Let $u(x,t)$ be the temperature along a 1-D rod, from $x=0$ to $x=L$.

$\frac{\partial u}{\partial t} = A \frac{\partial^2 u}{\partial x^2}+Bu$, where A and B are constants.

Initial condition is $u(x,0)=0$.

Boundary conditions are:

$\frac{\partial u}{\partial x}|_{x=0}=f(t)$ (arbitrarily prescribed heat influx on the left side of the rod)

$\frac{\partial u}{\partial x}|_{x=L}=0$ (insulated on the other end).

I failed to solve this problem using separation of variables, and would like to know: is there any other analytic methods I can use before I try numerical solutions?


  • $\begingroup$ You can use separation of variables. Try it one last time. $\endgroup$ – Chinny84 Sep 24 '14 at 19:28

Sure, you should take try to take the inverse Laplace transform of $ f(t) $. Here is a sketch. for simplicity, I'll flip the conditions so that $ \partial_x u|_{x = 0} = 0 $ and $ \partial_x u_{x = L} = f$.

Let $ \psi_k(x,t) = \frac{-1}{k \sin(k)} \cos(k x) e^{-\lambda_k t} $ where $ \lambda = A k^2 + B $ (mod some negative signs). Then you see that $ \psi_k(x,t) $ solves the heat equation with boundary conditions $ \partial_x \psi_k|_{x=0} = 0 $ and $ \partial_x \psi_k|_{x=L} = e^{-\lambda_k t} $. So, if you can find $ g $ so that $ f(t) = \int_{0}^{\infty} g(k) \; e^{-\lambda_k t} \; dk $ then you're done: $ u = \int_{0}^{\infty} g(k) \; \psi_k(x,t) \; dk $. The way to do that is to take inverse laplace transform, which I'll let you look up.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.