Chapter 3: Methodology for Finite Element Analysis

23

Chapter 3

METHODOLOGY FOR FINITE ELEMENT ANALYSIS

3.1 INTRODUCTION
Finite Element calculations more and more replace analytical methods especially if problems have to be solved which are adjusted to specific tasks. In many countries, a lot of efforts are carried out to get new code standards for the calculation of ultimate load capacity of single steel angles under eccentric loadings. All these calculation methods are based on linear descriptions of the material behavior. Concerning the non-linear and time dependent characteristics of materials, standard linear elastic finite element calculations in addition to code methods are often not suitable. Therefore, a new finite element model was developed to describe the real (elastic-plastic) behavior of the single steel angle under eccentric edge loads. Besides, an exact geometric modeling the description of the material behavior of all components is very important for the quality of performed analysis. This applies to analytical as well as to numerical methods. For components made of steel elastic or elastic-plastic material laws are able to simulate the real behavior of those parts in sufficient accuracy. The actual work regarding the finite element modeling of a single steel angle connected to end plates has been described in detail in this chapter. The representation of various physical elements with the FEM (Finite Element Modeling) elements, properties assigned to them, boundary conditions, material behavior and analysis types have also been discussed. The various obstacles faced during modeling, material behavior used and details of finite element meshing were also discussed in detail.

3.2 THE FINITE ELEMENT PACKAGES

A large number of finite element analysis computer packages are available now. They vary in degree of complexity and versatility. The names of few such

Chapter 3: Methodology for Finite Element Analysis

24

packages are: ANSYS (General purpose, PC and work stations) DYNA-3D (Crash / impact analysis) SDRC/I-DEAS (Complete CAD / CAM / CAE packages) NASTRAN (General purpose FEA on main frames) ABAQUS (Non-linear and dynamic analyses) COSMOS (General purpose FEA) ALGOR (PC and work stations) PATRAN (Pre / post processor) Hyper Mesh (Pre / post processor) Of these packages ANSYS10.0 has been chosen for its versatility and relative ease of use. ANSYS is capable of modeling and analyzing a vast range of two- dimensional and three-dimensional practical problems. Buckling analysis of a real structure (calculation of buckling loads and determination of the buckling mode shape) can be performed quite satisfactorily by means of this software. . Both linear (eigenvalue) buckling and nonlinear buckling analyses are possible

3.3 FINITE ELEMENT MODELING OF THE STRUCTURE

End plate Steel angle

Applied force Figure 3.1: General sketch of a single steel angle with end plates at its both ends subjected to eccentric load.

Chapter 3: Methodology for Finite Element Analysis

25

3.3.1 Modeling of Steel Angle and End Plates

To facilitate the non-linear buckling analysis of the whole system, modeling procedure has been simplified by eliminating bolts and considering end plates at the two ends of the steel angle. Since the whole modeling was performed in 3-dimension, the element used here is 3-D in nature. For representing both the steel angle and the end plates, SHELL-181(a 4 node structural shell element) has been used. Discussion about the element is shown below in details:

SHELL181 Element Description

SHELL181 is suitable for analyzing thin to moderately-thick shell structures. It is a 4-node element with six degrees of freedom at each node: translations in the x, y, and z directions, and rotations about the x, y, and z-axes. (If the membrane option is used, the element has translational degrees of freedom only). The degenerate triangular option should only be used as filler elements in mesh generation. SHELL181 is well-suited for linear, large rotation, and/or large strain nonlinear applications. Change in shell thickness is accounted for in nonlinear analyses. In the element domain, both full and reduced integration schemes are supported. SHELL181 accounts for follower (load stiffness) effects of distributed pressures.

Figure 3.2: SHELL181 Geometry

Chapter 3: Methodology for Finite Element Analysis

26

xo x

= Element x-axis if ESYS is not provided. = Element x-axis if ESYS is provided.

SHELL181 Input Data

The geometry, node locations, and the coordinate system for this element are shown in "SHELL181 ". The element is defined by four nodes: I, J, K, and L. The element formulation is based on logarithmic strain and true stress measures. The element kinematics allows for finite membrane strains (stretching).The thickness of the shell may be defined at each of its nodes. The thickness is assumed to vary smoothly over the area of the element. If the element has a constant thickness, only TK(I) needs to be input. If the thickness is not constant, all four thicknesses must be input. A summary of the element input is given in below (Table 3.1). Table 3.1: SHELL181 Input Summary Element name Nodes SHELL181 I, J, K, L

Degrees of Freedom

UX, UY, UZ, ROTX, ROTY, ROTZ if KEYOPT (1) = 0 UX, UY, UZ if KEYOPT (1) = 1

Real Constants TK(I), TK(J), TK(K), TK(L), THETA, ADMSUA E11, E22, E12, DRILL, MEMBRANE, BENDING EX, EY, EZ, (PRXY, PRYZ, PRXZ, or NUXY, NUYZ, NUXZ), ALPX, ALPY, ALPZ (or CTEX, CTEY, CTEZ or THSX, THSY, THSZ), DENS, GXY, GYZ, GXZ

Material Properties

Chapter 3: Methodology for Finite Element Analysis

27

SHELL181 Assumptions and Restrictions:

Zero area elements are not allowed (this occurs most often whenever the elements are not numbered properly).

Zero thickness elements or elements tapering down to a zero thickness at any corner are not allowed (but zero thickness layers are allowed).

In a nonlinear analysis, the solution is terminated if the thickness at any integration point that was defined with a nonzero thickness vanishes (within a small numerical tolerance).

This element works best with full Newton-Raphson solution scheme. The through-thickness stress, SZ, is always zero.

3.3.2 Material properties

Bilinear Kinematic Hardening The Bilinear Kinematic Hardening (BKIN) option assumes the total stress range is equal to twice the yield stress, This option is recommended for general small-strain use for materials that obey von Mises yield criteria (which includes most metals). It is not recommended for large-strain applications.

y 1
Stress,

E2

T1

E1 1 y Strain, Figure 3.3: Bilenear kinematic hardening

Chapter 3: Methodology for Finite Element Analysis

28

Where,

y = Yield stress
y = strain corresponding to yield stress E1 = Modulus of elasticity upto yield point E2 = Modulus of elasticity after exceeding yield point T1 = Temperature for material 1.

3.4 TYPES OF BUCKLING ANALYSES

Two techniques are available for performing buckling analyses - nonlinear buckling analysis and eigenvalue (or linear) buckling analysis. These two methods frequently yield quite different results.

3.4.1 Nonlinear Buckling Analysis

Nonlinear buckling analysis is usually the more accurate approach. It is recommended for design or evaluation of actual structures. This technique employs a nonlinear static analysis with gradually increasing loads to seek the load level at which a structure becomes unstable. Using the nonlinear technique, models can include features such as initial imperfections, plastic behavior, gaps, and largedeflection response. In addition, using deflection-controlled loading, the postbuckled performance of structures (which can be useful in cases where the structure buckles into a stable configuration, such as "snap-through" buckling of a shallow dome)can be obtained (Figure 3.4: Buckling Curves (a)).

3.4.2 Eigenvalue Buckling Analysis

The second method, eigenvalue buckling analysis, predicts the theoretical buckling strength (the bifurcation point) of an ideal linear elastic structure. This method corresponds to the textbook approach to elastic buckling analysis: for instance, an eigenvalue buckling analysis of a column will match the classical Euler solution. However, imperfections and nonlinearities prevent most real-world structures from achieving their theoretical elastic buckling strength. Thus, eigenvalue buckling analysis often yields unconservative results, and is not generally used in actual day-to-day engineering analyses (Figure 3.4: "Buckling Curves (b)).

Chapter 3: Methodology for Finite Element Analysis

29

(a) Nonlinear load-deflection curve (b) Linear (Eigenvalue) buckling curve Figure 3.4: Buckling Curves The Arc-Length Method: One major characteristic of nonlinear buckling, as opposed to eigenvalue buckling, is that nonlinear buckling phenomenon includes a region of instability in the post-buckling region whereas eigenvalue buckling only involves linear, prebuckling behavior up to the bifurcation (critical loading) point (Figure 3.5).

Figure 3.5: Nonlinear vs. Eigenvalue Buckling Behavior The unstable region above is also known as the snap through region, where the structure snaps through from one stable region to another. To illustrate, consider the shallow arch loaded (Figure 3.6) may be considered.

Chapter 3: Methodology for Finite Element Analysis

30

Figure 3.6: Snap Through Buckling For most nonlinear analyses, the Newton-Raphson method is used to converge the solution at each time step along the force deflection curve. The Newton-Raphson method works by iterating the equation [KT]{u}={Fa}-{Fnr}, where {Fa} is the applied load vector and {Fnr} is the internal load vector, until the residual, {Fa}-{Fnr}, falls within a certain convergence criterion. The NewtonRaphson method increments the load a finite amount at each substep and keeps that load fixed throughout the equilibrium iterations. Because of this, it cannot converge if the tangent stiffness (the slope of the force-deflection curve at any point) is zero (Figure 3.7).

Figure 3.7: Newton - Raphson Method To avoid this problem, the arc-length method should be used for solving nonlinear post-buckling. To handle zero and negative tangent stiffness, the arclength multiplies the incremental load by a load factor, , where is between -1 and

Chapter 3: Methodology for Finite Element Analysis

31

+1. This addition introduces an extra unknown, altering the equilibrium equation slightly to [KT]{u} = {Fa}-{Fnr}. To deal with this, the arc-length method imposes another constraint, stating that

(3.1)

throughout a given time step, where is the arc-length radius. Figure 3.8 illustrates this process.

Figure 3.8: Arc-Length Methodology The arc-length method therefore allows the load and displacement to vary throughout the time step as shown (Figure 3.9).

Figure 3.9: Arc-Length Convergence Behavior

Chapter 3: Methodology for Finite Element Analysis

32

3.5 FINITE ELEMENT MODEL PARAMETERS

In this analysis small deflection and plastic materials properties (material nonlinearity) are considered. The following properties are used in the modeling. Table 3.2: Various input parameters Parameter Angle dimension in X-direction Angle dimension in Y-direction Thickness of angle Thickness of end plate Corner dimension of end plate excluding the angle (on each side) Center of gravity of bolt pattern Youngs modulus of elasticity Yields stress for the angle Poisons ratio Applied load 38 mm 200 KN/mm2 .3259 KN/mm2 .3 Slightly greater than critical buckling load for the steel angle Reference value 102 mm 102 mm 6 mm 25 mm 25 mm

3.6 MESHING 3.6.1 Meshing of the End Plate

The end plate is divided along both of its axes (X-axis and Y-axis in the global co-ordinate system). Individual division is rectangular. Number of division is chosen in such a way that the aspect ratio is of the element is reasonable.

3.6.2 Meshing of the Steel Angle

The length of the steel angle (in Z-direction) perpendicular to the axis of the end plate is divided into sufficient number of divisions. Individual division is rectangular. Number of division is chosen in such a way that the aspect ratio is of the element is reasonable.

Chapter 3: Methodology for Finite Element Analysis

33

3.7 BOUNDARY CONDITIONS 3.7.1 Restraints

In case of the bottom end plate, corresponding node at the location of center of gravity of the bolt pattern, which simulate the contact, is kept restrained in all direction(in X,Y and Z-directions). But, in case of the the top end plate, the node at the same location, was restrained in X and Y direction only to allow the free deflection in the Z direction (along the length of the steel angle).At the right most corner of the top end plate, only the deflection in the X direction has been kept restrained. In all cases, the whole model is kept unrestrained against rotation. These options are allowed to facilitate the non-linear buckling analysis of the system.

3.7.2 Load
In this case, the steel angle is given an applied load slightly higher than its critical buckling load at node at the location of center of gravity of the bolt pattern on the top end plate along the length of the steel angle.

Figure 3.10: Preliminary model of a single steel angle connected to end plates at its both ends (prior to meshing)

Chapter 3: Methodology for Finite Element Analysis

34

Figure 3.11: Finite elements mesh of the steel angle with end plates at its both ends

Figure 3.12: Finite elements mesh with loads and boundary conditions

Chapter 3: Methodology for Finite Element Analysis

35

Figure 3.13: Typical deflected shape of the model

450 400 350 300 10 40

l/r
20 90 130

Load, kN

250 200 150 100 50 0 0.00

110

1.00

2.00

3.00

4.00

5.00

6.00

Displacement, mm

Figure 3.14: Typical load vs deflection curve for different slenderness ratio obtained from
non-linear buckling analysis of L102x102x6 steel angle.