Formulation of Harmonic Analysis

Introduction, Complex Variables and Notation,

Displacement-Velocity-Acceleration, Formulation
and Derivation, Discussion

Solid Mechanics III – Methods of Solving Problems

Forced Frequency Response-Harmonic Analysis
Introduction

• Harmonic analyses are used to determine the steady-

state response of a structure to loads that vary
sinusoidally (harmonically) with time.
• This enables the verification of a design, providing
insight such as;
‐ Will resonance occur, which could result in excessive motion,
stress, noise and vibration.
‐ Will the cyclically repeating load result in repetitive stress,
resulting in fatigue damage and potential failure.

2
 Introduction

• Dynamic problems can be solved in the time domain or the frequency domain;
1 Time Domain 2 Frequency Domain
• The input and responses are arbitrary • Inputs and Reponses are single frequency
• We typically are investigating the transient response sinusoidal so we solve for the response for a
• To reach a steady state response is computationally particular excitation frequency using 1 solution
expensive as we have to solve for the transient iteration
response (multiple iterations) to get to the steady • We sweep over a frequency range of interest
state • Solving for the steady state response

Input Response

3
Introduction

• An important property of a harmonic analysis is when the system

has a steady state sinusoidal input or excitation, the output is also
sinusoidal, but with varying amplitude and phase angle.
• There can be multiple loadings, but all are sinusoidal, and the
excitation and response is steady state oscillatory behavior.
‐ A harmonic excitation is typically of this form below , but sometimes in
listed in complex notation (which we will discuss next).
‐ The harmonic function is comprised of 3 main variables;
1 Frequency
2 Amplitude 𝐹𝑖 = 𝐹𝑖 sin Ω𝑖 𝑡 + 𝜃𝑖
3 Phase angle
‐ For a given time (t), the corresponding force can be computed.

4
 Introduction

1 Frequency of Excitation 𝐹𝑖 = 𝐹𝑖 sin Ω𝑖 𝑡 + 𝜃𝑖

• The imposed circular frequency, Ω𝑖 is a simple expression and typically
has units of radians/sec. Beware… It is common mistake to mix up Ω where
and 𝑓. Ω𝑖 = Imposed Circular Frequency
1 𝑓 = Imposed frequency
Ω = 2𝜋𝑓 𝑓 = 𝑇𝑓 = Period
𝑇𝑓
𝐹𝑖 = Amplitude
2 Amplitude 𝜃𝑖 = Phase Angle
𝑡 = Time
• The peak magnitude of the forcing excitation. Note that harmonic
motion excites in both positive and negative directions. 1.50
𝑃ℎ𝑎𝑠𝑒 𝑆ℎ𝑖𝑓𝑡
1.00
3 Phase Angle
0.50
Phase Angle (0 deg)
𝐹𝑖
• Typically necessary when multiple loads are present to control their Phase Angle (90 deg)

Load
0.00
relative phase or timing -0.50

Substitute any value for time, and the equation computes the resulting -1.00
𝑇𝑓
Force at that time as can see in the graph -1.50
0 0.0005 0.001 0.0015 0.002 0.0025
Time

5
Complex Variables and Notation
• Before we dive further into the formulation of harmonic analysis, let’s cover a bit about complex
variables which will be helpful for the derivation.
• Why do we concern ourselves with complex variables to begin with? While they are somewhat
abstract, their use facilitates the mathematics of the solution.
• While it is beyond this course to fully explore complex variables, we will touch upon the
fundamentals and use this complex plane Argand diagram for reference.
• A complex variable z is shown as a vector in the plot and can be written as;
𝑖 = −1 = 𝑖𝑚𝑎𝑔𝑖𝑛𝑎𝑟𝑦 𝑛𝑢𝑚𝑏𝑒𝑟
𝑧 = 𝑥 + 𝑖𝑦 where 𝑥 = 𝑟𝑒𝑎𝑙 𝑝𝑎𝑟𝑡 𝑜𝑓 𝑧
𝑦 = 𝑖𝑚𝑎𝑔𝑖𝑛𝑎𝑟𝑦 𝑝𝑎𝑟𝑡 𝑜𝑓 𝑧
• From the Argand diagram and simple trigonometry, we can observe that;
𝑥 + 𝑖𝑦 = A(cos 𝜙 + 𝑖 sin 𝜙)
• So, depending on the angle 𝜙, we have varying real and imaginary quantities, the animation
(phasor plot) shows the tracing of the real portion (cosine) of the function.
• Notice how it is an oscillatory steady state behavior smoothly transitioning from max to min.

6
Complex Variables and Notation
• Complex variables can be written in two common forms;

𝑧 = 𝑥 + 𝑖𝑦 = 𝐴𝑒 𝑖𝜙 𝐴= 𝑥2 + 𝑦2 Amplitude
with
𝑦
𝑧 = 𝑥 + 𝑖𝑦 = A(cos 𝜙 + 𝑖 sin 𝜙) 𝜙 = tan−1 Phase angle
𝑥

• While the trigonometric form intuitively illustrates that the function has a sinusoidal shape as shown previously,
the compact, mathematically equivalent and elegant exponential form will be used going forward.

7
Displacement, Velocity, and Acceleration
• While we have discussed force and displacement, other quantities of interest include the velocities and
accelerations. So far, we have just discussed the force and displacement.
• We know from physics the following;
𝑑𝑢 𝑑𝑢ሶ
𝑢ሶ = = 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑢ሷ = = 𝐴𝑐𝑐𝑒𝑙𝑒𝑟𝑎𝑡𝑖𝑜𝑛
𝑑𝑡 𝑑𝑡

• So given that we have a complex form of the displacement that represents our sinusoidal displacement;
𝑢 = 𝐴𝑒 𝑖Ω𝑡
• The velocity and Accelerations are computed by differentiating the exponential form;
𝑢ሶ = 𝑖Ω𝐴𝑒 𝑖Ω𝑡 𝑢ሷ = −Ω2 𝐴𝑒 𝑖Ω𝑡
• To visualize this, let’s momentarily revert to the trigonometric form of the equations;

𝑢 = 𝐴 (cos Ω𝑡 + 𝑖 sin Ω𝑡) 𝑢ሶ = Ω𝐴 (𝑖cos Ω𝑡 − sin Ω𝑡) 𝑢ሷ = −Ω2 𝐴 (cos Ω𝑡 + 𝑖 sin Ω𝑡)

8
Displacement, Velocity, and Acceleration

• Here we can see easily see graphically how a 1.50 8.00E+03

6.00E+03
harmonic steady state response varies over time. 1.00
4.00E+03

Displacement
0.50
• Notice how the velocity is shifted to be 90 degrees 2.00E+03

Velocity
0.00 0.00E+00
out of phase from the displacements. -2.00E+03
-0.50
• When the displacement is at maximum or -1.00
Displacement -4.00E+03
Velocity -6.00E+03
minimum, the velocity is zero. -1.50 -8.00E+03
• Also notice how acceleration is shifted 180 0 0.0005 0.001
Time
0.0015 0.002 0.0025

degrees out of phase from the displacements (or

90 degrees from velocity). 1.50 5.00E+07
4.00E+07
• Additionally when the displacement is at a 1.00 3.00E+07
2.00E+07
maximum, the acceleration is at a minimum, and

Displacement

Acceleration
0.50
1.00E+07
vice-versa. 0.00 0.00E+00
-1.00E+07
-0.50 -2.00E+07
Displacement
-1.00 -3.00E+07
Acceleration -4.00E+07
-1.50 -5.00E+07
0 0.0005 0.001 0.0015 0.002 0.0025
Time

9
 Formulation

• Starting with the matrix form of the governing equations of motion, let’s explore
a harmonic solution to this equation.
𝑀 𝑢ሷ + 𝐶 𝑢ሶ + 𝐾 𝑢 = 𝐹 where
ψ = Force phase shift
• The Forcing Function 𝐹 and displacement 𝑢 are of the below form. Recall
ϕ = Displacement phase shift
that these can be visualized as sinusoids and simply given a value for time (t)
and circular frequency (Ω) we can compute the complex values of the force or Ω = Imposed Circular Frequency
displacement at each node, once we have of course solved for the unknown max 𝜔 = Natural Circular Frequency
values and phase shifts for each node in the model. ζ = Damping Ratio
𝐹 = 𝐹𝑚𝑎𝑥 𝑒 𝑖𝜓 𝑒 𝑖Ω𝑡 𝑢 = 𝑢𝑚𝑎𝑥 𝑒 𝑖𝜙 𝑒 𝑖Ω𝑡 𝑢1 = 𝑅𝑒𝑎𝑙 𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 𝑣𝑒𝑐𝑡𝑜𝑟
𝑢2 = 𝐼𝑚𝑎𝑔𝑖𝑛𝑎𝑟𝑦 𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 𝑣𝑒𝑐𝑡𝑜𝑟
• We differentiate the displacements to get velocity and acceleration, substitute
𝐹1 = 𝑅𝑒𝑎𝑙 𝑓𝑜𝑟𝑐𝑒 𝑣𝑒𝑐𝑡𝑜𝑟
and simplify resulting in the harmonic equations of motion
𝐹2 = 𝐼𝑚𝑎𝑔𝑖𝑛𝑎𝑟𝑦 𝑓𝑜𝑟𝑐𝑒 𝑣𝑒𝑐𝑡𝑜𝑟
−Ω2 𝑀 + 𝑖Ω 𝐶 + 𝐾 ሼ𝑢1 ሽ + 𝑖 ሼ𝑢2 ሽ = ሼ𝐹1 ሽ + 𝑖 ሼ𝐹2 ሽ
𝑦𝑗𝑐 = 𝐶𝑜𝑚𝑝𝑙𝑒𝑥 𝑀𝑜𝑑𝑎𝑙 𝐶𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒𝑠
• Using the mode superposition method and omitting all the details, the equation 𝑓𝑗𝑐 = 𝐶𝑜𝑚𝑝𝑙𝑒𝑥 𝑀𝑜𝑑𝑎𝑙 𝐹𝑜𝑟𝑐𝑒
can be converted into modal form.
−Ω2 + 𝑖2𝜔𝑗 Ω𝜁𝑗 + 𝜔𝑗2 𝑦𝑗𝑐 = 𝑓𝑗𝑐
Formulation

• The modal form of the equation (repeated here) has the complex modal coordinates 𝑦𝑗𝑐 as the unknowns, which
are fewer to solve than the physical DOFs as unknowns.
𝑓𝑗𝑐
−Ω2 + 𝑖2𝜔𝑗 Ω𝜁𝑗 + 𝜔𝑗2 𝑦𝑗𝑐 = 𝑓𝑗𝑐 𝑦𝑗𝑐 =
(𝑤𝑗2 −Ω2 ) + 𝑖(2𝑤𝑗 Ω𝜁𝑗 )
• It is computationally efficient to compute since the equations are uncoupled (only one dependent variable per
equation).

• From the mode superposition method we then compute the displacement contribution from each mode;

𝐶𝑗 = 𝜙𝑗 𝑦𝑗𝑐 Where 𝐶𝑗 = Contribution of mode j

• Using 𝑦𝑗𝑐 and the mode shapes (eigenvectors) 𝜙𝑗 we compute the complex displacements by summing over the
n modes.
𝑛

𝑢𝑐 = ෍ 𝐶𝑗
𝑗=1 Please see the Mode Superposition Section which gives further insights into the method
Discussion (Damping and Resonance)

• A few points can be made by looking at the modal coordinates equation.

𝑓𝑗𝑐
𝑦𝑗𝑐 =
(𝑤𝑗2 −Ω2 ) + 𝑖(2𝑤𝑗 Ω𝜁𝑗 )

• When the excitation frequency (Ω) is equal to the natural frequency 𝜔 , and the damping zeta (𝜁) = 0, then the
modal coordinate (𝑦𝑗𝑐 ) is infinite.

• In a physical system, exiting at the natural frequency or resonance (Ω = 𝜔) may happen, but there should
always be some amount of damping in the system to prevent predicting a non-physical infinite response.

• In addition, when exiting at the natural frequency or resonance (Ω = 𝜔) the phase angle of the response to the
input is 90 degrees. We will discuss this in more detail in the next section.
 Discussion (Amplitude and Phase vs. Real and Imaginary)

• As discussed, prior, the output from the harmonic solution is the complex displacement vector.
𝑛

𝑢𝑐 = ෍ 𝐶𝑗
𝑗=1
• This provides the real and imaginary values of the result, but often a more useful form of the results are the
amplitude and phase.
• Recall from the discussion on complex notation the relationship, where x is the real value and y is the imaginary value.
𝑧 = 𝑥 + 𝑖𝑦 = 𝐴𝑒 𝑖𝜙 𝐴= 𝑥2 + 𝑦2 Amplitude
𝑦
𝑧 = 𝑥 + 𝑖𝑦 = A(cos 𝜙 + 𝑖 sin 𝜙) 𝜙 = tan−1 Phase angle
𝑥
• We can therefore compute the amplitude of the steady state response, which is typically more useful than knowing
the real or imaginary results.
• In one way the real and imaginary can be thought of as the portion of the amplitude at a given time based on the
angle as we saw in the phasor plot during the discussion on complex notation. As we march along in time, the
quantity Ω𝑡 is changing resulting in different values of the real (cos Ω𝑡) and imaginary (sin Ω𝑡) components of the
displacement.
𝑢 = 𝑢𝑚𝑎𝑥 𝑒 𝑖𝜙 𝑒 𝑖Ω𝑡 Complex displacement equation for referencea

Complex displacement equation for reference

Discussion (Stress and Strain Output)

• Up until now we have discussed the results quantities of displacement, velocity and acceleration.
• But as we have seen in other sections, the strains are derived from the displacements and the stresses
from the strains and the constitutive model (in the case of linear elastic materials – Hooke’s law)
• It is often of interest to understand the harmonic (sinusoidal) stress levels as these will be repeating at
known frequencies and using these results, fatigue damage and life predictions can be made.
• Finally, while the stress and strain components (normal and shears) are harmonic (sinusoidal), popular
derived quantities like principal stresses or equivalent stresses are not.

Complex displacement equation for reference