You are on page 1of 16

16

Lumped versus Distributed Parameter Systems


16.1 Procedure of Analysis 16.2 Continuous Mass Matrix Method
Member under Axial Displacement Member under Bending along Its Major Moment of Inertia Axes Dynamic Member Stiffness Matrix for Plane and Space Frames

16.3 16.4 16.5 16.6 16.7 16.8

Consistent Mass Matrix Method Lumped Mass Matrix Method Free Vibration of Frames Forced Vibration Practical Applications Linear Structures without Additional Effects
Single Beams Frames Single Beams Frames Offshore Structures

Bulent A. Ovunc
University of Louisiana at Lafayette

16.9 Linear Structures with Additional Effects

The lumped, consistent, and distributed (or continuous) mass methods are the main methods for dynamics and vibration analyses of structures. In the continuous mass method the equations of motion are satised at every point of the structure. In the consistent and lumped mass methods they are satised only at the joints of the structures. In consistent mass the displacements within members are assumed as static displacements. The lumped mass method considers the members as massless springs. The lumped and consistent mass methods are simple and fast; they are fairly approximate, but their accuracy decreases for structures subjected to the effects of the shear and rotatory inertia, member axial force, elastic medium, and so on. The continuous mass method provides accurate results under the assumptions made.

16.1 Procedure of Analysis


For frames, the general formulation is based on the vector of displacement {ao( y, t)} at a time t and at a point y on the center line of its constitutive members. The same formulation is valid for the above mentioned three methods, only the vector of center line displacement {ao( y, t)} depends on the assumptions made for each method. For materially and geometrically linear frames, the vector {a0( y, t)} is made of four independent components,

2005 by CRC Press LLC

{ao(y, t)}T = [u(y, t) v(y, t) w(y, t) J( rQ, t)]

(16.1)

where u(y, t), v(y, t), and w(y, t) are the displacements along the member axes x, y, z; J(rQ, t) is the twist rotation about the y axis, and rQ = (x2 + y2)1/2. It is assumed that the material properties are independent of time, and that the external disturbances applied to a structure are proportional to a same-time variable function. Thus, the displacement function {ao(y, t)} can be written in separable variable form. The integration of the differential equations and elimination of the integration constants gives {ao( y, t)} = [N( y)]{d(t)} = [N( y)]{d}f(t) (16.2)

where {d(t)} = {d}f(t), [N( y)], and {d} are time-independent matrix-of-shape functions and vector-ofmember displacements, and f(t) is a time-variable function of external disturbances. See Figure 16.1. The strains {e } at a point within the cross-section of a member can be obtained from the center line displacements {ao( y, t)} [Przemieniecki, 1968] as {e} = []{ao( y, t)} = [B( y)]{d}f(t) where [B( y)] = [][N( y)], and []is the matrix of differential operators. The stresses {s } are determined from the stressstrain relationship as {s} = [E]{e} = [E][B( y)]{d}f(t) (16.4) (16.3)

The expressions of the strain energy, Ui, and kinetic energy, , as well as the work done by damping forces, WD, and by externally applied loads, We, are written as Ui = 1 1 {e}T {s}dV = ( f (t ))2 {d}T [B( y)]T[E ][B( y)]dV {d} v 2 v 2

o }T {ao }dV = -cf (t ) f (t ){d}T WD = - c{a v

[N(y)] [N(y)]dV {d}


T v

1 = 2

1 o } {a o }dV = ( f (t ))2 {d}T m [N ( y)]T (N ( y)]dV {d} m{a v 2 v


T

(16.5)

We = fe (t ) {Po }T {ao }dy = fe (t ) y

{P } [N(y)]dy {d}
T y o

where fe(t) and {Po} are the time-dependent and -independent parts of the vector of externally applied joint forces. For a member, stiffness [k], mass [m], and damping [c] matrices are determined by substituting the strain energy Ui, damping energy WD, kinetic energy , and external energy We into the Lagrangian dynamic equation [Ovunc, 1974], U i WD d K We + = d j d j dt d j d j which provides the equation of forced vibration for a member as (t ) = {P } f (t ) [k ]{d} f (t ) - 2n o[m]{d} f (t ) + [m]{d} f o e (16.7)

(16.6)

where the damping matrix [c ] is assumed to be proportional to mass matrix [m] and no is the damping coefcient.
2005 by CRC Press LLC

z, d9, f9 d12, f12 Z z, d3, f3 d6, f6 y, d8, f8 d11, f11 x, d7, f7 d10, f10

y, d2, f2 d5, f5 X d4, f4

x, d1, f1

FIGURE 16.1 Coordinate axes systems.

For the free vibration, {P} = 0, the equation of motion is divided into two parts time-independent and time-dependent: ([k] - w2[m]){d} = 0 d 2 f (t ) df (t ) + 2nw + w 2 f (t ) = 0 dt dt 2 (16.8)

(16.9)

The time-dependent part, f(t), Equation (16.9), is the same for all four independent cases, and n = now [Paz, 1993]. Although the lumped mass method was developed long before the continuous mass method, the continuous mass method is herein explained rst. The consistent and lumped mass methods are presented as particular cases of the continuous mass method.

16.2 Continuous Mass Matrix Method


For the materially and geometrically linear frame, an arbitrary vibration of its member is obtained by combining the four independent components: axial displacement, torsional rotation, and bending in two orthogonal planes.

Member under Axial Displacement


For members with constant section, the time-independent part of the differential equation of free vibration has been given [Ovunc, 1985]. (See Figure 16.2.) d 2Y ( y) + a 2Y ( y) = 0 dy 2

(16.10)

where a2 = (w2m - Cfp)/EA; and where m = (Ar + q)/g, p, and cf. are the mass, the peripheral area, and the friction coefcient of the elastic medium per unit length of the member, respectively. The time-dependent part, f(t), is the same for all four independent cases.
2005 by CRC Press LLC

z, w z, w q B B P A Ct (pdy)v dy A dy 2y g t2 B P P+ dy y y, v A A M V
2 A dy w g t2

q B M+

M dy y y, v

V V+ dy y A Cs(pdy)w dy

FIGURE 16.2 Axial force member.

FIGURE 16.3 Bending member.

The time-independent part of axial displacement function Y(y) is obtained by integrating Equation (16.10), through the elimination of the integration constants {C}, by the boundary conditions; thus one has Y ( y) = {f ax ( y)} T[L]-1 {dax } = {N ax ( y)} T {dax } The nature of the shape function {Nax( y)} depends on the sign of parameter a2. {N ax ( y)} T = (1 / sin a l){(sin a l cos a y - cos a l sin a y) sin a y} for a 2 > 0 sinh a y} for a 2 > 0 (16.11)

{N ax ( y)} T = (1 / sinh a l){(sinh a l cosh a y - cosh a l sinh a y)

Member under Bending along Its Major Moment of Inertia Axes


For members with constant section, the time-independent part of the differential equation of free vibration (see Figure 16.3) has been given by Ovunc [1985] as d4Z d2Z + 2k 2 2 - b 4 Z = 0 4 dy dy (16.12)

where 2k2 = P/EI, b4 = (w2m - CsP)/EI; and where I, Cs, and where p are the moment inertia of the cross-section, the subgrade coefcient of the elastic medium, and the projected area of the cross-section, respectively. The shape function {Nbd(y)} due to deection, Z(y) = {fbd(y)}T[L}-1{dbd} = {Nbd(y)}T{dbd } (16.13)

is obtained in a similar manner as in the case of axial displacement, Equation (16.11). The nature of the 2 and a 2 shape function {Nbd(y)} and its component {fbd(y)}T depend on the parameters a 1 2 , which are 2 4 expressed in terms of k and b : a1 = [(b4 + k4) 1/2 + k2] 1/2, a2 = [(b4 + k4) 1/2 - k2] 1/2 Thus, for b4 > 0 and P > 0 (compression is positive), {fbd(y)}T = (sin a 1y cosa1y sinh a2y cosh a2y) (16.14)

The above expression remains the same when the axial force P is tension, except a1 and a2 must be interchanged. For the combination of b4 < 0 and P < 0 or P > 0, the expression of {fbd(y)} can be determined in a similar manner.
2005 by CRC Press LLC

The member stiffness matrices for the twist rotation and the bending in the Oxy plane are obtained by following similar steps as in the previous cases. The stiffness matrix for the space frame is determined by combining the stiffness matrices of all four independent cases.

Dynamic Member Stiffness Matrix for Plane and Space Frames


The dynamic member stiffness matrix {kdy] for either a plane frame or a space frame is obtained by substituting either the shape functions for axial displacement and bending in Oyz plane [Equation (16.11) and Equation (16.13)] or the shape functions of all the four independent cases, in the Lagrangian dynamic equation [Equation (16.6)]. Integrating [Equation (16.7)] one has [kdyn] = [k] - w2[m] (16.15)

The continuous mass method has also been extended to frames with tapered members [Ovunc, 1990].

16.3 Consistent Mass Matrix Method


In the consistent mass matrix method, the deformations within a member are static deformations. The shape function for each independent component {N( y)} is a static displacement due to its corresponding independent cases. Thus, the shape functions for axial displacement, {Nax( y)}, and for bending in the Oyz plane, {Nbn(y)}, are given as {Nax( y)}T = ((1 - h) h) {N bn( y)} T = ((1 - 3h2 + 2h3 ) (h - 2h2 + h3 )l (3h2 - 2h3 ) (- h2 + h3 )l) (16.16) (16.17)

where h = y /l. The shape functions for twist rotation and bending in the Oxy plane are obtained in similar manner. The member stiffness [k] and mass [m] matrices are evaluated by substituting the shape functions in the Lagrangian dynamic equation [Equation (16.6)]. Herein, the stiffness matrix [k] is a static stiffness matrix and the mass matrix [m] is a full matrix [Przemieniecki, 1968; Paz, 1993]. Moreover, the member stiffness matrix [k] and the mass matrix [m] for the consistent mass matrix method can be obtained as the rst three terms of the power series expansion of the dynamic member stiffness matrix [kdyn] for the continuous mass matrix [Paz, 1993].

16.4 Lumped Mass Matrix Method


The lumped mass method is obtained from the continuous mass matrix method by considering the limit when the mass of the members tends to zero. Thus, the shape functions [Equation (16.17)] and the member stiffness matrix [k] are the same as in the consistent mass method. But the mass matrix [m] is diagonal [Paz, 1993].

16.5 Free Vibration of Frames


The stiffness coefcients for the frames are evaluated from those of its members as K i,j =

r ,s

Mi,j =

r ,s

Pi =

where r and s are the member freedom numbers corresponding to the i and j of the structure freedoms. The equation of free vibration is obtained from those of members [Equation (16.8)]
2005 by CRC Press LLC

([K] - w2[M]){D} = {0}

(16.18)

where {D} is the vector of the structure displacements. The natural circular frequencies, wi, and the corresponding modal shapes, {Di}, are calculated from the equation of free vibration [Equation (16.18)], also called the frequency equation.

16.6 Forced Vibration


The second-order differential (n simultaneous equations with n unknowns) (t )} + 2n [M ]{D (t )} + [K ]{D(t )} = {P(t )} = {P } f (t ) [M ]{D o o e (16.19)

is converted to n separate, second-order, single-variable differential equations through the two orthogonality conditions, which proves that each modal shape vector {Di} is independent of the others. The ith participation factor i is dened as the component of given forced vibration on the ith modal shape vector {Di}. Thus, any arbitrary motion can be determined by considering the summation of its components i on each modal shape vector {Di} [Clough and Penzien, 1993; Paz, 1993; Ovunc, 1974]. If the time variable factor fe(t) of the external disturbance [Equation (16.2)] is periodic (pulsating), the forced vibration can be directly determined without the calculation of participation factors i [Paz, 1993].

16.7 Practical Applications


For any structure, the natural circular frequency w can be expressed in terms of a parameter C as [Paz, 1993] EI j m j l4 j

w =C

(16.20)

where E is the Youngs modulus, and Ij, mj, and l j are the moment of inertia, mass, and span length of a selected member j. Substituting the natural circular frequency w (its expression in terms of C) into the frequency equation [Equation (16.18)] gives -C 2[M ] + [K ] = 0 (16.21)

where the general terms Mrs and Krs of the mass and the stiffness matrix are constant and expressed as M rs = M rs /m j l j and K rs = K rs l j / EI j (16.22)

It can be easily seen that the determinant of the frequency equation [Equation (16.21)] is independent of the member characteristics (E, m, l, I) but depends on the parameter C. If, in a frame, the same characteristics of the members are multiplied by the same factor, the magnitude of the parameter C remains unchanged. However, a natural circular frequency wi corresponding to Ci changes [Equation (16.20)]. If the characteristics of some members change and those of the others remain constant, the parameter Ci is affected. The advantages of one method over the others and the limits on their accuracy depend on the number and type of the members in the structures and whether the structure is subjected to additional effects. The type of member depends on the ratio of the thickness t (of its cross-section) to its span length l: g = t /l.

2005 by CRC Press LLC

If the ratio g = O(1/100) (in the order of 1/100), the effect of bending is negligible. The member is a very thin member, called cable. If the ratio g = O(1/10), axial force, torsion, and bending are affecting the member. The member is a thin member. If the ratio g = O(1), the member is considered a deep beam. The inclusion of the variation of the width or the thickness of the member, the effect of member axial force, shear, rotatory inertia, vibration of the member within an elastic medium, and so on constitutes the additional effects. The dynamic responses of beams and frames are evaluated with or without the additional effects by lumped and consistent mass methods. The results are compared with those evaluated by continuous mass.

16.8 Linear Structures without Additional Effects


The behavior of a structure is linear when its stresses are within their elastic limits and its deformations are innitesimal. The rst one constitutes the geometrical linearity, the second is the material linearity.

Single Beams
The data related to a cantilever beam, considered as an example, composed of a single element or with two, three, or ve subelements, are given in Figure 16.4. The dynamic responses of each beam have been determined by the lumped mass and the consistent mass matrix methods. The rst two natural circular frequencies, w1 and w2, and the vertical displacement dv11, rotation dr11, and shearing force V11 at the free end 1 as well as the bending moment M21 at the xed end 2 due to the rst mode w1 are furnished in Table 16.1 for the lumped mass method, for the consistent mass method, and for the continuous mass method. In the lumped mass method, increasing the number of subelements improves the accuracies of the natural circular frequencies w 1, w2, and wi and those of the vertical displacements, dv11, and rotations, dr11, when their magnitudes are compared to their magnitudes obtained by the continuous mass method. For the consistent mass method, the beam has better approximations when it is subdivided into two subelements.
Pofe(t) 1 /2 /2 (a) Single Element 2 1 Pofe(t) 3 /2 /2 (b) Two sub-elements 2

Pofe(t) 1 3 4 2 /3 /3 /3 (c) Three sub-elements

Pofe(t) 1 /5 3 4 5 6 /5 /5 /5 (d) Five sub-elements 2 /5

fe(t) 1. .2 t

Steel beam: W10 45, span length, = 15 ft External force: P(t) = Pofe(t) Po = 5 k

FIGURE 16.4 Cantilever beam formed by (a) a single member or (b) two, (c) three, or (d) ve subelements and subjected to a force P(t) = Po (t)fe(t).

2005 by CRC Press LLC

TABLE 16.1 Comparison of Analyses


Member w1 w2 dv11 101 dr11 102 V11(k) M21(k/ft)

Analysis by Lumped Mass Method Single Two Three Five 66.2265 85.3437 90.4627 93.3437 0.00 439.6537 510.7243 560.500 0.51534 0.39661 0.37075 0.35681 0.0 0.38314 0.34954 0.33090 9.490 1.515 1.061 0.625 71.714 30.179 29.104 28.548

Analysis by Consistent Mass Method Single Two Three Five 95.4062 94.9687 94.9687 94.9687 925.5823 595.0385 592.2191 590.7352 0.35050 0.34934 0.34882 0.34831 0.32200 0.32070 0.32023 0.32014 2.000 1.317 0.954 0.611 26.180 28.052 28.177 28.190

Analysis by Continuous Mass Method Single 95.0762 595.8353 0.34962 0.32080 1.312 28.025

weight = .788 k/ft P7(t) P5(t) P3(t) 1 1 30' 2 7 7 4 5 3 9 weight = 1.616 k/ft 6 weight = 1.728 k/ft 3 8 8 6 5 4 2 15' 10' 10'

Member 1, 2 4, 5, 7, 8 3, 6, 9

Size W10 45 W10 21 W24 84 W21 62 fe(t) 1. .2

External Forces P1(t) = P01fe(t) P01 = 5 k, P02 = 4 k P03 = 2.5 k

FIGURE 16.5 Three-story steel frame.

Frames
Three-story frames have been selected as an example to compare responses obtained by lumped, consistent, and continuous mass methods [Ovunc, 1980]. The data for the three-story frames are given in Figure 16.5. The dynamic responses of the three-story frame for three different ratios of a = Ib/Ibo are selected, where Ib and Ibo are new and actual moments of inertia of the beams. The moments of inertia Ibo are for thin members with a depth-to-span ratio g in the order of 1/10. Only the sizes of the beams have been varied. The frames rst two natural circular frequencies (w1, w2), horizontal displacements (dh71, dh72) at joint 7, and bending moments (M11, M12) at joint 1 are given in Table 16.2. The responses obtained by the lumped mass and consistent mass methods are close to each other for any magnitude of a (or g). However, they are roughly approximate compared to those obtained by the continuous mass method for low order of the ratio a = O(.01) (or g) that is, when the beams are very thin. The responses obtained by lumped mass, consistent mass, and continuous mass are very close for the actual or higher order of the ratio a O(1) (or g) that is, for thin and deep beams.

16.9 Linear Structures with Additional Effects


The additional effects included in the analysis of structures with linear behavior are the member axial force, soil structure interaction, and the additional mass of the members [Ovunc, 1992]. The parameters related to the member axial forces, soilstructure interaction, and additional masses of the members are
2005 by CRC Press LLC

TABLE 16.2 Dynamic Responses of a Three-Story Frame


Method Lumped mass a = Ib/Ibo 0.01 1.00 100.00 0.01 1.00 100.00 0.01 1.00 100.00 w1 2.9275 7.5657 8.2351 2.9099 7.5652 8.2429 2.1738 7.3964 8.2402 w2 10.9188 22.3703 23.7597 10.5168 22.3942 23.7951 10.0834 21.8749 23.7688 dh71(ft) 0.94994 0.11420 0.09400 0.95120 0.11440 0.09382 1.19336 0.11410 0.09370 dh72(ft) 0.02215 0.00105 0.00091 0.02279 0.00106 0.00092 0.00236 0.00106 0.00091 M11(k/ft) 163.871 81.711 78.053 162.792 81.672 78.032 228.538 81.849 78.110 M12(k/ft) 15.2073 1.2817 1.2048 15.5520 1.2886 1.2148 39.703 1.317 2.216

Consistent mass

Continuous mass

z, w B P P+ dy y y, v A (2 y)vdy dy A

B P

A + q dy d2v g dt2

FIGURE 16.6 Axial force element.

introduced in the differential equations of motion. The stability of the structures, including the additional effects, is also analyzed by means of the continuous mass matrix method. It is assumed that the material is linear elastic and the member axial forces are static. The compaction of the soil is due to deformations and the effect of pore pressure is neglected. Depending on the type of soil, the variation of the soil subgrade reaction is given as (Figure 16.6) [APJ RP-2A]: For cohesive soils, CS = kL (y/L)n k/ft3 For cohesionless soils, CS = k (y/d) k/ft3 where kL is the value of CS at the tip of the pile (y = L), n and k are empirical constants that depend on the type of soil, and d is the diameter of the pile. Under the axial displacements, the equation of motion of an innitesimal element of the member can be written as (Figure 16.7) C - C fb y P Cdv md 2 v - p C fb + ft dy dt 1 dt 2

(16.23)

where C and p are the damping coefcient and unit peripheral area of the pile and Cft and Cfb are the skin frictions at the top and the bottom of the pile. The mass per unit length of the pile is m = (ar + q)/g. By expressing the axial force P in terms of displacement v(y, t), and assuming the displacement v(y, t) can be written in separable variable form,
2005 by CRC Press LLC

z, w

M P

B V P

d2v A + q dy dt2 g

B M+

V V+ dy y A (4 2 y)w

M dy y P P+ dy y

y, v

FIGURE 16.7 Bending element.

v(y, t) = Y(y)f(t) then substituting them in Equation (16.23), one has the temporal part, d 2f dt + 2nw + w 2f = 0 2 dt dt d2 Y + (a 2 - gy )Y = 0 dy 2 (16.24)

(16.25)

where n = C/2wm, a2 = (mw2 pCfb)/AE, g = p(Cft Cfb)/AEl, and w is the natural frequency of the vibration. The integration of the differential equation [Equation (16.25)], for the spatial part of can be written as Y(y) = {(t)}T{C} (16.26)

The vector of the integration constants {C} is determined from the boundary conditions. The dynamic member stiffness matrix can be determined either by writing the relations among the member end forces {f} and the member end displacements {d}, or by using the Lagrangian Dynamic Equation. Thus one has, {f} = [kA]{d} (16.27)

where [kA], is the dynamic, axial displacement member stiffness matrix. Noticing the similarity between the twist rotation and axial displacement functions, and the similarity between the stressstrain relationships in both cases, the dynamic member stiffness matrix for torsion can be written by analogy from that of the axial force member by simply substituting the axial force rigidity EA by the torsional rigidity GJ and the parameters a, by their expression for torsion a2 = (mw2/GA pR2Cfb/GJ) g = p R2 (Cft Cfb)/GJl where R is the mean radius of the cross-section. The vibration of a member under the bending in the Oyz plane is obtained from the differential of motion as follows (Figure 16.7) EI x C - C st dw d2 w d2 w 4w C P m y = 0 + + + + p C sb - sb 4 2 2 dt 1 dy dy dt (16.28)

2005 by CRC Press LLC

where P is the member axial force, positive when it is in compression; p is the projection of the crosssection per unit length of the member on the plane subjected to soil pressure; and since the variation of the soil modula is assumed to be linear, Cst and Csb are the soil subgrade modulus at the top and at the bottom of the pile. Assuming that the deection function w(y, t) can be written in separable variable form: w(y, t) = Z(y) f(t) and substituting them into the differential equation of motion (Equation 16.28), for the spatial part one has d4 Z d2 Z + 2k 2 2 - (b 4 + g 2 )Z = 0 4 dy dy (16.29)

where k2 = P/2EIx, 4 = (mw2 p Csb)/EIx, and g2 = p (Csb Cst)/EIxl. The time-dependent part f(t) is the same for all four independent vibrations [Equation (16.24)]. By using the Lagrangian Dynamic Equation, the member stiffness matrix [kB, Oyz ] subjected to bending in the Oyz plane can be obtained as {f} = [kB,Oyz]{d} (16.30)

The nature of the shape function and the stiffness matrix depend on the values of the parameters and g and the member axial force P, whether it is in tension or compression [Ovunc, 1985]. The dynamic member stiffness matrix for the bending in the Oxy plane is obtained following the same steps mentioned for bending in the Oyz plane. The dynamic stiffness matrices for all four independent cases are combined, according to the type of structure: either truss or frames in planes or in space. The dynamic stiffness matrix [K] for the structure is obtained from the dynamic stiffness matrices of its constituent members as {F} = [K] {D} (16.31)

The dynamic stiffness matrix [K] is a transcendental function of the natural circular frequency w of the structure.

Single Beams
The cantilever beam subdivided into three subelements is subjected to a static axial force fax at its free end 1 (Figure 16.6). The effect of member axial force fax appears in the equations of free vibration as an additional matrix [NLM] and [NCS] for the lumped and consistent mass methods [Equation (16.18)]: ([K] + [NLM] - w2[MLM]){D} = {0} ([K] + [NCS] - w2[MCS]){D} = {0} (16.32) (16.33)

In the continuous mass method, the member axial force fax appears in the argument of the trigonometric or hyperbolic functions [Equation (16.14)]. The dynamic responses of the cantilever beam (Figure 16.6) have been computed by the lumped and consistent mass method by only changing the magnitude of the member axial force fax from zero to its critical value ( fax)crit [Equation (16.23) and Equation (16.24)]. The same computations have been performed using the continuous mass method [Equation (16.14)]. The ratios for f,

2005 by CRC Press LLC

fmi,mj = w1,mi/w01,mj

(16.34)

of the rst natural frequency by method mi (with the effect of member axial force) versus that of method mj (without the effect of member axial force) are plotted in Figure 16.7. The index mi or mj designates LM, CS, and CT, the lumped mass, consistent mass, and continuous mass methods. The comparison of the variations of the ratios fLM,LM with fLM,CT [Equation (16.24)] exhibits rough approximation. The approximation involved in the variations of ratios fCS,CS and fCS,CT is very close to the actual one. The comparison of the variations of the ratios fLM,CT and fCS,CT with fCT,CT exhibits some degree of approximation.

Frames
All the columns of the three-story steel frame are assumed to be subjected to a static axial force, fax, of the same magnitude. The dynamic responses of the three-story frame were evaluated by lumped, consistent, and continuous mass methods. Two different cases were considered. In the rst case, only the magnitudes of the weights acting on the beams have been increased by a factor m. In the second case, the magnitudes of the member axial force, fax, and the weights acting on each beam have been increased by px and m, in such a way that both factors have the same magnitude, px = m. For a same-method mj, the ratio ai,mj = (wI/woi)mj (16.35)

of the ith natural frequency wi (including the effect of member axial force and/or additional mass, only on the beams) versus the ith natural frequency woi (excluding all the additional effect) is plotted in Figure 16.8. The ratio ai,CT for rst, second, and third natural frequencies computed by the continuous mass method is also shown in Figure 16.8. The sways at the oor level Dmj, including and excluding the effect of the axial force fax, are computed by the lumped, consistent, and continuous mass matrix methods. The variations of the sway at the oor levels Dmj are plotted in Figure 16.9. When the effects of member axial forces are excluded, the sways at the oor levels obtained by the lumped, consistent, and continuous mass methods are almost the same. The effect of member axial force has shown small variations in the oor sways evaluated by the lumped and consistent mass methods. But the variation in the sways at the oor levels computed by the continuous mass method is large. Although the rst buckling mode for lumped and consistent mass methods occurs by increasing sways from lower to upper oors, for the continuous mass method the rst buckling mode occurs between the base and the rst oor. The relative displacement of second and third oors with respect to the displacement of the rst oor tends to zero.

Offshore Structures
The dynamic behavior of the plane frame of the offshore platform similar to Conocos Main Pass 296A is selected as a structural system that includes the effects of member axial force and soilstructure interaction. Thus, the results obtained through the computations can be easily compared with those obtained previously [Ovunc, 1984 and 1985]. The static loads on the members vibrating within the plane
fax Pofe(t) 1 /3 3 /3 4 /3 2

FIGURE 16.8 Beam subjected to axial force.

2005 by CRC Press LLC

1.00

mi, mj

0.75

0.50

0.25

+ P Pcrit

0.

0.25 CT,CT

0.50

0.75 LM,CT

1.00

LM,LM CS,CS CS,CT

FIGURE 16.9 Effect of member axial force.

of the frame are considered as distributed on these members, whereas the static loads acting on the transversal members are lumped at the joints of the plane frame. Load factors m and l are introduced on the distributed and lumped loads. When the distributed load factor m tends to zero, the continuous mass matrix approach tends to the lumped mass matrix approach. Two different boundary conditions are taken into account at the bottom tips of the piles: xed and free tips. The ratio of the rigidity of the beams to columns and the embedded length of the piles can be selected as additional parameters. A dimensionless parameter ai is dened as ai = wpi/woi the ratio of the natural circular frequencies of the ith mode wpi and woi, corresponding to the structures with and without soilstructure interaction and/or with or without the member axial force, respectively. The data related to the structural system are assumed to be similar to those of Conocos offshore platform. The data for the wave forces and soil characteristics are collected from information related to the Gulf of Mexico.

Dening Terms
Damping Results from the internal friction within the material or from system vibration within another material. Damping energy Work done by the internal friction within the material as a result of the motion. Kinetic energy Work done by a mass particle as a result of its motion. Stiffness coefcient Ki,j Force or moment in the direction of the rst index (i) required to maintain the equilibrium of the body due to a unit displacement or rotation in the direction of the second index (j), while all the other specied displacements and rotations are equal to zero. Strain energy Work done by a particle due to its stress and strain. External energy Work done by an external force due to a displacement in its direction.

References
Clough, R. W. and Penzien, J. 1993. Dynamic of Structures. McGraw-Hill, New York. Ovunc, B. A. 1974. Dynamics of frameworks by continuous mass method. Compt. Struct. 4:10611089.
2005 by CRC Press LLC

Ovunc, B. A. 1980. Effect of axial force on framework dynamics. Compt. Struct. 11:389395. Ovunc, B. A. 1985. Soil-structure interaction and effect of axial force on the dynamics of offshore structures. Compt. Struct. 21:629637. Ovunc, B. A. 1990. Free and Forced Vibration of Frameworks with Tapered Members, Struceng and Femcad Conference, Grenoble, France, Oct. 1718, pp. 341346. Paz, M. 1993. Structural Dynamics, Theory and Computations. Van Nostrand Reinhold, New York. Przemieniecki, J. S. 1968. Theory of Matrix Structural Analysis. McGraw-Hill, New York.

Further Information
Paz, M. 1986. Microcomputer Aided Engineering: Structural Dynamics, Van Nostrand Reinhold, New York. Ovunc, B. A. 1972. The dynamic analysis of space frameworks by frequency dependent stiffness matrix approach. In International Association for Bridges and Structural Engineering, vol. 32/2, Zurich, Switzerland, pp. 137154. Ovunc, B. A. 1986. Offshore platforms subjected to wave forced. In Recent Applications in Computational Mechanics, (ed. D. L. Karabalis), ASCE, September, pp. 154169. Ovunc, B. A. 1985. STDYNL, a code for structural systems. In Structural Analysis Systems (ed. Niku Lari), Pergamon Press, Oxford, vol. 3, pp. 225238. Ovunc, B. A. 1992. Dynamics of Offshore Structures Supported on Piles in Cohesionless Soil. ASME, European Joint Conference on Engineering Systems Design and Analysis, Istanbul, Turkey, ESDA 1992, June 29July 4, ASME PD, vol. 475, pp. 1118. Ovunc, B. A. 1990. Vibration of Timoshenko Frames Including Member Axial Force and Soil-Structure Interaction. FEMCAD & OPTIMIZATION Conference, Los Angeles, Nov. 56, pp. 359364.

1.00

i, mj

0.80

p = 0, m 0 1,CT: 2, CT: 3,CT: i,LM i, CS p= m 0 1,CT:

0.60

i,LM i, CS (i = 1, 2, 3, ...)

0.40

0.20

0.0

px, m 10. 20. 30. 40.

FIGURE 16.10 Additional effects of member axial force.

2005 by CRC Press LLC

10'

10'

Axial force excluded DLM Dcs DCT: 15' Axial force included DCT: DLM: DCS:

FIGURE 16.11 Sways at oor levels.


35' 35' 35'

16.00' 36.50' 60.00'

80.00' 212.00' 82.00'

230.00'

73.69' 73.69' 73.69' Soil Modes Cohesionless: First : Cohesive : Second :

Hard soil

Soft soil

FIGURE 16.12 Offshore platform and its modal shapes.

2005 by CRC Press LLC

i 1.00 i i = 1, i = 2, i = 3, 1.00 i = 1, i = 2, i = 3,

.80

.50 .60 .60 .40 .40


Cohesionless: Cohesive: Cohesionless: Cohesive:

.20

LogC3 .0 .01 .05 .10 Free Tip Piles .50 1.0 1.2 .005 .01 .05 .10 Fixed Tip Piles

LogC3 .50 1.0

FIGURE 16.13 The variation of natural circular frequencies of the platform, supported by free and xed tip piles vs. the soil characteristics.

2005 by CRC Press LLC

You might also like