Higher-Order Linear Differential Equations

Introduction and engineering context

Higher-order differential equations model systems whose present behavior depends on several derivatives of a state variable. In mechatronics, these derivatives commonly represent velocity, acceleration, jerk, current rate of change, or actuator dynamics. A mass–spring–damper system, for example, is governed by a second-order equation, while flexible structures, coupled motors, filters, and multivariable control systems often lead to third- or higher-order equations.

A general nth-order linear ordinary differential equation (ODE) is

Here, y is the dependent variable, t is the independent variable, y^(k) is the kth derivative of y, a_k(t) are coefficient functions, and g(t) is the forcing function. The equation is linear because y and its derivatives occur only to the first power and are not multiplied together. It is homogeneous when g(t) = 0 and nonhomogeneous when g(t) ≠ 0.

This chapter develops methods for constant- and variable-coefficient equations, including characteristic roots, undetermined coefficients, variation of parameters, and power-series solutions. The methods are connected to vibration, electrical circuits, and actuator dynamics.

Learning objectives

By the end of this chapter, you should be able to:

1. Classify higher-order linear ODEs and state their initial and boundary conditions.
2. Solve constant-coefficient homogeneous equations using characteristic roots.
3. Construct particular solutions by undetermined coefficients.
4. Use variation of parameters when the forcing function is not suitable for undetermined coefficients.
5. Apply power-series methods to variable-coefficient equations.
6. Interpret roots in terms of stability, oscillation, damping, and resonance.
7. Formulate and solve engineering models involving mechanical, electrical, and electromechanical systems.

1. Structure, linearity, and solution spaces

For a linear equation, the superposition principle applies. If y₁ and y₂ solve the homogeneous equation, then any linear combination c₁y₁ + c₂y₂ also solves it. If y_p is one particular solution of the nonhomogeneous equation, then

is the complete solution, where y_h is the general homogeneous solution. The constants in y_h are determined by initial or boundary conditions.

An nth-order equation normally requires n independent conditions. Initial conditions are specified at one point, such as y(0), y'(0), and y''(0). Boundary conditions are specified at different points, such as y(0) and y(L). A well-posed problem requires suitable coefficient continuity and conditions that produce a unique solution.

For the normalized equation

where the coefficients are continuous on an interval, existence and uniqueness are guaranteed locally when n initial values are specified at a point in that interval.

2. Constant-coefficient homogeneous equations

Consider

where the a_k are constants and a_n ≠ 0. Try an exponential solution y = e^{rt}. Since y^(k) = r^k e^{rt}, substitution gives

Because e^{rt} is never zero, r must satisfy the characteristic equation

The roots determine the form of the solution.

Root cases

Complex roots occur in conjugate pairs when the coefficients are real. Euler’s identity, e^{iβt} = cos βt + i sin βt, converts complex exponential solutions into real solutions.

Stability interpretation

For a free response, a root with positive real part produces exponential growth; a negative real part produces decay; and a zero real part produces sustained or marginal behavior. Thus, asymptotic stability requires every characteristic root to have negative real part. A pair α ± iβ gives oscillation at angular frequency |β| radians per unit time with an envelope proportional to e^{αt}.

Worked Example 1: repeated and complex roots

Solve

The characteristic polynomial is

The roots are −1, −1, i, and −i. Therefore,

The terms containing e^(−t) decay, while the sine and cosine terms represent undamped oscillations. Consequently, the homogeneous system is not asymptotically stable because two roots lie on the imaginary axis.

3. Nonhomogeneous equations and undetermined coefficients

For

where L is a constant-coefficient differential operator, the method of undetermined coefficients is effective when g(t) is a finite combination of exponentials, polynomials, sines, and cosines.

The trial form must contain all functions generated by differentiation of the forcing function. Typical choices are:

If any part of the trial solution is already in y_h, multiply the entire trial by t^s, where s is the multiplicity of the corresponding characteristic root.

Worked Example 2: forced actuator model

Consider

The characteristic equation is r² + 3r + 2 = (r + 1)(r + 2), so

Because e^(−t) is already part of y_h, use

Then y_p' = A(1 − t)e^(−t) and y_p'' = A(t − 2)e^(−t). Substitution gives

Thus A = 4, and

The factor t is required because the forcing frequency corresponds to a simple characteristic root. This is the mathematical signature of resonance-like amplification in the operator response.

4. Resonance and engineering interpretation

For an undamped mechanical oscillator,

where m is mass in kilograms, k is stiffness in newtons per metre, F₀ is force amplitude in newtons, and ω is forcing angular frequency in radians per second. The natural frequency is

For ω ≠ ω_n, a harmonic particular solution has bounded amplitude. At ω = ω_n, the forcing frequency matches a homogeneous frequency and the particular solution takes the form

under zero damping. Its amplitude grows proportionally to t in the ideal model. Real damping, nonlinearities, actuator saturation, and structural losses limit this growth.

For the damped model

c is the viscous damping coefficient in N·s/m. The damping ratio is

The response is underdamped for 0 < ζ < 1, critically damped for ζ = 1, and overdamped for ζ > 1. These classifications follow from the roots of mr² + cr + k = 0.

5. Variation of parameters

Undetermined coefficients is limited to special forcing functions. Variation of parameters applies more broadly. For a second-order normalized equation

suppose y₁ and y₂ are linearly independent solutions of the homogeneous equation. Their Wronskian is

A particular solution is

This formula assumes the equation has first been divided by the coefficient of y''. The integrals may be definite or indefinite; constants of integration are absorbed into y_h.

For an nth-order equation with fundamental solutions y₁, …, y_n, one may set y_p = Σu_i(t)y_i(t), impose the standard auxiliary conditions, and solve the resulting linear system for u_i'. This is systematic but often algebraically intensive.

Worked Example 3: a forcing function unsuitable for undetermined coefficients

Solve

on an interval that does not cross a point where cos t = 0. The homogeneous solutions are y₁ = cos t and y₂ = sin t, with W = 1. Therefore,

Since sin t tan t = sin²t/cos t = sec t − cos t and cos t tan t = sin t,

and ∫sin t dt = −cos t. Hence

The final two trigonometric terms cancel, giving

Thus y = C₁ cos t + C₂ sin t − cos t ln|sec t + tan t|. The interval restriction is essential because tan t and the logarithm are not defined everywhere.

6. Systems, circuits, and actuator dynamics

Mechanical vibration

Newton’s second law yields

The state variables may be selected as x₁ = x and x₂ = x'. Then

The second-order ODE and the first-order state model are mathematically equivalent. The state form is particularly useful for simulation and control design.

Series RLC circuit

For a series resistor–inductor–capacitor circuit, Kirchhoff’s voltage law gives

where q is charge in coulombs, q' is current in amperes, L is inductance in henries, R is resistance in ohms, C is capacitance in farads, and E is applied voltage in volts. Dividing by L gives a second-order linear ODE. The characteristic roots determine whether the circuit is overdamped, critically damped, or underdamped.

DC motor actuator

A simplified armature-controlled DC motor can be represented by

Here J is rotor inertia, b is viscous friction, K_t is torque constant, i is current, L is armature inductance, R is resistance, v is applied voltage, K_e is back-emf constant, and θ is angular position. Eliminating i produces a higher-order equation in θ. In SI units, K_t and K_e have numerically compatible values under standard motor conventions, although their physical units differ.

7. Variable-coefficient equations and power series

When coefficients depend on t, the characteristic-root method generally does not apply. A power-series method assumes

where a_n are constants and t₀ is the expansion point. Derivatives are obtained term by term within the radius of convergence:

The procedure is to substitute the series, shift indices so all terms use the same power, equate coefficients, derive a recurrence relation, and use initial conditions to determine the arbitrary constants.

Worked Example 4: series solution

Find a series solution near t = 0 for

Let y = Σ a_n t^n. Then

and

Equating coefficients gives a₂ = 0 and, for n ≥ 1,

The coefficients divide into three index chains. With a₀ and a₁ arbitrary, the first terms are

This solution is locally valid near the expansion point and, for this entire-coefficient equation, the resulting series has an extensive convergence domain.

Ordinary and regular singular points

For

a point t₀ is ordinary if p and q are analytic there. If coefficients are singular but (t−t₀)p(t) and (t−t₀)²q(t) are analytic, t₀ is a regular singular point. At a regular singular point, the Frobenius method assumes

where r is determined by an indicial equation. Frobenius analysis is important in models involving cylindrical, spherical, or boundary-layer geometries.

8. Illustrative case study: positioning actuator with flexible load

Background

Consider an illustrative servo actuator positioning a lightweight arm. The motor drives the arm through an elastic transmission. The motor inertia and elastic flexibility produce a dominant second-order mode, while electrical inductance adds a third dynamic state.

Problem

The designer must predict whether a commanded sinusoidal position will cause excessive oscillation near the flexible mode. A simplified mechanical transfer equation is represented by

where τ_c is commanded torque. Suppose J = 0.02 kg·m², b = 0.04 N·m·s/rad, and k = 8 N·m/rad.

Analysis

The undamped natural frequency is

The damping ratio is

The system is lightly underdamped. Its damped frequency is

A command near 20 rad/s will therefore produce a large response unless damping or control compensation is increased.

Approach and solution

The engineering response should include: identifying the characteristic roots; measuring the actual frequency response; adding velocity feedback or a notch filter near the flexible resonance; limiting command bandwidth; and checking actuator torque and sensor noise. The ideal differential equation is only an approximation, so validation against experimental data is necessary.

Outcome

If velocity feedback increases effective damping while preserving adequate bandwidth, the resonance peak is reduced. A notch filter can attenuate the narrow frequency range, but excessive filtering may slow positioning. The final design must balance tracking accuracy, robustness, noise sensitivity, and saturation limits.

Learning points

1. Characteristic roots predict decay and oscillation.
2. Parameter units are essential: k/J has units s⁻², so its square root has units s⁻¹.
3. A mathematically stable model can still perform poorly near resonance.
4. Model-based predictions must be verified experimentally.

9. Method selection, limitations, and best practice

Common errors include omitting repeated-root factors, failing to normalize an equation before variation of parameters, using an incompatible trial form, confusing angular frequency with frequency in hertz, and applying a local series outside its convergence interval.

Computer algebra and numerical simulation are valuable for factoring high-degree characteristic polynomials, checking derivatives, and comparing analytical and numerical responses. They do not replace model validation, dimensional analysis, or engineering judgment. In safety-critical systems, assumptions, parameter uncertainty, sensor limitations, and failure modes must be documented. A controller should not be approved solely because an idealized ODE has stable roots.

Chapter Summary

Higher-order linear ODEs consist of a homogeneous response and, when forcing is present, a particular response. Constant-coefficient homogeneous equations are solved through the characteristic polynomial. Distinct roots produce exponential terms, repeated roots require powers of t, and complex roots produce damped or growing oscillations. Root locations provide a direct interpretation of stability and resonance.

Undetermined coefficients is efficient for standard forcing functions but requires modification when the trial duplicates a homogeneous term. Variation of parameters is more general and uses the fundamental homogeneous solutions and their Wronskian. Variable-coefficient equations are often treated with power series or the Frobenius method. These tools support analysis of vibration systems, RLC circuits, and electromechanical actuators. Reliable engineering practice combines exact analysis with dimensional checks, numerical verification, experimental validation, and explicit attention to limitations.

Key Terms

Linear differential equation: An ODE in which the dependent variable and its derivatives appear linearly.

Order: The highest derivative appearing in an ODE.

Homogeneous equation: A linear equation whose forcing term is zero.

Nonhomogeneous equation: A linear equation with a nonzero forcing term.

Characteristic equation: The polynomial obtained by substituting e^{rt} into a constant-coefficient homogeneous ODE.

Characteristic root: A value of r satisfying the characteristic equation.

Repeated root: A characteristic root occurring more than once.

Complex conjugate roots: A pair α ± iβ arising from real coefficients.

Particular solution: Any one solution of a nonhomogeneous equation.

Superposition: Addition and scalar multiplication of solutions of a linear homogeneous equation.

Resonance: Large response caused by forcing near a natural frequency.

Wronskian: Determinant W = y₁y₂' − y₁'y₂ used to test independence and construct variation-of-parameters solutions.

Variation of parameters: A method that replaces constants in homogeneous solutions with functions.

Undetermined coefficients: A method using a trial function with unknown coefficients.

Power-series solution: A solution represented as an infinite polynomial series.

Frobenius method: A series method using powers (t−t₀)^{n+r} near a regular singular point.

Natural frequency: The characteristic oscillation rate of an unforced system.

Damping ratio: A dimensionless measure comparing actual damping with critical damping.

State variable: A variable that, together with other state variables, specifies the system’s future evolution.

Review Questions

5 Short Answer Questions

1. What distinguishes a homogeneous linear ODE from a nonhomogeneous linear ODE?
2. Why are factors of t introduced for repeated characteristic roots?
3. State the condition for asymptotic stability in terms of characteristic roots.
4. When is undetermined coefficients generally applicable?
5. What is the purpose of the Wronskian in variation of parameters?

5 Long Answer Questions

1. Derive the characteristic equation for a general constant-coefficient homogeneous ODE and explain the solution forms associated with real, repeated, and complex roots.
2. Explain the method of undetermined coefficients and demonstrate how resonance changes the trial solution.
3. Derive the variation-of-parameters formula for a normalized second-order equation.
4. Explain how a power-series solution is constructed for a variable-coefficient ODE, including index shifting.
5. Compare the mechanical vibration and series RLC circuit models, identifying corresponding parameters and stability interpretations.

3 Analytical / Application Questions

1. Solve y'' − 4y' + 4y = 0 and classify the stability of the homogeneous response.
2. For 2x'' + 0.6x' + 18x = 5 cos(3t), calculate the undamped natural frequency and damping ratio, then state whether the system is underdamped, critically damped, or overdamped.
3. A servo has J = 0.01 kg·m² and k = 4 N·m/rad. Determine its undamped natural frequency and explain why a command near that frequency may be problematic.

Multiple Choice Questions with Answers

Question 1

For the characteristic equation [[M:r^2 + 5r + 6 = 0]], what is the homogeneous solution?

A. [[M:C_1e^{-2t} + C_2e^{-3t}]]

B. [[M:C_1e^{2t} + C_2e^{3t}]]

C. [[M:C_1cos(2t) + C_2sin(3t)]]

D. [[M:(C_1+C_2t)e^{-3t}]]

Correct Answer: A. [[M:C_1e^{-2t} + C_2e^{-3t}]]

Question 2

What factor is required in a trial particular solution when the forcing term corresponds to a simple characteristic root?

A. 1/t

B. t

C. e^t

D. t^2

Correct Answer: B. t

Question 3

For a real characteristic root r = −2 with multiplicity three, which homogeneous contribution is correct?

A. [[M:C_1e^{-2t}]]

B. [[M:(C_1+C_2t+C_3t^2)e^{-2t}]]

C. [[M:(C_1+C_2t)e^{-2t}]]

D. [[M:C_1cos(2t)+C_2sin(2t)]]

Correct Answer: B. [[M:(C_1+C_2t+C_3t^2)e^{-2t}]]

Question 4

The roots α ± iβ produce which real solution form?

A. [[M:e^{αt}(C_1cos(βt)+C_2sin(βt))]]

B. [[M:e^{βt}(C_1cos(αt)+C_2sin(αt))]]

C. [[M:C_1e^{αt}+C_2e^{βt}]]

D. [[M:C_1t^α+C_2t^β]]

Correct Answer: A. [[M:e^{αt}(C_1cos(βt)+C_2sin(βt))]]

Question 5

For [[M:m x'' + c x' + kx = 0]], which condition gives asymptotic stability when m, c, and k are positive constants?

A. At least one characteristic root has positive real part

B. All characteristic roots have negative real parts

C. The roots must be purely imaginary

D. The forcing function must be sinusoidal

Correct Answer: B. All characteristic roots have negative real parts

Question 6

Which method is generally most suitable for a forcing function such as [[M:tan t]]?

A. Undetermined coefficients

B. Variation of parameters

C. Characteristic roots alone

D. Polynomial division

Correct Answer: B. Variation of parameters

Question 7

For a mass–spring system, the undamped natural frequency is which expression?

A. [[M:m/k]]

B. [[M:k/m]]

C. [[M:√(k/m)]]

D. [[M:√(m/k)]]

Correct Answer: C. [[M:√(k/m)]]

Question 8

In a series RLC circuit, what physical quantity is represented by [[M:q']]?

A. Voltage

B. Resistance

C. Current

D. Inductance

Correct Answer: C. Current

Question 9

For the actuator case with J = 0.02 kg·m² and k = 8 N·m/rad, what is [[M:ω_n]]?

A. 2 rad/s

B. 8 rad/s

C. 20 rad/s

D. 40 rad/s

Correct Answer: C. 20 rad/s

Question 10

At an ordinary point of a variable-coefficient ODE, which representation is commonly assumed for a local solution?

A. A finite Fourier series only

B. A power series in powers of [[M:(t−t_0)]]

C. A constant only

D. A characteristic polynomial with constant coefficients

Correct Answer: B. A power series in powers of [[M:(t−t_0)]]