Introduction to Structural Dynamics / Edition 1

Introduction to Structural Dynamics / Edition 1

Pub. Date:
Cambridge University Press


Current price is , Original price is $180.0. You
Select a Purchase Option (New Edition)
  • purchase options
    $146.70 $180.00 Save 19% Current price is $146.7, Original price is $180. You Save 19%.
  • purchase options


Introduction to Structural Dynamics / Edition 1

This textbook provides the student of aerospace, civil, or mechanical engineering with all the fundamentals of linear structural dynamics analysis and scattered discussions of non-linear structural dynamics, it is designed to be used primarily for a first-year graduate course. This textbook is a departure from the usual presentation of this material in two important respects. First, descriptions of system dynamics throughout are based on the simpler-to-use Lagrange equations of motion. Second, no organizational distinction is made between single- and multiple-degree-of-freedom systems. In support of these two choices, the first three chapters review the needed skills in dynamics and finite element structural analysis. The remainder of the textbook is organized mostly on the basis of first writing structural system equations of motion, and then solving those equations. The modal method of solution is emphasized, but other approaches are also considered. This textbook covers more material than can reasonably be taught in one semester. Topics that can be put off for later study are generally placed in sections designated by double asterisks or in endnotes. The final two chapters can also be deferred for later study. The textbook contains numerous example problems and end-of-chapter exercises.

About the Author:
Bruce K. Donaldson, At the University of Maryland he became a professor of aerospace engineering and then a professor of civil engineering

Product Details

ISBN-13: 9780521865746
Publisher: Cambridge University Press
Publication date: 07/31/2006
Series: Cambridge Aerospace Series
Edition description: New Edition
Pages: 539
Product dimensions: 6.97(w) x 9.96(h) x 1.38(d)

About the Author

Bruce K. Donaldson was first exposed to aircraft inertia loads when he was a carrier-based U.S. Navy anti-submarine pilot. He subsequently worked in the structural dynamics area at the Boeing Co. and at the Beech Aircraft Co. in Wichita, KS before returning to school and then embarking on an academic career in the area of structural analysis. He became a professor of Aerospace Engineering, and then a professor of Civil and Environmental Engineering at the University of Maryland. Professor Donaldson is the recipient of numerous teaching awards, and has maintained industrial contacts, working various summers at government agencies and for commercial enterprises, the last being Lockheed Martin at Fort Worth, Texas.

Read an Excerpt

Introduction to structural dynamics
Cambridge University Press
978-0-521-86574-6 - Introduction to structural dynamics - by Bruce K. Donaldson

1 The Lagrange Equations of Motion

1.1 Introduction

A knowledge of the rudiments of dynamics is essential to understanding structural dynamics. Thus this chapter reviews the basic theorems of dynamics without any consideration of structural behavior. This chapter is preliminary to the study of structural dynamics because these basic theorems cover the dynamics of both rigid bodies and deformable bodies. The scope of this chapter is quite limited in that it develops only those equations of dynamics, summarized in Section 1.10, that are needed in subsequent chapters for the study of the dynamic behavior of (mostly) elastic structures. Therefore it is suggested that this chapter need only be read, skimmed, or consulted as is necessary for the reader to learn, review, or check on (i) the fundamental equations of rigid/flexible body dynamics and, more importantly, (ii) to obtain a familiarity with the Lagrange equations of motion.

   The first part of this chapter uses a vector approach to describe the motions of masses. The vector approach arises from the statement of Newton’s second and third laws of motion, which are the starting point for all the material in this textbook. These vector equations of motion are used only to prepare the way for the development of the scalar Lagrange equations of motion in the second part of thischapter. The Lagrange equations of motion are essentially a reformulation of Newton’s second law in terms of work and energy (stored work). As such, the Lagrange equations have the following three important advantages relative to the vector statement of Newton’s second law: (i) the Lagrange equations are written mostly in terms of point functions that sometimes allow significant simplification of the geometry of the system motion, (ii) the Lagrange equations do not normally involve either external or internal reaction forces and moments, and (iii) the Lagrange equations have the same mathematical form regardless of the choice of the coordinates used to describe the motion. These three advantages alone are sufficient reasons to use the Lagrange equations throughout the remaining chapters of this textbook.

1.2 Newton’s Laws of Motion

Newton’s three laws of motion can be paraphrased as (Ref. [1.1]):

  1. Every particle continues in its state of rest or in its state of uniform motion in a straight line unless it is compelled to change that state by forces impressed upon it.
  2. The time rate of change of momentum is proportional to the impressed force, and it is in the direction in which the force acts.
  3. Every action is always opposed by an equal reaction.

These three laws are not the only possible logical starting point for the study of the dynamics of masses. However, (i) these three laws are at least as logically convenient as any other complete basis for the motion of masses, (ii) historically, they were the starting point for the development of the topic of the dynamics, and (iii) they are the one basis that almost all readers will have in common. Therefore they are the starting point for the study of dynamics in this textbook.

   There are features of this statement of Newton’s laws that are not immediately evident. The first of these is that these laws of motion are stated for a single particle, which is a body of very, very small spatial dimensions, but with a fixed, finite mass. The mass of the jth particle is symbolized as mj. The second thing to note is that momentum, which means rectilinear momentum, is the product of the mass of the particle and its instantaneous velocity. Of course, mass is a scalar quantity, whereas velocity and force are vector quantities. Hence the second law is a vector equation. The third thing to note is that the second law, which includes the first law, is not true for all coordinate systems. The best that can be said is that there is a Cartesian coordinate system “in space” for which the second law is valid. Then it is easy to prove (see the first exercise) that the second law is also true for any other Cartesian coordinate system that translates at a constant velocity relative to the valid coordinate system. The second law is generally not true for a Cartesian coordinate system that rotates relative to the valid coordinate system. However, as a practical matter, it is satisfactory to use a Cartesian coordinate system fixed to the Earth’s surface if the duration of the motion being studied is only a matter of a few minutes. The explanation for this exception is that the rotation of the Cartesian coordinate system fixed at a point on the Earth’s surface at the constant rate of one-quarter of a degree per minute, or 0.0007 rpm, mostly just translates that coordinate system at the earth’s surface in that short period of time. See Figure 1.1(a).

Figure 1.1. (a) Valid and invalid coordinate systems for Newton’s second law, both moving at constant speed. (b) Illustration of the right-hand rule for r x F = M = r F sin αn.

Image not available in HTML version

   As is derived below, when Newton’s second law is extended to a mass m of finite spatial dimensions, which is subjected to a net external force of magnitude1 F, then Newton’s second law can be written in vector form as follows:

Display matter not available in HTML version

where P = mv is the momentum vector, v is the velocity vector of the total mass m relative to the valid coordinate origin, t is time, and a is the acceleration vector, which of course is the time derivative of the velocity vector. The velocity vector is not the velocity of all points within the mass m relative to the valid coordinate system. Rather, it is the velocity of the one point called the center of mass, which is defined below. Further, note that the mass of the system of particles whose motion is described by this equation is the mass of a fixed collection of specific mass particles. Hence, even though the boundary surface that encloses these specified mass particles may change considerably over time, the mathematical magnitude of the mass term is a constant. Those mass particles that are included within the mass, or alternately, enclosed by the boundary surface of the mass system, are defined by the analyst as the “mass system under study.”

   The above basic result, Eq. (1.1), can be derived as follows. Consider a collection of, that is, a specific grouping of, N particles of total mass m = Σmj, where all such sums run from j = 1 to j = N, where N can be a very large number. Again, it is not essential that there be any particular geometric relationship between the N particles. Newton’s second law applies to each of these N particles. To write Newton’s second law in a useful way, let each of these N particles be located by means of its own position vector rj(t) originating at the origin of a valid coordinate system. Note that if the time-varying spatial position of the ith particle in terms of the valid Cartesian coordinates is [xi(t), yi(t), zi(t)], then the position vector can be written as ri(t) = xi(t)i + yi(t)j + zi(t)k. Since the differential quantity dri is tangent to the path of the ith particle, the velocity vector is always tangent to the particle path. However, because the forces applied to the particle are not necessarily tangent to the particle path, neither is the acceleration vector, d2r/dt2. Thus the path of the particle need not be straight.

   The statement of the second law for the individual ith particle now can be written as

Display matter not available in HTML version

where Fiex is the vector sum of all the forces acting on the ith particle that originate from sources outside of this collection of N particles (to be called the net external force acting on the ith particle), and Fiin is the vector sum of all the forces acting on ith particle that originate from interactions with the other N-1 particles (i.e., the net internal force acting on the ith particle). From Newton’s third law, each of the N-1 components of the net internal force acting on the ith particle can be associated with an equal and opposite force acting on one of the other particles in the collection of N particles. Hence, summing all such Eqs. (1.2) for the N particles leads to the cancellation of all the internal forces between the N particles, with the result

Display matter not available in HTML version

   Again, the total mass m is defined simply as the scalar sum of all the mi. That is m = Σmj. The location of the center of mass of the total mass m is identified by introducing the center of mass position vector, r(t) (without a subscript). Since this position vector goes from the coordinate origin to the center of mass, this vector alone fully describes the path traveled by the center of mass as a function of time. The center of mass position vector r at any time t is defined so that

Display matter not available in HTML version

This definition means that the center of mass position vector is a mass-weighted average of all the mass particle position vectors. This definition can also be viewed as an application of the mean value theorem. Differentiating both sides of the definition of the center of mass position vector with respect to time twice and then substituting into the previous equation immediately yields Eq. (1.1): F = mrma. Again, the force vector F, without superscripts and subscripts, is the sum of all the external forces. Note that external forces can arise from only one of two sources: (i) the direct contact of the boundary surface of the N particles under study with the boundary of other masses or (ii) the distant action of other masses, in which case they are called field forces. Gravitational forces are an example of the latter type of action.

1.3 Newton’s Equations for Rotations

A knowledge of the motion of the center of mass can tell the analyst a lot about the overall motion of the mass system under study. However, that information is incomplete because it tells the analyst nothing at all about the rotations of the mass particles about the center of mass. Since rotational motions can be quite important, this aspect of the overall motion needs investigation.

   Just as the translational motion of the center of mass can be viewed as determined by forces, rotational motions are determined by moments of forces. Recall that the mathematical definition of a moment about a point, when the moment center is the origin of the valid coordinate system, is

Display matter not available in HTML version

Recall that reversing the order of a vector cross product requires a change in sign to maintain an equality. Further note that it is immaterial where this position vector intercepts the line of action of the above force vector because the product of the magnitude of the r vector and the sine of the angle between the r and F vectors is always equal to the perpendicular distance between (i) the line of action of the force and (ii) the moment center.

   Structural engineers are more familiar with moments about Cartesian coordinate axes than moments about points. The relation between a moment about a point and a moment about such an axis can be understood by reference to Figure 1.1. (b). This figure illustrates that the moment resulting from the cross product of the r vector and the F vector, by the rules of vector algebra, is in the direction of the unit vector n, which is perpendicular to the plane formed by the r and F vectors. The positive direction of n is determined by the thumb of the right hand after sweeping the other four fingers of the right hand from the direction of r, the first vector of the cross product, through to the direction of F. In terms of α, the angle between these two vectors in the plane formed by the two vectors

Display matter not available in HTML version

Like any other vector, the vector M has components along the Cartesian coordinate axes. In terms of the components of the force F and the position vector r, the moment about a point can be written, using vector algebra, as follows:

Display matter not available in HTML version

Considering the last equation, it is clear that moments about axes are simply components of moments about points.

   When describing the rotation of the mass m, it is often convenient to consider a reference point P that is other than the valid coordinate origin, which is here called the point O. See Figure 1.2. Let the this new reference point P move in an arbitrary fashion relative to the coordinate origin, point O, in a fashion defined by the position vector rP(t). Introduce the vector quantity LPj(t) which is to be called the angular momentum about point P, or, more descriptively, the moment of momentum of the mass particle mj about the arbitrary point P. That is, the angular momentum about point P of the jth mass particle is defined as the vector cross product of (i) the position vector from point P to the particle mj and (ii) the momentum vector of mj where the associated velocity vector is that relative to point P rather than the origin of the coordinate system, point O. Thus, in mathematical symbols, relative to point P, the angular momentum of the jth particle, and the angular momentum of the total mass m are, respectively,

Figure 1.2. Vectors relevant to the rotational motion of a mass. Point P has an arbitrary motion relative to point O.

Image not available in HTML version
Display matter not available in HTML version

Differentiating both sides of the total angular momentum with respect to time, and noting that the cross product of the relative velocity vector (rj - rP) with itself is zero, yields the following result:

Display matter not available in HTML version

From the original statement of Newton’s second law, it is possible to substitute in the above equation the net external and internal forces on the jth particle for mj(d2/dt2)rj. The result is

Display matter not available in HTML version

The term involving the net internal forces sums to zero because all the component internal forces are not only equal and oppositely directed, but, by the strong form of Newton’s third law, they are also collinear. See Exercise . The remaining portion of the first term, that involving the net external forces on the N particles, sums to MP, called the moment about point P of all the external forces acting on the mass m. The last term in the above sum can be simplified by noting that

Display matter not available in HTML version

Thus the final result for the time derivative of the angular momentum of the mass m is

Display matter not available in HTML version

In other words, with reference to Figure 1.2,

Display matter not available in HTML version

Clearly, if point P is coincident with the center of mass (called the center of mass or CG case, where rP = r), or if the relative position vector rP-r and the acceleration vector (d2/dt2)rP are collinear (unimportant because it is unusual), or if point P is moving at a constant or zero velocity with respect to point O (called, for simplicity, the fixed point or FP case), then the rotation equation reduces to simply

Display matter not available in HTML version

   Note that the above vector equation is the origin of the static equilibrium equations, which state that “the sum of the moments about any axis is zero.” That is, when the angular momentum relative to the selected point P is zero or a constant, then the three orthogonal components of the total moment vector of the external forces acting on the system about point P are zero. These three orthogonal components are the moments about any three orthogonal axes.

   The above rotational motion equation, Eq. (1.3b) is not as useful as Eq. (1.1), the corresponding translational motion equation. In Eq. (1.1), the three quantities force, mass, and acceleration are individually quantifiable. In Eq. (1.3b), while the moment term is easily understood, the time rate of change of the angular momentum needs further refinement so that perhaps it too can be written as some sort of fixed mass type of quantity multiplied by some sort of acceleration. Recall that for the mass system m, the total angular momentum relative to point P, is defined as the sum of the moments of the momentum of all the particles that comprise the mass m. That is, again

Display matter not available in HTML version

   From the previous development, that is, Eqs. (a,b), there are two simplifying choices for the reference point P: the FP (so-called fixed point) case and the CG (center of mass) case, where the time derivative of the angular momentum is equal to just the moment about point P of all the external forces. First consider the FP case, where point P has only a constant velocity relative to the coordinate origin, point O. Then, from Exercise , either point P or point O is the origin of a valid Cartesian coordinate system. Since these two points are alike, for the sake of simplicity, let the reference point P coincide with the origin of the coordinate system, point O. Again, this placement of point P at point O does not compromise generality within the FP case because when point P is only moving at a constant velocity relative to point O, point P can also be an origin for a valid coordinate system. Then with rP = 0, and because the ei vectors of Figure 1.2. originate at the center of mass, the total angular momentum becomes

Display matter not available in HTML version

To explain why the second and third terms of the above second line are zero, recall the definition of the center of mass position vector, r. That mean value definition is mr ≡ Σmi ri. Since ri = r+ei, mr ≡ Σ mirmiei. Since r is not affected by the summation over the N particles, it can be factored out of the first sum on the above right-hand side. The result is mrmr + Σmiei or 0 = Σ miei. Furthermore, because the mass value of each particle is a constant, the time derivative of this last equation shows that 0 = Σ miei. This is just an illustration of the general fact that first moments, that is, multiplications by distances raised to the first power, of mass or area, or whatever, about the respective mean point are always zero. Multiplications of mass by distances with exponents other than one lead to terms which are generally not zero.

   In the above FP equation, Eq. (1.4a), for the angular momentum, the first term depends only on the motion of the center of mass relative to the Cartesian coordinate origin. Even if the mass is not rotating relative to the Cartesian coordinate origin, this term is generally not zero. The second part of the angular momentum exists even if the center of mass is not moving. This second part accounts for the spin of the mass about its own center of mass.

   The CG case is where the reference point P is located at the center of mass, point C, rather than at the coordinate origin, point O, as in the FP case. In this CG case, r=rP and ri-rP=ei. Substituting these vector relationships into the expression for LP immediately leads to the same result for the angular momentum, as was obtained for the FP case, except that the first of those two terms is absent. Hence the mathematics of the CG case are included within that of the FP case, and therefore the CG case does not need a parallel development.

1.4 Simplifications for Rotations

Since Newton’s second law is a vector equation, it has been convenient to derive its rotational corollaries by use of vector algebra in three-dimensional space. However, it is no longer convenient to pursue the subject of rotations using three-space vector forms because, in general, the rotations themselves about axes in three dimensions (as opposed to moments about axes in three dimensions) are not vector quantities. For a quantity be classified as a vector, the order of an addition has to be immaterial; that is, it is necessary that A + B = B + A, which is called the commutative law for vector addition. In contrast, as Figure 1.3. illustrates, the order of addition of rotations in three-space can greatly change the final orientation of the mass whenever the rotational angles involved are large, like the 90° angles selected for Figure 1.3. There are two simple ways of circumventing this difficulty. The first simplifying approach is to restrict the rotational motion equations to a single plane. In a single plane, all rotations simply add or subtract as scalar quantities. This is a wholly satisfactory approach for most of the illustrative pendulum problems considered in the next chapter. The second option for simplification is to retain rotations about more than one orthogonal axis but limit all those rotations to being small. Here “small” means that the tangent of the angle is closely approximated by the angle itself.2 As is explained in Ref. [1.2], p. 271, in contrast to larger angles, angles about orthogonal axes of these small magnitudes can be added to each other as vector quantities. This approach of restricting the rotations to either being small or lying in a single plane would not be adequate for formulating a general analysis of the motion of bodies of finite size, which is not a present concern. However, this is a satisfactory approach for almost all structural dynamics problems because structural rotations due solely to the vibrations of a flexible structure are almost always less than 10° or 12°. Therefore, to repeat and thus underline this important point, for the present purposes of structural dynamics, it is often satisfactory only to look at rotations in a single plane or restrict the analysis to small rotations, which can be added vectorially.

Figure 1.3. Proof that, generally, rotations are not vectors because the order of the rotations is not irrelevant.

Image not available in HTML version

   To further the discussion, consider all rotations confined to a single plane that, for the sake of explicitness, is identified as the z plane. To reflect the change from three to two dimensions, the notation FP for a fixed point in three-dimensional space, transitions to FA for a fixed axis perpendicular to the z plane. This simplification from a general state of rotations to those only about an axis paralleling the z axis allows the introduction of a pair of convenient unit vectors in the z plane called p1 and q1 such that p1 is directed from the origin toward the center of mass and q1 is rotated 90° counterclockwise from p1. These two unit vectors rotate in the z plane as the center of mass moves in that plane. In terms of the fixed-in-space Cartesian coordinate unit vectors, i, j, as shown in Figure 1.4. (a),

Figure 1.4. (a) The relationship between the rotating unit vectors and the fixed unit vectors, i and j. (b) Use of unit vectors to locate the ith mass particle.

Image not available in HTML version
Display matter not available in HTML version

Again, even though p1 and q1 have a fixed unit length, they have time derivatives because their orientation in the z plane varies with time as the angle ϕ changes with time. The above equations show that the time derivatives of these rotating unit vectors are

Display matter not available in HTML version

This unit vector pair p,q can be used with both the position vector for the center of mass and the vector from the center of mass to the ith mass particle. That is, as illustrated in Figure 1.4. (b),

Display matter not available in HTML version

As the final limitation on the dynamics equations to be developed, let the geometry of the total mass be restricted to small changes in overall shape so that the rotation angle for the jth mass about the center of mass differs so little from that average rotation that the average rotation ϕ2 can be used as the rotation angle about the center of mass for all the mass particles that are included within the boundary of the total mass. This is a rather minor limitation, if any at all, for almost all structures.

© Cambridge University Press

Table of Contents

Preface for the Student     xi
Preface for the Instructor     xv
Acknowledgments     xvii
List of Symbols     xix
The Lagrange Equations of Motion     1
Introduction     1
Newton's Laws of Motion     2
Newton's Equations for Rotations     5
Simplifications for Rotations     8
Conservation Laws     12
Generalized Coordinates     12
Virtual Quantities and the Variational Operator     15
The Lagrange Equations     19
Kinetic Energy     25
Summary     29
Exercises     33
Further Explanation of the Variational Operator     37
Kinetic Energy and Energy Dissipation     41
A Rigid Body Dynamics Example Problem     42
Mechanical Vibrations: Practice Using the Lagrange Equations     46
Introduction     46
Techniques of Analysis for Pendulum Systems     47
Example Problems     53
Interpreting Solutions to Pendulum Equations     66
Linearizing Differential Equations for Small Deflections     71
Summary     72
**Conservation of Energy versus the Lagrange Equations**     73
**Nasty Equations of Motion**     80
**Stability of Vibratory Systems**     82
Exercises     85
The Large-Deflection, Simple Pendulum Solution     93
Divergence and Flutter in Multidegree of Freedom, Force Free Systems     94
Review of the Basics of the Finite Element Method for Simple Elements     99
Introduction     99
Generalized Coordinates for Deformable Bodies     100
Element and Global Stiffness Matrices     103
More Beam Element Stiffness Matrices     112
Summary     123
Exercises     133
A Simple Two-Dimensional Finite Element     138
The Curved Beam Finite Element     146
FEM Equations of Motion for Elastic Systems     157
Introduction     157
Structural Dynamic Modeling     158
Isolating Dynamic from Static Loads     163
Finite Element Equations of Motion for Structures     165
Finite Element Example Problems     172
Summary     186
**Offset Elastic Elements**     193
Exercises     195
Mass Refinement Natural Frequency Results     205
The Rayleigh Quotient     206
The Matrix Form of the Lagrange Equations     210
The Consistent Mass Matrix     210
A Beam Cross Section with Equal Bending and Twisting Stiffness Coefficients     211
Damped Structural Systems     213
Introduction     213
Descriptions of Damping Forces     213
The Response of a Viscously Damped Oscillator to a Harmonic Loading     230
Equivalent Viscous Damping     239
Measuring Damping     242
Example Problems     243
Harmonic Excitation of Multidegree of Freedom Systems     247
Summary     248
Exercises     253
A Real Function Solution to a Harmonic Input     260
Natural Frequencies and Mode Shapes     263
Introduction     263
Natural Frequencies by the Determinant Method     265
Mode Shapes by Use of the Determinant Method     273
**Repeated Natural Frequencies**     279
Orthogonality and the Expansion Theorem     289
The Matrix Iteration Method     293
**Higher Modes by Matrix Iteration**     300
Other Eigenvalue Problem Procedures     307
Summary     311
**Modal Tuning**     315
Exercises      320
Linearly Independent Quantities     323
The Cholesky Decomposition     324
Constant Momentum Transformations     326
Illustration of Jacobi's Method     329
The Gram-Schmidt Process for Creating Orthogonal Vectors     332
The Modal Transformation     334
Introduction     334
Initial Conditions     334
The Modal Transformation     337
Harmonic Loading Revisited     340
Impulsive and Sudden Loadings     342
The Modal Solution for a General Type of Loading     351
Example Problems     353
Random Vibration Analyses     363
Selecting Mode Shapes and Solution Convergence     366
Summary     371
**Aeroelasticity**     373
**Response Spectrums**     388
Exercises     391
Verification of the Duhamel Integral Solution     396
A Rayleigh Analysis Example     398
An Example of the Accuracy of Basic Strip Theory     399
Nonlinear Vibrations     400
Continuous Dynamic Models     402
Introduction     402
Derivation of the Beam Bending Equation     402
Modal Frequencies and Mode Shapes for Continuous Models     406
Conclusion     431
Exercises     438
The Long Beam and Thin Plate Differential Equations     439
Derivation of the Beam Equation of Motion Using Hamilton's Principle     442
Sturm-Liouvilie Problems     445
The Bessel Equation and Its Solutions     445
Nonhomogeneous Boundary Conditions     449
Numerical Integration of the Equations of Motion     451
Introduction     451
The Finite Difference Method     452
Assumed Acceleration Techniques     460
Predictor-Corrector Methods     463
The Runge-Kutta Method     468
Summary     474
**Matrix Function Solutions**     475
Exercises     480
Answers to Exercises     483
Solutions     483
Solutions     486
Solutions     494
Solutions     498
Solutions     509
Solutions     516
Solutions     519
Solutions     525
Solutions     529
Fourier Transform Pairs     531
Introduction to Fourier Transforms     531
Index      537

Customer Reviews

Most Helpful Customer Reviews

See All Customer Reviews