STUDY OF SHEAR LAG BEHAVIOUR OF UMINATED COMPOSITE BOX-BEAMS

Abstract

Fiber reinforced plastic (FRP) is a material that has been in use since the 1940s but only recently has won the attention of engineers involved in the construction of civil structures. The effective properties of the FRP laminate vary with the orientation, thickness, and stacking sequence of the individual layers. The large number ofvariables available with composites also means that the use of computational procedures, and artificial intelligence should be considered in improving designs. Need of appropriate structural form with FRP is required to utilize its advantages. Use of FRP in box beam form combines the advantages of both the material and form and results in an efficient generic structural element. In addition to this, the designer can change material properties for different parts of the beam crosssection. This enables the shape of the cross-section to be exploited to the fullest by judicious arrangement of unidirectional plies within the laminated composite panels, which form the beam. This tailoring develops high degree orthotropy in the beam panels causing unexpected enhancement in non-classical effects like shear lag. The distributions of bending stress across wide flanges of a beam cross-section are nonuniform. This phenomenon is usually called shear lag. Shear lag necessarily involves loss of efficiency of the material put in the flange. Especially in laminated composites due to anisotropy the phenomenon gets enhanced. In addition to this it exists till failure, hence increases the importance of shear lag studies in case of composites. In the present work, the shear lag behaviour of simply supported laminated composite box beam subjected to mid-span point load has been investigated. In 11 particular, the effect of orthotropy of FRP composites on shear lag behaviour has been explored. In this context, large number of numerical studies has been conducted using Finite element method (FEM). Finite element method has been employed to analyze the laminated composite box beams as it remains the most convenient method of treating complex composite structures. Shear lag behaviour has been investigated for • Symmetrical single-cell FRP box beams; • Un-symmetrical single-cell FRP box beams, and • Symmetrical multi-cell FRP box beams. Key parameters have been identified which govern the shear lag phenomenon for each case. Effect of these parameters on shear lag phenomenon has also been studied. It has been observed that three key parameters, ratio of span to width of beam (L/B), orthotropical parameter (<y,) and cross sectional parameter (/cr,), govern the shear lag behaviour of symmetrical laminated composite box beam. Additional key parameters, 77, (ratio of smeared longitudinal extensional stiffness per unit width of top and bottom flanges) for un-symmetrical box beam and n(number of cells) for multi-cell box beam are required to explain the shear lag effect. It is observed that shear lag orthotropical parameter (co{) depends upon material properties and fiber orientations, whereas, cross-sectional parameter (/c,) is sensitive to geometrical parameters and fiber orientations. It is also seen that in general, an increase in parameters &>, and v, result decrease in effective width ratio of top flange {BJB), that is shear lag enhances with the increase in these parameters. It is further noticed that ±45° fibers orientation significantly reduce shear lag effect. In addition to these it is found that with increase in/c, or with decrease inw,, the effectiveness of cells, in reduction of shear lag behaviour, reduces, in case of multi-cell box beams. iii Onthe basis of FEM studies a large data base has been generated, which will be helpful in designs. Finally using database the three ANNs are trained to predict shear lag with reasonable accuracy. It is observed that effective width prediction by ANN takes fraction of second while the corresponding FEM analysis takes two to six minutes. This proves the computation efficiency of developed artificial neural networks. In nut shell the identified shear lag parameters and developed database as well as trained ANN will be quite useful in the prediction of shear lag behaviour of FRP box-beam. IV