Sei sulla pagina 1di 7

Construction

and Budding

Vol. 9, NO. 6, pp. 325-331, 1995 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 095&0618/95/$10.00+0.00

Materials,

0950-0618(95)00051-8

Local buckling of pultruded beams nonlinearity, anisotropy and inhomogeneity


Lawrence C. Bank, Jiansheng Yin and Murali Nadipelli
The Catholic University of America, Washington, DC

Department of Civil Engineering, 20064, USA Received 21 October

1994; accepted

13 February

1995

Local buckling of pultruded fibre-reinforced plastic beams is discussed. The paper focuses on three issues related to the prediction of buckling loads both from experimental data and from analytical and numerical approaches, viz. nonlinearity, anisotropy and inhomogeneity. Experimental data obtained from full-scale buckling tests are reviewed and a method proposed for estimating the buckling stress in pultruded beams. Analytical studies based on classical orthotropic plate buckling theory are used to determine the edge restraint coefficient for pultruded beams and also to show the influence of the in-plane material properties on the buckling loads. Numerical studies using the finite element method in which inhomogeneous material properties in the beam cross-section are considered are used to give predictions of buckling loads of the beams. Although the paper focuses on local buckling of pultruded beams, it raises issues which are relevant to the analysis of pultruded material structures of all types. Keywords: pultruded beams; fibre-reinforced plastics; finite element analysis

Pultruded fibre-reinforced plastic (FRP) profile sections have been produced for over 20 years in the United States and in Europe. In recent years there has been increased interest in the use of pultruded profiles in the construction industry for load-bearing structural applications in both building systems and bridges. Primarily motivated by their resistance to corrosion, engineers are now discovering additional benefits of pultruded composites which may include light weight, electromagnetic transparency, damage tolerance, formability and installed cost. Recent improvements in manufacturing technology, particularly in the pultrusion method (although also in the various resin transfer moulding methods), have enabled the production of large crosssections (in the range of 1 m by 1 m) with relatively thick walls (in the range of 25 mm) at relatively low cost ($4 to $6 per kg). Such large sized members are needed for full-scale load-bearing civil engineering applications. The low cost coupled with potentially low installation costs due to light weight and customized profile shapes can make pultruded composites viable structural materials. Regardless of the form of the eventual profile, prior to use as a structural component, a pultruded profile currently needs to be tested at full-scale to validate its structural performance. This is because at the present time reasonable doubt exists as to whether or not structural behaviour can be predicted from knowledge of the profile geometry and coupon test data. The purpose of the current paper is to attempt to shed some light on why such doubts exist. Recently, a series of
Construction

controlled and extensively instrumented full-scale flexural tests on a variety of pultruded beams was conducted at the Catholic University of AmericazA. The tests were conducted to investigate local buckling and ultimate failure of the compression flange of commercially produced wide flange I-shaped beam sections when loaded in flexure. The purpose of the testing was to develop a robust test methodology, to obtain test data on local buckling and failure behaviour, and to determine whether analytical and numerical methods could be used to predict buckling loads based on knowledge of the profile geometry and coupon test data. Although prior flexural testing of pultruded beams has been conducted this work does not contain the experimental detail needed in order to gain an indepth understanding of the phenomena discussed in this paper. Since local buckling is one of the governing design criteria for pultruded beams it was chosen for this in-depth study. Other design criteria of importance are lateral-torsional buckling6, ultimate failure, deflection, vibration, connection capacity and local bearing capacity. To meet these objectives a set of studies was conducted to assess the significance of a number of factors related to the local buckling characteristics of the pultruded beams tested. These were (i) the determination of buckling load and buckling stress from experimental nonlinear flange strain data, (ii) the effect of anisotropy of the pultruded material and the edge restraint provided by the web/flange junction, and (iii) the effect of
and Building Materials 1995 Volume 9 Number 6 325

Local buckling

of pultruded

beams:

L.C. Bank et al.

in-plane material inhomogeneity webs of the pultruded profile.

in the flanges and the

Nonlinearity
Tests were conducted on three types of standard pultruded FRP wide flange I-beams: (i) vinylester beams of the 203 mm (8) high and wide by 9.5 mm (3/8) thick series, (ii) polyester beams of the 203 mm (8) high and wide by 9.5 mm (3/8) thick series, and (iii) vinylester beams of the 203 mm (8) high and wide by 12.7 mm (112) thick series. For identification purposes these will be referred to as V 318, P 3/8, and V l/2, respectively. Tests were conducted in a fixture especially designed to produce local buckling in the beams in a controlled fashion and to prevent undesirable local instability and failure modes from occurring. The beams were simply supported and tested in four-point bending at a span of 2743 mm (9 ft) with an interior constant moment span of 1219 mm (4 ft). The beams were monitored with strain and displacement indicators. Full details of the testing methodology and descriptions of the buckling and failure behaviour of the beams are described in References 2-4. The load at which local buckling occurred in the experiments was determined from strain data obtained from back-to-back strain gauges bonded on the top and bottom surfaces of the compression flanges. The difference between the top and bottom surface strain gauge data provides a convenient way of detecting the instability in the compression flange. A typical plot of the difference in strain data (S2-S-3) from back-to-back strain gauges, one upperside (S-2) and one on the under-side (S-3) of the compression flange, is shown in Figure 1 for one of the P 318 beams (identified as P8-1 in Figure I). As can be seen from Figure 1 the bifurcation in the strain data occurs gradually over a range of load values and a method is needed to precisely determine the buckling load from plots of this type. Two methods were considered: an estimation method and an analytical method. In the estimation method the buckling load is determined graphically (or numerically) from the knee in the difference strain versus the load plot. The location of the knee is determined by extending the initial linear

portion of the load-strain curve until it intersects the straight line extending from the final post-buckled (large deformation) linear portion of the curve. The load thus obtained is referred to as the estimated buckling load. It is this load that has been previously reported as the buckling loads3 for the beams tested. Buckling data for pultruded column tests have also been obtained using this method7. Previously reported local buckling data for beams was obtained by visual observation of large post-buckled deformation and therefore also represents this estimated value. Following careful examination of photographic data of the buckled beams, such as that shown in Figure 2, and considering the fact that the load versus strain difference has a gradual slope change as the buckling phenomenon initiates and then progresses into the postbucked regime, it was felt that a procedure was needed to determine the initial buckling load. This initial buckling load was, therefore, determined analytically from the point at which the change in the initial linear slope of the load versus strain difference plot exceeded a predetermined value. A value of 10% slope change was used in this study. This value was the lowest value that could confidently be used to identify buckling that was not adversely influenced by the noise of the strain data obtained from the experiments. A typical plot of the change of slope versus load is shown in Figure 3. Average values of the estimated and the initial buckling loads for the three types of beams tested are given in Tuble 1. From the data in Tuble 1 it can be seen that the percentage difference between the initial buckling loads and the estimated buckling loads is 22%, 33% and 27% for beams V 3/X, P 318 and V l/2, respectively. Photographs (taken during the testing) of the postbuckled deformation (for example Figure 2) show between three and four buckle half-wavelengths in the

Figure 2

Post-buckled deformation of a pultruded beam

120 110 100 90 60 5 -0 d s 70 60 50 40 30 20 10 0 0 -

201

estimated buckling ______________--___~_-----~__--_________---------~ _.m initial buckling

Pa- 1 -f--------------------------------------------------------: f 100 1 200 I 300 I 400 I 500 1 600 1 700 I 600 I 900 1000 . s-s3

-20

/ 10 20

/ 30

1 40

I 50

I 60

! 70

/
90 10

60

Microstrain
Figure 1

Load
Figure 3

(kN)

Load YL~SUS strain difference plot

Change of slope VP~SU.S plot load

326

Construction

and Building

Materials

1995 Volume 9 Number 6

Local buckling
Table 1

of pultruded

beams:

L.C. Bank et al.

Experimental loads and stresses Initial buckling load, P,, CkN) 71.1 66.9 140.1 Estimated buckling
load, P,,,

Beam

Ultimate failure
load, P,,,

(W 87.3 89.0 177.9

(kN 98.3 96.1 189.4

Initial buckling stress, 4, WPa) 66.6 62.6 102.7

Estimated buckling stress, cr,,, WPa) 121.0 123.1 192.5

v 3/8 P 318 v II2

constant

moment region of the beams. It is felt that the lower initial buckling load realistically represents the onset of elastic buckling, while the estimated buckling load is that which exists when the section has been loaded into the irreversible post-buckled regime (see the large post-buckled flange deformation shown in Figure 2). Loading, unloading and reloading tests show that the buckling load is significantly reduced when the beam is reloaded after having been loaded into the postbuckled regime2. It can therefore be concluded that the nonlinearity seen in the expermental data is caused by both material nonlinearity due to accumulating material damage and geometric nonlinearity. Although the pultruded beams may appear to be undamaged following unloading after an excursion into the post-buckled regime, this is not the case, and designers are cautioned against utilizing the post-buckling capacity. Although deflection limits will usually govern beam design in typical applications it is nevertheless vital to understand the phenomenon of local buckling in pultruded beams so that design codes can be developed for pultruded structures. Based on the above, it is recommended that the appropriate method for determining buckling loads from strain data of this type be the analytical method and not the estimation method. A simplified graphical version of this method can also be used whereby the buckling load is determined approximately by observing where the onset of nonlinearity in the initial linear portion of the load-strain difference plot occurs. For comparison purposes the average ultimate failure loads of the beams are also shown in Tuble 1. Since the beams fail in a number of different modes2s3 it is not possible to directly compare the buckling loads to ultimate loads for the purposes of safety factor estimation; however, it is important to note that the estimated buckling loads are very close to the ultimate loads in all cases. Consequently, design approaches based on the estimated buckling loads will have lower margins of safety and should be avoided. Following determination of the buckling loads a procedure is required to obtain the buckling stresses. Based on experimental data in which it has been observed that the distribution of axial strain across the compression flange is nearly uniform, the classical beam flexure formula is used, viz. CJ = o,,, = MC/Z, where cr,,,, is the maximum flange compressive stress (shown throughout this paper with a positive value for ease of presentation), M is the bending moment at the section, c is the distance from the neutral axis to the compression face in the elastic section, and Z is the second moment of area of the elastic section. However, since a postbuckled configuration exists at the estimated buckling Construction

load, a post-buckled version of the classical formula is used in this case, viz. cr = MylZ*, where y is the distance to the face of the compression flange from the neutral axis in the post-buckled section, and I* is the second moment of area of the post-buckled section. The geometry of the post-buckled section is obtained using an effective width concept and experimental data for the location of the post-buckled neutral axis. For the fourpoint-bend test geometry used M = Vu, where V = PI2 and c1 is the moment arm equal to 762 mm. The geometric properties of beams V 318 and P 318 are I = 4.13 x 10 mm4, I* = 3.24x 10 mm4, c = 101.6 mm, y = 117.9 mm and for beams V l/2 are, I= 5.28 x 10 mm, I* = 4.15 x 10 mm4, c = 101.6 mm, y = 117.9 mm. The calculated initial and estimated buckling stresses are shown in Table 1. As can be seen from Tuble I the estimated stresses are significantly higher than the initial stresses. As mentioned previously the initial stresses should be considered to be the actual buckling stresses for the pultruded beams, while the estimated buckling stresses give a measure of the flange stress in the post-buckled configuration close to failure of the beam. Therefore, while the initial buckling stresses may be predicted using linear elastic analysis, prediction of the estimated buckling stresses will require geometrically and materially nonlinear analysis.
Anistropy

In order to determine whether or not the experimental initial buckling loads could be predicted from knowledge of the section geometry and material properties an analytical study was first conducted. Classical linear elastic orthotropic plate buckling analysis was performed in which the compression flange was modelled as an orthotropic plate free on its longitudinal outer edge and elastically restrained at its longitudinal inner edge. This approach was chosen, as opposed to the approach in which the buckling of the full-section is modelled, since it enables one to obtain a numerical estimate of the extent of the web/flange interaction through the edge restraint coefficient. The classical solution for an orthotropic plate which is simply supported on its loaded edges is well documented3a4 and will not be repeated in detail. The solution is given in terms of the four in-plane independent elastic constants of the orthotropic plate, E,,, E12, G,,, v,~, the nondimensional buckling coefficient, K, = N,b2/tiD,, the nondimensional edge restraint coefficient, R = SblD,, the mode number, m, and the plate aspect ratio, $ = u/b. N, is the in-plane load per unit width, b
and Building Materials 1995 Volume 9 Number 6 327

Local buckling

of pultruded

beams:

L.C. Bank et al.


Table 2

is the plate width, D2 = &t3/{ 12(1 - vlzv,,)} is the plate

Material property data Vinylester (V 318, V 112) Polyester (P 318)

transverse rigidity, S is the restraining moment along the rotationally restrained edge per unit length per unit rotation and a is the plate length. The buckling stress is obtained as, CT=NJt, where t is the plate thickness. The classical solution is obtained via the Levy solution technique and results in a nonlinear transcendental equation which is solved numerically. Typical results are shown in Figure 4, which shows the dependence of the buckling coefficient on the aspect ratio and the edge restraint coefficient. For the beams tested in the experimental study2-4 4 = 12. In order to perform the above calculations the inplane material properties of the pultruded plate material are needed. Results presented in this paper are given for two different sets of material properties: those based on properties provided by the manufacturer of the pultruded sections and identified as design data and those based on properties obtained from coupon tests conducted at the Georgia Institute of Technology16 and called Georgia data. The two sets of data are given in Table 2. As can be seen from Table 2, these two sets of data are different. This is not uncommon and stems

Property Design data Georgia data Design data Georgia data -%, (GPa) J% (GPa) Gr2 (GPa) 12 17.0 8.0 2.5 0.3 24.6 10.3 3.7 0.33 17.0 7.0 2.0 0.34 24.0 1.5 2.6 0.31

Data not provided by manufacturer and obtained from other sources3

E =24.6GPa G:;=3.66GPa

E =10.27GPa v:~=O.333

0.0

1
0

10

12

14

Length-width
Figure 4

Ratio # (a/b)

Buckling curve for elastically restrained orthotropic plate

from the fact that manufacturer data is given for design purposes and is generally conservative. Recent results have indicated that the inhomogeneity of pultruded materials, which is due to the manufacturing process, can lead to significant differences in property data depending on coupon size and location within the section. As can be seen from the buckling curve shown in Figure 4, prediction of the buckling stress depends on the value of R, the edge restraint coefficient. The buckling stress for the two limiting cases of a simply-supported edge (R = 0) and a clamped edge (R = -) can, however, be obtained. Comparison between the experimental buckling stresses and the limiting cases is shown in Table 3 for the different beams and material property data. From Table 3 it can be seen that the expermental buckling stress falls between the two limiting cases. For design purposes the assumption of a simply-supported edge is a conservative assumption. Since stability will often dictate the design of pultruded structures it may be desired to take account of the edge restraint provided by the web/flange intersection to increase the design load on the beam. In order to obtain the value of the edge restraint coefficient from the experimental data the buckling curves can be plotted in an alternative form as shown in Figure 5. From plots such as these the value of the edge restraint coefficient and the number of buckle half-wavelengths (mode number) can be obtained from the experimentally determined buckling stress. In Figure 5 the specific case of a V l/2 beam is shown. From the experimentally determined buckling stress of 102.7 MPa (see Table I) an edge restraint coefficient of 1.06 is obtained. In addition, it can be seen that the analytical solution predicts four buckle halfwavelengths (m = 4) which agrees with the expermental

Table 3

Comparison between theoretical and experimental buckling stresses $=12,t=9.5mm %Wrlmrnl (MPa)
R=O R=m R=O R=cm

@= 12, I = 12.7 mm %WUnrnl (MPa)

Georgia data

Vinylester Polyester Vinylester Polyester

36.8

146.4

260.3

26.8

114.2

62.2

47.6

203.0

NA

Design data

102.7

20.7

93.4

62.2

38.0

171.7

NA

328

Construction

and Building

Materials

1995 Volume 9 Number 6

Local buckling
250

of pultruded

beams:

L.C. Bank et al.

225

200 3 $?
b-

175

2102.6 2 100 Lz 75

w
rn=5

-1

experimentally3 that such customizing of the fillet region significantly increases the buckling stress in pultruded beams. Inhomogeneity Composite materials are by definition inhomogeneous, consisting of fibres in a surrounding matrix. Theoretical analysis of composite materials on the macroscale is based on the assumption of statistical homogeneity in a representative volume element and the theory of laminated plates. For composite materials made by precise lay up of prepregs these two assumptions appear to be valid. However, for pultruded composite materials, these assumptions cannot be made. Inspection of a through-the-thickness section of a pultruded material shows inhomogeneity on the scale of the part itself, consisting of clusters of roving reinforcements in arbitrary locations both in and out of the plane of the material. Although some layering due to the presence of continuous stand mats does exist it is not precise as in the case of a laid-up composite plate. This is due to the nature of the pultrusion process which has only approximate control over the placement of reinforcements (fibres, mats and fabrics) entering the die and no control during the curing processes within the die. In particular, inspection of a profile section shows that the reinforcements tend to be most arbitrary at the free edges and at the junctions between plates. Often the region near a free edge (up to 10 mm) may have fewer roving bundles than the rest of the plate, while the reinforcement mats in the region of the junctions may be folded over in a bunched fashion. Therefore test coupons are not taken from these regions and their properties are unknown. However, it can be assumed that the properties of these sections are lower than the average properties of the pultruded material in the flat parts of the profile. It has

In=8

Vinylester

Design data E,,=17.00GPa E,,=8.00GPa G,z=2.48GPa =0.3 a 9 b=12 b=95.25mm t=12.7mm

25 10-l loo 10' lo2

lOA

Coefficient

of Restraint

Figure 5 Buckling coefficient observations.

curve used for determination

of the edge restraint

of the edge restraint conumber, m, for the beams tested are given in Tuble 4. Experimental mode numbers between 3 and 4 were obtained for all beams tested2s3. The fact that the predictions with the design data show mode numbers of 5 for the V 3/8 and P 3/8 beams suggests that the design data are in fact lower than the true material data. However, since the photographs of the buckled deformation were taken in the post-buckled range it is possible that the wavelengths may have increased (hence the decrease in mode number) from those at the initiation of buckling due to damage of the web/flange junction. A parametric study was conducted to study the influence of the orthotropic material properties and the edge restraint coefficient on the buckling behaviour. Figure 6 shows the value of the buckling stress as a function of the variation in individual properties. The effect of relative changes in properties from the baseline properties defined by the Georgia data (identified with superscript 0) are shown. As can be seen from the figure the inplane shear modulus has the greatest effect on the buckling stress. The effect of the edge restraint coefficient can also be seen. The edge restraint coefficient can be increased by increasing the size of the fillet region between the web and the flange. It has been shown
Predictions the efficient,

R, and

mode

1.20 1.25 1.20 1.15 ObX A, b 0 3 l+z 1.05 1.00 110

Table 4
number

Analytical

predictions

of edge restraint

coefficient

and mode

l
t (mm)
Gpcrrment
m

ii

0.95

v)

0.90
0.85 0
80

Wa)
66.6 102.7 62.6 66.6 102.7 62.6 3 3 4 5 4 5 0.75 0.19

2 ._
al

Georgia data

Vinylester Polyester

9.5 12.7 9.5 9.5 12.7 9.5

9 S

I .06
2.18 I .06 2.81

0.75 0 70 0.5 0 6 Vanable 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 s/s') -J 1.5

Design data

Vinylester -~ Polyester

Ratio

(EI1/EP,,

E,,/Eiz,

GI,/G,l,

Figure 6

Parametric representation of buckling stress

Construction

and Building Materials

1995 Volume

9 Number 6

329

Local buckling

of pultruded

beams:

L.C. Bank et al.

additionally been shown I7 that even in the flat parts of the section variations in stiffness properties may be as much as 65% depending on location in the cross-section. In order to account for the influence of the inhomogeneity on the local buckling behaviour of the pultruded beam an elastic buckling analysis using the COSMOS finite element code was performed. The finite element method was chosen since it permits the user to vary the material properties throughout the cross-section in an arbitrary manner and also allows for precise modelling of the beam support and loading conditions. Analytical studies of in-plane inhomogeneous composite plates are few and have considered only simplified distributions of the in-plane properties, primarily for theoretical purposes8*9. The beams were modelled using anisotropic plate elements measuring 25.4 mm by 25.4 mm. Each cross-section was divided into 24 elements. Lengthwise the 3048 mm long beam was divided into 120 elements giving a total of 2880 elements in the beam model. To investigate the effect of plate inhomogeneity different distributions of in-plane properties were chosen. The property distributions chosen for the inhomogeneous cases are shown in Figure 7. The distributions were chosen in an

attempt to account for lower properties at the edges and the intersections and to investigate the influence of the longitudinal properties versus the transverse properties. They were refined by a trial-and-error method for purposes of demonstrating the dependence of the buckling load on material inhomogeneity. In all the distributions shown in Figure 7 a constraint condition was chosen such that the average property over the plate was equal in value to the Georgia data. A baseline case in which the average property was used throughout the cross-section (uniform case) was used for comparison purposes. Models were run in which either all or selected material properties were varied while the remaining properties were held uniform. In one model only the longitudinal modulus, E,,, was varied according to the distributions shown in Figure 7, while in the other case the transverse and Poisson properties were varied while the longitudinal modulus was held uniform. Such arbitrary choices of property variations are theoretically possible since all four constants are independent; however, in reality it is likely that there would be a relationship between these properties due to the microstructure of the material. Results of the finite element analyses are given in Table 5. A number of interesting features can be seen in

I
8 I I

Gz#e2 Figure 7 Distribution of properties in the beam cross-sections

Case 3

Table 5

Buckling loads (kN) of beams with variable material properties V 318 Experimental buckling load 71.1
FEM

P 3/8

v 112

69.9 analysis buckling load (uniform properties) 70 I analysis buckling load (variable properties)

88
FEM

Case 1 Case 2 Case 3

70.0 62.0 69.0

80.0 89.0 84.6

67.7 70.0 16.2

47.7 56.2 59.5

156.6 141.5 160.0

330

Construction

and Building Materials

1995 Volume 9 Number 6

Local buckling

of pultruded

beams:

L.C. Bank et al.

these results. The assumption of uniform properties throughout the cross-section leads to FEM predictions which are greater than the experimental loads. The FEM predictions in which variable properties are used can be very close to the experimental values (this, of course, depends on the distributions chosen in Figure 7). The influence of the transverse property variations is much more significant than of the longitudinal property variation. In fact, the low predicted values of the P 318 beam in contrast to the V 3/8 beams (all properties varied) is almost entirely a function of the transverse property variation since the longitudinal modulus of these beams is very similar (see Table 2). The results suggest that material inhomogeneity should be included in finite element analyses of pultruded beams, and that variation of the properties in the edges and junctions of the beams can have a significant effect on the overall performance of pultruded beams. It remains to be determined if the property variations chosen in this study are indeed a true representation of the actual material property variations

Acknowledgement
Support for this study was provided by the National Science Foundation under grant number MSM-9015502 (Dr K. Chong, Program Director). The pultruded beams used in the experimental study were provided by Creative Pultrusions, Inc.

References
Head, P.R. Design methods and bridge forms for the costeffective use of advanced composites in bridges. In Advanced Composites in Bridges and Structures, eds. K.W. Neale and P. Labossiere, The Canadian Society for Civil Engineering, 1992, pp. 15-30 2 Bank, L.C., Nadipelli, M. and Gentry, T.R. Local buckling and failure of pultruded fiber-reinforced plastic beams. ASME J. Eng.
1 Mater. Technol. 1994, 116, 233-231

Bank, L.C., Gentry, T.R. and Nadipelli, M. Local buckling of pultruded FRP beams - analysis and design. In Proc. 49th
Annual SPI Conference, Composites Institute, Society for the Plastics Industry, Cincinnati, OH, February 7-10. 1994, Session

Conclusion
Prediction of the local buckling loads in pultruded fibre reinforced plastic beams subjected to flexural loading is influenced by a number of factors arising from the nonlinearity in expermental data and from the anisotropic and inhomogeneous material properties of the beams. It has been shown that nonlinear experimental data obtained from full-scale tests on pultruded beams must be correctly interpreted so as not to overestimate the buckling capacity of these beams. An argument has been made for estimating buckling loads at the very onset of nonlinearity in the test data, and not at the point at which large and visible post-buckling deformation occurs. It has also been shown that the value of the edge restraint coefficient for a pultruded beam can be obtained from a combination of test data and classical orthotropic plate theory. Parametric studies have been presented which show that the in-plane shear modulus has the largest influence on the buckling load. The importance of utilizing the actual material properties in analytical or numerical studies has also been emphasized. It has been demonstrated that material inhomogeneity in pultruded beams can be a cause of discrepancies between experimentally determined buckling loads and those predicted by finite element methods. The assumption of uniform material properties throughout the beam may not be appropriate for the analysis of pultruded beams.

8-D, pp. 1-6 Nadipelli, M. Expermental investigation of local buckling and failure of pultruded fiber reinforced plastic beams. Master of Civil Engineering Thesis, The Catholic University of America, Washington, DC, 1994 5 Barbero, E.J. and Raftoyiannis, I.G. Local buckling of FRP beams and columns. ASCE J. Mater. Civil Eng. 1993, 5 (3),
339-355

6 I

Mottram, J.T. Lateral-torsional


Composites, 1992, 23 (2) 81-92

buckling of a pultruded I-beam.

Yoon, S.J., Scott, W. and Zureick, A. An experimental investigation of the behavior of concentrically loaded pultruded columns. In Advanced Composites in Bridges and Structures, eds. K.W. Neale and P. Labossiere, The Canadian Society for Civil Engineering, 1992, pp. 309-317 8 Mosallam, A.S. Short and long-term behavior of a pultruded fiber reinforced plastic frame. PhD Thesis, The Catholic University of America, Washington, DC, 1990 9 Mosallam, AS. and Bank, L.C. Short-term behavior of pultruded fiber-reinforced plastic frame. ASCE J. Struct. Eng. 1992, 118 (7), 1937-1954 IO Lekhnitskii, SC. Anisotropic Plates. Translated from the 2nd Russian Edition (1956) by SW. Tsai and T. Cheron, Gordon and Breach, NY, I968 Lee, D.J. The local buckling coefficient for orthotropic structural sections. Aero J., 1978, 82 (81 l), 313-320 Bulson, P.S. The Stability of Flat Plates. Chatto and Windus, London, 1970 Shuleshko, P. A reduction method for buckling problems of orthotropic plates. Aero. Q., 1957, 8, 145-156 Brunelle, E.J. and Ayibo, G.A. Generic buckling curves for specially orthotropic rectangular plates. AIAA J., 1983, 21 (8) Alum Bank, PA, 1988 16 Zureick, A. Personal communication, 17 Wang, Y. and Zureick, A. Characterization of the longitudinal tensile behavior of pultruded I-shape structural members using coupon specimens. Comp. Struct. 1994, 29 (4) 463472 I8 Leissa, A.W. and Martin, A.F. Vibration and buckling of rectangular composite plates with variable fiber spacing. Comp.
Struct. 1990, 14 (4), 339-357

I5

1150-1156 Creative Pultrusions Design Guide, Creative Pultrusions,

I9

Pandey, M.D. and Sherbourne, A.N. Stability of inhomogeneous fibrous composite plates. Int. .I Sol. Struct., 1993, 30 (I), 37-60

Construction

and Building

Materials

1995 Volume

9 Number 6

331

Potrebbero piacerti anche