Matrix Analysis of Beams and Grids 51 INTRODUCTION The basic formulation and steps involved in the application of matrix methods (stiffness and {exibility methods) were demonstrated in the previous Chapters with reference to one- dimensional axial struetures, plane trusses and space trusses. This Chapter is a continuation ‘ite same theme, with the application directed to beams. Towards the end of the Chapter, ® shall also discuss ‘grids’, which are spatial skeletal structures involving a planar network ‘Tteams that may resist torsion, in addition to bending and shear. __ The beam is a skeletal element that resists external loads by undergoing bending (or qt") whereby the primary internal force resultant (which can vary along the length of the nett) is the bending moment, M(x), and the corresponding deformation, over an elemental sacha is a rotation d@ = pdx, where (x) is the curvature. Assuming linear elastic avatun (Evlet-Bemoulli theory), the bending moment M(x) varies linearly with the Amen, {22 and the constant of proportionality is the flexural rigidity EY of the beam In addition to bending moment, there is another internal force resultant, called shear * 5), which is equal to the rate of change of bending moment; a322 Advanced Structural Analysis: normal, well-proportioned beams, we assume that shear ee MS are Deplipit Shall see later how to account shear deformations in the analysis, w' WEVET required Wy ' i i At any point along the beam centre-line, there we “wo displace, t va interest to us, viz., the deflection A(x) and its Panes eae If we py Values of these two quantities at the two ends of any prismatic 7 element (y hing Part of a continuous beam system), then it should be pastinie & Bit dics he intermediate location. This forms the basis of the stiffness formulation, whereby fel of independent displacement coordinates (ie., degrees of freedom) in any prismagete Clement is limited to four. It is assumed that the loads in the form of direct Aton applied at the joints; otherwise, if there are intermediate loads, these shall be cone equivalent joint loads, Thus, in the conventional stiffness formulation, the clemen 4 four degrees of freedom (two translational and two rotational), whereby the stiffness te of the element (in local coordinates) is of order 4 x 4, This is discussed in Section 59 In the reduced Stiffness formulation (and the flexibility formulation), the Aegrey freedom are reduced to ‘wo, eliminating the translational degrees of freedom, Tn this K the stiffness matrix of the element (in local coordinates), and its inverse, the flexbifn “®® ( is of order 2x2. In the reduced stiffness formulation, we can also reduce the de + : : in which the extreme ends aa 5.2 CONVENTIONAL ‘STIFFNESS METHOD APPLIED TO BEAMS 5.2.1 Four Degrees of Freedom Consider a prismatic (initially straight) beam clement i, having a length L, and a crose-secoe with second moment of area 4. Let the bendit (start node) be M(x*), i and rotation (2+) =e The curvature of the given by ox)» SS for small deformations, as explained in Section 1.5.3. Assumits d plane sections before bending to remain plane after bending # eeoe y the slope (rotation), O(x*), will have a se cubic Variation along the length of the ve 8 cds : Thus, intuit *) wil a i following form, expressible in terms of four constants, icra ce 2 AGM) =C, + C02) +0, (08)? +C,(24)) 6.1) dA ON =F G 426,08) 430,008) (52) Choosing the coordinate system as shown in Fi ig. 5.la, the kinematic boundary conditions at the two ends of the beam, can be expressed as follows: *=0 > A=Di; 6=Di, - _ * (5.3) =L > A=Dj; @=p), Applying these kinematic boundary conditions, and solving the resulting equations, ve can express the four constants in Eq. 5.1 in terms of the four end displacements, D;, D,, Dyand De: Cc, =Di, C=D, = 3D i + Dh) (20%. + Ds) Ee Z (5.4) 2h Pe) +(De + Dis) [a ‘Thus, we have established that the deflection and rotation at any location x* in a ‘Prismatic «tam element, subject to shear forces and moments at the two ends (without any late loading), can be expressed in terms of the four degrees of freedom at the two inthe beam, Applying the Principle of Superposition, we can show that the deflection A); Totation’ OG) and curvature o(<*) et aay distance = fron Ge mar node of the can be generated as follows :324 Advanced Structural Analysis . Di AG | A(t) gC) ACM) AO] py O(x*) $=] Ox") 8,00) 0%) 0,0") D. ex] [AG GG) HO) HO? D. the deflection, rotation and cury, ti where A,(x*), @,(2*) and @,(x*) denote respectively, ents restrained. at x*, corresponding to D), =1 and all other end displacem 5.2.2 Stiffness Matrix of Beam Element with Four Degrees of Freedom ‘The stiffness matrix k. of the beam element, i, with four degrees of freedom, as show, Fig. 5.1a, has an order of 4x4. By definition, the first column of this stiffness mary, corresponds to the displacement condition, ‘Dl. =1 (unit vertical translation at the start nog. with the other three degrees of freedom restrained. ‘The deflected profile is as shown i, Fig. 5.1b. The stiffness coefficients, kr fand Ki, correspond respectively to the (upway and the (anti-clockwise) end moment, Fj,, at the start node, while k,,, and kine correspond respectively to the (upward) vertical force, F-» and the (anti-clockwise) ex moment, Fi., at the end node. The resulting bending moment diagram is also depicted a Fig. 5.1b. Similarly, the other stiffness coefficients are defined, corresponding 10 D. = (Fig. 5.1c), Dj. =1 Fig. 5.16), and Dj. =1 (Fig. 5.1¢), with the other degrees of freedon restrained. ‘There are many alternative ways of generati adopt either a ‘displacement approach’ or a ‘force approact vertical force, Fi. ng the stiffness coefficients. Wet h’ for this purpose. Displacement-based Formulation leflection function defined by B45! In the displacement approach, we work with the cubic d and derive the following expression for curvature, (x*) = A") - 58) xt) = A(x*) = 2C, +6C,(x*) Now, invoking the linear relationship between bending moment and curvature, M(x) =(El), 002") = (ED), [2G +60] cu ‘As expected, we have a linear variation of bending moment along the length of beam element, whereby the shear force, S(x*), will be constant, and given bY: dM 5.8) St) = Fo (B66) :oe Hee Eee Hee Kop Bw a, =12(Et), /Es b =6(EN)/ Gs 6 » & & git h 4 | 1), 1 k, - . a ye = aT ey se ; bee IG Ty ° | we | ten, =1(e1 ; penta vo i | L | oa oo | de, | pba 4). (Go =a ry | mcg. os (@ Sttiness coefficients due to Ds. =1 | (psitiness coefficients due to Di=l Dh, = Di. = = ( 0) (2: 0) 4 ) =0) (.=0) Dy=1 ee ED, ye Mf) ee a 5 p _ 64), ? s() [rE c ae 2(EN), 222E), ae Di. ine th LB | aoe A (et wa (et), jp! E 2(El), 4(E1), | Bow 22) foc (e) Stitiness coefficients due to Di. = | (@Stiiness coetficients due to Dy. =1 Figure 5.1. Generation of 4x4 stiffness matrix for a beam element Wd can now invoke the static boundary conditions, relating the beam end forces to force and bending moment at x* = 0 and x* = Ly following the sign convention we te opted earlier, for the condition Dj, = 1:2 P, 2A 326 __ Advanced Structural Analysis Sy Using this ‘displacement approach’, we can derive the following ex, deflection a,(2*), rotation @,(x*) and curvature @,(x*) and the Stiffness Cont 6x* 12(x*)? Ee 6. 12(EN), , _ (EL), a) B ikea 6 + > Marsa 4x* 3(x*) poor sceer{i-£) = Bey=1- + = = AG) =—7 SD, ED, (EN), 7 SOF hee A he Biker 21 1m) fen —- z 2a gE my es 6x*_6(x*)° 6 12x* ace=3(2) +o) > See > BO%=— (65.10) 12(EL 6(EI), 12(E1), 6(Et), re A). Kaye --, Kae . Kaes ER x* 2a (x4 = 2, 6x* (»£] > SaaS EF OG= ae 3108 Os, » SE) , 4(El), fons at These solutions are indicated in Fig. 5.1. It may be noted that the deflections oF mid-span (x* = 0.57,), oa Ponding to the various displacement profiles, are ! 4,=4,=05 and a, =0.125L; A, =-0.1251,. The resulting element stiffness matrix takes the following form:Matrix Analysis of Beams and Grids __ 327 YK, 12/8 ot, (ED) bh 4 Of, 2 TL |-1/G -/, 12/h -6/1, Gh 2 0b 8 (5.41) n cad we the stiffness coefficients using the energy formulation described in ge % 75, This constitutes an alterative ‘displacement approach’. For this purpes®, we 2" Tmulate an expression for the strain energy U, in the element, in terms of the four ae af which is usually negligible in well-proportioned beams, the expression for strain nts, and generate kj. = Ignoring the contribution of shear strain vu, aD,aD),* El), 7 EDT econ -9,(29 0), +9,62)D,+9,0 0D. de (5-12) 24 jer the various expressions for 9,(x*) are listed in Eq. 5.10. Expanding Eq. 5.12 and png Eq 1.65, itcan be shown that . =x = (El), Joe, Cae (5.13) For example, ay, “FD, 4 eh (al 4 al acet 3672+ 48] = =(El), locotas=(en, ay a eae es EE ee 12(EI), cD eeu (EN, Jostmendeaten I en b Similarly, all other stiffness coefficients can be generated. sre tis Det rroach, we consider equilibrium of the free-body, corresponding to any of the vations i Fig, 5.1. By considering a section of the beam at a distance x* from the * We can express the bending moment at this section as follows: M(x) =-,. +b, (x*) ea Mea~ 328 Advanced Structural Analysis M We can now generate an expression for curvature 9 (= ue) and by (EI), ite pray oy, generate expressions for rotation @,(x*) and deflection A, (*): 2 1 (*) sya | kd, Cat) thie |G 80°) an| ie 2) + Rey > | 7 : sec (is) +) =| “kt + Cy (x*)+ 4,G*) , Kaep @ w By applying the two kinematic boundary conditions at x* =0, the two Eq. 5.15 reduce to: C, = 8,;(x*=0) and Cy =4,(*=9)- Then, the remaining two kinem” boundary conditions at x* = L;, can be solved to yield the two stiffness coefficients (a 4, start node): 12(EI),[G/6 -L,/2 6,(x*=L,)-8, ("= 0) BE [u/2 -t |laGtat)-4/@=9-L8,0%= 0) 6.19 The remaining two stiffness coefficients (at considering equilibrium of the overall free-body: the end node) can be obting (5.17) =1 (Fig. 5.1b), (= 0) =1, whereby, applying Eqns 5.16 and 5.17, For example, considering D;. Cy, =0,(x*=0)=0 and Cy, Kune] _12(ED,[E/6 -4/2]{ 0-9 |. 6(E1),/ Kan Leal te} “TE ie “1 Testa nena isan deflection A(x") 2 ® for deflections des Substituting these values in Eq. 5.15, the expression for generated. Thus, the various stiffness coefficients and expressions in Egns 5.10 and 5.11 can be generated using the ‘force approach’. 4 x4 Flexibility Matrix not Possible! Itis interesting to note that the determinant of the 4 x 4 stiffness matrix Kk: in Ba sl Ee (rank = 2'), whereby the matrix is not invertible. This means that a 4x4 flexibility (inverse of the stiffness matrix) for the element shown in Fig. 5.1a cannot be ; ee i * For example, the third column is obtainable by iplyi row) bY ft * yy multiplying the first column (Of TOY. iy, thereby is linearly dependent on it. Similarly, the third row can be generated by multiplyins row by-1.Mat | trix Analy ie Got this element is unstable and cannot resis Yai of Beane and Grite__329 Fy, and 1 the ones ba In order to action of any arbi s B ms yy arbitrary combination ee degrees of freedom. The resulting flexi Cniy mato seseeery 9 lity matri “ 'Y matrix is a 2 x 2 matrix, to be defined coordinate Transformations and Structure Stiftness Matrix ion Matrix is beam system, the local coordi 3 continuous al coordinate x*- and y*-i beam 8 on cwith 4 may of ope 1m) can be conveniently aa pdr! same setio“ 5 a a aod areas egetuae ae e four local coordinates, fmbered i : a ha as shown in Fig, 5.1a, can be directly linked in the global axes ee, 3S 1, 2, 3 and 4, to appropriate global coordinates (say, | m, n and p) at the same cons in the ‘continuous beam. Compatibility of displacement requires D, =D} = P+ =D), and D,=D,=Dj,.. The 4x4 element transformation matrix displacement and force vector components, Jobal axes system to the local coordinate system. identity matrix (T' =1 27"). It is cor 7,-0,= Pe Dis 4, relating the (FE FO in the gi (D,,D;,D,,0,)_ and pi=T'D' and mnvenient tO eT, 2s defined in Chapter 3, will be an sak the global coordinates in parenthesis as follows: 100 0] qi |eeu ol 18) 0 0 1 0]@ 0 0 0 1) stem is given by the transformation trix in the global axes 89 jess mal ces to k' =k: whereby: The element stiffne ), which in this case, v1 KT (refer Eq. 3- Sifiness coefficients © coordinate locations. In translational and a rotationaller cont: tog e inates must be towards the end, with the active coording ten < the beginning. “ph Displacement Transformation Matrix As an alternative, and to avoid the additional work involved in keeping HACK of he linkages with the global coordinates (J, m, n and p) connected to a Particular beam «)%% We can generate the displacement transformation matrix, Ty of order 4 x (n n, vee the number of active global coordinates (equal to the degree of kinematic indeterming."* rs the number of restrained coordinates. Typically, if there are m elements acon) beam system, there will be m + 1 joints, and hence, n + r=2(m+1). This Matrix, con Moy clements that are either zero or 1, relates the element local coordinates 1a all a the coordinates: {085 = [loca | oe] 6m Using the element stiffness matrix k! given in Eq. 5.11, the product kt! is fay Somputed for each beam element and thereafter, the structure stiffness matrix k is assemble, directly by summing up the stiffness coefficients contributions from. the various bean elements using the following transformation: al eset fle fre] EEE] ea Feleaenn * fag lalla. leaden, 5.2.4 Fixed End Forces such as the ones listed in Table 1.3, can be conveniently used to generate the fet moments, and using equilibrium equations, the corresponding fixed end reactions ( o forces at the two ends) can be calculated. Care must be taken to assign the appropriate (Positive or negative), while generating the fixed end force vector for each