The gateway to modeling oscillations, damping, and dynamic systems
Chapter 2 introduces the general second-order linear differential equation and teaches you three complementary methods to solve it. By the end of this chapter, you will be able to solve any non-homogeneous linear second-order DE with constant coefficients, and understand how to model real-world systems like springs, dampers, and electrical circuits.
While first-order DEs model systems with a single rate of change (like exponential growth or radioactive decay), second-order DEs model systems with two interacting effects: inertia and restoring forces. These are ubiquitous in engineering:
All problems in this chapter boil down to solving:
where \(a, b, c\) are constants (with \(a \neq 0\)), \(g(t)\) is the external forcing function, and \(y\) is the unknown solution we seek.
The brilliant feature of linear differential equations is that the general solution decomposes into two independent pieces:
\(y_c\): The natural frequency of the string (determined by its length, tension, and material).
\(y_p\): The specific note you hear when you pluck it in a particular way (the input matters).
\(y_c\): How the springs and dampers naturally respond to a sudden impulse.
\(y_p\): How the suspension responds to the road bumps you're driving over.
Here's the magic: \(y_c\) and \(y_p\) are independent. The natural oscillation of the system (y_c) doesn't interfere with its response to forcing (y_p). This is why linear systems are so easy to work with — you can find each piece separately, then add them together. That's superposition.
This chapter is organized as a logical progression. Each section builds on the previous one and addresses a different aspect of the problem:
Find \(y_c\): the natural dynamics of the system using the characteristic equation.
Go to 2.1Find \(y_p\) when \(g(t)\) is "nice" (polynomial, exponential, sinusoid). Fast method for standard inputs.
Go to 2.2Find \(y_p\) for ANY forcing function \(g(t)\). The general method. Uses the Wronskian.
Go to 2.3When the system itself changes: coefficients depend on the independent variable.
Go to 2.4How the methods connect:
2.2 is faster when \(g(t)\) is simple. 2.3 works for any \(g(t)\). 2.4 handles variable coefficients.
The reason we can decompose \(y = y_c + y_p\) is that our equation is linear. This means the system respects the superposition principle:
If \(y_1(t)\) and \(y_2(t)\) are solutions to a linear differential equation, then any linear combination \(C_1 y_1(t) + C_2 y_2(t)\) is also a solution.
Applied to our problem: because the equation is linear, we can find \(y_c\) (the solution to the homogeneous equation \(ay'' + by' + cy = 0\)) and \(y_p\) (any particular solution to the non-homogeneous equation) separately, then add them to get the general solution to the full problem.
This is profound: the natural behavior of the system and its forced response are independent. A guitar string's natural vibration doesn't "know about" how you pluck it. A car's suspension has the same natural oscillation whether you drive over a pothole or a brick. The system decomposes cleanly.
This principle applies to all linear systems — mechanical, electrical, thermal, chemical. It's one of the most beautiful and practical ideas in mathematics.
The quintessential second-order model is the mass-spring-damper system, which appears (in different forms) in virtually every engineering discipline:
where:
Describes what happens when \(F(t) = 0\). The solution depends on the discriminant \(\Delta = c^2 - 4mk\):
Describes the system's steady-state response to the specific forcing \(F(t)\). For example:
When you design a car suspension, you choose \(m\), \(c\), and \(k\) so that \(y_c\) decays quickly (you want the oscillation to stop fast). But you also need \(y_p\) to be small (you don't want the car to bounce violently over bumps). This is the engineer's balancing act — two competing goals, both captured by our decomposition!
The methods in sections 2.1, 2.2, and 2.3 all assume that \(a\), \(b\), and \(c\) are numbers (constants). This means the system itself is fixed — it doesn't change over time. The characteristic equation, the method of undetermined coefficients, and the Wronskian all depend on this simplicity.
In section 2.4, we'll encounter equations like:
Here, the coefficients depend on the independent variable \(x\). The system itself is changing! This is harder because:
But don't worry — the fundamental ideas (finding \(y_c\) and \(y_p\), superposition) still apply.
"The study of differential equations is the study of the universe itself, for every natural phenomenon can be expressed through the language of change and motion."
— Joseph-Louis Lagrange (1736–1813), pioneer of differential equations and mechanics
To make the most of this chapter, follow these steps in order:
Each section is self-contained but builds on the previous one. Don't skip 2.1 — the characteristic equation is absolutely essential.