Welcome to our community

Be a part of something great, join today!

[SOLVED] Infinite domain to finite plate by a change of variables

dwsmith

Well-known member
Feb 1, 2012
1,673
Consider the following solution to the steady state heat diffusion problem on an infinite y domain.
\[
T(x, y) = \sum_{n = 1}^{\infty}c_n\exp\left(-\frac{\pi n}{\ell} y\right)
\sin\left(\frac{\pi n}{\ell}x\right)
\]
How does one obtain the results of finite plate by making the change of variables \(d - y\) for \(y\) and considering the linit as \(d\to\infty\)?

Making that sub we have \(\exp(-\lambda_nd)\exp(\lambda_ny)\). If we take the limit as d goes to inifinity, we get 0. Therefore, \(T(x,y) = 0\). This doesn't seem correct.
 

dwsmith

Well-known member
Feb 1, 2012
1,673
Consider \(\exp(d-y) = \sinh(d-y) + \cosh(d-y)\).

Then \(\exp\left(\frac{\pi n}{\ell}(d-y)\right) = \sinh\left(\frac{\pi n}{\ell}(d-y)\right) + \cosh\left(\frac{\pi n}{\ell}(d-y)\right)\).
\[
T(x,y) = \sum_{n = 1}^{\infty}c_n\sin\left(\frac{\pi n}{\ell}x\right)\left[\sinh\left(\frac{\pi n}{\ell}(d-y)\right) + \cosh\left(\frac{\pi n}{\ell}(d-y)\right)\right]
\]
General the solution to the Laplace on a rectangle will only have sinh or cosh, but if it has both, they should have their own Fourier coefficients. I don't see away to re-write this any further though. I still don't understand the \(d\to\infty\) part.

If we break \(T\) up, we have
\[
T_1 = T(x,y) = \sum_{n = 1}^{\infty}c_n\sin\left(\frac{\pi n}{\ell}x\right)\sinh\left(\frac{\pi n}{\ell}(d-y)\right)
\]
which implies a boundary condition of \(T(x, d) = f(x)\), and
\[
T_2 = T(x,y) = \sum_{n = 1}^{\infty}c_n\sin\left(\frac{\pi n}{\ell}x\right)\cosh\left(\frac{\pi n}{\ell}(d-y)\right)
\]
which implies a boundary condition of \(T'(x, d) = g(x)\).
Is there more to say about \(T\)?

If \(d\to\infty\), we are back to the infinite domain problem.
 
Last edited:

dwsmith

Well-known member
Feb 1, 2012
1,673
\[
T(x,y) = \sum_{n = 1}^{\infty}c_n\sin\left(\frac{\pi n}{\ell}x\right)\left[\sinh\left(\frac{\pi n}{\ell}(d-y)\right) + \cosh\left(\frac{\pi n}{\ell}(d-y)\right)\right]
\]
Should this have be written automatically as
\[
T(x,y) = \sum_{n = 1}^{\infty}\sin\left(\frac{\pi n}{\ell}x\right)\left[A_n\sinh\left(\frac{\pi n}{\ell}(d-y)\right) + B_n\cosh\left(\frac{\pi n}{\ell}(d-y)\right)\right]
\]
instead?

This would then make more since for the boundary conditions. We would have the first equal to say \(f(x)\) and the second would be the derivative equal to \(g(x)\).
 
Last edited: