TWO-DIMENSIONAL MATRIX STIFFNESS ANALYSIS Analytical Model As you are aware, matrix methods do not involve any new fundamental principles. The fundamental relationships of equilibrium, compatibility and constitutive (member force-displacement) relationships are simply expressed in the form of matrix equations so that the numerical computations can be efficiently performed on a computer. 1 Again, in matrix stiffness analysis, the structure is modeled as an assemblage of straight members connected at their ends to joints or nodes. Kassimali (1999) defines a member (or element) as: part of the structure for which the member force-displacement relations are valid. A joint is defined to be a point used to idealize a structural connection. A node can be a joint or simply a mathematical break in the member in order to model a 2 change in geometry, loading or simply to provide intermediate displacement and force calculated results. Prior to performing an analysis of the structure, an analytical or mathematical model of the structure must be defined. The model is represented by a line diagram of the structure on which all nodes and elements are identified by numbers. NOTE: Element force-displacement relations herein are exact for piecewise prismatic members only.3 Figure 17.1 – Frame Structure (Kassimali, 1999) 4 1 Overall geometry and behavior of a structure is referenced to a global or structural Cartesian coordinate system XYZ – plane of the structure lying in the XY plane. However, the basic force-displacement relationships are derived in terms of forces and displacements aligned with the element (member), which is referred to as the local coordinate system xyz. Furthermore, design forces are defined in terms of the local 5 coordinate system. Displacement degrees of freedom are numbered starting at the lowest node number and proceeding sequentially to the highest node number. Order of the nodal degrees of freedom is: X-axis, Yaxis and rotation about Z-axis. If a DOF does not exist, e.g., continuous beam model that neglects axial deformation or truss structure in which the joint rotations are not defined since truss joints are presumed to be frictionless hinges, then that DOF is eliminated from 7 the sequence. x Y Global and Local Coordinate Systems y frame element X z Z (a) Global Coordinate System (b) Local Coordinate System Figure A – 2D Stiffness Analysis Coordinate Systems In Fig. A: x-axis aligned with the element centroidal axis y-axis normal to the element x-axis and consistent with the right hand rule positive coordinate sign convention z-axis out of plane local axis parallel and in the same direction as the global Z-axis; y-axis is normal to 6 the x-z plane using the right hand rule positive sign convention Furthermore, if a displacement is known, then that DOF is also not numbered. Figure 17.2(a) shows a continuous beam subjected to transverse (lateral) loading only. For such a beam, the axial deformation is zero and the corresponding analytic model is shown in Fig. 17.2(b). 8 Figure 17.2 Continuous Beam Structure 2 Similarly, Fig. 17.3 shows a truss structure and Fig. 17.1 shows an example frame structure. NOTE: The frame structure element includes axial and bending deformation, which results in three displacement degrees of3 freedom per node. (a) Truss Figure 17.3 Truss Structure (17.1) 10 (a) Frame Member: Global and Local Coordinate Systems (17.2a) Q 2 k 21u1 k 22 u 2 k 23u 3 k 24 u 4 (17.2b) k 25u 5 k 26 u 6 Qf 2 Q3 k 31u1 k 32 u 2 k 33u 3 k 34 u 4 (17.2c) k 35u 5 k 36 u 6 Qf 3 Q 4 k 41u1 k 42 u 2 k 43u 3 k 44 u 4 k 45u 5 k 46 u 6 Qf 4 [S] {d} = {P} – {Pf} 9 Considering a plane frame element with three nodal degrees of freedom ( NNDF) and six element degrees of freedom ( NEDF) as shown in Fig. C, the element stiffness equations are k15u 5 k16 u 6 Qf1 As stated previously, the nodal displacement vector {d} for the analytical model of the structure is obtained by solving a system of simultaneous equations The structure stiffness matrix [S] is obtained by assembling the stiffness matrices for the individual elements of the structure. (b) Analytical Model Q1 k11u1 k12 u 2 k13u 3 k14 u 4 Member Stiffness Equations in Local Coordinates (17.2d) (b) Frame Member End Displacements and Forces Figure C. Frame Element Terminology and Deformation Behavior 12 3 Q5 k 51u1 k 52 u 2 k 53u 3 k 54 u 4 (17.2e) k 55u 5 k 56 u 6 Qf 5 Q6 k 61u1 k 62 u 2 k 63u 3 k 64 u 4 (17.2f) k 65u 5 k 66 u 6 Qf 6 Recall that kij represents the force at the location and in the direction of Qi required, along with other end forces, to cause a unit value of the displacement uj while all other end displacements are zero. Such forces per unit displacement are defined as stiffness coefficients – first subscript identifies the force and the second subscript identifies the displacement. Equations (17.2) can be expressed in matrix form as Q1 Q 2 Q3 Q 4 Q5 Q6 k11 k 21 k 31 k 41 k 51 k 61 k12 k 22 k13 k 23 k14 k 24 k15 k 25 k 32 k 33 k 34 k 35 k 42 k 43 k 44 k 45 k 52 k 53 k 54 k 55 k 62 k 63 k 64 k 65 Qf1 Qf 2 Qf 3 Qf 4 k16 k 26 k 36 k 46 k 56 k 66 Qf 5 u1 u 2 u3 u 4 u5 u 6 Qf 6 T (17.3) 13 The element stiffness matrix of (17.3) expresses the end forces in terms of the element material and geometric properties (stiffness coefficients) times the element displacements ( u i , i 1,2,,6 ) plus the contribution of the element loads in the kinematically determinate state (fixed-end forces Qfi , i 1,2,,6 ). As shown in Fig. C, the element end displacements are measured relative to the undeformed position of the element in the local coordinate system. 15 14 Translational forces and displacements are positive when they are in the positive x or y coordinate directions and counterclockwise moments and rotations are positive. A more succinct representation of (17.3) is (17.4) {Q} [k]{u} {Qf } {Q} local coordinate element force vector {Qf } local coordinate element fixed-end force vector {u} local coordinate element displacement vector [k] local coordinate element stiffness matrix 4 Stiffness Coefficient Derivations The frame element stiffness equations of (17.3) can be rewritten as {Qa }2x1 [k aa ]2x2 [k ab ]2x4 {u a }2x1 {Q}6x1 {Qb }4x1 [kba ]4x2 [kbb ]4x4 {ub}4x1 {Q } fa 2 x1 {Qfb }4 x1 (1) Where {Q a } Q1 Q 4 T = axial member end force vector; {Qb } = Q2 Q3 Q5 Q6 T = bending member end force vector; [kaa] = axial member stiffness matrix; 17 Equation (1) simply rearranges the degrees of freedom listed in (17.3) into axial and bending components and clarifies that the two deformation modes are uncoupled ([kab] = [kba]T = [0]2x4) for straight members in the local xy coordinate system. Since the frame element is straight, the axial and bending deformation modes are uncoupled. Thus, the derivation of the stiffness coefficients can be carried out separately. 19 k11 k14 [kbb] = beam member k ; 41 k 44 stiffness matrix = k 22 k 32 k 52 k 62 k 23 k 33 k 53 k 25 k 35 k 55 k 63 k 65 k 26 [k ] = [k ]T = ab ba k 36 axial-beam ; k 56 coupling k 66 stiffness matrix = [0]2x4; {ua} = axial member displacement vector = <u1 u4>T; {ub} = beam member displacement vector = <u2 u3 u5 u6>T; subscript f ≡ fixed-end member force vector; and subscripts 1, 2, …, 6 coincide with the degrees of freedom shown in 18 Fig. 17.4 and Eq. (17.3). Axial Stiffness Coefficients Consider the frame member of Fig. C subjected to u1 = 1 with all other displacements equal to zero as shown in Fig. A.1 below. Figure A.1 u1 1; u 2 u 3 u 4 u 5 u 6 0 The force Q1 (= k11) to cause u1 = 1 with all others equal to zero can be obtained using Hooke’s law (see your Mechanics of Deformable Solids textbook): 20 5 x E x (a) k 41 Q 4 u 1 1 Substituting x Fx A Remember Q1 ; A x L L u1 L 1 L leads to Q1 x A E x A EA L Since u1 = 1 with all others equal to zero no bending deformation is introduced as shown in Fig. A.1: k 21 k 31 k 51 k 61 0 EA k11 Q1 u 1 1 L Similarly, imposing u4 = as shown in Fig. A.2 leads to: Stiffness coefficient k41 at end e is obtained from equilibrium: Fx 0 k11 k 41 k 41 k11 EA L Figure A.2 u 4 1; u1 u 2 u 3 u 5 u 6 0 21 EA k 44 Q4 u 1 4 L Fx 0 k14 k 44 k14 k 44 EA Q1 u 1 4 L k 24 k 34 k 54 k 64 0 22 Beam Stiffness Coefficients You may recall from your Mechanics of Deformable Solids course that the differential equation of a beam composed of linearly elastic, homogenous material and loaded in the plane of symmetry of it crosssection can be expressed as d 2u y dx 2 M EI (2) in which uy = deflection of the beam’s centroidal axis (which coincides with the neutral axis) in 23 24 6 the y direction at a distance x from the origin of the x-y coordinate system; and M = internal bending moment at the same location x and follows the usual beam sign convention. To calculate the stiffness coefficients in the first column of [kbb], subject the member to u2 = 1, u3 = u5 =u6 = 0 as shown in Fig. B.1. (NOTE: Axial displacements are not considered since they are uncoupled from the bending displacements.) Recall that kij = force at dof i due to a unit displacement at dof j with all other displacements equal to zero. Taking a cut at an arbitrary distance x in Fig. 1 results in M k 32 k 22 x (3) Substituting (3) into (2): d 2u y Figure B.1 – First Column of [kbb] = {kbb}1 dx 1 x2 k 32 x k 22 C1 dx EI 2 uy 1 (k 32 k 22 x) EI 25 Integrating (4) twice with respect to x leads to du y 2 (5a) 1 x2 x3 k 22 k 32 C1 x C2 (5b) EI 2 6 where C1 and C2 are constants of integration. A total of four unknowns are included in (5a, b) – the two constants of integration and the two stiffness coefficients. Referring to Fig. B.1 and imposing the displacements provides four boundary conditions, i.e. 27 (4) 26 End b, x 0 : (x 0) u 3 0 C1 u y (x 0) u 2 1 C2 (6a) End e, x L : L2 (x L) u 6 0 k 32 L k 22 C1 2 (6b) 2 3 1 L L u y (x L) u 5 0 k 32 k 22 EI 2 6 C1 L C2 Equations (6a) show that C1 = 0 and C2 = 1 for Fig. B.1. Substituting these results into (6b) leads to 28 7 k 22 12EI k 32 6EI (7a) L3 (7b) L2 The stiffness coefficients at end e can be calculated directly from equilibrium of the beam element, i.e. Fy 0 k 52 k 22 12EI (7c) L3 M e 0 k 62 k 32 k 22 L 6EI L2 (7d) All four beam bending cases are in your class notes. Pick one of the 29 remaining three for me to solve. Combining the derived axial and beam stiffness coefficients, the element stiffness matrix coefficients of (17.2) can be expressed in matrix form as A 0 0 A 0 0 L L 6I 12I 6I 0 12I3 0 L L2 L3 L2 6I 4I 0 0 6I2 2I L L L2 L [k] E A 0 A 0 0 0 L L 0 12I 6I 0 12I 6I L3 L2 L3 L2 6I 2I 0 6I2 4I 2 0 L L L L (17.5) 31 Combining (7), (12), (17) and (22) from your class notes leads to k 22 k [k bb ] 32 k 52 k 62 12 3 L 6 L2 EI 12 L3 6 L2 k 23 k 25 k 33 k 53 k 63 k 35 k 55 k 65 6 2 L 4 L 6 L2 2 L 12 3 L 6 L2 12 L3 6 L2 k 26 k 36 k 56 k 66 6 L2 2 L 6 L2 4 L (17.6) which is obviously symmetric, i.e., kij = kji . Equation (17.6) is the 30 beam element stiffness matrix. Derivation of Frame Member Fixed-End Forces The fixed-end force vector of (1) has contributions due to axial ({Qfa}) and bending ({Qfb}) member loads. Consider the beam shown in Fig. B.5, where px is the distributed axial load and py is a uniform transverse load. Figure B.5 – Frame Element Subjected to Uniform Axial and Transverse Loadings 32 8 Fixed-End Axial Forces Fixed-end axial forces are obtained by solving the following differential equation du x Q x dx EA (23) x du x () d d x Q x (x) Qf1 p x d 0 where ux = x-axis (axial) displacement; and Qx = axial force (or axial stress resultant). The solution of (23) is expressed as where Qa(x) depends on the distribution of the axial for local xaxis loading. For the uniformly distributed axial loading of Fig. B.5: Q x ( ) d EA (24) Qf1 p x x (25) Substituting (25) into (24) and integrating over the element length L: u x (x L) u x (x 0) x Qf1 L p x L2 EA 2EA 33 (26) 34 where ux(x=0) = u1; and ux(x=L) = u4. Imposing the boundary conditions u1 = u4 = 0 on (26) and solving for Qf1: p L (27) Qf1 x 2 Fixed-End Bending Forces Fixed-end bending forces are obtained by solving the following differential equation Fixed-end force Qf4 is obtained from equilibrium: The solution of (29) is expressed as Fx 0 Qf1 Qf 4 px L p L Qf 4 x 2 (28) d2u y dx 2 M EI d 2 u y () x d 2 d (29) M() d EI x d 2 u (x) M() y d dx d dx EI dx 2 x x x x 35 (30a) (30b) 36 9 where M() depends on the distribution of the bending or y-axis loading. For example the uniformly distributed loading py shown in Fig. B.5 results in the following moment expression M(x) Qf 2 x Qf 3 p y x2 2 (31) Substituting (31) into (30a) and integrating results in (x L) (x 0) 3 L L (Qf 2 L 2Qf 3 ) py 2EI 6EI (32a) p y L2 12 and Qf 2 pyL 2 u y (x L) u y (x 0) 2 (32b) 4 pyL L (Qf 2 L 3Qf 3 ) 6EI 24EI where uy(x=0) = u2; and uy(x=L) = u5. Imposing the boundary conditions u3 = u6 = 0 on (32a) and u2 = u5 = 0 on (32b) leads to p y L3 L (Qf 2 L 2Qf 3 ) 2EI 6EI (33a) p y L4 L2 (Qf 2 L 3Qf 3 ) 6EI 24EI (33b) Fixed-End Forces: Thermal Loading Solving (33a, b) for Qf3 and Qf2 gives Qf 3 du where y ; (x=0) = u3; and dx (x=L) = u6. Similarly, substituting (31) into (30b) and integrating: (34) Axial Deformation Fixed-end forces Qf5 and Qf6 are obtained from equilibrium: pyL (35) Fy 0 Qf 5 2 M e 0 Qf 2 L Qf 3 Qf 6 p y Qf 6 Qf 2 L Qf 5 L2 2 p y L2 2 2 pyL 12 (36) 39 Figure B.6 – Axial Force Member Subjected to Uniform Temperature Loading 40 10 The prismatic bar of Fig. B.6(a) is subjected to a constant temperature load of Ta. For this case x L L Ta L LTa if permitted to freely expand as shown in Fig. B.6(a) and = linear coefficient of thermal expansion. If the expansion is prevented, as shown in Fig. B.6(b), the force required to prevent the free expansion is Q EA L EATa L Thus, the fixed-end force at end e of the member is Qf 4 Q EATa and the fixed-end force at the beginning of the member (b) is obtained from x-axis equilibrium to give Qf1 Qf 4 EATa (37b) These axial fixed-end forces are independent of the length of the member. 41 Bending Deformation (37a) 42 In the beam bending case, the beam of Fig. B.7 is presumed to be doubly symmetric (bisymmetrical) and prismatic. The entire bottom surface of the beam is heated to Tl and the top surface Tu, which produces a linear temperature gradient Tb = (Tl – Tu) through the depth of the beam. If free to expand, the axis of the beam would elongate L = L(Tl + Tu)/2 Figure B.7 – Thermal Gradient 43 Through Beam 44 11 The axial fixed-end force can be calculated in this case by substituting (Tl + Tu)/2 for Ta in (37a, b). Neglecting second-order effects, which is the case for linear analysis 1 d (Tl Tu ) (38) dx dx h For the bending case, the beam will curve into a circular arc as shown in Fig. B.7(b). Figure B.7(c) is an enlarged view of Fig. B.7(b) from which it can be seen that where = radius of curvature and = curvature. Thus (Tl Tu ) d dx h d (Tl Tu ) Tb dx h h 2 d d u y M dx EI dx 2 Substituting (38) into (39) and solving for M leads to Qf 3 Qf 6 M 45 (39) EI(Tl Tu ) (40) h 46 12