Simulation of the Resin Film Infusion Process Based on the Finite Element Method

Vol.21 No.4Joumal of Wuhan Univrsity of Technology - Mater. Sei. Ed.Dec. 2006Simulation of the Resin Film Infusion Process Based onthe Finite Element Method*YANG Mei，YAN Shilin' ，TAN Hua(School of Science, Wuhan University of Technology, Wuhan 430070, China)Abstract: A phsically accurale and computationally efectire pure finite element method ( FEM) 1 de-tveloped to simulate the isothermal resin infising process. The FEM was based on conseration of resin mass at anyinstant of time and was objective of rein film infusin ( RFT) fiber impregnatin and mold filing. The deelopedcomputer code uoas able to simulate the resin infusing risually. A numerical exumple presented here demnonstratedthat comnpared with tradional finite element/ control olumne ( FE/ CV) , and FEM was phsially accurate andcompuatinally eficien.Key words: resin film infusion ; finie element mehod; control wolune/ fnite elementof the resin flow front and the pressure distribution'1 IntroductionThe FE/CV approach makes the resin impregnation pro-cess be regarded as a quasi-steady process by assuming aResin film infusion (RFT) is a new cost-effctivesteady state condition at each time step even though it is afabrication technique. In this process, there are a largetransient process. The selection of the time step incrementnumber of material properties and processing parametersfor each of the quasi-steady state is based on the consider-that must be specified and contolled during resin iniltra-ation that the time increment used allows only one controlion make trial-and-error procedures of determining thevolume region to be completely flled ( this restriction ofprocessing cycle extremely inefficient. Analytical and nu-the time increment ensures the stability of the quasi-merical models are clearly superior altemnatives for deter-steady-state approximation). However, this may becomemination of optimal processing cycles.cumbersome and time-consuming when higher-order nodesThe success of RFI depends largely on the successfulare involved.impregnation of the preform by the resin. Prediction of theA pure finite element method ( FEM) was proposedlocation of the flow front and pressure distribution duringin this study to calculate the flow front movement for moldthe mold flling is useful in addressing several critical is-flling of resin film infusion. FEM is based on the conser-sues of the process. It will allow the designer to optimizevation of resin mass at any instant of time, so it is physi-parameters such as compaction pressure, infusion time ,ally accurate. Furthermore, this methodology primarilyperform layup, etc， in order to achieve an acceptableworks with the finite element mesh geometry, and theflow patterm before the mold is built and thus avoid poten-control volume regions associated with nodes need not betial problems.specified, so it is computationally fficient+Mold flling of resin flm infusion process involves amoving flow front, and so presents a moving boundary2 Resin Flow Modelproblem. Approaches to solve this problem can be classi-fied into either fixed mesh or moving meshd . The majorAn analysis impregnation is based on the conserva-method currenly being used to simulate the polymer resintion of the resin mass at any instant of time flowingflow movement of resin film infusion is the finite elemenU/through a porous prefon. A fll factor is defined to repre-control volume (FE/CV) method that is of fixed mesh.sent the amount of resin inside the preform and its distri-For example，Joohyuk Park and Moon Koo Kang proposedbution at any time. Considering for a general Euleriana mumerical algorithm for simulating RFI process for thinmold domain几, a fill factor(中) of one means that thepanels with sifeners based on the FE/CV method withregionis flled out with resin, while a fll factor of zeroconsidering the ffcts of compaction pressure on the pre-means that the region contains no resin, regions with fillform*; Alfored C Loos and John D MacRae et al chosefactors between zero and one are partly flld. Since thethe finite element control volume technique to develop amass conversation is maintained at any instant of time ，two-dimension resin flow model to determine the positionthe continuity equation based on it can be witten as;(Received: Feb. 15,2005; Accepted: Apr. 23 ,2006)[a(φ)' 中国煤化工)d =0(1)YANG Mei(杨梅): E mail: yangmei@ mail. whut. edu.cnwhereCNMH(or resin, n the vector*YAN Shilin(晏石林): Prof. ;E mil: yanshl@ mail. whut. edu. cn.MH(T),在the velocity* Funded by the National Natural Science Foundation of China(No. .vector. If the density is assumed to be constant based on50573060)Vol.21 No.4YANG Mei et al :Simulation of the Resin Film Infusion Process Based on the....181the incompressible condition, employing the Gauss theo-Then Eq. (8) can be given by:rem, then equation (1) can be predigested:[M]°●({φ|"+l_ {φ}")+ Ot[K]°●{P| = Ot{f|°(11)dQ+p(σ●i)d2 =0(2)The discretized finite element system is then ob-For describing the polymer resin flow through fibroustained:reinforcements as in RFI, Darcy's law is the most com-[M]●({φ}+1- {φ}") + Ot[K]●{P} = Ot{f}monly used equation, which can be expressed as:(12)百=-LAVP(3)4 Verificationwhere, [ K ] is the permeability tensor for porous media,P the resin pressure，and μ the resin viscosity. Substitut-In order to make the present numerical developmentsing Eq. (3) into Eq. (2) yields:and to demonstrate the applicability to RFT simulation in-中= [中(v .Kv P)dnvolving isothermal conditions a blade-stiffened panel ( seeFig. 1) was considered. The criteria included predictionsIf the pressure gradients are assumed to be negligibleof the resin flow front and the pressure distribution in thein the ufilled and partially flled regions where 0< φ