You are on page 1of 36

Stress and stiness analysis

of
beam-sections
Lars Damkilde

Department of Structural Engineering and Materials
Technical University of Denmark
DK-2800 Lyngby
November 2000

Summary
The report gives an introduction to stiness and stress analysis of beam-sections.
The purpose is to estimate the cross-sectional stiness parameters i.e. axial,
bending, shear and torsional stiness from the geometry and material properties
of the beam section. Furthermore the stress distribution in the beam-section
from the internal forces i.e. axial force, bending moments, shear forces and
torsional moment is calculated.
Axial force and bending moments are treated based on the assumption of "plane
sections remain plane". Methods to nd the elastic center and principal axis
are given including a computer oriented formulation based on triangularization
of the beam-section. The stress distribution - Navier's formula - is formulated
including beam-sections build of dierent materials. Examples illustrate the
methods.
The theoretical shear stress distribution for a rectangular prole from shear
forces is presented. The Grashof formula for calculating shearing forces is presented and illustrated with examples. The torsional problem is not treated in
detail and merely an overview of the problem is given. The torsion problem is
divided up into: open proles, closed proles and solid sections. Examples on
each type is given and comparisons are made.
The report has 2 appendices. The rst lists a MatLab procedure for calculating
sectional properties for a general polygonal cross-section. The second gives
solutions for a number of parametric proles e.g. I, U and Z-proles.

i

Preface
The notes are used for a preliminary course in Structural Analysis. The aim is to give a
sucient background for analysis of stiness and stresses in beam-sections. The theory
is formulated as direct as possible and often illustrated by an example. The methods are
general but emphasize has been on explaining the dierent steps.
The axial and bending problems are solved based on the assumption of "plane sections
remain plane". Methods to calculate the elastic center and principal axis are given including
a computer algorithm. The stress distribution is given through Navier's formula in a
formulation that allows for dierent materials in a cross-section.
The shear and torsion problem is treated in less detail. Grashof's formula for shearing
forces is given and illustrated by examples. The torsion problem is divided up into: open
proles, closed proles and solid sections. Only results are given, and for a more detailed
discussion the reader has to consult other textbooks.
Lyngby, November 2000
Lars Damkilde

ii

. . . . . . . . . . . . . . . 3. . Shear forces . . . . . . . . . . . . . . . . . . . . 2 Bending moments and axial force 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 3. . . . . . . . .5 2. . . . . . . Torsional stiness and shear stresses 3. . . . . . . .2 2. . . . . . . . . . . . .Grashof . . . . . . . . . . . . . . . . . . .1 Open prole . . General cross-section . . . . .4. . . . . . . . . . . . Normal and shear stresses . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 3. . . . . . 3 Shear forces and torsional moment 3. . . . . . . . . . . . . . y -axis . . . . . . . . . . . . . . . . . .1 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example on shear stresses . . . . . . . . . . . . . . . . . . . .Table of Contents 1 Beam theory 1. . . . . . . . . . . . . . . . . . . . . . . .Navier . . . . . . . . . . . . . . . . . . . . . . . .3 Solid section . . . . . .6 2. . Numerical methods . . . . 3. . . . . . .4 2. . . . . .2 3. .4. . . . . . . . .2 Closed prole . . . . . .2 1 Stiness properties . . . . 16 18 19 21 21 23 24 References 27 A MatLab procedure for polygon 28 B Solutions for some parametric proles 30 iii . . . Cross-sections of dierent materials . . . . . . . . . . . . . . . . . . . . Reference system .7 5 Basic assumption . . . . . . . Example: Cross-section symmetric about Stress distribution . . .3 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 2 3 5 6 7 9 10 12 14 16 Shear in rectangular cross-section . . . . . . . . .1 1. . . .

and the dierences lie in the simplications. and simplied continuum mechanics solutions are again applied. Figure 1. In all beam theories the cross-sectional stiness properties have to be determined. Mz . The internal forces also have to be translated into stress distributions over the cross-section. The internal forces . The nal solutions are not entirely consistent as they are a result of approximations but in many practical applications it is of no importance. The internal forces represent normal and shear stress distributions over the cross-section. The cross-sectional geometry is represented by a number of scalar quantities such as area and moments of inertia. 1 . and the beam axis denoted x is in the length direction. and the displacements in the cross-section are direct related to the deformation of the beamaxis. length.Chapter 1 Beam theory Beams are structural elements where the length is several times larger than the dimensions in any of the two other directions. loads and boundary conditions. The validity of the beam theory depends on the cross-sectional geometry. and simplied continuum mechanics solutions are used. shear forces Vz . All beam theories describe the beam as a 1-dimensional structure.1: Sectional forces Beam theory is a simplication of the 3-dimensional continuum mechanics problem. The displacements and rotations describe the deformation of the beam-axis. Mx . N . My . Vy and bending/torsion moments. Several dierent beam theories exist.axial force.1.are also related to the beam-axis as shown in Figure 1.

so-called warping. see Grashof's formula in Chapter 3. closed prole and solid sections . but normally this has no importance. and in many cases the extra cross-sectional parameters can be hard to access.g. The theory treats axial stiness and bending stiness but disregards deformations due to shear forces. The deformation state consists of both a rotation of the cross-section and axial deformation . and it is necessary to dene a new cross-sectional stiness parameter.2 four dierent types of loadings are shown. More advanced beam theories allow deformation in the cross-sectional plane. Torsion is treated separately and is discussed later.called Timoshenko beam theory . Many of these solutions will probably in the future be solved more economically by a full 3-Dimensional Finite Element solution based on shell or solid elements. The theory is widely used. 2 . where E is the elastic modulus for the material and A the cross-sectional area. i. In case a) the loading gives axial force.1 Stiness properties Analysis of statically indeterminate frames requires information of the stiness of the individual beam elements.e. Vlasov developed a theory which could describe non-uniform torsion of a beam including non-uniform warping of the cross-section. Torsion is in both Euler-Bernoulli and Timoshenko beams treated as a separate problem. 1. The simplication is often reasonable as deformations due to bending normally are much larger. The torsional stiness depends very much on the cross-section.or deformation-method. the so-called warping stiness.The most simple theory which is called Euler-Bernoulli theory assumes that the crosssection remains orthogonal to the beam axis. In all of the above theories it is assumed that the cross-section retains its form. A more sophisticated beam theory . e. The shear forces can not be determined via strains and have to be found through equilibrium conditions. For thin-walled structures torsion is a more complicated problem.are discussed. The complexity of the theories become high.open prole. and this allows a direct representation of the shear forces. and the stiness parameter is EA. The shear strain distribution over the cross section is uniform and more advanced beam theories allow a more realistic distribution. in manual calculation of frames via the force. The idea is to allow a deformation between the beam axis and the cross-section. A shear exible beam theory is relevant for low ratios of beam length to beam height or for materials with a low shear stiness or for thin-walled proles with a thin web. The non-uniform warping results in normal stresses. Deformation in the beam sections own plane can be of importance if the prole has parts with small thickness. In this way a shear strain is dened. and is suciently as long as the cross-section is small and compact. Torsion is in this context treated as homogeneous torsion. This solution does not comply with the boundary conditions. In some cases the contribution to the axial deformation is disregarded.takes the shear deformation into account. that each cross-sectional part deforms equally. and in chapter 3 dierent types . In Figure 1. The Vlasov beam theory is important for open thin-walled steel proles especially in connection with stability phenomena such a lateral buckling.

In this respect it is important to know the corresponding distribution of stresses. In lateral buckling the critical load is a function of both the torsional stiness and the bending stiness about the week axis. Chapter 2 deals with calculation of the principal moments of inertia. The stiness parameters are EIz and EIy . where Iy and Iz are the moments of inertia about the principal axis in the crosssection. no shear force. The shear deformations are often neglected. The stiness parameter is GAk . and is the most economically way to exploit the structure. The stresses in a cross-section can be divided into normal-stresses. where G is the shear modulus of the material and Iv the torsional moment of inertia.and y -axis are often described as bending about strong or week axis. the part of the area that carries the shear force. In chapter 3 solutions for open proles. The corresponding stiness parameter is GIv .Figure 1. i. which are perpendicular to the cross-section and shear stresses which act in the cross-sectional plane.e. Bending about the strong axis has the largest moment of inertia. In case d) the loading is torsion. Based on these informations the strength of the cross-section is evaluated.g. In chapter 3 an example with shear deformation is given. but torsion can be part of a stability problem. The normal stresses correspond to axial force and bending moments as shown in Figure 3 .2 Normal and shear stresses The result of a frame analysis is the internal forces in each beam-section. It is important to notice that a closed prole has a much larger stiness than an open prole. Structures normally do not carry load by torsional moments. closed proles and solid sections are given. Bending about the z . where G is the shear modulus of the material and Ak the so-called shear area i. Bending about the week axis is often only due to secondary loadings or due to stability problems e. 1.2: Beam stiness In case b) the loading is pure bending. In case c) the loading is bending and shear. but can be of importance for thin-walled proles or for a low ratio of beam length to beam height. lateral buckling.e.

closed prole and solid section . 4 .open prole.are presented. which gives the normal stress distribution on the cross-section. Figure 1. Figure 1.3. In Chapter 3 solutions for three dierent types . The distribution of shear stresses from a shear force is shown in Figure 1.5. Figure 1.3: Normal stresses from axial force and bending moments The shearing stresses correspond to shear forces in two directions and a torsional moment. Chapter 2 deals with analysis of the stress distribution from axial force and bending moments.1.4: Shear stresses from shear force The distribution of shear stresses form a torsional moment on a solid section is shown in Figure 1. The so-called Navier's formula is derived.5: Shear stresses from torsional moment The general distribution of shear stresses form torsional moments is complicated.4. Chapter 3 deals with calculation of the shear stresses based on Grashof's formula.

(2. 5 . The orientation of the axis set is determined so that a curvature about the z -axis only gives moment about the z -axis and similarly for the y -direction. The curvature κy is due to movements in the z -direction and gives normal strains which vary linear with z . The principal moments of inertia are dened and illustrated by examples including cross-sections made up of dierent materials. Finally an algorithmic formulation for a triangulated prole is given. The sign is determined so that a positive curvature gives compression in the top of the beam (positive y ). The curvature κz is due to displacements in the y -direction and gives normal strains which vary linear with y . First a prole symmetric about the z -axis is examined and afterwards a general cross-section is treated. The basic assumption gives a linear strain distribution over the cross-section. The sign is according to practice chosen positive.Chapter 2 Bending moments and axial force This chapter deals with the stress distribution caused by bending moments and axial force. The axial elongation is given by εC where C refers to the elastic center of the cross-section. This has no physical importance but denes how the curvature and displacement in y -direction are related." This assumption implies that the normal strain distribution over the cross-section also has to be linear. The stress distribution over the cross-section is given by the so-called Navier's formula.1) εx = εC − κz y + κy z In order to simplify the formulas for stress distribution over the cross-section a coordinate system through the elastic center is used. The elastic center and principal axis are introduced in two steps. 2. The magnitude of the normal strain depends on the curvatures of the beam and the axial elongation. The stresses will be normal stresses perpendicular to the cross-section.1 Basic assumption The basic assumption in beam theory is that: "Plane sections remain plane during deformation.

The static moment of the whole cross-section can be taken as a sum of individual contributions.3) The static moments about the original axes y and z can be calculated by dividing the cross-section into simple shapes e.2) (z − zC )dA = 0 The coordinates of the elastic center can then be calculated by: R ydA R A zdA = A A yC = zC A (2. and a system through the so-called elastic center y 0 . The contribution for a part of the cross-section can be given as the area of the part.and z .1 shows a cross-section with two sets of axis. yP . 2. In the following sections a general method for establishing this reference system is given. The position C = (yC . The distance can be both positive and negative. The task is to determine the elastic center. Figure 2. zC ) is dened so that the static moment about the y 0 .1: Cross-section Z ZA A (y − yC )dA = 0 (2.and z 0 . multiplied by the distance from the center of gravity of the part to the axis.g.4) ∆Sz = ∆A yP 6 .2 Reference system Figure 2. which can be compared to the center of gravity.and z 0 -axes through C is 0. ∆A. rectangles and triangles.The direct coupling between axial force/axial strain and bending moments/curvatures which will be shown in a following section implies that a proper reference coordinate system has to be chosen. (2. An original system y . Also formulas for calculating moment of inertia are presented.

Sz = 0. Only the upper and lower parts contribute to the static moment about the z -axis.1h = 0.5h + 1 1 0.0275 bh2 2 2 7 (2. The area of the cross-section is given by A = b 0.1h + 0.5b and the web with thickness 0.8) . Iz0 z0 is given by: Z Iz0 z0 = A (y − yC )2 dA (2.5 h + 0.6) where ∆Izl zl is the parts moment of inertia calculated about a local axis zl through the elastic center of the part.and z 0 -axis respectively.1h and width b respectively 0.25 bh (2.7) Due to symmetry zC must be 0.1 h) = 0.3 Example: Cross-section symmetric about y -axis In Figure 2.1 h) − 0.2 a cross-section symmetric about the y -axis is shown. The integral can again be calculated by dividing into simpler parts.1b h + 0. Figure 2.5) The moment of inertia about the y 0 -axis is dened similarly.5b 0. In order to calculate yC the static moment about the z -axis is calculated.1 bh(0.05 bh(0.the upper and lower part with thickness 0. The contribution from a part of the cross-section can be given as: ∆Iz0 z0 = ∆A yP2 + ∆Izl zl (2. The aim is to determine the elastic center and the moments of inertia about y 0 .1b and height h.The moment of inertia about the z 0 -axis.2: Symmetric cross-section The cross-section is divided into three parts . 2.

1h)3 + 0. The moment 1 of inertia for a rectangular cross-section is given by 12 bh3 .009458 hb3 12 12 12 (2.15) The radius of inertia iy is found similarly to iz .5b (0.1b)3 + 0. The moment of inertia Iy0 y0 is also calculated as a sum of contributions from the upperand lower-ange and the web.0508 bh3 (2. The upper ange: ∆Iz0 z0 = 1 b (0.55h + yC )2 = 0.1 bh3 + 0. The radius of inertia has to be smaller than the maximum distance from the elastic center to any point in the cross-section. This is important for the following stress analysis of the cross-section. i2y = Iy 0 y 0 = (0. Iy0 y0 = 1 1 1 0.9) The moment of inertia Iz0 z0 is calculated by three individual contributions. see Equation (2.01944 bh3 12 (2.02182 bh3 12 (2.1 bh yC2 = 0.14) The radius of inertia can be viewed as an average distance for all parts of the cross-section from the elastic center.55h − yC )2 = 0.11) 1 0. 8 .00954 bh3 12 (2.10) The lower ange: ∆Iz0 z0 = 1 0.13) The radius of inertia.The y -coordinate of the elastic center is given by yC = Sz = 0.1 hb3 + h (0.1 bh (0.12) The web: ∆Iz0 z0 = The total moment of inertia Iz0 z0 is given as the sum of the three contributions: Iz0 z0 = 0. iz is dened as: i2z = Iz0 z0 = (0.14). In the example the anges have the major contribution to the moment of inertia and therefore the value of iz is close to h/2.1h)3 + 0.05 bh (0.16) As the section is symmetric about the y -axis the so-called centrifugal moment of inertia R Iy0 z0 = A y 0 z 0 dA = 0.451 h)2 A (2.1945 b)2 A (2.1 h (0.11h A (2.5b)3 = 0.

The stress distribution .21) Through the formulas (2.19)-(2. and similar for a bending moment My .18) σx = Eεx The axial force.21) the strain distribution from (2.17) yzdA = Iyz = 0 The normal stress σx is given by Hooke's law as: (2. In the previous section the mixed moment of inertia is 0 due to symmetry.22) This formula shows that a bending moment Mz only gives a curvature κz .Navier's formula . This decoupling is a consequence of the chosen coordinate system. N .is given by: σx = N Mz My − y+ z A Izz Iyy (2. This means that the axis set has origo in the elastic center. is Z N= Z A σx dA = A (2. As the axes are principal axes we have: Z ZA Z A A ydA = 0 zdA = 0 (2.and z -axis are principal axes.4 Stress distribution . see Equation (2.23) 9 .19) Eεx dA = EAεC The moment about the z -axis is given by Z Mz = Z A −yσx dA = A −yEεx dA = EIzz κz (2. In order to nd the resulting force and moments the stress distribution is integrated over the cross-section.2.Navier In this section the y .20) The moment about the y -axis is given by Z My = A Z zσx dA = A zEεx dA = EIyy κy (2. The axes in this system are therefore described as principal axes. and in the following section a general method for establishing a set of principal axes is given. and that the mixed moment of inertia Iyz = 0.1).1) can be rewritten as: εx = N Mz My − y+ z EA EIzz EIyy (2.

3: General cross-section The elastic center is dened by Equation (2.2h2 × 0.5 General cross-section The method is illustrated by a non-symmetric prole shown in Figure 2. For non-symmetric proles the moment of resistance is not very useful as maximum points can be found for dierent numeric values of the z .12h2 × 0. For symmetric proles it is often convenient to introduce the moment of resistance.3h = 0.2h2 × 0.The maximal stresses are found on the contour of the prole.26) Mmax = Wz σY where σY denes the maximum allowable stress for the material.2h.and z -axes are: Sz = 0. The static moments about the y .12h2 × 0.6h × 0. Wz . 2.or y -coordinate. The moment of resistance for bending about the y -axis.3.176h3 10 (2.27) .208h3 Sy = 0. Figure 2. Wy .7h + 0. The cross-section is divided into 2 parts: A rectangle with dimensions h × 0.25) or the moment capacity of a cross-section can be written as (2. as: Wz = Izz ymax (2.2).5h + 0. is dened in a similar manner.2h and a rectangle with dimensions 0.24) The numerical value of the maximum stress from a bending moment Mz is given by: σbending = Mz Wz (2.9h = 0.

Iz0 z0 and Iy0 y0 are: 1 1 0.2h2 + 0.2h)3 + (0. y 00 = y 0 cos v + z 0 sin v z 00 = −y 0 sin v + z 0 cos v (2.6h(0.12h2 = 0.28) The moments of inertia about axes through the elastic center and parallel to the y .2h)3 + 12 12 (0.31) ∆Iz0 y0 = ∆A yP zP where ∆A is the area of the part and the elastic center for the part has coordinates (yP .12h2 = 0. and the transformation formulas gives the relations. The coordinate transformation is inserted in the denition of Iy00 z00 as: Z A Z y 00 z 00 dA = A [(−y 02 + z 02 ) cos v sin v + y 0 z 0 (cos2 v − sin2 v)]dA = 0 11 (2.33) The idea is to determine the angle v so that the centrifugal moment disappears.02907h4 1 1 = h(0. and the elastic center is dened by: yC = Sz = 0. v .9h − yC )2 0.34) .and z -axis.5h − yC )2 0.7h − zC )(0.30) The integral can be calculated by dividing the cross-section into simpler parts.e. The coordinates in the rotated system are denoted y 00 and z 00 .65h A and zC = Sy = 0.32h2 .6h)3 + 12 12 (0.01627h4 Iz0 z0 = Iy0 y0 (2.3h − zC )2 0.29) The centrifugal moment of inertia Iz0 y0 is dened by: Z Iy0 z0 = A y 0 z 0 dA (2. is measured anti-clockwise i.9h − yC ) = −0. from the y -axis towards the z -axis. Contribution from a rectangular part with sides parallel to the coordinate axes is: (2.5h − yC ) + 0.012h4 (2.12h2 (0. The rotation angle.32) In order to handle the stress distribution in a simple manner the coordinate system through the elastic center is rotated so that the centrifugal moment of inertia Iy0 z0 vanishes.2h2 + 0.2h(0.55h A (2.7h − zC )2 0.2h2 (0.The cross-section has A = 0.2hh3 + (0. zP ). In the example Iy0 z0 is given by: Iy0 z0 = 0.3h − zC )(0.

6 Cross-sections of dierent materials In Figure 2. Iz0 z0 and Iy0 z0 together with sin 2v = 2 sin v cos v and cos 2v = cos2 v − sin2 v Equation (2.00907h4 (2.96o (2. but the nal result namely strains and stresses are independent.02907h4 − 0.39) 12 .38) These two moments of inertia are the principal moments of inertia.4 a cross-section made up of dierent materials is shown. From Equation (2. is given below.35) The moment of inertia Iy00 y00 and Iz00 z00 can be found be inserting the coordinate transformation in the denition: Iz00 z00 = Iz0 z0 cos2 v + Iy0 y0 sin2 v + Iy0 z0 sin 2v Iy00 y00 = Iy0 y0 cos2 v + Iz0 z0 sin2 v − Iy0 z0 sin 2v (2. The section is symmetric about the y -axis which makes the calculation easier but the principles are the same for more general cross-sections.36) In the example the angle v is determined as: tan 2v = −2 0. These properties depend on the reference. and it is seen that stier parts contribute relatively more than more exible parts. The transformed area. At . At = E1 E2 b1 t1 + hb0 + b2 t2 E0 E0 (2. Use of dierent materials in a cross-section is normal in concrete and timber structures and also in new composite materials.01627h4 ⇒ v = −30.37) and the moments of inertia as Iz00 z00 = 0.012h4 0. and calculate the so-called transformed cross-sectional properties. The basic idea in the analysis is to choose a reference E -modulus. The transformation is equivalent to the transformation between stresses and principal stresses.36) it is easily seen that the sum of the moments of inertia is identical in the two dierent coordinate systems y 0 − z 0 and y 00 − z 00 .03627h4 and Iy00 y00 = 0.Using the denitions of Iy0 y0 . 2. In this example the reference is chosen as E0 .34) can be stated as: tan 2v = 2Iy0 z0 Iz0 z0 − Iy0 y0 (2.

The stresses in the dierent parts of the material depend on the E -modulus. The moment of inertia Iz0 z0 t around an z 0 -axis through the elastic center is calculated as: Iz0 z0 t = E1 1 1 1 ( b1 t31 + b1 t1 ( (h + t1 ) − yC )2 + ( b0 h3 + b0 hyC2 + E0 12 2 12 E2 1 1 ( b2 t3 + b2 t2 ( (h + t2 ) + yC )2 E0 12 2 2 (2.40) The coordinate of the elastic center yC is given as: yC = Szt At (2. In material i the stress is: σx = Ei N Mz 0 My ( − y + z) E0 At Iz0 z0 t Iyyt (2.4: Cross-section with dierent materials The transformed static moment Szt is calculated along the same line as: Szt = E1 1 E2 1 b1 t1 ( (h + t1 )) − b2 t2 ( (h + t2 )) E0 2 E0 2 (2.44) 13 .42) The linear strain distribution is related to the internal forces N . Mz and My as: εx = N Mz My − y0 + z 0 0 E0 At E0 Iz z t E0 Iyyt (2.43) where the coordinate y 0 is measured from the z 0 -axis through the elastic center.41) It is noticed that the coordinate of the elastic center does not depend on the choice of reference E -modulus.Figure 2.

5: Triangular element The area of the triangle. The stresses are independent of the chosen reference E -modulus. which otherwise would have lead to a physically unsound method.and y -axis are given by: Z 1 (y1 + y2 + y3 )A 3 A Z 1 zdA = (z1 + z2 + z3 )A 3 A ydA = (2.The stress-distribution is linear within the individual parts. A. The static moment about the z . More details on this can be found in various textbooks on the Finite Element Method. is given by: 1 A = ((y2 − y1 )(z3 − z1 ) − (z2 − z1 )(y3 − y1 )) 2 (2. but not over the entire crosssection due to discontinuities in the E -modulus. Figure 2.5 are given below. The formulas for the basic triangular element shown in Figure 2.g (Zienkiewicz 1977).46) The elastic center (yC .7 Numerical methods A computer method for calculating the cross-sectional properties can be based on a subdivision of the cross-section into triangles. Dierent material properties in the cross-section can be handled by adding individual E -modules to each triangle. The numbering of the nodal points in the triangle denes the orientation. zc ) is given by: 1 yC = (y1 + y2 + y3 ) 3 and 1 zC = (z1 + z2 + z3 ) 3 14 (2. and in order to get a positive area the numbering should be anti-clockwise.45) The following formulas are calculated by means of the so-called area coordinates or triangular coordinates. Direct integration in the y − z system is a tedious process. see e. 2. The fundamental assumption that a plane in the cross-section remains plane is still valid.47) .

Iyy . Determine the elastic center and make a coordinate transformation which places Origo in the elastic center. 2. The polygon should be numbered anti-clockwise.The moments of inertia. In Figure 2. 4. 3. Iyy and Iyz are given by: Z 1 2 (y1 + y22 + y32 + 9yC2 )A 12 A Z 1 2 2 z dA = (z1 + z22 + z32 + 9zC2 )A 12 A Z 1 yzdA = (y1 z1 + y2 z2 + y3 z3 + 9yC zC )A 12 A y 2 dA = (2.35-2. Calculate the angle for the rotated principal system and the corresponding principal moments of inertia.6: Triangulated cross-sections In the automatic triangularization the starting point of the polygon is used as the rst point in each triangle. Loop over all triangles to nd the moments of inertia. Izz . whereas the triangles in the last example can be constructed automatically using only information on the nodal points on the contour. In this process some of the triangles can lie outside the perimeter but it is accounted for by other triangles. 1. see Equations (2. The 2 following points are for the rst triangle taken as point 2 and 3 and for next triangle as point 3 and 4 until the last triangle which is dened by the points 1. 15 . The procedure can be extended to cover more polygons with an individually assigned E -modulus. n-1 and n.6 examples of triangulated cross-sections are shown. Loop over all triangles to nd the area and static moments about the initial coordinate axis. In the rst example the triangles are constructed manually. It should be noted that the integrals can be regarded as sum of contributions of each triangle similar to the previous examples.36).48) The algorithm works in the same procedure as the previous examples. Izz . Figure 2. where n is the number of points in the polygon. The area of the triangles depend on the orientation of the triangle. Iyz . In Appendix A a procedure written in MatLab is presented which calculates the sectional properties of a polygon.

16 . based on a Finite Element discretization or some approximate solution. height h and width b is considered. see e. The rest will be based on approximate methods. Figure 3. (Gere and Timoshenko 1991) or for a more modern approach (Krenk 1989).g.1: Shear stresses on rectangular cross-section The beam is loaded with a shear force Py in the free end corresponding to x = ` and xed against displacement and rotation for x = 0. The rst deals with shear in a rectangular cross-section. and the second with torsion in an elliptic cross-section.1 Shear in rectangular cross-section As introduction to shear a rectangular beam with length `. The material is linear-elastic with Young's modulus E and Poisson's ratio ν = 0. In this context two analytical solutions are presented. (Seely and Smith 1967). 3.Chapter 3 Shear forces and torsional moment Calculation of the shear stress distribution due to a shear force or a torsional moment is a complicated problem which involves solution of a 2-dimensional boundary value problem. In the literature some exact solutions are given. There is only a very limited number of analytical solutions and for general cross-sections it is necessary to use a numerical method e.g (Timoshenko and Goodier 1951).

P σxx = Eεxx = − (` − x)y I P 5 5 y 2 3 P y 2 σxy = 2 Gεxy = ( − ( ) )= (1 − ( )) Ak 4 4 h/2 2 A h/2 (3.x ) = ( − ( )) 2 2 GAk 4 4 h/2 εxx = ux. G = 2 (1+ν) The displacements over the cross-section means that "plane sections remain plane" is no longer valid. The method is based 17 .1) 1 where the shear area Ak = 56 bh and the moment of inertia I = 12 bh3 . Without this orthogonality a false value of Ak is determined. The displacement mode which is non-planar is often described as crosssectional warping. The shear stress has a parabolic variation over the cross-section as shown in Figure 3.The displacement in direction of the beam axis is denoted ux and the displacement perpendicular to the beam-axis and along the y−axis is denoted uy . see (Gere and Timoshenko 1991) or (Krenk 1989).2) The only non-vanishing stress components are the normal stress σxx and the shear stress σxy . see Chapter 2. A discussion of energy orthogonality can be found in (Krenk 1989).1).y + uy. Integration of the shear stress over the cross-section gives the shear force P . As ν = 0 there will be no displacements in the z−direction. The warping part in Equation (3.1) is non-unique. The only non-vanishing strain components are the normal strain εxx and the shear strain εxy . The shear stiness can be calculated in an alternative way only relying on the distribution of shear stresses.1. P 1 P 1 5 1 (`x − x2 )y + ( y − y3 ) EI 2 GAk 4 12 (h/2)2 P P 1 2 1 3 x ( `x − x ) + = EI 2 6 GAk ux = − uy (3.3) The normal stress is seen to correspond to a linear variation in the bending moment.x = − εxy (3. see (Timoshenko and Goodier 1951) which gave a value of 23 A. P (` − x)y EI 1 1 P 5 5 y 2 = (ux. but by enforcing energy orthogonality between normal stress from bending and normal stress from nonuniform warping the displacement is uniquely determined. The equivalent shear area Ak denes the shear stiness together with the shear modulus E . An exact solution to this problem is given in Equation (3.

xz (3.x is the variation along the beam axis of the bending moment about the z−axis. and only the results are given. In this way an estimate of the shear stress is given by: τaverage = H t (3. This is useful when the exact solution is not known. In the cutting surfaces equivalent sectional forces are introduced equivalent to the stresses. 3. and the method will be applied in a following section dealing with I-proles. and the shearing force H the shear stresses along the cut.5) z where M. Figure 3. The part of the total moment is given by the ratio between the static moment about the z−axis and the moment of inertia.6) If the thickness of the cut.x dx I (3.2: Equilibrium .Grashof From beam theory the shear force is given by the bending moment as: Vy = M.4) Through the shear distribution an estimate of the shear area Ak can be given. Figure 3. t. Equilibrium of the part gives the so-called Grashof's formula: ∆A dF z Sz H= = M. The forces F and F +dF represent the normal stresses.on complementary energy.2 shows a part of the beam where a cut is made. is reasonably small the shearing force can be assumed to be constant.shearing force The force F represents the part of the bending moment M z which is due to the area ∆A.2 Shear forces . The complementary energy for a section of a beam is given by: Z V τ Ec = V dx = τ dAdx G Ak A G (3.7) 18 .

3. The shear force in a beam theory secures both equilibrium for the vertical direction (y−axis) and equilibrium due to changes in bending moments. But due to the symmetric stress tensor which states that σxy = σyx this is also the shear stress in the y−direction. The shearing force is: h V h H = 2 t( − zd ) × 4 2 Izz (3. It is due to the double contribution to the static moment and that the thickness is only half of the ange thickness. The shear stiness is estimated by formula (3. Even though the shearing force H is determined exact Grashof's formula is still an approximate method as the force has to be distributed over a thickness. The stress for y = h2 is four times the size in the ange.It should be noted that Grashof's formula determines the shear stresses directed in the beam axis. 19 .3. Izz = 1 h h 7 t h3 + 2 × (2 t) ( )2 = t h3 12 2 2 12 (3.3 in Appendix B. In the rst example a shear force V directed opposite to the y -axis is acting on an I-prole with the dimensions shown in Figure 3. Figure 3.4). and the value for zd = h4 is 0.10) The shear stress in the web is found by the same procedure.9) The variation of the shearing force is linear. The magnitude is determined by using Grashof's formula on the upper left ange cut for z = zd .3 Example on shear stresses In the following examples Grashof's formula is used to give an estimate of the shear stress distribution.8) The shear stress in the anges will be directed in the z−direction. The shear stress is assumed constant over the cut.3: Shear stresses in I-prole The moment of inertia about the z−axis is found by means of Table B. and the shear stress for zd = 0 is: τf = 2 t h h 4 2 V 7 th3 12 2t =3 V 14 h t (3.

The shear stiness can be calculated by means of Equation (3.11) The shearing force will vary parabolic along the web. and in the example this would underestimate the maximum shear stress with 7 %. Another common approximation is to assume a constant shear stress along the web. The rst eect is that the distribution along the web becomes more uniform than for a beam.5h 2 V = 1.3 the shear stress distribution is shown.14) Ak = 0.5h G −0.12) The remaining part of the shear stress distribution can be found by symmetry. and the value of the shear stress in the middle (yd = 0) is: 5 τw = t h2 8 V 7 th3 12 t = 15 V 14 h t (3. and it is noticed that the integral of the shear stress in the y−direction equals the shear force V .The shearing force for a cut positioned in y = yd is: 1 1 h h V H = ( t h2 + ( − yd )t( + yd )) 2 2 2 2 Izz (3.966 h t The normal assumption Ak = Aweb is seen to be very accurate.5h τ2 V 1h 1 y 2 2 2 2 dA = ( ) [4 × 2t 3 + ) ) t dy] (15 − 3( G 14 h t 34 0.4). The shear stresses in the z−direction have no resultant force as expected. In Figure 3.13) The shear area is then found to be: (3. and the second eect which is much smaller is the contribution to the strain energy. Z A Z 0. The eect of the shear stresses in the ange is two-fold.0347 Ght (3.4: Shear stresses in I-prole with dierent materials 20 . Figure 3.

Similar to the shear deformations plane sections no longer remain plane.g.g. 3. lateral or torsional buckling. In this context only results are presented and the literature contains many textbooks on thin-walled beams. For thin-walled open and closed proles the approximate results are very accurate and the methods are general. The three cuts are the critical sections for shear. The shear center will for a U-prole as shown in Figure 3. (Vlasov 1961). see (Timoshenko and Gere 1961). 21 .4 is typical for stressed-skin structures build of timber and ply-wood. An I-prole shown in Figure 3. The center of rotation is called the shear center and has the same meaning for shear forces and torsion as the elastic center has for normal forces and moments. In this context an analytical solution for an elliptic cross-section is presented together with results for a rectangular cross-section 3. A shear force through the elastic center will therefore give torsion. Shear forces acting through the shear center do not create torsion.4 Torsional stiness and shear stresses The torsion problem depends very much on the type of prole. The closed proles are a lot stier and the maximum shear stress smaller. The only dierence is that the contribution form each part should be weighted according to its stiness. Torsion in open proles is of special interest in connection with stability e. It is noticed that the choice of reference modulus has no inuence on the shear stress. For cut I-I we get: τ= tI−I Szt V 1 Izzt tI−I (3.15) tI−I where Szt is the transformed static moment about the z−axis for part I-I. (Krenk 1989). For solid sections the general problem has to be solved numerically e. Methods to calculate the shear center are outside the scope of this text but more information can be found in textbooks on thinwalled structures or e. The anges are typically somewhat stier than the timber part.1 Open prole Torsion in open proles like the U-prole shown in Figure 3.g. The torsional resistance is very dierent for the two types of proles. so-called warping of the cross-section. In Grashof's formula the static moment and moment of inertia should be replaced by the transformed (or weighted) parts.4. which theoretically rst was treated by Vlasov. Izzt the transformed moment of inertia for the cross-section. V the shear force and tI−I the thickness of the cut.5 be situated under the horizontal ange.5 results in a rotation of the prole in the y −z -plane and axial displacements in the direction of the beam axis. There is only a few analytical solutions and no simple approximation exists.For cross-sections build of dierent materials the methodology is the same. by a 2-dimensional Finite Element method.

For open proles stability often limits the load carrying capacity.g (Timoshenko and Gere 1961). Figure 3. see e.17) For the example shown in Figure 3. Each small section contributes to the torsional moment.16) The maximum shear stress will be in the parts with largest thickness.4b + 0.5: Shear stresses in open proles The shear stresses form torsion will be distributed as shown in Figure 3.6 the torsional moment of inertia is: Figure 3. The torsional moment of inertia is given as: 1 Z IV = t(s)3 ds 3 C (3.8t)3 + bt3 ) = t3 (1.512h) 3 3 (3.5. 22 . and the value is: τmax = MV t IV (3.2bt3 + 2 × h(0.The warping displacements dene the so-called warping stiness of the cross-section.18) The torsional moment of inertia is low as it depends on t3 . but the eectiveness is small as it depends on the thickness of the wall.6: Open U-prole 1 1 IV = (2 × 0.

2 Closed prole In the closed proles the shear stresses can act more eciently as shown in Figure 3.6 the torsional moment of inertia is: Figure 3. The integral of t−1 along the contour of the prole C denes an average of the thickness. τmax = MV 2AC tmin (3. It should be noted that the shear ow is constant around the prole. In the example shown in Figure 3. The torsional moment of inertia is given by: A2C −1 C t (s)ds (3.7.7: Shear stresses in closed proles 3. The warping stiness is considerably larger and therefore the warping part is of less interest.20) Similar to the open proles closed proles subjected to a torsional moment will rotate about the shear center and have axial displacements (warping). The shear stresses are constant through the thickness and create a shear ow around the prole.19) IV = 4 R where AC is the enclosed area and t is the thickness.Figure 3.20).8: Closed rectangular prole 23 .4. In this way the resulting torsional moment is proportional to the enclosed area. The maximum shear stress is in the parts with smallest thickness as given in (3.

3279( t ) IV (3.22) The dierence between an open and a closed prole is illustrated by calculating IV and τmax for an open prole with a similar geometry as in Figure 3. sin θ) (3.8 t (3. 3.8. 2π]. A parametric denition of the contour of the ellipse is: (y.4.8t + 2b 1t (3.IV = 4 b2 h2 1 2h 0. The distance from origo r is given by: r2 (θ) = a2 b2 b2 cos2 θ + a2 sin2 θ (3.5b the ratios between the torsional moments of inertia and shear stresses are: IVclosed b 2 open = 1.4911 b τmax (3.3 Solid section For an elliptic cross-section it is possible to construct an analytic solution. The shear stresses are on the other hand much larger for the open proles which is due to the less eective way of carrying stresses shown in Figure 3.28) 24 . The ellipse is dened by the two half axes a and b.7. z) = (a cos θ.26) and It is seen that the torsional stiness is many times higher for closed proles than for open proles.25) closed τmax t open = 0.21) and the maximum shear stress will be in the parts with smallest thickness: τmax = MV 2 b h 0.5 in opposition to the stress pattern shown in Figure 3. The cut that makes the closed prole open can be placed anywhere on the prole. b sin θ) = r(θ) (cos θ.23) and the shear stress is: τmax = MV t IV (3. The torsional moment of inertia is: 1 IV = (2 × h(0.27) where θ is an angle in the interval [0.24) For h = 1.8t)3 + 2 × bt3 ) 3 (3.

b) (3.30) The maximum shear stress is found at the boundary in the point with smallest distance to the center. For a circular section the warping part vanishes. On the contour the resulting shear strain is perpendicular to the normal of the contour. uθ is the displacement perpendicular to the r−direction and ux is the displacement in the direction of the beam axis.29) where ur is the displacement in the r−direction. The value is given by: τmax = 2 MV 1 π a b min (a. and the derivative of the torsion angle φ.29) as: IV = π a3 b3 a 2 + b2 (3. The angle of torsion is denoted φ. Figure 3. The displacement is a rotation around origo and a displacement in the direction of the beam axis (warping). The only non-vanishing strains are the shear strains in the θ − x plane and in the r − x plane. For a = b the section is circular and the result form Equation (3.30) corresponds to the result given in Table B. In this way it can be proven that the resulting stress distribution is equivalent to a torsional moment.9: Shear stresses in solid section 25 .2 in Appendix B. The torsional stiness is deduced from the displacement eld in (3.x 2 a + b2 (3.The displacements are conveniently dened in a polar coordinate system as: ur = 0 uθ = r φ 1 a2 − b2 ux = − r 2 2 sin 2θ φ. The maximum shear stress for a circular prole is 2 MV /(π r3 ). This corresponds to a linear shear stress distribution along the radius starting with 0 at the center.31) where MV is the torsional moment.x denes the magnitude of the torsional moment.

see e. and the solution is given as a series expansion. Similar to the elliptic cross-section the maximum shear stress is found at the point on the contour with smallest distance to the center.018( )5 ) 3 h h (3.33) The cross-sectional warping stiness is much larger than for the open proles.21 + 0. and therefore the warping part is of less interest. The shear ow is illustrated in Figure 3. The torsional moment of inertia is given by: 1 b b IV = hb3 ( − 0. 26 .g (Timoshenko and Goodier 1951).8b MV h2 b2 (3.9 and the shear stresses are largest at the contour.32) and the maximum shear stress as τmax = 3h + 1.For a rectangular section the shear distribution is more complicated.

(1951). S. The Finite Element Method. Theory of Elastic Stability. John Wiley & Sons. Theory of Elasticity. Israel Program for Scientic Translations. M. and Timoshenko. Technical University of Denmark. McGraw-Hill. F. S. Thin-Walled Elastic Beams. (1961). McGraw-Hill. C. Seely. P. part 1 and 2. Chapman and Hall. S. J. 2. Zienkiewicz. S. P. Timoshenko. (1977). J. O.References Gere. London. and Goodier. Series F 114-115. Mechanics of Materials. Timoshenko. B. V. and Smith. Jerusalem. J. Vlasov. Krenk. N. (1989). J. M.. O. Z. Advanced Mechanics of Materials. ed. Department of Structural Engineering. and Gere. 27 . McGraw-Hill. (1961). (1967). P. (1991). Three-dimensional elastic beam theory.

28 .1). 8]. % % Loop over n-2 triangles % for i=1:n-2 y2 = YZ(i+1. % % Skew profile Example section 2. 8.2).5 h = 10 % YZ = [ 0. Sy = 0.2). yg = (y1 + y2 + y3)/3. y1 = YZ(1. 10. 0. 6. z3 = YZ(i+2. zg = (z1 + z2 + z3)/3.1). 6. 8. Area = 0.5*((y2-y1)*(z3-z1)-(z2-z1)*(y3-y1)). 10.1).1).Appendix A MatLab procedure for polygon % % Cross : Calculates sectional properties for a polygon % % Coordinates for polygon. 0. z2 = YZ(i+1.2). dA = 0. % % Number of points in polygon % n = size(YZ. z1 = YZ(1. 0. 8. Sz = 0. y3 = YZ(i+2.

1) = YZ(:. 29 .2)-zc.1). z1 = YZ(1. Iyz = Iyz +dIyz.5*((y2-y1)*(z3-z1)-(z2-z1)*(y3-y1)). Iyy = Iyy +dIyy. end % % Elastic Center % yc = Sz/Area. Iyz = 0. Iyy = 0.2) = YZ(:. angle = v*180/pi. yg = (y1 + y2 + y3)/3. dIyz = (y1*z1 + y2*z2 + y3*z3 + 9*yg*zg)/12*dA.2).1). YZ(:.5*atan(2*Iyz/(Izz-Iyy)). zc = Sy/Area. z3 = YZ(i+2. dA = 0. zg = (z1 + z2 + z3)/3. dIzz = (y1^2 + y2^2 + y3^2 + 9*yg^2)/12*dA.2). % % Coordinate transformation to elastic center % YZ(:. z2 = YZ(i+1. Izz = 0. Izz = Izz +dIzz. y1 = YZ(1. Sy = Sy + dA*zg. Sz = Sz + dA*yg.2). y3 = YZ(i+2.1). end % % Transformation % v = 0. dIyy = (z1^2 + z2^2 + z3^2 + 9*zg^2)/12*dA.Area = Area + dA. I1 = Izz*cos(v)^2 + Iyy*sin(v)^2+Iyz*sin(2*v). % % Loop over n-2 triangles % for i=1:n-2 y2 = YZ(i+1. I2 = Izz*cos(v+pi/2)^2 + Iyy*sin(v+pi/2)^2+Iyz*sin(2*v+pi).1)-yc.

shear A = 2 (h tw + b tf ) Bending strong axis Izz ≈ 16 tw h3 + 12 b tf h2 Wz ≈ 13 tw h2 + b tf h Bending weak axis Iyy ≈ 12 h tw b2 + 16 tf b3 Wy ≈ h tw b + 31 tf b2 Torsion t t Iv = 2 h2 b2 h tfw+bf tw Ak = 56 A √h 12 iy = √b 12 1 b 4 ( ) )) 12 h Ak ∼ 2 h tw Table B. shear A=hb Bending strong axis 1 Izz = 12 b h3 Wz = 61 b h2 Bending weak axis 1 Iyy = 12 h b3 Wy = 16 h b2 Torsion Iv = 13 hb3 (1 − 0.1: Parametric rectangular cross-sections 30 iz = .627 hb (1 − Normal force.Appendix B Solutions for some parametric proles Geometry Cross-sectional properties Normal force.

shear A ≈ πdt Bending both axes Izz ≈ π8 t d3 Wz ≈ π4 t d2 Torsion Iv ≈ π4 t d3 Ak = iz = 96(1+ν)2 A 117+244ν+148ν 2 ∼ = 0.2: Parametric circular cross-sections Geometry Cross-sectional properties Normal force Shear z Shear y Elastic center A ≈ h tw + b1 t1f + b2 t1f Ak ∼ h tw Ak ∼ b1 t1f + b2 t1f b1 t1 −b2 t1 f f yc = A Bending strong axis 1 Izz ≈ 12 h t3w + b1 t1f ( h2 − yc )2 + b2 t2f ( h2q+ yc )2 Izz Wz ≈ h +|y iz = IAzz | 2 c Bending weak axis 1 (t1f (b1 )3 + t2f (b2 )3 ) Iyy ≈ 12 Wy ≈ max2I(byy1 .3: Parametric I-section 31 q Iyy A . shear A = π4 d2 Bending both axes π 4 d Izz = 64 π 3 Wz = 32 d Torsion π 4 Iv = 32 d Normal force.b2 ) iy = Torsion Iv = 13 (h (tw )3 + b1 (t1f )3 + b2 (t2f )3 ) Table B.8A d 4 Ak ∼ 12 A iz = d √ 2 2 Table B.Geometry Cross-sectional properties Normal force.

4: Parametric U/C-section 32 q Izz A q iy = Iyy A .Geometry Cross-sectional properties Normal force A ≈ h t w + 2 b tf Shear z Ak ∼ h tw Shear y A k ∼ 2 b tf Elastic center b2 t zc = A f Bending strong axis 1 3 Izz ≈ 12 h tw + 12 b tf h2 Wz ≈ 16 h2 tw + b tf h iz = Bending weak axis Iyy ≈ h tw zc2 + 16 tf b3 + 2 b tf ( 2b − zc )2 Iyy Wy ≈ b−z c Torsion Iv = 13 h t3w + 2 b t3f Table B.