Experimental investigation on the moment-rotation performance of pultruded FRP web-flange junctions

Abstract

This paper aims to present an experimental investigation on the behavior of web-flange junctions (WFJs) rotational stiffness of pultruded fiber-reinforced polymer composites (FRP). Channels and I-sections were tested using a simple set-up, which was developed in order to experimentally characterize the junctions in a direct manner. The Digital Image Correlation (DIC) technique was used, allowing overall deflections and relative rotations between web and flange to be monitored. The WFJs' imperfections were analyzed through an optical microscope and correlated with the cracks' formation. Further, damage thresholds are identified using available stress equations for curved composite members and lower bound functions are proposed to simulate the junction stiffness retention. Finally, two Equations are developed in order to analytically predict pultruded junctions' rotational stiffness per unit of width. In general, the theoretical and experimental results agreed fairly well, with a maximum difference of 24% for I-sections and 38% for channels.

Introduction

The use of pultruded fiber-reinforced polymers (FRP) composites in structural applications has greatly increased in recent decades, due to their several advantages over traditional materials, such as its lightweight, corrosion resistance and ability to tailor geometry. Researchers worldwide have dedicated significant efforts to better understand the performance of pultruded FRP members, especially related to instability problems. In this context, the behavior of the junctions between adjacent plates comprising the cross section became a matter of interest due to its influence on the buckling response of pultruded members. The local failure at these regions affects the overall structure behavior, having influence on its stiffness and strength [1,2].

Previous works [[3], [4], [5]] have shown that web-flange junctions (WFJs) are characterized by a non-linear semi-rigid constitutive law, which affects the local buckling capacity. Cintra et al. [6] compared the critical loads for pultruded GFRP columns having semi-rigid and rigid junctions between web and flange, reporting reductions up to 9.4% for the former. Liu and Harries [7] investigated the flange local buckling behavior of pultruded box beams and found out that the experimental results were less than one-third of the theoretical predictions using rigid-junction expressions available in FRP of design codes [8]. The authors observed that analytical equations considering constituent plates simply supported along edges between adjacent plates lead to better predictions, ratifying the fact that junctions do not present a rigid behavior. Mosallam and Bank [2] also reported results that have indicated that the web-flange intersection could only offer a partial restraint to the compressed flange.

The composites failure is often related to the junctions, which are an important point of weakness [4,7,9,10] due to their curved shape and particular fiber architecture with incidence of mat wrinkling and presence of resin-rich areas [1,11]. Such defects are cited in literature as some of main factors that lead to premature failure of web-flange junctions [12]. The junction is, thus, a potential area for the onset of failure that might lead to separation of web and flange, controlling the overall behavior of the composite until its premature rupture [2,12,13]. Bank and Yin [14] stated that this specific failure mode is a matter of concern, once it affects the geometry integrity of the member cross-section, which will behave “as a collection of individual and separate plates” [14] that can no longer bear the applied load. The authors cited the geometry, fiber architecture and material properties of junctions' region as elements that play an important role in determining the dominating failure mode. In this context, the junctions’ characterization became a matter of great relevance.

Many authors have devoted attention to develop appropriate techniques to determine the moment-rotation relations (M – θ) and to obtain the WFJs rotational stiffness [2,3,[9], [10], [11], [12], [13]]. Bank et al. [15] have conducted four-point bending tests to characterize the junctions indirectly, through the measurements of the half-wave lengths. Values of 5.04 and 3.29 kN/rad were reported for the rotational stiffness per unit of width, considering properties obtained from characterization tests and from the lay-up information provided by manufacturer, respectively. Turvey and Zhang [16] developed three-point bending apparatus to characterize WFJs rotational stiffness of an I-section. Analytical models were used to obtain transverse bending modulus, as well as the relation between moments M and web rotations θ. The same behavior was assumed for both junctions of cross-section and values of 6.83 kN/rad and 6.57 kN/rad were found for rotational stiffness. The authors reported “a relative lack of consistency” [16] in the results and attributed it to the fact that fiber architecture in web-flange junctions are much less ordered than other parts of the profile. In addition, the failure process was observed to initiate through a delamination at the triangular core rovings. Mosallam et al. [1] also developed a test apparatus to evaluate rotational stiffness of web-flange junctions of H-sections and angles. The bottom flange and web were clamped using thick steel angles, while rotation was applied to the upper flange. However, according to Yanes-Armas et al. [3], this test fixture may not have represented the actual boundary conditions, due to pre-compression in specimens through thickness direction, affecting results and failure mode. Xin et al. [12] developed a test apparatus to obtain the moment-rotation relation. The test fixture consisted in clamping the specimen flanges with screws on a base, while a load P is applied on the web from a distance d of the junction. The set-up prevented the web's flexure, simplifying issues related to the obtaining of deflections due to bending, however demanding a more elaborate test configuration. The authors also evaluated the influence of critical parameters on the WFJs behavior, concluding that greater values of flange/web thickness and fillet radius result in higher junction's rotational stiffness. Yanes-Armas et al. [3] conducted an experimental study on WFJs of a bridge deck system and determined the M-θ relation through two analytical models. Results ranging from 178 to 628 kN/rad were found for junction's rotational stiffness, reaching significantly higher values in some cases. These high values were attributed to the different decks' fiber architecture, which includes triaxial multi-ply fabrics besides rovings and non-structural mats. On the other hand, some lower values were attributed to nonlinearities due to existing pre-cracks.

Based on researches available on literature, the angle of rotation θ, needed for the construction of the M-θ relation, is usually obtained through indirect manners, i.e., using the elastic curve equation to previously determine additional parameters – such as the transverse modulus E_t – through specific characterization tests [16,18]. Test setups using an especial apparatus to avoid influence of flexural deformation in the measurement of θ can also be found in literature [1,13]. Therefore, the present work proposes the characterization of the rotational stiffness by a direct method, where the relative rotation between web and flange is measured directly during test using Digital Image Correlation technique [17]. This procedure allows a simple setup to be used - without the need of a special apparatus – and simplifies issues related to the gap between the clamp tips and the junction that arise when using elastic curve approaches.

Moreover, this paper aims to discuss the influence of imperfections on the WFJ's performance and to fill existing gaps in literature by proposing three complementary analyses for WFJs' behavior: i) a simple methodology to determine the stresses on crack onset region based on the curved beam theory for anisotropic material [18]; ii) a lower-bound function to model stiffness retention based on experimental results; and iii) a theoretical expression for prediction of the rotational stiffness of WFJs.