to rapidly characterize feasible domains for complex phenomena involving multiscale and Multiphysics interactions. We augment the diffusion map procedure with a stochastic sampler guaranteed to sample on the manifold, thus allowing us to impute a very large sample that is consistent with the statistics of the original dataset and its learned intrinsic features.

We propose a comprehensive framework for uncertainty management within the manufacturing process of non-crimp fiber composites (NCF), including the forming, resin injection, curing, and distortion. The challenge of meeting performance requirements while having incomplete knowledge about the fundamental physical processes is addressed with the objective of proposing manufacturing guidelines that are agnostic to these uncertainties. We accomplish this by making the functional dependence of uncertainties in the performance metrics and uncertainties in the various parameters and models explicit. We tackle the issues associated with dimensionality, which has hampered similar efforts in the past, through a basis adaptation procedure that permits the development of functional dependencies, for realistic systems, without any loss of accuracy. These representations are uniquely suited for design optimization as they provide explicit, yet highly accurate, stochastic reduced order models (SROM) that can be analytically differentiated and integrated. We compute several Quantities of Interest (QoI) as functions of random variables and processes of material properties and process conditions. During the forming stage, we consider the mechanical properties of the fibers and the local fiber directions to be random. The deformation of the fabric was computed via a reduced model consisting of an effective shell with embedded upscaling algorithms at the integration points. Forming induces stochastic fluctuations in the relative shearing angles of the fabric, which are mapped, through a stochastic model, into spatial fluctuations of the permeability field. The simulations were carried out using the PAM-COMPOSITESâ„¢

Non-crimp fabric (NCF) preforms are an attractive alternative to traditional preimpregnated tapes due to their low manufacturing cost and high efficiency. The orientation of the fibers in each layer can be tailored, independent of the other layers, to optimize the required load carrying capacity in that particular direction, making them capable of improved performance. Stitches help to keep the fiber tows in the NCF fabric straight during handling; however, the stitches prevent the fibers from reorienting easily to accommodate complex shapes without wrinkling. In order for NCF fabrics to be used to create complex geometric shapes, their draping behavior needs to be understood with respect to different fabric variables so that the draping performance can be maximized. To date, the draping behavior of NCF fabrics has been only sparsely investigated in contrast to the amount of research reported on woven fabrics. This paper presents an investigation on the role of fabric architecture in the formability of NCF fabrics. This study is a subset of a broad study conducted under the purview of a Department of Energy project funded to General Motors for developing state of the art computational tools for integrated manufacturing and structural performance prediction of carbon fiber composites. For modeling the draping behavior, fabric characterization tests such as bending and bias-extension evaluations were conducted for NCF fabrics with varying areal weights and construction. Taking advantage of ESI’s PAM-FORM material model, asymmetric shear behavior was included in conjunction with different membrane and bending behavior to model the draping behavior of these fabrics. For this study, the fabric characterization data was first used to calibrate the draping models in simple shear and bending tests. Later these models were validated against the deformation of the fabrics when they were formed using a truncated pyramid tool designed at the General Motors Research Labs to assess the drapeability of the dry fabrics.

Composite materials are being used at an increasing rate in the automotive industry due to their superior mechanical properties at low densities leading to lightweight components. Continuing this trend requires the identification of manufacturing processes for these components that have short cycle times along with delivering reproducible parts. The modeling of these processes is essential in order to prevent the excessive experimentation currently required to develop the process parameters. One of the manufacturing processes currently under consideration is compression resin transfer molding (C-RTM). Recently, General Motors and ESI, NA Group have been working together in developing a state of the art computational tool for process simulation of composites in a project supported by the Department of Energy (DOE). This section of work on the development and validation of a simulation tool for the CRTM process is under the broad scope of the above DOE project. Current development for C-RTM simulation introduces two new technologies embedded into the current PAM-RTM solver: - Fluid-Solid mechanics coupling to determine the preform deformations during the compression-injection process, and - An Inter-Penetrating Mesh providing the ability to handle the vanishing gap during the closing of the mold. This paper will focus on the development and validation of the proposed simulation software using resin flow experiments in a truncated pyramid tool manufactured at General Motors Research and Development Labs.

Stokes, Darcy and solid mechanics coupling is a matter of interest in many domains of engineering such as soil mechanics, bio-mechanics, and composites. The aim of this paper is to present a robust iterative method to deal with this coupling for low permeability media within the framework of industrial simulation, and especially for composite manufacturing processes. Stokes and Darcy problems are solved using a mixed velocity–pressure finite element using a mini-element formulation, and coupled together by the so-called Beavers–Joseph–Saffman conditions through their interface. This fluid formulation is then coupled to a non-linear solid mechanics formulation in finite deformations using Terzaghi׳s law at the pore level, and an explicit dependence of permeability with respect to porosity that is exactly computed from the solid mechanics kinematics. Then, those formulations are validated with test-cases and by the Method of the Manufactured Exact Solution (MMES) (Knupp and Salari, 2003[1]). Finally, a 3D curved transient example of application is presented.

Today, LRI is a proven manufacturing technology for both small and large scale structures (e.g. sailboats) where, in most cases, experience and limited prototype experimentation is sufficient to get a satisfactory design. However, large scale aerospace (and other) structures require reproducible, high quality, defect free parts, with excellent mechanical performance. This requires precise control and knowledge of the preforming (draping and manufacture of the composite fabric preforms), their assembly and the resin infusion. The INFUCOMP project is a multi-disciplinary research project to develop necessary Computer Aided Engineering (CAE) tools for all stages of the LRI manufacturing process. An ambitious set of developments have been undertaken that build on existing capabilities of leading drape and infusion simulation codes available today. Currently the codes are only accurate for simple drape problems and infusion analysis of RTM parts using matched metal moulds. Furthermore, full chaining of the CAE solution will allow results from materials modelling, drape, assembly, infusion and final part mechanical performance to be used in subsequent analyses. Although the materials and manufacturing methods in INFUCOMP are specific to aerospace structures, it is expected that the work would be of great value to other industries, including energy (windmill), rail, sea, advanced automotive and manufacturing. INFUCOMP has built on PAM-RTM, an existing simulation software, to provide a full solution chain for LRI composites; including fabric modelling, drape, assembly, infusion, cost and final part performance prediction. Simulation tools will avoid costly and time consuming prototype testing, will allow the CAE design of alternative manufacturing routes and enable cost effective, efficient LRI composite structures to be designed and manufactured. This paper presents the work carried out during industrial validation phase of the project on simplified industrial components and an industrially relevant LRI aircraft sub-structure. This work has used several specific developments including numerous enhancements to the state-of-the-art for resin infusion simulation; in particular, better viscosity models and essential developments to run under DMP (Distributed Memory Processing) to take advantage of new generation cluster computers and massive parallel computing. Some details about coupling of modelling and monitoring allowing a combination of predictive capabilities provided by simulation with the capability of detecting unexpected events and variations in real time provided by process monitoring will be presented. Introduction Within the INFUCOMP project, an infusion-compression tool has been developed to fully simulate the infusion process, taking into account the deformations of the preform within PAM-RTM, an ESI Group software. Validations on small scale singularities that are representative of industrial issues have been conducted and are presented in this paper. After an initial description of the test and monitoring set-ups, we will present details of the simulations including input data and results. Finally, experiment results are compared with simulation results. Following those representative tests, an industrial benchmark will be detailed including defect prediction by simulation and correlation with experimental results.

With the increasing use of composite materials in the aerospace industry, composite parts suppliers show a growing interest about process simulation. Among the industrial processes used in the production phase, resin infusion appears more and more as an economical alternative for manufacturing large parts with an important fiber fraction (wind turbine blade, aircraft wing...). However, the lack of control on the final properties of the part, implying long and expensive process tuning, significantly reduces the above-mentioned advantages. So, a full model coupling fluid/solid/porous mechanics is proposed to simulate the liquid resin infusion (LRI) process, in order to anticipate the potential problems numerically.

Infusion processes are a cheaper alternative to the usual injection process to realize big composite parts. However the bad control of the final properties of the molded part is a disincentive to their democratization at industrial level. We are presenting an advanced modeling method, coupling fluid flows with finite deformation solid mechanics, in order to anticipate as best as possible thickness variation of the part and infusion time.