*393*
*58*
*16MB*

*English*
*Pages 0
[429]*
*Year 2016*

*Table of contents : Introduction to finite element method and ANSYS --Bar and beam structures --Solid mechanics --Vibration --Heat transfer --Fluid mechanics --Multiphysics --Meshing guide.*

ENGINEERING – GENERAL “A must read for those interested in exploring the possibilities of using the finite element method (FEM) as a practical analysis tool for structural mechanics, stress analysis, vibration, heat transfer, and fluid dynamic problems. … Alawadhi systematically introduces the theory of FEA and demonstrates a step-by-step procedure for practical analysis using ANSYS. This book is a must-have reference for students, academics, and practitioners in the field of mechanical, civil, environmental, and design engineering.” —Dr. Arun Arjunan, University of Wolverhampton, UK

Uses a Step-By-Step Technique Directed with Guided Problems and Relevant Screen Shots Simulation use is on the rise, and more practicing professionals are depending on the reliability of software to help them tackle real-world mechanical engineering problems. Finite Element Simulations Using ANSYS, Second Edition offers a basic understanding of the principles of simulation in conjunction with the application of ANSYS. Employing a step-by-step process, the book presents practical end-of-chapter problems that are solved using ANSYS and explains the physics behind them. The book examines structure, solid mechanics, vibration, heat transfer, and fluid dynamics. Each topic is treated in a way that allows for the independent study of a single subject or related chapter.

Finite Element Simulations Using ANSYS Second Edition

What’s New in the Second Edition:

• Introduces the newest methods in modeling and meshing for finite element analysis • Modifies ANSYS examples to comply with the newest version of ANSYS • Replaces many ANSYS examples used in the first edition with more general, comprehensive, and easy-to-follow examples • Adds more details to the theoretical material on the finite element • Provides increased coverage of finite element analysis for heat transfer topics • Presents open-ended, end-of-chapter problems tailored to serve as class projects Finite Element Simulations Using ANSYS, Second Edition functions as a fundamental reference for finite element analysis with ANSYS methods and procedures, as well as a guide for project and product analysis and design.

K24312 ISBN: 978-1-4822-6197-4

Esam M. Alawadhi

90000

9 781482 261974

K24312_Cover_PubGr.indd All Pages

8/10/15 11:35 AM

Finite Element Simulations Using ANSYS Second Edition

This page intentionally left blank

Finite Element Simulations Using ANSYS Second Edition Esam M. Alawadhi Kuwait University Safat, Kuwait

CRC Press Taylor & Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL 33487-2742 © 2016 by Taylor & Francis Group, LLC CRC Press is an imprint of Taylor & Francis Group, an Informa business No claim to original U.S. Government works Version Date: 20150728 International Standard Book Number-13: 978-1-4822-6198-1 (eBook - PDF) This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint. Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright.com (http:// www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com and the CRC Press Web site at http://www.crcpress.com

Contents

Preface ix Acknowledgments xi About the author xiii 1 Introduction to finite element method and ANSYS 1.1 1.2 1.3 1.4 1.5

The finite element method and structural analysis Stress analysis of pin-jointed bar The finite element method and thermal analysis Heat transfer through a composite wall Introduction to ANSYS software

2 Bar and beam structures

1 1 5 6 8 10 19

2.1 Finite element method for a horizontal bar element 19 2.2 Finite element method for an arbitrary-oriented bar element 23 2.3 Analyzing a plane bar structure 26 2.4 Analyzing a plane bar structure using ANSYS 29 2.5 Finite element method for a horizontal beam element 42 2.6 Analyzing a horizontal beam structure 49 2.7 Analyzing a horizontal beam structure using ANSYS 51 2.8 Development of an arbitrary-oriented beam element 67 2.9 Distributed load on a beam element 70 2.10 Analyzing beam structure under a transient loading using ANSYS 72

v

vi

Contents

3 Solid mechanics 3.1 Stress–strain relations 3.2 Development of triangular plane stress and plane strain element 3.3 Analyzing rectangular plate subjected to forces 3.4 Development of rectangular plane stress element 3.5 Analyzing a plate with a hole subjected to tensile pressure using ANSYS 3.6 Axisymmetric elements 3.7 Displacement analysis of a vessel under transient loading with ANSYS 3.8 Contact element analysis 3.9 Two horizontal cylinders in contact analysis using ANSYS 4 Vibration

95 95 100 105 109 112 126 130 145 147 171

4.1 Vibration analysis 171 4.2 Development of stiffness equations for axial vibration of a bar element 172 4.3 Natural frequencies of axial vibration of a bar element 173 4.4 Development of stiffness equations for flexural vibration of a beam element 175 4.5 Natural frequencies of the flexural vibration of a beam element 177 4.6 Natural frequencies of the flexural vibration of beam element using ANSYS 179 4.7 Development of stiffness equations for vibration of an oriented beam element 190 4.8 Harmonic vibration of a plate with holes using ANSYS 192 4.9 Three-dimensional vibration of shaft with disks using ANSYS 207 5 Heat transfer 5.1 Introduction to heat conduction 5.2 Finite element formulation for conductive heat transfer 5.3 Finite element method for one-dimensional heat conduction 5.4 Heat transfer through a composite wall 5.5 Finite element method for two-dimensional heat conduction 5.6 Heat conduction in a solid plane 5.7 Thermal analysis of fin and chip using ANSYS 5.8 Finite element method for transient heat transfer 5.9 Unsteady thermal analyses of a masonry brick using ANSYS

225 225 228 231 233 235 238 243 264 265

vii

Contents

6 Fluid mechanics 6.1 Governing equations for fluid mechanics 6.2 Finite element method for fluid mechanics 6.3 Entrance length in developing flow in a channel using ANSYS 6.4 Studying flow around a half cylinder in a channel using ANSYS 7 Multiphysics 7.1 Introduction 7.2 Thermal and structural analysis of a thermocouple using ANSYS 7.3 Chips cooling in a channel using ANSYS 8 Meshing guide 8.1 8.2 8.3 8.4

Mesh refinement Element distortion Mapped mesh Mapped mesh with ANSYS

293 293 296 298 318 343 343 345 361 393 393 395 396 398

Bibliography 411 Index 413

This page intentionally left blank

Preface

Due to the complexity of modern-day problems in mechanical e ngineering, engineers, in practice, seldom rely only on theory or experiments. The use of engineering software is becoming prevalent among academics as well as practicing engineers. Commercial finite element software such as ANSYS (ANSYS, Canonsburg, Pennsylvania) and Abaqus (Dassault Systèmes, Vélizy Villacoublay, France) for analysis and design is commonly found in use in universities, industry, and research centers. This software has become more reliable, reputable, easy to use, and trustworthy, and it allows much time to be saved. This book focuses on the use of ANSYS in solving practical mechanical engineering problems. ANSYS is extensively used in the design cycle by industry leaders in the United States and around the world. Additionally, ANSYS is available in computer labs for students in most universities around the world. Courses such as computer-aided design, modeling, and simulation, and major design courses all utilize ANSYS as a tool for analyzing various mechanical components. Graduate students also use ANSYS in their finite element courses as a complement for the theoretical study of the finite element method. This book provides mechanical engineering students and engineers with the fundamental knowledge of numerical simulation using ANSYS. The book serves most of the disciplines of mechanical engineering: structure, solid mechanics, vibration, heat transfer, and fluid dynamics, with adequate background material to explain the physics behind the computations. Each physical phenomenon is treated independently in a way that enables readers to pick out a single subject or a related chapter and study it. Instructors can cover appropriate chapters depending on the objectives of the course. The required basic knowledge of the finite element method relevant to each physical phenomenon is illustrated at the beginning of the respective chapter. The general theory of the finite e lement, however, is presented briefly and concisely because the theory is well documented by other finite element books. ix

x

Preface

For example, in the heat transfer chapter, the theory is first explained, the governing equations are derived, the modeling techniques are presented, and finally practical problems are solved using ANSYS in a step-by-step technique. Each chapter independently discusses a single physical phenomenon, while the last chapter is devoted to multiphysics analyses and problems. The finite element solution is greatly affected by the quality of the mesh, and therefore, a separate chapter on meshing is included as a guide that emphasizes the basics of the meshing techniques. Practical end-of-chapter problems are provided in each chapter to challenge the reader’s understanding. Undergraduate and graduate engineers will use this book as a part of their courses, either when studying the basics of applied finite elements, or in mastering practical tools of engineering modeling. Engineers in industry can use this book as a guide for better design and analysis of their products. In all mechanical engineering curricula, junior- and senior-level courses use some type of engineering modeling software, and, oftentimes, this software is ANSYS. Senior students also use ANSYS in their senior design projects. Graduate-level finite element courses frequently use ANSYS to complement the theoretical analysis of finite elements. The course that uses this book should be taken after the introduction to design courses and the basic thermal–fluid courses. Courses such as that in senior design can be taken after this course. In this second edition of the book, new sections are added, and ANSYS examples are modified to be in compliance with the new version of ANSYS. Most ANSYS examples in the first edition are replaced by more general, comprehensive, and easy-to-follow examples. In the finite element theoretical part, more details are added, especially for the heat transfer chapter. Additionally, open-ended problems are added at the end of each chapter, which can serve as class projects.

Acknowledgments

I profited greatly from discussion with faculty members and engineers at Kuwait University. I want to mention particularly Professor Ahmed Yigit.

xi

This page intentionally left blank

About the author

Esam M. Alawadhi is a professor of mechanical engineering at Kuwait University, Kuwait City, Kuwait. He earned his doctor of philosophy in mechanical engineering from Carnegie Mellon University, Pittsburgh, Pennsylvania in May 2001. His research focuses on renewable energy, thermal management of electronics devices, energy conservation for buildings, fluid flow stability, and phase-change heat transfer.

xiii

This page intentionally left blank

Cha p t e r ON E

Introduction to finite element method and ANSYS

1.1

The finite element method and structural analysis

The finite element method was developed in the 1950s and has been continuously enhanced since then. Rapid advances in computing power and the drastic drop in cost make the finite element method affordable. Now, it is a commonly used method for solving a wide range of problems, and the potential of the finite element method is enormous. The finite element method is typically found in the aerospace, automotive, electrical, hydraulic, biomedical, nuclear, and structural engineering fields, among many others. The first step in the finite element solution procedure is to divide the domain into elements, and this process is called the domain discretization. The variable distribution across each element can be defined by linear, quadratic, or trigonometric function. The elements distribution in the domain is called the finite element mesh. The elements are connected at points called nodes. For example, consider a section of a bridge, as shown in Figure 1.1a. The bridge is divided into linear elements and connected by nodes, as shown in Figure 1.1b. After the region is discretized, the governing equations for the element must be established for the required physics. Material properties, such as modulus of elasticity for structural analysis, should be available. The equations are assembled to obtain the global equation for the mesh, which describes the behavior of the body as a whole. Generally, the global governing equation has the following form: [K]{U} = {F}

(1.1)

where [K] is called the stiffness matrix; {U} is the nodal degree of freedom, such as the displacements for structural analysis; and {F} is the nodal external force, such as forces for structural analysis. The [K] matrix is a singular matrix, and therefore it cannot be inverted.

1

2

Finite Element Simul ations Using ANSYS

Node Element

(b)

(a)

FIGURE 1.1 (a) A section of a bridge and (b) finite element mesh.

Consider a one-dimensional bar with initial length L that is subjected to a tensile force at its ends, as shown in Figure 1.2a. The cross-sectional area of the bar is A. The element can be modeled using a single element with two nodes, i and j, as shown in Figure 1.2b. Assume that the displacement function of the bar u(x) varies linearly along the length of the bar. The expression of the displacement is represented as u(x) = a + bx

(1.2)

The displacement at nodes i and j are ui and uj, respectively. Then ui(x) = a + bxi (1.3) uj(x) = a + bxj (1.4) where xi is the x-coordinate for node i, and xj is the x-coordinate for node j. Solving for constants a and b, it is found that a = (uixj − ujxi)/L (1.5) b = (uj − ui) / L

(1.6)

where L is the initial length of the element and is equal to (xi − xj). Substituting constants a and b into the displacement equation (Equation 1.2) and rearranging, the displacement function becomes u (x) =

F

xj − x x − xi ui + u j (1.7) L L Element

F x

L (a)

Node i

Node j (b)

FIGURE 1.2 A one-dimensional bar element. (a) A single one-dimensional bar element and (b) an element with two nodes.

3

Introduc tion to finite element me thod and ANSYS

or u(x) = Niui − Njuj (1.8) where Ni and Nj are called the shape functions of the element. When the element is loaded, it will be in an equilibrium position. The sum of the strain energy γ and work w done by external forces is the potential energy π of the element. The potential energy at the equilibrium position has the minimum value, and the potential energy is defined as π = γ − w

(1.9)

For a single bar element, the strain energy in the bar is given by

1 γ= 2

xj

∫ σ ε A dx (1.10) xi

Since the strain is related to the stress by the modulus of elasticity (σ = Eε), the strain energy is expressed as xj

AE 2 γ= ε dx (1.11) 2

∫ xi

The strain is equal to the elongation of the element in the x-direction ε = (uj − ui) / L

(1.12)

Then, the strain energy in Equation 1.11 becomes

γ=

2 AE u j − u i ) (1.13) ( 2L

The strain energy can be expressed in the matrix form as follows:

γ=

u i 1 AE u i u j 1 −1 = {U}T [K ]{U} (1.14) −1 1 u j 2 2L

where [K] is the stiffness matrix for the element. The external forces are forces acting at the nodes. The work done by the applied forces at the nodes is expressed as w = uiFi + ujFj = {U}T{F} (1.15)

4

Finite Element Simul ations Using ANSYS

Finally, the work done and strain energy are substituted into the potential energy for a single element. The potential energy is expressed as

π=

1 {U}T [ K ]{U} − {U}T {F } (1.16) 2

For the minimum potential energy, the displacement must be equal to zero:

∂π = 0 (1.17) ∂{U}

or

∂π = 0 and ∂ui

∂π = 0 (1.18) ∂u j

Equation 1.16 is differentiated using the expressions (1.18), and the results are

AE ( ui − u j ) − Fi = 0 (1.19) L

AE ( − ui + u j ) − Fj = 0 (1.20) L Equations 1.19 and 1.20 are expressed in the matrix form as follows:

−1 u i Fi − = 0 (1.21) 1 u j Fj

AE 1 L −1

and symbolically, [K]{U} − {F} = 0

(1.22)

where

[K ] =

AE 1 L −1

−1 (1.23) 1

The derivation of the stiffness matrix is valid only for one bar element. In practice, a model consists of many elements of different properties. The total potential energy of a number of elements is E

π=

∑(γ e =1

e

− w ) (1.24)

Introduc tion to finite element me thod and ANSYS

5

Minimizing Equation 1.24, the result is E

∑ K {U} − {F} = 0 (1.25) e

e =1

1.2

Stress analysis of pin-jointed bar

Use the finite element method to calculate the maximum displacement of the bar assembly shown in Figure 1.3a. Given that A1 = 90 mm2, A2 = 75 mm2, A3 = 50 mm2, E1 = 100 GPa, E2 = 110 GPa, and E3 = 180 GPa. Before starting to solve the problem, all bar elements and nodes of the assembly should be numbered, as suggested in Figure 1.3b. There are three main steps for solving this problem using the finite element method: (1) the stiffness matrix for all bar elements is determined, and then stiffness matrices are assembled to obtain the global stiffness matrix; (2) the boundary conditions are applied to the global stiffness matrix; and (3) rows and a column of the global stiffness matrix are eliminated to remove the singularity of the stiffness matrix, and finally we solve for the displacements. Step 1: Construct the stiffness matrix for all bar elements The stiffness matrix for element 1, which has nodes 1 and 2, is calculated using matrix (1.23):

K (1) =

(

)

90 × 10 −6 100 × 10 9 1 −1 3 −3 = 108 −3 30 × 10 −1 1 −3 3

The stiffness matrix for element 2, which has nodes 2 and 3, uses

K ( 2) =

(

)

75 × 10 −6 110 × 10 9 1 −1 1.5 −1.5 = 108 −3 55 × 10 −1 1 −1.5 1.5

(1)

(2)

30.00 mm

(3)

55.00 mm

50 kN

45.00 mm

(a)

Element (1) Node (1)

Element (2)

Node (2)

Element (3) Node (3)

Node (4)

(b)

FIGURE 1.3 (a) Bar assembly. (b) Elements and nodes for the bar assembly.

6

Finite Element Simul ations Using ANSYS

The stiffness matrix for element 3, which has nodes 3 and 4, uses K (3) =

(

)

50 × 10 −6 180 × 10 9 1 −1 45 × 10 −3

2 −1 = 108 1 −2

−2 2

Assembling the elements’ stiffness matrices to form the global stiffness matrix,

[K ] =

∑ K e =1

e=3

(e)

3 −3 [ K ] = 108 0 0

−3 4.5 −1.5 0

0 −1.5 3.5 −2

0 0 −2 2

Step 2: Apply the boundary conditions to the global stiffness matrix using Equation 1.22

108

3 −3 0 0

−3 4.5 −1.5 0

0 −1.5 3.5 −2

0 0 −2 2

u1 = 0 u 2 u3 u 4

F1 F2 = 0 = F3 = 0 F4 = 50

Step 3: Eliminate the first row and column from the global stiffness matrix to remove the singularity

4.5 108 −1.5 0

−1.5 3.5 −2

0 −2 2

u2 u 3 u 4

0 = 0 50

Solve for displacements: u2 = 1.667 × 10 −7 m u3 = 5.0 × 10 −7 m u4 = 7.5 × 10 −7 m (Maximum)

1.3

The finite element method and thermal analysis

Heat transfer in a one-dimensional space can be easily analyzed using the finite element method. Consider a wall that is made by common bricks, as shown in Figure 1.4. The outdoor surface temperature is Ti, while the indoor surface temperature is Tj.

Introduc tion to finite element me thod and ANSYS

Ti

7

Tj

Wall

Indoor

Outdoor

FIGURE 1.4 One-dimensional heat conduction in a wall.

L

The Fourier law indicates that the heat flow across the wall Q is proportional to temperature gradient dT/dx in the direction of heat transfer. That is,

Q = − kA

dT (1.26) dx

where k is the thermal conductivity of the wall and A is the area perpendicular to the heat flow. For steady-state conditions, the temperature varies linearly along the length of the wall as follows:

dT Tj − Ti (1.27) = dx L

Hence, the heat flow across the indoor and outdoor surfaces of the wall can be expressed as, respectively,

Q i = − kA

Tj − Ti (1.28) L

Q j = − kA

Ti − Tj (1.29) L

Equations 1.28 and 1.29 are expressed in the following matrix form:

kA 1 L −1

−1 Ti Q i − = 0 (1.30) 1 Tj Q j

This is analogous to the bar element equation (Equation 1.21). The nodal forces are substituted by nodal heat flow, nodal displacements by nodal temperatures, and constant AE/L by kA/L; therefore, [K]{T} − {Q} = 0

(1.31)

8

Finite Element Simul ations Using ANSYS

where

[K ] =

kA 1 L −1

−1 (1.32) 1

Heat transfer through a composite wall

1.4

Consider a wall that is constructed by stone bricks at the outdoor surface (k = 1.25 W/m · °C), common brick (k = 0.75 W/m · °C), and plaster at the indoor surface (k = 0.9 W/m · °C), as shown in Figure 1.5. The outdoor surface temperature is To = 45°C, while the indoor surface temperature is Ti = 25°C. Determine the temperature at the interfaces, and the results should be per unit area of the wall. Step 1: Construct the stiffness matrix for all elements The stiffness matrix for element 1, which has nodes 1 and 2, is calculated using expression (1.32):

1.25 × 1 1 K (1) = 5 × 10 −2 −1

−1 25 = 1 −25

−25 25

The stiffness matrix for element 2, which has nodes 2 and 3, uses

0.75 × 1 1 K ( 2) = 15 × 10 −2 −1

Outdoor (To)

Stone brick

Common brick

−1 5 = 1 −5

−5 5

Indoor (Ti) Plaster

5 cm

15 cm

3 cm

(a) Element (1) Node (1)

Element (2)

Node (2)

Element (3) Node (3) Node (4)

(b)

FIGURE 1.5 The wall composition and finite element mesh. (a) The wall composition and (b) the finite element presentation.

9

Introduc tion to finite element me thod and ANSYS

The stiffness matrix for element 3, which has nodes 3 and 4, uses

0.9 × 1 1 K (3) = 3 × 10 −2 −1

−1 30 = 1 −30

−30 30

Assembling the elements’ stiffness matrices to form a global stiffness matrix,

[K ] =

∑ K e =1

e=3

(e)

25 −25 [ K ] = 108 0 0

−25 30 −5 0

0 −5 35 −30

0 0 −30 30

Step 2: Apply the boundary conditions to the global stiffness matrix using Equation 1.31

25 −25 0 0 −25 30 −5 0 −5 35 −30 0 0 0 −30 30

T1 = 45 Q1 T2 Q2 = 0 = T 3 Q3 = 0 T4 = 25 Q 4

Then Q1 and Q2 are required to maintain the constant temperature at the external surfaces. Step 3: Eliminate the rows and columns, numbers 1 and 4, from the global stiffness matrix to remove the singularity from the stiffness matrix The temperatures at nodes 1 and 4 are known. Hence, only Equations 2 and 3 are needed to solve the problem:

25 −5

−5 T2 35 T3

25 × 45 = 30 × 25

Solving for temperatures, T2 = 42.07°C T3 = 27.44°C

10

Finite Element Simul ations Using ANSYS

1.5

Introduction to ANSYS software

The ANSYS software is the most advanced package for single- and multiphysics simulations, offering enhanced tools and capabilities that enable engineers to complete their jobs in an efficient manner. ANSYS includes significant capabilities, expanding functionality, and integration with almost all computer-aided design (CAD) drawing software, such as pro-engineers, AutoCAD, and Solid Edge. In addition, ANSYS has the best-in-class solver technologies, a coupled physics for complex simulations, meshing technologies customizable for physics, and computational fluid dynamics. ANSYS can solve problems in structural, thermal, fluid, acoustics, and multiphysics: Structural: • Linear • Geometric and material nonlinearities • Contact • Static • Dynamic • Transient, natural frequency, harmonic response, response spectrum, random vibration • Buckling • Topological optimization Thermal: • Steady state or transient • Conduction • Convection • Radiation • Phase change Fluid mechanics: • Steady state or transient • Incompressible or compressible • Laminar or turbulent • Newtonian or non-Newtonian • Free, forced, or mixed convection heat transfer • Conjugate solid/fluid heat transfer • Surface-to-surface radiation heat transfer

Introduc tion to finite element me thod and ANSYS

11

• Multiple species transport • Free surface boundaries • Fan models and distributed resistances • Stationary or rotating reference frames Acoustics: • Fully coupled fluid/structural • Near and far field • Harmonic, transient, and modal Multiphysics: • Thermal–mechanical • Thermal–electric • Thermal–electric–structural • Piezoelectric • Piezoresistive • Peltier effect • Thermocouple effect • Electromechanical circuit simulator ANSYS is not a single program, but it is a family of CAD programs, as shown in Figure 1.6. The ANSYS/Multiphysics package consists of ANSYS/Emag, which is for magnetic field analysis, ANSYS/Flotran for fluid mechanics, ANSYS/LS-DYNA for dynamics analysis, and ANSYS/Mechanical for structural and thermal analyses.

ANSYS/ Emag

ANSYS/ Mechanical

ANSYS/ Multiphysics

ANSYS/ Flotran

ANSYS/ LS-DYNA

FIGURE 1.6 The ANSYS family.

12

Finite Element Simul ations Using ANSYS

To start ANSYS: Double click on the Mechanical APDL Product Launcher icon in the desktop or, go to Start > All Programs > ANSYS > Mechanical APDL Product Launcher The ANSYS Product Launcher window will appear as shown. First, select the ANSYS Multiphysics in License. Then, type the working directory path for ANSYS files in the Working Directory. All ANSYS files will be stored in this directory, including images and AVI movies. The session name is specified in the Job Name. Finally, click on Run to start the ANSYS.

A select ANSYS Multiphysics in License B change Working Directory to C:\, or any directory C change the initial Job Name to Example, or any name Run The ANSYS interface has the following components: • • • • • •

Utility menu Toolbar Main menu View tools Graphics Comment prompt

Introduc tion to finite element me thod and ANSYS

13

In addition, the ANSYS Output window will show up. The output window is dynamically listing important information during the preprocessor, solution, and postprocessor. Additionally, the calculated results in the postprocessor are shown in this window. Warnings in the ANSYS Output should be carefully considered to avoid unexpected errors.

14

Finite Element Simul ations Using ANSYS

Utility menu

The Utility menu is used for the file operations, listing and plotting, and changing display options. In the pull-down File menu, the following tasks can be performed:

A Clearing current job and starting a new job B Resuming a job, and the database file can be located C Saving ANSYS job D Reading ANSYS input file E Listing and file operations F Importing and exporting files G Generating a summary file H Exiting ANSYS job The pull-down List menu is for listing the model components, such as keypoints, areas, and nodes. In addition, the properties of the material can be listed. The boundary conditions imposed on the model can also be listed.

Introduc tion to finite element me thod and ANSYS

15

The pull-down Plot menu is similar to the List menu. Plotting g eometry’s components, such as keypoints and areas, can be performed in this menu. In the PlotCtrls menu, printing the model in the ANSYS graphics, changing the style of the ANSYS graphics, or changing the quality of the graphics can be done. The Workplane is for the grids setup. ANSYS Main Menu Most ANSYS jobs are done using the Main Menu, from building the model to viewing the results. In the preprocessor, the material properties and real constants are specified, element type is selected, and modeling operations and meshing tools are available. In the solution, the boundary conditions are imposed, and the solution setup parameters are specified. In the postprocessor, the ANSYS results are presented. List, plot results, and path operations can be performed.

The three tasks are summarized as follows: Preprocessor:

1. Element type 2. Material properties 3. Real constants 4. Modeling 5. Meshing

Solution:

1. Boundary condition 2. Solution setup parameters

Postprocessor:

1. Plot results 2. List results 3. Path operation

PROBLEM 1.1 Use the finite element method to calculate the maximum displacement of the bar assemblies shown in Figures 1.7 to 1.9.

16

Finite Element Simul ations Using ANSYS

E = 155 GPa A = 100 mm2

FIGURE 1.7 Two bars assembly with a tensile force for at one end.

E = 150 GPa A = 75 mm2

E = 200 GPa A = 125 mm2

35.00 mm

E = 180 GPa A = 50 mm2

50 kN

55.00 mm

E = 150 GPa A = 75 mm2

150 kN

150 kN 55.00 mm

35.00 mm

55.00 mm

FIGURE 1.8 Three bars assembly with a tensile force for at both ends.

100 kN E = 150 GPa D = 0.45 cm E = 150 GPa D = 1.1 cm

2 cm

E = 150 GPa D = 1.75 cm E = 150 GPa D = 2.25 cm

1 cm 0.5 cm 0.25 cm

FIGURE 1.9 A structure with a vertical tensile force.

PROBLEM 1.2 Consider a wall shown in Figure 1.10. The outdoor surface temperature is To = 42°C, while the indoor surface temperature is Ti = 23.5°C. Determine the temperature at the interfaces, and the results should be per unit area of the wall. The thermal conductivity of stone brick, thermal insulation, common brick, and indoor finishing is 1.25 W/m · °C, 0.85 W/m · °C, 0.2 W/m · °C, and 1.75 W/m · °C, respectively.

17

Introduc tion to finite element me thod and ANSYS

Insulation

Outdoor (To)

Stone brick

Common brick

Indoor (Ti) Indoor finishing

10.5 cm

5.2 cm

22.5 cm

1.25 cm

FIGURE 1.10 Heat transfer in a wall composed of four different materials. Al 205 W/m . ºC

Fe 73 W/m .ºC

Cu 386 W/m .ºC T2 = 25˚C

T1 = 100˚C

2.5 cm

5.5 cm

4.5 cm

FIGURE 1.11 Heat transfer in a rod composed of three different materials.

PROBLEM 1.3 Consider a thermal rod that is composed of three different materials, as shown in Figure 1.11. The left circular area is maintained at a temperature of T1 = 100°C, while the right one is maintained at a temperature of T2 = 25°C. The entire lateral surface is well insulated. Determine the heat flow through the rod, if its diameter is 1.5 cm.

This page intentionally left blank

Cha p t e r T WO

Bar and beam structures

2.1

Finite element method for a horizontal bar element

The bar structures are composed of straight bars and welded at the interactions. The bar structures can be found in bridges and roofs. The derivation of the stiffness matrix for a bar element is applicable to the pin-connected structures. The bar element is assumed to have a constant cross-sectional area A, uniform modulus of elasticity E, and initial length L. The bar is subjected to tensile forces along the local axis and applied at its ends. There are two coordinate systems: local (x, y) and global (x,y) coordinates. Figure 2.1 shows the local and global coordinate systems. For each element, there are two local forces acting on the nodes: F1x and F1y are acting on node 1, and F2x and F2y are acting on node 2. The nodal degrees of freedom are the four local displacements, two at each node: d1x and d1y are at node 1, and d2x and d2y are at node 2. Figure 2.2 shows a bar element lying along the local x-coordinate. The strain–displacement relationship is obtained from Hooke’s law as follows:

σ x = Eε x (2.1)

where

εx =

du (2.2) dx

and therefore,

Aσ x = Fx (2.3)

Then u is the local axial displacement in the x-direction and Fx is the tensile force in the local x-direction. Note that the bar element cannot sustain shear forces. Substituting σx and εx into Hooke’s law (2.1) yields:

du d AE = 0 (2.4) dx dx 19

20

Finite Element Simul ations Using ANSYS

y y x

θ

FIGURE 2.1 The local and global coordinate systems.

FIGURE 2.2 A bar element lying along the local x-coordinate.

Fx1

Node 1

A, E x

x

Node 2

Fx2

L

Assuming a linear displacement along the local x-axis of the bar, the displacement function is expressed as

u ( x ) = a1 + a 2 x (2.5)

The displacement values at nodes are used to determine the values of the constants a1 and a2. Now, the displacement function is expressed as a function of the nodal displacement d1x and d2x, which can be achieved by evaluating u at the nodes, and solving for a1 and a2 as follows:

u ( 0 ) = d1x = a1 (2.6)

u ( L ) = d2x = a 2 L + d1x (2.7)

and solving for a2,

a2 =

d2x − d1x (2.8) L

Hence, the displacement function (2.5) becomes

d −d u ( x ) = 2x 1x x + d1x (2.9) L The displacement function (2.9) is expressed in the following form:

u ( x ) = N1 d1 + N 2 d2 (2.10)

where N1 and N2 are called the shape functions of the displacement function, which is associated with degrees of freedom d1 and d2. The shape functions are defined as

Bar and be a m struc tures

N1 = 1 −

21

x (2.11) L

and

x (2.12) L

N2 =

The displacement function can be written in the following matrix form:

d1x u = [ N1 N 2 ] d2x

(2.13)

The strain is the elongation in the x-direction, which is equal to the derivative of the displacement function as follows:

u ( x + ∆x ) − u ( x ) du = (2.14) ∆x dx

εx =

In terms of the nodal displacement, the strain is written as

εx =

du d2x − d1x = (2.15) dx L

The strain (2.15) is substituted into Hooke’s law (2.1) to obtain the force:

d − d1x Fx = AE 2x (2.16) L

The nodal force at node 1 is acting in the negative x-direction and should have a negative sign as follows:

F1x = − Fx (2.17) Hence, the nodal force at node 1 becomes

F1x =

(

)

AE d1x − d2x (2.18) L

The nodal force at node 2 is acting in the positive x-direction and should have a positive sign as follows:

F2x = Fx (2.19)

F2x =

(

)

AE d2x − d1x (2.20) L

22

Finite Element Simul ations Using ANSYS

The nodal force in the x-direction is expressed in a matrix form as follows:

AE 1 −1 d1x = L −1 1 d2x

F1x F2x

(2.21)

and similarly for the nodal forces in the y-direction:

F1y F2y

AE 1 −1 d1y = L −1 1 d2y

(2.22)

{}

Since {F } = K d , the stiffness matrix for a bar element in local coordinates can be written as

AE 1 −1 K = (2.23) L −1 1

The stiffness matrix can be determined using the strain energy expression along with Castigliano’s theorem for a bar. The strain energy of a bar in tension is expressed as follows: L

U=

∫ 0

Fx2 dx (2.24) 2EA

In the theorem, the change of strain energy is equal to the virtual displacement times the external force at a point as follows:

∂U = Pi ∂ di (2.25) Equation 2.25 in terms of external force is

Pi =

∂U (2.26) ∂ di

The axial force can be written in terms of displacement function as

Fx = σ x A = E ε x A = EA

∂u (2.27) ∂x

Using Equation 2.10 in Equation 2.27, the axial force is expressed as

(

)

Fx = EA N1′ d1x + N ′2 d2x (2.28) The strain energy of a bar (2.24) in tension is expressed as

U=

EA 2

L

∫ ( N′ d 1

0

1x

)

2

+ N ′2 d2x dx (2.29)

23

Bar and be a m struc tures

Using Castigliano’s first theorem, the nodal force in the x-direction at node 1 is written as δ ∂U F1x = = EA ∂ d1x

L

∫ ( N′ d 1

1x

)

+ N ′2 d2x N1′ dx

0

L L = EA N1′ N1′ dx d1x + EA N1′ N ′2 dx d2x 0 0

∫

(2.30)

∫

and at node 2, δ ∂U F2x = = EA ∂ d2x

L

∫ ( N′ d 1

1x

)

+ N ′2 d2x N ′2 dx

0

L L = EA N1′ N ′2 dx d1x + EA N ′2 N ′2 dx d2x 0 0

∫

(2.31)

∫

Nodal forces in the x-direction in the matrix form are expressed as F1x F2x

k11 k12 = k 21 k 22

d1x d2x

(2.32)

Or, symbolically,

{F } = [K ]{d } (2.33) x

x

where [K] is the stiffness matrix, in which the matrix coefficients are defined as L

∫

k ij = EA N ′i N ′j dx (2.34) 0

Substituting the shape functions into Equation 2.32, the stiffness equation is expressed as

2.2

F1x F2x

EA 1 −1 d1x = L −1 1 d2x

(2.35)

Finite element method for an arbitrary-oriented bar element

The local coordinate system is always chosen to represent an individual element, whereas the global coordinate system is chosen for the whole structure. In order to relate the global displacement components to a local one, the transformation matrix is used. Figure 2.3 shows a bar element lying along the local x-axis. The local x-axis is oriented at an angle θ

24

Finite Element Simul ations Using ANSYS

Fy2, dy2

No

A,

No

Fy1, dy1

de

1

E,

L

de

, d x2 F x2

2

Fx2, dx2

y

θ

x

, d x1 F x1

Fx1, dx1

FIGURE 2.3 Relationship between the local and global displacements.

measured counterclockwise from the global x-axis. In the local coordinate, each node has a nodal force Fx and nodal displacement dx along its axis. In the global coordinate, each node has a horizontal force Fx, vertical force Fy, horizontal displacement dx, and vertical displacement dy. Hence, each element has four degrees of freedom. From Figure 2.3, the displacement in the x-direction for nodes 1 and 2 in a local coordinate system can be obtained from its global displacements as follows:

d1x = d1x cos θ + d1y sin θ (2.36)

d2x = d 2x cos θ + d 2y sin θ (2.37)

and in the y-direction,

d1y = − d1x sin θ + d1y cos θ (2.38)

d2y = − d 2x sin θ + d 2y cos θ (2.39)

The transformation for nodal displacements for each element can be written in matrix form as follows:

d1x d1y d2x d 2y

d 0 0 1x cos θ sin θ − sin θ cos θ 0 0 d1y = 0 cos θ sin θ d2x 0 0 − sin θ cos θ 0 d2y

(2.40)

25

Bar and be a m struc tures

and symbolically,

{d} = [ T ]{d} (2.41)

The nodal force is transformed from the global to the local coordinate system in a similar manner:

F1x F1y F2x F2y

0 F1x cos θ sin θ 0 − sin θ cos θ 0 F1y 0 = 0 cos θ sin θ F2x 0 − sin θ cos θ F2y 0 0

(2.42)

and symbolically,

{F} = [ T ]{F} (2.43)

The global element nodal force vector is related to the global displacement vector using the global stiffness matrix as follows:

F1x F1y F2x F2y

d1x d1y = [K ] d2x d 2y

(2.44)

where

1 AE 0 [K ] = L −1 0

{}

0 0 0 0

−1 0 1 0

0 0 0 0

(2.45)

{}

Substituting d = [ T ]{d} and {F } = [ T ]{F } into {F } = K d , the following equation is obtained:

[ T ]{F } = K [ T ]{d} (2.46)

By multiplying both sides of the above equation by the inverse of the transformation matrix, the nodal force vector becomes

{F } = [ T ]−1 K [ T ]{d} (2.47)

26

Finite Element Simul ations Using ANSYS

Since the transformation matrix is orthogonal, its transport and inverse are equal. The nodal force vector is expressed as

{F } = [ T ]T K [ T ]{d} (2.48)

Since the global coordinate force equation {F} = [K]{d} is equal to nodal force in Equation 2.48, the global stiffness matrix for an element is written as

[ K ] = [ T ]T K [ T ] (2.49) Expanding the stiffness matrix yields

cos θ cos θ cos θ sin θ − cos θ cos θ − cos θ sin θ AE cos θ sin θ sin θ sin θ − sin θ cos θ − sin θ sin θ [K] = L − cos θ cos θ − sin θ cos θ cos θ cos θ sin θ cos θ − cos θ sin θ − sin θ sin θ sin θ cos θ sin θ sin θ (2.50) Assemble the global stiffness and force matrices using the direct stiffness method to obtain matrices for the entire domain:

[K ] =

N

∑ K

(e)

e =1

{F } =

N

(2.51)

∑ {F } (2.52) (e)

e =1

{d} =

N

∑ {d } (2.53) (e)

e =1

The global stiffness matrix is related to the global nodal force matrix and global displacement matrix for the whole structure using the following expression: {F}=[K]{d}

2.3

(2.54)

Analyzing a plane bar structure

The bar structure shown in Figure 2.4a is composed of three bars. All bars have the same length, cross-sectional area, and modulus of elasticity. A horizontal force is applied at Point 2 with F = 100 kN.

27

Bar and be a m struc tures

F3y, d3y Node 3

3

Element 2

Element 3

F1y, d1y 1

2

F

F1x, d1x

F3x, d3x

F2y, d2y

Node 1

Element 1

(a)

Node 2

F2x, d2x

(b)

FIGURE 2.4 (a) A bar structure and (b) nodes and elements.

Determine the horizontal and vertical displacements at all connections, given that E = 100 GPa, A = 1.0 × 10 −4 m2, and L = 1 m. Before starting to solve the problem, all elements and nodes of the structure should be numbered as suggested in Figure 2.4b. Step 1: Construct the stiffness matrix for each element The stiffness matrix for Element 1, which has nodes 1 and 2 with θ = 0o, is calculated using the following matrix (2.50): 1

AE K (1) = 4L

1

4 0 −4 0

2

2

0 −4 0 0 0 4 0 0

0 0 0 0

1 1 2 2

The stiffness matrix for Element 2, which has nodes 1 and 3 with θ = 60o, is calculated as follows:

1

AE K ( 2) = 4L

−

1

3

3

−1 − 3 1 −3 1 3 3 − 3 −1 − 3 1 3 3 −3 3 3 3 3 1

3

28

Finite Element Simul ations Using ANSYS

The stiffness matrix for Element 3, which has nodes 2 and 3 with θ = 120°, is calculated as follows: 2

AE K (3) = 4L

3

2

1 − 3 − 3 3 − 3 −1 3 −3 −

3

3 2 3 −3 2 1 − 3 3 3 3 3

−1

Step 2: Assemble the elements’ stiffness m atrices to form a global stiffness matrix using Equation 2.51 1

[K ] =

AE 4L

1

2

2

3

3

−4 −1 − 3 1 5 3 0 −3 1 3 3 0 0 − 3 −1 0 5 − 3 3 2 −4 −3 2 0 0 − 3 3 3 −1 3 2 0 3 −1 − 3 −3 −3 3 0 6 3 − 3

Step 3: Apply the boundary conditions to the global stiffness matrix using Equation 2.54

F1x F1y = 0 F2x = 10 5 F2y = 0 F3x F3y

d1x = 0 d1y d 2x = [K ] d = 0 2y d 3x = 0 d =0 3y

Step 4: Eliminate rows and columns from the global stiffness matrix to remove the singularity Since the stiffness matrix is singular, at a minimum, one equation must be eliminated to remove the singularity. The first, fourth, fifth, and sixth columns are multiplied by zero displacements, and therefore, they can be deleted with the corresponding rows, and the result is

(

)

F1y = 0 1 × 10 −4 100 × 10 9 3 0 d1y = 0 5 5 1× 4 d 2x F2x = 10

Bar and be a m struc tures

29

Step 5: Solve for d1y and d2x d1y = 0 m d2x = 0.008 m

2.4

Analyzing a plane bar structure using ANSYS

The bar structure shown in Figure 2.4a is composed of three bars. All bars have the same length, cross-sectional area, and modulus of elasticity. A horizontal force is applied at Point 2, F = 100 kN. Determine using ANSYS the horizontal and vertical displacements at all connections and the reaction at the supports, given that E = 100 GPa, A = 1.0 × 10 –4 m2, and L = 1 m. Double click on the Mechanical APDL Product Launcher icon

A select ANSYS Multiphysics in License B change Working Directory to C:\ or any directory C change the initial Job Name to bar, or any name Run This example is limited to structural analysis. Hence, select Structural in Preferences and leave other physics unselected. This will reduce selection options in the preprocessor and solution tasks.

30

Finite Element Simul ations Using ANSYS

Main Menu > Preferences

A select Structural in Preferences for GUI Filtering OK Next, the element type is selected. For the present problem, the element type is Link, which is equivalent to a bar. Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add...

Bar and be a m struc tures

31

The selected link element will not support pressure on the elements or moment on the nodes. However, this element can be used for threedimensional analyses.

A select Link B select 3D finit stn 180 OK

Close The cross-sectional area of the bars is required for the analysis. The initial strain is zero by default.

32

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Real Constants > Add/Edit/Delete

Add...

Notice that the selected element type is shown in the Element Type for Real Constants. Here, the cross-sectional area of the bars is specified. OK

Bar and be a m struc tures

33

A type 1e-4 in Cross-sectional area AREA OK

Close Material properties of the bars are specified in the following steps. The bars are elastic and independent of the direction isotropic. Only the modulus of elasticity is required and the zero Poisson ratio, or any value, is just to avoid an error message from ANSYS.

34

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic

A type 100e9 in EX B type 0 in PRXY OK Close the Material Model Behavior window The process of modeling the bar structure is started here. First, three nodes are created, followed by creation of the elements. The x- and y-coordinates of each node are specified in ANSYS. The coordinate for node 1 is (0,0), for node 2 is (1,0), and for node 3 is (0.5,0.866). The Apply button will not close the window, allowing for additional inputs, while the OK will close the window.

Bar and be a m struc tures

Main Menu > Preprocessor > Modeling > Create > Nodes > In Active CS

A type 1 in Node number B type 0 and 0 in X,Y,Z Location in active CS Apply

A type 2 in Node number B type 1 and 0 in X,Y,Z Location in active CS Apply

A type 3 in Node number B type 0.5 and 0.866 in X,Y,Z Location in active CS OK

35

36

Finite Element Simul ations Using ANSYS

ANSYS graphics show the created nodes Main Menu > Modeling > Create > Elements > Auto Numbered > Thru Nodes Click on node 1 then 2. Then in Elements from Nodes window, click on Apply Click on node 2 then 3. Then in Elements from Nodes window, click on Apply Click on node 3 then 1. Then in Elements from Nodes window, click on OK

ANSYS graphics show the three elements The preprocessor task is ended at this point. The solution task starts here. Nodal forces and displacements are applied. Starting with forces or

Bar and be a m struc tures

37

displacements will not affect the solution. A zero nodal displacement at a certain direction means that the node is fixed at that direction. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Nodes Click on node 1. Then in Elements from Nodes window, click on OK

A select UX in Lab2 DOFs to be constrained B type 0 in Displacement value Apply Click on node 2. Then in Elements from Nodes window, click on OK

A select UY in Lab2 DOFs to be constrained B type 0 in Displacement value Apply

38

Finite Element Simul ations Using ANSYS

Click on node 3. Then in Elements from Nodes window, click on OK

A select UX in Lab2 DOFs to be constrained B select UY in Lab2 DOFs to be constrained C type 0 in Displacement value OK Main Menu > Solution > Define Loads > Apply > Structural > Force/Moment > On Nodes Click on node 2. Then in Apply F/M on Nodes window, click on OK

A select FX in Direction of force/mom B type 100e3 in Force/moment value OK

Bar and be a m struc tures

39

The positive FX means that the force at node 2 is in the positive x-direction. The ANSYS graphics will show the applied force and its direction with a red arrow. Reapplying the force at node 2 will automatically delete the force and apply the new value.

ANSYS graphics show the force with d irection and the displacement constraints on nodes The final step is to initiate the solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the equations. Main Menu > Solution > Solve > Current LS

OK

Close

40

Finite Element Simul ations Using ANSYS

Results can be viewed in the general postprocessor task. Inspecting the deformation of the bar structure will help to identify if the problem is solved correctly. Node 2 should move to the right as a result of the applied force, while node 3 should be stationary. Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + unreformed OK

ANSYS graphics show the bar before and after deformation From the above figure, the result is as expected. Nodes 1 and 3 are fixed, while node 2 is moved in the direction of the applied force. In the following steps, the nodal results are presented. First, the nodal displacements are listed, and second, the reactions at the supports are listed.

Bar and be a m struc tures

41

Main Menu > General Postproc > List Results > Nodal Solution

A click on Nodal Solution > DOF Solution > X-Component of displacement OK

A list of nodal displacement in the x-direction is shown. In addition, the maximum nodal displacement is shown at the end. The nodal displacement in the y-direction can be determined by selecting Y-Component in the Nodal Solution. The results are identical to the previous example, and the reactions at the supports are listed as follows: Main Menu > General Postproc > List Results > Reaction Solution

A select All items OK

42

Finite Element Simul ations Using ANSYS

As shown, the forces are balanced in the x- and y-directions. The sum of the forces in the x-direction is equal to −100 × 103, which is equal to the applied force, and the sum of the forces in the y-direction is equal to zero.

2.5

Finite element method for a horizontal beam element

A beam is defined as a long and slender structural member that can be subjected to transverse loadings. Therefore, the beam can be twisted and bended. The bar elements can sustain the transverse loadings only. Consider a horizontal beam element, as shown in Figure 2.5. The beam has an initial length L, modulus of elasticity E, and moment of inertia I. The local displacement and rotation at nodes 1 and 2 are (dx1, θ1) and (dx2, θ2), respectively. The local nodal force and bending moment at nodes 1 and 2 are (F1y, M1) and (F2y, M2), respectively. All moments and rotations are positive if their direction is counterclockwise and negative if their direction is in a clockwise direction. The horizontal forces are positive if the direction is positive in the x-direction and negative if the direction is negative in the x-direction. The vertical forces are positive if the direction is positive in the y-direction and negative if the direction is negative in the y-direction. Consider a differential beam element, as shown in Figure 2.6.

y 1

I, E

2

M1, θ1 L Fy1, dy1

Fy2, dy2

FIGURE 2.5 Beam element subjected to forces and moments.

x

M2, θ2

43

Bar and be a m struc tures

M V

M + dM a

V + dV

b

FIGURE 2.6 A differential beam element.

dx

Applying the moments balance at Point b yield,

− Vdx + dM = 0 (2.55)

or

V =

dM (2.56) dx

The curvature of the beam is related to the applied moment. The θ is the rotation and the function v(x) is the transverse displacement in the y-direction. Figure 2.7 shows a section of a horizontal deflected beam. The rotation of the deflection is expressed as

θ =

dv ( x ) (2.57) dx

The axial strain is related to the axial displacement using the following relationship: du (2.58) dx On the other hand, the axial displacement is related to the transverse displacement by

εx =

dv (2.59) dx Substituting the relationship (2.59) into (2.58) yields u = −y

ε x = −y

d2v (2.60) dx 2

y

θ

v(x) dx

x

FIGURE 2.7 Horizontal beam.

deflected

44

Finite Element Simul ations Using ANSYS

For a cross section of the beam shown in Figure 2.8, the differential force and moment along the center of the beam are expressed as, respectively, dF = σx dA

(2.61)

dM = σx ydA

(2.62)

Using Hooke’s equation, the differential moment becomes

dM = Eε x ydA = Ey 2

d2v dA (2.63) dx

Integrating Equation 2.63 for the beam section yields

M = E

d 2v y 2 dA (2.64) dx 2

∫

Since the first moment of inertia is equal to ∫ y2 dA, Equation 2.64 is expressed as

M = EI

d2v (2.65) dx 2

Also, the shearing force (2.56) is expressed as

V = EI

d3 v (2.66) dx 3

The variations of beam curvature are approximated by a third-order polynomial, and the constants in the polynomial are determined using the boundary conditions as follows: v(x) = a1x3 + a2x2 + a3x + a4 (2.67) Using expression (2.57), the rotation of the deflection can be determined as θ(x) = 3a1x2 + 2a2x + a3 (2.68) Applying the boundary conditions, v(0) = dy1 : a4 = dy1 (2.69) v(L) = dy2 : a1L3 + a2L2 + a3L + a4 = dy2 (2.70)

dF y

FIGURE 2.8 Beam cross section.

45

Bar and be a m struc tures

θ(0) = θ1 : a3 = θ1 (2.71) θ(L) = θ2 : 3a1L2 + 2a2L + a3 = θ2 (2.72) Solving Equations 2.69–6.72 for a1, a2, a3, and a4 and substituting the solution into Equation 2.67 yields

2 1 v ( x ) = 3 ( d1y − d 2y ) + 2 ( θ1 + θ2 ) x 3 L L 3 1 + − 2 ( d1y − d 2y ) − ( 2θ1 − θ2 ) x 2 + θ1x + d1y L L

(2.73)

or v ( x ) = d1y + θ1x −

3x 2 2x 2 3x 2 x2 d1y − θ1 + 2 d 2y − θ2 2 L L L L

2x 2 2x 2 x3 x3 + 3 d1y + 2 θ1 − 3 d 2y + 2 θ2 L L L L

(2.74)

The displacement function v(x) can be expressed in terms of shape functions as v(x) = N1d1y + N2θ1 + N3d2y + N1θ2 (2.75) and the shape functions are 2

3

x x N1 = 1 − 3 + 2 (2.76) L L

x3 x2 N 2 = x − 2 + 2 2 (2.77) L L

x x N 3 = 3 − 2 (2.78) L L

x 2 x3 N 4 = − + 2 (2.79) L L

2

3

Finally, the element stiffness matrix is derived using the equilibrium approach. The nodal shear force and bending moment are calculated using Equations 2.65 and 2.66, respectively: d 3 v ( 0 ) EI = 3 (12d1y + 6Lθ1 − 12d 2y + 6Lθ2 ) (2.80) dx 3 L

F1y = EI

M1 = − EI

d 2 v ( 0 ) EI = 3 6d1y + 4L2 θ1 − 6Ld 2y + 2L2 θ2 (2.81) dx 2 L

(

)

46

Finite Element Simul ations Using ANSYS

F2y = − EI

M2 = EI

d 3 v ( L ) EI = 3 ( −12 d1y − 6Lθ1 + 12d 2y − 6Lθ2 ) (2.82) dx 3 L

d 2 v ( L ) EI = 3 6 Ld1y + 2L2 θ1 − 6Ld 2y + 4L2 θ2 (2.83) dx 2 L

(

)

The nodal forces and moments can be expressed in matrix form, {F} = [K]{d}, as follows:

F1y 12 M1 EI 6L = F2y L3 −12 6L M 2

6L −12 6L d1y 4L2 −6L 2L2 θ1 −6L 12 −6L d 2y 2L2 −6L 4L2 θ2

(2.84)

where

EI [K ] = 3 L

12 6L −12 6L 6L 4L2 −6L 2L2 (2.85) −12 −6L 12 −6L 6L 2L2 −6L 4L2

Assemble the stiffness matrices and force and displacement vectors:

[K ] =

N

∑ K

(e)

e =1

(2.86)

N

{F } =

∑ {F } (2.87) (e)

e =1

{d} =

N

∑ {d } (2.88) (e)

e =1

The global stiffness matrix is used to relate the nodal force vector to nodal displacement vector for the whole structure: {F} = [K]{d}

(2.89)

The stiffness equation for the horizontal beam element can also be obtained by using Castigliano’s theorem. The strain energy for a beam with uniform material properties and cross-sectional area is L

U=

L

2

1 EI ∂2 v ( M )2 dx = dx (2.90) 2EI 2 ∂x 2

∫ 0

∫ 0

47

Bar and be a m struc tures

The second derivative of displacement function is

v ′′ ( x ) = N1′′ d1y + N ′′θ 2 1 + N 3′′ d 2y + N ′′θ 4 2 (2.91)

and the second derivative of the shape functions is

N1′′= −

6 x + 12 3 (2.92) L2 L

N ′′2 = −

4 x + 6 2 (2.93) L L

N 3′′ =

N ′′4 = −

6 x − 12 3 (2.94) L2 L 6 x + 6 2 (2.95) L L

In Castigliano’s first theorem, the change of strain energy is equal to the virtual displacement times external force at a point. Applying Castigliano’s theorem at node 1 gives

∂U ∂ F1y = = ∂d1y ∂d1y

EI L ∂2 v 2 dx (2.96) 2 ∂x 2 0

∫

Substituting the second derivative of the displacement function into Equation 2.80 gives

∂ EI F1y = ∂d1y 2

L

∫ ( N′′d 1

0

1y

2 + N ′′θ 2 1 + N 3′′ d 2y + N ′′θ 4 2 ) dx (2.97)

Simplifying expression (2.97) yields L

EI F1y = 2 ( N1′′d1y + N ′′θ 2 1 + N 3′′ d 2y + N ′′θ 4 2 ) N1′′ dx (2.98) 2

∫ 0

and finally, F1y = k11v1 + k12θ1 + k13v2 + k14θ2 (2.99) where L

∫

k ij = EI N ′′i N ′′j dx (2.100) 0

48

Finite Element Simul ations Using ANSYS

The nodal force in the y-direction at node 1 is

F1y =

EI (12 d1y + 6Lθ1 − 12d2y + 6Lθ2 ) (2.101) L3

If instead of virtual displacement, the virtual rotation ∂θ at a point where moment M is applied, the moment in terms of strain energy is

M1 =

∂U (2.102) ∂θi

Applying Castigliano’s theorem at node 1 gives

L 2 ∂U ∂ EI ∂2 v M1 = = dx (2.103) 2 ∂θ1 ∂θ1 2 ∂x 0

∫

Substituting the second derivative of the displacement function into Equation 2.103 gives

M1 =

∂ EI ∂θ1 2

L

∫ ( N′′ d 1

0

1y

2 + N ′′θ 2 1 + N 3′′ d 2y + N ′′θ 4 2 ) dx (2.104)

Since the modulus of elasticity and moment are independent of x, Equation 2.104 becomes L

M1 =

EI 2 ( N1′′d1y + N ′′θ 2 1 + N 3′′ d 2y + N ′′θ 4 2 ) N ′′ 2 dx (2.105) 2

∫ 0

or M1 = k21v1 + k22θ1 + k23v2 + k24θ2 (2.106) where L

∫

k ij = EI N ′′i N ′′j dx (2.107) 0

Finally, the nodal moment at node 1 is expressed as

M1 =

EI 6 d1y + 4L2 θ1 − 6Ld 2y + 2L2 θ2 (2.108) L3

(

)

By using the same procedure for nodal force and moment at node 2, the stiffness equation for a horizontal beam can be expressed as

F1y 12 M1 EI 6L = 3 F2y L −12 6L M 2

6L −12 6L d1y 4L2 −6L 2L2 θ1 −6L 12 −6L d 2y 2L2 −6L 4L2 θ2

(2.109)

49

Bar and be a m struc tures

2.6

Analyzing a horizontal beam structure

A horizontal beam structure, shown in Figure 2.9, is made of two solid cylinders with different materials and radii. Determine the displacement and slope at the points where force or moment is applied. For Beam 1: E = 210 GPa, D = 5 cm; and for Beam 2: E = 180 GPa, D = 4 cm, where D is the diameter of the cylinder. There are unlimited options for elements and nodes distribution, and some of these options are shown in Figure 2.10. Increasing the number of elements will definitely enhance the accuracy of the results, but only up to a certain number of elements. After this number, the results become independent of the number of elements. The first mesh contains just two elements, which is the minimum to solve this problem. The second mesh contains 3 elements, the third mesh contains 6 elements, the fourth mesh contains 12 elements, and the fifth mesh contains 24 elements. For an illustration purpose, the first mesh is selected because it has the minimum number of elements. The first moments of inertia of the first and second beams are required to solve the problem, and they are

I1 =

π 4 π R = 2.5 × 10 −2 4 4

(

)

4

= 3.067 × 10 −7 m 4

500 N Beam 1

Beam 2 250 N . m

2m

1m

FIGURE 2.9 Beam structure. F2y

Mesh 1

Element 1 Node 1

Mesh 2 Mesh 3 Mesh 4 Mesh 5

FIGURE 2.10 Meshes for the beam structure.

Element 2 Node 2

Node 3

M3

50

Finite Element Simul ations Using ANSYS

I2 =

π 4 π R = 2.0 × 10 −2 4 4

(

)

4

= 1.256 × 10 −7 m 4

First, the stiffness matrix for each element is obtained using Equation 2.85. For the first element, which has nodes 1 and 2, the stiffness matrix is

9 −7 210 × 10 3.067 × 10 ( ) [K (1) ] = 23

12 12 12 16 −12 −12 12 8

1

1

2

2

9.66 9.66 −9.66 9.66

9.66 12.88 −9.66 6.44

−9.66 −9.66 9.66 −9.66

9.66 6.44 −9.66 12.88

[K (1) ] = 10 4

−12 12 −12 8 12 −12 −12 16

1 1 2 2

For the second element, which has nodes 2 and 3, the stiffness matrix is

9 −7 180 × 10 1.256 × 10 ( ) [K ( 2) ] = 13 2

[K ( 2) ] = 10 4

2

12 6 6 4 −12 −6 6 2 3

−12 6 −6 2 12 −6 −6 4

3

27.13 13.56 −27.13 13.56 2 13.56 9.04 −13.56 4.52 2 −27.13 −13.56 27.13 −13.56 3 13.56 4.52 −13.56 9.04 3

Assembling [K(1)] and [K(2)] using Equation 2.86 yields: F1y 9.66 9.66 M 1 F2y = −500 4 −9.66 10 = 9.66 M2 = 0 0 F3y = 0 0 M 250 = − 3

9.66 12.88 −9.66 6.44 0 0

d =0 0 0 1y −9.66 9.66 0 0 θ1 = 0 −9.66 6.44 36.79 3.9 −27.13 13.56 d 2y 3.9 21.88 −13.56 4.56 θ2 −27.13 −13.56 27.13 −13.56 d 3y 13.56 4.52 −13.56 9.04 θ 3

The first and second columns and rows are deleted to remove the singularity from the stiffness matrix, and it becomes

51

Bar and be a m struc tures

F2y = −500 M2 = 0 F3y = 0 M3 = −250

36.79 3.9 −27.13 3.9 21.92 −13.56 4 = 10 −27.13 −13.56 27.13 13.56 4.52 −13.56

13.56 d 2y 4.56 θ2 −13.56 d 3y 9.04 θ3

There are four equations and four unknowns, and solving for displacements and rotations, the results are d2y = –0.0279 m d3y = –0.0562 m

θ2 = –0.0227 rad

θ3 = –0.0337 rad

2.7

Analyzing a horizontal beam structure using ANSYS

The horizontal beam structure shown in Figure 2.9 is made of two solid cylinders with different materials and radii. Determine the maximum displacement and reactions at the support using ANSYS. Divide each meter of the beam by 50 elements. For Beam 1: E = 210 GPa, D = 5 cm, and for Beam 2: E = 180 GPa, D = 4 cm. Double Click on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select Structural OK

52

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add... The beam element can support the moment and the selected element has two nodes.

A select Beam B select 2 node 188 OK

Bar and be a m struc tures

53

Options… The illustrated theory for the beam in this chapter is based on the third-order polynomial for deflection function. Therefore, in options, the element behavior should be changed to a cubic form.

A select Cubic Form. in Element behavior K3 OK

54

Finite Element Simul ations Using ANSYS

Close The beam cross-sectional area is specified in the beam section. In this example, two beams with different radii and material properties will be modeled. Main Menu > Preprocessor > Material Props > Material Models

A Click on Structural > Linear > Elastic > Isotropic The following window will appear, the modulus of elasticity for the first beam will be specified, and any value for the Poisson ratio is required to avoid an error message.

Bar and be a m struc tures

55

A type 210e9 in EX B type 0 in PRXY OK In the following steps, the material of the second beam will be specified. The ID for the first beam is one, while the ID for the second beam is two. In the Define Material Models Behavior

A click on Material, then New Model

A type 2 in Define Material ID OK

56

Finite Element Simul ations Using ANSYS

A click Material Model Number 2 B click on Structural > Linear > Elastic > Isotropic The following window will appear, the modulus of elasticity for the second beam will be specified, and any value for the Poisson ratio is required to avoid an error message.

A type 180e9 in EX B type 0 in PRXY OK

Bar and be a m struc tures

57

Close the Material Model Behavior window Main Menu > Preprocessor > Sections > Beam > Common Sections

A select the solid cylinder in Sub-Type B type 0.025 in R Apply Meshview

ANSYS graphics show the properties of the solid cylinder

58

Finite Element Simul ations Using ANSYS

A type 2 in ID B select the geometry of a solid cylinder in Sub-Type C type 0.02 in R OK The process of modeling the beam structure is started here. First, three keypoints are created followed by creating the lines. The x- and y-coordinates for all keypoints are specified: (0,0) is the coordinate for Keypoint 1, (0,2) for Keypoint 2, and (0,3) for Keypoint 3. Main Menu > Preprocessor > Modeling > Create > Keypoints > In Active CS

A type 1 in Keypoint number B type 0 and 0 in X,Y,Z Location in active CS Apply

Bar and be a m struc tures

59

A type 2 in Keypoint number B type 2 and 0 in X,Y,Z Location in active CS Apply

A type 3 in Keypoint number B type 3 and 0 in X,Y,Z Location in active CS OK Main Menu > Modeling > Create > Lines > Lines > Straight Line Click on Keypoint 1 then 2, and then click on Keypoint 2 then 3. In Create Straight Line window, click on OK

ANSYS graphics show the three keypoints and two lines The modeling task is ended at this point. Before meshing the lines, the radius and material properties should be specified for the two lines. All lines by default have ID number 1. Hence, only properties of Beam 2 should be changed to ID number 2. Each meter of the beam is divided into 50 elements. Therefore, the first beam will be 100 elements, while the second beam will be 50 elements.

60

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Meshing > Mesh Tool

A select Lines B click on Set Select Line 2 only. In Line Attributes window, click on OK The following Line Attributes window will show up. By selecting number 2 for material number and element section, the properties of number 2 are assigned to the line number 2.

A select 2 in Material number B select 2 in Element section OK

Bar and be a m struc tures

61

Main Menu > Preprocessor > Meshing > Mesh Tool

A click on Set in Lines In Element Size on Picked Lines window, click on Pick All In Element Size on Picked Lines window, there are two options: specify the length of elements or the number of element divisions for a line. For the present example, each meter of the beam is divided into 50 elements. Or, the length of the elements is 0.02.

A type 0.02 in Element edge length OK

62

Finite Element Simul ations Using ANSYS

A click on Mesh In Mesh Lines window, click on Pick All Utility Menu > Plot > Nodes

ANSYS graphics show the nodes Modeling and meshing are completed at this point. Next, the boundary conditions are applied starting with the support and then the force and moment. This order is not important for the solution. Loads are

Bar and be a m struc tures

63

applied at the keypoints, which will automatically be transferred to the nodes. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Keypoints Click on Keypoint number 1. In Apply U,ROT on KPs window, click on OK

A select All DOF B type 0 in Displacement value OK Main Menu > Solution > Define Loads > Apply > Structural > Force/Moment > On Keypoints Click on Keypoint 2. Then, in Apply F/M on KPs window, click on OK

64

Finite Element Simul ations Using ANSYS

A select FY in Direction of force/mom B type –500 in Force/moment value Apply Click on Keypoint 3. Then, in Apply F/M on KPs window, click on OK

A select MZ in the Direction of force/mom B type –250 in the Force/moment value OK

ANSYS graphics show the nodal force and moment with direction The final step is to run the ANSYS solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the problem.

Bar and be a m struc tures

65

Main Menu > Solution > Solve > Current LS

OK

Close Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

ANSYS graphics show the beams before and after applying the loads

66

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > List Results > Nodal Solution

A click on Nodal Solution > DOF Solution > Y-Component of displacement. OK

A list of nodal displacements in the y-direction is shown. The maximum nodal displacement is shown at the end of the file, and node 102 has the highest deflection of –0.057388. Comparing with only two elements

Bar and be a m struc tures

67

mesh solution in the previous example, maximum nodal displacement was −0.0562 m, which is close to the ANSYS result. The nodal reactions are determined in the following step: Main Menu > General Postproc > List Results > Reaction Solution

A select All items OK

The results are for node 1 only, the fixed node boundary, and the moment reaction is listed. Notice that the reaction force and moment at the fixed boundary are equal to in negative value the applied force and moment, which ensures the balance of force and moment in the structure.

2.8

Development of an arbitrary-oriented beam element

Practically, the beam elements are not horizontal, but they can be arbitrary oriented in two- or three-dimensional space to form a complex structure. For two-dimensional space, each node possesses three

68

Finite Element Simul ations Using ANSYS

degrees of freedom: x-displacement, y-displacement, and rotation. Additionally, there are forces in the x- and y-directions and a bending moment at each node. Figure 2.11 shows an arbitrary-oriented beam element at an angle φ with the global x-axis. At node 1, the local x-direction and y-direction forces are F1x and F1y, respectively, and the bending moment is M1. At node 2, the local x-direction and y-direction forces are F2x and F2y, respectively, and the bending moment is M2. These forces and bending moments cause the beam to bend. Node 1 is displaced in the local x- and y-directions d1x and d1y, and rotated at an angle θ1, while node 2 is displaced in the local x- and y-directions d2x and d2y, and rotated at an angle θ2. The beam has modulus of elasticity E, uniform cross-sectional area A, moment of inertia I, and initial length L. Since the solution for axial displacements, the transverse deflection, and rotation is independent, the bar and horizontal beam are combined to form an axial flexural beam element. The bar element has a stiffness equation of

F1x F2x

AE 1 −1 d1x = L −1 1 d2x

(2.110)

and a horizontal beam element has a stiffness equation of

F1y 12 6L −12 6L M1 EI 6L 4L2 −6L 2L2 = 3 F2y L −12 −6L 12 −6L 6L 2L2 −6L 4L2 M 2

d1y θ1 (2.111) d2y θ2

Fy2, dy2 No

A,

E,

L,

I

Fx2, dx2 de

2

M2, θ2

y

Fy1, dy1 No

Fx1, dx1

de

1

ϕ

M1, θ1

FIGURE 2.11 An arbitrary-oriented beam element at an angle φ.

x

69

Bar and be a m struc tures

Hence, the stiffness equations of bar and beam are combined as EA EA 0 0 0 0 − L L 12 EI 6 EI 12 EI 6 EI d F1x 0 0 − L3 L2 L3 L2 1x F1y 6 EI 4 EI 6 EI 2 EI d1y 0 − M 0 1 L2 L L2 L θ1 = EI F2x − EI 0 0 0 0 d2x L F2y L d2y 12 EI 6 EI 12 EI 6 EI 0 − 2 θ2 M 2 0 − 3 − 2 3 L L L L 6 EI 2 EI 6 EI 4 EI 0 − 2 0 2 L L L L (2.112) The axial flexural beam element is lying along the local x-axis and the local x-axis is oriented at an angle φ measured counterclockwise from the global x-axis. The transformation matrix [T] is used to transfer the local nodal forces and moments, and nodal displacements and rotation from local coordinate to the global one. The nodal displacements and rotation vector in a symbolic form are as follows:

{d} = [ T ]{d} (2.113)

and the nodal forces and bending moment vector in a symbolic form are

{F} = [ T ]{F} (2.114)

where

cos θ sin θ 0 0 0 0 − sin θ cos θ 0 0 0 0 0 0 1 0 0 0 [ T ] = 0 0 0 cos θ sin θ 0 0 0 0 − sin θ cos θ 0 0 0 0 0 1 0

(2.115)

{}

Since {F } = K d , and using the transformation matrix, the following equation is obtained:

[ T ]{F } = K [ T ]{d} (2.116)

70

Finite Element Simul ations Using ANSYS

By multiplying both sides of Equation 2.116 by the inverse of the transformation matrix, the global force vector is expressed as

{F } = [ T ]−1 K [ T ]{d} (2.117)

Since the transformation matrix is orthogonal, its transport and inverse are equal, and therefore,

{F } = [ T ]T K [ T ]{d} (2.118)

Since the global force vector, {F} = [K]{d}, is equal to the global force vector in Equation 2.118, the global stiffness matrix can be written as

[ K ] = [ T ]T K [ T ] (2.119)

The stiffness matrices, force vectors, and displacement vectors are assembled as follows:

[K ] =

N

∑ K

(e)

e =1

(2.120)

N

{F } =

∑ {F } (2.121) (e)

e =1

{d} =

N

∑ {d } (2.122) (e)

e =1

Finally, the global stiffness matrix is used to relate the global nodal force vector to global displacement vector for the entire structure by {F} = [K]{d}

2.9

(2.123)

Distributed load on a beam element

Unlike the bar elements, beam elements can support a distributed load. The distributed load can be uniform, varying linearly, or randomly distributed. The work equivalence method is used to replace the distributed load by concentrated nodal forces and moments. Consider the beam subjected to a distributed load, as shown in Figure 2.12a. The work due to the distributed load is expressed as L

∫

Wdist = P ( x ) v ( x ) dx (2.124) 0

71

Bar and be a m struc tures

y

P(x)

F1y

x

F2y

M1

M2 L

L

(b)

(a)

FIGURE 2.12 (a) Beam that is subjected to a distributed load and (b) its equivalent nodal forces and moments.

where P(x) is the distributed load on beam with length L and v(x) is the transverse displacement of a beam. The work due to the discrete nodal forces and moments is given by Wdisc = M1θ1 + M2θ2 + F1yd1y + F2yd2y (2.125) Hence, the distributed load can be replaced by equivalent nodal forces and moments, as shown in Figure 2.12b. M1, M2, F1y, and F2y are determined by setting Wdist = Wdisc, as follows: L

2

∫P(x) L (d 3

0

1y

− d 2y ) +

1 3 1 θ + θ2 ) x 3 + − 2 (d1y − d 2y ) − (2θ1 − θ2 ) x 2 2( 1 L L L

+ θ1x + d1y dx = M1θ1 + M 2 θ2 + F1y d1y + F2y d 2y

(2.126) For example, consider the case where the applied distributed pressure is uniform on a horizontal beam, P(x) = −Po. Integrating Equation 2.126 with respect to x between 0 and L gives the following:

2 1 − 3 ( d1y − d 2y ) + 2 ( θ1 + θ2 ) L L L4 3 1 (2.127) − − ( d1y − d 2y ) − ( 2θ1 − θ2 ) 4 L2 L L3 L2 − θ1 − d1y L = M1θ1 + M2 θ2 + F1y d1y + F2y d 2y 3 2

then,

F1y = −

Po L (2.128) 2

M1 = −

Po L2 (2.129) 12

72

Finite Element Simul ations Using ANSYS

F2y = −

M2 =

Po L (2.130) 2

Po L2 (2.131) 12

Or, in matrix form,

F1y M {Fdist } = 1 = F2y M2

Po L 2 Po L2 − 12 Po L − 2 Po L2 12 −

(2.132)

In general, the concentrated nodal forces and moments, which are equivalent to the distributed load, are added to the applied nodal forces and moments to obtain the actual global nodal forces and moments as follows: [K]{d} = {F} + {Fdist} (2.133) For a linearly increasing pressure along a beam, P(x) = −Pox, the distributed load can be replaced by equivalent nodal forces and moments, M1, M2, F1y, and F2y. Equation 2.126 is integrated, and the equivalent nodal forces and moments are determined by setting Wdist = Wdisc, and the final result is

3 PL − 20 F1y PL2 M1 30 {Fdist } = + (2.134) F2y − 7PL 20 M2 5PL2 − 96

2.10

nalyzing beam structure under a transient loading A using ANSYS

For the three-dimensional beam structure shown in Figure 2.13, a transient harmonic force, with a frequency of f = 1/500 s−1 and amplitude of A = 100 kN, is applied at the shown location. Use ANSYS to determine

73

Bar and be a m struc tures

9

Fx(t)

4m 8

7

5

6

4m

2

m

z

x 4m

4

1

3

4

y

FIGURE 2.13 A structure under transient loading.

the displacement at the point where the force is applied as a function of time. Also, create an animation file. The total time for the loading process is 2500 seconds. The beams are made of circular cross-sectional pipe, E = 200 GPa, and with inner and outer radii of Ri = 0.12 m, Ro = 0.15 m, respectively. Mesh each beam by 10 elements. Use the following formula to simulate the applied transient force: Fx (t) = A sin (2πft) Double clicks on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select Structural OK

74

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add... The beam elements can support the moment and have three- dimensional capability.

A select Beam B select 2 node 188 OK

Bar and be a m struc tures

Options…

A select Cubic Form. in Element behavior K3 OK

75

76

Finite Element Simul ations Using ANSYS

Close For material properties, only the modulus of elasticity is required. In this example, all beams have the same geometry and material properties. Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic

Bar and be a m struc tures

77

A type 200e9 in EX B type 0 in PRXY OK Close the Material Model Behavior window The beam cross-sectional properties are specified in sections. The geometry of the cross-sectional area of the pipe will be selected and radii are specified. The number of divisions along the circumference is 20 to ensure a smooth circular cross section. Main Menu > Preprocessor > Sections > Beam > Common Sections

A B C D

select the cross-sectional geometry of a pipe type 0.12 in Ri type 0.15 in Ro type 20 in N

Apply Meshview OK

78

Finite Element Simul ations Using ANSYS

ANSYS graphics will show geometrical properties of the pipe The process of modeling the beam structure is started here. First, keypoints are created, followed by creating the line connecting the keypoints. The x- and y-coordinates of each keypoint are specified. The coordinate for Keypoint 1 is (0,0,0). Main Menu > Preprocessor > Modeling > Create > Keypoints > In Active CS

A type 1 in Keypoint number B type 0, 0, and 0 in X,Y,Z Location in active CS Apply Similarly for other keypoints, (4,0,0) is the coordinate for Keypoint 2; (4,0,–4) is for Keypoint 3; (0,0,–4) is for Keypoint 4; (0,4,0) is for Keypoint 5; (4,4,0) is for Keypoint 6; (4,4,–4) is for Keypoint 7; (0,4,–4) is for Keypoint 8; and (2,8,–2) is for Keypoint 9. To view the keypoints at different directions, the Pan-Zoom-Rotate is used:

Bar and be a m struc tures

79

Utility Menu > PlotCtrls > Pan-Zoom-Rotate

A B C D E

this group is for viewing a geometry at different directions this group is for zooming at a specific region in ANSYS graphics this group is for zooming and panning this group is for rotating the geometry this group is for fitting the geometry, resetting, closing, and helping

Close Main Menu > Modeling > Create > Lines > Lines > Straight Line click on Keypoint 1 then 5 click on Keypoint 2 then 6 click on Keypoint 3 then 7 click on Keypoint 4 then 8 click on Keypoint 5 then 6 click on Keypoint 6 then 7 click on Keypoint 7 then 8 click on Keypoint 8 then 5 click on Keypoint 5 then 9 click on Keypoint 6 then 9 click on Keypoint 7 then 9 click on Keypoint 8 then 9 OK

80

Finite Element Simul ations Using ANSYS

ANSYS graphics show the keypoints and lines The modeling task is ended at this point. Before meshing the lines, each meter of the beam is divided into 10 elements. Therefore, the beams’ element length is 0.1 m. Main Menu > Preprocessor > Meshing > Mesh Tool

A click on Set in Lines In Element Sizes on Picked Lines window, click on Pick All The following window will show up. In Element Sizes on Picked Lines window, there are two options, specifying either the length of

Bar and be a m struc tures

81

e lements or the number of element divisions. For the present example, each meter of the beam should be divided into 10 elements. Alternatively, the length of the element is 0.1.

A type 0.1 in Element edge length OK

A click on Mesh In Mesh Lines window, click on Pick All

82

Finite Element Simul ations Using ANSYS

Utility Menu > Plot > nodes Modeling and meshing is completed at this point. Next, the solution is switched to transient, and unsteady parameters will be specified. The boundary conditions are applied starting with the supports, and then the transient force. This order is not important for solution. Main Menu > Solution > Analysis Type > New Analysis

A select Transient OK

OK The total time duration for the process is 2500 seconds. To ensure that the obtained results are accurate, the total time duration is divided into 100 time steps. Hence, the time step for this problem is 25 seconds. Results for all time steps are stored by selecting Write every substep option in Frequency. Otherwise, only the result at time step 2500 seconds will be stored.

Bar and be a m struc tures

83

Main Menu > Solution > Analysis Type > Sol’n Control

A B C D

type 2500 in Time end of loadstep select Time increment type 25 in Time step size select Write every substep

OK The function editor is used to apply transient force formula as a nodal force. This technique is simple and convenient for this problem since an equation for the force is given. The function is plotted to ensure correct typing. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Keypoints Click on Keypoint numbers 1, 2, 3, and 4. In Apply U,ROT on KPs window, click on OK

A select All DOF B type 0 in Displacement value OK

84

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Define Loads > Apply > Functions > Define/Edit

A type the equation: 100e3*sin(2*3.1415/500*{TIME}) B click on GRAPH

A type 0 and 2500 in X-Axis Range B type 100 in Number Of Points Graph Close Then 0 and 2500 are the range of the data on the x-axis, while 100 is the number of points to be plotted. Number Of Points has nothing to do with the accuracy of the results, and a higher number will just create a smooth plot. The equation should be saved.

Bar and be a m struc tures

85

ANSYS graphics show the oscillating function of the force In Equation Editor window, click on File then Save Save the file as ForceX, and this file name is optional. After saving the function, it is required to load it to the ANSYS solution using the read file. Save Close the Function Editor window Main Menu > Solution > Define Load > Apply > Functions > Read File Select the file ForceX from Open window, and then click on Open

86

Finite Element Simul ations Using ANSYS

A type ForceX in the Table parameter name. This name is optional, and should not be the same as the file name of the function. OK Utility Menu > Plot > Lines Main Menu > Solution > Define Loads > Apply > Structural > Force/Moment > On Keypoints Click on Keypoint number 9. In Apply F/M on KPs window, click on OK

A select FX in the Direction of force/mom B select Existing table OK The following window will show up to select the function.

A select FORCEX OK The final step is to run the ANSYS solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the problem.

Bar and be a m struc tures

87

Main Menu > Solution > Solve > Current LS

OK

Close An animation of the deformed structure from time = 0 to 500 seconds can be easily accomplished using animate in the PlotCtrls option in Utility Menu. The Number of the frames in the animation over time is the number of pictures in the avi file, while the animation time delay is the display period between two pictures. Fifty frames produce a good resolution file. With 0.5 s, the file duration is 250 s (500 frames × 0.5 delay). Main Menu > General Postproc Utility Menu > PlotCtrls > Animate > Over time …

A type 50 in Number of animation frames B select Time Range

88

Finite Element Simul ations Using ANSYS

C type 0 and 500 in Range Minimum, Maximum D select DOF solution E select Deformed Shape OK ANSYS creates an animation for the deformation. The displacement at any point in the structure can be displayed as a function of time using the time history postprocessor. For this problem, the displacement at the point where the force is applied as a function of time is required. Main Menu > TimeHist Postpro

A click on the green + button

A click on Nodal Solution > DOF Solution > X-Component of displacement. OK

Bar and be a m struc tures

89

Click on the Keypoint number 9, where the force is applied, and in Node for Data window, click on OK

A click on the graph button OK

ANSYS graphics show the displacement history of the selected location

90

Finite Element Simul ations Using ANSYS

As shown, the maximum displacement in the x-direction is 0.00420747 m. PROBLEM 2.1 For the bar structures shown in Figure 2.14a through 2.14d, determine the horizontal and vertical nodal displacements, using the finite element method, given that E = 210 GPa, A = 2.5 × 10 –4 m2, F = 0.5 kN, and L = 0.5 m. PROBLEM 2.2 For the bar structures shown in Figure 2.14a through 2.14d, determine the horizontal and vertical nodal displacements, using ANSYS, given that E = 210 GPa, A = 2.5 × 10 –4 m2, F = 0.5 kN, and L = 0.5 m. PROBLEM 2.3 The horizontal beam structures, shown in Figure 2.15a through 2.15d, are made of different materials and cross-sectional areas. Determine the nodal displacements and slopes, using the finite element method, given E1 = 270 GPa, E2 = 215 GPa, E3 = 170 GPa, I1 = 2.5 × 10 –6 m4, I2 = 5 × 10 –6 m4, and I3 = 7.5 × 10 –6 m4. L/2

3 L L

L

F

45˚ 2×F

(a)

(b) 2×F

L

L/2 L

L F

(c)

FIGURE 2.14 Bar structures.

(d)

3×F

91

Bar and be a m struc tures

E1, I1

E1, I1

1.25 kN . m 2.75 m

3.25 m (a) 0.5 kN

0.1 kN

E1, I1

E2, I2

2.5 m

2.5 m (b)

1.25 kN

0.75 kN

E1, I1

E2, I2

E1, I1

1.0 m

1.0 m

1.0 m

(c) 1.5 kN E1, I1

E2, I2

E3, I3

1.25 kN . m 1.5 m

3.0 m

1.0 m

(d)

FIGURE 2.15 Horizontal beam structures.

PROBLEM 2.4 The beam structures, shown in Figure 2.16a through 2.16d, are made of solid cylinders with different materials and radii. Determine the maximum nodal displacement and reaction at the support(s) using ANSYS. Divide each meter of the beam by 25 elements, given E1 = 170 GPa, E2 = 210 GPa, E3 = 270 GPa, D1 = 2 cm, D2 = 5 cm, and D3 = 7 cm.

92

Finite Element Simul ations Using ANSYS

0.5 kN

0.75 kN

E1, D1

E2, D2

120 cm

120 m

(a) 0.75 kN E1, D1

E2, D2

E3, D3 1.25 kN . m

125 cm

250 cm

100 cm

(b) 2.5 kN 0.1 kN 0.2 kN . m

All beams E1, D1

100 cm

75 cm

150 cm

75 cm

(c) 1.5 kN . m D2

D2 75 cm

D2

D2

D1

D1

75 cm

75 cm

All beams E1

0.1 kN

(d)

FIGURE 2.16 Two-dimensional beam structures.

PROBLEM 2.5 For the three-dimensional beam structure, shown in Figure 2.17, transient harmonic forces with a frequency of f = 1/250 s−1 and amplitude of A = 0.85 kN are applied at the shown locations. Use ANSYS to determine the displacement at the points where the forces are applied as a function of time. Also, create an animation for the loading process. The total time duration for the loading process is 1250 seconds. The beams are made of rectangular cross-sectional pipe, as shown in Figure 2.17, and E = 180 GPa. Mesh each beam by 10 elements. Use the following formula to simulate the applied transient forces: Fx (t) = A sin (2 π f t)

93

Bar and be a m struc tures

Fx(t) Fx(t)

5m 10 cm

1.0 cm

10 cm Beam cross section

5

m

5m

5m

FIGURE 2.17 Three-dimensional beam structure with transient forces.

PROBLEM 2.6 For the beam structure shown in Figure 2.18, two transient harmonic forces are applied at the shown locations. Use ANSYS to determine the displacement at the points where the forces are applied as a function of time. Also, create an animation for the loading process. The total time duration for the loading process is 5000 seconds. The beams are made of circular cross-sectional pipe, E = 200 GPa, with inner and outer radii of R i = 0.1 m, Ro = 0.11 m, respectively. Mesh each beam by 10 elements. Use the following formulas to simulate the applied transient forces:

2 πt F1 ( t ) = 100 sin 500

2 πt F2 ( t ) = 200 sin 1000 F1(x)

1.25 m

2.5 m

FIGURE 2.18 Horizontal beam structure with transient forces.

F2(x)

1.25 m

This page intentionally left blank

Cha p t e r TH REE

Solid mechanics

3.1

Stress–strain relations

In this chapter, finite element development of two-dimensional solid elements is described. The development is limited to plane stress–strain elements. The plane stress is defined as the state of stress in which the normal and shear stresses are perpendicular to the plane. Alternatively, the loads on the body are in the xy-plane only. The plane strain is defined as the state of strain in which the normal and shear strain are perpendicular to the plane. Stress is defined as the magnitude of force ΔF on a unit area ΔA, as the unit area approaches zero,

∆F (3.1) ∆A → 0 ∆A

σ = lim

Figure 3.1 shows a two-dimensional stress acting on an element with width dx and height dy. The element is treated as two-dimensional with a unit depth. The normal stresses σx and σy are acting in the x- and y-directions, respectively. Shear stress τxy is acting in the y-direction and normal to the x-plane. Shear stress τyx is acting in the x-direction and normal to the y-plane. From the moment of equilibrium, τxy must be equal to τyx. The equilibrium equation for two-dimensional stresses can be obtained, and the net force in the x-direction should be zero:

∂τ yx ∂σ x dx − τ yx dx (3.2) σ x + dy − σ x dy + τ yx + ∂x ∂ y Simplifying Equation 3.2 yields

∂σ x ∂τ yx + = 0 (3.3) ∂x ∂y

Similarly, the equilibrium equation for two-dimensional stresses in the y-direction is

∂σ y ∂τ xy + = 0 (3.4) ∂x ∂y

95

96

Finite Element Simul ations Using ANSYS

σy + dσy τyx + dτyx τxy + dτxy σx

σx + dσx τxy

y

FIGURE 3.1 Two-dimensional state of stress.

τyx

x

σy

Strain is defined as the magnitude of elongation Δδ due to stresses divided by its original length ΔL, as ΔL approaches zero, ∆δ (3.5) ∆L → 0 ∆L

ε = lim

Figure 3.2 shows a state of two-dimensional strain. There are two types of strain: longitudinal strain due to stress, as shown in Figure 3.2a, and shearing strain due to shear stress, as shown in Figure 3.2b. The shearing strain is defined as the change in value of the originally right angle in an unstrained state. In Figure 3.2a, u is the displacement in the x-direction and v is the displacement in the y-direction. From Figure 3.2a, the strain in the x-direction can be expressed as u + du dx − u du dx ε x = lim = (3.6) dx → 0 dx dx

dv dy dy

du dy dy dy

v

y

dy u dx

du dx dx

dv dx dx

y dx x

x (a)

FIGURE 3.2 (a) Longitudinal and (b) shear strains.

(b)

97

Solid mechanics

Similarly in the y- and z-directions, the strains are, respectively,

dv v + dy dy − v dv ε y = lim = (3.7) dy → 0 dy dy w + dw dz − w dw dz εz = lim = (3.8) dz → 0 dz dz

As shown in Figure 3.2b, the shearing strain in the xy, xz-, and yz- planes is equal to the sum of changes in angle, respectively,

γ xy = γ yx

γ xz = γ zx

γ yz = γ zy

∂u ∂v dy dx ∂v ∂ u ∂y (3.9) = ∂x + = + dx dy ∂x ∂ y ∂w ∂u dx dz ∂w ∂ u x ∂ (3.10) + ∂z = = + dx dz ∂x ∂z ∂w ∂v dy dz ∂v ∂w ∂y (3.11) = ∂z + = + dz dy ∂z ∂ y

The stresses and strains can be presented in a vector form as follows:

{σ } =

{ε} =

σx σy σ z (3.12) τ xy τ xz τ yz

εx εy εz γ xy γ xz γ yz

=

∂u ∂x ∂v ∂x ∂w ∂z ∂ u ∂v + ∂ y ∂x ∂ u ∂w + ∂z ∂x ∂w ∂v + ∂ y ∂z

(3.13)

98

Finite Element Simul ations Using ANSYS

According to Hooke’s law, the stress and strain are related by the modulus of elasticity E. The stress–strain relation in the x-direction is expressed as σx = Eεx (3.14) Elongations in the y- and z-directions are caused by stress in the x-direction and stress lateral direction. Using Poisson’s ratio v, which is defined as the lateral strain over the axial strain, the elongation in the y- and z-directions due to x-direction stresses can be determined as follows:

ε y = ε z = − vε x = − v

σx (3.15) E

The shearing stress and the shearing strain are related by modulus of shear G. For the xy-plane, τxy = Gγxy (3.16) where

G=

E (3.17) 2 (1 + v )

Hence, the generalized Hooke’s laws are written as

(

)

εx =

1 σ x − v ( σ y + σ z ) (3.18) E

εy =

1 (σ y − v (σ x + σ z )) (3.19) E

εz =

1 σ z − v ( σ x + σ y ) (3.20) E

γ xy =

2 (1 + v ) τ xy (3.21) E

γ yz =

2 (1 + v ) τ yz (3.22) E

γ zx =

2 (1 + v ) τzx (3.23) E

(

)

Expressing Equations 3.18–3.20 and 3.21–3.23 in terms of strain components as

σx =

vE E ε x + ε y + εz ) + ε x (3.24) ( 1 v 1 2v 1 v + − + ( )( )

σy =

vE E ε x + ε y + εz ) + ε y (3.25) ( 1+ v (1 + v )(1 − 2v )

σz =

vE

(1 + v )(1 − 2v )

( εx + ε y + εz ) + 1 +E v εz (3.26)

99

Solid mechanics

τ xy =

E γ xy (3.27) 2 (1 + v )

τ yz =

E γ yz (3.28) 2 (1 + v )

τzx =

E γ zx (3.29) 2 (1 + v )

For plane stress cases, all stresses in the z-direction are assumed zero, σz = τyz = τzx, and strain–stress relations are expressed as

εx =

1 (σ x − vσ y ) (3.30) E

εy =

1 (σ y − vσ x ) (3.31) E

εz =

−v (σ x + σ y ) (3.32) E

γ xy =

2 (1 + v ) τ xy (3.33) E

Stress–strain relations are expressed as

σx =

E ( εx + vε y ) (3.34) 1 − v2

σy =

E ( ε y + vεx ) (3.35) 1 − v2

τ xy =

E γ xy (3.36) 2 (1 + v )

For the plane strain case, all strains in the z-direction are assumed to be zero, εz = γyz = γzx, and strain–stress relations are expressed as vE E εx + εy ) + ε x (3.37) ( 1+ v (1 + v )(1 − 2v )

σx =

σy =

σz =

vE = v ( σ x + σ y ) (3.39) (1 + v )(1 − 2v )

τ xy =

E γ xy (3.40) 2 (1 + v )

vE

(1 + v )(1 − 2v )

( εx + ε y ) + 1 +E v ε y (3.38)

Stress–strain relations are expressed as

εx =

1+ v ((1 − v ) σ x − vσ y ) (3.41) E

100

Finite Element Simul ations Using ANSYS

1+ v ((1 − v ) σ y − vσ x ) (3.42) E

εy =

γ xy =

2 (1 + v ) τ xy (3.43) E

In matrix form, the stress–strain relations in the state of plane stress and plane strain are expressed as, respectively, σx σy τ xy

σx σy τ xy

E = − 1 v2

(

)

1 v 0 v 1 0 − 1 v 0 0 2

εx εy γ xy

(3.44)

1− v v 0 E v 1− v 0 = − 1 2v + − 1 v 1 2v ( )( ) 0 0 2

εx εy γ xy

(3.45)

The stress–strain relationships are symbolically expressed as {σ} = [D]{ε} (3.46) where [D] is called the stress–strain matrix, or simply the D matrix.

3.2

evelopment of triangular plane stress D and plane strain element

Figure 3.3a shows a plate subjected to tensile stress. The plate is divided into linear triangular elements, as shown in Figure 3.3b. The linear triangular element has three nodes and nodes can be displaced, but the element’s sides remain straight. As shown in Figure 3.3c, the nodes are named i, j, and m. At each node, there are two degrees of freedom, displacement in the x- and y-directions. The coordinates for the three nodes i, j, and k are (xi,yi), (xj,yj), and (xk,yk), respectively. The nodal displacement vector is given by

{d} =

ui vi uj vj um vm

(3.47)

Therefore, each element has six degrees of freedom. Since it is a l inear element, linear displacement functions are selected for the x- and y-directions as follows: u(x,y) = a1 + a2x + a3y (3.48)

101

Solid mechanics

(a)

(b) vm (xm, ym)

um

vi (xi, yi)

ui

vj

x

(xj, yj)

y

uj

(c)

FIGURE 3.3 (a) Plate subjected to tensile stress, (b) dividing the plate into linear triangular elements, and (c) description of a linear triangular element.

v(x,y) = a4 + a5x + a6y (3.49) The displacement equations can be solved because there are six constants, a1 to a6, and six equations. The nodal x- and y-displacements for all nodes are expressed in the following matrix form:

ui uj um

1 xi = 1 xj 1 xm

y i a1 y j a2 y m a 3

(3.50)

vi vj vm

1 xi = 1 xj 1 xm

yi a 4 y j a5 y m a 6

(3.51)

102

Finite Element Simul ations Using ANSYS

and solving for a’s,

a1 a2 a3

αi 1 = βi 2A γ i

α j α m ui β j βm u j γ j γ m u m

(3.52)

αi 1 = βi 2A γ i

α j α m vi β j βm v j γ j γ m v m

(3.53)

and

a4 a5 a6

where A is the area of the element, which is equal to

A=

1 x i ( y j − y m ) + x j ( y m − y i ) + x m ( y i − y j ) (3.54) 2

The α, β, and γ are defined as αi = xj ym − yj xm

αj = xm yi − ym xi

αm = xi yj − yi xj

βi = yj − ym

βj = ym − yi

βm = yi − yj (3.55)

γi = xm − xj

γj = xi − xm

γm = xj − xi

Substituting the values of α’s, β’s, and γ’s into Equations 3.48 and 3.49 yields

u ( x, y ) =

1 ( α i + β i x + γ i y ) u i + ( α j + β jx + γ j y ) u j (3.56) 2A + ( α m + β m x + γ m y ) u m

v ( x, y ) =

1 ( α i + β i x + γ i y ) v i + ( α j + β jx + γ j y ) v j (3.57) 2A + ( α m + β m x + γ m y ) v m

The strain vector for a two-dimensional element is given in Equation 3.13. Using the displacement in the v- and the u-directions, the strain vector is expressed as εx {ε} = ε y γ xy

β i u i + β ju j + β m u m 1 γ i v i + γ jv j + γ m v m = 2A γ u + β v + γ u + β v + γ u + β v i i j j j j m m m m i i

(3.58)

The strain vector can be expressed in a matrix form as follows:

{ε} =

β 1 i 0 2A γ i

0 γi βi

β j 0 βm 0 γj 0 γ j βj γ m

0 γm βm

ui vi uj vj um vm

(3.59)

103

Solid mechanics

The matrix (3.59) is symbolically expressed as {ε} = [B]{d}

(3.60)

The stress–strain relationship is symbolically expressed as {σ} = [D][B]{d}

(3.61)

The principle of the minimum potential energy theory is used to obtain the stiffness matrix [K]. The total potential energy is the sum of the strain energy U and the potential energy of the total applied nodal forces Fp. The strain energy is given as

∫

U = {ε}T {σ } dV (3.62) V

Using Equations 3.60 and 3.61 in Equation 3.62, the strain energy becomes

U=

∫ [ B] {d} [ D][ B]{d} dV (3.63) T

T

V

The total applied nodal forces can be expressed as {Fp} = −{d}T{F} (3.64) Applying Castigliano’s first theorem,

Fi =

∂U (3.65) ∂d i

Using expressions (3.63) and (3.64) in the definition of Castigliano’s first theorem yields

∂U ∂{d}

∫ [ B] {d} [ D][ B]{d} dV = {d} {F} (3.66) T

T

T

V

Taking the first variation of the potential energy with respect to the nodal displacement vector yields

∫ [ B] [ D][ B]{d} dV = {F} (3.67) T

V

Since the nodal displacements vector is related to the nodal forces vector by [K]{d} = {F}, the stiffness matrix is expressed as

∫

[ K ] = [ B]T [ D ][ B] dV (3.68) V

104

Finite Element Simul ations Using ANSYS

For a constant thickness element t, Equation 3.68 is rewritten as

∫

[ K ] = [ B]T [ D ][ B] t dx dy (3.69) A

The matrices [D] and [B] are not functions of x or y, and thus the integration of dx dy is simply the area of the element A as follows: [K] = tA[B]T[D][B] (3.70) The global stiffness matrix is obtained by assembling the stiffness matrices for all elements as follows:

[K ] =

N

∑ K

(e)

e =1

(3.71)

and the nodal forces are also assembled to form a global force vector as follows:

[F ] =

N

∑ F

(e)

e =1

(3.72)

Also, the global displacements vector can be obtained by

[d] =

N

∑ d

(e)

e =1

(3.73)

Finally, {F} = [K]{d}

(3.74)

To determine the nodal displacements, the global stiffness matrix must be formulated. The following steps are for solving solid mechanics problems:

a. Divide the domain into elements and create a finite element mesh. b. Formulate the [B] matrix for all elements. c. Formulate the [D] matrix for all elements. d. Formulate the [K] matrix for all elements using calculated [B] and [D] matrices. e. Assemble the [K] matrices to create the global [K] matrix. f. Formulate the stiffness equation {F} = [K]{d}. g. Apply the boundary conditions to the stiffness equation to remove the singularity. h. Solve the stiffness equations to determine the unknowns.

105

Solid mechanics

3.3

Analyzing rectangular plate subjected to forces

The thin plate, as shown in Figure 3.4a, is subjected to two forces at its right edges. Determine the nodal displacements at points where forces are applied, given E = 200 GPa, v = 0.3, and t = 0.01 m. Consider the problem as a plane stress. 50 kN

1m

100 kN

2m (a)

(b) (0,1) 3

(2,1) 4 Element 2

y

Element 1

1 (0,0) x (c)

2 (2,0)

FIGURE 3.4 (a) Thin plate subjected to two forces, (b) element distribution for example, and (c) finite element mesh.

106

Finite Element Simul ations Using ANSYS

There are unlimited options for elements and nodes distributions, and some of these options are shown in Figure 3.4b. An increasing number of elements will definitely enhance the accuracy of the results, but up to a certain number of elements. After which, the results become independent of the number of the elements. The first mesh contains just two elements, which is the minimum number of elements to solve this problem. The second mesh contains four elements, the third mesh also contains four elements, but with a different elements distribution. The last mesh has random elements distribution. For illustration purposes, the first mesh is selected for this example, because it has the least number of elements. Figure 3.4c shows the finite element mesh for the present problem. First, the [B], [D], and [K] matrices for Element 1 are formulated. Element 1 has the coordinate shown in Figure 3.4c. Node 1 is named (i) and its coordinate is (0,0), Node 2 is named (j) and its coordinate is (2,0), and Node 3 is named (m) and its coordinate is (0,1). The β’s and γ’s are required for the [B] matrix, and they are calculated using Equations 3.55; A is calculated using (3.54). We have βi = yj − ym = 0 − 1 = −1 βj = ym − yi = 1 − 0 = 1 βm = yi − yj = 0 − 0 = 0 γi = xm − xj = 0 − 2 = −2 γj = xi − xm = 0 − 0 = 0 γm = xj − xi = 2 − 0 = 2 2A = 1 m2 Then, the [B] matrix is formulated using (3.59):

[ B] =

−1 1 0 2 −2

0 −2 −1

1 0 0

0 0 1

0 0 2

0 2 0

and [D] matrix is formulated using (3.44):

1 0.3 0 200 × 10 9 0.3 1 0 [D] = 2 1 − 0.3 1 − 0.3 0 0 2

107

Solid mechanics

Finally, the [K] matrix for Element 1 is obtained using expression (3.70):

1) 11 ( K = 10

1

1

2

2

3

3

1.318 0.714 –0.549 –0.385 –0.769 –0.330

0.714 2.390 –0.330 –0.192 –0.385 –2.198

–0.549 –0.330 0.549 0 0 0.330

–0.385 –0.192 0 0.192 0.385 0

–0.769 –0.385 0 0.385 0.770 0

–0.330 –2.198 0.330 0 0 2.198

1 1 2 2 3 3

Element 2 has the coordinate shown in Figure 3.4c. Node 2 is named (i) and its coordinate is (2,0), Node 4 is named (j) and its coordinate is (2,1), and Node 3 is named (m) and its coordinate is (0,1). The β’s and γ’s are βi = yj − ym = 1 − 1 = 0 βj = ym − yi = 1 − 0 = 1 βm = yi − yj = 0 − 1 = −1 γi = xm − xj = 0 − 2 = −2 γj = xi − xm = 2 − 0 = 2 γm = xj − xi = 2 − 2 = 0 2A = 1 m2 Then the [B] matrix is formulated using (3.59):

[ B] =

0 1 0 2 −2

0 −2 0

1 0 2

0 2 1

−1 0 0

0 0 −1

and [D] matrix is formulated using (3.44):

1 0.3 0 200 × 10 9 0.3 1 0 [D] = 2 1 − 0.3 1 − 0.3 0 0 2

108

Finite Element Simul ations Using ANSYS

Finally, the [K] matrix for Element 2 is obtained using expression (3.70): 2 K ( 2) = 1011

2

4

4

3

3

2 0.769 0 –0.769 –0.385 0 0.385 2 0 2.198 –0.329 –2.198 0.330 0 4 –0.769 –0.330 1.318 0.714 –0.549 –0.385 –0.385 –2.198 0.714 2.390 –0.330 –0.192 4 0 0.330 –0.549 –0.330 0.549 0 3 0.385 0 –0.385 –0.192 0 0.192 3

Using Equation 3.71, the global matrix is formulated as shown:

11 [ K ] = 10

1

1

1.318 0.714 –0.549 –0.385 –0.769 0.330 0 0

0.714 2.390 –0.330 –0.192 –0.385 –2.198 0 0

2 –0.549 –0.330 1.319 0 0 0.715 –0.769 –0.385

3

2 –0.385 –0.192 0 2.390 0.714 0 –0.330 –2.198

–0.769 –0.385 0 0.715 1.319 0 –0.549 –0.330

3

4

4

1 –0.330 0 0 1 –2.198 0 0 0.715 –0.769 –0.385 2 0 –0.329 –2.198 2 0 0.549 –0.330 3 2.390 –0.385 –0.192 3 –0.385 1.318 0.714 4 –0.192 0.714 2.39 4

The force–displacement equation is assembled using expression (3.74):

F1x F 1y F2x = 100 × 10 3 F2y = 0 F3x F3y F4x = 0 F4y = 50 × 10 3

= [K ]

d1x = 0 d1y = 0 d 2x d 2y d 3x = 0 d 3y = 0 d 4x d 4y

Eliminating the first, second, fifth, and sixth columns and rows, F2x = 100 × 10 3 F2y F4x F4y = 50 × 10 3

1.319 0 −0.769 −0.385 0 2.39 −0.329 −2.198 11 = 10 0.769 0.330 1.318 0.714 − − 2.39 −0.385 −2.198 0.714

d 2x d 2y d 4x d 4y

Finally, solving the above equations, the nodal displacements are d2x = 1.665 × 10 –6 m

109

Solid mechanics

d2y = 2.836 × 10 –6 m d4x = 1.156 × 10 –8 m d4y = 3.082 × 10 –6 m

3.4

Development of rectangular plane stress element

The rectangular element has four nodes, and nodes can be displaced but sides remain straight. It is assumed that the stress components are constant within each element. At each node, there are two degrees of freedom, displacement in the x- and y-directions, as shown in Figure 3.5. Each node is subjected to two forces: vertical and horizontal. The width and height of the element are a and b, respectively. Node 1 has coordinate (0,0) and displacement (u1,v1), Node 2 has coordinate (a,0) and displacement (u2,v2), Node 3 has coordinate (a,b) and displacement (u3,v3), and Node 4 has coordinate (0,b) and displacement (u4,v4). The nodal displacement vector is given by

{d} =

u1 v1 u2 v2 u3 v3 u4 v4

(3.75)

Fy4, v4

Fy3, v3

Fx4, u4

Fx3, u3

b y Fx1, u1

x Fy1, v1

Fx2, u2 a Fy2, v2

FIGURE 3.5 Rectangular element with displacements and forces at the nodes.

110

Finite Element Simul ations Using ANSYS

Therefore, there are eight degrees of freedom for each element, and the displacement functions for the x- and the y-directions are u(x,y) = a1 + a2x + a3y + a4xy (3.76) v(x,y) = a5 + a6x + a7y + a8xy (3.77) The values of constants a1 to a8 are obtained by substituting the eight nodal coordinates in the displacement functions in terms of element dimensions, a and b. Rearranging the displacement functions as u(x,y) = f1u1 + f2u2 + f3u3 + f4u4 (3.78) v(x,y) = f1v1 + f2v2 + f3v3 + f4v4 (3.79) where functions f1, f2, f3, and f4, are defined as follows:

x y f1 = 1 − 1 − (3.80) a b

f2 =

y x 1 − (3.81) a b

f3 =

xy (3.82) ab

f4 =

x y 1 − (3.83) b a

The strain vector for a two-dimensional element is given in Equation 3.13. Using the displacements v and u, the strain vector is expressed as

εx {ε} = ε y γ xy

ρ1 0 ρ2 0 ρ3 0 ρ4 0 = 0 µ1 0 µ 2 0 µ 3 0 µ 4 µ1 ρ1 µ 2 ρ2 µ 3 ρ3 µ 4 ρ4

u1 v1 u2 v2 u3 v3 u4 v4

(3.84)

where

1 y ρ1 = − 1 − (3.85) a b

ρ2 =

1 y 1 − (3.86) a b

ρ3 =

y (3.87) ab

111

Solid mechanics

y (3.88) ab

ρ4 = −

1 x µ1 = − 1 − (3.89) b a

µ2 = −

x (3.90) ab x µ3 = (3.91) ab µ4 =

1 1− b

x (3.92) a

The strain vector (3.84) is symbolically expressed as {ε} = [B]{d}

(3.93)

and the stress–strain relationship is expressed as {σ} = [D][B]{d}

(3.94)

The principle of the minimum potential energy theory is used to obtain the stiffness matrix [K]. The total potential energy πp is the sum of the strain energy U and the potential energy of the total applied nodal forces Fp. For a constant thickness element t, the stiffness matrix is rewritten as

∫

[ K ] = [ B]T [ D ][ B] t dx dy (3.95) A

Using Equation 3.95, the stiffness equation for the rectangular element with four nodes in the state of plane stress element is calculated as C1 C2 C4 C5 − C1 /2 − C2 − C5 C7 C C C C C C /2 C C8 − − − 2 3 5 6 2 3 5 C C1 C7 C5 C2 − C5 − C2 − C1 /2 4 C5 C2 C6 C3 C8 − C2 − C5 − C3 /2 [K ] = C4 C5 C7 C1 C2 − C5 − C1 /2 − C2 − C2 − C3 /2 C5 C8 C2 C3 C6 − C5 C5 − C1 /2 C2 C4 − C5 C1 − C2 C7 −C C8 C2 C5 C6 C3 − C3 /2 − C2 5

(3.96) where

b 1 − v a Et C1 = + (3.97) 3a 6 b 1 − v2

112

Finite Element Simul ations Using ANSYS

v 1 − v Et C2 = + (3.98) 4 8 1 − v2

b 1 − v b Et C3 = + (3.99) 3a 6 a 1 − v2

b 1− v C4 = − + 3a 12

v 1 − v Et C5 = + (3.101) 4 8 1 − v2

a 1 − v b Et C6 = + (3.102) 6b 6 a 1 − v2

b 1 − v a Et C7 = + (3.103) 6a 6 b 1 − v2

a 1− v C8 = − + 3b 12

a Et (3.100) b 1 − v2

b Et (3.104) a 1 − v2

The global stiffness matrix is obtained by assembling the stiffness matrices for all elements as follows:

[K ] =

N

∑ K

(e)

e =1

(3.105)

and the nodal forces are assembled to form a global force vector as follows:

[F ] =

N

∑ F

(e)

e =1

(3.106)

Also, the global displacement vector is obtained by N

[ d ] = ∑ d(e) (3.107) e =1

Finally, {F} = [K]{d}

3.5

(3.108)

nalyzing a plate with a hole subjected A to tensile pressure using ANSYS

The square plate with a hole shown in Figure 3.6a is subjected to tensile pressure at both vertical sides. Use ANSYS to determine the maximum stress in the x-direction. Also, compare the ANSYS result with maximum stress using the stress concentration factor chart. The applied tensile

113

Solid mechanics

ϕ 1.25 cm P

P

σ(ave)

P

σ(max)

5 cm

(a)

(b)

K 3.0 2.75 0.025 m

2.5

P

2.25 2.0

R 0.00625 m 0.1

0.2

0.3

0.4

0.5

(c)

0.6

0.7

0.8

0.9

1.0

d/H

0.025 cm

(d)

FIGURE 3.6 (a) A plate with a hole subjected to tensile pressure. (b) A plate with hole under stress, and stress distribution near the hole. (c) Stress concentration factor for a plate with a hole. (d) Symmetry boundary conditions for the problem.

pressure is 100 kN/m2, and let E = 270 GPa, v = 0.3, and consider the plate as a plane stress with thickness of t = 0.005 m. For the plate with a hole shown in Figure 3.6a, the stress along the vertical symmetry line is typically assumed uniform. However, from the experimental observations, the stress is not uniform, but it has a maximum value near the hole, and it is greater than the average stress, as illustrated in Figure 3.6b. The complexity of maximum stress can be conveniently treated using the stress concentration factor K. The maximum stress is equal to the average stress multiplied by the stress concentration factor. The definition of the stress concentration factor is

K=

σ ( max ) (3.109) σ ( ave )

The average stress is calculated as

σ ( ave ) =

P×H

(H − d)

(3.110)

where H is the height of the plate, 5 cm for the present problem; P is plate thickness; and d is the hole diameter. The stress concentration for a solid plate with a hole can be obtained using Figure 3.6c. The x-axis represents the ratio of the diameter of the hole to the height of the plate,

114

Finite Element Simul ations Using ANSYS

while the y-axis is the corresponding stress concentration factor. Note that decreasing the diameter of the hole increases the stress concentration up to three times the average stress. ANSYS is employed to determine the maximum stress. This example is limited to structural analysis. Hence, select Structural only in the Preferences. Solid element is used, and its shape is rectangular with four nodes. Double click on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select Structural OK Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add…

Solid mechanics

115

A select Solid B select Quad 4 node 182 OK The Quad 4 node 182 is a rectangular-shaped element with four nodes. Triangular elements are not available in the ANSYS elements library. A quadratic element with eight nodes is available. The plate has a thickness. In the option, select plane stress with thickness. In Real Constant window, the thickness is specified.

Options…

A select Plane strs w/thk OK

116

Finite Element Simul ations Using ANSYS

Close the Element Type window Close Main Menu > Preprocessor > Real Constants > Add/Edit/Delete

Add…

OK

Solid mechanics

117

A type 0.005 in Thickness OK Close the Element Type for Real Constants window Close For the material properties, the modulus of elasticity and Poisson’s ratio are required to solve the problem. Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic

A type 270e9 in EX B type 0.3 in PRXY OK

118

Finite Element Simul ations Using ANSYS

Close the Material Model Behavior window The geometry is modeled by creating a square and a circle. Boolean operation is utilized to remove the circle from the square using subtraction. The advantage of symmetry in the problem is considered. Only the upper right quarter is considered. Figure 3.6d shows the considered geometry for the problem. Notice that the imposed boundary condition at the vertical symmetry line is zero displacement in the x-direction and zero displacement in the y-direction at the horizontal symmetry line. Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

A type 0 in WP X B type 0 in WP Y C type 0.025 in Width D type 0.025 in Height OK

Solid mechanics

119

Main Menu > Preprocessor > Modeling > Create > Areas > Circle > Solid Circle

A type 0 in WP X B type 0 in WP Y C type 0.00625 in Radius OK

ANSYS graphics show the square and the circle Now, the circular area should be subtracted from the square area to form a plate with a hole. In Boolean operations, there are a number of operations such as adding, subtracting, dividing, and many more.

120

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Modeling > Operate > Booleans > Subtract > Areas Click on square area, and then in Subtract Areas window, click on Apply Click on circular area, and then in Subtract Areas window, click on OK A free mesh is generated using the smart mesh option, and the mesh refinement is 1. The high stresses are expected to be at a region close to the hole, and therefore, more elements will be added at that region using refinement at the line. Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 Mesh

Solid mechanics

121

In Mesh Areas window, click on Pick All

A select Lines B click on Refine Click on the curved line at the hole, and then in Refine Mesh at Line window, click on OK

A turn on the Yes to activate Advanced options OK

122

Finite Element Simul ations Using ANSYS

A type 3 OK

ANSYS graphics show the mesh Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Lines In the ANSYS graphics, click on the left vertical line, where zero x-direction displacement is applied. Then, in Apply U,ROT on Lines window, click on OK

OK A select UX B type 0 in Displacement value Apply

Solid mechanics

123

In the ANSYS graphics, click on the horizontal bottom line, where zero y-direction displacement is applied. Then, in Apply U,ROT on Lines window, click on OK

A select UY B type 0 in Displacement value OK Pressure is applied at the right vertical side of the plate. The negative pressure means that the pressure is tensile. Main Menu > Solution > Define Loads > Apply > Structural > Pressure > On Lines In the ANSYS graphics, click on the right vertical line. Then, in Apply PRES on lines window, click on OK

A type –100e3 in Load PRES value OK

124

Finite Element Simul ations Using ANSYS

The final step is to run the solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the problem. Results can be plotted and listed in the General Preprocessor task. Main Menu > Solution > Solve > Current LS

OK

Close The above windows indicate that the solution task is completed successfully. The next step is for getting the results. The first figure shows the plate before and after deformation. Inspecting this figure is very important to determine if the problem is solved correctly. The second figure is the stress contours in the x-direction. This figure is used to determine the maximum stress in the x-direction, and the stress concentration factor. Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

Solid mechanics

125

ANSYS graphics show the plate before and after deformation The above figure indicates that the left vertical and bottom horizontal lines are fixed, while the right vertical line is moved to the right in the direction of the applied pressure. The result is as expected. Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click on Nodal Solution > Stress > X-Component of stress OK

ANSYS graphics show the x-direction stress contours

126

Finite Element Simul ations Using ANSYS

The stress contours show the location of the maximum stress in the x-direction. As expected, it is at the top of the hole. The accuracy of the result can additionally be improved if the number of elements in the model is increased. As shown in the stress contours, the maximum stress is 360,874 N/m2. The maximum stress can also be obtained using the stress concentration in Figure 3.6c. The average stress is σ ave =

100 × 10 3 × 0.025 = 133333.33 N/m 2 0.025 − 0.00625

Using Figure 3.6c, the stress concentration should be K = 2.34 and the maximum stress is calculated as follows: σmax = Kσave = 2.34 × 133333.33 = 312,000 N/m2 Comparing the ANSYS maximum stress with maximum stress obtained from Figure 3.6c, the error is calculated as

3.6

Error =

312000 − 360874 (100 ) = 15.6% 312400

Axisymmetric elements

An axisymmetric solid that is subjected to axisymmetric loading is commonly found, and it is considered as a special case. Hence, a complex three-dimensional solid model is simplified to a two-dimensional model. Moreover, there are no additional degrees of freedom in the simplification. The axisymmetric solid model to be considered in this illustration is a cylinder with rectangular cross-sectional area, as shown in Figure 3.7a. Because of symmetry in the geometry and loads, no tangential degrees of freedom are assumed and, therefore, all derivations are independent of the circumferential angular variables. The plane strain finite element that is developed in this chapter can be extended to an axisymmetric element, by replacing the integration along the thickness to along the circumference. The cross-sectional area of the axisymmetric solid model is shown in Figure 3.7b. A linear triangular element is used to mesh the cross-sectional area, as shown in Figure 3.7c. The nodal displacement functions in the r- and the z-directions are u(r,z) = a1 + a2r + a3z (3.111) w(r,z) = a4 + a5r + a6z (3.112)

127

Solid mechanics

Z Ro Ri

Z Ro Ri wm um

H wi

H r

r

(a)

ui

z

wj uj

r

(b)

(c)

FIGURE 3.7 (a) Axisymmetric solid model, (b) cross-sectional area of the axisymmetric solid model, and (c) an axisymmetric element.

The cylindrical coordinates for the three nodes are (r1, z1), (r2, z2), and (r3, z3). The displacement functions can be expressed in matrix form for all nodes as

u1 u2 u3 z1 z2 z3

=

1 r1 1 r2 1 r3 0 0 0 0 0 0

z1 z2 z3 0 0 0

0 0 0 0 0 0 1 r1 1 r2 1 r3

0 0 0 z1 z2 z3

a1 a2 a3 a4 a5 a6

(3.113)

and symbolically: {d} = [B]{a}

(3.114)

The constants a’s can be obtained by inverting the matrix [B] as follows: {a} = [B]−1{d} (3.115) The inverse of the matrix [B] is calculated as

1 −1 [ B] = γ

ω1 ω4 ω7 0 0 0

ω2 ω5 ω8 0 0 0

ω3 ω6 ω9 0 0 0

0 0 0 ω1 ω4 ω7

0 0 0 ω2 ω5 ω8

0 0 0 ω3 ω6 ω9

(3.116)

128

Finite Element Simul ations Using ANSYS

where ω1 = r2z3 − r3z2

ω2 = r3z1 − r1z3

ω3 = r1z2 − r2z1

ω 4 = z 2 − z 3

ω5 = z3 − z1

ω6 = z1 − z2 (3.117)

ω7 = r3 − r2

ω8 = r1 − r3

ω9 = r2 − r1

γ = r1 (z2 − z3) + r2 (z3 − z1) + r3 (z1 − z2) The unit elongation of the element in radial direction is expressed as

εr =

∂u (3.118) ∂r

and the total tangential strain is given by

εθ =

u ∂v (3.119) + r r ∂θ

where the second term is the tangential strain due to the displacement in the y-direction. The shearing strain is expressed as

γ rθ = −

v ∂ u ∂v (3.120) + + r r ∂θ ∂r

For an axisymmetric deformation, the displacement in the θ-direction is equal to zero, and hence, the strain and shearing strain are expressed as

εr =

∂u = a 2 (3.121) ∂r

εθ =

u a1 az = + a 2 + 3 (3.122) r r r

εz =

∂w = a 6 (3.123) ∂z

γ rz =

∂ u ∂w = a 3 + a 5 (3.124) + ∂z ∂r

γrθ = 0

(3.125)

γθz = 0

(3.126)

Strain and shearing strain are expressed in matrix form as

εr εθ εz γ rz

=

0 1 0 0 0 0 1/r 1 z /r 0 0 0 0 0

0 0

0 1

0 0 1 0 1 0

a1 a2 a3 a4 a5 a6

(3.127)

129

Solid mechanics

Equation 3.127 is symbolically expressed as {ε} = [G]{a}

(3.128)

Using Equation 3.115 in Equation 3.128 yields {ε} = [A]{d}

(3.129)

[A] = [G][B]−1 (3.130) For an isotropic material, and the case of thermal strain due to temperature rise ΔT, the strain–stress relationship is expressed as

σr σθ σz τ rz

1− ν ν ν ν 1− ν ν E = ν ν − ν 1 (1 + ν )(1 − 2 ν ) 0 0 0

1 − ν 2 0 0 0

εr εθ εz γ rz

α ∆T α ∆T − α ∆T 0

(3.131)

where α is the thermal coefficient of expansion. The symbolic expression for Equation 3.131 is {σ} = [C][A]{d} − [C]{εi} (3.132) Castigliano’s first theorem, which is the same as Equation 3.74 with an additional term for thermal strain, is expressed as follows: {F} = [K]{d} − {Fi} (3.133) where the stiffness matrix is obtained as

∫

[ K ] = [ A ]T [ C ][ A ] dV (3.134) Putting Equation 3.130 into Equations 3.133 and 3.134 yields

(

F i = [ B]−1

) ∫ [G] T

T

[ C ]{ε i } dV (3.135)

Finally, the stiffness matrix is

[ K ] = ([ B]−1 ) [ G ]T [ C ][ G ] dV [ B]−1 (3.136) T

∫

130

Finite Element Simul ations Using ANSYS

3.7

isplacement analysis of a vessel under D transient loading with ANSYS

The vessel shown in Figure 3.8a is subjected to internal transient pressure loading. Determine the maximum displacement in the radial direction of the vessel as a function of time. Take the advantage of symmetry in the problem to reduce the computational size. The inner and outer radii are 0.25 and 0.265 m, respectively, and the height of the vessel is 0.75 m. Let E = 180 GPa and v = 0.33. The upper and lower horizontal surfaces are fixed. The total time duration for the process is 1000 seconds. The internal pressure is a function of time according to the following equation:

2π P ( t ) = 100 × 10 3 + 100 × 10 3 sin ( t + 75) 100

Double click on the Mechanical APDL Product Launcher icon This example is limited to structural analysis. Hence, select Structural only. Solid element is used, and its shape is rectangular with four nodes. The following steps are for selecting the element type and its behavior:

Ro Ri

Ro

dx = 0

Ri

H

dy = 0

H/2

P(t) y

x dy = 0

(a)

(b)

FIGURE 3.8 (a) A vessel subjected to time-dependent pressure at the inner surface. (b) The axisymmetric model for the problem.

Solid mechanics

Main Menu > Preferences

A select Structural OK Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add…

A select Solid B select Quad 4 node 182 OK

131

132

Finite Element Simul ations Using ANSYS

In the element type, Quad means that a quadratic element is selected. The first digit of number 42 is the number of nodes in the element, and the second digit is the number of degrees of freedom in each node, which are the x- and y-displacements. In the following steps, the element behavior is changed from plane stress to axisymmetric. The results from the axisymmetric analysis should be more accurate than the results from three-dimensional elements, and the computational time will be less.

Options…

A select Axisymmetric in Element behavior OK Close the Element Types window Close

Solid mechanics

133

For the material properties, the elasticity and Poisson ratio are required to solve the problem. Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic The following windows will show up.

A type 180e9 in EX B type 0.33 in PRXY OK

134

Finite Element Simul ations Using ANSYS

Close the Material Model Behavior window The geometry is simply a rectangle. The advantage of symmetry in the problem is considered, and only the cross-sectional area of the vessel is considered. Figure 3.8b shows the considered geometry used to solve the problem. Notice that the axis of rotation is the y-axis, and no geometry is allowed in the negative x-axis region. The geometry rotates about the positive y-axis. Additionally, only the upper half of the vessel will be modeled. Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

A type 0.25 in WP X B type 0 in WP Y C type .015 in Width D type .375 in Height OK

Solid mechanics

135

Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 Mesh In Mesh Areas window, click on Pick All

ANSYS graphics show the mesh of the cross-sectional area of the vessel

136

Finite Element Simul ations Using ANSYS

Modeling and meshing are completed at this point. Next, the solution is switched to transient, and unsteady parameters are specified. The boundary conditions are applied starting with the displacements, then the transient pressure. This order is not important for solution. Main Menu > Solution > Analysis Type > New Analysis

A select Transient OK

OK The total time duration for the process is 1000 seconds and to ensure obtaining accurate results, the process duration is divided into 200 time steps. Hence, the time step for this problem is 5 seconds. Results for all time steps are stored by selecting the Write every substep option in Frequency.

Solid mechanics

137

Main Menu > Solution > Analysis Type > Sol’n Control

A B C D

type 1000 in Time at end of loadstep select Time increment type 5 in Time step size select Write every substep

OK The function editor is used to apply the transient pressure equation at the inner surface. This technique is simple and convenient for this problem because an equation for the pressure is given. The upper horizontal line is fixed, while the lower horizontal line is fixed in the y-direction only due to the symmetry. Transient pressure is applied at the left vertical line. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Lines In the ANSYS graphics, click on the top lines that are fixed, and in Apply U,ROT on Lines window, click on OK

A select All DOF B type 0 in Displacement value Apply

138

Finite Element Simul ations Using ANSYS

In the ANSYS graphics, click on the bottom horizontal line, and in Apply U,ROT on Lines window, click on OK

A select UY B type 0 in Displacement value OK Main Menu > Solution > Define Loads > Apply > Functions > Define/Edit

A Type the equation: 100e3+100e3*sin(2*3.1415/100*({time}+75)) B File → Save

Solid mechanics

139

Save the file as PreTran, and this file name is optional. After saving the function, it is required to load it to the ANSYS solution using the read file, and then close the Function Editor window. Main Menu > Solution > Define Loads > Apply > Functions > Read File Select the file PreTran, then open, and the following window will show up.

A type PreTran in Table parameter name and this name is optional, and it should not be the same as the file name of the function. OK Main Menu > Solution > Define Loads > Apply > Structural > Pressure > On Lines In the ANSYS graphics, click on the left vertical line where pressure is applied, and then in Apply PRES on lines window, click on OK

140

Finite Element Simul ations Using ANSYS

A select Existing table OK The following window will show up to select the function.

A select PRETRAN OK The final step in the solution task is to run the solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the problem. Results can be plotted and listed in the General Preprocessor task.

Solid mechanics

141

Main Menu > Solution > Solve > Current LS

OK

Close The above window indicates that the solution task is successfully accomplished. The next step is for getting the results. First, the results at time of 75 seconds is uploaded, and the deformed shape and displacement contours in the x-direction are displayed. Main Menu > General Postproc > Read Results > By Pick

A select set 15 Read Close

142

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

ANSYS graphics show the vessel wall before and after deformation The above figure indicates that the upper line is fixed, while the b ottom line is moved to the outside. The maximum displacement is exactly at the middle of the vessel, as expected. Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

Solid mechanics

143

A click on Nodal Solution → DOF Solution → X-Component of displacement OK

ANSYS graphics show the contours of the displacement in the x-direction The maximum displacement is equal to 0.231 × 10 –5 m. To visualize the results in three-dimensional space, the vessel wall is partially expanded in 180° as follows: Utility Menu > PlotCtrls > Style > Symmetry Expansion > 2D Axi-Symmetric

A select 1/2 expansion OK

144

Finite Element Simul ations Using ANSYS

Utility Menu > PlotCtrls > Pan-Zoom-Rotate Click on the isometric view.

ANSYS graphics show the x-direction displacement contours in three dimensions Animation of the deformation from time = 0–100 seconds, one cycle, can be easily done using animate in the PlotCtrol options. The Number of the frames in the animate over time is the number of pictures in the avi file, while the animation time delay is the display period between two pictures. Fifty animation frames produce a good resolution animation file and with 0.5 second, the animation file duration is 25 seconds (50 frames × 0.5 delay). Main Menu > General Postproc Utility Menu > PlotCtrls > Animate > Over time …

A type 50 in Number of animation frames B select Time Range C type 0 and 100 for Range Minimum, Maximum

145

Solid mechanics

D select DOF solution E select Translation UX OK ANSYS creates an animation file for the deformation. 3.8

Contact element analysis

In this section, the contact pressure distribution between two bodies held in contact is determined. Figure 3.9 shows two parallel cylinders of length L in contact. The upper cylinder has a radius of r1, while the lower cylinder has a radius of r2. Force F is applied to the upper and lower cylinders. At the contact region, a semielliptical contact pressure distribution is developed within each cylinder. The width of the contact line is 2a, and the contact pressure varies from zero at the end of the contact line to the largest value Po at the center of the contact line. The contact width and maximum contact pressure can be expressed, respectively, as 1/2

4Fr1r2 1 − v12 1 − v 22 a= + E 2 πL ( r1 + r2 ) E1

Po =

(3.137)

2 F (3.138) π aL

where E is the modulus of elasticity and v is Poisson’s ratio. If the two cylinders are made of the same material and Poisson’s ratio is equal to 0.3, the expressions (3.137) and (3.138) are simplified as follows: 1/ 2

F r1 r2 a = 1.52 EL r1 + r2

(3.139)

F r1

Po 2a

r2 F

FIGURE 3.9 Two cylinders in contact.

146

Finite Element Simul ations Using ANSYS

1/2

FE r1 + r2 Po = 0.418 (3.140) L r1r2

Expressions (3.139) and (3.140) can be used to determine the contact width and maximum pressure of a contact problem of a circular cylinder and a flat surface, as shown in Figure 3.10. If the radius of the second cylinder is very large, r2 = ∞, the contact width and maximum contact pressure are expressed as 1/2

Fr a = 1.52 1 EL

FE Po = 0.418 (3.142) Lr1

(3.141) 1/2

When two spheres are in contact with different materials and loaded with force F, the contact area will have a circular shape of diameter a. The contact area and maximum stress can be expressed as

(

)

(

)

1/3

3F 1 − v12 /E1 + 1 − v 22 /E 2 a= (3.143) 1/r1 + 1/r2 4

Po = 1.5

F (3.144) πa 2

If the sphere of radius r1 contacting a large flat body of the same material, the expressions of the contact area, and the maximum stress for the contact can be obtained by substituting r2 = ∞ in Equation 3.143, and then the contact diameter and maximum pressure become 1/3

3Fr a = 0.88 1 E

FE 2 Po = 0.62 2 4r

(3.145)

1/3

(3.146)

F

FIGURE 3.10 A cylinder and a flat plate in contact.

r1

147

Solid mechanics

3.9

Two horizontal cylinders in contact analysis using ANSYS

Two horizontal cylinders are placed close to each other, as shown in Figure 3.11. The two cylinders have the same radius, r = 0.01 m, and force of 10 kN is applied to upper and lower cylinders, as shown in the figure. As a result, the cylinders will move and meet. Determine the maximum pressure at the contact region, given that E = 200 GPa and v = 0.3 for both cylinders. Contact problems are highly nonlinear and require large computer resources to solve. Additionally, the physics behind the contact problem is relatively complex, and it is important to understand the physics of the problem to solve it as accurately as possible. Too many parameters are required and just using the ANSYS default settings will give a reasonable result. Contact problems present two main difficulties. First, the regions of contact cannot be accurately predicted. The contact region depends on the loads, materials, boundary conditions, and other factors. Second, most contact problems need to account for friction. There are several friction models to choose from, and all are nonlinear, which makes solution convergence even more difficult. The basic steps for performing a contact analysis are as follows:

1. Create the model geometry and mesh. 2. Identify the contact pairs. The contact pairs are the region in which contact might occur during the deformation of the model. Once potential contact surfaces are identified, they must be defined via target and contact elements. 3. Select contact and target surfaces. Contact elements are constrained against penetrating the target surface, but target elements can penetrate through the contact surface. If a convex surface is

F r

r F

FIGURE 3.11 A contact element problem.

148

Finite Element Simul ations Using ANSYS

expected to meet a flat surface, the convex surface should be the contact surface. If one surface is smaller than the other, the smaller surface should be the contact surface.

4. Apply the boundary conditions. Any type of boundary conditions can be applied in the contact problems.

5. Define solution options and load steps. Convergence behavior for contact problems depends on the geometry and boundaries of the problem. The time step size must be small enough to ensure the convergence. The time step size is specified by a number of steps or the time step size itself. However, the best way to set an accurate time step size is to turn on the automatic time stepping in ANSYS solution options.

6. Solve the contact problem. Ensure that the problem is fully converged. 7. Finally, get the contact results. Results from a contact analysis consist mainly of displacements, stresses, strains, reaction forces, and the contact information, such as contact pressure. Double click on the Mechanical APDL Product Launcher icon This example is limited to structural analysis. Hence, select Structural in Preferences. Solid element is used, and its shape is rectangular with four nodes. The contact simulation requires another two element types: the target and contact elements. These two elements will be introduced into the model when the contact wizard tool is activated. Main Menu > Preferences

A select Structural OK

Solid mechanics

Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add...

A select Solid B select Quad 4 node 182 OK

Close

149

150

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic The following window will show up. For the material properties, the modulus of elasticity and Poisson’s ratio are required to solve the problem.

A type 200e9 in EX B type 0.3 in PRXY OK Close the Material Model Behavior window Next, the geometry of the problem is created, which is two circles. The expected contact surfaces should be defined in the contact wizard. Additionally, the contact region should have a fine mesh for accurate numerical predication. Hence, a small circle of a radius of 0.002 m is created at the contact region and will have a fine mesh.

Solid mechanics

Main Menu > Preprocessor > Modeling > Create > Areas > Circle > Solid Circle

A type 0 in WP X B type 0.01 in WP Y C type 0.01 in Radius Apply

151

152

Finite Element Simul ations Using ANSYS

A type 0 in WP X B type –0.01 in WP Y C type 0.01 in Radius Apply

A type 0 in WP X B type 0 in WP Y C type 0.002 in Radius OK

ANSYS graphics show the created three circles

Solid mechanics

153

First, the three areas are overlapped. Second, since there is s ymmetry in the problem, a vertical line is created to split the entire mode by half. Only the right symmetry side is considered. Finally, extra areas are deleted. Main Menu > Preprocessor > Modeling > Operate > Booleans > Overlap > Areas In Overlap Areas window, click on Pick All ANSYS Main Menu > Preprocessor > Modeling > Create > Lines > Lines > Straight Line Create a vertical line by clicking on two keypoints. The first at the top of the upper circle, and the second at the bottom of the lower circle, and then in create Straight Line window, click on OK

ANSYS graphics show the created three overlapped circles and a vertical line ANSYS Main Menu > Preprocessor > Modeling > Operate > Booleans > Divide > Area by Line To select all three circles, in Divide Area by Line window, click on Pick All Click on the created vertical line. In Divide Area by Line window, click on OK Main Menu > Preprocessor > Modeling > Delete > Area and below Select the left areas and the small area on the right side that is between the two circles. Then, in Delete Area and Below window, click on OK

154

Finite Element Simul ations Using ANSYS

ANSYS graphics show the final geometry Main Menu > Preprocessor > Meshing > Mesh Tool

A Click on Set in Lines

Solid mechanics

155

Select the lines that are far away from the contact area, as illustrated in the following figure.

In Element Sizes on Picked Lines window, click on Apply

A type 100 in No. of element divisions Apply

156

Finite Element Simul ations Using ANSYS

Select the lines that are close to the contact area, as illustrated in the following figure.

In Element Sizes on Picked Lines window, click on OK

A type 25 in No. of element divisions OK

157

Solid mechanics

A Mesh In Mesh Areas, click on Pick All

ANSYS graphics shows the mesh

158

Finite Element Simul ations Using ANSYS

In this problem, the top surface of the lower circle is the target surface because it is stationary, while the bottom surface of the upper circle is the contact surface because it is moving. Both contact and target surfaces are associated with the deformable bodies. These two surfaces together comprise the contact pair. The Contact Manager is very effective in defining, viewing, and editing the contact pairs. In addition, all contact pairs for the entire model can be managed. The Contact Wizard, which is accessed from the Contact Manager, makes the process of creating contact pairs very efficient. Utility Menu > Plot > Lines Main Menu > Preprocessor > Modeling > Create > Contact Pair

A click on the contact wizard The contact wizard is used to select the target and contact surfaces. First, the target surface should be selected and then the contact surface.

A click on Lines B click on Pick Target Click on the top surfaces of the lower circle, as shown in the following figure.

Solid mechanics

159

In Select Lines for Target window, click on OK Then in the Contact Wizard window, click on Next >

A click on Lines B click on Pick Contact Click on the bottom surfaces of the upper circle, as shown in the following figure.

160

Finite Element Simul ations Using ANSYS

In Select Lines for Contact window, click on OK Then in the Contact Wizard, click on Next >

Create >

Solid mechanics

161

Finish

ANSYS graphics show the contact region Close the Contact Manager window Utility Menu > Plot > Lines Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Lines In the ANSYS graphics, click on all vertical left lines, where zero x-displacement is applied, and then in Apply U,ROT on Lines, click on OK

162

Finite Element Simul ations Using ANSYS

A select UX B type 0 in Displacement value OK Forces are applied at the top keypoint of the upper circle and bottom keypoint of the lower circle. Main Menu > Solution > Define Loads > Apply > Structural > Force/Moment > On Keypoints In the ANSYS graphics, click on the bottom keypoint at the lower circle, where the force is applied, and then in Apply F/M on KPs window, click on OK

A select FY in Direction of force/mom B type 10e3 in Force/moment value Apply In the ANSYS graphics, click on the top keypoint at the upper circle, where the force is applied, and then in Apply F/M on KPs window, click on OK

A select FY in Direction of force/mom B type −10e3 in Force/moment value OK

Solid mechanics

163

Main Menu > Solution > Solve > Current LS

OK

Yes

Close The above window indicates that the solution task is accomplished successfully. The next step is getting the results. The scale is changed to true scale for actual dimensional display in ANSYS graphics. Utility Menu > PlotCtrls > Style > Displacement Scaling

A select 1.0 (true scale) OK

164

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click on Nodal Solution > Stress > Y-Component of stress OK

ANSYS graphics show the contours of stress in the y-direction The above figure indicates that the maximum stress in the y-direction is equal to 0.274 × 109 Pa. The maximum stress in the y-direction can be theoretically obtained using expression (3.140) as follows: Po = 0.418

(

)

10 × 10 3 200 × 10 9 0.01 + 0.01 FE r1 + r2 = 0.418 L r1r2 1 0.01 × 0.01

= 0.264 × 10 9 Pa

The theoretical maximum stress is very close to the result obtained by ANSYS with marginal error of 3.78%.

165

Solid mechanics

PROBLEM 3.1 The thin plates, as shown in Figure 3.12a and 3.12b, are subjected to force. Determine the nodal displacements at the point where the force is applied, using the finite element method, given E = 180 GPa, v = 0.33, and t = 0.02 m. Consider the plates as plane stress problems, and use only one triangular element. PROBLEM 3.2 The thin plates, as shown in Figure 3.13a and 3.13b, are subjected to force(s). Determine the nodal displacements, using the finite element method, given E = 240 GPa, v = 0.3, and t = 0.01 m. Use linear triangular elements and the suggested two-element mesh. Consider the plates as plane stress problems. PROBLEM 3.3 The thin plates, as shown in Figure 3.14a and 3.14b, are subjected to force(s). Determine the nodal displacements, using the finite element method, given E = 195 GPa, v = 0.3, and t = 0.015 m. Use linear triangular elements and the suggested finite element mesh. Consider them as plane stress problems. PROBLEM 3.4 The square plate with a hole, as shown in Figure 3.15, is subjected to tensile pressure at both vertical sides. Use ANSYS to determine the maximum stress in the x-direction. Also, compare the ANSYS result with maximum stress using the stress concentration factor figure. The applied pressure is 75 kN/m2, and let E = 230 GPa, v = 0.30, and consider the plate as a plane stress with thickness of t = 0.01 m. Take the advantage of symmetry in the problem to reduce the computational size.

1m

100 kN

2m (a)

FIGURE 3.12 Thin plate subjected to force.

100 kN

1.5 m

1.5 m (b)

166

Finite Element Simul ations Using ANSYS

50 kN 0.25 m 100 kN 0.75 m (a)

0.1 m

15 kN 0.1 m

(b)

FIGURE 3.13 Thin plate subjected to force.

0.25 m

85 kN 0.25 m

0.25 m (a) 75 kN

0.5 m 75 kN 1.0 m

1.0 m (b)

FIGURE 3.14 Thin plate subjected to force.

167

Solid mechanics

ϕ 1.0 cm

P

P

FIGURE 3.15 A plate with a hole subjected to tensile pressure.

5 cm

PROBLEM 3.5 A notched rectangular plate is in tension, as shown in Figure 3.16. Pressure (P) is applied at the left and right vertical sides of the plate, given applied pressure P = 200 kPa, E = 180 GPa, v = 0.3, and t = 0.05 m, where t is the thickness of the plate. Determine maximum displacement in the x-direction and calculate the stress concentration factor (K) using the ANSYS result. Take advantage of symmetry in the problem to reduce the computational size. PROBLEM 3.6 A flat plate is axially loaded, as shown in Figure 3.17. Pressure P = 150 kPa is applied at the left and right vertical sides of the plate. E = 210 GPa, v = 0.33, and t = 0.01 m, where t is the thickness of the plate. Determine using ANSYS the maximum displacement in the x-direction and the stress concentration factor (K). Take the advantage of symmetry in the problem to reduce the computational size.

0.06 m

0.04 m

0.02 m

0.04 m

FIGURE 3.16 A notched rectangular plate subjected to tensile pressure.

168

Finite Element Simul ations Using ANSYS

R 1.0 cm

3.5 cm

3.5 cm

4.0 cm

3.5 cm

FIGURE 3.17 A plate subjected to tensile pressure.

PROBLEM 3.7 Determine the maximum displacement in the radial direction of the vessel shown in Figure 3.18. Take the advantage of symmetry in the problem to reduce the computational size. The inner and outer radii are 0.15 and 0.1625 m, respectively, and the height of the vessel is 0.95 m. Let E = 230 GPa and v = 0.3. Pressure of P = 1.5 MPa is applied at the inner surface of the vessel. The upper and lower horizontal surfaces are fixed. PROBLEM 3.8 The square plate with a hole shown in Figure 3.15 is subjected to transient tensile pressure at both vertical sides. Use ANSYS to determine the maximum stress in the x-direction as a function of time, and let E = 230 GPa, v = 0.30, and consider the plate as a plane stress with

Ro Ri

H

FIGURE 3.18 A vessel subjected to pressure at the inner surface.

169

Solid mechanics

thickness of t = 0.01 m. Take the advantage of symmetry in the problem to reduce the computational size. The total time duration for the process is 500 seconds. The applied pressure (in N/m2) is a function of time according to the following expression:

2π P ( t ) = 75 × 10 3 + 75 × 10 3 sin t 50

PROBLEM 3.9 The square plate geometry with a hole shown in Figure 3.16 is subjected to transient tensile pressure at both vertical sides. Use ANSYS to determine the maximum stress in the x-direction as a function of time, and let E = 180 GPa, v = 0.33 and consider the plate as a plane stress with thickness of t = 0.02 m. Take the advantage of symmetry in the problem to reduce the computational size. The total time duration for the process is 2500 seconds. The applied pressure (in N/m2) is a function of time according to the following expression:

2π P ( t ) = 125 × 10 3 + 125 × 10 3 sin t 500

PROBLEM 3.10 A horizontal cylinder is placed on a flat surface, as shown in Figure 3.19. The cylinder has a radius of r = 0.025 m, and a force of 25 kN is applied to the cylinder. The cylinder will compress and penetrate the flat surface. Determine the maximum stress at the contact region, given that E = 180 GPa and v = 0.3 for the cylinder and flat plate. PROBLEM 3.11 Two horizontal cylinders are placed close to each other, as shown in Figure 3.20. The two cylinders have different radii, r1 = 0.01 m and

F r

FIGURE 3.19 A horizontal cylinder on a flat plate.

170

Finite Element Simul ations Using ANSYS

F r1

r2

FIGURE 3.20 A contact element problem.

r2 = 0.02 m, and force of 50 kN is applied only to the upper cylinder, while the lower cylinder is placed on a flat surface. As a result, the upper cylinder will move downward and meet the lower cylinder. Determine the maximum stress at the contact region. Given that E = 180 GPa and v = 0.33 for both cylinders.

Cha p t e r F OUR

Vibration

4.1

Vibration analysis

In this chapter, the basic formulation of the motion of a single degree of freedom is presented. The aim of this introduction is to discuss some important concepts necessary for solving and understating vibration problems. Figure 4.1 shows a mass–spring system that is subjected to a time-dependent force F(t), where m is the mass of the system and k is the stiffness of the spring. Applying Newton’s second law, the equation of motion of the spring– mass element in the x-direction is expressed as

+ kx = F ( t ) (4.1) mx Equation 4.1 can also be symbolically expressed as

[ m ]{x} + [ k ]{x} = {F } (4.2)

where [k] is the stiffness matrix, [m] is mass matrix, {x} is the nodal x} is the acceleration vector. If the applied displacement vector, and { force F(t) is equal to zero, then Equation 4.1 becomes a homogenous ordinary differential equation. The solution of this equation gives the natural frequency of the oscillating mass: k , (4.3) m where ω is called the natural frequency of the free vibration of the mass. The natural frequency depends on the stiffness of the spring and the mass of the system. Substituting (4.3) into the equation of motion of free vibration (4.1) yields

ω2 =

x + ω 2 x = 0 (4.4) The solution of Equation 4.4 is

x(t) = A cos(ωt) + B sin(ωt) (4.5) where the constants A and B can be determined using the initial conditions. 171

172

Finite Element Simul ations Using ANSYS

k m

F(t)

FIGURE 4.1 Spring–mass system subjected to a time-dependent force. 4.2

evelopment of stiffness equations for D axial vibration of a bar element

Lumped mass formulation is one of the common ways to formulate the stiffness matrix for dynamics analysis. Figure 4.2 shows a bar with initial length L, modulus of elasticity E, cross-sectional area A, and density ρ. To account for the mass inertia of the bar, the mass is distributed and concentrated in the nodes, as shown in Figure 4.2. The inertia forces are added to the nodal forces for the bar element equation, which is used in static analyses, as follows:

ρAL u1 F1 − 2 F2 − ρAL u2 2

EA 1 −1 d1 = L −1 1 d 2

(4.6)

The negative values of the inertial forces indicate that they are resisting the elastic forces F1 and F2. Equation 4.6 is rearranged as follows: F1 F2

EA 1 = L −1

−1 d1 1 d 2

ρAL 1 + 2 0

d 0 1 1 d2

(4.7)

or symbolically,

{F } = [ k ]{d} + [ m ]{d} (4.8)

For free vibration, the nodal force vector is equal to zero. For a harmonic motion, it is assumed that the displacements vary harmonically with respect to time with frequency ω: {d} = {U} sin(ωt) (4.9) The second derivative of the displacement vector is the acceleration:

{d} = −ω

2

{U} sin ( ωt ) (4.10)

Expression (4.9) into (4.10) yields

{d} = −ω {d} (4.11) 2

173

Vibr ation

Node 1

F1

A, ρ

Node 2

F2

L m1 =

ρAL

m2 =

2

ρAL

FIGURE 4.2 Bar element and the concentrated mass at the nodes.

2

Substituting the acceleration vector in Equation 4.7 yields F1 EA 1 −1 d1 = F2 L −1 1 d 2

2 ρAL 1 −ω 2 0

0 d1 1 d 2

(4.12)

Equation 4.12 can be rearranged as F1 EA 1 −1 ρAL 1 0 d1 − ω2 (4.13) = 2 0 1 d 2 F2 L −1 1

The global stiffness matrix can be obtained by assembling the stiffness matrix for each element as follows:

[K ] =

N

∑ K

(e)

e =1

(4.14)

The nodal forces are also assembled to form a global force vector:

[F ] =

N

∑ F

(e)

e =1

(4.15)

Similarly for the displacement vector, N

[ d ] = ∑ d(e) (4.16) e =1

Finally, the nodal forces and the displacement relationship is expressed as {F} = [K]{d}

4.3

(4.17)

Natural frequencies of axial vibration of a bar element

Determine the first two natural frequencies of a cantilever bar, as shown in Figure 4.3. The total length of the bar is 2 m. Model the bar with two elements only, and let E = 200 GPa, A = 2 × 10 –3 m2, and ρ = 5000 kg/m3.

174

Finite Element Simul ations Using ANSYS

E, A, ρ

2m

FIGURE 4.3 Cantilever bar and Node 1 finite element mesh.

Element 1

Element 2 Node 2

Node 3

Element 1 has nodes 1 and 2, and its length is L = 1.0 m. Using Equation 4.13 for nodal forces and displacement relationship,

F1 EA 1 −1 ρAL 1 0 d1 − ω2 = 2 0 1 d 2 F2 L −1 1

Nodal displacement at node 1 and nodal force at node 2 are equal to zero. Element 2 has nodes 2 and 3, and its length is L = 1 m. Using Equation 4.13,

ρAL 1 0 d 2 F2 EA 1 −1 − ω2 = 2 0 1 d 3 F3 L −1 1

Nodal forces at node 3 are equal to zero. Assembling the nodal and displacement vectors and stiffness matrix for both elements yields F1 F2 = 0 F3 = 0

1 0 0 d1 = 0 1 −1 0 EA 2 ρAL 0 2 0 d 2 −1 2 −1 − ω = 2 L 0 −1 1 0 0 1 d 3

Since the first column is multiplied by zero displacement, it can be deleted. The first row can also be deleted, and then

0 EA 2 −1 ρAL 2 0 d 2 − ω2 = 2 0 1 d 3 0 L −1 1 The above equations are simplified as

0 2 = 0 −1

2 −1 − λ 1 0

0 1

175

Vibr ation

where λ is the eigenvalue, λ = ω2

ρL2 2E

then, 0 2 − 2λ −1 = 1− λ 0 −1

The determinant of the above matrix yields the following equation: 2λ2 − 4λ + 1 = 0 And the solution for the above equation is λ1 = 0.29289 and λ2 = 1.70711. The natural frequencies are calculated as ω1 =

2λ1E = 4840.58 Hz ρL2

ω2 =

2λ 2 E = 11686.23 Hz ρL2

The exact solution for the present problem is ω=

nπ 2L

E ρ

where n is the mode number. The exact solutions for the first and the second frequencies are ω1 = 4967.15 Hz and ω2 = 9934.3 Hz, respectively. The exact results are close to the finite element solution with error in the first and the second frequencies of 2.55% and 17.64%, respectively. The error can be reduced if the number of elements is increased.

4.4

evelopment of stiffness equations for D flexural vibration of a beam element

Figure 4.4 shows a horizontal beam element. The variation of curvature of the beam is assumed to be a polynomial of the third order, and the constants in the polynomial are determined using the boundary conditions. There are four degrees of freedom for the element, two vertical displacements and two rotations. The displacement function is expressed as dy(x) = a1x3 + a2x2 + a3x + a4 (4.18) Using expression (2.68), the rotation is expressed as θ(x) = 3a1x2 + 2a2x + a3 (4.19)

176

Finite Element Simul ations Using ANSYS

y E, I, A, ρ

FIGURE 4.4 Horizontal beam element.

1

θ1

L

d1y

x 2

θ2 d2y

The constant a’s are obtained by using the boundary conditions at the nodes as follows: dy(x = 0) = d1y (4.20) dy(x = L) = d2y (4.21)

θ(x = 0) =

∂v = θ1 (4.22) ∂x

θ(x = L) =

∂v = θ2 (4.23) ∂x

Therefore, the displacement function (4.18) is expressed as d y ( x ) = d1y + θ1x −

2x 2 3x 2 x2 3x 2 d − θ + d − θ2 2y 1y 1 L2 L L2 L

2x 2 2x 2 x3 x3 + 3 d1y + 2 θ1 − 3 d 2y + 2 θ2 L L L L

(4.24)

Rearranging the displacement function (4.24) gives d(x) = f1(x)d1 + f2(x)θ2 + f3(x)d2 + f4(x)θ2 (4.25) where f1(x), f2(x), f3(x), and f4(x) are shape functions, and they are expressed as 2

3

x x f1 ( x ) = 1 − 3 + 2 (4.26) L L

x2 x3 f2 ( x ) = 1 − 3 + 2 2 (4.27) L L

x x f3 ( x ) = 3 − 2 (4.28) L L

x 2 x3 f4 ( x ) = − + 2 (4.29) L L

2

3

177

Vibr ation

The kinetic and strain energies are, respectively, L

T=

2

ρA ∂v dx (4.30) 2 ∂x

∫ 0

L

2

EI ∂2 v U= dx (4.31) 2 ∂x 2

∫

0

Using the Lagrange equation, d ∂T ∂T = F (4.32) + dt ∂θ ∂θ

Substituting kinetic and strain energies into the Lagrange equation gives

[ m ]{d} + [ k ]{d} = {F ( t )} (4.33)

{}

In the case of free vibration, and with the expression d = −ω 2 {d} , the Lagrange equation (4.33) becomes −ω2[m]{d} + [k]{d} = {0}

(4.34)

Expanding Equation 4.34 yields

156 22L 54 −13L d1y ρAL 22L 4L2 13L −3L2 θ1 −ω 2 420 54 13L 156 −22L d 2y −13L −3L2 −22L 4L2 θ2 12 / L 6 / L −12 / L EI 6 / L 4 −6 / L + L −12 / L2 −6 / L −12 / L2 6/L 2 −6 / L 2

4.5

2

6 / L 2 −6 / L 4

d1y θ1 d 2y θ2

=

(4.35) 0 0 0 0

atural frequencies of the flexural N vibration of a beam element

Determine the first natural frequency of the beam shown in Figure 4.5a. The beam is 2 m in length and fixed at both ends. The beam has a square cross section with height and width of 0.05 m, ρ = 5000 kg/m3, and E = 200 GPa. The minimum number of nodes required to determine the natural frequency is 3, because two of the nodes are fixed. In this example,

178

Finite Element Simul ations Using ANSYS

E, A, I, ρ Element 1 2m

Node 1

Element 2 Node 3

Node 2

(a)

(b)

FIGURE 4.5 (a) Horizontal beam with fixed ends and (b) finite element mesh.

the beam is divided into two elements only. The finite element mesh for the horizontal beam is shown in Figure 4.5b. In the next ANSYS example, the beam is meshed with 50 elements. Element 1 has a length of L = 1 m, nodes 1 and 2, and its stiffness and mass matrices are, respectively,

ρAL 156 420 −22L

EI 12 / L2 K (1) = L −6 / L

[ m1 ] =

−22L 4L2 −6 / L 4

Element 2 has a length of L = 1 m, nodes 2 and 3, and its stiffness and mass matrices are, respectively,

[m2 ] =

22L 4L2

ρAL 156 420 22L

EI 12 / L2 K ( 2) = L 6 / L

6/L 4

Applying Equation 4.34:

−ω 2

ρAL 312 420 0

0 d 2y 8L2 θ2

EI 24 / L2 + 0 L

0 d 2y 8 θ2

0 = 0

Simplifying the above equation yields

−ω 2

ρAL 312 420 0

0 EI 24 / L2 + 8L2 L 0

0 = 0 8 0

179

Vibr ation

Solving for the first natural frequency, ω1 =

5.68 EI ( L )2 ρA

1/ 2

The exact solution for the first natural frequency is given by ω1 =

5.59 EI ( L )2 ρA

1/ 2

which is very close to the finite element result.

4.6

atural frequencies of the flexural vibration N of beam element using ANSYS

For a horizontal beam shown in Figure 4.5a, determine the first five natural vibration frequencies. Also, create animation for the third mode. Mesh the beam with 50 elements. The beam has a square cross section with height and width of 0.05 m, ρ = 5000 kg/m 3, and E = 200 GPa. Double click on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select Structural OK

180

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add...

A select Beam B select 2 node 188 OK

Options…

Vibr ation

181

The illustrated theory for the beam in this chapter is based on the third-order polynomial for displacement function. Therefore, in the option, the element behavior should be changed to cubic form.

A select Cubic Form. in Element behavior K3 OK Close the Element Type window Close The beam cross section is specified in sections. For material properties, the modulus of elasticity and density are required. Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic The following window will appear, the Young modulus for the beam is specified, and any value for the Poisson ratio is required to avoid an error message. Also, the density of the beam is required.

182

Finite Element Simul ations Using ANSYS

A type 200e9 in EX B type 0.3 in PRXY OK

A click on Structural > Density

A type 5000 in DENS OK Close the Define Material Model Behavior window The cross section of the beam is a square. In the cross section, the square section type is the default section. The width and height of the beam’s cross section should be specified.

183

Vibr ation

Main Menu > Preprocessor > Sections > Beam > Common Sections

A type 0.05 in B B type 0.05 in H Apply Review OK

ANSYS graphics will show the geometrical properties of the beam cross section Modeling the beam structure takes place in the following steps. First, two keypoints are created. Then, a line that connects the keypoints is created. The x- and y-coordinates of each keypoint are specified for ANSYS. Then, (0,0) is the coordinate for Keypoint 1, and (0,2) is the coordinate for Keypoint 2.

184

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Modeling > Create > Keypoints > In Active CS

A type 1 in Keypoint number B type 0 and 0 in X,Y,Z Location in active CS Apply

A type 2 in Keypoint number B type 2 and 0 in X,Y,Z Location in active CS OK Main Menu > Modeling > Create > Lines > Lines > Straight Line Click on Keypoints 1 and 2. In Create Straight Line window, click on OK Main Menu > Preprocessor > Meshing > Mesh Tool

Vibr ation

185

A click Set in Lines In Element Sizes on Picked Lines window, click on Pick All The following window will show up. In Element Sizes on Picked Lines window, there are two options, either specifying the length of the elements or the number of element divisions. For the present example, the beam is divided into 50 elements.

A type 50 in No. of element divisions OK

A click on Mesh In Mesh Lines window, click on Pick All

186

Finite Element Simul ations Using ANSYS

Utility Menu > Plot > Nodes

ANSYS graphics show the nodes The analysis type is changed from Static to Modal in New Analysis. The required number of free vibration modes is five, and this is requested in analysis options. Main Menu > Solution > Analysis Type > New Analysis

A select Modal OK Main Menu > Solution > Analysis Type > Analysis Options

Vibr ation

187

A type 5 in No. of modes to extract B type 5 in NIMODE No. of modes to expand OK

OK Boundary conditions are specified in solution. Both ends of the beam are fixed and no forces are applied because it is a free vibration problem. The displacement is applied on the keypoints. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Keypoints Click on Keypoints 1 and 2. Then, in Apply U,ROT on KPs window, click on OK

A select UX, UY, UZ, ROTX, ROTY, and ROTZ B type 0 in Displacement value OK The final step is to run the ANSYS solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the problem.

188

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Solve > Current LS

OK

Close In the following steps, the natural frequencies of the beam are listed. Notice that the frequencies 1 and 2 are the same, because the beam is vibrating similarly in the y- and the z-directions. Main Menu > General Postproc > Results Summary

Close The exact solution for the first natural frequency for this geometry is given by ω1 =

5.59 EI ( L / 2 )2 ρA

1/ 2

where L is the beam length. The exact solution for the present geometry is ω1 = 510.377 rad/s or 81.23 Hz, which is very close to the ANSYS results. Next, the deformation shape of the third mode is displayed.

Vibr ation

189

Main Menu > General Postproc > Read Results > By Pick

A select Set 3 Read Close Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

ANSYS graphics show the beam before and after deformation The animation of the third vibration mode is displayed in the following steps. An AVI file is created and stored in the working directory. The number of frames is the number of screen shoots for the animation file. The 50-frame file is specified for this example and a higher value

190

Finite Element Simul ations Using ANSYS

will enhance the resolution of the animation, but the size of the file will be increased. The time delay is the time between each screen shot, and increasing the time delay will increase the duration of the animation. Utility Menu > PlotCtrls > Animate > Mode Shape

A type 50 in No. of frames to create B select Def + undeformed OK Animation will be shown on the screen for the third mode. 4.7

evelopment of stiffness equations for D vibration of an oriented beam element

Figure 4.6 shows an axial flexural frame element oriented at an angle θ in the xy-plane. There are three degrees of freedom at each node, displacement in x-direction, displacement in y-direction, and rotation. For a static beam, the stiffness equation is expressed as {F} = [T]T[K][T]{d} (4.36) The stiffness equation is extended for vibration by adding the ω 2 [ m ] matrix to the stiffness matrix as follows:

{F } = [ T ]T ([ K ] − ω 2 [ m ])[ T ]{d} (4.37)

where

{F } =

F1x F1y M1 (4.38) F2x F2y M 2

191

Vibr ation

Fy2, dy2

No

L

Fx1, dx1

No

M1, θ1

de

1

, ,E

I, A

,ρ

M2, θ2

e2

d

Fx2, dx2

y

θ

x

Fy1, dy1

FIGURE 4.6 An axial flexural frame element.

cosθ sinθ −sinθ cosθ 0 0 T = [ ] 0 0 0 0 0 0

EI K = [ ] L

0 0 0 12 / L2 0 6/L 0 0 0 −12 / L2 0 6/L

0 0 0 0 0 0 1 0 0 0 cosθ sinθ 0 −sinθ cosθ 0 0 0 0 6/L 4 0 −6 / L 2

0 0 0 0 0 1

(4.39)

0 0 0 −12 / L2 0 −6 / L 0 0 0 12 / L2 0 −6 / L

0 6/L 2 0 −6 / L 4

(4.40)

and

ρAL [m] = 420

0 0 0 0 156 22L 0 22L 4L2 0 0 0 0 54 13L 0 −13L −3L2

0 0 0 −13L 0 54 0 13L −3L2 0 0 0 0 156 −22L 0 −22L 4L2

(4.41)

The equation of motion for the two-dimensional flexural beam element is

{F } = ([ K ] − ω 2 [ m ]) {d} (4.42)

192

Finite Element Simul ations Using ANSYS

4.8

Harmonic vibration of a plate with holes using ANSYS

The rectangular plate with six holes, as shown in Figure 4.7a, is subjected to forcing function. The forcing function is shown in Figure 4.7b. Create a graph showing the relationship between the displacement at the point where the force is applied and the load’s frequency. Let t = 0.001 m, E = 210 GPa, v = 0.25, and ρ = 5000 kg/m3. The frequency of the load is varied between 1 and 2000 Hz, and the number of subsets is 100. Double click on the Mechanical APDL Product Launcher icon This example is limited to vibration analysis. Hence, select Structural. The geometry is meshed with quadratic four-node elements. Main Menu > Preferences

A select Structural OK F(t) R 0.025 m

0.1 m

0.075 m

0.075 m

0.1 m 0.65 m F, kN

(a)

0.1

(b)

1.0

t, sec

FIGURE 4.7 (a) A plate with six holes and (b) forcing function.

Vibr ation

193

Material properties of the structure are defined in the following steps. The modulus of elasticity and density are required to solve this problem. In addition, the plate has a thickness, and the thickness is specified in real constants. Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add...

A select Solid B select Quad 4 node 182 OK

Options…

194

Finite Element Simul ations Using ANSYS

A select Plane strs w/thk in Element behavior K3 OK Close the Element Type window Close Main Menu > Preprocessor > Real Constants > Add/Edit/Delete

Add...

OK

Vibr ation

A type 0.001 in Thickness THK OK Close the Real Constants window Close Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic

A type 210e9 in EX B type 0.25 in PRXY OK

195

196

Finite Element Simul ations Using ANSYS

Main Menu > Material Props > Material Models

A click on Structural > Density

A type 5000 in DENS OK Close the Define Material Model Behavior window The geometry is modeled by creating a rectangle and circles. Boolean operation is utilized to subtract the circles from the rectangle. Alternatively, the circles can be overlapped and then deleted. Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

Vibr ation

A B C D

197

type 0 in WP X type 0 in WP Y type .65 in Width type .1 in Height

OK Main Menu > Preprocessor > Modeling > Create > Areas > Circle > Solid Circle

A type .075 in WP X B type .05 in WP Y C type .025 in Radius OK The other circles are created using the copy area in the modeling. The number of circles including the original circle is six, and the distance between the circles is 0.1 m. Main Menu > Preprocessor > Modeling > Copy > Areas In the ANSYS graphics, select the solid circular area, and then in Copy Areas window, click on OK

198

Finite Element Simul ations Using ANSYS

A type 6 in Number of copies B type .1 DX X-offset in active CS OK Main Menu > Preprocessor > Modeling > Operate > Booleans > Subtract > Areas Click on the center of the rectangular area to select it. Then, in Subtract Areas window, click on OK Click on all circular areas to select them. Then, in Subtract Areas window, click on OK

ANSYS graphics showing the final geometry

Vibr ation

199

The geometry is meshed with quadratic four-node elements. A free mesh is generated using the smart mesh option. The mesh refinement is 1. Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 Then, in Mesh Tool, click on Mesh In Mesh Areas, click on Pick All

200

Finite Element Simul ations Using ANSYS

The analysis type is changed from Static to Harmonic. The frequency of the load is varied between 1 and 2000 Hz, and the number of subsets is 100. Main Menu > Solution > Analysis Type > New Analysis

A select Harmonic OK Boundary conditions are applied as follows. The left vertical line of the beam is fixed, while the other lines are free. Force in the negative y-direction is applied at the keypoint at the upper right corner. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Lines In the ANSYS graphics, click on the left line where displacement is applied. Then, in Apply U,ROT on Lines window, click on OK

A select All DOF B set VALUE to 0 OK

Vibr ation

201

The Function Editor is used to apply the transient force formula as a nodal force. This technique is simple and convenient for this problem because an equation for the force is given. Notice that the force unit given in Figure 4.7b is in kN. Main Menu > Solution > Define Loads > Apply > Functions > Define/Edit

A type the equation: −100−100*{TIME} In the Equation Editors click on File then Save Save the file as FY_Time, and the file name is optional. After saving the function, it has to be loaded to the ANSYS solution using the read file. Save Close the Function Editor Main Menu > Solution > Define Loads > Apply > Functions > Read File Select the file FY_Time.func then Open

202

Finite Element Simul ations Using ANSYS

A type FY_Time in Table parameter name, and this name is optional OK Main Menu > Solution > Define Loads > Apply > Structural > Force/Moment > On Keypoint Click on keypoint where the force is applied, and then in Apply F/M on KPs window, click on OK

A select FY in Direction of force/mom B select Existing table OK

Vibr ation

203

The following window will show up to select the function.

A select FY_TIME OK Main Menu > Solution > Load Step Opts > Time/Frequency > Freq and Substeps

A type 0 and 2000 in Harmonic freq range B type 100 in Number of substeps C select Stepped OK The final step is to run the ANSYS solution. ANSYS will assemble the stiffness matrices, apply the boundary conditions, and solve the problem.

204

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Solve > Current LS

OK

Close A graph that shows the displacement at a specific location in the plate as a function of frequency can be done with the Time history. In the following steps, the displacement at the node where the force is applied is plotted as a function of frequency. Main Menu > TimeHist Postpro

A click on the green + button

Vibr ation

205

A click on Nodal Solution > DOF Solution > Y-Component of displacement OK In the ANSYS graphics, click on the node at the upper right corner of the plate and then in Node for Data window, click on OK

A click on the graph button

206

Finite Element Simul ations Using ANSYS

ANSYS graphics show the y-displacement as a function of frequency The jumps in the displacement shown in the above figure occurred at the natural frequency of the first and the second modes of the plate. Next, an animation is created for the frequency of 1000 Hz. Main Menu > General Postproc > Read Results > By Time/Freq

A type 1000 in TIME Value of time or freq OK Utility menu > Plot > Areas Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

Vibr ation

207

ANSYS graphics show the plate before and after deformation Utility Menu > PlotCtrls > Animate > Mode Shape

A type 50 in No. of frames to create B select Def + undeformed OK Animation will be shown in ANSYS graphics. 4.9

Three-dimensional vibration of shaft with disks using ANSYS

For the geometry shown in Figure 4.8, the ends of the shaft are fixed, while the two disks and shaft are vibrating freely. The total shaft length is 0.655 m. Determine the first five natural vibration frequencies.

208

Finite Element Simul ations Using ANSYS

R 0.3 R 0.03 Fixed

Fixed

0.125 m

0.25 m 0.015 m

FIGURE 4.8 Shaft and two disks.

Also, create animation for the third mode. Let m, E = 180 GPa, v = 0.3, and ρ = 7000 kg/m3. Use free mesh with Tet 4 node element and set smart size to 5. Double click on the Mechanical APDL Product Launcher icon This example is limited to structural analysis. Hence, select Structural only. The three-dimensional solid element must be used and the type of the element is tetrahedral with four nodes.

Vibr ation

Main Menu > Preferences

A select Structural OK Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add...

A select Solid B select Tet 4 node 285 OK

209

210

Finite Element Simul ations Using ANSYS

Close For the material properties, the modulus of elasticity and density are required to solve the problem. Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic The following window will show up:

Vibr ation

211

A type 180e9 in EX B type 0.3 in PRXY OK Main Menu > Material Props > Material Models

A click on Structural > Density

A type 7000 in DENS OK Close the Material Model Behavior window The used technique for modeling the geometry is quite simple. The following steps are performed: 1. The shaft is modeled as a long, solid cylinder. 2. The first disk is modeled as a thin, solid cylinder and then it is moved to its position in the shaft. 3. The second disk is created by copying the first disk. 4. The shaft and two disks are added to create one volume.

212

Finite Element Simul ations Using ANSYS

ANSYS Main Menu > Preprocessor > Modeling > Create > Volumes > Cylinder > Solid Cylinder

A B C D

type 0 in WP X type 0 in WP Y type 0.03 in Radius type 0.655 in Depth

OK ANSYS Utility Menu > PlotCtrls > Pan-Zoom-Rotate …

Vibr ation

213

Click Iso button to show the isometric view of the shaft and zoom in and out.

ANSYS graphics show the isometric view of the shaft ANSYS Main Menu > Preprocessor > Modeling > Create > Volumes > Cylinder > Solid Cylinder

A B C D

type 0 in WP X type 0 in WP Y type 0.3 in Radius type 0.015 in Depth

OK The disk is not in its correct position, and it has to be moved along the z-axis to its position at a distance of 0.25 m from its current position.

214

Finite Element Simul ations Using ANSYS

ANSYS Main Menu > Preprocessor > Modeling > Move/Modify > Volume Click on the disk and then in Move Volumes window, click on OK

A type 0 in DX X-offset in active CS B type 0 in DY Y-offset in active CS C type 0.25 in DZ Z-offset in active CS OK The second disk is created by copying the first disk. The offset distance between the two disks is the spacing plus the thickness of the disk, which is 0.14 m. ANSYS Main Menu > Preprocessor > Modeling > Copy > Volumes Click on the disk, and then in Copy Volumes window, click on OK

A type 0 in DX X-offset in active CS B type 0 in DY Y-offset in active CS C type 0.14 in DZ Z-offset in active CS OK

Vibr ation

Next, the three volumes are added to create one volume. ANSYS Main Menu > Preprocessor > Modeling > Operate > Booleans > Add > Volumes In Add Volumes windows, click on Pick All Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 5 Mesh In Meshed Areas window, click on Pick All

215

216

Finite Element Simul ations Using ANSYS

ANSYS graphics show the mesh The analysis type is changed from Static to Modal. The required number of free modes to be calculated is five. Main Menu > Solution > Analysis Type > New Analysis

A select Modal OK Main Menu > Solution > Analysis Type > Analysis Options

A type 5 in No. of modes to extract B type 5 in NIMODE No. of modes to expand OK

Vibr ation

217

OK The left and right surfaces of the shaft are fixed, while the other surfaces are free. No forces are applied. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Areas In the ANSYS graphics, click on the left and right surfaces of the shaft, which are fixed. Then in Apply U,ROT on Areas window, click on OK

A select All DOF B type 0 in Displacement value OK Main Menu > Solution > Solve > Current LS

Close

218

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Results Summary

ANSYS lists the natural frequencies of the shaft with two disks. Next, the deformed shape of the shaft for the mode number 4 is displaced, followed by an animation. Main Menu > General Postproc > Read Results > By Pick

A select Set 4 Read Close Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

219

Vibr ation

ANSYS graphics show the vibrating shaft with two disks before and after deformation Utility Menu > PlotCtrls > Animate > Mode Shape

A type 50 in No. of frames to create B select Def + undeformed OK Animation will be shown for the third mode. PROBLEM 4.1 Determine the first natural vibration frequency of a horizontal cantilever bar using finite element method, as shown in Figure 4.9. The total bar length is 3 m. Model the bar with three elements only and let E = 180 GPa, A = 4 × 10 –3 m2, and ρ = 7000 kg/m3.

E, A, ρ 3m

FIGURE 4.9 A simple cantilever bar.

220

Finite Element Simul ations Using ANSYS

PROBLEM 4.2 A horizontal cantilever bar is composed of two different bars of different cross sections and lengths, as shown in Figure 4.10. Determine the first natural vibration frequency of the bar using the finite element method. Model the bar with three elements only, as suggested in the figure, and let E = 210 GPa, ρ = 4500 kg/m 3, A1 = 4 × 10 –3 m 2, and A 2 = 2 × 10 –3 m 2. PROBLEM 4.3 A horizontal beam structure is composed of three different beams of different material properties, as shown in Figure 4.11. Determine the first natural vibration frequency of the structure using the finite element method. Model the structure with three elements only and let E1 = 205 GPa, ρ1 = 4250 kg/m3, E2 = 270 GPa, ρ2 = 3000 kg/m3, E3 = 180 GPa, ρ3 = 4000 kg/m3, and the beams have a square crosssectional area with height and width of 0.1 m. PROBLEM 4.4 A horizontal cantilever bar structure is composed of two different bars of different cross-sectional areas and lengths, as shown in Figure 4.10. Determine the first five natural vibration frequencies of bar structure and let E = 210 GPa, ρ = 4500 kg/m3; A1 = 4 × 10 –3 m2, and A2 = 2 × 10 –3 m2. PROBLEM 4.5 A horizontal beam structure is composed of three different beams of different material properties, as shown in Figure 4.11. Determine the first five natural vibration frequencies of beam structure using ANSYS. Also, create an animation for the second mode. Mesh each

E, A1, ρ

E, A2, ρ

1m

2m

Element 1 Node 1

Element 2

Node 2

Element 3

Node 3

Node 4

FIGURE 4.10 A cantilever bar made of two different bars.

E1, ρ1 0.5 m

E2, ρ2 0.5 m

E3, ρ3 0.5 m

FIGURE 4.11 A horizontal beam structure made of three different beams.

221

Vibr ation

beam with 50 elements and let E1 = 205 GPa, ρ1 = 4250 kg/m 3, E2 = 270 GPa, ρ2 = 3000 kg/m3, E3 = 180 GPa, ρ3 = 4000 kg/m3, and the beam has a square cross-sectional area with height and width of 0.01 m. PROBLEM 4.6 For the plate with circular holes, as shown in Figure 4.12, the left vertical line is fixed. Determine the first five natural vibration frequencies of structure using ANSYS. Also, create an animation for the third mode. Let t = 0.002 m, E = 200 GPa, and ρ = 7500 kg/m3. PROBLEM 4.7 For the rectangular cross-sectional plate with square holes, as shown in Figure 4.13, both ends are fixed. Determine the first five natural vibration frequencies of the structure using ANSYS. Also, create an animation for the third mode. Let t = 0.001 m, E = 270 GPa, and ρ = 6000 kg/m 3. PROBLEM 4.8 The geometry shown in Figure 4.14a is subjected to the transient force. The relationship between force and time is shown in Figure 4.14b. Create a graph showing the relationship between the displacement R 0.025 m

0.1 m

0.05 m

0.1 m

0.05 m

FIGURE 4.12 A plate with circular holes.

0.04 m

0.01 m

0.02 m

0.02 m

FIGURE 4.13 A rectangular plate with square holes.

222

Finite Element Simul ations Using ANSYS

F(t)

0.08 m

0.02 m

0.02 m

0.44 m (a) F, N 250

(b)

1.0

t, sec.

FIGURE 4.14 (a) A rectangular plate with a rectangular hole and (b) forcing function.

at the point where the force is applied and the load’s frequency. Let t = 0.002 m, E = 230 GPa, and ρ = 4000 kg/m3. The frequency of the load is varied between 1 and 5000 Hz, and the number of subset is 50. PROBLEM 4.9 For the geometry shown in Figure 4.15, the ends of the shaft are fixed, while the disk and the shaft are vibrating freely. Determine the first five natural vibration frequencies for the shaft. Also, create an animation for the third mode. Let m, E = 180 GPa and ρ = 7000 kg/m3. Use free mesh with Tet 4 node element, and set smart size to 5. PROBLEM 4.10 For the airplane wing shown in Figure 4.16, determine the first five natural vibration frequencies. Let E = 180 GPa, v = 0.25, and ρ = 3500 kg/m3. The wingspan is 5 m, and the left surface of the wing is fixed. Solve the problem as a three-dimensional problem. Use Tet 10-node elements with smart size set to 1.

223

Vibr ation

R 0.2 R 0.05

Fixed

0.2 m

0.15 m

FIGURE 4.15 A shaft and disk.

R 0.4 m

2.5 m

FIGURE 4.16 Airplane wing.

0.025 m

R 0.025

Fixed

This page intentionally left blank

Cha p t e r FIV E

Heat transfer

5.1

Introduction to heat conduction

Whenever a temperature gradient exists in a solid, heat will flow from the high-temperature region to the low-temperature region. The basic governing heat conduction equation is obtained by considering a plate with a surface area A and thickness Δx, as shown in Figure 5.1. One side is maintained at temperature T1 and the other side is at temperature T2. Experimental observation indicates that the rate of heat flow is directly proportional to the area and temperature difference, but inversely proportional to the plate thickness. The proportionality sign is replaced by an equal sign by introducing the constant k as follows:

Q = kA

T1 − T2 (5.1) ∆x

The constant k is the thermal conductivity of the plate. This property depends on the type of the plate’s material. Equation 5.1 is also called Fourier’s law. Fourier’s law can be expressed in differential form in the direction of the normal coordinate:

Q = − kA

dT (5.2) dn

Also, Fourier’s law can be expressed for multidimensional heat flux flow as follows:

∂T ˆ ∂T ˆ ∂T ˆ q = −k i+ j+ k (5.3) ∂y ∂z ∂x

An energy balance can be applied to a differential volume dx dy dz, for conduction analysis in a Cartesian coordinate, as shown in Figure 5.2. The objective of this energy balance is to obtain the temperature distribution within the solid. The temperature distribution is used to determine the heat flow at a certain surface, or to study the thermal stress. The heat flux perpendicular to the surface of the control volume is indicated 225

226

Finite Element Simul ations Using ANSYS

Δx

T1 T2

FIGURE 5.1 Heat through a plate.

transfer

qy

qx

FIGURE 5.2 Differential control volume for the energy balance.

qz+dz qz

qx+dx

qy+dy

by the terms, qx, qy, and qz. The heat flux at the opposite surface can then be expressed using the Taylor series expansion of the first order as follows:

q x + dx = q x +

∂q x dx (5.4) ∂x

q y + dy = q y +

∂q y dy (5.5) ∂y

q z + dz = q z +

∂q z dz (5.6) ∂z

Energy can be generated in the medium, and the expression of the heat generation is

E gen = q dx dy dz (5.7)

where q is the generated heat per unit volume, W/m3. If the heating process is unsteady, the total energy of the control volume can be increased or decreased. The energy storage term is expressed as

∂T dx dy dz (5.8) E st = ρCP ∂t

227

He at tr ansfer

The sum of the energy generation in the control volume and net heat flow should be equal to the energy stored in the control volume. The energy conservation can be expressed in the following mathematical form:

(

)

E gen + E in − E out = E st (5.9)

Substituting expressions (5.4–5.8) into (5.9), the energy conservation equation becomes ∂q y ∂q ∂q ∂T q dx dy dz + x dy dz + dx dz + z dx dy = ρCP dx dy dz ∂y ∂z ∂t ∂x (5.10) The Q ′′, y and Q z′′ are obtained from Fourier’s law (5.2) as follows: x Q ′′,

Q ′′x = − k

dT dx (5.11) dx

Q ′′y = − k

dT dy (5.12) dy

Q z′′ = − k

dT dz (5.13) dx

Finally, the conduction energy equation per unit volume, in a Cartesian coordinate, is expressed as

∂ ∂T ∂ ∂T ∂ ∂T ∂T (5.14) k + k k + + q = ρCp ∂x ∂x ∂ y ∂ y ∂z ∂z ∂t

When the system reaches the steady-state condition, the term ∂T/∂t is equal to zero. If the thermal conductivity is independent of the direction, the conduction energy equation can be written in a simpler form as

∂2 T ∂2 T ∂2 T q ρCP ∂T (5.15) + + + = ∂x 2 ∂ y 2 ∂z 2 k k ∂t

The energy equation is a partial differential equation with second order in space and first order in time. The boundary conditions along its surface as well as the initial condition must be specified. For the initial condition, the temperature distribution in the system must be provided. In heat transfer problems, there are three types of boundary conditions: temperature, heat flux, and convection. The constant temperature, also called the Dirichlet condition, corresponds to a situation for which the surface is maintained at a fixed temperature. The mathematical expression for this boundary condition is

T (x, t) = Ts (5.16)

228

Finite Element Simul ations Using ANSYS

The second type of boundary condition, which is called the Neumann condition, corresponds to a constant heat flux applied to a surface. The heat flux is related to the temperature gradient at the surface by Fourier’s law as follows: −k

∂T = q (5.17) ∂x

A special case of this boundary condition is an insulated boundary condition, and the heat flux should be zero: −k

∂T = 0 (5.18) ∂x

The third boundary condition corresponds to convection at a surface. The conduction–convection heat balance at the wall surface must be satisfied. The heat transfer coefficient (h) should be known, as well as the fluid bulk temperature (T∞): −k

5.2

∂T = h [ T∞ − T ( x, t )] (5.19) ∂x

Finite element formulation for conductive heat transfer

The finite element method is an efficient way to solve conduction problems. The heat transfer solution can be used to estimate the heat flow at the boundaries or to determine the temperature distribution for thermalstress analysis. In this chapter, the variation formulation is used to obtain the conductive equation. It is accomplished by minimizing the following potential function: 1 I= 2

2 ∂T 2 ∂T k x + k y dV − Q T dV − q T dS ∂ y ∂x V V S

∫

∫

∫

(5.20)

1 2 + h ( T − T∞ ) dS 2

∫ S

where the first term is heat conduction in the solid, the second term is volumetric heat generation, the third term is surface heat flux, and the fourth term is surface heat convection. For any surface, either surface heat flux or convection is applied. The temperature function T within each element can be expressed in terms of shape functions as

{T} = N1 N 2 N 3

... N i

(5.21) Ti

T1 T2 T3

229

He at tr ansfer

and the heat flux in the x- and y-directions is given by q x q y

kx = − 0

0 ky

(5.22)

∂T ∂x ∂T ∂y

The temperature gradient vector can be expressed in terms of shape functions as {g} =

∂T ∂x ∂T ∂y

∂ N1 ∂x = ∂ N1 ∂ y

∂N 2 ∂x ∂N 2 ∂y

∂N3 ∂x ∂N3 ∂y

... ...

∂N i ∂x ∂N i ∂y

(5.23) Ti

T1 T2 T3

or symbolically, {g} = [B] {T}

(5.24)

where [B] is a derivative of the shape functions. The thermal conductivity matrix is kx

0 (5.25) ky

[D] = 0

Hence, the heat flux vector can be written in terms of the temperature gradient vector and thermal conductivity matrix as follows: q x q y

= [ D ]{g} (5.26)

Potential function (5.20) can be expressed as I=

1 T {g} [D]{g} dV − {T}T [ N ]T Q dV − {T}T [ N ]T qdS 2

∫

∫

V

+

∫

V

∫ (

)

S

(5.27)

1 T h {T}T [ N ] − T∞ dS 2 2

S

Using the expression of the temperature gradient (5.24) and expanding the surface convection term, the potential function (5.27) can be written as I=

1 { T }T 2 +

∫ [ B] [ D][ B]{T} dV − {T} ∫ [ N ] Q dV − {T} ∫ [ N ] q dS T

T

T

V

V

(

T

T

S

)

1 T T h {T}T [ N ] [ N ]{T} − {T}T [ N ] + [ N ]{T} T∞ − T∞2 dS 2

∫ S

(5.28)

230

Finite Element Simul ations Using ANSYS

Minimizing the potential equation (5.28) with respect to temperature vector, it gives ∂I =0= ∂{T}

∫ [ B] [ D][ B]{T} dV − ∫ [ N ] Q dV − ∫ [ N ] q dS T

T

V

T

V

∫

S

(5.29)

+ h [ N ] [ N ]{T} − [ N ] T∞ dS T

T

S

Rearranging Equation 5.29 as

∫ [ B] [ D][ B]{T} dV + ∫ h [ N ] [ N ]{T} dS T

T

V

S

=

∫ [ N ] Q dV + ∫ [ N ] q dS + ∫ h [ N ] T dS T

V

T

S

T

(5.30)

∞

S

Equation 5.30 can be expressed in terms of conductive matrix, nodal temperature, and thermal forces as [K] {T} = {FQ} + {Fq} + {Fc} (5.31) where

∫

∫

[ K ] = [ B]T [ D ][ B] dV + h [ N ]T [ N ] dS (5.32) V

S

The first term for the conductive matrix (5.32) is for axial conduction through the element, while the second term is convection at the external surface of the element, and the thermal forces are

{FQ } = ∫ [ N ]T Q dV (5.33) V

{Fq } = ∫ [ N ]T q dS (5.34) S

∫

{Fc } = h [ N ]T T∞ dS (5.35) S

The term {FQ} is volumetric heat generation in the element, {Fq} is applied heat flux on the external surface of the element, and {Fc} is applied convection on the external surface of the element. The integrations are over the surface where convection or heat flux is applied. A symbolic expression for element conduction equation is [K] {T} = {F}

(5.36)

231

He at tr ansfer

where {F} = {FQ} + {Fq} + {Fc} (5.37) The global conductive matrix is obtained by assembling the conductive matrix for each element as follows:

[K ] =

N

∑ K

(e)

e =1

(5.38)

The thermal force vectors are assembled to form a global one as follows: N

{F } =

∑ {F } (5.39) (e)

e =1

Also, the temperature vectors are assembled to form a global one as follows: N

{ T} =

∑ {T } (5.40) (e)

e =1

5.3

F inite element method for one-dimensional heat conduction

Consider one-dimensional heat conduction, as shown in Figure 5.3, where the length of the element is L. The temperature function is similar to the displacement function, which is T(x) = N1T1 + N2T2 (5.41) where T1 and T2 are nodal temperatures, and N1 and N2 are shape functions, which are defined as

N1 = 1 −

N2 =

x (5.42) L

x (5.43) L

The shape functions can be presented in matrix form as

x

x

[ N ] = N1 N 2 = 1 − (5.44) L L

Node 1

A, k x

L

Node 2

FIGURE 5.3 One-dimensional heat conduction.

232

Finite Element Simul ations Using ANSYS

The thermal conductivity of the material is assumed to be isotropic. The [B] and [D] matrices are calculated as ∂ N1 ∂x

[ B] =

[D] = 0

kx

1 = − L

∂N 2 ∂x

0 k = k y 0

1 (5.45) L

0 (5.46) k

In Equation 5.32, and since the element has a uniform cross-sectional area, the integral over differential volume dV is replaced by A dx, where A is the cross-sectional area of the element, and the integral over differential surface dS is replaced by P dx, where P is the circumference of the element. The element conductive matrix (5.32) is expressed as

∫

T

x k L 0

x

[K ] = 1 − L x + h 1− L

∫

T

x 0 1− k L

x x 1− L L

x A dx L

(5.47)

x dA L

After performing matrix operations and integration, the element conductive matrix (5.47) is expressed as

[K ] =

Ak 1 L −1

−1 hPL 2 + 1 6 1

1 1 + hA 2 0

0 (5.48) 0

The first term of the conductive matrix is axial conduction, the second term is perimeter convection, and the third term is convection at the surface of node 1. If the convection is applied at node 2, the conductive matrix (5.48) becomes

[K ] =

Ak 1 L −1

−1 hPL 2 + 1 6 1

0 1 + hA 2 0

0 (5.49) 1

The thermal nodal forces can be calculated using expressions (5.33–5.35) as follows: T

{FQ } = ∫

x 1− L

QAL 1 x (5.50) QA dx = 2 1 L

{Fq } = ∫

x 1− L

1 qPL 1 x (5.51) + qA qP dx = 1 2 L 0

T

233

He at tr ansfer

∫

T

1 hT∞ PL 1 x + hT∞ A T∞ P dx = 2 1 L 0

x

{Fc } = h 1 − L

(5.52) In Equations 5.51 and 5.52, the first term after integration is for heat flux or convection applied at the perimeter and the second term is for heat flux or convection applied at node 1. If the convection is applied at node 2, Equations 5.51 and 5.52 become, respectively,

{Fq } = ∫ {Fc } =

∫

T

x 1− L

0 qPL 1 x (5.53) + qA qP dx = 2 1 L 1

x h 1− L

0 hT∞ PL 1 x + h T∞ A T∞ P dx = 2 1 L 1

T

(5.54) 5.4

Heat transfer through a composite wall

Consider a composite wall shown in Figure 5.4. The wall is composed of three layers with different thermal conductivities: k A = 0.1 W/m · °C, k B = 0.2 W/m · °C, and kC = 0.25 W/m · °C. A convection boundary condition is applied at the left surface, h = 15 W/m2 · °C and Ti = 25°C, while the right surface is at a fixed temperature, To = 50°C. Calculate the temperature at the interfaces and heat flow through the wall. Given that LA = 5 cm, LB = 15 cm, and LC = 2.5 cm. The heat conduction in the wall is a one-dimensional heat transfer problem and can be solved using the finite element method using three elements only, as shown in Figure 5.4. Since the temperature variation is linear in each layer, having more than one element in each layer would not affect the results. The convection thermal load is applied at node 1 and temperature load at node 4. The results are per unit area A.

To

Ti

kA

kB

kC

LA

LB

LC

Element 1

Element 2

Element 3

h

Convection

Node 1

Node 2

Node 3 Node 4

FIGURE 5.4 Heat conduction in a composite wall and finite element mesh.

234

Finite Element Simul ations Using ANSYS

For element 1, convection load is applied at node 1, and no nodal thermal load is applied at node 2, there is no convection or heat flux applied at the surface of the element, and no heat is generated in the element. The conductive matrix and nodal thermal load vector, respectively, are 1 −1 + 0 + h 1 0

k K (1) = A 1 L A −1 =

c

1

0 17 = 0 −2

1 −1 + 15 1 0

{F( ) } = {F( ) } = hT 1

0.1 1 0.05 −1

∞

0 0 −2 2

1 375 = 0 0

For elements 2 and 3, no nodal thermal loads are applied on nodes, there is no convection or heat flux applied at the element, and no heat is generated in the element. The conductive matrix and nodal thermal load vector are for elements 2 and 3, respectively, k K ( 2) = B 1 L B −1

−1 0.2 1 = 1 0.15 −1

−1 1.33 = 1 −1.33

k K (3) = C 1 L C −1

−1 0.25 1 = 1 0.025 −1

−1 10 = 1 −10

{F( ) } =

0 0

{F( ) } =

0 0

2

3

−1.33 1.33 −10 10

Assembling the conductive matrix and nodal thermal forces and temperatures yields, it gives

17 −2 0 0

−2 3.33 −1.33 0

0 −1.33 11.33 −10

0 0 −10 10

T1 T2 T3 T4

375 0 = 0 Q 4

where Q4 is heat flow at node 4, which is required to maintain the temperature at node 4 at 50°C. Since the temperature at node 4 is known, T4 = 50°C, the last equation is modified, so that T4 has a value of 50°C.

235

He at tr ansfer

Consequently, the third equation number 3 is also modified. The conductive matrix and nodal thermal forces and temperatures are expressed as 17 −2 0 0

−2 3.33 −1.33 0

0 −1.33 11.33 0

0 0 0 1

T1 T2 T3 T4

375 0 = 500 50

Then, there are three equations and three unknowns, the solution for the equations is T1 = 26.175°C T2 = 34.981°C T3 = 48.237°C

5.5

F inite element method for two-dimensional heat conduction

The simplest type of element used for heat conduction is a three-node triangular element, as shown in Figure 5.5. The nodes are named i, j, and m, and the temperature is linearly varying across the element. The temperature function is the same as the displacement function used in structural analysis:

{T} = N i N j N m

Ti Tj (5.55) Tm

and the shape functions are given as

Ni =

1 (α i + βi x + γ i y ) (5.56) 2A

Node m (xm, ym)

(xi, yi) Node i

y x

Node j (xj, yj)

FIGURE 5.5 Three-node triangular element.

236

Finite Element Simul ations Using ANSYS

Nj =

1 (α j + β jx + γ jy ) (5.57) 2A

Nm =

1 (α m + βm x + γ m y ) (5.58) 2A

where α, β, and γ are defined as

α i = x j y m − y jx m

α j = xm yi − ym xi

α m = xi y j − yi x j

βi = y j − ym

β j = ym − yi

βm = yi − y j

γ i = xm − x j

γ j = xi − xm

γ m = x j − xi

(5.59)

The [B] matrix is calculated using Equation 5.23 as

∂N i ∂x [ B] = ∂N i ∂ y

∂N j ∂x ∂N j ∂y

∂N m ∂x ∂N m ∂y

= 1 βi 2A γ i

βj γj

βm (5.60) γm

where A is the area of the element. The thermal conductivity [D] matrix is

kx

0 (5.61) ky

[D] = 0

The thermal conductive matrix (5.32)

∫

∫

[ K ] = [ B]T [ D ][ B] dV + h [ N ]T [ N ] dS

V

S

Assuming the element has a constant thickness t, the thermal conductive matrix can be written for a triangular element as

2 hLt [ K ] = tA [ B] [ D ][ B] + 1 6 0 T

1 2 0

0 0 0

(5.62)

where A is the area of the element and L is the element side length where the convection is applied. The second term in Equation 5.62 represents convection on the side of the element, and the volumetric heat generation can be considered as a nodal thermal force as follows:

{FQ } =

∫ V

[ N ]T Q dV =

1 QAt 1 (5.63) 3 1

237

He at tr ansfer

where the heat generation is equally distributed on the nodes. The heat flux on a surface is considered as nodal thermal load as

{Fq } = ∫ [ N ]T q dS = ∫ N i S

S

T

N j N m q dS (5.64)

If the heat flux is applied on a specific side of the element, the heat flux vector is

{Fq } =

q L ij 1 1 if heat flux applied between nodes i and j 2 0 0 1 1

{Fq } = qL2jm {Fq } =

(5.65)

if heat flux applied between nodes j and m (5.66)

1 qL mi 0 if heat flux applied between nodes m and i 2 1

(5.67)

where Lij is the side length between nodes i and j, Ljm is the side length between nodes j and m, and Lmi is the side length between nodes m and i. The convection on a surface is considered as nodal thermal load as follows:

∫

∫

T

{Fc } = [ N ]T h T∞ dS = N i N j N m h T∞ dS (5.68) S

S

If the convection is applied on a specific side of the element, the convection vector is 1 hT∞ L ij {Fc } = 1 if convection applied between nodes i and j 2 (5.69) 0 0 hT∞ L jm 1 if convection applied between nodes j and m 2 (5.70) 1

{Fc } =

1 hT∞ L mi 0 if convection applied between nodes m and i 2 (5.71) 1

{Fc } =

238

Finite Element Simul ations Using ANSYS

5.6

Heat conduction in a solid plane

The two-dimensional body, shown in Figure 5.6a, is subjected to convection at the vertical left side, with T∞ = 50°C and h = 20 W/m2 · °C. At the right side, a fixed temperature boundary is imposed, with To = 100°C, while the two horizontal sides are well insulated. The sides’ length of the body is 2 m, and it has a thermal conductivity of 25 W/m · °C. Determine the temperature at the center of the body. There are unlimited options for elements distributions, and some of these options are shown in Figure 5.6b. Increasing the number of elements will definitely enhance the accuracy of the results, but up to a certain number of elements. After which, the results become independent of the number of the elements. The first mesh contains two elements, which is the minimum number of elements required to solve this problem. However, this mesh cannot predict the temperature at the center of the body because there is no node at the center of the body. The second mesh contains 4 elements, the third mesh contains 8 elements, and the fourth mesh contains 16 elements. For an illustration purpose, the second mesh is selected because it has the least number of elements. Figure 5.6c shows the nodes and elements distribution for the geometry, and it consists of four elements and five nodes. First, the [B], [D], and [K] matrices for element 1 are formulated. The name of node 1 is (i) and its coordinate is (0,0), the name of node 2 is (j) and its coordinate is (2,0), and the name of node 5 is (m) and its coordinate is (1,1). The values of β’s and γ’s are required for the [B] matrix and are calculated using Equation 5.59, and A is the area of the element, so we have βi = yj − ym = 0 − 1 = −1 βj = ym − yi = 1 − 0 = 1 βm = yi − yj = 0 − 0 = 0 γi = xm − xj = 1 − 2 = −1 γj = xi − xm = 0 − 1 = −1 γm = xj − xi = 2 − 0 = 2 A = 0.5(2)1 = 1 m2 Then, the [B] matrix is formulated using Equation 5.60: 1

[ B] = −1 2 −1

1 −1

0 2

and the [D] matrix is formulated using Equation 5.61:

25 0

[D] =

0 25

239

He at tr ansfer

Insulated

T∞ = 50˚C h = 20 W/m2 ·˚C

To = 100˚C

2m (a)

(b) Node 4

Node 3 Elem. 3

Elem. 2

Node 5 Elem. 4

y Elem. 1 Node 1

x

Node 2 (c)

FIGURE 5.6 (a) The two-dimensional body subjected to convection and fixed temperature boundary conditions, (b) suggested finite element meshes, and (c) the elements distribution.

240

Finite Element Simul ations Using ANSYS

No convection is applied, and therefore, the convection term in Equation 5.62 is equal to zero. Finally, the [K] matrix for element 1 is calculated using Equation 5.62: 1 12.5 K = 0 − 12.5 (1)

2

5

0 12.5 −12.5

−12.5 −12.5 25

1 2 5

For element 2, the [B], [D], and [K] matrices are formulated. The name of node 1 is (i) and its coordinate is (0,0), the name of node 5 is (j) and its coordinate is (1,1), and the name of node 4 is (m) and its coordinate is (0,2). The values of β’s and γ’s are required for the [B] matrix, and calculated using Equation 5.59, and A is the area of the element. We then have βi = yj − ym = 1 − 2 = −1 βj = ym − yi = 2 − 0 = 2 βm = yi − yj = 0 − 1 = −1 γi = xm − xj = 0 − 1 = −1 γj = xi − xm = 0 − 0 = 0 γm = xj − xi = 1 − 0 = 1 A = 0.5(2)1 = 1 m2 Then, the [B] matrix is formulated using Equation 5.60: 1 [ B] = −1 2 −1 2 −1 0 1 and the [D] matrix is the same as for element 1. No convection is applied, and therefore, the convection term in Equation 5.62 is equal to zero. Finally, the [K] matrix for element 2 is calculated using Equation 5.62: 1 12.5 K = −12.5 0 ( 2)

5

4

−12.5 25 −12.5

0 −12.5 12.5

1 5 4

For element 3, the [B], [D], and [K] matrices are formulated. The name of node 4 is (i) and its coordinate is (0,2), the name of node 5 is (j) and its coordinate is (1,1), and the name of node 3 is (m) and its coordinate is (2,0). The values of β’s and γ’s are required for

241

He at tr ansfer

the [B] matrix, and are calculated using Equation 5.59, and A is the area of the element. We have βi = yj − ym = 1 − 2 = −1 βj = ym − yi = 2 − 2 = 0 βm = yi − yj = 2 − 1 = 1 γi = xm − xj = 2 − 1 = 1 γj = xi − xm = 0 − 2 = −2 γm = xj − xi = 1 − 0 = 1 A = 0.5(2)1 = 1 m2 Then, the [B] matrix is formulated using Equation 5.60: 1 [ B] = −1 0 1 2 1 −2 1 and the [D] matrix is the same as for element 1. No convection is applied, and therefore, the convection term in Equation 5.62 is equal to zero. Finally, the [K] matrix for element 3 is calculated using Equation 5.62: 4

12.5 K (3) = −12.5 0

5

3

−12.5 25 −12.5

0 −12.5 12.5

4 5 3

For element 4, the [B], [D], and [K] matrices are formulated. The name of node 2 is (i) and its coordinate is (2,0), the name of node 3 is (j) and its coordinate is (0,2), and the name of node 5 is (m) and its coordinate is (2,0). The values of β’s and γ’s are required for the [B] matrix, and calculated using Equation 5.59, and A is area of the element. We have βi = yj − ym = 2 − 1 = 1 βj = ym − yi = 1 − 0 = 1 βm = yi − yj = 0 − 2 = −2 γi = xm − xj = 1 − 2 = −1 γj = xi − xm = 2 − 1 = 1 γm = xj − xi = 2 − 0 = 2 A = 0.5(2)1 = 1 m2

242

Finite Element Simul ations Using ANSYS

Then, the [B] matrix is formulated using Equation 5.60: 1 [ B] = 1 1 −2 − 1 1 2 2 and the [D] matrix is the same as for element 1. Convection is applied on the side i–j of the element, and therefore, the convection term in Equation 5.62 is not equal to zero. Finally, the [K] matrix for element 4 is calculated using Equation 5.62: 2

12.5 −12.5 0

3

5

−12.5 25 −12.5

0 −12.5 12.5

2 13.3 2 3 + 6.67 0 5

3

5

66.7 13.3 0

0 0 0

2 3 5

Finally, the [K] matrix for the entire body is obtained by adding [K] matrices for elements 1, 2, 3, and 4 using the expression (5.38), and the result is 1 5

[K] =

∑ i =1

25 0 (i) [K ] = 0 0 −25

2

3

4

5

0 38.33 6.67 0 −25

0 6.67 38.33 0 −25

0 0 0 25 −25

−25 −25 −25 −25 100

1 2 3 4 5

Convection is applied on the side i–j of the element 4, the convection force vector is

1 1000 hT∞ L ij {Fc } = 1 = 1000 2 0 0

Using total conductive matrix, [K], and nodal force in expression (5.36) yields [K]{T} = {F}

25 0 0 0 0 38.33 6.67 0 6.67 38.33 0 0 0 0 0 25 −25 −25 − 25 − 25

−25 −25 −25 −25 100

T1 T2 T3 T4 T5

=

Q1 1000 1000 Q4 0

243

He at tr ansfer

where Q1 and Q4 are heat flow at nodes 1 and 4, respectively, which are required to maintain the temperature at the nodes. The temperature of nodes 1 and 4 are known, and therefore, the above equations should be modified. In nodal force vector, Q1 and Q4 are replaced by 50°C, and consequently, equations 1 and 4 are modified. The conductive matrix and nodal thermal forces and temperatures are expressed as

1 0 0 0 0

0 38.33 6.67 0 −25

0 6.67 38.33 0 −25

0 0 0 1 0

0 −25 −25 0 100

T1 T2 T3 T4 T5

=

100 1000 1000 100 5000

There are three equations and three unknowns. The T5 is the temperature at the center of the body, which is 84.62°C. The temperature at nodes 2 and 3 are equal to 69.33°C. 5.7

Thermal analysis of fin and chip using ANSYS

The fin shown in Figure 5.7 is used to manage the temperature of an electronic chip that generates heat. Heat is generated in the heat source within the chip, and its value is 25 Watts. The heat transfer process is steady. Heat convection is applied at the entire external surface, h = 15 W/m2 · °C and To = 20°C, while the bottom surface of the chip is well insulated. Determine the maximum and average temperatures at the bottom surface of the chip. The thermal conductivities of the used materials: kfin = 110 W/m · °C, kchip = 1.25 W/m · °C, and khs = 2.5 W/m · °C. Double click on the Mechanical APDL Product Launcher icon This example is limited to thermal analysis. Hence, select Thermal in the preferences. The Solid element is used, and its shape is Triangle with six nodes.

1.0 cm 5.0 cm 9.0 cm

Convection

Convection

Fin

Chip

3.0 cm

6.0 cm

Convection

Heat source Insulation

FIGURE 5.7 Heated chip with fin, and the boundary conditions.

244

Finite Element Simul ations Using ANSYS

Main Menu > Preferences

A select Thermal OK Main Menu > Preprocessor > Element Types > Add/Edit/Delete

Add...

He at tr ansfer

245

A select Solid B select Triangl 6node 35 OK

Close Only the thermal conductivity is required to solve the problem. Note that the thermal conductivities of the fin, chip, and heater are different. By default, all areas will be assigned to material number 1. In this problem, 1 is the material number of the fin, 2 for the chip, and 3 for the heater.

246

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Material Props > Material Models

A click on Thermal > Conductivity > Isotropic

A type 110 in KXX OK In the Define Material Models Behavior: Material > New Model

Make sure that the number 2 is in Define Material ID OK

He at tr ansfer

247

A select Material Model Number 2 B click on Thermal > Conductivity > Isotropic

A type 1.25 in KXX OK In the Define Material Models Behavior: Material > New Model

Make sure that the number 3 is in Define Material ID OK

248

Finite Element Simul ations Using ANSYS

A select Material Model Number 3 B click on Thermal > Conductivity > Isotropic

A type 2.5 in KXX OK Close the Material Model Behavior window The thermal conductivities are listed using list properties in utility menu. Utility Menu > List > Properties > All Materials

He at tr ansfer

249

The geometry is created using the ANSYS graphics. Setting up the workspace is done using the WP Settings. Snap is enabled to allow the mouse-click on the ANSYS graphics with an increment. Spacing is the distance between the vertical or horizontal grids. The size of the grids is specified in the Minimum and Maximum. The space is divided into squares with side length of 0.01 m. Both the total width and the height of the grids are 0.09 m. This setup makes the modeling easy by creating keypoints, lines, and areas in the ANSYS graphics. The grids should be first activated by selecting Display Working Plane in the Utility Menu. ANSYS Utility Menu > WorkPlane > Display Working Plane Utility Menu > WorkPlane > WP Settings …

A select Grid Only B type 0.01 in Snap Incr C type 0.01 in Spacing D type 0 in Minimum E type 0.09 in Maximum OK

250

Finite Element Simul ations Using ANSYS

ANSYS Utility Menu > PlotCtrls > Pan-Zoom-Rotate … Click on zoom in and out, until the ANSYS graphics show all the grids.

ANSYS graphics show the grids The keypoints are created first, and then lines. Finally, fin, heater, and chip areas are created. The thermal conductivities of the chip, heater, and fin are assigned using element attribute in the meshing tools window. ANSYS Main Menu > Preprocessor > Modeling > Create > Keypoints > On Working Plane Click on the ANSYS graphics window at the location of the keypoints, as shown in the following figure. OK Now, the grids are deactivated by selecting Display Working Plane in Utility Menu. ANSYS Utility Menu > WorkPlane > Display Working Plane

ANSYS graphics show the keypoints

251

He at tr ansfer

ANSYS Main Menu > Preprocessor > Modeling > Create > Lines > Lines > Straight Line Click on two keypoints to create a line, and continue to create all lines. The created lines are shown in the following figure. OK

ANSYS graphics show the lines ANSYS Main Menu > Preprocessor > Modeling > Create > Areas > Arbitrary > By Lines Click on chip lines to create a large rectangle area, and then in Create Area by Lines window, click on Apply Click on heater lines to create a small rectangle area, and then in Create Area by Lines window, click on Apply Click on fin lines to create the fin area, and then in Create Area by Lines window, click on OK

ANSYS graphics show the areas

252

Finite Element Simul ations Using ANSYS

The area between the heater and the chip is separated, and the area of the heater is just over the area of the chip. The two areas should be connected by common lines. Overlap merges the two areas, and the boundary lines of the heater are sheared with the chip. Main Menu > Preprocessor > Modeling > Operate > Booleans > Overlap > Areas In Overlap Areas window, click on Pick All Main Menu > Preprocessor > Meshing > Mesh Tool

A select Area B click on Set

He at tr ansfer

253

Click the chip’s area only

The above window will show up. Because the chip and heater have the same centroid, ANSYS is inquiring about which area should be selected. The chip or heater selection can be switched by clicking on Prev and Next buttons in the Multiple_Entities window. Make sure that the chip’s area is highlighted. In Multiple_Entities window, click on OK In Area Attributes window, click on OK By selecting number 2, the properties of number 2 in the material model are assigned to the chip. The fin by default has the properties of number 1 in the material model.

A select 2 in Material number Apply

254

Finite Element Simul ations Using ANSYS

Click the heater area only

The chip or heater selection can be switched by clicking on Prev and Next buttons in the Multiple_Entities window. Make sure that the heater’s area is highlighted. OK In Area Attributes window, click on OK By selecting number 3, the properties of number 3 in the material model are assigned to the heater.

A select 3 in Material number OK To ensure that thermal conductivity of the fin and chip are assigned correctly, the components are colored according to their material number.

He at tr ansfer

255

The material coloring has no effect on the solution. The geometry is meshed with triangular six-node elements. A free mesh is generated using the smart mesh option. The mesh refinement is 1. Utility Menu > PlotCtrls > Numbering …

A select Material numbers B select Colors only OK Utility Menu > Plot > Areas

ANSYS graphics show the heater, fin, and chip with different colors

256

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 Mesh In Mesh Areas window, click on Pick All Close the Mesh Tool window Close Boundary conditions are applied in the solution task. A volumetric heat generation is applied to the heater only. The convection boundary condition is applied to the fin surface and the vertical sides of the chip. The bottom surface of the chip is well insulated, and the

He at tr ansfer

257

zero heat flux simulates the insulation boundary condition. However, if no boundary condition is specified at any external surface, ANSYS will consider it as an insulated boundary condition. No boundary conditions are applied at a common line between the chip and fin, and the chip and heater. The heat generation must be per unit volume. The applied heat generation is divided by the area of the heater because the problem is two dimensional. The heater volumetric heat generation is calculated as

Q=

25 = 50,000 W /m 3 0.05 × 0.01 × 1

Main Menu > Solution > Define Loads > Apply > Thermal > Heat Generat > On Areas Click on the heater area, where heat generation is applied.

Because the chip and heater are overlapped, and having the same centroid, ANSYS is inquiring about which area should be selected. The chip or heater selection can be switched by clicking on Prev and Next buttons in the Multiple_Entities window. Make sure that the heater’s area is highlighted. OK In Apply HGEN on areas window, click on OK

A type 50000 in Load HGEN value OK

258

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Define Load > Apply > Thermal > Convection > On Lines Click on fin external surfaces and two vertical surfaces of the chip where the convection boundary condition is applied. Then in Apply CONV on lines window, click on OK

A type 15 in Film coefficient B type 20 in Bulk temperature OK The preprocessor and solution tasks are completed. Main Menu > Solution > Solve > Current LS

OK

He at tr ansfer

259

Close The solution is successfully completed, and no error messages are posted. In the postprocessor, the temperature contours should be carefully inspected to ensure that the boundary conditions are applied correctly. Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click on Nodal Solution > DOF Solution > Nodal Temperature OK

ANSYS graphics show the temperature contours

260

Finite Element Simul ations Using ANSYS

The temperature contours indicate that the maximum temperature is located at the bottom surface of the chip, and is equal to 26.431°C. The default number of contours is 9, and this number can be increased to a higher value for better data analysis. First, the graphics device must be changed to win32C, and then the number of contours can be increased up to 114 contours. Notice that increasing the number of contours does not mean that the accuracy of the result is improved. The vector plot showing the heat flow from the chip to the fin can be presented. The red arrow is for high value of the heat flux. The average temperature at the base of the chip is calculated using the path operation. To create a path, there are two options: Arbitrary and Circular paths. The Arbitrary path can be made from straight-line segments by clicking on the ANSYS graphics, and the grids should be enabled. For this example, the Arbitrary path is utilized. Main Menu > General Postproc > Path Operations > Define Path > On Working Plane

OK Click on the ANSYS graphics window at the right and left bottom corners of the chip, as shown in the following figure.

He at tr ansfer

261

A click on left bottom corner of the chip B click on right bottom corner of the chip Then in On Working Plane window, click on OK

A name the path as base B type 100 in Number of divisions OK The name of the path is optional. The number of the data set is the maximum number of field variables. The number of division is 20 by default, and increasing this number to 100 will produce a smoother plot. Next, the field variable is assigned to the path for plotting. This can be accomplished by using the Map onto Path in the path operation. Only one variable can be selected. For this example, the temperature is selected. Main Menu > General Postproc > Path Operations > Map onto Path

A select Temperature TEMP OK

262

Finite Element Simul ations Using ANSYS

Now, the temperature is ready to be plotted. In the Plot path item, there are two options. The temperature can either be plotted or listed. The list results can be exported to another software, such as EXCEL. Main Menu > General Postproc > Path Operations > Plot path Items > On Graph

A select TEMP OK

ANSYS graphics show the temperature distribution at the base The temperature at the exit is perfectly parabolic due to the symmetry in the problem. The above graph indicates that the maximum temperature at the bottom surface of the chip is 26.079°C. The average temperature at the base can be determined using the integration in the path operation. The value of the integration must be divided by the path length to get the average value of the variable, and the path length

He at tr ansfer

263

is 0.09 m. Hence, number 11.111 is entered in the FACT and will be multiplied by the integration result. Selecting S in the Lab2 means that the integration is performed along the path. Main Menu > General Postproc > Path Operation > Integrate

A type 11.1111 in FACT Factor B select TEMP in Lab1 1st Path item C select S in Lab2 2nd Path item OK

The ANSYS output window shows the value of the average temperature at the base which is 25.076°C.

264

Finite Element Simul ations Using ANSYS

5.8

Finite element method for transient heat transfer

Finite element formulation for transient conduction heat transfer is presented in this section. The conduction energy equation in per unit volume and in Cartesian coordinate can be expressed as

∂ ∂T ∂ ∂T ∂ ∂T ∂T + kz (5.72) ky kx + + q = ρCp ∂x ∂x ∂ y ∂ y ∂z ∂z ∂t

where Cp is the specific heat of the material. The finite element method is an efficient way to solve transient conduction problems. The heat transfer solution can be used to estimate the heat flow at the system’s boundary, or to determine the temperature distribution for thermal-stress analysis. In this section, the variation formulation is used to obtain the conductive equation, and it is accomplished by minimizing the following potential function: I=

2 ∂T 1 ∂T 1 2 h(T − T∞ ) dS k x dV − Q −ρCp T dV − q T dS + ∂t 2 ∂x 2 v v s s

∫

∫

∫

∫

(5.73) The second term in Equation 5.73 adds an additional contribution for transient as follows:

ΩQ = −

∫ ( Q − ρC T ) T dV (5.74) p

V

where T = ∂T / ∂ t. The temperature at nodes can be expressed in terms of the shape functions as follows:

{T} = [ N i ]{T i } (5.75)

Substituting temperature function (5.75) into the transient term in the heat conduction equation (5.74) yields

ΩQ = −

∫ ([ N ]{T } Q − ρ C [ N ]{T }[ N ]{T }) dV (5.76) i

i

p

i

i

i

i

V

Minimizing the transient term with respect to the nodal temperature is as follows:

∂Ω Q =− ∂{Ti }

∫ ([ N ] Q − ρ C [ N ][ N ]{T }) dV (5.77) i

V

p

i

i

i

265

He at tr ansfer

The second term in Equation 5.77 is an additional term that should be added to the conductive matrix, and it is expressed as

∫ (ρC [ N ][ N ]) dV {T } = [ m ]{T } (5.78)

p

i

i

i

i

V

where [m] is the element mass matrix. Equation 5.78 is added to the conductive matrix to account for transient heat transfer as

[ K ]{T} + [ m ]{T i } = {F } (5.79)

where

∫

∫

[ K ] = [ B]T [ D ][ B] dV + h [ N ]T [ N ] dS (5.80) V

S

and {F} = {FQ} + {Fq} + {Fc} (5.81) {FQ} is volumetric heat generation in the element, {Fq} is applied heat flux on the external surface of the element, and {Fc} is applied convection on the external surface of the element. The integration is over the surface in which convection or heat flux is applied. The global conductive matrix can be obtained by assembling the conductive matrixes as follows: N

[K ] =

∑ [K

(e)

] (5.82)

e =1

and the nodal heat force are also assembled to form a global one as follows: N

{F } =

∑ {F

(e)

} (5.83)

e =1

Also, the global temperature vector can be obtained as N

{ T} =

∑ {T

(e)

} (5.84)

e =1

5.9

Unsteady thermal analyses of a masonry brick using ANSYS

The cross-sectional area of a masonry brick, as shown in Figure 5.8a, is made of cement and hollow cylinders. The purpose of the numerical

266

Finite Element Simul ations Using ANSYS

To and ho

8 × D 0.04 m 0.04 m

Air

0.08 m

Cement

0.04 m

R 0.02 m

Cement

0.08 m

Air 0.04 m 0.04 m

0.08 m (a)

0.04 m

Ti and hi

(b)

FIGURE 5.8 (a) Masonry brick with boundary conditions. (b) The considered geometry for numerical simulation.

simulation is to study the heat flow from an outdoor to the indoor space. The initial temperature of the brick is 25°C. Property

Air

Cement

Density (kg/m3) Conductivity (W/m · K) Specific heat (J/kg · K)

1.125 0.025 1005

2500 0.8 750

At the indoor surface of the brick, the convection boundary condition is applied with time-independent temperature and heat transfer coefficient, Ti = 20°C and hi = 10 W/m2 · °C. At the outdoor surface, the heat transfer coefficient is time independent, ho = 25 W/m2 · °C, but the temperature is time dependent. The following expression is used to simulate the outdoor temperature variation:

2π T ( t ) = 35 + 5 Sin t 86400

To ensure having a periodic condition, the simulation is kept running for 3 days, and the results for the last day are presented. Since there is symmetry in the geometry and boundary conditions, only one-eighth of the geometry is considered. Figure 5.8b shows the considered geometry for numerical simulation. Outdoor convection is applied at the upper lateral line, and indoor convection is applied at the lower lateral line. The two vertical lines are well insulated, which show temperature history at the indoor and outdoor surfaces of the brick for 1 day.

He at tr ansfer

267

Double click on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select Thermal OK Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add…

A select Solid B select Triangl 6node 35 OK

268

Finite Element Simul ations Using ANSYS

Close For the material properties, thermal conductivity, specific heat, and density are required to solve the problem because the problem is unsteady. Note that the material properties of the air and cement are different. By default, all areas will be assigned to material number 1. In this problem, the material number of the air is 1, and 2 for the cement. Main Menu > Preprocessor > Material Props > Material Models

A click on Thermal > Conductivity > Isotropic

A type 0.025 in KXX OK

He at tr ansfer

269

Click on Thermal > Specific Heat

A type 1005 in C OK Click on Thermal > Density

A type 1.125 in DENS OK In the Define Material Models Behavior: Material > New Model

Make sure that the number 2 is in Define Material ID OK

270

Finite Element Simul ations Using ANSYS

A select Material Model Number 2 B click on Thermal > Conductivity > Isotropic

A type 0.8 in KXX OK Click on Thermal > Specific Heat

A type 750 in C OK

He at tr ansfer

271

Click on Thermal > Density

A type 2500 in DENS OK The geometry is modeled by creating a rectangle and circles. The Boolean operation is utilized to merge the areas. Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

A B C D OK

type 0 in WP X type 0 in WP Y type 0.04 in Width type 0.16 in Height

272

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Modeling > Create > Areas > Circle > Solid Circle

A type 0.04 in WP X B type 0.04 in WP Y C type 0.02 in Radius Apply

A type 0.04 in WP X B type 0.12 in WP Y C type 0.02 in Radius OK

273

He at tr ansfer

ANSYS graphics show the areas The cement and the air areas are separated, and the area of the air is just over the area of the cement. The two areas should be connected by common lines. Overlap operation merges the areas, and the boundary lines of the air will be sheared with the cement. Then, the extra areas of the air are deleted. Main Menu > Preprocessor > Modeling > Operate > Booleans > Overlap > Areas In Overlap Areas window, click on Pick All Main Menu > Preprocessor > Modeling > Delete > Area and Below Click on extra air areas to select them, and in Delete Area and Below window, click on OK

ANSYS graphics show the final geometry

274

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Meshing > Mesh Tool

A select Areas B click on Set Select the brick area only. In Area Attributes window, click on OK The following windows will show up. By selecting number 2, the properties of number 2 in the material model are assigned to the cement. The air by default has the properties of number 1 in the material model.

A select 2 in Material number OK

He at tr ansfer

275

To ensure that thermal conductivities of the air and cement are assigned correctly, the components are colored according to their material number. This has no effect on the solution. The geometry is meshed with triangular six-node elements. A free mesh is generated using the smart mesh option. The mesh refinement is 1. Utility Menu > PlotCtrls > Numbering …

A select Material numbers B select Colors only OK Main Menu > Preprocessor > Meshing > Mesh Tool

276

Finite Element Simul ations Using ANSYS

A select Smart Size B set the level to 1 Mesh In Mesh Areas window, click on Pick All Close the Mesh Areas window Close

ANSYS graphics show the brick and air with different colors, and finite element mesh The type of the analysis is changed to Transient, and a Full solution method is selected. This solution method fully imposes the boundary conditions at time 0. The solution control window is used to specify the total time duration, time step, and number of output results to be stored for the postprocessor. The total time duration is 3 days, or 259,200 seconds, and the total time duration is solved by approximately 1000 steps, or the s olution time step size is 250. If less time step size is used, such as 125 seconds, the results will be more accurate, but the computational time will be doubled. Data will be stored at every two subset. Hence, there will be 518 data sets available for the postprocessor task. Main Menu > Solution > Analysis Type > New Analysis

A select Transient OK

He at tr ansfer

277

OK Main Menu > Solution > Analysis Type > Sol’n Controls

A type 259200 in the Time at end of loadstep B select Time increment C type 250 in Time step size D select Write every Nth substep E type 2 in where N = F click on Transient tab In the Transient tab, selection between stepped loading or ramped loading is specified. If stepped loading is selected, loads are fully applied at the time = 0. If ramped loading is selected, loads are linearly increasing over the entire process. In this example, the stepped loading is selected.

278

Finite Element Simul ations Using ANSYS

A select Stepped loading. OK Boundary conditions and initial conditions are applied in the solution task. The function editor is used to apply a transient temperature formula for a convective boundary condition on the upper horizontal surface. This technique is convenient for this problem since an equation for the temperature is given. ain Menu > Solution > Define Loads > Apply > Functions > M Define/Edit

A type the equation: 35+5*sin(2*3.1415/86400*{TIME}) In Equation Editor, click on File then Save Save the file as Tout, and this file name is optional. After saving the function, it is required to load it to the ANSYS solution using the read file. Save

He at tr ansfer

279

Close the Equation Editor window Main Menu > Solution > Define Loads > Apply > Functions > Read File Select the file Tout.func, then Open

A type Tout in the Table parameter name, and this name is optional, and should not be the same as the file name of the function. Main Menu > Solution > Define Loads > Apply > Thermal > Convection > On Lines Click on the bottom line where the indoor convective boundary condition is applied, and then in Apply CONV on lines window, click on OK

A type 10 in the VALI Film coefficient B type 20 in the VAL2I Bulk temperature OK

280

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Define Loads > Apply > Thermal > Convection > On Lines Click on the upper line where the outdoor convective boundary condition is applied, and then in Apply CONV on lines window, click on OK

A type 25 in Film coefficient B select Existing table OK The following window will show up to select the function.

A select TOUT OK

He at tr ansfer

281

Main Menu > Solution > Define Loads > Apply > Initial Condit’n > Define To select all nodes in the domain, in Define Initial Conditions window, click on Pick All

A select TEMP in DOF to be specified B type 25 in the VALUE Initial value of DOF OK The solution task is now completed, and the model is now ready to be solved. During the solution task, ANSYS output windows will show the progress of the solution, and carefully monitor the run. Main Menu > Solution > Solve > Current LS

OK

Close

282

Finite Element Simul ations Using ANSYS

The temperature contours are presented at 12 PM of the third day, or at time = 216,000 seconds. First, the time step is loaded, and then the temperature contours are plotted. Main Menu > General Postproc > Read Results > By Time/Freq

A type 216000 in TIME Value of time or freq OK Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click on Nodal Solution > DOF Solution > Nodal Temperature OK

He at tr ansfer

283

ANSYS graphics show the temperature contours at time = 216,000 seconds Determining the temperature history at a specific location in the domain is required. Here, the temperature history at the upper and bottom surfaces is presented in graphical form. Results are presented for the third day only. Hence, the time range is from 172,800 to 259,200 seconds as follows: Main Menu > TimeHist Postpro

A click on the data properties button

284

Finite Element Simul ations Using ANSYS

A B C D

select X-Axis select Specified type 172800 in Min type 259200 in Max

OK

A click on green + button

A click on Nodal Solution > DOF Solution > Nodal Temperature Apply Click at right bottom corner, and in Node for Date window, click on Apply

He at tr ansfer

285

A click on Nodal Solution > DOF Solution > Nodal Temperature OK Click at right upper corner, and in Node for Date window, click on OK

A click on Ctrl key, and then click on TEMP_3 and TEMP_4 to select both temperatures B click on the graph button OK

286

Finite Element Simul ations Using ANSYS

ANSYS graphics show the temperature history at the selected locations Animation of the temperature contours from time = 172,800 to 259,200 seconds can be easily accomplished using animate in the PlotCtrls. The number of the frames in the animate over time is the number of pictures in the avi file, while the animation time delay is the display period between two pictures. Main Menu > General Postproc Utility Menu > PlotCtrls > Animate > Over time …

A B C D

type 50 in Number of animation frames select Time Range type 172800 and 259200 in Range Minimum, Maximum select Temperature TEMP

OK The ANSYS will show an animation of the heating process of the brick. The animation file will be stored in the working directory, and its format is avi.

287

He at tr ansfer

PROBLEM 5.1 Consider a composite wall shown in Figure 5.9. The wall is composed of four layers with different thermal conductivities: k A = 0.22 W/m · °C, k B = 0.16 W/m · °C, kC = 0.25 W/m · °C, and k D = 0.35 W/m · °C. A convective boundary condition is applied at the left surface, hi = 7.5 W/m2 · °C and Ti = 22°C, and at the right surface, h = 15 W/m2 · °C and Ti = 45°C. Calculate the temperature at the interfaces, and heat flow through the wall using the finite element method. Given that LA = 2.5 cm, LB = 15 cm, LC = 20 cm, and LD = 2.5 cm. PROBLEM 5.2 Consider a composite pipe shown in Figure 5.10. The pipe is composed of three pipes with different thermal conductivities (k A = 0.22 W/m · °C, k B = 0.16 W/m · °C, and kC = 0.25 W/m · °C), and radii (rA = 0.03 m, rB = 0.02 m, and rC = 0.01 m). A convective boundary condition is applied at the right surface: h1 = 10 W/m2 · °C and T1 = 25°C, and fixed temperature at the left surface, T2 = 45°C. The external surface of the pipe is well insulated. Calculate the temperature at the interfaces, and heat flow through the pipe using finite element method. Given that LA = 0.05 m, LB = 0.075 m, and LC = 0.025 m. PROBLEM 5.3 The two-dimensional body shown in Figure 5.11 is subjected to a fixed temperature boundary at the left and right vertical sides, T L = 100°C and T R = 50°C, respectively. The two horizontal sides are well insulated. The side length of the body is 1 m, and it has thermal conductivity of 10 W/m · °C. Determine the temperature at the center of the body using the finite element method. Consider the suggested four elements mesh.

Ti hi

kA

kB

kC

kD

LA

LB

LC

LD

kA

kB

kC

LA

LB

LC

T1

To ho

FIGURE 5.9 Heat conduction in a composite wall.

T2

h1

FIGURE 5.10 Heat conduction in a composite pipe.

288

Finite Element Simul ations Using ANSYS

Insulated

TL = 100˚C

TR = 50˚C

1m

FIGURE 5.11 Two-dimensional body subjected to fixed temperature boundary conditions and suggested finite element mesh.

PROBLEM 5.4 The two-dimensional body shown in Figure 5.12 is subjected convection at the inclined side, with TR = 50°C and h = 20 W/m2 · °C. At the left vertical side, a fixed temperature boundary is imposed, TL = 100°C, while the two horizontal sides are well insulated. It has thermal conductivity of 10 W/m · °C. Determine the temperature at the inclined surface of the body using the finite element method. Consider the suggested three elements mesh. PROBLEM 5.5 The fin shown in Figure 5.13 is used to manage the temperature of an electronic chip that generates heat. Heat is only generated in the chip, and its value is 15 Watts. The heat transfer process is steady. Heat convection is applied along the entire external surfaces, h = 10 W/m 2 · °C and To = 25°C, while the bottom surface of the chip is well insulated. Determine the maximum and average temperatures at the bottom surface of the chip, given k fin = 75 W/m · °C, and kchip = 0.95 W/m · °C. PROBLEM 5.6 Four cylindrical heaters are embedded in a conductive epoxy for heating purposes. An aluminum alloy fin is used to enhance heat flow out of the heaters. Figure 5.14 shows the geometry and the thermal conductivity of the components. Each heater generates 12.5 Watts, and free convection, h = 10 W/m2 · °C and T = 20°C, is applied at all external surfaces. Determine:

1. Maximum operating temperature in the device. 2. Temperature distribution a long path A–A. 3. Average temperature a long path A–A.

289

He at tr ansfer

0.5 m Insulated TR = 50°C h = 20 W/m2 °C 0.5 m

TL = 100°C

1m

FIGURE 5.12 Two-dimensional body subjected convection and fixed temperature boundary conditions and the suggested finite element mesh.

Convection

3.0 cm

Fin

Convection

6.0 cm

Convection

Chip

Insulation

9.0 cm

FIGURE 5.13 Heated chip with fin, and the boundary conditions.

2 cm

1.5 cm

4 × R 0.5 cm

Fin (85 W/m K)

Epoxy (15 W/m K)

Heater (25 W/m K) A

1.5 cm 2 cm

1 cm

1 cm A 3 cm

3 cm

FIGURE 5.14 Cylindrical heaters with fin.

290

Finite Element Simul ations Using ANSYS

PROBLEM 5.7 A masonry brick, as shown in Figure 5.15, is made of cement and air holes. At the indoor surface, convective boundary conditions are applied with time-independent temperature and heat transfer coefficient, Ti = 22.5°C and h i = 7.5 W/m 2 · °C. At the outdoor surface, heat transfer coefficient is time independent, ho = 27.5 W/m 2 · °C, but the temperature is time dependent. The initial temperature of the brick is 20°C. Property

Air

Cement

Density (kg/m3) Conductivity (W/m · K) Specific heat (J/kg · K)

1.125 0.025 1005

2400 0.82 740

The following expression simulates the outdoor temperature:

2π T ( t ) = 30 + 10 Sin t 86400

To ensure having a periodic condition, the simulation should be kept running for 3 days, and the results for the last day should be presented. Show temperature histories at the indoor and outdoor surfaces of the brick for the last day. PROBLEM 5.8 For the electronic board shown in Figure 5.16, the initial temperature of the entire system is 20°C. At time = 0, 24 Watts are generated in the processor, and convection is applied to the boundaries with h = 5 W/m2 · ˚C and To = 20°C, while the board bottom surface is well insulated. The total To and ho 0.05 m

Air

0.10 m Cement

0.05 m

0.10 m

0.05 m

Ti and hi

FIGURE 5.15 Masonry brick and the boundary conditions.

291

He at tr ansfer

0.005 m

0.005 m

Fin 0.045 m

Processor

Chip

Board 0.01 m 0.015 m

0.01 m 0.005 m 0.015 m

0.065 m

0.035 m

0.015 m

0.04 m

FIGURE 5.16 Electronic board.

time duration of the device is 10 hours. Solve this transient process using a time step of 25 seconds. Determine the following: a. Does the system reach the steady-state condition? b. What is the maximum temperature of the processor and chip at time = 5 hours? c. Show a temperature history at the center of the processor and chip. d. Show the temperature distribution along the external surface of the chip at time = 5 and 7.5 hours. Property Density (kg/m ) Conductivity (W/m · K) Specific heat (J/kg · K) 3

Chip

Fin

Processor

Board

900 2.1 2200

2050 45 4500

920 2.1 1300

910 3.2 1150

This page intentionally left blank

Cha p t e r SIX

Fluid mechanics

6.1

Governing equations for fluid mechanics

The mass conservation equation in a differential form can be obtained by applying the mass conservation principle on a differential control volume as shown in Figure 6.1. Considering the control volume, the net mass flow rate in the x-, y-, and z-directions can be expressed as

x-direction:

∂ (ρu ) dx dy dz (6.1) ∂x

y-direction:

∂ (ρv ) dx dy dz (6.2) ∂y

z-direction:

∂ (ρw ) dx dy dz (6.3) ∂z

The rate of change of mass inside the control volume can be obtained from the Reynolds transport theory as follows:

∂ρ

∂ρ

∫ ∂t dV = ∂t dx dy dz (6.4)

CV

The net mass flux into the control volume should be equal to the rate of change of mass inside the control volume. The mass conservation in differential form can be expressed as

∂ρ ∂ + (ρu ) + ∂∂y (ρv ) + ∂∂z (ρw ) = 0 (6.5) ∂ t ∂x

Newton’s second law on a differential control volume, as shown in Figure 6.1, can be used to obtain the conservation of momentum equation. The net forces on the control volume should be balanced with the acceleration of the control volume times its mass as follows:

a dm =

∑ dF (6.6) 293

294

Finite Element Simul ations Using ANSYS

my

mx

FIGURE 6.1 Fixed control volume.

mz+dz mz

mx+dx

my+dy

The acceleration vector field a is obtained from the total time derivative of the velocity vector

dV du dv dw a= = i+ j+ k dt dt dt dt

(6.7)

Each component of the velocity fields is a function of the space and time. Using the chain rule, the scalar time derivative can be obtained

du ( x, y,z, t ) ∂ u ∂ u dx ∂ u dy ∂ u dz (6.8) = + + + ∂ t ∂x dt ∂ y dt ∂z dt dt

where u = dx/dt is the local velocity component in the x-direction, v = dy/dt is the local velocity component in the y-direction, and w = dw/dt is the local velocity component in the z-direction. The total derivative of u is the acceleration in the x-direction

ax =

du ( x, y,z, t ) ∂ u du du du = +u +v +w (6.9) dt ∂t dx dy dz

The acceleration in the y- and z-directions can be expressed as, respectively,

ay =

dv ( x, y,z, t ) ∂v dv dv dv (6.10) = +u +v +w dt ∂t dx dy dz

az =

du ( x, y,z, t ) ∂w dw dw dw = +u +v +w (6.11) dt ∂t dx dy dz

The mass of the control volume must be equal to volume times the density as follows: dm = ρ dx dy dz (6.12)

295

Fluid mechanics

The forces on the control volume are of two types: body and surface. The body force is due to the gravity:

dFb = ρg dx dy dz (6.13)

while the surface forces are due to the surface stresses, including normal and parallel stresses. The surface stresses in the x-, y-, and z-directions are as follows:

∂τ yx ∂τ zx ∂σ dFsx = xx + + dx dy dz (6.14) ∂y ∂z ∂x

∂σ yy ∂τ xy ∂τzy dFsy = + + dx dy dz (6.15) ∂y ∂z ∂y

∂τ yz ∂σ ∂τ dFsz = zz + xz + dx dy dz (6.16) ∂y ∂z ∂x

The equations of motion in the x-, y-, and z-directions can be expressed as, respectively,

∂u ∂u ∂u ∂u x-direction: ρ +u +v +w ∂ t ∂ x ∂ y ∂z ∂τ yx ∂τ zx ∂σ = ρgx + xx + + ∂x ∂y ∂z ∂v ∂v ∂v ∂v y-direction: ρ +u +v +w ∂ t ∂ x ∂ y ∂z ∂σ yy ∂τ xy ∂τ zy = ρg y + + + ∂x ∂y ∂z ∂w ∂w ∂w ∂w z-direction: ρ +u +v +w ∂ t ∂ x ∂ y ∂z ∂τ yz ∂σ ∂τ = ρgz + zz + xz + ∂x ∂y ∂z

(6.17)

(6.18)

(6.19)

For Newtonian fluid, the stress components are obtained from the theory of elasticity, and they are

σ xx = − P + 2µ

∂u (6.20) ∂x

σ yy = − P + 2µ

∂v (6.21) ∂y

296

Finite Element Simul ations Using ANSYS

∂w (6.22) ∂z

σ zz = − P + 2µ

∂ u ∂v τ xy = τ yx = µ + (6.23) ∂ y ∂x

∂v ∂w τ yz = τ zy = µ + (6.24) ∂z ∂ y

∂w ∂v τzy = τ yz = µ + (6.25) ∂ y ∂z Substituting the stress equations into the equations of motion, we have

∂u ∂2 u ∂2 u ∂2 u ∂u ∂u ∂u ∂P ρ +u +v +w =− + µ 2 + 2 + 2 + ρgx ∂x ∂y ∂z ∂x ∂y ∂z ∂t ∂x (6.26)

∂v ∂2 v ∂2 v ∂2 v ∂v ∂v ∂v ∂P ρ +u +v +w =− + µ 2 + 2 + 2 + ρg y ∂x ∂y ∂z ∂y ∂y ∂z ∂t ∂x

(6.27) ∂w ∂2 w ∂2 w ∂2 w ∂w ∂w ∂w ∂P ρ +u +v +w =− + µ 2 + 2 + 2 + ρgz ∂x ∂y ∂z ∂z ∂y ∂z ∂t ∂x

(6.28) Equations 6.26–6.28 are called the Navier–Stokes equations. They are nonlinear and nonhomogenous partial differential equations.

Finite element method for fluid mechanics

6.2

The finite element method is utilized to solve the governing equations and to discretize the computational domain for fluid flow problems. Four nodes of quadrilateral elements were used for the numerical discretization. ANSYS has only this type of element in its library for fluid dynamics. Weighted integral statements of the mass, momentum, and energy conservations over a typical element Ω are given by

∫ w f dΩ (6.29) 1 1

Ω

∫ w f dΩ (6.30) 2 2

Ω

297

Fluid mechanics

∫ w f dΩ (6.31) 3 3

Ω

where f1, f2, and f3 are mass, momentum, and energy conservations, respectively. Then w1, w2, and w3 are weight functions, which are equal to the interpolation functions. The choice of the weight function is restricted to the spaces of approximation functions used for pressure, velocity fields, and temperature. The pressure, velocity fields, and temperature are approximated as follows:

P ( x, t ) =

4

∑Φ ( x) P ( t ) = Φ l

l

T

{P} (6.32)

l =1

u i ( x, t ) =

4

∑Ψ (x) u ( t ) = Ψ n

n i

T

{u i } (6.33)

n =1

T ( x, t ) =

4

∑Θ

m

( x ) Tm ( t ) = Θ T {T} (6.34)

m =1

where Φ, ψ, and θ are vectors of the shape functions, and P, ui, and T are vectors of nodal value of the pressure, velocity components, and temperature, respectively. The weight functions have the following correspondence: w1 ≅ Φ, w2 ≅ Ψ, w3 ≅ Θ (6.35) The mass and momentum conservations can be symbolically written in the following matrix form: [AT] {u} = 0

(6.36)

[C] {u} + [K] {u} – [A] {P} = {F}

(6.37)

[D] {T} + [L] {T} = {G}

(6.38)

The coefficients are defined by

Ai =

∂ψ

∫ ∂x Φ dΩ (6.39) Ω

T

i

∫

Ci ( u j ) = ρψ u i Ω

∂ψ T dΩ (6.40) ∂x i

298

Finite Element Simul ations Using ANSYS

∫

K ij = µ Ω

∂ψ ∂ψ T dΩ (6.41) ∂x j ∂x i

∫

Fi = ψτ i dΓ (6.42) Γ

∫

D i ( u j ) = ρ CP ψu i Ω

∫

L ij = k Ω

∂Θ T dΩ (6.43) ∂x i

∂Θ ∂Θ T dΩ (6.44) ∂x i ∂x j

∫

∫

G = Θ Q ′′′dΩ + Θ q ′′dΓ (6.45) Ω

Γ

where k is the thermal conductivity, Cp is the specific heat, and τi is the component of the total boundary stress, which is the sum of the viscous boundary stress and the hydrostatic boundary stress. The q″ is the heat flux applied at the boundary of the elements.

6.3

Entrance length in developing flow in a channel using ANSYS

Water flow in a two-dimensional channel, as shown in Figure 6.2, develops in the axial direction. At the inlet, the flow is uniform and the flow velocity is 0.005 m/s. The exit condition is a reference zero pressure. The density and viscosity of water are ρ = 998.3 kg/m3 and μ = 1.002 × 10 –3 (1.002e-3) Pa · s, respectively. Determine the entrance length and show that the mass conservation principle is satisfied.

Double click on the Mechanical APDL Product Launcher icon

This example is limited to fluid analysis. Hence, select Fluid FLOTRAN. The 2D FLOTRAN 141 element is used, and its shape is rectangle with Wall Uniform flow in

Flow 0.05 m out Wall 1.0 m

FIGURE 6.2 Channel with developing flow.

Fluid mechanics

299

four nodes. The 3D FLOTRAN 142 element is for three-dimensional analysis. The density and viscosity are the only needed properties to solve this problem. Main Menu > Preferences

A select FLOTRAN CFD OK Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add…

300

Finite Element Simul ations Using ANSYS

A select 2D FLOTRAN 141 OK

Close Main Menu > Preprocessor > Material Props > Material Models

A click on CFD > Density

Fluid mechanics

A type 998.3 in DENS OK Click on CFD > Viscosity

A type 1.002e-3 in VISC OK

301

302

Finite Element Simul ations Using ANSYS

Close the Define Material Model Behavior window Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

A B C D

type 0 in WP X type 0 in WP Y type 1 in Width type 0.05 in Height

OK The smart mesh number of one will produce insufficient mesh density to have accurate results. The elements in the domain can be additionally increased by using the lines size control in the Mesh Tool. Lines are divided into segments, which will be elements in these lines. The lines are divided by either specifying the number of divisions or the lengths of the segments. In this example, the number of divisions is specified. The vertical and lateral lines will be divided by 25 and 500 segments, respectively.

Fluid mechanics

303

Main Menu > Preprocessor > Meshing > Mesh Tool

A click on Set in Lines Click on two horizontal lines. Then, in Element Sizes on Picked Lines window, click on OK

304

Finite Element Simul ations Using ANSYS

A type 500 in NDIV No. of element divisions Apply Click on two vertical lines. Then, in Element Sizes on Picked Lines window, click on OK

A type 25 in NDIV No. of element divisions OK

Fluid mechanics

Main Menu > Preprocessor > Meshing > Mesh Tool

A click on Mesh In Mesh Areas window, click on Pick All

305

306

Finite Element Simul ations Using ANSYS

ANSYS graphics show the mesh Zero velocity is imposed at the upper and lower walls to simulate wall boundary condition. At the inlet, the velocity is uniform. At the exit, zero pressure is imposed to simulate the exit boundary condition. Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Velocity > On Lines Click on the two horizontal lines. Then in Apply VELO load on lines window, click on OK

A type 0 in VX Load value B type 0 in VY a Load value OK

Fluid mechanics

307

Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Velocity > On Lines Click at the inlet line. Then, in Apply VELO load on lines window, click on OK

A type 0.005 in VX Load value B type 0 in VY a Load value OK Main Menu > Solution > Define Load > Apply > Fluid/CFD > Pressure DOF> On Lines Click on the exit line, and then in Apply PRES on lines window, click on OK

A type 0 in PRES Pressure value OK

308

Finite Element Simul ations Using ANSYS

The present problem is steady state and adiabatic. Hence, keep the default setting in execution control. Notice that ANSYS is capable of simulating turbulent and compressible flow. The maximum number of iterations is 1000, and an additional 1000 iterations are required if the termination criterion is not satisfied. The termination criterion for the velocity components and pressure is 1 × 10 –6. The iterations will stop if the maximum number of iterations is reached or the termination criterion is satisfied. The material properties are MP, which means that the ANSYS will use properties stored in the Material Properties in the preprocessor task to solve the problem. Main Menu > Solution > FLOTRAN Set Up > Execution Ctrl

A B C D OK

type 1000 in EXEC Global iteration type 1e-6 in the termination criterion for VX Velocity component type 1e-6 in the termination criterion for VY Velocity component type 1e-6 in the termination criterion for PRES Pressure

Fluid mechanics

Main Menu > Solution > FLOTRAN Set Up > Fluid Properties

A select MP table in Density B select MP table in Viscosity OK

309

310

Finite Element Simul ations Using ANSYS

A confirmation window will show up. Read it carefully to avoid an unexpected error. The value –1 indicates that the property is not available. The conductivity and specific heat are not required to solve the problem. OK The following step will initiate the numerical iterations. Carefully examine the normalized rate of change for all field variables. The normalized rate of change should reach termination criterion for all field variables to declare the convergence. Otherwise, additional iterations are required. When the normalized rate of change is decreasing, the solution process is approaching the convergence. If the solution is diverged, there are either incorrect boundary conditions or fluid properties. Main Menu > Solution > FLOTRAN Set Up > Run FLOTRAN

Close

ANSYS graphics show the solution convergence The normalized rate of change reaches the termination criterion. Hence, the convergence is reached. The Last Set is selected in the Read Results to ensure that the data from the last iteration set are loaded for the postprocessor. Otherwise, no results will be shown in the postprocessor. A plot of velocity vectors is presented in the postprocessor task, followed by velocity profile along the centerline of the channel and at the exit.

Fluid mechanics

311

Main Menu > General Postproc > Read Results > Last Set Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A select X-Component of fluid velocity OK

ANSYS graphics show the vector for the velocity The red contours are for the maximum velocity in the channel, while the blue contours are for minimum velocity. The velocity is maximum at the centerline and zero at the wall. The developing region is clearly visible at the entrance region of the channel. Notice that the velocity has a parabolic velocity profile at the exit of the channel. To determine

312

Finite Element Simul ations Using ANSYS

the entrance length, a plot of the x-velocity component along the channel’s centerline is created using the path operation in the postprocessor. The x-velocity component should be increased from its initial value at the inlet until it becomes unchanged. The path is created by specifying two points, one at the inlet and the other at the exit along the centerline. The number of division is the number of points used to create the plot. A higher number of divisions will create a smoother plot. ANSYS Utility Menu > WorkPlane > Display Working Plane Utility Menu > WorkPlane > WP Setting

A select Grid Only B type 0.025 in Snap Incr C type 0.025 in Spacing OK

Fluid mechanics

313

Main Menu > General Postproc > Path Operations > Define Path > On Working Plane The path is arbitrary. In On Working Plane window, click on OK Click on the ANSYS graphics window at the right and left ends of the centerline of the channel as shown below.

A click on left end of the centerline B click on right end of the centerline In On Working Plane window, click on OK

A type center in Define Path Name; the name of the path is optional B type 100 in Number of divisions OK

314

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Path Operation > Map onto Path

A select Velocity VX OK The velocity in the x-direction is ready to be plotted. In the Plot Path Item, there are two options. The velocity can be either plotted or listed. The list results can be exported to another graphical software such as EXCEL. First, the grids are removed from ANSYS graphics. ANSYS Utility Menu > WorkPlane > Display Working Plane Main Menu > General Postproc > Path Operation > Plot Path Item > On Graph

A select VX OK

Fluid mechanics

315

ANSYS graphics show the x-velocity profile along the centerline of the channel The figure indicates that the flow is fully developed at the exit of the channel. The x-velocity component becomes invariant at a distance of 0.75 m from the entrance. Experimentally, the entrance length can be determined using the following equation: L = 0.056 Re H where Re is the Reynolds number and H is the channel’s height. The Reynolds number is calculated using the following expression:

Re =

ρVH 998.3 × 0.005 × 0.05 = = 249.07 µ 1.002 × 10 −3

Then, the entrance length is 0.697 m, which is close to the ANSYS results. The error between the two methods can be additionally reduced if a finer mesh is used. To determine the average velocity at the exit, the path operation is used to plot the velocity profile with integration. Utility Menu > Plot > Areas Main Menu > General Postproc > Path Operation > Define Path > On Working Plane The path is arbitrary. In On Working Plane window, click on OK Click on the ANSYS graphics window at the top and bottom c orners at the exit of the channel, and then in On Working Plane window, click on OK

316

Finite Element Simul ations Using ANSYS

A type Exit in Define Path Name; the name of the path is optional B type 100 in Number of divisions OK Main Menu > General Postproc > Path Operation > Map onto Path

A select Velocity VX OK Main Menu > General Postproc > Path Operation > Plot Path Item > On Graph

A select VX OK

Fluid mechanics

317

ANSYS graphics show the x-velocity profile at the exit of the channel The velocity profile at the exit is perfectly parabolic. This graph can be compared to the analytical velocity profile of a fully developed flow to ensure that the obtained solution is accurate. The average velocity at the exit can be determined using the integration in the path operation. The value of the integration must be divided by the path length to get the average value. The number 20 in the Factor is the inverse of the path length. Selecting S in the Lab2 means that the integration is performed along the path. Main Menu > General Postproc > Path Operation > Integrate

A type 20 in FACT Factor B select VX in 1st Path item C select S in 2nd Path item OK

318

Finite Element Simul ations Using ANSYS

The ANSYS Output window shows the average velocity at the exit, which is 0.48 × 10 –2 m/s. The mass flow rate at the exit and inlet should be the same as follows:

i =m e m

or (ρVA)i = (ρVA)e Since the density and cross-sectional area of the inlet and exit are the same, then Vi = Ve Comparing the inlet velocity, which is 0.005 m/s, to the exit velocity, which is 0.48 × 10 –2 m/s, the two velocities are close to each other with an error of 4%.

6.4

S tudying flow around a half cylinder in a channel using ANSYS

Air at low velocity enters a channel as shown in Figure 6.3. A half cylinder with a circular cross section is placed at the bottom wall of the channel. The inlet velocity has fully developed velocity profile with an average velocity of Um = 0.005 m/s. The exit condition is a reference zero pressure. Air’s density and viscosity are ρ = 1.25 kg/m3 and y x

0.15 m

Flow out

Half cylinder

0.1 m

FIGURE 6.3 Flow over a half cylinder in a channel.

0.75 m

H = 0.2 m

Fluid mechanics

319

μ = 17.7 × 10 –6 Pa · s, respectively. Determine the pressure drop in the channel and drag force on the half cylinder. Use the following equation for the velocity profile at the inlet:

u ( y) =

2 3 2y U m 1 − 2 H

Double click on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select FLOTRAN CFD OK Main Menu > Preprocessor > Element Type > Add/Edit/Delete

Add…

320

Finite Element Simul ations Using ANSYS

A select 2D FLOTRAN 141 OK

Close Since this problem is fluid dynamics, density and viscosity are required to solve this problem. The geometry is modeled by creating a rectangle and a circle. The Boolean operation is utilized to remove the circle from the rectangle using subtraction. Main Menu > Preprocessor > Material Props > Material Models

A click on CFD > Density

Fluid mechanics

A type 1.25 in DENS OK Click on CFD > Viscosity

A type 17.7e-6 in VISC OK

321

322

Finite Element Simul ations Using ANSYS

Close the Define Material Model Behavior window Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

A B C D

type 0 in WP X type –0.1 in WP Y type 1 in Width type 0.2 in Height

OK Main Menu > Preprocessor > Modeling > Create > Areas > Circle > Solid Circle

A type 0.2 in WP X B type –0.1 in WP Y C type 0.05 in Radius OK

Fluid mechanics

323

Main Menu > Preprocessor > Modeling > Operate > Booleans > Subtract > Areas Click on rectangle area, and then in Subtract Area window, click on Apply Click on circular area, and then in Subtract Area window, click on OK A free mesh is generated using the smart mesh option, and the mesh refinement is 1. More elements will be added to the computational domain using refinement at elements. Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 C Click on Mesh In Mesh Areas window, click on Pick All

324

Finite Element Simul ations Using ANSYS

D select Elements E click on Refine In Refine Mesh at Area window, click on Pick All

OK

ANSYS graphics show the final mesh The modeling and meshing tasks are now completed. The hydrodynamics boundary conditions are applied. At the inlet, Vx is a function of the y-direction and the Vy is 0. At the exit, zero pressure is imposed to simulate a free exit boundary condition. Zero velocity components are applied at the surface of the cylinder and lateral surfaces of the channel to simulate a wall boundary condition. The inlet velocity profile can be applied easily with the function editor of ANSYS. ANSYS will create a data table from the velocity function and apply it to the selected lines.

Fluid mechanics

325

Main Menu > Solution > Define Loads > Apply > Functions > Define/Edit

A type the equation 1.5*0.005*(1−(2*{Y}/0.2)^2) In Equation Editor window, click on File then Save Save the file as VinY. The file name is optional. Save Close the Function Editor window After saving the function, it is required to load the function to the ANSYS solution using the read file in function. Main Menu > Solution > Define Loads > Apply > Functions > Read File Select VinY.func Open

326

Finite Element Simul ations Using ANSYS

A type VinY in the Table parameter name. The name of the table is optional. The function “VinY” will be shown later when the boundary conditions are applied. OK Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Velocity > On Lines Click on the channel entrance line, and then in Apply VELO load on lines window, click on OK

A select Existing table in Apply VX load as a B type 0 in VY a Load value OK

Fluid mechanics

327

A select VINY OK Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Velocity > On Lines Click on the upper and lower lines of the channel, and the surface of the cylinder, and then in Apply VELO load on lines window, click on OK

A type 0 in VX Load value B type 0 in VY a Load value OK

328

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Pressure DOF > On Lines Click on the channel’s exit line, and then on Apply PRES on lines window, click on OK

A type 0 in PRES Pressure value OK The maximum number of iterations is 1000, and an additional 1000 iterations is required if the termination criterion was not satisfied. The termination criterion for the velocity components and pressure is 1 × 10 –6. The iterations will stop if the maximum number of iterations is reached or the termination criterion is satisfied for all field variables. Main Menu > Solution > FLOTRAN Set Up > Execution Ctrl

A B C D OK

type 1000 in EXEC Global iterations type 1e-6 in VX Velocity component type 1e-6 in VY Velocity component type 1e-6 in PRES Pressure

Fluid mechanics

329

By default, the shear stress will not be calculated by ANSYS, which is required to calculate the lift and drag forces. To calculate the shear stress, it must be selected in the additional output in the FLOTRAN setup. Main Menu > Solution > FLOTRAN Set Up > Additional Out > RFL Output Derived

A select TAUW Output wall shear stress? OK Main Menu > Solution > FLOTRAN Set Up > Fluid Properties

A select MP table in Density B select MP table in Viscosity OK

330

Finite Element Simul ations Using ANSYS

OK Main Menu > Solution > FLOTRAN Set Up > Run FLOTRAN

OK

ANSYS graphics show the solution convergence

Fluid mechanics

331

As shown in the normalized rate of change, the solution did not reach the maximum number of iterations, which is 1000, and all field variables reach the termination criterion, which is 1 × 10 –6. Therefore, the solution is converged. A plot of velocity vectors is presented in the postprocessor task, followed by pressure drop, and lift and drag calculations. Main Menu > General Postproc > Read Results > Last Set Main Menu > General Postproc > Plot Results > Vector Plot > Predefined

OK

ANSYS graphics show the vector for the velocity

332

Finite Element Simul ations Using ANSYS

The red arrows are for the maximum velocity in the channel, while the blue arrows are for minimum velocity. The wake flow behind the cylinder is visible in the figure, and the flow is very slow at this region. The flow velocity is maximum above the cylinder. Notice that the velocity has a parabolic profile at the inlet and exit of the channel. To calculate the pressure drop, the average pressure at the inlet is determined using path operation in the postprocessor. The path at the inlet is created by specifying two points. The grids should be enabled. Utility Menu > Plot > Areas Utility Menu > WorkPlane > Display Working Plane Utility Menu > WorkPlane > WP Setting

A select Grid Only B type 0.01 in Snap Incr C type 0.01 in Spacing OK

Fluid mechanics

333

Main Menu > General Postproc > Path Operation > Define Path > On Working Plane

OK ANSYS Utility Menu > PlotCtrls > Pan-Zoom-Rotate … Click on zoom in and out until the ANSYS graphics show all grids. Click on the ANSYS graphics window at the top and bottom corners at the inlet of the channel, and then in On Working Plane window, click on OK

A type inlet in Define Path Name. The name of the path is optional B type 100 in Number of divisions OK

334

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Path Operation > Map onto Path

A select Pressure PRES OK Utility Menu > WorkPlane > Display Working Plane Main Menu > General Postproc > Path Operation > Plot path Item > On Graph

A select PRES OK

ANSYS graphics show the pressure distribution at the inlet of the channel

Fluid mechanics

335

The average pressure at the inlet can be determined using the integration in the path operation. The value of the integration must be divided by the path length to get the average value of the variable. The number 5 in the Factor is the inverse of the path length. Selecting S in the Lab2 means that the integration is performed along the path. Main Menu > General Postproc > Path Operation > Integrate

A type 5 in FACT Factor B select PRES in Lab1 1st Path item C select S in Lab2 2nd Path item OK

The ANSYS Output window shows the average pressure at the inlet, which is 0.34299 × 10 –4 N/m 2. At the exit, the pressure is specified as a boundary condition and equal to zero. Hence, the pressure drop in the channel is equal to 0.34299 × 10 –4 N/m 2. To determine the lift and drag forces on the surface of the cylinder, nodes along

336

Finite Element Simul ations Using ANSYS

the surface of the cylinder should be selected first. Then, ANSYS automatically integrates the pressure and shear stress along the surface of the cylinder. The nodal selection is performed in the following steps: Utility Menu > Plot > Areas Utility Menu > PlotCtrls > Pan-Zoom-Rotate … Zoom the area around the cylinder. Utility Menu > Select > Entities

A select Lines OK Carefully select the two lines at the surface of the cylinder, and then in Select Lines window, click on OK

Fluid mechanics

Utility Menu > Select > Entities

A select Nodes B select Attached to C select Lines, all OK

337

338

Finite Element Simul ations Using ANSYS

Utility Menu > Plot > Nodes

ANSYS graphics show the nodes at the cylinder Main Menu > General Postproc > Nodal Calcs > Surface Integral

A select Both OK

ANSYS shows the net pressure in the x- and y-directions, which are Px = 1.534646078 × 10 –6 N/m2 Py = –6.556592423 × 10 –7 N/m2 Notice that Py has a negative value because the pressure in the y-direction is downward. The net shear stresses in the x- and y-directions are τx = 6.823577481 × 10 –7 N/m2 τy = 2.67938876 × 10 –7 N/m2

339

Fluid mechanics

Hence, the net drag force is (Px + τx) × S, and the net lift force is (Py + τy) × S, where S is the circumference of the half cylinder, and it is equal to 0.05π. To visualize the flow streamlines, the flow is traced with particles. First, the trace points will be defined at the entrance and behind the cylinder to capture the wake flow. Then, the trace particles are plotted. Finally, an animation is created. Utility Menu > Select > Everything Utility Menu > Plot > Areas Main Menu > General Postproc > Plot Results > Defi Trace Pt Click on all grid points at the entrance and just one grid point behind the cylinder, and then in Define Trace Point window, click on OK Main Menu > General Postproc > Plot Results > Plot Flow Tra

A select VSUM OK

ANSYS graphics show the streamlines

340

Finite Element Simul ations Using ANSYS

Utility Menu > PlotCtrls > Animate > Particle Flow

A type 50 in No. of frames to create B type 0.25 in Particle Size factor C select VSUM in Plot Flow Trace OK ANSYS graphics show the animation of the particles flow in the channel. PROBLEM 6.1 Air at low velocity enters a channel as shown in Figure 6.4. An object is placed in the channel, at the location shown. The inlet velocity is uniform, 0.005 m/s. The exit condition is a reference zero pressure. Let ρ = 1.25 kg/m3 and μ = 17.7 × 10 –6 Pa · s. Determine the pressure drop in the channel and net drag force on the object. Also show that the mass conservation principle is satisfied. PROBLEM 6.2 Water at low velocity enters a channel as shown in Figure 6.5. A triangular cross-sectional area cylinder is placed in the channel, at the location shown. The inlet velocity has a fully developed profile. The exit condition is a reference zero pressure. The properties of water are

Flow in

R0.03 0.15 m

Flow out

R0.03 0.09 m

FIGURE 6.4 Flow over an object in a channel.

0.5 m

0.12 m

341

Fluid mechanics

y Flow in

x

Flow out

1.0 cm

4 cm

1.0 cm

4 cm 25 cm

FIGURE 6.5 Channel with a triangular cylinder.

ρ = 998.3 kg/m3 and μ = 1.002 × 10 –3 Pa · s. Use the following equation for the velocity profile at the inlet:

2 3 2y u ( y ) = 0.0001 1 − 2 H

where H = 0.04 m. Determine the pressure drop in the channel, and drag and lift forces on the cylinder, and show that the mass conservation principle is satisfied. PROBLEM 6.3 Two cylinders in a wind tunnel, as shown in Figure 6.6, are used to simulate heat exchanger tubes. The inlet velocity is 0.0025 m/s, and air is the working fluid (μ = 20 × 10 –6 Pa · s, ρ = 1.25 kg/m3). The exit condition is a reference zero pressure. Determine the following: a. Is mass balance satisfied? b. What is the maximum shear stress on the first and second cylinders? c. What is the average shear stress on the first and second cylinders? d. What is the pressure drop in the channel? e. Is the exit flow fully developed? PROBLEM 6.4 Water at low velocity enters a backward-facing step channel as shown in Figure 6.7. The height in the channel is increased at a distance 0.5 m from the entrance as shown in the figure. The inlet velocity has a fully 2 × R 0.05

Flow in

Flow out

0.2 m

0.4 m

0.3 m 2.5 m

FIGURE 6.6 Two cylinders in a wind tunnel.

0.4 m

342 Flow in

Finite Element Simul ations Using ANSYS y

Flow out

x 0.1 m

0.4 m

0.5 m 2.0 m

FIGURE 6.7 Flow over a backward-facing step.

developed profile. The exit condition is a reference zero pressure. Water is the working fluid, and its properties are ρ = 997.1 kg/m3 and μ = 1.003 × 10 –3 Pa · s. Use the following equation for the velocity profile at the inlet:

2 3 2y u ( y ) = 0.001 1 − 2 H

where H = 0.15 m. Flow over a backward-facing step generates a recirculation zone due to the separation flow obtained from the adverse pressure gradients in the fluid flow, and the flow reattached again. Determine the reattachment length.

Cha p t e r S E V EN

Multiphysics

7.1

Introduction

In practice, the engineering problems to be simulated are m ultiphysics, and there has not been sufficient capability in the past to address the full physics of the problems. Therefore, there has been a tendency to simplify the problems by either focusing on the primary physics or decoupling the physics. For example, for heat exchanger analysis, fluid flow is solved first, followed by the thermal solution. Recently, significant improvements have been achieved in both hardware and software capabilities that made the multiphysics simulations inexpensive and accurate. As computing power increases, the ability to model the full physics is becoming a practical possibility. Hence, there is an increasing demand for multiphysics simulation software. ANSYS has the capability to model such complicated simulations. ANSYS can effectively simulate the thermal–structural, thermal–fluid, and fluid– structural problems. In this chapter, the setting up of problems involving multiphysics is introduced. The level of coupling for multiphysics simulations is illustrated. Examples of thermal–structural and thermal–fluid systems are presented. The coupling between the physical phenomena can be classified into three levels: low, medium, and high. With a low level, it may be sufficient to use a simple one-way coupling with file transfer between two physics, as analysis is thermal–structural and thermal–fluid with temperature-dependent properties. In the t hermal–structural system, the thermal analysis is done first, and then the temperature distribution is transferred to the structural analysis as a boundary condition. This analysis is fast, and the thermal and structural analysis is completely separate, because the structural deformation has a negligible effect on the temperature distribution. The process is shown in Figure 7.1. The problem with a medium level of coupling is that the solution of each physics depends on another but requires no mesh movement, such as thermal–fluid analysis with temperature-dependent properties.

343

344

Finite Element Simul ations Using ANSYS

In the thermal–fluid analysis, the properties of the fluid, such as the density and viscosity, are a function of temperature. In this analysis, fluid is solved first, followed by thermal solution. The properties of the fluid are updated, and then the fluid is solved again. This process is continued until the solution is fully converged. The process is shown in Figure 7.2. If all properties of the fluid are independent of the temperature, the heat transfer and fluid flow can be solved separately, and the simulation is considered as a low-level analysis. The fluid–structural analysis is considered as a high-level analysis. It requires a degree of compatibility in the solver technologies and often involves a mesh movement, which may be accounted for using the Arbitrary Lagrangian Eulerian method. The computation time for this analysis is high, and it requires high-performance computers. The process is shown in Figure 7.3. There are two types of fluid–structural analyses. In the first type, the solid is moving in a specified way and the fluid pressure has no effect on the movement of the solid. In the second type, which is much more complex, the fluid pressure affects the movement of the structure, and possible structural deformation occurs.

Boundary conditions Geometry

Thermal

Thermal results

Structural

Structural results

Properties

FIGURE 7.1 Low-level analysis.

Boundary conditions Geometry

Fluid

Properties

Fluid results

Thermal

Thermal results

Structural

Structural results

Properties update

FIGURE 7.2 Medium-level analysis.

Boundary conditions Geometry

Fluid

Properties

FIGURE 7.3 High-level analysis.

Fluid results Geometry update

345

Multiphysics

7.2

Thermal and structural analysis of a thermocouple using ANSYS

Furnaces use thermocouples to control their temperature, and they come in various designs. The common design of thermocouples is the two-plate design. It consists of aluminum and carbon steel plates attached to each other as shown in Figure 7.4. The left end is fixed and maintained at 100°C, and other surfaces are exposed to a free convection boundary condition with h = 5 W/m 2 · °C and T∞ = 15°C. Determine the maximum displacement in the y-direction. The thermophysical properties of the aluminum and carbon steel are shown in Table 7.1. This problem is considered as multiphysics, and physics are not coupled because the structure and heat transfer are solved separately. The deformation of the thermocouple has no effect on the heat transfer. The heat transfer must be solved first, and then the structure. Double click on the Mechanical APDL Product Launcher icon A solid thermal element is used, and its shape is quadratic with four nodes. This element will be replaced by a structural element when the thermal part is completely solved.

T = 15°C, h = 5 W/m2 . °C Aluminum 0.005 m

T = 100°C

0.005 m Carbon steel T = 15°C, h = 5 W/m2 . °C 0.1 m

FIGURE 7.4 A thermocouple consists of aluminum and carbon steel plates.

Table 7.1 Thermophysical properties of aluminum and carbon steel Property

Aluminum

Carbon steel

Thermal conductivity (W/m · K) Young modulus (Pa) Poisson’s ratio Thermal expansion (1/K)

83 70 × 109 0.33 23 × 10–6

111 210 × 109 0.29 12 × 10–6

346

Finite Element Simul ations Using ANSYS

Main Menu > Preferences

A select Structural B select Thermal OK Main Menu > Preprocessor > Element Types > Add/Edit/Delete

Add...

Multiphysics

347

A select Solid in Thermal Mass B select Quad 4 node 55 OK

Close The thermal conductivity is required for the thermal part, and the modulus of elasticity and Poisson’s ratio are required for the structural part. Additionally, the thermal expansion of both m aterials is important to account for the deflection of the thermocouple. There will be two different sets of properties for aluminum and carbon steel. The material number 1 in the Define Material Model Behavior is for aluminum, while material number 2 is for carbon steel.

348

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Material Props > Material Models

A click on Structural > Linear > Elastic > Isotropic

A type 70e9 in EX B type 0.33 in PRXY OK Click on Structural > Thermal Expansion > Secant Coefficient > Isotropic

A type 23e-6 in ALPX OK

Multiphysics

349

Click on Thermal > Conductivity > Isotropic

A type 83 in KXX OK In the Define Material Models Behavior menu: Material > New Model OK Select material number 2 in the Material Models Defined, and then Click on Structural > Linear > Elastic > Isotropic

A type 210e9 in EX B type 0.29 in PRXY OK

350

Finite Element Simul ations Using ANSYS

Click on Structural > Thermal Expansion > Secant Coefficient > Isotropic

A type 12e-6 in ALPX OK Click on Thermal > Conductivity > Isotropic

A type 111 in KXX OK Close the Material Models Behavior window The geometry of the problem is relatively simple. Two rectangles are created, and then they are glued.

Multiphysics

Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle by 2 Corners

A B C D

type 0 in WP X type 0 in WP Y type 0.1 in Width type 0.005 in Height

Apply

A type 0 in WP X B type 0.005 in WP Y

351

352

Finite Element Simul ations Using ANSYS

C type 0.1 in Width D type 0.005 in Height OK Main Menu > Preprocessor > Modeling > Operate > Booleans > Glue > Areas In Glue Areas window, click on Pick All Main Menu > Preprocessor > Meshing > Mesh Tool

A select Areas in Element Attributes B click on Set in Element Attributes

Multiphysics

353

select the carbon steel area only, and then in Area Attributes window, click on OK

A select 2 in Material number OK Utility Menu > PlotCtrls > Numbering …

A select Material numbers in Elem/Attrib numbering B select Colors only in Numbering shown with OK

354

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 Mesh In Mesh Area window, click on Pick All Close the Mesh Tool window Close

355

Multiphysics

ANSYS graphics show the mesh Main Menu > Solution > Define Loads > Apply > Thermal > Temperature > On Lines Click on the two left vertical lines where the temperature boundary condition is applied, and then in Apply TEMP on Lines window, click on OK

A select TEMP B type 100 in VALUE Load TEMP value C select Yes OK

356

Finite Element Simul ations Using ANSYS

Main Menu > Solution > Define Loads > Apply > Thermal > Convection > On Lines Click on lines where the convection boundary condition is applied: the two external horizontal lines and the right two vertical lines. Do not click on the line between the two materials. Then in Apply CONV on lines window, click on OK

A type 5 in Film coefficient B type 15 in Bulk temperature OK Now, the problem is ready to be solved as a heat transfer problem. Main Menu > Solution > Solve > Current LS

Close

Multiphysics

357

The thermal solution is now completed. The thermal elements must be replaced by structural elements with the same element type. This can be done using the Switch element type in the preprocessor. The temperature contours are plotted to ensure that the thermal solution is correct before solving it as a structural problem. Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click Nodal Solution > DOF Solution > Nodal Temperature OK

ANSYS graphics show the temperature contours The temperature contours show the left side is maintained at 100°C, and the temperature is decreasing in the positive x-direction due to the convective cooling. In general, the temperature contours are as expected. Next, the element type is switched to structural type, and the nodal temperature solution from thermal analysis is loaded to the structural nodes.

358

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Element Type > Switch Elem Type

A select Thermal to Struc in Change element type OK

Close Main Menu > Solution > Define Loads > Apply > Structure > Temperature > from Thermal Analysis

A click on Browse … Select from the thermal results file that has an extension of rth, and then click on Open

OK

Multiphysics

359

The structural boundary condition is the zero displacement at the left side of the thermocouple. Main Menu > Solution > Define Loads > Apply > Structural > Displacement > On Lines In the ANSYS graphics, click on the two vertical left lines where zero displacement is applied. Then, in Apply U,ROT on Lines window, click on OK

A select All DOF B type 0 in VALUE Displacement value OK Main Menu > Solution > Solve > Current LS

OK

Close

360

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Plot Results > Deformed Shape

A select Def + undeformed OK

ANSYS graphics show the thermocouple before and after deformation The plot shows downward deflection of the thermocouple because the thermal expansion of the aluminum is higher than that of the carbon steel. Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

361

Multiphysics

A click on Nodal Solution > DOF Solution > Y-Component of displacement OK

ANSYS graphics show the displacement in the y-direction The ANSYS results indicate that the maximum displacement in the y-direction is equal to 0.755 × 10 –3 m.

7.3

Chips cooling in a channel using ANSYS

Study the thermal characteristics of two electronic chips mounted on a channel’s bottom wall using ANSYS. The configuration is shown in Figure 7.5. The working fluid is air, and thermophysical properties of the chips are listed in Table 7.2. The inlet velocity is 0.01 m/s, while the exit condition is a reference zero pressure. The inlet temperature is 25°C, and 5 W are generated in each chip. Consider the problem as a steady fluid flow and heat transfer. Determine the following:

1. Maximum velocity in the x-direction, and temperature in the domain 2. Satisfaction in mass and energy balances 3. Heat transfer coefficient distribution at the surface of the first chip

Electronic manufacturers are continuously providing the market with high-performance devices, but with high heat dissipation. Heat highly affects the performance and durability of electronic devices more than any other factor, and operating electronic devices at a temperature higher

Chips

Flow in

Flow out

0.025 m 0.1 m

0.05 m 0.05 m

FIGURE 7.5 Channel with two chips.

0.35 m

0.075 m

362

Finite Element Simul ations Using ANSYS

Table 7.2 Thermophysical properties of the air and electronic chips Property

Air

Chips

Density (kg/m3) Conductivity (W/m · °C) Specific heat (J/kg · °C) Viscosity (N · s/m2)

1.25 0.025 1006.2 18.5 × 10–6

2500 0.25 850 —

than its recommended operating value will significantly affect their reliability and functionality. Thermal analysis of a channel containing multiple heated blocks subjected to forced convection flow is extensively addressed in the literature because it simulates integrated circuit chips placed on a horizontal board. In this analysis, researchers are focusing on the temperature distribution within the chips, the maximum temperature of the chips, and distribution of the local Nusselt number along the chips’ surface. Temperature distribution within the chips is typically used to predict the reliability of some of its components, such as the solder joints, and to establish a guide for safe operating conditions. The Nusselt number is used to estimate the heat flow out of the chips and to determine the required cooling load. In addition, researchers are interested in the friction factors on the surface of the chips, and the pressure drop in the channel, which are used to measure the required fan power. Double click on the Mechanical APDL Product Launcher icon This problem is considered as multiphysics and physics are not coupled because the fluid flow and heat transfer are solved separately. ANSYS will solve the fluid flow first, and then it uses the velocity components to solve for the temperature field. ANSYS Flotran CFD can solve fluid dynamics problems, as well as the thermal problems. The selected FLOTRAN 141 element has temperature as a degree of freedom. Main Menu > Preferences

A select FLOTRAN CFD OK

Multiphysics

Main Menu > Preprocessor > Element Types > Add/Edit/Delete

Add...

A select 2D FLOTRAN 141 OK

Close

363

364

Finite Element Simul ations Using ANSYS

There will be different sets of properties for air and chips. The aterial number 1 in the Material Model Behavior is for the fluid only, m and material numbers 2 to 10 are for the solids. Main Menu > Preprocessor > Material Props > Material Models

A click on CFD > Conductivity > Density

A type 0.025 in KXX OK Click on CFD > Specific Heat

A type 1006.2 in C OK

Multiphysics

365

Click on CFD > Density

A type 1.25 in DENS OK Click on CFD > Viscosity

A type 18.5e-6 in VISC OK In the Define Material Model Behavior menu: Material > New Model

OK

366

Finite Element Simul ations Using ANSYS

A select Material Model Number 2 B click on CFD > Conductivity > Viscosity

A type 0.25 in KXX OK Click on CFD > Specific Heat

A type 850 in C OK

Multiphysics

367

Click on CFD > Density

A type 2500 in DENS OK Close the Define Material Model Behavior window The geometry is created by first creating a large rectangular area for fluid flow. Then, the two chips are created. Overlap is utilized to insert the chips into the fluid rectangular area. Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle by 2 Corners

A type 0 in WP X B type 0 in WP Y C type 0.6 in Width D type 0.075 in Height Apply

368

Finite Element Simul ations Using ANSYS

A B C D

type 0.1 in WP X type 0 in WP Y type 0.05 in Width type 0.025 in Height

Apply

A B C D OK

type 0.2 in WP X type 0 in WP Y type 0.05 in Width type 0.025 in Height

Multiphysics

369

Main Menu > Preprocessor > Modeling > Operate > Booleans > Overlap > Areas In Overlap Areas window, click on Pick All The following step is for changing the material properties of the chips from number 1 to 2. By selecting number 2, the properties of number 2 in the material model are assigned to the chips. Air by default has the properties of number 1 in the material model. Main Menu > Preprocessor > Meshing > Mesh Tool

A select Areas B click on Set Select both chips, and then in Area Attributes window, click on OK

370

Finite Element Simul ations Using ANSYS

A select 2 in Material number OK To ensure that the properties of air and chips are assigned correctly, the air and chips are colored according to their material number in the material model. This step has no effect on the solution. Utility Menu > PlotCtrls > Numbering …

A select Material numbers in Elem/Attrib numbering B select Colors only in Numbering shown with OK

Multiphysics

371

The smart mesh number 1 has insufficient mesh density to have a ccurate results for fluid dynamics problems. The number of elements in the domain is increased by using the area size control in the Mesh Tool. Main Menu > Preprocessor > Meshing > Mesh Tool

A select Smart Size B set the level to 1 C Mesh In Mesh Area window, click on Pick All

372

Finite Element Simul ations Using ANSYS

D select Areas in Refine at E click on Refine In Refine Mesh at Area window, click on Pick All

OK

ANSYS graphics show the mesh Using refine at elements instead of refine at areas will generate a similar mesh. Notice that the ANSYS properties of the material model behavior will be used for air and chips. Close the Mesh Tool Window Main Menu > Solution > Define Loads > Apply > Thermal > Temperature> On Lines Click at the inlet line of the channel to specify the inlet temperature, and in Apply TEMP on Lines window, click on OK

Multiphysics

373

A select TEMP B type 25 in VALUE Load TEMP value C select Yes OK The heat generation must be per unit volume. The applied heat generation is divided by the area of the chip because the problem is two dimensional. The chip’s volumetric heat generation is calculated as follows:

Q=

5 = 4000 W/m 2 0.05 × 0.025

Main Menu > Solution > Define Loads > Apply > Thermal > Heat Generat > On Areas Click on both chips. Then, in Apply HGEN on ARs window, click on OK

A type 4000 in Load HGEN value OK

374

Finite Element Simul ations Using ANSYS

The top and bottom boundaries are wall and insulated. Zero x- and y-velocities should be imposed. There is no need to impose a velocity boundary at the bottom surface of the chips. By default, any unassigned thermal boundary conditions will be considered as insulated. Do not impose any boundary conditions on the surface of the chips that are exposed to the flow. Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Velocity > On Lines Click on the inlet of the channel. Then in Apply VELO load on lines window, click on OK

A type 0.01 in VX Load value B type 0 in VY a Load value OK Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Velocity > On Lines Click on the top and bottom walls. Then, in Apply VELO load on lines window, click on OK

A type 0 in VX Load value B type 0 in VY a Load value OK

Multiphysics

375

Main Menu > Solution > Define Loads > Apply > Fluid/CFD > Pressure DOF > On Lines Click on channel’s exit. Then, in Apply PRES on lines window, click on OK

A type 0 in PRES Pressure value OK The fluid flow and heat transfer are solved at the same time. In s olution options, the problem will be solved as steady state, and the system is thermal. Flow can be changed from laminar to turbulent and from incompressible to compressible in this window. Main Menu > Solution > FLOTRAN Set Up > Solution Options

A change to thermal in TEMP Adiabatic or thermal? OK

376

Finite Element Simul ations Using ANSYS

In execution control, it is required to specify the number of iterations for fluid flow. Then 1000 iterations will be sufficient to reach the convergence criteria for all field variables, which is 1 × 10 –6 (1e-6). Main Menu > Solution > FLOTRAN Set Up > Execution Ctrl

A type 1000 in EXEC Global iterations B change the termination criteria for VX Velocity component to 1e-6 C change the termination criteria for VY Velocity component to 1e-6 D change the termination criteria for Pressure to 1e-6 E change the termination criteria for Temperature to 1e-6 OK Main Menu > Solution > FLOTRAN Set Up > Fluid Properties

Multiphysics

A B C D

377

select MP table in Density select MP table in Viscosity select MP table in Conductivity select MP table in Specific heat

OK

OK For the heat transfer coefficient calculation, the flow bulk temperature is required. In the reference condition, the bulk temperature is specified, which is equal to the flow inlet temperature. Main Menu > Solution > FLOTRAN Set Up > Flow Environment > Ref Conditions

A type 25 in Reference (bulk) temp OK

378

Finite Element Simul ations Using ANSYS

Main Menu > Solution > FLOTRAN Set Up > Run FLOTRAN

ANSYS graphics show a normalized rate of change for field variables

Close ANSYS graphics show the normalized rate of change of the field variables. Note the solution is fully converged. All field variables reach 1 × 10 –6 at the iteration number 185, which is less than the specified 1000 iterations. In read results, the last set must be read to have the results from the last iteration. Main Menu > General Postproc > Read Results > Last Set Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click on Nodal Solution > DOF Solution > X-Component of fluid velocity OK

Multiphysics

379

ANSYS graphics show the x-component velocity contours For the obtained results, the maximum and average velocity in the x-direction occurred above the chips, and the maximum velocity is 0.021396 m/s. This is expected because this region has the smallest cross-sectional area that forces the flow to accelerate. At the walls, the velocity is 0. Notice that the velocity has a parabolic profile at the exit of the channel. Main Menu > General Postproc > Plot Results > Contour Plot > Nodal Solution

A click on Nodal Solution > DOF Solution > Nodal Temperature OK

380

Finite Element Simul ations Using ANSYS

ANSYS graphics show the temperature contours As shown in the temperature contours, the maximum temperature is 73.2043°C, and this occurred at the second chip. To ensure that mass and energy balances are satisfied, the average velocity in the x-direction and temperature at the exit are determined using the path operation in the postprocessor. The path at the exit is created by specifying two points. ANSYS Utility Menu > WorkPlane > Display Working Plane Utility Menu > WorkPlane > WP Setting

A select Grid Only B type 0.025 in Snap Incr C type 0.025 in Spacing OK

Multiphysics

381

Utility Menu > Plot > Areas Utility Menu > PlotCtrls > Pan-Zoom-Rotate … Click on zoom in and out until the ANSYS graphics show all grids. Main Menu > General Postproc > Path Operations > Define Path > On Working Plane

OK Click on the ANSYS Graphics window at the exit’s top and bottom corners of the channel, and then in On Working Plane window, click on OK

A type Exit in Define Path Name; the name of the path is optional B type 100 in Number of divisions OK

382

Finite Element Simul ations Using ANSYS

Main Menu > General Postproc > Path Operations > Map onto Path

A select Velocity VX Apply

A select Temperature TEMP OK ANSYS Utility Menu > WorkPlane > Display Working Plane Main Menu > General Postproc > Path Operations > Plot Path Item > On Graph

A select VX OK

Multiphysics

383

ANSYS graphics show the velocity profile at the exit of the channel Main Menu > General Postproc > Path Operation > Plot Path Items > On Graph

A select TEMP OK

ANSYS graphics show the temperature profile at the exit of the channel

384

Finite Element Simul ations Using ANSYS

The average velocity and temperature at the exit are determined using the integration in the path operation. The value of the integration must be divided by the path length to get the average value of the variable. The number 13.333 in the Factor is the inverse of the path length. Selecting S in the Lab2 means that the integration is performed along the path. Main Menu > General Postproc > Path Operations > Integrate

A type 13.333 in Factor B select VX in 1st Path item C select S in 2nd Path item Apply

A type 13.333 in Factor B select TEMP in 1st Path item C select S in 2nd Path item OK

Multiphysics

385

The ANSYS Output window shows the average velocity and temperature at the exit, which are 0.95993 × 10 –2 m/s and 36.052°C, respectively. The mass flow rate at the exit and inlet should be the same as follows:

i =m e m

or ( ρ V A)i = ( ρ V A)e Since the density and cross-sectional area of the inlet and exit are the same, then Vi = Ve Comparing the inlet velocity, which is 0.01 m/s, to the exit velocity, which is 0.0095993 m/s, the two velocities are close to each other with an error of 4%. The heat gained by air in the channel can be determined using the first law of thermodynamics:

Q = m Cp ( Te − Ti )

or Q = ( ρ V A) Cp( Te – Ti ) Q = (1.25 × 0.01 × 0.075) × 1006.2 × (36.052 − 25) = 10.425 Watts The heat gained Q should be equal to the generated heat in the two chips. Comparing the heat gain to the generated heat in the two chips, which is 10 Watts, the energy balance is satisfied with an error of 4.25%. The error found in mass and energy balances can be reduced if the computational domain is meshed with a finer mesh. The path operation is performed in the following steps to obtain the distribution of the average heat transfer coefficient around the first chip. Here, the film coefficient is selected. Utility Menu > Plot > Areas ANSYS Utility Menu > WorkPlane > Display Working Plane Main Menu > General Postproc > Path Operations > Define Path > On Working Plane

386

Finite Element Simul ations Using ANSYS

OK Click on the ANSYS Graphics window at locations A, B, C, D, and E, as shown in the following figure, and then in On Working Plane window, click on OK

A type Chip1 in Define Path Name; the name of the path is optional B type 100 in Number of divisions OK

Multiphysics

387

Main Menu > General Postproc > Path Operations > Map onto Path

A select Other quantities B select Film coeff HFLM OK Now, the stored variable, HFLM, is ready to be plotted. In the Plot path item, there are two options. The stored data can be either plotted or listed. The list results can be exported into other software such as EXCEL. ANSYS Utility Menu > WorkPlane > Display Working Plane Main Menu > General Postproc > Path Operations > Plot of Path Item > On Graph

A select HFLM OK

ANSYS graphics show the heat transfer coefficient around the first chip

388

Finite Element Simul ations Using ANSYS

The average heat transfer coefficient at the chip surface can be d etermined using the integration in the path operation. The value of the integration must be divided by the path length to get the average value of the variable. Selecting S in the Lab2 means that the integration is performed along the path. Main Menu > General Postproc > Path Operation > Integrate

A type 10 in Factor B select HFLM in 1st Path item C select S in 2nd Path item OK

The ANSYS Output window shows the value of the average heat transfer coefficient around the first chip, which is 1.642 W/m2 · °C. PROBLEM 7.1 The thermocouple shown in Figure 7.6 consists of copper and carbon steel plates attached to each other. The left end is fixed and maintained at 200°C, and the thermocouple is exposed to a free convection boundary condition with h = 15 W/m2 · °C and T∞ = 20°C. Display the y-displacement contours and determine the maximum displacement in

389

Multiphysics

T = 20°C, h = 15 W/m2 . °C Copper 0.0005 m

T = 200°C

0.001 m Carbon steel T = 20°C, h = 15 W/m2 . °C 0.05 m

FIGURE 7.6 A thermocouple consists of copper and carbon steel plates.

the y-direction. The thermophysical properties of the used materials are shown Table 7.3. PROBLEM 7.2 A pipeline used for transporting highly corrosive liquid is made of concentric carbon steel and aluminum pipes as shown in Figure 7.7. The inner pipe is made of aluminum to prevent corrosion, while the outer pipe is made of carbon steel to add strength to the pipeline in Table 7.4. At the inner surface of the pipe, the temperature is 100°C, while on the outer surface the temperature is 25°C. Due to thermal expansion variation between the carbon steel and aluminum, a thermal stress is developed at the contact surface that could damage the pipeline. Determine the maximum stress in the pipe. The displacement at inner surface of the pipe is fixed. Table 7.3 Thermophysical properties of the used materials Property

Copper

Carbon steel

Thermal conductivity (W/m · K) Young’s modulus (GPa) Poisson’s ratio Thermal expansion (1/K)

401 140 0.34 16.5 × 10–6

111 210 0.29 12 × 10–6

Aluminum Carbon steel

8.0 cm 10.0 cm 11.0 cm

FIGURE 7.7 A pipeline.

composite

390

Finite Element Simul ations Using ANSYS

Table 7.4 Thermophysical properties of aluminium and carbon steel Property

Aluminum

Carbon steel

Thermal conductivity (W/m · °C) Young’s modulus (Pa) Poisson’s ratio Thermal expansion (1/K)

83 70 × 109 0.33 23 × 10–6

111 210 × 109 0.29 12 × 10–6

PROBLEM 7.3 Study the thermal characteristics of an electronic chip mounted on a channel using ANSYS. The configuration is shown in Figure 7.8. The working fluid is air, and thermophysical properties of the chip and air are listed in Table 7.5. The inlet velocity is 0.0125 m/s, while the exit condition is a reference zero pressure. The inlet temperature is 20°C, and 5 Watts are generated in a heat source. Consider the problem as a steadystate heat transfer. Determine the following: a. Maximum velocity in the x-direction and temperature in the domain b. Satisfaction of the mass and energy balances c. Heat transfer coefficient distribution at the surface of the chip. PROBLEM 7.4 The two cylinders in a wind tunnel, as shown in Figure 7.9, are used to simulate heat exchanger tubes. The inlet velocity and temperature are 0.002 m/s and 25°C, respectively. Air is the working fluid (μ = 20 × 10 –6 Pa · s, Cp = 1000 J/kg · °C, k = 0.025 W/m · °C, and ρ = 1.25 kg/m3). The surfaces of the cylinders are maintained at 50°C. The exit condition is a reference zero pressure.

Flow in

Chips

0.0175 m

0.015 m

0.005 m 0.05 m

Flow out

Heat source

0.025 m

0.25 m

FIGURE 7.8 A channel with a chip. Table 7.5 Thermophysical properties of the chip and air Property

Air

Chip

Density (kg/m3) Conductivity (W/m · °C) Specific heat (J/kg · °C) Viscosity (N s/m2)

1.25 0.025 1005 20 × 10–6

2250 0.4 950 —

0.045 m

391

Multiphysics

2 × R 0.05

Flow in

Flow out

0.2 m

0.4 m

0.4 m

0.3 m 2.5 m

FIGURE 7.9 Two cylinders in a wind tunnel.

a. Show that the mass balance is satisfied. b. Ensure maximum shear stress on the first and second cylinders. c. Find the average heat transfer coefficient on the first and second cylinders. d. Determine pressure drop in the channel.

PROBLEM 7.5 Water at low velocity enters a channel as shown in Figure 7.10. A triangular cross-sectional area cylinder is placed in the channel, at the shown location. The inlet velocity has a fully developed profile, and its temperature is uniform at 25°C. The exit condition is a reference zero pressure. The properties of water are ρ = 997.1 kg/m3, Cp = 4180 J/kg °C, k = 0.6076 W/m · °C, and μ = 0.891 × 10 –3 Pa · s. Use the following equation for the velocity profile at the inlet:

2 3 2y u ( y ) = 0.0001 1 − 2 H

where H = 0.04. Ten Watts are generated in the triangular cylinder, which has properties of ρ = 4600 kg/m3, Cp = 1055 J/kg · °C, and k = 76.5 W/m · °C.

a. Show that the mass balance is satisfied. b. Show that the energy balance is satisfied. c. Determine the maximum shear stress at the surface of the triangular cylinder. d. Determine the average heat transfer coefficient in a triangular cylinder. e. Determine pressure drop in the channel.

y Flow in

x

Flow out

1.0 cm 1.0 cm

4 cm 25 cm

FIGURE 7.10 Channel with a triangular object.

4 cm

This page intentionally left blank

Cha p t e r E IGHT

Meshing guide

8.1

Mesh refinement

The accuracy of the finite element model can be enhanced by increasing either the number or the order of the elements. An increasing number of elements is called h-refinement, while increasing the order of the elements is called p-refinement. Increasing the number of elements or their order will lead to a significant increase in the computational time and required memory to solve the problem. Consider the stress analysis of a plate shown in Figure 8.1a. A linear quadratic element is used to mesh the domain. The stress concentration factor is required, and the exact solution is available from theory of elasticity. Figure 8.1b shows the finite element mesh. The mesh consists of 55 elements. To study the effect of the number of elements on the solution, the number of elements is increased and the stress concentration is obtained for the corresponding elements number. Figure 8.2 shows the error of the stress concentration for different numbers of elements using linear quadratic elements. The domain is meshed with three different meshes, A, B, and C. The number of elements of mesh C is higher than that of mesh B, and the number of elements of mesh B is higher than that of mesh A. The figure indicates that as the number of elements increases, the relative error is decreased to approach a fixed value. Adding more elements to the mesh C will insignificantly reduce the error. When the number of elements has no effect on solution, the solution is called a mesh-independent solution. On the other hand, increasing the order of the elements reduces the error for the same mesh size. Figure 8.3 indicates that replacing the linear quadratic elements with cubical ones reduces the percentage of the error by about 10%. With high-order elements, the mesh-independent process is faster than with low-order elements, as shown in Figure 8.4. In addition, a mesh with the linear quadratic elements is becoming mesh independent at a very slow rate. The higher order elements lead to more accurate results than lower order elements. However, the time required to complete the solution is important, especially for very large meshes. In addition to time, the higher order elements need a large amount of memory that could not 393

394

Finite Element Simul ations Using ANSYS

(a)

(b)

Error (%)

FIGURE 8.1 (a) Geometry of a plate and (b) finite element mesh.

A

B

C

Number of elements

FIGURE 8.2 Error of the stress concentration for different mesh sizes using linear quadratic elements.

Error (%)

20

15

10

FIGURE 8.3 Percentage of error with different element orders.

be afforded. Figure 8.5 indicates that in spite of a low mesh convergence rate of linear quadratic elements, the time required to solve the problem is nearly independent of the number of elements in the mesh. In addition, increasing the order of the elements significantly increases the time required to solve the problem.

395

Error (%)

Meshing guide

A

B

C

Number of elements

Normalized computational time

FIGURE 8.4 Error of the stress concentration for different numbers of e lements and orders.

A

B

C

Number of elements

FIGURE 8.5 Normalized computational time for different quadrilateral elements. 8.2

Element distortion

When a complex geometry is meshed using an automatic mesh generator, elements at some critical regions in the geometry are distorted and the accuracy of the results is significantly decreased at these regions. Hence, this section concerns the element distortions and identifies these elements. In addition, guidelines about good element shapes are also introduced. In general, there are four types of elements distortion:

1. Mid-side node off-center 2. Taper 3. Skew 4. Aspect ratio

The mid-side node off-center type is when side nodes are shifted from the center of the side. As the node moves away from the center,

396

Finite Element Simul ations Using ANSYS

the errors increase. In practice, the mid-side node should not be shifted more than one-eighth of the side’s length. Rectangular elements can be skewed or tapered. The angles of the rectangular elements should be 90°. For a rectangular element, and when the height-to-width ratio exceeds 2, the element has an aspect-ratio problem. It turns out that the aspect ratio for a perfectly shaped rectangular element is 1 and 1.15 for an equilateral triangle element. Aspect ratios higher than the recommended value can only occur in noncritical regions of the model, such as in uniform temperature or stress regions. Figure 8.6 shows the major types of element distortions. Element distortions are not limited to two-dimensional spaces, they can also occur in three-dimensional spaces. The most common one is the warping. Warping occurs when the face of a solid element does not lie in the same plane, and the warping angle should not be more than 10°. 8.3

Mapped mesh

Meshes can be classified as free or mapped meshes. The elements of the free meshes are randomly distributed in the domain with different sizes and shapes, and element distortions are most likely to appear in these meshes. Figure 8.7 shows a free mesh of a corrugated channel used for fluid flow analysis. Most of the finite element software generates this type of mesh. Poor results are always associated with free meshes. The free meshes should be only limited to very complex geometries. Mapped meshes have organized distributed elements, and the meshes are almost free of the element distortions. Figure 8.8 shows a mapped mesh of a corrugated channel. The mesh designers can fully control the

(a)

(b)

(c)

FIGURE 8.6 Major types of element distortions: (a) mid-side node off-center, (b) taper and skew, and (c) aspect ratio.

397

Meshing guide

FIGURE 8.7 Free mesh of a corrugated channel.

FIGURE 8.8 Mapped mesh of a corrugated channel.

size of the elements and their distribution in the computational domain. These meshes are typically used in finite element simulations because of their flexibility and ability to mesh complex geometries. Mesh designers should have a strong understanding of the physics of the problem before generating the mapped meshes. For example, for solid mechanics problems, the areas with high and uniform stress gradients should be identified, and the mesh should be designed accordingly. For fluid problems, the regions close to the walls have very high velocity gradients due to shear forces, and elements should be dense at the region close to the walls. The geometry shown in Figure 8.9 is fin, and a thermal analysis is required. High temperature gradients at the upper region are expected. Therefore, elements must be concentrated at this region. The geometry is meshed with three different mapped meshes. The mesh in Figure 8.9a is unacceptable because the elements are not concentrated in the upper region, and poor results are expected. In the second mesh (Figure 8.9b), the upper region of the geometry has more

(a)

(b)

(c)

FIGURE 8.9 A fin (a) without elements concentration, (b) with only elements concentration at the upper region, and (c) with elements concentration at the upper region with gradual changes.

398

Finite Element Simul ations Using ANSYS

elements than the first mesh, but an abrupt change in element size should be avoided, because errors in the large elements will be transferred to the small elements. In addition, performing matrix operations of these meshes is difficult. The third mesh, Figure 8.9c, is the ideal mesh. Elements are concentrated at the upper region, and the change from the large elements in the lower region to small elements in the upper region is gradual. 8.4

Mapped mesh with ANSYS

The square plate with a hole shown in Figure 8.10 is subjected to tensile pressure at both vertical sides. Use ANSYS to create a mapped mesh of the plate and concentrate elements at the region close to the hole where high stress gradients are expected. Use quadratic elements with four nodes to mesh the plate. Double click on the Mechanical APDL Product Launcher icon Main Menu > Preferences

A select Structural OK

ϕ 1.25 cm P

FIGURE 8.10 Square plate with a hole.

P

5 cm

Meshing guide

399

Main Menu > Preprocessor > Element Types > Add/Edit/Delete

Add...

A select Solid B select Quad 4 node 182 OK

Close The geometry is modeled by creating a square and a circle. A Boolean operation is utilized to remove the circle from the square using subtraction. The advantage of symmetry in the problem is considered. Only the upper right quarter is considered. To create a mapped mesh with

400

Finite Element Simul ations Using ANSYS

ANSYS, the computational domain must be divided into areas, and each area must have four sides only. Main Menu > Preprocessor > Modeling > Create > Areas > Rectangle > By 2 Corners

A B C D

type 0 in WP X type 0 in WP Y type 0.025 in Width type 0.025 in Height

OK Main Menu > Preprocessor > Modeling > Create > Areas > Circle > Solid Circle

Meshing guide

401

A type 0.0 in WP X B type 0.0 in WP Y C type 0.00625 in Radius OK Main Menu > Preprocessor > Modeling > Operate > Booleans > Subtract > Areas Click on square area, and then in Subtract Areas window, click on Apply Click on circular area, and then in Subtract Areas window, click on OK Utility Menu > Plot > Lines

ANSYS graphics show the lines of the geometry As shown in the figure, the geometry is one area which is composed of five lines. Therefore, the mapped mesh cannot be created. The area should be divided into areas that each should be composed of four sides, and there are many options. As a suggestion for this example, the area is divided into two areas along the symmetry line of the geometry. First, a keypoint is created at the origin. Second, a line is created between the keypoints at the origin and upper right corner of the plate. Third, the area is divided by a line, which has just been created. The new Keypoint will have number 10, and this number is selected to avoid a conflict with existing keypoints.

402

Finite Element Simul ations Using ANSYS

Main Menu > Preprocessor > Modeling > Create > Keypoints > In Active CS

A type 10 in NPT Keypoint number B type 0 and 0 in X,Y,Z Location in active CS OK ANSYS Main Menu > Preprocessor > Modeling > Create > Lines > Lines > Straight Line Click on keypoints at the origin, and at the origin and upper right corner of the geometry. Then, in Create Straight Line window, click on OK

ANSYS graphics show the created line ANSYS Main Menu > Preprocessor > Modeling > Operate > Booleans > Divide > Area by Line Click on area to select it. Then, in Divide Area by Line window, click on Apply Click on the just created line to select it. Then, in Divide Area by Line window, click on OK

403

Meshing guide

Utility Menu > Plot > Areas

ANSYS graphics show the areas Now the geometry is ready to be meshed. It consists of two areas, and each area is composed of four lines. Elements should be concentrated at the region close to the hole since high stress gradients are expected at this region. The lines should be divided into segments, and the two opposite lines of an area must have the same number of division. Main Menu > Preprocessor > Meshing > Mesh Tool

A click on Set in Lines

404

Finite Element Simul ations Using ANSYS

The lines indicated in the following figure should have the same line division and spacing ratio.

Select the lines as shown in the figure. Then in Element Sizes on Picked Lines window, click on Apply

A type 25 in the NDIV The number 25 means that the selected lines will be divided into 25 segments, and each segment will be one element. Apply

Meshing guide

405

Select the lines, as shown in the following figure. Then in Element Sizes on Pick Lines window, click on

OK

A type 50 in NDIV B type 10 in SPACE The number 50 means that the selected lines will be divided into 50 segments, and the segments are gradually increasing by a spacing ratio of 10. OK

406

Finite Element Simul ations Using ANSYS

Utility Menu > Plot > Lines

ANSYS graphics show the lines division Notice that the division of the horizontal and vertical lines is wrongly implemented. The segment lengths are increasing toward the hole, and they should be reversed. The reversion can be easily done using flip in Mesh Tool as follows: Main Menu > Preprocessor > Meshing > Mesh Tool

A click on Flip in Lines Select the horizontal and vertical lines, which have a wrong division. Then in Flip Line Bias window, click on OK

407

Meshing guide

Utility Menu > Plot > Lines

ANSYS graphics show the lines division

A select Mapped Mesh In Mesh Areas window, click on Pick All

408

Finite Element Simul ations Using ANSYS

ANSYS graphics show a mapped mesh PROBLEM 8.1 Create a mapped mesh for the geometries shown in Figure 8.11, and the geometries are used for solid mechanics. F

R 1.0 cm

P 3.5 cm

P

1m F

3.5 cm

2m (a)

4.0 cm

3.5 cm

(b) F R 0.01 m

R 0.01 m

(c)

FIGURE 8.11 Geometries for solid mechanics: (a) a rectangular plate subjected to vertical and horizontal forces, (b) a notched rectangular plate subjected to a tensile pressure, and (c) two horizontal cylinders on a flat plate with a force.

409

Meshing guide

PROBLEM 8.2 Create a mapped mesh for the geometries shown in Figure 8.12, and the geometries are used for heat transfer. PROBLEM 8.3 Create a mapped mesh for the geometries shown in Figure 8.13, and the geometries are used for fluid mechanics. Insulated

T∞

To

h

2m

(a)

1.0 cm

Convection

Convection

Fin

Chip

3.0 cm

6.0 cm

Convection

5.0 cm 9.0 cm

Heat source

Insulation

(b) 0.005 m

0.005 m Fin

0.045 m

Processor

Chip

Board

0.01 m

0.015 m

0.01 m 0.005 m 0.015 m

0.065 m

0.035 m

0.04 m

0.015 m

(c)

FIGURE 8.12 Geometries for heat transfer: (a) two-dimensional body subjected to fixed t emperature and convection boundary conditions, (b) a chip and fin with heat generation in the heat source, and (c) electronic board assembly.

410

Finite Element Simul ations Using ANSYS

Flow in

Flow out

0.1 m

0.5 m

0.4 m

2.0 m (a)

y Flow in

Flow out

x

0.15 m

0.1 m

0.2 m

0.75 m (b)

y Flow in

x

Flow out

1.0 cm

4 cm

1.0 cm

4 cm 25 cm (c)

FIGURE 8.13 Geometries for fluid mechanics: (a) flow over a backward-facing step, (b) flow over a half cylinder in a channel, and (c) flow over a triangular cylinder in a channel.

Bibliography

V. A da m s and A . A s k e na z i. Building Better Product with Finite Element Analysis. ONWORD Press, Santa Fe, NM. A nsys Th e ory. ANSYS online manual. ANSYS, Canonsburg, PA. T. C h a n d ru pat l a and A . Be l e gu n du. Introduction to Finite Elements in Engineering. 3rd edition. Prentice Hall, Upper Saddle River, NJ. M. Faga n. Finite Element Analysis: Theory and Practice. Pearson Education, Harlow, Essex, England. S. M oav e n i. Finite Element Analysis: Theory and Application with ANSYS. Prentice Hall, Upper Saddle River, NJ. J. R e ddy. An Introduction to the Finite Element Method. 2nd edition. McGraw-Hill, New York, NY. E . Thom p s on. Introduction to the Finite Element Method: Theory, Programming, and Applications. Wiley, Hoboken, NJ. T. Ya ng. Finite Element Structural Analysis. Prentice Hall, Englewood Cliffs, NJ.

411

This page intentionally left blank

This page intentionally left blank

ENGINEERING – GENERAL “A must read for those interested in exploring the possibilities of using the finite element method (FEM) as a practical analysis tool for structural mechanics, stress analysis, vibration, heat transfer, and fluid dynamic problems. … Alawadhi systematically introduces the theory of FEA and demonstrates a step-by-step procedure for practical analysis using ANSYS. This book is a must-have reference for students, academics, and practitioners in the field of mechanical, civil, environmental, and design engineering.” —Dr. Arun Arjunan, University of Wolverhampton, UK

Uses a Step-By-Step Technique Directed with Guided Problems and Relevant Screen Shots Simulation use is on the rise, and more practicing professionals are depending on the reliability of software to help them tackle real-world mechanical engineering problems. Finite Element Simulations Using ANSYS, Second Edition offers a basic understanding of the principles of simulation in conjunction with the application of ANSYS. Employing a step-by-step process, the book presents practical end-of-chapter problems that are solved using ANSYS and explains the physics behind them. The book examines structure, solid mechanics, vibration, heat transfer, and fluid dynamics. Each topic is treated in a way that allows for the independent study of a single subject or related chapter.

Finite Element Simulations Using ANSYS Second Edition

What’s New in the Second Edition:

• Introduces the newest methods in modeling and meshing for finite element analysis • Modifies ANSYS examples to comply with the newest version of ANSYS • Replaces many ANSYS examples used in the first edition with more general, comprehensive, and easy-to-follow examples • Adds more details to the theoretical material on the finite element • Provides increased coverage of finite element analysis for heat transfer topics • Presents open-ended, end-of-chapter problems tailored to serve as class projects Finite Element Simulations Using ANSYS, Second Edition functions as a fundamental reference for finite element analysis with ANSYS methods and procedures, as well as a guide for project and product analysis and design.

K24312 ISBN: 978-1-4822-6197-4

Esam M. Alawadhi

90000

9 781482 261974

K24312_Cover_PubGr.indd All Pages

8/10/15 11:35 AM