Ordinary Differential Equations and Dynamical Models

Introduction

An ordinary differential equation (ODE) relates an unknown function of one independent variable to its derivatives. ODEs are central to engineering mathematics because physical laws usually describe rates of change: capacitor voltage changes with current, motor speed changes with torque, and a control-system output changes in response to an input. Unlike an algebraic equation, an ODE describes a process evolving in time or space.

A dynamical model consists of variables, parameters, governing equations, initial or boundary conditions, and an interpretation of the resulting solution. In electronics and communication engineering, ODE models arise in RC and RLC circuits, filters, oscillators, transmission systems, sensor interfaces, and feedback-control systems.

Learning Objectives

After studying this chapter, a student should be able to:

• classify first- and higher-order ODEs;
• solve separable, linear, exact, homogeneous and selected nonlinear equations;
• solve linear constant-coefficient equations and initial-value problems;
• formulate ODE models from electrical and mechanical laws;
• represent coupled equations as systems of first-order equations;
• analyze equilibrium points, phase behavior and stability;
• interpret eigenvalues in dynamical systems;
• recognize the limitations of idealized mathematical models.

1. Fundamental Concepts and Classification

A general nth-order ODE may be written as

Here, t is the independent variable, y(t) is the dependent variable, and y′, y″, … denote derivatives with respect to t. The order is the highest derivative present. The equation is linear if y and its derivatives occur only to the first power and are not multiplied together or placed inside nonlinear functions. Thus,

is linear, whereas y′ + y² = 0 and y′′ + sin y = 0 are nonlinear.

A homogeneous linear equation has g(t) = 0; otherwise it is nonhomogeneous. An initial-value problem (IVP) specifies values such as y(t₀) and y′(t₀) at one point. A boundary-value problem specifies conditions at different points, for example y(0) and y(L).

The general solution of an nth-order equation normally contains n arbitrary constants. Initial or boundary conditions determine these constants when a unique solution exists. The existence and uniqueness theorem for a first-order IVP

states that if f and ∂f/∂y are continuous in a region containing (t₀, y₀), then a unique local solution exists.

2. First-Order Differential Equations

2.1 Separable equations

An equation is separable when it can be expressed as

provided h(y) is nonzero on the interval under consideration. Separation gives

and integration produces an implicit or explicit solution.

Worked Example 1: Exponential charging model

Suppose a population or signal amplitude satisfies dx/dt = kx, where k is a constant with units of inverse time. Separating and integrating,

so ln|x| = kt + C and therefore

For x(0) = x₀, C₁ = x₀, giving x(t) = x₀e^{kt}. If k < 0, the response decays; if k > 0, it grows. The exponential argument kt is dimensionless, so k must have units s⁻¹ when t is measured in seconds.

2.2 First-order linear equations

The standard form is

where p and q are known functions. The integrating factor is

Multiplying the equation by μ gives

Thus,

The formula applies on an interval where p and q are continuous.

Worked Example 2: A first-order IVP

Solve y′ + 2y = 6, y(0) = 1. The integrating factor is μ = e^{2t}. Hence

or d(e^{2t}y)/dt = 6e^{2t}. Integration gives e^{2t}y = 3e^{2t} + C, so y = 3 + Ce^{-2t}. Applying y(0) = 1 gives C = −2. Therefore,

The steady-state value is 3, while the transient term −2e⁻²ᵗ decays with time constant 1/2 time units.

2.3 Exact equations

An equation

is exact if there is a potential function Φ(t,y) such that dΦ = Mdt + Ndy. The test is

on a simply connected region. The solution is Φ(t,y) = C. If the test fails, an integrating factor may sometimes convert the equation into an exact one.

2.4 Logistic growth

A useful nonlinear model is

where x is the population or amplitude, r is the intrinsic growth rate, and K is the carrying capacity. Both r and 1/t have the same units, while K has the units of x. For x₀ = x(0), the solution is

for x₀ ≠ 0. The equilibria are x = 0 and x = K. When r > 0, the first is unstable and the second is stable for positive initial values.

3. Higher-Order Linear Equations

A second-order constant-coefficient equation has the form

The associated homogeneous equation is solved by assuming y = e^{λt}. This produces the characteristic equation

The nature of its roots determines the response.

For a nonhomogeneous equation, the complete solution is

where y_h is the complementary or homogeneous solution and y_p is any particular solution. The method of undetermined coefficients is convenient when g(t) is a polynomial, exponential, sine or cosine. Variation of parameters is more general.

Worked Example 3: A damped oscillator

Consider

The characteristic equation is λ² + 4λ + 3 = (λ + 1)(λ + 3) = 0. Hence y = C₁e⁻ᵗ + C₂e⁻³ᵗ. The initial conditions give C₁ + C₂ = 1 and −C₁ − 3C₂ = 0. Solving yields C₁ = 3/2 and C₂ = −1/2. Therefore,

Both roots are negative, so the response approaches zero without sustained oscillation. The term e⁻ᵗ is the slower transient and dominates at large t.

3.1 Standard mechanical and electrical forms

The mass–spring–damper model is

where m is mass, c is damping coefficient, k is stiffness, and f is applied force. In an ideal series RLC circuit, Kirchhoff’s voltage law gives

where q is charge in coulombs, i = q′ is current in amperes, L is inductance in henries, R is resistance in ohms, C is capacitance in farads, and v_s is source voltage in volts. Dimensional consistency follows because Lq″, Rq′ and q/C all have units of volts.

The natural frequency and damping ratio of a normalized second-order system are

for a mechanical system, or

for a series RLC circuit. These expressions assume positive physical parameters and the standard second-order form. The system is underdamped when 0 < ζ < 1, critically damped when ζ = 1, and overdamped when ζ > 1.

4. Systems of First-Order ODEs

A second-order equation can be rewritten as two first-order equations by defining state variables. For an RLC circuit, let x₁ = q and x₂ = i. Then

In matrix form,

where x is the state vector, A is the system matrix, B is the input matrix, and u(t) is the input. The output may be written y = Cx + Du.

For a homogeneous system x′ = Ax, solutions can be expressed as

The matrix exponential may be evaluated through eigenvalues, diagonalization, or a power series. If A has independent eigenvectors vᵢ with eigenvalues λᵢ, then each mode has the form cᵢvᵢe^{λᵢt}.

Worked Example 4: Eigenvalue-based stability

Consider

The characteristic polynomial is det(A − λI) = λ² + 2λ + 3. Its roots are λ = −1 ± j√2. Both eigenvalues have negative real part, so every solution tends to the origin. The trajectories oscillate while their amplitude decays approximately as e⁻ᵗ.

5. Equilibria, Phase Behavior and Stability

For an autonomous system x′ = f(x), an equilibrium x* satisfies

In a scalar equation x′ = f(x), a phase line indicates the sign of f(x). Arrows point toward increasing x when f(x) > 0 and toward decreasing x when f(x) < 0. An equilibrium is stable if sufficiently small perturbations remain small; it is asymptotically stable if perturbations also decay to the equilibrium.

For a two-dimensional system x′ = f(x,y), y′ = g(x,y), trajectories in the x–y plane form a phase portrait. The nullclines are curves f(x,y) = 0 and g(x,y) = 0. Their intersections are equilibria. Direction fields and trajectories reveal oscillation, convergence, divergence and possible limit cycles.

For nonlinear systems, linearization near an equilibrium uses the Jacobian matrix

If all eigenvalues of J have negative real parts, the equilibrium is locally asymptotically stable. If at least one eigenvalue has positive real part, it is unstable. If eigenvalues have zero real parts, linearization alone is inconclusive and nonlinear terms must be examined.

For a two-dimensional linear system, the trace T and determinant D satisfy

A necessary and sufficient condition for asymptotic stability is T < 0 and D > 0. This is a special two-dimensional result; higher-dimensional systems require the eigenvalue or Routh–Hurwitz criteria.

6. Numerical Solution of Initial-Value Problems

Closed-form solutions are not always available. For y′ = f(t,y), the forward Euler method is

where h is the step size and t_{n+1} = t_n + h. Its local truncation error is O(h²) and its global error is O(h), assuming sufficient smoothness.

The classical fourth-order Runge–Kutta method uses four slope estimates:

The method has global error O(h⁴) under standard smoothness assumptions. Numerical stability is also important. For a rapidly decaying equation y′ = λy with λ < 0, Euler’s method is stable only when |1 + hλ| < 1. Thus a method can be accurate in principle but unstable for an unsuitable step size.

7. Detailed Application: RLC Transient Response

Background

A series RLC circuit contains a resistor, inductor and capacitor connected to a voltage source. Such circuits model resonators, filters and communication-system tuning networks.

Problem

Assume a source-free circuit with L = 1 H, R = 4 Ω and C = 1/3 F. Let the initial charge be q(0) = 1 C and initial current be i(0) = 0 A. Determine q(t) and describe the response.

Analysis and solution

Kirchhoff’s law gives

The characteristic roots are −1 and −3. Consequently,

Since q(0) = 1, C₁ + C₂ = 1. Because i = q′ and i(0) = 0,

Solving gives C₁ = 3/2 and C₂ = −1/2. Therefore,

and

The natural frequency is ω_n = 1/√(LC) = √3 rad/s. The damping ratio is ζ = R/2 × √(C/L) = 2/√3 ≈ 1.155, so the circuit is overdamped. It returns to equilibrium without oscillation. The result also illustrates that charge and current have different units and must not be interchanged in the model.

Outcome and learning points

The transient is a sum of two decaying modes. The slower mode, e⁻ᵗ, determines the long-time behavior. Increasing resistance generally increases damping, but excessive resistance can reduce selectivity in a resonant filter. In practical design, component tolerances, parasitic resistance, temperature and measurement loading alter the ideal response.

8. Engineering Practice, Limitations and Contemporary Tools

A reliable modeling workflow is: define system boundaries; identify states and inputs; apply conservation laws; check dimensions; specify initial conditions; solve or simulate; compare predictions with measurements; and document assumptions. Initial conditions must be physically compatible with stored energy. For example, capacitor voltage and inductor current cannot change discontinuously without an impulse source in the ideal model.

Ideal ODE models omit nonlinear saturation, switching, noise, quantization, parasitic elements and parameter uncertainty. Linearization is valid only near the selected operating point. Numerical results require convergence and sensitivity checks. In control and safety-critical applications, stability margins, sensor faults, actuator limits and fail-safe behavior must be considered.

Computer algebra and numerical software can calculate solutions, eigenvalues and phase portraits, but software output does not replace model validation. AI-assisted tools may help generate code or suggest model forms; engineers remain responsible for checking equations, units, boundary conditions, numerical stability, data provenance and reproducibility. In communication systems, privacy and secure handling of measured signals are additional professional responsibilities.

Chapter Summary

ODEs model quantities that vary with time or another single variable. First-order equations include separable, linear and exact forms; integrating factors solve first-order linear equations. Higher-order linear equations are solved through characteristic roots and particular solutions. The roots determine exponential decay, growth, repeated-mode behavior and oscillation.

Electrical and mechanical systems lead naturally to second-order equations and first-order state-space systems. Eigenvalues determine the modes and local stability of linear systems. Equilibria and phase portraits describe nonlinear behavior, while Jacobian linearization provides local conclusions except at nonhyperbolic equilibria. Euler and Runge–Kutta methods provide numerical approximations, but step size and stability must be considered. Sound engineering practice combines mathematical solution, dimensional checking, simulation, measurement and explicit recognition of model limitations.

Key Terms

ODE: An equation involving an unknown function of one independent variable and its derivatives.

Order: The highest derivative appearing in a differential equation.

Linear ODE: An ODE in which the unknown function and its derivatives occur linearly.

Initial-value problem: A differential equation together with conditions specified at one independent-variable value.

Separable equation: An equation whose variables can be placed in separate factors.

Integrating factor: A function that converts a first-order linear equation into an exact derivative.

Exact equation: An equation obtained as the total differential of a potential function.

Homogeneous solution: The solution of the associated equation with zero forcing.

Particular solution: Any solution of a nonhomogeneous differential equation.

Characteristic equation: The algebraic equation obtained by substituting an exponential trial solution.

Transient response: The component that changes or decays from initial conditions.

Steady-state response: The long-time response associated with persistent forcing or equilibrium.

State variable: A variable that, with other state variables, describes the system at an instant.

State-space model: A first-order vector representation of a dynamical system.

Equilibrium: A state at which all time derivatives are zero.

Phase portrait: A graphical representation of trajectories in state space.

Stability: The property that small perturbations do not produce unbounded deviation.

Jacobian: The matrix of first partial derivatives used in local linearization.

Damping ratio: A dimensionless measure of damping relative to critical damping.

Numerical stability: The ability of a numerical method to control errors during computation.

Review Questions

5 Short Answer Questions

1. Distinguish between a homogeneous and a nonhomogeneous linear ODE.
2. What is the purpose of an integrating factor?
3. State the characteristic equation for ay″ + by′ + cy = 0.
4. Define an equilibrium point of an autonomous system.
5. Why must the units of terms in an ODE be compatible?

5 Long Answer Questions

1. Derive the solution of a first-order linear equation using an integrating factor and illustrate the method with an IVP.
2. Explain the three possible root patterns of a second-order constant-coefficient equation and relate each to physical response.
3. Formulate a series RLC circuit as both a second-order equation and a first-order state-space system.
4. Explain how eigenvalues determine the stability and oscillatory behavior of a linear system.
5. Compare Euler’s method with fourth-order Runge–Kutta in terms of procedure, accuracy and stability.

3 Analytical / Application Questions

1. A series RLC circuit has L = 2 H, R = 6 Ω and C = 1/8 F. Classify its damping and determine the characteristic roots of its source-free charge equation.
2. For the system x′ = x(1 − x) and y′ = −2y, find the equilibrium points and analyze their local stability.
3. A sensor obeys τy′ + y = Ku(t). Explain how τ and K affect the response to a step input, and identify assumptions under which this first-order model may fail.

Multiple Choice Questions with Answers

Question 1

Which condition identifies a first-order linear ODE?

A. The dependent variable appears only inside a sine function.

B. The equation can be written as y′ + p(t)y = q(t).

C. The highest derivative is always second order.

D. The equation contains a product of y and y′.

Correct Answer: B. The equation can be written as y′ + p(t)y = q(t).

Question 2

What is the integrating factor for y′ + 3y = q(t)?

A. e^t

B. 3t

C. e^{3t}

D. e^{q(t)}

Correct Answer: C. e^{3t}

Question 3

The characteristic roots of y″ + 5y′ + 6y = 0 are

A. 1 and 6

B. −2 and −3

C. 2 and 3

D. −1 and −6

Correct Answer: B. −2 and −3

Question 4

A second-order system with characteristic roots −1 ± j4 has a response that is

A. Growing and nonoscillatory

B. Decaying and oscillatory

C. Constant and nonoscillatory

D. Growing and oscillatory

Correct Answer: B. Decaying and oscillatory

Question 5

For a series RLC circuit, the charge equation is

A. LCq″ + Rq′ + q = v_s(t)

B. Lq″ + Rq′ + (1/C)q = v_s(t)

C. Rq″ + Lq′ + Cq = v_s(t)

D. Cq″ + Lq′ + Rq = v_s(t)

Correct Answer: B. Lq″ + Rq′ + (1/C)q = v_s(t)

Question 6

An equilibrium of x′ = f(x) satisfies

A. f(x*) = 1

B. f(x) = x

C. f(x*) = 0

D. x* = t

Correct Answer: C. f(x*) = 0

Question 7

For a two-dimensional linear system, which condition guarantees asymptotic stability?

A. Trace positive and determinant positive

B. Trace negative and determinant positive

C. Trace zero and determinant negative

D. Trace positive and determinant negative

Correct Answer: B. Trace negative and determinant positive

Question 8

The global error order of the forward Euler method is

A. O(h)

B. O(h²)

C. O(h³)

D. O(h⁴)

Correct Answer: A. O(h)

Question 9

If ζ = 1 for a standard second-order system, the system is

A. Underdamped

B. Critically damped

C. Overdamped

D. Unstable by definition

Correct Answer: B. Critically damped

Question 10

For y′ = −2y, y(0) = 3, the exact solution is

A. 3e^{2t}

B. 2e^{−3t}

C. 3e^{−2t}

D. −3e^{−2t}

Correct Answer: C. 3e^{−2t}