ODE: Heat Conduction in a Rod

Understanding how heat diffuses through a solid object is a cornerstone of engineering design, from managing thermal stress in bridges to cooling electronic components. The one-dimensional heat equation provides a powerful mathematical model for this phenomenon, allowing you to predict temperature changes over time and space. Mastering its solution equips you to analyze not just thermal systems, but a wide array of diffusion-based processes, making it an indispensable tool in your engineering toolkit.

The Governing Equation and Boundary Conditions

The foundational model for one-dimensional heat conduction is the heat equation, a partial differential equation (PDE) derived from energy conservation and Fourier's law of heat conduction. For a homogeneous rod with constant thermal properties, it is expressed as:

$$\frac{\partial u}{\partial t}=\alpha \frac{{\partial }^{2}u}{\partial x^2}$$

Here, $u(x,t)$ represents the temperature at position $x$ and time $t$, and $\alpha$ is the thermal diffusivity, a material property governing how quickly heat spreads. To solve this PDE uniquely, we must specify both initial and boundary conditions.

Boundary conditions define the thermal interaction at the rod's ends. Two primary types are crucial:

1. Dirichlet (Fixed-Temperature) Condition: The temperature at an endpoint is held constant. For example, $u(0,t)=T_0$.
2. Neumann (Insulated) Condition: An insulated endpoint implies no heat flow, which translates to a zero spatial derivative (zero gradient). This is written as $\frac{\partial u}{\partial x}(0,t)=0$.

A typical mixed-boundary problem involves one end held at a fixed temperature and the other insulated. The initial condition, $u(x,0)=f(x)$, defines the temperature profile along the rod at time zero.

The Method of Separation of Variables

Separation of variables is the standard technique for solving linear PDEs like the heat equation with homogeneous boundary conditions. The core assumption is that the solution can be written as a product of a function of space alone and a function of time alone: $u(x,t)=X(x)T(t)$. Substituting this into the heat equation and rearranging leads to:

$$\frac{1}{\alpha }\frac{T^{\prime }(t)}{T(t)}=\frac{X^{\prime \prime }(x)}{X(x)}$$

Since the left side depends only on $t$ and the right side only on $x$, both sides must equal a constant, which we denote as $-\lambda$. This separates the single PDE into two ordinary differential equations (ODEs):

1. A spatial ODE: $X^{\prime \prime }(x)+\lambda X(x)=0$
2. A temporal ODE: $T^{\prime }(t)+\alpha \lambda T(t)=0$

The spatial ODE, along with the homogeneous boundary conditions (e.g., $u(0,t)=0$ and $\frac{\partial u}{\partial x}(L,t)=0$), forms a Sturm-Liouville eigenvalue problem. Only specific values of $\lambda$, called eigenvalues, yield non-trivial solutions for $X(x)$, known as eigenfunctions.

Steady-State Temperature Distribution

The steady-state temperature distribution, $u_s(x)$, is the solution that remains after all transient effects have decayed, meaning $\frac{\partial u}{\partial t}=0$. The heat equation then reduces to $u_s^{\prime \prime }(x)=0$. Solving this simple ODE yields a linear function, $u_s(x)=Ax+B$.

The constants $A$ and $B$ are determined solely by the non-homogeneous boundary conditions. For instance, if the boundary at $x=0$ is fixed at $T_0$ and the boundary at $x=L$ is insulated, the conditions are $u_s(0)=T_0$ and $u_s^{\prime }(L)=0$. Applying these gives $A=0$ and $B=T_0$, resulting in a uniform steady-state temperature equal to $T_0$ everywhere. The steady-state solution is independent of the initial condition and the thermal diffusivity $\alpha$.

Transient Temperature Profiles and the Full Solution

The transient temperature profiles represent the time-dependent part of the solution that bridges the initial condition to the steady state. The temporal ODE, $T^{\prime }(t)+\alpha \lambda T(t)=0$, has a general solution of the form $T_n(t)=C_n{e}^{-\alpha {\lambda }_{n}t}$, where ${\lambda }_n$ is the n-th eigenvalue.

For each eigenvalue-eigenfunction pair $({\lambda }_n,X_n(x))$, we have a product solution $X_n(x)T_n(t)$. The general solution to the original PDE is the infinite sum (or series) of all such product solutions:

$$u(x,t)=u_s(x)+\sum _{n=1}^{\infty }{C}_n{X}_n(x)e^{-\alpha {\lambda }_{n}t}$$

The coefficients $C_n$ are determined by the initial condition. At time $t=0$, the equation becomes $f(x)-u_s(x)={\sum }_{n=1}^{\infty }{C}_n{X}_n(x)$. Because the eigenfunctions $X_n(x)$ are orthogonal, we can use the inner product to "project" the initial temperature distribution onto each eigenfunction, calculating each $C_n$ via integration.

Physical Interpretation of Eigenfunction Decay Rates

The physical interpretation of eigenfunction decay rates is directly tied to the exponential term $e^{-\alpha {\lambda }_{n}t}$. Each mode (or term) in the infinite series decays at a rate proportional to $\alpha {\lambda }_n$. Higher eigenvalues ${\lambda }_n$ correspond to eigenfunctions with more spatial oscillations (more "bumps"), and these higher-frequency modes decay much faster.

For example, the first harmonic (n=1) decays as $e^{-\alpha {\lambda }_{1}t}$, while the second decays as $e^{-\alpha {\lambda }_{2}t}$, where ${\lambda }_2>{\lambda }_1$. Since ${\lambda }_2$ is larger, the exponent is more negative, causing the second mode to vanish more quickly. This explains why complex initial temperature profiles smooth out rapidly—the fine details (high-frequency modes) disappear first, leaving the dominant, slowly decaying mode to govern the approach to steady state. The thermal diffusivity $\alpha$ acts as a scaling factor; a higher $\alpha$ (e.g., in metals like copper) accelerates the entire decay process.

Common Pitfalls

1. Misapplying Boundary Conditions to the Steady-State Solution: A frequent error is forcing the steady-state solution $u_s(x)$ to satisfy all boundary conditions, including the homogeneous ones. The steady-state solution must satisfy only the non-homogeneous boundary conditions from the original problem. The homogeneous conditions are absorbed by the transient part of the solution ($X_n(x)$).

1. Confusing the Role of the Initial Condition: The initial condition $f(x)$ is used to find the series coefficients $C_n$ for the full solution $u(x,t)$. It is a mistake to apply it directly to the steady-state solution $u_s(x)$ or to the spatial eigenfunctions $X_n(x)$ in isolation. You apply it to the complete series at $t=0$: $u(x,0)=u_s(x)+\sum C_n{X}_n(x)=f(x)$.

1. Incorrectly Solving the Eigenvalue Problem: When deriving eigenvalues from boundary conditions like $X(0)=0$ and $X^{\prime }(L)=0$, it's easy to make algebraic or trigonometric errors. Always revisit the general solution to the spatial ODE ($X(x)=A\cos (\sqrt{\lambda }x)+B\sin (\sqrt{\lambda }x)$) and apply each boundary condition meticulously to derive the characteristic equation that defines ${\lambda }_n$.

1. Overlooking the Orthogonality Condition for Coefficients: The formula for $C_n$ involves an integral that leverages the orthogonality of the eigenfunctions. Forgetting the corresponding norm in the denominator (${\int }_0^L{X}_n^2(x)dx$) is a common calculation error that yields incorrect coefficient values.

Summary

• The one-dimensional heat equation, $u_t=\alpha u_{xx}$, models thermal diffusion. Its solution requires an initial temperature profile and boundary conditions, commonly fixed-temperature (Dirichlet) or insulated (Neumann).
• The separation of variables method reduces the PDE to two ODEs: a spatial eigenvalue problem that determines the eigenfunctions and eigenvalues (${\lambda }_n$), and a temporal ODE whose solution decays exponentially.
• The complete solution is $u(x,t)=u_s(x)+{\sum }_{n=1}^{\infty }{C}_n{X}_n(x)e^{-\alpha {\lambda }_{n}t}$, where the steady-state solution $u_s(x)$ is a linear function found from non-homogeneous boundary conditions.
• The transient solution is an infinite series where each term decays at a rate proportional to $\alpha {\lambda }_n$. Higher-frequency modes (larger ${\lambda }_n$) decay faster, explaining the rapid smoothing of temperature profiles.
• The coefficients $C_n$ are computed by projecting the initial condition (minus the steady state) onto the orthogonal set of spatial eigenfunctions $X_n(x)$.