Sie sind auf Seite 1von 8

6 Beam Elements

For the sake of simplicity we are restricting ourselves to plane problems in this section, i.e. the presented beam theories according to Timoshenko and (Euler-) Bernoulli establish a dimensional reduction from the two-dimensional plane stress problem to one spatial dimension. Thus, uniaxial bending is considered and torsion is excluded. Dimensional reduction is predominantly realized by means of kinematic assumptions. Some of these assumptions are identical for both theories (cross sections remaining plane during deformation), some are different (consideration or neglect of transverse shear strain effects). Section 6.1 introduces the more complex Timoshenko beam theory. This is done first because in fact in the context of a finite element formulation this model appears to be much simpler than the simpler beam theory according to Bernoulli (Section 6.2). The latter is usually applied for manual computations and classical static calculations. 6.1 Shear deformable (Timoshenko-) beam element

6.1.1 Governing equations The continuous, one-dimensional Timoshenko beam theory utilizes two degrees of freedom per material point, namely the deflection w and cross sectional rotation . Both quantities are independent of each other in the first place and are used to fully described the behavior of the two-dimensional body of the beam: w provides the deflection of the neutral axis (beam axis or center line), indicates the rotation of cross-sectional fibers which are initially normal to the center line. This means, that cross sectional areas (fibers) remain straight but they are not necessarily normal to the beam axis after deformation. The deviation from the normal diretion is the shear angle (125) which consequently consists of contributions from the gradient of the deflection (i.e. the rotation of a material fiber which is tangential to the mid axis) and the cross-sectional rotation . If both quantities have the same absolute value (with different signs, see definition of coordinate system in Figure 6.1) the shear angle is zero (cf. Equation (144) for the Bernoulli beam, where = 0 is an a priori assumption).

Fig. 6.1: Timoshenko beam Transverse normal stresses z are assumed to vanish as well. Figure 6.1 shows a typical, plane Timoshenko beam problem. Without loss of generality we are assuming displacement boundary conditions on the left and force boundary conditions at the right end of the beam. The kinematic quantities (strains) of Timoshenko beam theory are the curvature and the shear angle 47

(126) The corresponding material equations read (127) The shear stiffness is usually multiplied by a shear correction factor to take into account a realistic shear strain distribution on the cross sectional area. We are assuming here that this is already taken into account in the value A representing the cross sectional area. Along with displacement boundary conditions (128) and force boundary conditions (129) all equations which are necessary for the derivation of a finite element formulation are readily available. For the sake of simplicity, the equilibrium equations are given here as well (130) 6.1.2 Virtual work principle The virtual work principle for the Timoshenko beam depicted in Figure 6.1 reads

(131)

It can be derived from the differential equations with the help of the method of weighted residuals or if a potential exists from the principle of minimum potential energy, using methods of variational calculus. The underlying mathematical tools go beyond the scope of this course. However, as long as linear-elastic models in structural or solid mechanics are treated the virtual work principle can always be constructed according to the following rules: The internal virtual work equals the integral over the domain of the internal forces (here: stress resultants V and M) multiplied with their energetically conjugate virtual kinematic quantities (here: shear angle and curvature ). The (negative) external virtual work equals the integral over the domain of the external forces (here: the transverse load q and the distributed moment load m) multiplied with their energetically conjugate virtual displacements (here: transverse deflection w and cross sectional rotation ) plus the corresponding contributions from the force boundary conditions. Thus, in order to be able to specify the virtual work expression one only needs to know the involved variables (stress resultants, or stresses, kinematic variables or strains, forces, moments, displacements, rotations) and their energetic correspondence. Introducing material equations into the expression for the internal energy provides

48

(132)

Equation (132) will be used as a basis in the subsequent section for the finite element discretization. Expressing the kinematic quantities in terms of w and finally yields

(133)

representing the virtual work principle for the Timoshenko beam, expressed in the primal variables w and . This format of the equation is useful in the context of determining requirements for the finite element shape functions (cf. 2nd paragraph of the next section). 6.1.3 Finite element formulation As a prerequisite for a feasible finite element formulation we have to take two major decisions: Choose the quantities which ought to be directly discretized (i.e. the primary variables) and choose an appropriate parameterization by nodal values and shape functions. The primary variables have already been implicitly chosen to be w and in the previous section. For a feasible parameterization it is important to make sure that the integrals in Equation (133) exist. Achieving this is not very demanding in this case, because only first derivatives of the primal variables show. This is in fact the reason for the simplicity and attractivity of the Timoshenko beam formulation in the context of finite elements. Ensuring existence of the integrals will be more demanding in the case of the Bernoulli beam. Therefore, a discussion of this topic is postponed to Section 6.2.3. For discretization of both transverse deflection and cross-sectional rotation linear shape functions may be applied

(134) The corresponding finite element has two nodes with two degrees of freedom each (see Figure 6. 2).

Abb. 6. 2: Timoshenko beam element Like in the previous chapters we are utilizing the virtual work principle for one generic element with pure force boundary conditions for derivation of the expressions for element stiff49

ness matrix and vector of consistent nodal forces. In analogy to Equation (132) the virtual work principle for one single element reads

(135)

The left-hand side provides the expression for the element stiffness matrix and from the right hand side the vector of consistent nodal forces may be derived. All that is left to be done is to introduce the discretization (134). As usual, the discretized equations are expressed in matrix form, starting with Equation (134)

(136)

The kinematic equations then read

(137)

and the material equation is (138)

Finally, the prescribed body forces and nodal forces are collected within vectors

(139)

The derivatives of the shape functions, needed in the kinematic equations, are (140) (cf. Chapter 4). Transformation of the integral into the local coordinate system is also identical to the one already conducted in the context of one-dimensional elasticity problems. (141) After these formal preparations, the discretization (134) can be introduced into the virtual work principle (135) 50

(142) From this, the definitions of the element stiffness matrix and the vector of consistent nodal forces can be extracted as (143) The format of these equations is well-known and in fact identical to the one obtained in previous chapters. 6.2 Bernoulli Beam Element

6.2.1 Governing equations In contrast to the shear deformable formulation according to Timoshenko, Bernoulli beam theory (sometimes also termed Euler-Bernoulli beam theory) possesses only one single degree of freedom per material point, namely transverse deflection w. It is assumed that cross sectional areas (material fibers) remain straight and normal to the mid-axis during deformation. Due to this normality assumption, cross-sectional rotations directly depend on the gradient of the deflection (144) In other words, transverse shear strains are neglected (cf. Equation (126)2 with shear angle = 0). As in Timoshenko beam theory, transverse normal stresses z are neglected as well. In Figure 6.3 a typical Bernoulli beam problem is shown. It resembles in most parts the one for the Timoshenko model. The major difference is the fact that only one field variable w(x) exists. We also omit distributed bending moments here for the sake of simplicity. Moreover, there is a difference in the displacement boundary conditions, prescribing the first derivative of w rather than the rotation (which is not a free variable).

Fig. 6.3: Bernoulli beam The only kinematic quantity (strain) in the Bernoulli model is curvature sponding kinematic equation reads and the corre-

(145) The quantity which is energetically conjugate to the curvature is the bending moment (146) and the corresponding stiffness EI is called the bending stiffness of the beam. These are already all equations which are necessary for the construction of a finite element formulation. 51

It may be surprising that the transverse shear force V has not shown up so far. It does not appear in the material equations because its energetically conjugate quantity is assumed to vanish a priori. This comes along with the assumption of infinite transverse shear stiffness. The material equation therefore degenerates to something like infinity times zero an indeterminate expression. Transverse shear forces can be obtained from the equilibrium equations (147) though. The displacement boundary conditions (at and the force boundary conditions (at in our example) are (148) ) read (149) 6.2.2 Virtual work principle The virtual work principle for the Bernoulli beam model is obtained from the same considerations as in the case of Timoshenkos formulation in Section 6.1.2. (150) Introducing kinematic and material equations leads to (151) The fundamental difference of this expression in comparison with Equation (133) is the fact that second derivatives of the primal variable w show up. This results in stronger requirements on the shape functions used within the finite element formulation. 6.2.3 Finite element formulation Linear shape functions have been used for all element types discussed so far in this course. Obviously, this does not work in the case of the Bernoulli beam because then and thus strain energy was identically zero in Equation (151). Higher order polynomials are needed. The crucial condition, however, is compatibility between the elements. Assume element-wise linear (or quadratic, cubic, etc.) shape functions. When using displacements w as nodal degrees of freedom this will result in a deflection curve with kinks, i.e. the deflection itself is continuous, but its first derivative is not; it has jumps. Such functions are called C0continuous (a Cn-continuous function is a function with its nth derivative being continuous). Consequently, the second derivative will have infinite values at nodes which are shared by two elements and, as a result of this, the integral over the entire domain is not well defined. The shape functions therefore need to be C1-continuous. A different (but related) argument leading to the same conclusion can be put forward looking at the boundary conditions: We want to be able to prescribe displacements at a node in order to model a hinged support. In order to model a clamped support some rotational degree of freedom at the node is necessary. This additional degree of freedom, in turn, automatically ensures C1-continuity when assembling the system via the direct stiffness method. 52

Figure 6.4 depicts a Bernoulli beam element with two nodes and two degrees of freedom at each node. Thus, a total of four degrees of freedom per element are obtained, like in the case of the Timoshenko beam, however with a different physical meaning.

Fig. 6.4: Bernoulli beam element As a result of this choice of nodal degrees of freedom, a second fundamental difference to the finite element formulations discussed so for comes into play. The parameterization of a certain quantity (here: deflection w) is realized with nodal values that do not only represent this quantity, but also derived values (here: ). For the derivation of related shape functions (denoted as Hermitian polynomials) similar conditions apply as before. For shape function NK (K = 14) the Kth degree of freedom is 1 and all others are zero. Consequently, N1 is a function where the value at the left node (w1) is equal to one. The derivative at this node is zero. At the right node both the functional value and the derivative are zero. For N2 the function itself is zero at both nodes, but its gradient is one at the left node, etc. The shape functions are plotted in Figure 6.5.

Fig. 6.5: Shape functions for Bernoulli beam element The ansatz for the displacements is (152) or, in matrix form 53

(153)

For the finite element formulation the kinematic and material equations are cast into matrix form as well

(154)

(155) Finally, the prescribed body forces and nodal forces are written as vectors

(156)

The formulae for the element stiffness matrix and the vector of consistent nodal forces are then formally identical to the ones for the Timoshenko beam (157)

54

Das könnte Ihnen auch gefallen