SIMULATION BASED OPTIMISATION OF INDUSTRIAL MANUFACTURE OF LARGE COMPOSITE PARTS BY INFUSION

Anthony Pickett * ESI GmbH, Mergenthalerallee 15-21, 65760 Eschborn, Germany akp@esigmbh.de http://www.esi-group.com/ Tony Green Israel Aerospace Industries Ltd. Engineering Development Division. Material Engineering Department 4479 Ben Gurion International Airport, Israel 70100 agreen@iai.co.il

Abstract To date, manufacture of advanced composites in the Aerospace industry mostly uses pre-impregnated composite materials, tape laying technologies and autoclave curing for the production of large, high performance structures and components. These combined technologies allow toughened resins to be uniformly dispersed in a well controlled fibre system with a high fiber content, producing excellent mechanical stiffness, strength and fatigue resistance properties. However, there are drawbacks, including high material costs, limited shapeability, complex, expensive and time-consuming manufacturing, and short materials shelf life. As a consequence alternative manufacturing methods are being sought based on Liquid Resin Infusion (LRI) technologies in which the resin is infused only after all dry textiles are assembled to form the final composite component configuration. This assembly, prior to infusion, is called a preform. The advantages are lower material and material storage costs, indefinite shelf life (for the textiles) and the ability to manufacture integrated structures having complex geometries only limited by shapeabilty of the dry preforms. Currently, LRI of large composite structures require 'trial and error' testing and considerable experience on the part of designers and manufacturers to get the correct set-up. The high cost and risks involved will often lead to overly conservative infusion designs with associated cost and performance penalties; or may lead to alternative, less competitive, manufacturing technologies and materials being adopted. The scientific aim of the CEC INFUCOMP project (CEC, 2010) is to provide a full simulation chain for LRI manufacture of large aerospace composite structures dedicated to solutions required by the European Aircraft industry. Extensive materials testing for a range of dry fabrics and permeability characterisation is being conducted from which new constitutive laws will be developed. Software developments will be implemented into an existing infusion code PAM-RTMTM (ESI Group 2009), which has essentially been developed for Resin Transfer Moulding (RTM) processes. Some other specific developments include process optimisation, cost

Anthony Pickett and Tony Green

analysis and predictive tools to characterise imperfections such as porosity and residual stresses. One major software development is extension of current scalar computing capabilities so that advantage may be taken of State-of-the-Art massive parallel computers; this will allow a step change to full 3D infusion modelling involving tens of millions of elements. The new capabilities will be validated by a series of demonstration studies on representative parts using the different infusion technologies actively used at the four industrial partners sites.

Keywords Composites, Manufacturing, Infusion, Simulation, Aerospace. 1. Partnership of the Consortium The consortium is led by a software company who will further develop the existing PAM-RTM software code, based on scientific contributions from university, materials and research partners. An important aspect of the project is feedback on new developments from a series of demonstrator studies to be undertaken by industrial partners; this will help validate and industrialise the new software. In detail the consortium includes: Four aircraft manufacturers: Bombardier Aerospace, Belfast (UK), Piaggio Aero Industries S.p.A (Italy), Daher Aerospace (France) and Israel Aerospace Industries (Israel). One material manufacturer: Hexcel (France). One digital simulation software supplier: ESI Group (France) and ESI GmbH (Germany). An infusion sensor specialist INASCO (Greece). Four academic partners Cranfield University (UK), University of Patras (Greece), Ecole des Mines de Douai and Saint-Etienne (France), and Katholieke Universiteit Leuven (Belgium). Two institutes: The Institute for Aircraft Design, IFB (Germany) and SWEREA SICOMP (Sweden). 2. Infusion Processes Infusion of dry fabrics may be undertaken using a variety of processes. The first popular method is Resin Transfer Moulding (RTM) which is widely used for the manufacture of small to medium sized components and is especially suited to automation and the manufacture of relatively high production volumes. High injection pressures can be used to reduce infusion time, but is in practice limited by stability of the tooling and effects such as porosity and fibre distortion which may occur especially in the vicinity of the injection ports. The main limitations are high cost and size (weight) of the two part tooling, and relatively long cycle times needed to cure the resin before the part is sufficiently stable to be extracted from the mould. The key steps in RTM are shown in Figure 1 below and involve shaping the preform, extraction and trimming of the preform, and resin infusion in sealed matched (usually metal) tooling.

Fig. 1. Steps in pre-forming and RTM infusion (Rudd, et al. 1997).

As mentioned some limitations of RTM are high cost, weight and complexity of tooling. Liquid Resin Infusion can overcome these problems and requires only one sided tooling, which may be significantly lighter and therefore cheaper. Figure 2 below shows the essential features of LRI. A dry fabric is laid up in the tooling, usually with a low permeability flow medium to aid infusion, and a peel ply and/or release ply to help separation of the final composite and flow medium after curing of the final part. The complete setup is sealed in a vacuum bag that prevents air entering the system; vacuum is applied at one location (outlet port) and draws resin from an inlet port through the dry fabric. Large complex parts invariably need a system of inlet and outlet ports which may use synchronised opening and closing to enable complete infusion of the large volume. Also shown in Figure 2 is an intermediate stage of the infusion of an aircraft composite fairing, described in more detail (Alonim, 2005).

Fig. 2. The LRI infusion set-up and an industrial example (Courtesy IAI). The following example, Figure 3, illustrates a typical state-of-the-art resin infusion of a rotor blade pitch horn (Weiland et. al. 2009). The analysis model is separated into zones with each having assigned thickness and permeability data. In complex preforming and infusion processes, detailed effects of fabric shear and compaction on permeability distribution may require assessment.

Fig. 3. View of thickness and permeability zones (Courtesy Eurocopter). For the analysis, Figure 4, a simplified 2D shell element representation is used. The study investigated two infusion strategies; namely, using circumferential and axial inlet (blue) and outlet (green) lines with a 1 bar overpressure for each. Analysis results predict similar infusion times of 895 secs and 805 secs for the circumferential and axial injection respectively; however, the results indicate that the axial infusion strategy could present a higher risk as the flow front does not remain straight and there is a likelihood of air entrapment as the two flow fronts meet.

Anthony Pickett and Tony Green

Fig. 4. Example of infusion strategies and results (filling times). 3. Technical Challenges of the Project To date general purpose computational software to tackle LRI all use a Finite Element method based on a control volume approach. The constitutive law for rate of flow is governed by Darcy's law for flow through a permeable medium. A brief description of the theory is given in the following section. The method is well validated and does give acceptable results for RTM type processes; however, application of the method, particularly to large scale LRI processes does present special difficulties. First, the use of thick laminates and particularly the application of flow media causes resin flow to be a truly 3D process, with resin rapidly advancing through the low permeability flow media followed by slower entry into the preform in the through-thickness direction. Modelling this phenomenon necessitates the use of 3D solid elements. Whilst these element types are available, the problem becomes the huge number of elements required, particularly for large structures, and tens of millions of elements will be mandatory. Tackling model sizes on this scale will require a complete restructuring of the current software for massive parallel computer architectures; furthermore, and equally important, new pre- and post-processing codes will be required to handle this vast amount of data efficiently. Several important new developments have been identified to improve representation of the physics in the LRI infusion process. First, air flow though the dry fabric is considered to influence the infusion and, second, the flow velocity and capillary actions at the flow front are felt to be poorly described by simple Darcy's law. New constitutive laws will be developed for these effects from a planned test program. The preforming operation and loading in the infusion set-up causes fabric deformations that significantly modify permeability and porosity distribution. New work will test and characterise these effects experimentally, and provide methods to transfer permeability and porosity distributions of the loaded fabric to the infusion model for more accurate infusion simulation.

A primary aim of the simulation is minimisation of defects such as resin porosity. However, as in any practical manufacturing process a compromise is inevitable between acceptable manufacturing imperfections and final part cost and performance. Consequently, several tasks are dedicated toward this topic and will develop predictive models to estimate manufacturing voids and their effect on final part performance, and to determine residual stresses that result from the infusion and curing. Cost analysis to predict final part cost in LRI is another area of investigation. Finally, application and validation of the new software tools and methods as they are developed is planned to be undertaken by the industrial partners. This work will eventually lead to complex demonstrator parts representative of the different infusion processes used at the industrial partners sites.

4. Summary Current trends in the aircraft industry have placed an increased emphasis on cost reduction. This trend, however, does not permit any compromise with traditional quality parameters for components, such as weight, geometrical tolerances, material properties and defect free structure. For infusion technologies, achieving these aims relies on the skills of a small number of experienced experts using closely guarded company 'know-how'. As with any process that requires manual input, the development process for new parts is often imprecise with unpredictable aspects. This was also the situation some years ago for the prepreg/tape laying/autoclave technology. This drawback was recognised by the aircraft industry and led to the development of computer aided systems for all stages of the design, manufacturing and QA process. This has led to a step function improvement in the economics of composites in new aircraft designs. Infusion processing requires similar, automated, expert systems for component selection, definition and optimisation that take much of the human element out of the loop. The aim of INFUCOMP is to establish this infrastructure for the benefit of the European industry.

References Advani, S., Bruschke, M., & Parna, R. (1994). Resin transfer molding. In S. Advani, Flow and rheology in polymeric composites manufacturing (pp. 465526). Amsterdam: Elsevier Publishers. Alonim, J., Arnon E., David A., Gold E., Green A.K., Hackman N. and Leibovich H. (2005). Development of a Business Jet Component by Resin Infusion using a Stitched Preform, Proc. 26th Intl. SAMPE EC Conf., Paris, 9A, (pp. 476). CEC. (2010). Simulation based solutions for industrial manufacture of large infusion composite parts. INFUCOMP CEC project ACP8-GA-2009-233926. ESI Group. (2009). 99 Rue des Solets, Silic 112, 94513 Rungis-Cedex, France. Mathur et al. (2001). Flow front measurements and model validation in the vacuum assisted resin transfer molding proces. 22 (4), 477-490. Rudd, C., Long, A., Kendall, K., & Mangin, G. (1997). Liquid moulding technologies. Cambridge, England: Woodhead Publishing.