Tamkang Journal of Science and Engineering, Vol. 8, No 1, pp.

29-42 (2005)

29

Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities
Ying-Hui Lei* and Yu-Lin Chien
Department of Civil Engineering, Tamkang University, Tamsui, Taiwan 251, R.O.C.

Abstract
In this paper, the dynamic behavior of a group of transmission towers linked together through electrical wires and subjected to a strong ground motion will be investigated in detail. In performing the seismic analysis, the wires and the towers concerned are modeled, respectively, by using the efficient cable elements and the 3-D beam elements considering both geometric and material nonlinearities. In addition, to enhance the practical usage of the analytical scheme, the strength capacities and the fracture occurrences for the main members of the transmission tower will be examined with the employment of the appropriate strength interaction equations. It is expected that by aid of this investigation, those who are engaged in code constitution or practical designing of transmission towers may gain a better insight into the roles played by the interaction force between towers and wires and by the configuration of the transmission towers under strong earthquake. Key Words: Transmission Towers, Geometric and Material Nonlinearities

1. Introduction
The importance of the transmission tower on national economy and people’s living has been well recognized. During the attack of the Ji-Ji earthquake, with a size of 7.3 in Ritcher magnitude, in Taiwan on Sept. 21, 1999, over two thousand four hundred residents were killed. Besides, the strong vibration of the ground motion has caused the collapse of a main transmission tower located in the central region of the state. As a result, the government was forced to take measures of reducing electricity supply for more than six weeks. During this period, a great living inconvenience was brought to the people, and a huge commercial loss was incurred in the high-tech industry of the island. To achieve the aim of supplying electricity everywhere in a country, many transmission towers are hence built in the rugged circumstances of climbing mountains
*Corresponding author. E-mail: demanad@mail.tku.edu.tw

or crossing rivers. Accordingly, the elevation at which some tower structures are located may differ from that associated with other transmission towers. Moreover, the marching route of the tower procession in such circumstances may exhibit in an extremely irregular manner. This variation on either the elevation or the orientation of the geometric configuration for a group of transmission towers would certainly affect the interaction force between electrical wires and tower structures. The conventional seismic analysis of transmission towers is usually undertaken by taking each of towers as an isolated structure without taking the strong traction given by high-voltage electrical wires lining up in various directions in the air into account. Furthermore, many of structural engineers were used to simply ignore all wire mass or to take the wire mass as the lumped mass affiliated with the tower in seismic analysis. The results obtained from such analytical schemes would not be able to reflect effectively the actual forced conditions of the tower structure itself as well as the base foundation be-

(2) the joint slippage resulting from the providence of erection tolerance in the course of producing the bolt holes throughout tower members [1-3] and (3) the flexural deformations of the primary leg member.and ZL-axes. Analytical model of tower system. 2. with its two ends being denoted by I and J respectively. 2. a sophisticated seismic analysis considering both geometric and material nonlinearities will be performed in this research. In consequence. The YLZL-plane is defined as the vertical stretched Figure 1. It is expected that the adoption of these assumptions would not lessen much the value of the findings regarding the variation tendency of the internal forces acting on tower members under various configurations of a group of transmission towers. introduced from the secondary diagonal members jointed with the leg member by using bolted connections [2.30 Ying-Hui Lei and Yu-Lin Chien neath it. the phenomenon of material nonlinearity is often observed on the primary leg members especially for those in the lowermost part of the tower subjected to a strong ground motion. Accordingly.4]. On the other hand. the electrical wire should therefore be analyzed by taking the effect of geometric nonlinearity into account. Being slender and tall in appearance. To simplify the calculations involved. The cable element proposed by Jayaraman and Knudson and having been verified to be accurate and efficient in acquiring the stiffness matrix and elemental nodal-forces will be used for the modeling of electrical wires [5]. as sketched in Figure 2. as indicated in Figure 1. The complicacies are mainly due to (1) the flexibility of joints behaving nonlinearly from the very onset of loading [1]. tower structures and end-restraint springs. Since the transmission tower is usually constructed by using the rolled steel angles which are eccentrically connected one another. nonlinear seismic analysis with respect to such a structure is widely known extremely difficult when the flexural deformations of the angle-section members are intended to be taken into account. . The formulation involved in the modeling of each portion of the system will be described in the following. To describe properly the deformed behavior of the structural joints connecting tower members together is probably one of the most complicated tasks in tower analyses. the transmission tower is destined to be susceptible to the effect of geometric nonlinearity. a proof-loading or full-scale testing combined with a linear elastic analysis in which the assumption of axially loaded conditions is applied to all the component members has formed an integral part of the tower design in practice [4]. Modeling of Structural System All seismic analyses in this paper are undertaken with respect to the structural system composed of various portions including electrical wires. Consider a cable element in the local coordinate system composed of YL.1 Electrical Wires and End-restraint Springs Being highly flexible and undergoing significant deformations. the effects due to joint flexibility and bolt slippage will be neglected and each component member of the transmission tower is assumed to have a square shape of cross section in this paper.

Cable element in local coordinate system. namely the vertical axis of the global coordinate system. F4.TI2 ) + J 2 EAw w F4 + TJ ù 1 é 2 ê F4TJ + F2TI + F1 log ú TI . TI and TJ be the tensile forces acting at ends I and J respectively.F2 û ë EA w V= T . Lu and L be the unstressed and stressed lengths of the element respectively.F2 û 2 EAw ë (4) (5) (6) Figure 3. It is observed from the above equations that H and V can be written in terms of F1 and F2 only. Schematic presentation of error vector JiJ.F1 ê u + log 4 J ú TI . ai4 are correction factors. H and V be the distances between two ends measured along horizontal and vertical directions respectively. F3. ai3. the following iterative equations can be used: i F1i +1 = F1i + d F1i = F1i + a1id H i + a 2d V i i i F2i +1 = F2i + d F2i = F2i + a 3d H i + a 4d V i (13) (14) L =V2 + H2 where sinh 2 l l2 (1) in which ai1. and the positive direction of YL-axis is defined to be the same as that of Z-axis. w be the weight of the cable element per unit unstressed length. ai2. F4 = -F2 + wLu F3 = -F1 TI = ( F12 + F22 )1/ 2 TJ = ( F32 + F42 )1/ 2 (7) (8) (9) (10) Figure 2.Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities 31 The variables F1. the stiffness matrix of the cable element can be written as [5] l= F2 = w| H | 2 F1 w é cosh l ù -V + Lú 2ê sinh l ë û (2) (3) éL F +T ù 1 H = . TI and TJ are related by the following equations. one can thus express the stressed shape of the cable element by using the following Eqs. F2. L = Lu + . The iterative process is repeated until the error vector JiJ (referring to Figure 3) converges to a value lower than the prescribed tolerance limit. [5-7]: æ ¶H ö æ ¶H ö i i dHi = ç ÷ d F1 + ç ÷ d F2 è ¶F1 øi è ¶F2 øi æ ¶V ö æ ¶V ö i i dV i = ç ÷ d F1 + ç ÷ d F2 ¶F1 øi ¶F2 øi è è (11) (12) in which i represents the ith cycle of iteration. To update the values of F1 and F2. By making use of the relationship between elemental nodal-force and incremental-displacement. The infinitesimal variations in H and Vcan thus be approximated by their first order differentials as follows: plane containing the vector IJ and the force vector representing the self-weight of the element.TI 1 (TJ2 . Letting F1 and F2 be the nodal-force components at end I in YL and ZL directions respectively and F3 and F4 be those quantities at end J in YL and ZL directions respectively.

32 Ying-Hui Lei and Yu-Lin Chien é C -C ù [K ] = ê ú ë -C C û The matrix C is in the following form: é F1 2 2 ê . where l = sin b .2. scy and sty are the yielding stresses corresponding to tensile and compressive deformations respectively and a is called strain-hardening parameter. Strain distribution along element depth. one can get the strain e at a distance h from the centroidal axis of the cross section considered by using the following equation: e = Yh + e0 (18) where Y is the curvature of the cross section and e0 is the strain along centroidal axis. having recognizing that the neutral axis won’t be coincident with the centroidal axis due to the existence of internal axial force in the current cases. the stress at any depth of cross section can be written as ìa E (Yh + e 0 + e y ) . 2. m = cos b (17) and b is the inclination of the local YLZL-plane with the global YZ-plane. Furthermore. Model of elastic-strain hardening.a1l ê =ê ê ê SYM . To account for the material nonlinearity. measured in the same direction as that of the end-restraint spring. as illustrated in Figure 5.5 times the stiffness of a single tower.1 l 2 .e y ) + Ee y (21) . hty = 1 D (e y . The end-restraint springs in Figure 1 are used for modeling the tensile restraints provided by a series of transmission towers located beyond either end of the system.1 Formulation for Material Nonlinearity The component members of each transmission tower in the system will be modeled by using the 3D beam-column elements.2 Tower Structures 2. In the current study. the stiffness of any end-restraint spring placed along certain direction will be chosen as 1.Ee y (h £ hcy ) ï s = í E ( Yh + e 0 ) (hcy £ h £ h ty ) ï (h ³ h ty ) îa E (Yh + e 0 . one can obtain the unyielding depths corresponding to tensile and compressive deformations.a1m H F . Observing the strain distribution along the depth of the beam-column element.e 0 ) £ Y 2 1 D hcy = (-e y . the stress-strain relationship of all these elements will be assumed following the relationship as exhibited in an elastic-strain hardening model (referring to Figure 4). denoted as hty and hcy respectively. Figure 5. ê ê ë F4 lm . In Figure 4.e 0 ) ³ Y 2 (19) (20) Accordingly.H m . by making use of the relations in the following.a1m 2 H (15) [C ] 3´3 ù -a 2l ú ú -a 2 m ú (16) ú -a 4 ú ú ú û Figure 4.

With further calculations.y . namely in the case of yet £ yec and yet £ y £ ytc: in which B is the width of cross section.(1 .e y (24) ë + (1 . To make the expressions of s be more concise.÷ ç e ÷ y ø è ù ü (26) e0 ö ï 2 úý úï ûþ 1 .p ) + ay ] (33) 1.e0 ) 2 (2y + 1 + e0 ) 4y 2 æ e0 ç1 ç e y è é Yy æ e ê1 + ç1 .a)(1 . (26) and (27) leads to p = e0 + m=y- (1 .e y ù dh û The subscript y in Eq. Eq. (29) simultaneously gives p-m-y relationships in various conditions: (i) The tensile and compressive zones are all in initial elastic range.e0 4 ë (1 . with the use of the definition given by Introducing the dimensionless parameters p.Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities 33 in which E is the modulus of elasticity.a) é y .a ) Yh + e 0 . and e y is the yielding strain.a ) -Yh . y and e 0 into Eqs.y M = EI Y í1 2 Y ï î Yy (1 .e y ù h dh û (25) (ii) The compressive zone is all in initial elastic range whereas a portion of the tensile zone has been deformed into the subsequent strain-hardening range.1 + e0 in which (2y + 1 . (21) can be rewritten as s = E é Yh + e 0 .e 0 . the above two equations can be expressed in more concise forms: D ì P = EA íe 0 + (1 .a ) Yh + e 0 .(1 .e 0 . and also equating the total moment M with the resultant of the components for flexural force leads to P=ò D/2 -D/ 2 s dA = B ò D/2 -D/ 2 D/2 -D/ 2 s dh = BE ò é(Yh + e 0 ) .2 y[(1 .a ì Y ï (1 . (30) is used to denote the state of yielding. Eliminating the parameter e 0 by solving Eq.Y y ç1 + e 0 ç e ê y è ë ö ÷ ÷ ø 2 m=ywhere e0 = (1 . m.e0 4j 2 ë 2 2 .Y .0 ê 2Y ç e y è ë öù ü ï ÷ú ý ÷ú øû ï þ (34) (27) where .a ) -Yh . e0 = 0 py My Yy ey (30) Equating the total axial force P with the resultant of the components for axial force.e0 ) ù û Consequently.e0 ) 2 4y (32) æ .e y ù û (23) p= e p M Y .a ) Yh + e 0 . m= .a) é y .e y ë + (1 .y .1 .Y y ç1 .1 .a ) 8Y î é æ ê Y .a ) -Yh .a + y + ay .(1 .a ) 12 Y æ e0 ç1 + ç è ey ö ÷ ÷ ø ö ÷ ÷ ø 2 2 é Yy æ e ê1 ç1 + 0 ç ê 2Y è e y ë öù ÷ú ÷ú øû (iii) The tensile zone is all in initial elastic range whereas a portion of the compressive zone has been deformed into the subsequent strain-hardening range.a) (y .1 + e0 ) 2 (2y + 1 .1 + e0 ù û (28) 2 S ì S (0 £ S ) = í î0 (0 ³ S ) (2y + 1 + e0 ) (29) 2 (22) . y= . namely in the case of yec £ yet and yec £ y £ yct: m=y(1 . (28) and Eq.a) (y .e 0 .e y ë + (1 . s is the normal stress corresponding to e 0. the symbol áSñ is introduced. namely in the case of y £ yet and y £ yec: m=y (31) M =ò D/2 -D/ 2 sh dA = B ò D/2 -D/ 2 D/2 -D/ 2 sh dh = BE ò é(Yh + e 0 ) .a ) 1.1 .

8 0.8 (c) When a portion of the tensile zone has come to yield already but the compressive zone is at the moment just beginning to yield.ay + 2 y[(1 . It can be seen that the variation of the slope for the curves in the figure consists of three main stages: (1) a considerably large value of slope kept constant until the outer fiber of the section is on the edge of yielding (2) a medium value of slope varied gradually until the whole section has come to yield (3) a value of slope kept being zero lastingly. namely in the case of ytc £ y or yct £ y: m=y(1 .y . .4 p = 0. the curvature of y = Y/Yy Figure 6. would be yet = 1 .2.p ) + 4 pa 2 y M / M = m (40) 0 0 1 2 3 4 5 6 7 8 9 10 (d) When a portion of the compressive zone has come to yield already but the tensile zone is at the moment just beginning to yield.6 0. (a) When the whole compressive zone is still in elastic range but the tensile zone has been just beginning to yield.a)(1 . a Cartesian coordinate system composed of x-.2 Formulation for Geometric Nonlinearity In taking the effect of geometric nonlinearity into account. y.1 + e0 ) (2y + 1 . 2. denoted as yct. yec.a + ay (37) In the above discussion.6 1.a) (1 + p ) . corresponding to certain axial force. denoted as yec. will be adopted as the flexural rigidity of a single beam-column element. would be m = M / My yec = 1 + p (39) 1. each of the boundary curvatures yet.a) é(y .1 .34 Ying-Hui Lei and Yu-Lin Chien e0 = -1 + a . would be y tc = 2(1 .2 p = 0. Since the slope of any m-y curve is also the flexural rigidity of the cross section. The longitudinal direction of the element will be as1. the curvature of the cross section.p (38) Figure 6 shows the m-y relationships under various values of p with a = 0.p ) + ay ] (35) 1.2a + (1 + p ) 2 .6 p = 0.4 1. would be y ct = 2(1 .p ) . the curvature of the cross section. denoted as ytc.4 0. m-y relationship (a = 0).e0 ) ù û 2 (41) (36) where e0 = py 1 .a) (1 .2 (b) When the whole tensile zone is still in elastic range but the compressive zone has been just beginning to yield.axes as illustrated in Figure 7 is adopted as the local coordinate system of the beam-column element. denoted as yet. a modified flexural rigidity.e0 ) 2 (2y + 1 + e0 ) 4y 2 ë + (y .a the cross section.2 p = 0. EI.4 pa (iv) Both a portion of tensile and a portion of compressive zones have been deformed into the strain-hardening range. (EI)ave.and z.0 p = 0.2a + (1 . the curvature of the cross section. obtained by taking the average of the sum of the rigidities corresponding to five equally spaced sections. ytc and yct will be associated with a specific stress distribution as described in the following.0 0.

y. u10.Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities 35 sumed to be coincident with the x-axis. the nodal force corresponding to the nodal displacement ui (i = 1. y. …. 2) ê [k ] = ê L L ê L ê L ê k (12. 7) = k (2. u2 and u3 represent the translational displacements at the left end of the element in x-. 3. G is the shear modulus of elasticity. that is the effect of the axial deformation caused by flexural force.f sin f ) y y y (54) Figure 7. u4..f y sin f y ) (52) S2 y = (53) S3 y = ˆ ˆ ˆ ˆ f y (sin f y .5) = k (11.and z-directions respectively. 6) = k (12.12) = ( EI y ) ave L S4 y (50) (51) k (1.10) = -k (4.1) = k (7.f y sin f y ) ˆ ˆ ˆ (2 . u12 and u6 represent the rotational displacements at the right end of the element around x-. are derived to be: (i) When the internal axial force is in compression (F1 = -F7 > 0): S1 y = ˆ ˆ f3 sin f y y ˆ ˆ ˆ (2 . into account. …. S4 y = ˆ ˆ ˆ ˆ f y (f y . J is the polar moment of inertia. y.12) ù L k (2.8) = -k (2. Besides. u5 and u6 represent the rotational displacements at the left end of the element around x-. In Figure 7. 2) = k (8. 6) = k (2.and z-directions respectively. u7. 2) ë L L L L L L k (1.cos f y ) y ˆ ˆ ˆ (2 .9) = -k (9.2 cos f y . u1.2 cos f y . Nodal displacement of 3-D beam-column element.12) = -k (6. 4) in the above equations. y.and z-axes respectively.f y sin f y ) (55) .and z-axes respectively.and z-axes respectively. 2. the stiffness coefficient k(i. u8 and u9 represent the translational displacements at the right end of the element in x-.10) = k (5.8) = EA Rt L ( EI z ) ave S1z L3 (43) (44) where E is the modulus of elasticity.11) = k (6. 12) which includes the effect of both material and geometric nonlinearities can be obtained as follows: k (1. A and L are the cross-sectional area and length of the beam-column element respectively.f y cos f y ) ˆ ˆ ˆ (2 . and the two principal axes perpendicular each other over the cross section would be set in y.11) = k (6.11) = -k (5.5) = k (3.11) =( EI y ) ave L2 S2 y (46) k (4. used for taking the effect of the interaction between axial and flexural forces into consideration.and z-directions respectively. The stability functions Sjy and Sjz (j = 1.2 cos f y . k (2. 2) é k (1.8) = -k (8. The stiffness matrix of the element in Figure 7 would be in the form as written in the following.12) =( EI z ) ave S2 z L2 (45) k (3.12) ú û ( EI z ) ave S4 z L With the application of the beam-column approach.1) ê k (2. 2.1) k (12. j) (i.1) k (2. 2. j = 1. Iy and Iz are the moments of inertia corresponding to y. 7) = -k (1.2 cos f ..12) = GJ L (47) ( EI y ) ave L S3 y (48) (49) ( EI z ) ave S3 z L k (5.f y sin f y ) ˆ ˆ f3 (1 . Rt is the axial stiffness coefficient used for taking the bowing effect. 12) will be denoted by using the symbol Fi.12) ú ú L L ú (42) ú L L ú L k (12. 4) = k (10.

Flowchart of acquiring the nonlinear stiffness matrix.36 Ying-Hui Lei and Yu-Lin Chien Figure 8. .

all the investigations regarding to the internal-force variation of structural elements will be mainly focused upon the four leg members. As indicated in Figure 1.54 Region III Leg Diagonal Member Member 3.26 ´ 10-6 686.75 ´ 10-3 4.05 ´ 105 --1.50 ´ 10-3 2.29 ´ 10-7 1.75 ´ 10-3 2. Sectional properties of tower members & electric wires Sectional Parameter A (m2) E (N/m2) G (N/m2) P (kg/m3) r (kg/m) Iy or Iz (m4) J (m4) Py (N/m2) Mpy or Mpz (N-m) Region I Leg Diagonal Member Member Region II Leg Diagonal Member Member 2.90 ´ 104 --1.69 ´ 1010 7. Figure 10 shows the structural pattern of each tower in the system.69 ´ 1010 7.sinh f y ) ˆ ˆ ˆ (2 .1) y ˆ ˆ ˆ (2 .16 ´ 10-7 1. The electrical wires suspending between two towers is prescribed by a length of 360 meters and will be modeled by using four cable elements aforementioned.850 -5.500 8.00 ´ 1011 7.03 ´ 10-6 622.2 cosh f + f sinh f ) y y y (58) (59) S4 y = ˆ ˆ ˆ ˆ f y (f y sinh f y .062 ----Figure 9.00834 g is taken as the ground excitation acting on the structural system. each of the structural systems investigated in the following will include three tower structures. Figure 9 shows the Fourier spectrum for the acceleration.00 ´ 1011 7. Further- 3.850 ---1.00 ´ 1011 2.69 ´ 1010 7.000 18.993.116. in which the towers M and N are located at two ends. In undertaking the seismic time-history analysis. Leg B. recorded at Table 1.69 ´ 10-4 8. Numerical Examples The acceleration of Ji-Ji earthquake.850 7.69 ´ 1010 7.59 ´ 10-4 1.40 8.54 ´ 10-6 3.628 ----Ground Wire 1.30 ´ 10-3 2.500 1.63 ´ 10-6 1.97 2. (43)-(51) repeatedly at each time step until the attainment of convergence.04 ´ 10-6 874.850 -6.69 ´ 1010 7.02 ´ 10-6 2.647. (52)-(55) & (57)-(60) with Iz leads to the shifting of the quantities on the left-hand side of these equations from Sjy (j = 1.82 ´ 10-6 6.47 ´ 10-3 2.99 Conduct Wire 4.3.760. Newton-Raphson method will be adopted as the iterative scheme for modifying the stiffness coefficients in Eqs.69 ´ 1010 7.2 cosh f y + f y sinh f y ) (60) Replacing Iy in Eqs.500 686.653. named Leg A.49 ´ 10-3 2.54 17.26 ´ 10-6 3.233.09 ´ 10-6 1. Table 1 specifies the cross-sectional properties of tower members and wires. Taiwan. and the tower O is located in the middle of the system. The procedure for acquiring the nonlinear elemental stiffness matrix is illustrated by the flowchart shown in Figure 8. Fourier spectrum of acceleration.850 7. TCU084 station in Taichung.2.00 ´ 1011 7. at the bottom of tower N.076.750 3.f y ) ˆ ˆ ˆ (2 .2 cosh f y + f y sinh f y ) ˆ ˆ ˆ ˆ f y (f y cosh f y . 4.4) to Sjz. The distance between adjacent towers is chosen to be 350 meters.71 .250 7.Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities 37 in which ˆ fy = PL2 EI y (56) (ii) When the internal axial force is in tension (F1 = -F7 < 0): S1 y = ˆ ˆ f3 sinh f y y ˆ ˆ ˆ (2 . with the peak value of 1.29 ´ 10-7 1.2 cosh f y + f y sinh f y ) (57) S2 y = S3 y = ˆ ˆ f3 (cosh f y .00 ´ 1011 7.00 ´ 1011 2. For simplification of calculations.850 -1.993. Leg C and Leg D respectively (referring to Figure 11).

. the term “input angle of seismic force”. Figure 11. the beam-column element exhibiting an elastic-perfectly plastic behavior (that is the case of a = 0) and having a cross section in rectangle will reach its ultimate strength if the interaction-strength equations expressed in the following are satisfied [8]. will be introduced in this study. measured counterclockwise from the positive direction of X-axis. Structural pattern of transmission tower. Figure 13. but by the combined action resulting from various kinds of internal forces including axial. 3. seismic analy- 3. In actual calculation. is defined as the angle of seismic force. denoted by using the symbol 8. To describe the position of tower N relative to that of towers M and O in a tower group.2 Failure Index The failure of the transmission tower can be triggered by the fracture of any component main-member in the structure.38 Ying-Hui Lei and Yu-Lin Chien more. Description of horizontal angle 2. Figure 12. as illustrated in Figure 11. the parameters of q and f. called horizontal and vertical angles respectively. 8 will range from zero to 180 degrees with the angle increment of 15 degrees. Figure 10.1 Configurations of Transmission Towers In the following numerical examples. flexural and shear forces. This member fracture is usually not caused by the action of single type of internal force. in order to investigate the load effects due to the acting direction of earthquake. Legs at tower bottom and input angle of seismic force. ses will be performed with respect to a group of towers exhibited in various configurations. Description of vertical angle f. Figures 12 and 13 show the illustration of these parameters. According to the theory proposed by Chen and Atsuta.

Results and Discussion (63) To inspect the effect of wire action on the dynamic behavior of transmission towers. and each point on the surface would correspond to an identical value of 7 = 1. 3. mz = z Py M py M pz (64) in which P is the axial force acting on the cross section. In addition. the quantities on the left-hand side of Eqs.and z-axes respectively. m y = y . (61)-(63) will be given the name “failure index”. which is denoted by using the symbol 7. Figure 14. the shaded area in the figure would represent the interaction plane associated with the loading conditions in which the axial ratio p has a fixed value of 0. Failure surface of beam-column elements.2(1 . the dimensionless parameters of p. it is assumed that the global failure of a transmission tower arises whenever the failure index corresponding to any component member of this tower is found equal or greater than unity. my and mz are defined. It is observed in Figures 16 and 17 that with a few exceptions only.p ) û ë û (61) (62) tion. since the peak value of 7 for Leg D is found greater than unity.p ) : 3 3 3 p 2 + m y + mz2 £ 1 4 2 2 (ii) m y £ (1 . It is observed that the difference between the curves corresponding to these two cases is considerably significant. the variation of 7 for Legs C and D in the cases of the structures being either connected or not connected with electrical wires (Both wire action and wire mass are neglected in the latter case) is shown in Figure 15. 4. Since being a useful index in the judgment of member fracture. Py is the axial yielding strength. N and O would lie on the XY-plane and be symmetric with respect to the X-axis (referring to Figure 13). . so the investigation of 7 will be focused upon Legs C and D only. the larger the value of q is the larger the failure index will usually be. In undertaking the seismic analysis for the cases described in numerical examples.p ) and mz £ (1 . Accordingly. My and Mz are the applied moments around the principal y.and z-axes respectively. In Eqs.Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities 39 2 2 (i) m y ³ (1 . Effect on failure index caused by cable action (2 = f = 0°). respectively. the bases of towers M. as p= M P M .5.p ) and mz ³ (1 .p ) ú £ 1 4 ë 2(1 . the fracture of this member can then be expected according to the interpretation of the interaction equations in Sec. Mpy and Mpz are the ultimate plastic moments around the principal y.p ): 3 3 my ù é 9é mz ù p + ê1 ú ê1 . (61)-(63). In addi- 4.1 Variation of 7 due to q In the condition of f = 0°. the curved surface indicated in Figure 14 will be the enveloped surface of fracture.p ) : 3 3 3 2 p 2 + m y + mz £ 1 4 2 2 (iii) m y ³ (1 .2.p) and mz ³ (1 . Figure 15.

versus input angle of seismic force. 4. the failure index corresponding to 8 = 8* for Legs A and C would be equal to that corresponding to 8 = B . Figure 17. This implies that under certain prescribed value of f. the larger the peak failure index and the higher the probability of member fracture will usually be. is presented in Figures 24 and 25. To get a better understanding to the variation tendency of the curves in Figures 20-23. it is noticed that the peak failure index corresponding to f = 0° for Leg B would be larger than those corresponding to other values of f. In addition.40 Ying-Hui Lei and Yu-Lin Chien Figure 16. which passes through Leg B or C.2 Variation of 7 due to f In the condition of q = 0°. 7ave. as indicated in Figures 23 and 24. of the base of tower N.8* for Legs B and D respectively (referring to Figures 11-13). It is observed in the first two figures that under a varied q and a prescribed f. Figure 18. Figure 19. Figures 20-23 are presented for this purpose. Failure index of Leg C with q = 0° under varied f. Failure index of Leg D with f = 0° under varied q. the peak failure index would occur at the time when the seismic force is parallel to the diagonal. 8. however. a plot of average failure index. this specific value of f will be shifted from 0° to 40° for Leg C. Failure index of Leg B with q = 0° under varied f. It is observed in Figures 18 and 19 that no matter what value the parameter f is. in which the ordinate coordinate of any point on a curve is obtained by taking the average of the failure-index values corresponding to an identi- . Failure index of Leg C with f = 0° under varied q. the larger the value of f is. the peak values of failure index on each curve would occur at 8 = 135° and 45° for Legs B and C respectively. so the investigation of 7 will be focused upon Legs B and C only. Similar phenomenon will also be observed under a varied f and a prescribed q.3 Variation of 7 under Various Combinations of f and q To investigate the safety of tower members under all kinds of configurations of transmission towers. 4.

Figure 22. Conclusions Having undertaken the seismic analysis in consider- Figure 20. . Average failure index of Leg C with q = 15° and 45°. Figure 23. Average failure index of Leg C with f = 10° and 40°. Failure index of Leg C with f = 10° under varied q. the phenomenon observed in both figures matches fairly well with that observed in Figures 20-23. Figure 24. Figure 21. Failure index of Leg C with q = 45° under varied f. Figure 25. 5. Failure index of Leg C with q = 15° under varied f. Failure index of Leg C with f = 40° under varied q. Except for a few cases.Seismic Analysis of Transmission Towers Considering Both Geometric and Material Nonlinearities 41 cal 8.

2. Vol. Finite Element Analysis of Latticed Transmission Towers. B.. 120. Vol. 14.” Journal of Structural Division. Vol. 2004 . F.” Journal of Structural Engineering. Similar phenomenon will also be observed under a varied f and a prescribed q. “General Solution of Suspended Cable Problems. H. G.” Thin-Walled Structures.. pp. one may recognize that ignoring the wire-action effect on the dynamic behavior of transmission towers may cause significant errors in the results of member forces. Theory of Beam-ColumnsSpace Behavior and Design Vol. and Knudson W. A. the larger the value of f is.” Computers and Structures. 698-712 (1993). G. pp. (1977). G. M. Under a varied q and a prescribed f. A. pp. McGraw-Hill. 167-177 (2003). U. the peak failure index would occur at the time when the seismic force is parallel to the diagonal. 41. 15. 19. in the condition of q = 0°.” Journal of Structural Division.. of the tower base. pp. H. NSC 93-2211-E-032-008. Acknowledgment The work has been supported by research grants made by the National Science Council (Taiwan. F.A. R. [6] O’Brien.. C. pp. [7] Peyrot. “Effect of Bolt Slippage on The Ultimate Strength of Latticed Structures. and Atsuta. with respect to the structural system modeled by using beamcolumn elements. S. which passes through the leg member considered. and Kitipornchai. Vol. cable elements and end-restraint elements. T.S.. [8] Chen. T. Republic of China) under grant No. In the condition of f = 0°. G. New York.. 93. [4] Albermani. [5] Jayaraman. AlBermani. A. “Joint Effects on Behavior of Transmission Towers. 2004 Accepted: Nov. and Santhakumar A. F. 1-26 (1967). S. H. Vol. [2] Knight. S. A. “A Curved Element for the Analysis of Cable Structure. and Kitipornchai S. 325-333 (1981). [3] Kitipornchai. “Nonlinear Manuscript Received: Aug. “Analysis of Flexible transmission lines. pp. and Peyrot A. 259-269 (1993). 5.” Engineering Structure. 2281-2287 (1994). 763-779 (1978). Vol. On the other hand. 119.” Journal of Structural Engineering. W. Vol. 104.42 Ying-Hui Lei and Yu-Lin Chien ation of both material and geometric nonlinearities. M.. the larger the peak failure index and the higher the probability of member fracture will usually be. pp.. F.. and Goulois A. “Numerical Simulation of Structural Behaviour of Transmission Towers. the larger the value of q is the larger the failure index will usually be. References [1] Albermani.