Heuristic algorithm for determining optimal gate and vent locations for RTM process design

Use of Centroidal Voronoi Diagram to find optimal gate locations to minimize mold filling time in resin transfer molding

“…Previously, many local and global search techniques have been attempted to optimize gate location using mold filling simulations to replace exhaustive search methods [24][25][26][27][28][29][30][31][32][33]. Gradientbased methods (such as Quasi-Newton method) [29] have been used to optimize the gate locations for RTM.…”

“…However, there is a danger that the searching process can converge to a local optimum [25]. Therefore, Genetic Algorithm (GA) and Artificial Neural Networks that are less prone to be trapped in a local optimum have been implemented to optimize gate locations in RTM to minimize cycle time [24,[26][27][28][30][31][32]. However, the complexity in formulating performance index requires one to evaluate a large number of generations before it converges, and renders these method less effective to handle complicate geometries and material property.…”

“…Many models have been developed to correlate the fill time for each point on the part surface with its distance to the injection gate [30,31,36,39,40]. On flat surfaces, resin flow was assumed as the combination of radial flow on the surface and channel flow along the boundaries and resin time is formulated in terms of the Euclidian distance to the injection gate [39].…”

Robust design of composites manufacturing processes with process simulation and optimisation methods

“…However, the large number of simulation runs required in the problem makes implementing impractical. Ye et al (2004) developed an effective algorithm, known as graph-based two phases heuristic, for this type of problem. The algorithm consists of the following steps:…”

Analysis of tail probability in stochastic RTM process simulation of FRP

“…に加えて，樹脂注入口で高圧をかける HP-RTM(high pressure RTM)が注目されている (Bodaghi et al, 2016) ． RTM 成形プロセス設計における課題点に，未含浸によるボイドの発生 (Matsuzaki et al, 2015)や成形後の残留 ひずみ (Canal et al, 2015, Mouton et al, 2010もあるが，プロセスの最適化という観点から樹脂浸透時間の短縮 (Gupta et al, 2013)に高い関心が集まっている．このため，数値シミュレーションを用いた最適化手法が多く報 告されている．Kang ら (Kang et al, 2000)は複数の入口ごとの圧力値を設計パラメータとし，Han ら (Han et al, 2015)は樹脂排出口の数と強化布の特性を，Ye ら (Ye et al, 2004) 予測手法が提案されてきた (Takano et al, 2002, Bechtold and Ye, 2003, Endruweit et al, 2006, Matsuzaki et al, 2012, Ravey et al, 2014, Zeng et al, 2015 ．Loix ら ( Loix et al, 2008)は 不確かさの要因の一つである平織強化布の変形ま で考慮している．また，X 線 CT を用いた最近の例 (Straumit et al, 2016)では，ばらつきが議論されている． 低コスト成形の問題点に， プロセスに関与する不確かさや成形品の特性のばらつきがあげられる (Sriramulra andChryssanthopoulos, 2009, Mukhopadhyay andAdhikari, 2017) ．成形板の部位による樹脂浸透係数の分布を考慮すれば 樹脂含浸完了時間の変動係数が 3～12%になること (Endruweit et al, 2008)などが報告されてきた．本研究でも樹 脂含浸完了時間を評価項目とする．Zhang ら ( Zhang et al, 2011)は，Probabilistic Collocation Method を用いた確率 的 RTM 成形プロセスシミュレーションを行い，モンテカルロシミュレーション(Monte Carlo Simulation，以降 MCS と略記)と比較を行っている．MCS は，いかなる確率密度関数でも容易に扱える信頼性の高いシミュレー ション法であり，FRP のマクロ特性予測 (Dwaikat et al, 2012, Almasi et al, 2015, Stefanou et al, 2015)や，各種の 成形プロセスシミュレーション (Li et al, 2006, Skordos and Sutcliffe, 2008, Mesogitis et al, 2015 ( Takano et al, 2012, Slamet et al, 2014, Akimoto and Takano, 2016 法と名付けた (Takano et al, 2012, Slamet et al, 2014, Akimoto and Takano, 2016 Cross-section w.r.t. and…”