An experimental and simulative study on the defects appeared during tow steering in automated fiber placement

Abstract

In the present study, an extensive experimental investigation using various process parameters and steering radii is performed to gain a deeper understanding of the defect formation process during tow steering in Automated Fiber Placement (AFP). Five major defects appeared during the tow steering are identified and effective factors on the formation of each of them are described. Furthermore, a finite element modeling approach is presented for the global modeling of the prepreg deposition process. The required model parameters are experimentally characterized. The application of the proposed framework for capturing and modeling two major defects appeared during steering namely, the blisters and wrinkles is presented. Simulation results are presented and compared to their corresponding AFP trial. It is demonstrated that the modeling approach is very well capable of capturing the trends, patterns, and frequency of both wrinkles and blisters.

Introduction

Various advantages Automated Fiber Placement (AFP) offers, has led to its increasing application in the aerospace industry to manufacture high-quality aerospace parts. Higher productivity and part quality with fewer variations and uncertainties, as well as a decrease of labor intensity and minimization of the number of the required parts can be obtained through this technology. AFP offers some unique features that specifically make it desirable for manufacturing complex parts with superior material properties. One of these features is its ability to deliver up to 32 prepreg tows in a course with different speeds, called differential tow pay-out. Differential tow pay-out enables the AFP machines to laminate complex geometries and Variable Stiffness Panels (VSP). In parts with complex geometries, steering the fiber path away from the geodesic path of the mold to maintain the desired fiber orientation is unavoidable. Moreover, differential tow pay-out allows engineers to implement curvilinear fiber paths, instead of fixed fiber orientations, within each lamina which consequently results in exploiting the full potential of the directionality of composites [1], [2].

Despite its numerous advantages and capabilities, AFP comes with its own limitations. The productivity of the lamination process and the quality of final parts are significantly influenced by the complexity of the part, as well as the process parameters such as lay-up temperature and speed, compaction force and fiber tension. Several different defects may appear while using AFP (especially during fiber steering) such as out-of-plane wrinkles at the inside edge of the tow, blisters (circular delaminations in form of a blister) in the middle of the tow-width, tow pull ups and shearing effects. By creating resin rich areas and deviation in fiber orientation, these defects decrease the part quality and reproducibility [2], [3].

The key mechanism which resists against the formation of most layup defects is prepreg tack [4], [5], [6], [7], [8], [9]. Tack is a complex systemic property of prepreg-tool interface that can be reasonably controlled by process parameters. Experimental trials performed during this study as well as a review of the literature demonstrated that tack levels as a function of process parameters can be generally understood considering the formation of a strong Intimate Contact (IC), stress relaxation effects in the prepreg resin and the resin’s cohesive strength. Perhaps layup temperature is the most sensitive process parameter that affects the prepreg tack and layup quality. By increasing the temperature, the resin viscosity decreases and a better IC is achieved, resulting in higher tack. Experimental observations, however, have reported a peak in the tack of prepregs and pressure sensitive adhesives as a function of temperature. This is due to the fact that reduction of resin viscosity (with increasing temperatures) is accompanied by the decline of resin strength. Consequently, at very high temperatures, a good contact is formed, whereas the lower strength of the resin results in cohesive failure within the prepreg (as opposed to interfacial failure at the prepreg-tool interface for lower temperatures) [6], [7], [9]. It is worthwhile to mention that the cohesive failure for prepreg materials is characterized by an observable amount of residual resin on the tool. Layup speed in the AFP is inversely proportional to the contact time. As such, the lower speed means more time for IC formation, lower strain rates and more time for relaxation of stresses in the prepreg resin, all of which result in higher tack levels. Increasing compaction force, again, helps to establish a good IC and increases the prepreg tack. A peak is also observed for the tack as a function of compaction force since further increase of the force will result in redistribution of the resin content and a contraction after the loading is released. Both these effects facilitate detachment of the prepreg from the tool surface (i.e. lower tack) [9]. Tack also has a peak around moderate values of relative humidity (50–60%), which is generally attributed to the changes in resin viscosity and strength [8]. Another important factor is the age of prepreg [4]. As the material ages, the crosslinking reaction occurs in the resin and the degree of cure increases. This will result in a stiffer material with lower tack. Using fresh prepregs within their shelf life helps in having a better quality layup. Other parameters such as tool’s surface roughness, and, resin and tool surface energies also play a role in both establishing the IC and the amount of work of adhesion, however, their macroscopic effects on prepreg tack is not very well understood at the moment and further investigation is required.

Among the different defects that limit the potential of the AFP’s process and applicability, the out-of-plane wrinkles have received special attention. Researchers have used experimental and modeling approaches to tackle the problem of wrinkle formation during tow steering. Essentially, trial and error is used to find the critical steering radius which results in a defect-free layup, in the experimental approaches [10], [11], [12], [13]. However, these methods are very expensive, time-consuming and drastically increase material wastage. Moreover, the critical steering radius significantly depends on the material system, tow width and manufacturing method used. Consequently, using different modeling techniques appears to be indispensable in efficiently finding the steering limit.

Modeling approaches [14], [15], [16], [17] take advantage of the incremental deposition of prepreg tows in the AFP and assume that wrinkles are formed individually, in isolation from one another. Therefore, each wrinkle can be considered as an orthotropic plate with one free edge and three clamped (or simply supported) edges which is resting on an elastic foundation (Fig. 1). Then, the critical buckling load of the representative plate can be linked to the critical radius of the steered tow. It has been suggested that the implicit assumption that the wrinkles appear frequently‌ (with constant wavelength) is not necessarily accurate and can be an important source of discrepancies. Recently, Hörmann [14] has recommended changing the modeling strategy for the future work by considering the overall behavior of the prepreg tow, instead of using the local approach.

In the present paper, an extensive experimental investigation using various different process parameters and steering radii was performed in order to gain a deeper understanding of the defect formation during tow steering and the steered tow quality, in general. Five predominant defect types, namely in-plane fiber waviness, sheared fibers, tow pull up (bridging), blisters and out-of-plane wrinkles, that occur during the steering are identified, and the mechanisms with which they appear are described. Moreover, a finite element modeling framework for simulating the prepreg deposition process in the AFP is presented in the commercial finite element software, ABAQUS. The cohesive zone modeling technique with a bilinear traction-separation law is used to represent the prepreg tack at the prepreg-tool interface. Using the finite element modeling, the global behavior of the prepreg tow can be analyzed as opposed to the local approach. The required model parameters are experimentally characterized. The application of the proposed approach for modeling the out-of-plane wrinkles at the inside edge of the tow, as well as the blisters in the middle of the tow-width is presented. Although it should be noted that the application is not limited to the case of fiber steering and can be generalized to different scenarios. An excellent agreement between the experiments and simulations was found.

The organization of the paper is as follows: The AFP machine and the trial procedure are introduced in Section 2. Moreover, important experimental observations are reported and a discussion on major types of defects occurring during the tow steering is presented. In Section 3, the strategies implemented for modeling the prepreg deposition process, modeling the prepreg tack and solving the models are presented. After the model parameters are identified, the experimental procedures implemented to characterize the material properties and to determine required parameters are briefly described in Section 4. The results of the models are presented in Section 5 and finally, the paper is concluded in Section 6.