FINITE ELEMENT ANALYSIS IN DESIGN ~ Outline 17.4 17.2 | 17.3 17.4 17.5 17.6 17.7 17.8 Introduction Stiffness Matrix for Axial Elements Formulation of the Finite Element Method and Its Application to Trusses Beam and Frame Elements Properties of Two-Dimensional Elements ‘Triangular Element Plane Stress Case Studies Axisymmetric ElementCHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN m5 17.1. INTRODUCTION In real design problems, generally structures are composed of a large assemblage of vari- ‘ous members. In addition, the built-up structures or machines and their components involve complicated geometries, loadings, and material properties. Given these factors, it becomes apparent that the classical methods can no longer be used. For complex structures, the designer has to resort to more general approaches of analysis. The most widely used of these techniques is the finite element stiffness or displacement method. Unless otherwise specified, we refer to it as the finite element method (FEM). Finite element analysis (FEA) is a numerical approach and well suited to digital com- puters. The method is based on the formulations of a simultaneous set of algebraic equa- tions relating forces to corresponding displacements at discrete preselected points (called nodes) on the structure. These governing algebraic equations, also referred to as force- displacement relations, are expressed in matrix notation [1]. With the advent of high-speed, large-storage capacity digital computers, the finite element method gained great promi- nence throughout the industries in the solution of practical analysis and design problems of high complexity. The literature related to the FEA is extensive (for example, [2-20)). Numerous commercial FEA software programs are available, including some directed at the learning process. Most of the developments have now been coded into commercial programs. ‘The basic concept of the finite element approach is that the real structure can be dis- cretized by a finite number of elements, connected not only at their nodes but along the in- terelement boundaries as well. Usually, triangular or rectangular shapes of elements are used in the finite element method. Figure 17.1 depicts how a real structure is modeled using triangular element shapes. The types of elements commonly employed in structural ideal- ization are the truss, beam, two-dimensional elements, shell and plate bending, and three~ dimensional elements. A solid model of an aircraft structure created using beam, plate, and shell elements is shown in Figure 17.2. Note that, in the analysis of large structural systems, such as ships, multistory buildings, or aircraft, the storage capacity of even general- purpose programs can sometimes be overrun. It may be necessary to use the method of sub- structures that divides the original system into smaller units. Once the stiffness of each substructure has been determined, the analysis of the system follows the familiar procedure of matrix methods used in structural mechanics. Figure 17.1 Tapered plate bracket and its triangular finite Figure 17.2 _ Finite element model of fuselage element model. and a wing (4).716 PART2 © APPLICATIONS ‘To adequately treat the subject of the FEA would require a far more lengthy presenta- tion than could be justified here. Nevertheless, the subject is so important that any engineer ‘concerned with the analysis and design of members should have at least an understanding of FEA. The fundamentals presented can clearly indicate the potential of the FEA as well as its complexities. It can be covered as an option, used as a “teaser” for a student's advance study of the topic, or as a professional reference. For simplicity, only four basic structural elements are discussed here: the one-dimensional axial element or truss element, the beam. element or plane frame element, two-dimensional element, and the axisymmetric element, Sections 17.3, 17.5, and 17.8 present the formulation and general procedure for treating typical problems by the finite element method. Solutions of axial stress, plane stress, and axisymmetrical problems are demonstrated in various examples and case studies. 17.2 | STIFFNESS MATRIX FOR AXIAL ELEMENTS ‘An axial element, also called a truss bar or simply bar element, can be considered as the simplest form of structural finite element. An element of this type with length L, modulus of elasticity E, and cross-sectional area A, is denoted by e (Figure 17.3). The two ends or Joints or nodes are numbered 1 and 2, respectively. It is necessary to develop a set of two ‘equations in matrix form to relate the joint forces (F, and F2) to the joint displacements Gi, and ii). Direct EQUILIBRIUM METHOD The following derivation by the direct equilibrium approach is simple and clear. However, this method is practically applicable only for truss and frame elements. The equilibrium of the x-directed forces requires that F, = —F, (Figure 17.3). Because AE/L is the spring rate of the element, we have AE A= 7th ~in). This may be written in matrix form Ai) _AE[ 1 -1] fi Al _AE (17.43) tae La “fe, i (Fhe = Eelelitle (17.16) ‘The quantity [k]. is called the stiffness matrix of the element. or symbolically Figure 17.3 Axial (truss bar or bar) clement.CHAPTER 17 © Furr ELEMENT ANALYSIS IN DESIGN ENERGY METHOD ‘The energy technique is more general and powerful than the direct approach just discussed, especially for sophisticated types of finite elements, To employ this method, it is necessary to first define a displacement function for the element (Figure 17.3): Say tat (17.2) in which a, and a2 are constants. Clearly, Eq. (17.2) represents a linear displacement vari- ation along the x axis of the element. The axial displacements of joints 1 (at ¥ = 0) and 2 (at ¥ = L), respectively, are therefore =a, ea tab Solving the preceding expressions, ay = ity and a; = ~(ity ~ iia)/L. Carrying these into Eq, (17.2), we have i=(1-*)a4 2a (7. a= (1-F)mt+ pie (17.3) ‘Then, by Eq. (3.60), the strain is 1 Z pCi + i) (17.4) So, the element axial force: AE. ag F = (Ee) = (i tik) (175) ‘The strain energy in the element is obtained by substituting Eq. (17.5) into Bq, (5.10) in the form ey epee ) 2AE Applying Castigliano’s first theorem, Eq. (5.54), we obtain (17.6) 7-aU AE (ii, ~ 2) ity +) ‘The matrix forms of the preceding equations are the same as those given by Eqs. (17.1). ‘GLOBAL STIFFNESS MATRIX We now develop the global stiffness matrix for an element oriented arbitrarily in a two- dimensional plane. The local coordinates are chosen to conveniently represent the individ- ual element, whereas the global or reference coordinates are chosen to be convenient for the whole structure. We designate the local and global coordinates systems for an axial element by ¥, 5 and x, y, respectively (Figure 17.4). 7ne 1 Fy om Ft Vey uy Figure 17.4 Local (&, 7) and global (x,y) coordinates for a typical axial element e. Al forces and displacements have a positive sense. ‘The figure depicts a typical axial element e lying along the Z axis, which is oriented at an angle 6, measured counterclockwise, from the reference axis x. In the local coordinate system, each joint has an axial force F,, a transverse force F, an axial displacement i, and a transverse displacement 3. Referring to Figure 17.4, Eq. (17.1a) is expanded as 1 (17.78) Qees 10 -1 0 00 00 -10 10 00 00 [Phe = Eele(8be (17.76) Clearly, {5}. represents the nodal displacements in the local coordinate system. ‘We see from Figure 17.4 that the two local and global forces at joint 1 may be related by the following expressions: Fig = Fiz 0080 + Fiy sind Fiy = —Fix sin6 + Fy cos8 Similar expressions apply at joint 2. For brevity we designate ¢Scos6) ands =sind ‘Thus, the local and global forces are related in the matrix form Px cs 0 07 (Fx fy -sc 0 0||fy Ft 7.88) Pe 00 ¢s|)Fe 7 hy), £00 -» ellmJ, ‘or symbolically {Fe = (TMF le (17.86)CHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN In the foregoing, [7] is the coordinate transformation matrix: es 0 0 —-s ¢ 00 m=1§ 5 2° «79 00 -s ¢ and {F}_ represents the global nodal force matrix: Fs =}Fy Fle= yr (17.10) Fry), Inasmuch as the displacement transforms in the same manner as forces, we have cy uy Bh ery (7.41) fig 1" 5 a), wl or Ble = (Te 7am) Here, (5) is the global nodal displacements. Carrying Eqs. (17.11b) and (17.8b) into (17.7) leads to or (Fe = (Tele Note that the transformation matrix [7] is an orthogonal matrix; that is, its inverse is the same as its transpose: [T]~' = [7']", where the superscript T denotes the transpose. The global force-displacement relations for an element e is (Fhe = [ke{5}e (17.12) where (ke = (TY [17] (17.13) Finally, to evaluate the global stiffness matrix for the element, we substitute Eq. (17.9) and {ke from Eq. (17.7a) into Eq. (17.13): 2 es ct -es cs, —c pes AE} cs st cs —s?|_AE| 5? cy —s? Ue ee as | ead feet pete ee Symmetric s? This relationship shows that the element stiffness matrix depends on its dimensions, orien- tation, and material property. ng720 PART 2 © APPLICATIONS [AXIAL FORCE IN AN ELEMENT Reconsider the general case of an axial element oriented arbitrarily in 2 two-dimensional plane depicted in Figure 17-4. t canbe shown that equation for the axial force is expressed ‘in matrix form raw AE Aft - “| p= Sle 5 «7 7 L Uy — UL. > ‘This may be written for an element with nodes (jas follows: AE uy — Mi 2a) la) Myo. (17246) [A positive (negative) value obtained for Fy indicates thatthe element isin tension (com- pression). The axial stress inthe element is given by aij = Fi/A- 47.3. FORMULATION OF THE FINITE ELEMENT METHOD AND ITS APPLICATION TO TRUSSES Development ofthe governing equations appropriate to truss demonstrates the formulation of the structural stiffness method, or the finite element method. As noted previously, a truss ts an assemblage of axial elements that may be differently oriented. To derive truss equa- tions, the global element relations given by Eq, (17.13) must be assembled. The preceding Tends tothe following force-displacement relations for the entre truss, the system equations: {F} = [K18} 17.17) ‘The global nodal matrix (F) and the global stifness matrix (Kare (F)= DF (17.488) 7 tk = tee (17.188) T Here ¢ designates an element and 1 is the number of elements making up the truss, tis noted that [K] relates the global nodal force {F} to the global displacement {8} for the entire truss. METHOD OF ASSEMBLAGE OF THE VALUES OF [K] “The element stiffness matrices in Eq, (17.18) cannot be directly added together or supe, imposed, To carry out proper summation, convenient method isto label the columns and sao of each element stiffness matrix according to the displacement components asso vied with it In so doing, the truss stiffness matrix [K] is obtained simply by adding Ieo® froam the individual element stiffness matrix into their corresponding locations in ThisCHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN approach of assemblage of the element stiffness matrix is given in Case Study 17-1. An alternative way is to expand the (k], for each element to the order of the truss stiffness matrix by adding rows and columns of zeros. However, for the problem involving a large number of elements, it becomes tedious to apply this approach. PROCEDURE FOR SOLVING A PROBLEM We now illustrate the use of the equations developed in the preceding paragraphs. The gen- eral procedure for solving a structural problem by application of the finite-element method may be summarized as shown in Figure 17.5. This outline is better understood when applied to planar structures, as shown in the solution of the following sample problem. DEFINE ANALYSIS PROBLEM ‘Choose a finite element model plan the mesh forthe mode! 1 INPUT DATA Plan refined ‘Materials, node and element definition, |¢——— mesh for the ‘boundary conditions, loads model 1 FORM ELEMENT (dl. Calculate element stiffness matrix [Saeleseneg eera | FORM SYSTEM [] Assemble elements [k], to form the system stiffness matrix ¥ FORM SYSTEM (F) AND (8) Apply boundary displacement and force conditions r ‘COMPUTE DISPLACEMENTS ‘Solve the system equations (F) = (8108) forthe displacements (6) =F) i ‘COMPUTE STRESSES Calculate stresses (oF forces) in elements ¥ EVALUATE RESULTS Ts refined mesh for the model required? PRESENT RESULTS Figure 17.5 Finite element block diagram [13] 721722 PART2 © APPLICATIONS Case Study 17-11 A three-bar truss 123 (Figure 17.6a) is subjected to a horizontal force P acting at joint 2. By using the finite el- ement method, analyze the truss and calculate the re- quired cross-sectional area of each member. Assumptions: All members will have the same length Land axial rigidity AE. Use a factor of safety of = 1.5 on yielding. Given: 5,=240MPa, P = 200KN Solution: The reactions are noted in Figure 17.6a. The node numbering is arbitrary for each element. Input Data, At each node there are two displace- ments and two nodal force components (Figure 17.66). Recall that 0 is measured counterclockwise from the pos- itive x axis to each element (Table 17.1). Inasmuch as the terms in [k, involve c, s*, and cs, achange in angle from 6 to6 +7, causing both c and s to change sign, does not affect the signs of the terms in the stiffness matrix. For Table 17.1 Data for the truss of Figure 17.6 Hemant 9 ¢ os @ = 1 eo1 o 4 o 0 2 1-1/2 3/24 W484 3 oor 23524 3/48 Figure 17.6 Basic plane truss. ANALYSIS AND DESIGN OF A TRUSS example, in the case of member 3, 6 = 60° if measured counterclockwise at node 1 or 240° if measured counter. clockwise at node 3. However, by substituting into Eq, (17.14), [ele remains unchanged, Element Stiffness Matrix. Using Eq. (17.14) and ‘Table 17.1, we have for the elements 1, 2, and 3, respec- tively, mom 10 -1 07m oo 0 0|™ i -10 10; 00 0 Oly a 1 -v3 -1 0 V3 , AE|-J3 3) V3 -3. |» Par} -1 v3 1 -v3}us 3 3 V3 3 Iu ms 1 VB -t -v3T AE] V3 30 -v3_ 3 | Us 4 _ -1 -V3 1 V3 |w -¥3 3 VB 3 ds ‘Note that the column and row of each stifness matrix are labeled according to the nodal displacements associated with them, aLCHAPTER 17. © FINITE ELEMENT ANALYSIS IN DESIGN 723 Case Study (CONTINUED) System Stiffness Matrix. There are a total of six components of displacement forthe truss before boundary con- straints are imposed, Therefore, thé order of the truss stiffness matrix must be 6 x 6. Subsequent to addition of the terms from each clement stiffness matrices into their corresponding locations in [K], we readily obtain the global stiffness ‘matrix for the truss: mm Cs us % 441 04+V3 -4 0 -1 v3 Jum O+v3 043 0 0 3 3 oy i=42| -4 0 441 0-V3 0-1 V3 | (a) “a| 0 0 o-v3 043 V3 -3 fu -1 0 -v3 =I v3 141 V3=v3 | ws V3 3 Va -3) V3-v3 343 Ju ‘System Force and Displacement Matrices. Accounting for the applied load and support constraints, with ref- erence to Figure 17.6, the truss nodal force matrix is. Fix Ry A] |o ae |_]P edt = Ye ® Px Ru Psy Roy Similarly, accounting for the support conditions the truss nodal displacement matrix, My 0 wa] _ Jus =r ayG © uw} |o w) bo Displacements. Substituting Eqs. (a), (b), and (c) into Eq. (17.18), the truss force-displacement relations are siven by R 5S V3 -4 0 -1 -v37 0 0 v3 3 0 0 -v3 -3 | Ju P 4 0 5 =v -1 V3 R o 0 -¥8 3 v3 -3 |jo Ry -1 -v3 -1 v3 2 0 0 Ry —v3 3 V3 -3 0 6 0 ‘To determine v and us, only the part of Eqs. (d) relating Solving preceding equations simultaneously or by matrix to these displacements is considered. We then have inversion, the nodal displacements are obtained: 0) _ AE[3 0} (us nj _ 40 fs 07/0 set (0) is (}-4 0 5} lu uJ SAE LO 3) |p)” 5Ak | (continued)724 PART2 © APPLICATIONS Case Study (CONCLUDED) Reactions. The values of v; and wp are used to de- acl 1 3] f- termine reaction forces from Eq. (A) as follows: Ps=F|-3 R Ro -4 R{_ ae] 0 -v3|fuy_P ]- ry =42[1 3] Rx ~ aE | -v3 -1 | wf 5 BT [2 2 Jlol Roy 3 3 ‘The results may be verified by applying the equations of Stresses in Elements. Dividing the foregoing equilibrium to the free-body diagram of the entire truss, element forces by the cross-sectional area, we have Figure 17.6a. 012 =4P/SA, om = 2P/5A, and ois Axial Forces in Elements. Using Eqs. (17-16) Required Cross-Sectional Areas of Elements. and (e) and Table 17.1, we obtain ‘The allowable stress is oy = 240/1.5 = 160 MPa. We AE ae then have A; = 0.8(200 x 10°)/160 = 1000 mm?, A> = roe En of 8} 500 mm?, and Ay = any area. 147.4 BEAM AND FRAME ELEMENTS Here, we formulate stiffness matrices for flexural or beam elements and axial-flexural or plane frame elements. Consider first an initially straight beam element of constant flexural rigidity E/ and length L, as depicted in Figure 17.7. Such an element has a transverse deflection i and a slope @ at each end or node. Corresponding to these displacements, a transverse shear force F, and a bending moment M act at each node. The deflected config- uration of the beam element is shown in Figure 17.8. The linearly elastic behavior of a beam element is governed according to Eq. (4.16c) as d‘v/dx* = 0. The right-hand side of this equation is 0 because in the formulation of the stiffness matrix equations, we assume no loading between nodes. In the elements where there is a distributed load, the equivalent nodal load components are used. The solution is taken to be a cubic polynomial function of x, Da, + az task +agk? @ Fay Figure 17.7 Beam clement; all forces Figure 17.8 Deformed beam element. and displacements have a positive sense.CHAPTER 17 ° Fuvrre ELeMEnt ANALYSIS IN DESIGN ‘The constant values of a are obtained by using the conditions at both ends. The stiffness matrix can again be obtained by the procedure discussed in Section 17.2. Itcan be verified that [8] the nodal force-displacement relations in the matrix form are fy 12 6L My| _ EL aL? -12 -6L 12 M3, Symmetric or symbolically {Fle = Eleldle ‘The matrix (represents the force and moment components. Equation (17.19b) defines the stiffness matrix [k], for a beam element lying along a local coordinate axis x. Having developed the stiffness matrix, formulation and solution of problems involving beam elements proceeds as discussed in Section 17.3. 6L 2? (17.198) (17.196) 725 Determining Displacements and Forces in a Statically Indeterminate Beam A propped cantilevered beam of flexural rigidity EY is subjected to end load P as shown in Fig- ure 17.9a. Using the finite element method, find (a) The nodal displacements. (b) The nodal forces and moments, @ Lats L ¥ a © ig at EPL © ae Figure 17.9 Example 17.2: (2) load diagram; (6) shear diagram; (c) moment diagram. EXAMPLE 17.1726 PART2 © APPLICATIONS Solution: We discretize the beam into elements with nodes 1, , and 3, as shown in Figure 17.9 By Eq, (17.19), e ny & wm & 2 6b -12 6L Jw EI 4L2 —6L 207 | & oh = Ts 12 -6L | ‘symmetric ao le mn mB 2 6b -12 GL» EL 4Lt —6L 20? | & Dae 12 -6L | ws ‘Symmetric av? | 65 (a) The global stiffness matrix of the beam can now be assembled: [K] = [Kk]; + (tls. The ‘governing equations for the beam are then Py 1 6L -122 SL 0 0 7F/u mM 42 -6L 202 0 0 | Jo Py 2% 0 12 6b | Ju M aL? -6L 20? | |e ae By 12-6 | Jos Ms Symmetric av J la (F} = [KM8} (17.20b) ‘The boundary conditions are v» = 0, 6 = 0, and vs = 0. Partitioning the first, second, and fourth of these equations associated with the unknown displacements: oF: 12 6L 6L 1 0 6L 21? 81? | (a Solving for nodal displacements, we obtain TPL 3PL? per aE” m (b) Introducing these equations into Eq. (17.20a), after multiplying, the nodal forces and ‘moments are found as, Note that My and Mo are 0, since no reactive moments are present on the beam at nodes Land 2.CHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN Comments: In general, itis necessary to determine the local nodal forces and moments associated with each element to analyze the entire structure, For the case under consideration, it may readily be observed from a free-body diagram of element | that (Mz); = —PL. Hence, we obtain the shear and moment diagrams for the beam as shown in Figures 17.9b and 17.9¢, respectively. 727 ARBITRARILY ORIENTED BEAM ELEMENT In plane frame structures, the beam elements are no longer horizontal. They can be oriented in a two-dimensional plane as shown in Figure 17.10. So, it is necessary to expand [k]. to allow for the displacements transforming into u and v displacements in the global system. The moments are unaffected. Referring to the figure, the global force and displacement matrices are, respectively, Fis Fy M Fax Fry MJ. (Fhe = (17.21) rn 4 aJH Be= yh 721) » 3, Following a procedure similar to that described in Section17.2, the coordinate transforma- tion matrix now becomes (T= (17.22) cocode cocoons coorce etnoce enncce mH ococe where, as before, ¢ = cos@ and s = sin 9. Substituting [7] from Eq. (17.22) and [k]e from Eq. (17.19) into Eq. (17.13), the global stifiness matrix is formed: 12s? -12es -6Ls ~12s? 12cs —6Ls 12 6Le Wes. —12e? 6Le El 4? 6Ls —6Le 2L? Ue= Tr 12s? —12cs 6Ls si 12? -6Le ‘Symmetric 4L?728 PART2 @ APPLICATIONS Fry Mt Figure 17.10 Global forces and Figure 17.11 Local forces acting on arbitrarily oriented displacements acting on an arbitrarily axiallexural elements. oriented beam element. ARBITRARILY ORIENTED AXIAL-FLEXURAL BEAM OR FRAME ELEMENT When a horizontal axial element (Figure 17.3) and a horizontal beam element (Fig- ure 17.7) are combined, we obtain the axial-flexural beam element. In this case, the solu- tion for the axial displacements and the transverse deflections and rotations can be carried out separately and independently. Local nodal forces acting on an axially flexural beam or frame clement oriented in the two-dimensional plane with an angle 6 with the x axis is shown in Figure 17.11, For this element, the stiffness matrix must undergo the routine co- ordinate transformation procedure described previously. In so doing, we obtain the global stiffness matrix for the element that contains the axial force, shear force, and bending mo- ‘ment effects [7]: ace the (4B Js - (104+ Fe) -(4- Fe arate Be -(a- Sr )es tnt ar a, 2 acs le 4. Symmetic (17.24, The global force and displacements are again given by Eqs. (17.21). Analysis and de- sign of rigid-jointed frameworks can be undertaken by applying Eqs. (17.23) or (17.24). From the latter equation, we observe that the element stiffness matrix of a frame in general is a function of E, A, L, 1, and the angle of orientation 9 of the element with respect to the global-coordinate axes. With the element stiffness matrix developed, formulation and solution of a frame problem proceeds as discussed in Section 17.3. The following example illustrates the procedure.cHaPTE ER 17 Funire ELEMENT ANALYSIS IN DESIGN 729 Determination of Displacements in a Frame A planar rectangular frame 1234 is fixed at both supports 1 and 4 (Figure 17.12). The load on the frame consists of a horizontal force P acting at joint 2 and a moment M applied at joint 3. By the finite element analysis, calculate the nodal displacements. L a 3 o Tl © || + Tu Figure 17.12 Example 17.3. Plane frame. Table 17.2 Data for the frame of Figure 17.12 Blement 1 2 3 a 30° ° 210" ¢ ° 1 o : 1 ° a 2/L? 04 02 04 oy R 6 2 EL sx 5x10 5x10 = 5 ft, E =30 x 10° psi, and A = 5 in? for all elements; 60 in for element 2 Solution: ‘The global coordinate axes xy are indicated in Figure 17.12. Through the use of Eq, (17.24) and Table 17.2, the element stiffness matrices are obtained as mom 8 04 0 -12 O80 =12. 0 480 Hh =500)) 4 9 12 0 -5 0 -12 0 240 tv 50.50 0 02 6 0 6 240 th=S00)} ¢ 9 9 0 -02 -6 0 6 120 uy 04 0 12 Oso 2 0 480 Wh =500) 94 9-12 0 -5 0 12 0 240 ~04 0 2 04 =12 0 240 2 0 480 w 4 ees Se v% 4 P55 4 EXAMPLE 17.2PART2 © APPLICATIONS ‘We superpose the element stiffness matrices and apply the boundary conditions: w= =8=0, m=%=%=0 ‘at nodes 1 and 4. This leads to the following reduced set of equations: tb wm 4000. 54.0 12 -5 0 OF fu 0 52.6 0 -02 6 ||u Oe 20 0 120 | Ja Of 54.0 12 | )us 0 | 52-6] |u 2000. Symmetric 720 tos Solving, the nodal deflections and rotations are a 18.208 in, w 0.582 in, | _|0271] 9-1) mad asf =) 17.408f 0° in % 0.582 in, % —0.248 rad ‘The negative sign indicates a downward displacement or clockwise rotation. 17.5 PROPERTIES OF TWO-DIMENSIONAL ELEMENTS So far we have dealt with only line elements connected at common nodes, forming trusses and frames. In this section, attention is directed toward the properties of two-dimensional finite elements of an isotropic elastic structure and general formulation of the finite element ‘method for plane structures. To begin with, the plate shown in Figure 17.13a is discretized, as depicted in Figure 17.13b. The finite elements are connected not only at their nodes but along the interelement boundaries. All formulations are based on a counterclockwise la~ beling of the nodes i,j, and m. The simplest constant strain triangular finite element is used to clearly demonstrate the basic formulative method. The nodal displacements, represented by w and v in the x and y directions, respectively, are the primary unknowns, PoE ZZ | Figure 17.13 Plate in tension: (a) before and (b) after division into finite elements,CHAPTER 17. © FINITE ELEMENT ANALYSIS IN DESIGN DISPLACEMENT FUNCTIONS Consider a typical finite element e with nodes i, j, and m (Fig. 17.13b). The nodal dis- placements are expressed in the convenient matrix form: {bhe= (17.25) ‘The displacement functions, describing the displacements at any point within the element, (fle is represented by Uke u(x, ») (reat 17.260 (fle =I }e (17.266) In the foregoing, the matrix [/V] is a function of position, to be obtained in the next section. STRAIN, STRESS, AND DISPLACEMENT MATRICES ‘The strain and stress are defined in terms of displacement functions. The strain matrix may be written: ou ie ax au {ele = [ by | = y (17.27) Yale ay. aw ax * ay. or {ele = [BN {S}e (17.27b) in which [B] is also obtained in the next section. Ina like manner, the stresses throughout the element are, by Hooke’s law, efi» 0 & {o}e 3 [; 1 0 le] (17.28a) “Loo a-v/2J bed. {o}e =([Dile}e (17.286) or Clearly, the elasticity matrix is ge fi» 0 (Di= vt 0 (17.298) 1-19 9 (= v)/2 731732 PaRT2 © APPLICATIONS Table 17.3 Elastic constants for two-dimensional problems Quantity Plane strain Plane stress za-¥) T+ya-m a Da v Ds In general, we write Dn 0 1 0 | (17.296) 0 Ds Recall from Section 3.18 that two-dimensional problems are of two classes: plane stress and plane strain. The constants 4, Dr, and D3; for a plane problem are defined in Table 17.3 [9, 10]. FINITE ELEMENT METHOD FOR TWO-DIMENSIONAL PROBLEMS ‘Through the use of the principle of minimum potential energy, we can develop the ex- pressions for a plane stress and plane strain element. For this purpose, the total potential ‘energy IT (see Section 5.8) is expressed in terms of two-dimensional element properties. ‘Then, the minimizing condition, 9T1/8(6}- = 0, results in [10] {Fhe = UlelS}e (17.12) This is of the same form as obtained in Section 17.2 and (5}- represents element nodal displacement matrix. However, the element stifiness matrix (k]e and element nodal force matrix (F'}_ are now given by a. = frerrouaiev (17.30) tnie= footie 730 where p= boundary surface forces per unit area 5 = boundary surface over which the forces p act V= volume of the element T = transpose of a matrixCHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN ‘We next assemble the element stiffness and nodal force matrices. This gives the fol- lowing global governing equations for the entire member, the system equations: {F} [K1(8} 17.17) where (=). 1K) = Doe (17.18) T T as before. Now, n represents the number of finite elements making up the member. Note that, in the preceding formulations, the finite element stiffness matrix has been derived for a general orientation of global coordinates (x, y). Equation (17.17) is therefore applicable to all elements. Hence, no transformation from local to global equations is necessary. The ‘general procedure for solving a problem by the finite element method is already shown in Figure 17.5. This outline is better understood when applied to a triangular element in the section to follow. 733 17.6 TRIANGULAR ELEMENT We now develop the basic constant strain triangular (CST) plane stress and strain element. Boundaries of iegularly shaped members can be closely approximated and the expressions related to the triangular elements are simple. The treatment given here is brief. Various types of two-dimensional finite elements yield better solutions. Exam- ples include linear strain triangular (LST) elements, triangular elements with addi- tional side and interior nodes, rectangular elements with comer nodes, rectangular elements with additional side modes, and so on [4, 8, 14]. The LST element has six nodes: usual corner nodes and three additional nodes conveniently located at the mid- points of the sides. Hence, the element has 12 unknown displacements. The procedures for development of the equations for the LST element follow the same steps as that of the CST element. DISPLACEMENT FUNCTION Consider the triangular finite element i, j, m shown in Figure 17.14. The nodal displace- ment matrix (8). is given by Eq. (17.25). The displacements w and v throughout the clement can be assumed in the following linear form: em (ete | (17.32) v(x, y) [ag + Osx + Chey where the « represents constants. The foregoing expressions ensure that the compatibil- ity of displacements on the boundaries of adjacent elements are satisfied.734 PART2 © APPLICATIONS oa 4, Figure 17.14 Basic triangular element. The nodal displacements of the element are Mj =O + 2X} + O35 Uj = Og + 5x; + WY; uy = a + px) + 059, YS a4 bast; +069) Um A + O2%m + 13 Yon Um = 04 + O5Xm + O6Ym Solving these equations gives [7, 9] (: 1 [: 4 Om) (us b= |b by bm |} uy as) 7A Le cj cmd atm Gn} (% bn “| cm J bn a) Pag as = so] br by a6, ag ‘The quantity A represents the area of the triangle: A= 5Lei() — Ym) +2) 0m — 1) + Xm — YD and = X]Ym —Yj%m_—— = Yim — Xin b= Yi Im bj = Yn Yi 1 = Xm — 4 oy =X Substituting Eq, (a) into Eq, (17:32), the displacement function is provided by a [" 0 Nj; 0 Mm °| i Om ON 0 Nal pup Te (a) (17.33) (17.34) (17.38)CHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN in which 1 N= sala thx tay) aya +byx + oy) 17.36) Na = Fyn + bax + ny) ‘The strain matrix is obtained by carrying Eq, (17.35) into Eq. (17.27a): & 1 fh 0 b 0 bn 0 ft =a] 9 co 0 G 0 cn) Be (17.37) Yodo ce be cy by cm Bm Introducing Eq, (17.37) into Eq. (17.27b), we have 1 [2% Of b OF bn 0 wm=3[8 ci 0 Gi 0 al (17.38) 2A Gb cy byl Cm Bm or [8] = [B51] (97.386) where 1f% ° ipa ° 1 [em 0 toa= 35° <| m= 358 «|: [Brel al? «| 17.39) Alb, Ale by, 2Al oben Clearly matrix [B] depends only on the element nodal coordinates, as seen from Eq. (17.34). Hence, the strain (and stress) is observed to be constant and, as already noted, the element of Figure 17.14 is called a constant strain triangle. ‘THE STIFFNESS MATRIX For an element of constant thickness ¢ the stiffness matrix can be obtained from Eq, (17.30) as follows: (ke = [BILD IBA (17.40) ‘This equation is assembled together the elasticity matrix [D] and [B] given by Eqs. (17.29) and (17.38). Expanding the resulting expression, the stiffness matrix is usually written in a partitioned form of order 6 x 6. We point out that the element stiffness matrix is generally developed in most computer programs by performing the matrix triple products 735736 PART2 © APPLICATIONS Table 17.4 Nodal forces Fj and Fn of 2 CST element of thickness t due to some common loadings [10] ‘A. Linear load p(y) per unit area 2 At B= Sep) + pm) nt Eble Fa = pm + PD Fa 1 pple Loe y 7 Ty 7 1 fnoa 2 7 Ae ie ae[ 2689 te” i ile” el) 5-2 (Lo.- ae ’ =a (== dy) GE tom f shown by Eq. (17.40). The explicit form of the stiffness matrix is rather lengthy and given in the specific publications on the subject, see, for example, [6]. ELEMENT NobaL Forces DUE To SURFACE LOADING ‘The nodal force attributable to applied external loading may be obtained either by evaluat- ing the static resultants or applying Eq. (17.31). Some examples, found by the former approach, are given in Table 17.4. An expanded form of Eq. (17.40), together with those expressions given for the nodal forces, characterizes the CST element. The unknown dis- placements, strains, and stresses may now be determined applying the general outline given in Figure 17.5. The basic procedure employed in the finite method using CST or any other element is illustrated in the next section. 17.7 PLANE STRESS CASE STUDIES Here, we present four case studies limited to plane stress situations and CST finite ele- ‘ments, A plate under tension, a deep beam or plate in pure bending, a plate with a hole sub- {jected to an axial loading, and a disk carrying concentrated diametral compression are theCHAPTER 17 @ FINITE ELEMENT ANALYSIS IN DESIGN members analyzed. There are very few elasticity or “exact” solutions to two-dimensional problems, especially for any but the simplest forms. As will become evident from the fol- lowing discussion, the designer and stress analyst can reach a very accurate solution by applying proper techniques and'modeling. Accuracy is usually limited by the willingness to model all the significant features of the problem and pursue the analysis until conver- gence is reached. It should be mentioned that an “exact” solution is unattainable using the finite ele- ment method, and we seek instead an acceptable solution. The goal is then the establish- ment of a finite element that ensures convergence to the exact solution. The literature contains many comparisons among the various elements. The efficiency of a finite element solution can, in certain situations, be enhanced using a “mix” of elements. A denser mesh, for instance, within a region of severely changing or localized stress may save much time and effort. 737 IN TENSION Given: p=4ksi, £=30x 10° psi, rating L=20in, —h=10in. 03, ‘The plate is divided into two CST Assumptior elements. T EHP 10 ; -—z—__| #20 in —|? @ o ure 17.15 Cantilever plate (a) before and (b) after being discretized. Case Study 17-2] ANALYSIS OF STRESSES AND DISPLACEMENTS IN A PLATE A cantilever plate of depth h, length L, and thickness Solution: The discretized plate is depicted in Fig- # supports a uniaxial tension load p as shown in ure 17.15b. The origin of coordinates is placed at node 1, Figure 17.15a, Outline the determination of deflections, for convenience; however, it may be located at any point strains, and stresses. in the xy plane. The area of each element is eee hb = 5(00)20) = 100 ‘The statically equivalent forces at nodes 2 and 3, 4(10 x 4)/2 = 1Okips, are shown in the igre. For plane stress, elasticity matrix [D] is given by Eq. (17.29a). Faq = 1 kis Fac= lkips (continued)738 PART2 © APPLICATIONS Case Study (CONCLUDED) Stiffness Matrix. For element a, on assigning i =1, j =3, and m = 4, Eq, (17.34) gives by = ys ye = 10-10 =0 bs = —y1 =10-0 bb=y— 0-10 @ cy) = x4 — x3 = 0-20 =a 0-0=0 ce =a — x) =20-0=20 Substitution of these and the given data into Eq, (17.40), after performing the matrix multiplications, results in stiffness matrix [k].. Similarly, for element b, assignment off = 1, j =2,and m =3 into Eq, (17.34) leads t0 b=n-y=0-10 by -y =10-0=10 b=n—m=0-0=0 a 4 — x. = 20-20=0 =m - 5 = 0-20 o=m—m =20-0=20 and [k], is determined. The displacements wz, v2, and ‘ug, ve are not involved in elements a and b, respectively. So, before summing [A], and (kl, to form the system matrix, rows and columns of zeros must be added to each element ‘matrix to account for the absence ofthese displacements, as ‘mentioned in Section 17.3. Finally, superimposition of the resulting matrices gives the system matrix [X]. Nodal Displacements. The boundary conditions are uy = v1 = us = v4 = 0. The force-displacement rela- tionship of the system is, Rc 0 Ry 0 10 we 0 » 0 f as 0 v5 Re 0 Ry 0 | Next, to compare the quantities involved, we introduce the results without going through the computation of the {K1. It cam be verified [8] that the preceding derivations yield 10 48 00-28 14.7 fur of _is7s| 0 87 12-80] Je 10f~ O91 | -28 12 48 -26 | Jus 0 4 -80 -26 87 | by Solving, mo) (2.4383 ve | _ ]0.0163] 4 5 [= }2.6548 (1 ws} lo.a261 Stresses. For element a, carrying Eqs. (a) and (b) into (17.37), we obtain the strain matrix {¢},. Equation (17.28), [D]fe}o, then result in oe 4020 oy | = 41204} psi ty), (9.6 Element b is treated in a like manner. Comments: Due to constant x-directed stress of 4000 psi applied on the edge of the plate, the normal stress is expected to be about 4000 psi in the element a (or 6). The foregoing result for 0, is therefore quite good. Interestingly, the support of the element a at nodes 1 and 4 causes a relatively high stress of a, = 1204 psi. Also note that the value of shear stress ‘Tay is negligibly small, as anticipated. The effect of ele- ‘ment size on solution accuracy will be demonstrated in the next example.CHAPTER 17. & Fire ELEMENT ANALYSIS IN DESIGN 739 Case Study 17-3 | A deep beam (or plate) is loaded by couples M, acting about its axis at the ends as shown in Figure 17.16a. Determine the stresses and displacements, using the meth- ds of theory of elasticity and finite element. Figure 17.16 Beam in pure bending. L=162mm, h=508mm, thickness 54mm, p=6895kPa, = 207 GPa, ‘The weight of the member is negligible. Elasticity or “Exact” Solution To begin with we replace the end moments with the stati- cally equivalent load per unit area p = Mh/T, as depicted in Figure17.16b, The stresses are then 0 (17.41) By Eqs. (2.6) and (3.612), au__ yp au _wyp ay, aw ox Eh’ By ER’ Bx By It.can be shown that, satisfying the conditions u(0, 0) = 0 and u(L., 0) = 0, we have Pcl a yt v ER” ty’) (17.42) Introducing the data into Eqs. (17.41) and (17.42) gives 1 =~ 95508)? (0.0762, ~0.0508) = 2.54 x 10°* m Oman = 6895 kPa o 10(0,0762, 0) = 1.904 x 10-6 m ANALYSIS OF A DEEP BEAM IN PURE BENDING BY THE ELASTICITY AND FINITE ELEMENT METHODS Finite Element Solution Based on symmetry and antisymmetry, only a quadrant (Gndictated by the shaded part in the figure) of the beam need be analyzed. Boundary Conditions. The discretized quarter- plate is composed of 12 triangular elements (Fig- ure 17.17a). Since no deformation takes place along the x and y axes, the boundary conditions are My Hay = uy Ug = uy =U =0, v4 =0 Nodal Forces. Applying the equation in Case A of ‘Table 17.4 to Figure 17.16b and inserting data given, we have 0.0254 x 0.0254 6 = 18535 N 0.0254 x 0.0254 6 0.0254 x 0.0254 6 For (2 x 6895 + 3447.5) Fis (2 x 3447.5 + 6895) (2 x 3447.5 +0) = 2204N ‘The other nodal forces are 0. Results [10]. The values of the stresses, obtained by a procedure identical to that ofthe preceding example, are shown in Figure 17.17b, Note the considerable differ- cence between the exact solution, Eq. (c), and that resulting from the coarse mesh arrangement used. To illustrate the effect of element size and orientation, the results corre- sponding to different grid configurations are given in Fig- ures 17.17c and 17.174. The displacements corresponding to Figures 17.17b through 17.17¢ and the exact solution are listed in Table 17.5 Comments: Figures 17.17b and 17.17c show the effect of clement orientation for the same number of elements (continued)740 PART 2 © APPLICATIONS Case Study (CONCLUDED) estate 25.4 fo 254 Gals elements ) Figure 17.17 Table 17.5 Displacements Number Disblacement (10 m) Case of nodes ta te Figure 17.17 2 1547 2133 Figure 17.176 n 174s 2.062 Figure 17.174 15 1s 1976 Exact solution - 1.905 2540 16 elements @ Effect of element size and orientation on stress in beam shown in Figure 17.16, Note: Stress (megapascals) obtained for an element is assigned to the centroid. and node locations. Observe that elements having large differences between their sides, or so-called weak ele- ments, such as Figure 17.174, yield less accurate results; even the number of nodes is larger than those of Fig- ures 17.17b and 17.17c. Using equilateral or nearly equi- lateral, well-formed elements of finer mesh gives solu- tions approaching the exact values. Case Study 17-4 | A thin plate containing a small circular hole of radius a is subjected to uniform tensile load of intensity 0, at its edges, as shown in Figure 17.18a, Apply the finite element ANALYSIS OF STRESS CONCENTRATION IN A PLATE WITH A HOLE SUBJECTED TO UNIAXIAL TENSION analysis to determine the theoretical stress concentration factor.CHAPTER 17 © Finrre ELEMENT ANALYSIS IN DESIGN 7a Case Study (CONCLUDED) i a4 hea, = 6ksi | Braet = : a solution : 7 i oh demens [>| % 2 Ll 4 + = ofA} (SS I h Figure 17.18 Solution: Owing to the symmetry, only any one- quarter of the plate need be analyzed, Figure 17.18b. ‘The solution for the case in which the quarter plate dis- cretized to contain 202 CST elements is given in [4]. ‘The roller boundary conditions are also indicated in (@) Circular hole in a plate under uniaxial tension (b) One-quadtrant plate model (c) Uniaxial stress (x, distribution. the figure. The values of the normal edge stress 0, ob- tained by the finite element method and the theory of elasticity {9}, are plotted in Figure 17.18e for compari- son. We see from the figure that the agreement is rea- sonably good, The stress concentration factor for oy is K, 305/00 =3.742 PART2 © APPLICATIONS Case Study 17-5 | A circular disk of constant thickness ¢ is subjected to a pair of compression forces, as shown in Figure 17.19. Demonstrate the accuracy of the finite element solution using well-formed (ie., nearly equilateral) CST elements Given: The radius a = 81 and v = 0.3. Assumption: The disk is in a state of plane stress * 7 Vr Pp ® Figure 17.19 diametral load. (b) One-quarter circular disk mode! with a coarse mesh, (a) Circular thin disk subjected to @ ANALYSIS OF STRESS DISTRIBUTION IN A CIRCULAR Disk WITH A DIAMETRAL LOAD Solution: Because of symmetry, only a quarter of the disk need be considered (Figure 17.19b). The exact ana- lytical as well as 30 and 109 CST element solutions of the normal stress are available [11, 12]. A plot of these results is shown in Figure 17.20. Note the considerable difference between the exact solution and the results for 30 elements but the particularly good agreement between the results for 109 elements and the exact solution. Exact solution 005 oT a 3 4 3 6 TOR Figure 17.20 Normal stress (o,) distribution along the xaxis of the circular disk 1 17.8 AXISYMMETRIC ELEMENT This section can provide only a brief introduction to the finite element method for determi- nation of strain and stress in an axisymmetrically loaded member, formed as a solid of revo- lution with material properties, support conditions, and loading, all of which are symmetri- cal about the z axis. The finite elements of the body of revolution or rings are used to discretize the axisymmetric member. We use a simple element ¢ of triangular cross section and cylindrical coordinates (r, 8, z), shown in Figure 17.21. Note that a node now is a circle. Therefore, the elemental volume dV appearing in Eq, (17.30) is the volume of the ring ele- ment (i.¢., 2rr dr dz). Although the element lies in three-dimensional space, any of its arbi- trarily selected vertical cross sections is a plane triangle. As observed earlier, no tangential displacement can exist in the axisymmetrical member, v = displacement u and axial displacement w in a (rz) plane are present, the expressions for dis placements for plane strain may readily be extended to the axisymmetric analysis. | Inasmuch as only the radialCHAPTER 17 @ FINITE ELEMENT ANALYSIS IN DESIGN Figure 17.21 Symmetric solid finite element with triangular cross section. Consider the triangular finite element i, j, m depicted in Figure 17.21. Properties be- longing to the element are denoted by e. Referring to Eqs. (17.27) and (16.3), we define the strain-displacement matrix of the form au wr & aw & he= ye | =] % 743) Yre r du | aw a Orie ‘The nodal displacement matrix for the element is represented by uy uy e= 4 (17.44) Lid wm Je ‘The element strain-nodal displacement matrix may be expressed (ele = [B(B}e (17.45) In the preceding, [B] is yet to be determined. The stress-strain relationship is as follows: a; (ole=] 2%} =[Ditele (17.48) % tide ‘The elasticity matrix is given by l-v v 0 E v 0 = aypa>m ney 0 17.47) ‘Symmetric (1 = 2v)/2, 743744 PART 2 © APPLICATIONS Following a procedure basically identical to that described in Section 17.6, it can be shown that [10] b, bj bm 0 0 0 _1]0 0 0 a G om (l= Fad dj dn 0 0 0 (17.48) Gj mB Oj Dm ‘The quantity A represents the area of the triangle defined by Eq. (17.33) and bi = 2) im by sin m=") g=nntn (1749) Oy dy = “ ++ (n =i, j,m) r r Note that the matrix [B] includes the coordinates r and z. Therefore, the strains are not con- stant, as in the case of plane strain and plane stress. The stiffness matrix can now be ob- tained through the use of Eq. (17.30): [Ke = f [By (DB av (17.50) Vv This may be rewritten, after integrating along the circumferential or ring boundary, as follows: [ke = 2 i r{B)" [D](B] dr dz 7.51) where the matrices [D] and [B] are defined by Eqs. (17.47) and (17.48), respectively. Let the radial and axial components of force per unit length be denoted by f, and f,, respectively, of the circumferential boundary of a node of a radius r. The total radial and axial nodal forces are then FE =2arf, Fe = 2arf, (17.52) Equations (17.51) and (17.52) characterize the triangular element for an axisymmetric problem. These are introduced into Eq. (17.18) and subsequently into Eq. (17.17) to eval- uate the nodal displacements by satisfying the prescribed boundary conditions. The strains and stresses are then found using Eqs. (17.45) and (17.46). Most of what was learned in studying the two-dimensional problems in Sections 17.5 through 17.7 also applies to axisymmettical problems. Case Study 17-6 | A thick-walled pipe with the inner and outer radii a and b, the finite element and exact approaches of analysis, across respectively, is subjected to extemal pressure pp (Fig- the wall of the cylinder. ture 17.22a). Sketch the stress distribution, as obtained by STRESSES IN A THICK-WALLED HIGH-PRESSURE STEEL CYLINDER,CHAPTER 17 © FintTe ELEMENT ANALYSIS in DESIGN 748 Case Study (CONCLUDED) Pe Figure 17.22 (a) Thickwalled cfinder under external pressure and ‘modeling of ts section 4, (b) Distribution of tangential and radial stresses in the lider for b = 28 Given: b= 2a, 03 Solution: Only a slice of the cylinder needs be analyzed (Figure 17.222). A total of 20 triangular elements, with gradually decreasing size toward the loaded region, are used. The boundary conditions for the mesh system may be represented by discrete rollers at the nodes along the top and bottom faces of the slice; that is, at these nodes, the w nodal displacements are to be constrained to vanish, Using a general-purpose finite element program [4 14-20}, the tangential and radial stresses, 09 and o, are obtained. The nondimensional results are plotted in Fig- ure 17.22b. Also shown are the exact results given by Eqs. (16.18a) and (16.18b). We see excellent agreement ‘between the solutions determined by the two methods. REFERENCES 1. Gere, J. M., and W. Weaver Ir. Matrix Methods for Engineers. New York: Van Nostrand, 1966. 2. Pao, ¥. C. Elements of Computer-Aided Design and Manufacturing. New York: Wiley, 1984. 3. Szilard, R. Theory and Analysis of Plates: Classical and Numerical Methods. Upper Saddle River, NJ: Prentice Hall, 1974.746 PART2 © APPLICATIONS Yang, T. Y. Finite Element Structural Analysis. Upper Saddle River, NI: Prentice Hall, 1986, Weaver, W., I, and P. R, Johnston. Finite Element for Structural Analysis. Upper Saddle River, NJ: Prentice Hall, 1984. 6. Gallagher, R. H. Finite Element Analysis—Fundamentals. Upper Saddle River, NJ: Prentice Hall, 1975. 7. Martin, H. C., and G. F, Carey. Introduction to Finite Element Analysis. New York: McGraw. Hill, 1973. 8. Logan, D. L.A First Course in the Finite Element Method, Boston: PWS-Kent, 1986, 9. Knight, E. The Finite Element Method in Mechanical Design. Boston: PWS-Kent, 1993. 10. Ugural, A. C., and S. K. Fenster. Advanced Strength and Applied Elasticity, 4th ed. Upper Seddle River, NI: Prentice Hall, 2003. 11. Boresi, A. P, and K. P. Chung. Elasticity in Engineering, 2nd ed. New York: Wiley, 2000. 12. Boresi, A. P, and D, M. Sidebottom. Advanced Mechanics of Materials, 3rd ed. New York Wiley, 1985, 13, Ugural, A. C. Stresses in Plates and Shells, 2nd ed. New York: McGraw-Hill, 1999. 14, Zienkiewitez, O..C., and R. L Taylor. The Finite Element Method, ath ed. vol. 2. Solid and Fluid ‘Mechanics, Dynamics and Non-Linearity. London: McGraw-Hill, 1991 15. Cook, R. D. Concepts and Applications of Finite Element Analysis, 2nd ed. New York: Wiley, 1980. 16, Segerlind, L. J. Applied Finite Element Analysis, 2nd ed. New York: Wiley, 1984. 17. Bathe, K. I. Finite Element Procedures in Engineering Analysis, Upper Saddle River, NI: Prentice Hall, 1996. 18, Bemadou, M. Finite Element Methods for Thin Shell Problems. Chichester, UK: Wiley, 1996. 19, Dunham, R. $., and R. E, Nickell, “Finite Element Analysis of Axisymmetric Solids with Arbitrary Loadings” Report AD 655 253. Springfield, VA: National Technical Information Service, June 1967. 20. Utku, S. “Explicit Expressions for Triangular Torus Element Stitfness Matrix.” Journal of the American Aeronautics and Astronautics 6, n0. 6, (June 1968), pp. 1174-76. PROBLEMS Sections 17.1 through 17.3 17.1 A planar truss consisting of five members is supported at joints 1 and 4 as shown in Fig- ure P17.1. Determine the global stiffness matrix for each element, ‘Assumption: All bars have the same axial rigidity AE. Figure P17.1CHAPTER 17 © FINITE ELEMENT ANALYSIS IN DEsIGn 172 through 17.4 The plane truss is loaded and supported as shown in Figures P17.2 through 75 176 7 178 PI7A4, Determine (@) The global stiffness matrix for each element. (®) The system matrix and the system force-displacement equations. Assumption: The axial rigidity AF is the same for each element, Figure P17.4 ‘A vertical load 10 KN acts at joint 2 of the two-bar truss 123 shown in Figure PI7.5, Determine (@) The global stiffness matrix for each member. (®) The system stiffness matrix. (©) The nodal displacements. (d) The reactions. (©) The axial forces in each member and show the results on a sketch of each member. Assumption: The axial rigidity AE = 30 MN is the same for each bar. Redo Problem 17.5 for the structure shown in Figure P17.6, with A= 1.8 in? and E 30 x 10° psi. Solve Problem 17.5 for the structure shown in Figure P17.7, with AE = 10 MN for each bar, Resolve Problem 17.5 for the truss shown in Figure P17.8, with AE = 125 MN for each member. 747748 PART 2 179 17.40 © Arpticarions Fig Figure P17.7 Figure P17.8 ‘The two-bar plane structure shown in Figure P17.9, due to loading P = 100 KN, settles an amount of wy = 25 mm downward at support 1. Determine (@) The global stiffness matrix for each member, (®) The system matrix. (©. The nodal displacements. (@)_ ‘The reactions. (©) The axial forces in each member. Given: E=210GPa, A =5 x 10~* mi for each bar. A plane truss is loaded and supported as shown in Figure P17.10. Determine (@) The global stiffness matrix for each member. (b) The system stiffness matrix. (©) The nodal displacements. (a) The reactions (©) The axial forces in each member. Assumption: The axial rigidity AE = 20 MN is the same for each bar.CHAPTER 17 © FINITE ELEMENT ANALYSIS IN DESIGN Figure P17.9 Sections 17.4 through 17.8 17.11 A propped cantilever beam of constant flexural rigidity ET with a vertical load of 10 kips at its midspan is shown in Figure P17.11. Determine (@) The stiffness matrix for each element. (©) The system stiffness matrix and nodal displacements. (©) ‘The member end forces and moments. (d) Sketch the shear and moment diagrams. Given: EI = 216 x 10° thin? 10kips ay aie aa Figure P17.11 17.12. A plane frame 123 with hinged supports at joints 1 and 3 is subjected to a horizontal load of 30 KN (Figure P17.12), Determine (@) The global stifiness matrix for each member. (®) The system stiffness matrix. (©) The displacements ua, v2, and 6. Design Assumptions: Members 12 and 23 are identical with a square cross-sectional area of A=b xh = 900mm? and E = 70 GPa. 749750 PART 2 1743 1714 17s 17.16 © APPLICATIONS Figure P17.12 A frame 123 is fixed at supports 1 and 2 as shown in Figure P17.13. A horizontal load of 40 KN acts at joint 2. Determine (@) The global stiffness matrix of each member. () The system stiffness matrix. (©) Displacements ua, vn and Given: E = 200GPa, fh =S¥2x 10° mm‘, hh =Sx 10mm, A, =2.5V2 x 10 mm?, A; =2.5 x 10° mm? 2 _40kN Figure P17.13, Redo Case Study 17-2, by discretizing the plate into four elements, as shown in Figure P17.14. Use a computer program with CST (or LST) elements. ‘Verify the results introduced in Case Study 17-5, using a computer program with CST (or LST) elements, A steel plate with a hole is under a uniform axial tension loading P (Figure P17.16). The di- mensions are in millimeters (@) Analyze the stresses using a computer program with the CST (or LST) elements. (®) Compare the stress concentration factor K, obtained in part a with that found from Figure CS. Given: P = 4 KN and plate thickness ¢ = 10 mm.qt 17.8 179 CHAPTER 17. © FINITE ELEMENT ANALYSIS IN DESIGN 10 kit 2? ns : Te may] 10in ° ie 10 Kips p34 fe10%n.-+h10%n.—f 150 Figure P17.14 Figure P17.16 Redo Problem 17.16 for the plate shown in Figure P17.17. -—3—4 +—— 150 Figure P17.17 ‘A simply supported beam is subjected to a uniform loading of intensity w (Figure P17.18). ‘Take L = 10h, 1 = 1, and v = 0.3. Refine the meshes to calculate the stress and deflection within 5% accuracy, by using a computer program with the CST (or LST) elements. Given: The “exact” solution [9] is e 3wi? SwLt _ 3wi+v) mes ape’ TG ERS SEht in which £ represents the thickness. th a a I = ster An os }— r— Figure P17.18 Redo Problem 17.18, for the case in which a cantilevered beam is under a uniform loading of intensity w (Figure P17.19). Given: The “exact” solution [9] has the form Bw! aw +») 16Eth® SE where ris the thickness, 751782 PART 2 17.20 17.21 17.22 17.23 © AppLIcarions L- Figure P17.19 Verify the results introduced i with CST (or LST) elements. Case Study 17-6, using a finite element computer program, ‘Accast iron cylinder (v = 0.2) of a hydraulic device having inner radius a and outer radius 4a is subjected to an external pressure p,. Using a finite element computer program with CST (or LST) elements, determine the distribution of tangential and radial stresses. Redo Problem 17.21, ifthe cylinder is under an internal pressure p;.Compare the results with the exact solution shown in Figure 16.3. Resolve Problem 17.21 for the case in which the cylinder is subjected to internal pressure p, and external pressure pp = 0.5p;.