Genişletilmiş polipropilen (EPP) köpükler; otomotiv, HVAC (Isıtma, Havalandırma ve İklimlendirme), gıda, savunma ve spor ekipmanları gibi çeşitli endüstrilerde yaygın olarak kullanılmaktadır. Bu çalışmanın amacı, EPP köpük enjeksiyon proses parametrelerinin çok yanıtlı optimizasyonunu incelemek ve en iyi performansı elde etmek amacıyla optimal parametrelerin doğrulanmasını sağlamaktır. Köpük enjeksiyon sürecinde operatörler 11 farklı parametreyi ayarlamaktadır. Mevcut durumda, operatörler bir ürünün parametresini geçmiş deneyimlerine dayanarak deneme yanılma yöntemi ile belirlemektedir. Bu çalışmada, Taguchi yöntemi temel alınarak L27 ortogonal dizisi ile deneyler tasarlanmıştır. Sabit ve hareketli taraf buhar süresi, soğutma süreleri, köpük basıncı süresi ve dolum basınçları gibi proses parametrelerinin çevrim süresi ve çarpıklık seviyesi üzerindeki etkileri analiz edilmiştir. İlk olarak proseste kullanılan tüm parametreler sisteme dahil edilmiş ve deneyler gerçekleştirilmiştir. İki çıktının da minimum seviyede tutulabilmesi amaçlandığından Gri Tabanlı Taguchi yöntemi kullanılarak proseste en etkili parametreler analiz edilmiştir. Alınan sonuçlar sabit ve hareketli taraf buhar süresi, otoklav süresi, soğutma süresi ve köpük basıncı soğutma süresinin en etkili parametreler olduğunu göstermektedir. Çalışmanın devamında etkisiz parametrelerin değerleri sabit tutularak çalışmaya belirlenen beş parametre üzerinden devam edilmiştir. Beş parametre için L27 ortogonal diziye göre deneyler tekrarlanmıştır. Her bir deney kombinasyonu için üç tekrarlı deney yapılmıştır. Her bir çıktı için ayrı ayrı Taguchi yöntemi uygulanarak optimal deney kombinasyonları belirlenmiştir. İki çıktı için kombinasyonlar farklılık gösterdiğinden Gri Tabanlı Taguchi metodu ile iki çıktının da optimum olduğu deney düzeneği belirlenmiş ve sonuçlar Varyans Analizi (ANOVA) yöntemi ile incelenmiştir. Optimal deney kombinasyonu, sabit ve hareketli taraf buharlama sürelerinin orta seviyelerde, otoklav süresinin yüksek seviyede ve soğutma sürelerinin düşük seviyelerde tutulmasını içermektedir. İkinci aşamada deneyler için harcanan zamanın ve maliyetin azaltılması için prosesi modelleyen bir yapay sinir ağı oluşturulması amaçlanmıştır. Yapay Sinir Ağları (YSA) yöntemi ile çevrim süresi ve çarpıklık seviyesinin tahminlemesini sağlayan bir model oluşturulmuştur. Çapraz doğrulama metodu ile hiperparametre optimizasyonu yapılmış ve en iyi parametrenin belirlenmesi amaçlanmıştır. Yapılan çalışmada en iyi tahmin sonucunu veren model parametrelerinin düzenleme değeri: 0,0001, gizli katman boyutu: 150, öğrenme hızı: 'adaptive' ve çözücü: 'lbfgs' olarak belirlenmiştir Alınan sonuçlarda modelin çevrim süresinin %99,8 ve çarpıklık seviyesinin %99,7 oranlarında tahminleme yapabildiği görülmüştür.
The responses of companies to social, environmental, and economic challenges represent one of the most critical issues of our time. Faced with rapidly evolving technologies and the pressures of globalization, businesses are now compelled to undergo transformation by developing strategies that are tailored to the specific needs of their regions. This transformation process is increasingly gaining importance with the aim of enhancing efficiency in industrial applications and providing a competitive advantage. In recent years, there has been a significant increase in the use of expanded polymer foam beads. With the widespread adoption of electric vehicles, the importance of reducing vehicle weight has grown substantially, particularly within the automotive sector. Expanded polypropylene (EPP) foams are extensively utilized across a range of industries, including automotive, HVAC (Heating, Ventilation, and Air Conditioning), food services, defense, and sports equipment. Materials with a closed-cell structure, such as Expanded Polypropylene (EPP), are considered the most rational choice for designing protective products or energy-absorbing components. The energy absorbed is equivalent to the work done by forces that cause deformation of the material (such as crushing or breaking). Expanded polymer foams, such as EPP, play a vital role in modern engineering and manufacturing due to their versatility and resilience. Beyond automotive applications, EPP's lightweight yet durable structure has made it a popular material in the design of protective equipment, packaging solutions, and even energy-efficient construction materials. In the automotive sector, the rising adoption of electric vehicles emphasizes the need for innovative materials like EPP that can reduce overall vehicle weight while meeting stringent safety standards. This material's closed-cell structure allows it to absorb significant amounts of energy upon impact, making it ideal for various crash-protection applications. The global shift toward sustainable practices further underscores the importance of EPP as it is reusable, recyclable, and offers substantial energy-saving benefits in production and application. Closed-cell EPP foam finds extensive use in automotive bumpers to absorb impact energy. Variations in foam density and thickness influence energy absorption properties, thereby optimizing bumper system performance. Additionally, it has recently been increasingly used in vehicle seat components, including the cushion, backrest, and headrest sections. Due to its low weight and high energy absorption capabilities, EPP is widely applied in the aerospace and automotive industries. Compared to plastic injection processes, the EPP foam injection process is more complex and influenced by numerous factors. EPP exhibits a homogeneous, closed-cell structure due to its favorable mechanical properties and stable thermal conductivity, even in humid environments. For this reason, EPP can function as a high-performance thermal insulation material. Its degradation performance surpasses that of PS and polyethylene (PE) foams. However, PP resin has a high crystallinity and low melt viscosity, making it difficult to control compared to amorphous polymer foams. Additionally, pure PP is typically a linear polyolefin, and such polymers do not exhibit the high strain hardening required to withstand the stretching forces that occur during the later stages of bubble growth. Consequently, it is challenging to produce foams with a smooth, fine pore structure, high expansion ratios, and low open-cell content. Optimizing the EPP foam injection process requires overcoming several unique challenges due to the material's complex physical properties. EPP's high crystallinity and low melt viscosity make it difficult to manage during injection molding, as the material demands precise control over temperature, pressure, and expansion. Traditional trial-and-error methods, often based on operator experience, are time-intensive and may lead to inconsistent results. To address these challenges, process optimization techniques, including statistical methods like the Taguchi method and advanced analytical approaches, have been employed. By carefully adjusting parameters such as steam duration, cooling times, and foam pressure, companies aim to achieve consistent quality, minimize cycle times, and reduce waste. These optimizations are particularly valuable in the competitive manufacturing landscape, where efficiency gains can translate directly to cost savings and improved market responsiveness. In EPP foam bead production, steam serves as both a heat source and an expanding agent. In the initial phase, steam enters the mold, heating the polymer matrix and bonding the beads together. During depressurization, the steam within the beads transitions to a gas, leading to bead expansion and pressure formation within the mold. Water cooling solidifies the product, while air cooling stabilizes it. Once the pressure drops sufficiently, the mold opens, and the expanded foam bead product is ejected. Throughout the foam injection process, several factors impact part quality, including the pressure of the steam introduced into the system, steam duration, cooling times, internal bead pressure upon mold opening, and filling pressures. Determining parameters involves not only maintaining an appropriate quality level but also optimizing production costs. Shorter processing times lower production costs, provide flexibility in capacity utilization, and make cycle time a critical factor. Therefore, optimizing EPP production parameters is essential. The primary goal of this study is to elucidate the complex structure of the expanded polypropylene foam injection process, analyze the effects of parameters on output, and optimize the process considering these effects. In the current system, parameters are set experimentally based on operator experience, leading to significant time losses. To maintain competitiveness, the company aims to reduce cycle times to lower energy costs, enhance capacity flexibility, and minimize quality-related costs, adhering to a "right first time" approach. In line with this philosophy, it also targets reducing the most common defect, warpage. The study consists of two stages. First, the effects of parameters on output were identified, followed by an analysis of influential parameters to achieve optimal settings. Due to limited information in the literature on EPP foam injection molding processes, all parameters used in production were included in the scope of the study. During the foam injection process, operators adjust 11 different parameters. In this study, experiments were designed using the L27 orthogonal array based on the Taguchi method. The effects of process parameters, such as steam duration on both fixed and moving sides, cooling times, foam pressure duration, and filling pressures, on cycle time and warpage level were analyzed. Initially, all parameters used in the process were included, and experiments were conducted. Since the study aimed to minimize both outputs, the Taguchi-based Grey Relational Analysis (TB-GRA) method was applied to identify the most influential process parameters. Results indicated that steam durations on the fixed and moving sides, autoclave time, cooling time, and foam pressure cooling time were the most impactful parameters. In the continuation of the study, values for non-influential parameters were held constant, and further analysis was conducted with the selected five parameters. Experiments for the five parameters were repeated according to the L27 orthogonal array. Each experimental combination was conducted in triplicate. Optimal experimental combinations for each output were identified using the Taguchi method separately. Since combinations varied across outputs, the optimal setup was determined using the Taguchi-based Grey Relational Analysis method, and results were analyzed through Analysis of Variance (ANOVA). The parameters that yielded optimal results for both outputs were identified as follows: the cross-steam duration on the fixed side is 5.5 seconds, while the cross-steam duration on the moving side is also 5.5 seconds. The autoclave duration is set to 11 seconds. The cooling durations for both the fixed and moving sides are maintained at 65 seconds each. Lastly, the foam pressure cooling duration is established at 30 seconds. The integration of digital tools, such as artificial neural networks (ANNs), marks a pivotal shift in manufacturing, moving away from traditional, labor-intensive practices toward a more predictive and data-driven approach. Digitalization not only enhances process efficiency but also fosters adaptability, as ANN models can continuously learn from new data and refine production settings accordingly. In the context of EPP foam production, the ANN model developed in this study enables manufacturers to predict key outcomes like cycle time and warpage with high accuracy, streamlining decision-making and minimizing variability in production. This transition to a digitalized process aligns with the broader Industry 4.0 movement, which emphasizes smart manufacturing and sustainable practices. As such, the adoption of ANN and other digital tools positions companies to remain competitive in a rapidly evolving industrial landscape, ultimately paving the way for a more resilient and efficient manufacturing sector. In the second stage, an artificial neural network (ANN) model was developed to predict cycle time and warpage levels. Hyperparameter optimization was conducted using cross-validation, aiming to identify the best parameters. In the study, the best model parameters were found to be: regularization value: 0.0001, hidden layer size: 150, learning rate: 'adaptive,' and solver: 'lbfgs.' The 81-sample dataset was randomly split into 80% training and 20% test sets, and the model was run again. Results showed that cycle time and warpage level predictions were achieved with 99.8% and 99.7% accuracy, respectively. Future optimization studies can now focus on analyzing effective parameters, thus preventing time losses. The developed ANN model, by guiding parameter determination, can be regarded as foundational for transitioning from trial-and-error methods to a digitalized process.
Except where otherwise noted, this item's license is described as http://creativecommons.org/licenses/by/4.0/
