I am trying an analytical solution for a 3D Laplace equaition, and just wanted to know if I could use the symbolic calculation capabilities of Mathematica to calculate the Fourier coefficients

Context (Please bear with the length)

On the last step of solving a three-dimensional Laplace equation,($\nabla^2T=0$, where $\nabla^2 = \frac{\partial^2}{\partial x^2} + \frac{\partial^2}{\partial y^2} + \frac{\partial^2}{\partial z^2}$) with BC(s) as

$T(0,y,z) = T(L,y,z) = T_a$,

$T(x,0,z) = T(x,l,z) = T_a$,

$\frac{\partial T(x,y,0)}{\partial z} = p_c\bigg(e^{-b_cy/l}\left[T_{ci} + \frac{b_c}{l}\int_0^y e^{b_cs/l}T(x,s,z)ds\right] - T(x,y,0)\bigg)$

$\frac{\partial T(x,y,w)}{\partial z} = p_h\bigg(e^{-b_hx/L}\left[T_{hi} + \frac{b_h}{l}\int_0^x e^{b_hs/l}T(s,y,z)ds\right] - T(x,y,w)\bigg)$

On using a solution form

$T(x,y,z)=\sum_{m,n=1}^{\infty}(A_{nm}e^{\gamma z} + B_{nm}e^{-\gamma z})\sin(\frac{n\pi x}{L})\sin(\frac{m\pi y}{l})+T_a.$

I need to evaluate the Fourier Coefficients $A_{nm}$ and $B_{nm}$ using the following equation which was obtained after applying the $z=0$ BC

$$ \frac{1}{p_c}\sum_{n,m=1}^\infty\sin(\frac{n\pi x}{L})\sin(\frac{m\pi y}{l})\gamma(A_{nm}-B_{nm}) = \delta e^{\frac{-b_c y}{l}} + P-Q+R-S \rightarrow (\mathrm{1}) $$

$P =\sum_{n,m=1}^\infty\frac{b_c^2}{b_c^2+(m\pi)^2}(A_{nm}+B_{nm})\sin(\frac{n\pi x}{L})\sin(\frac{m\pi y}{l})$

$Q =\sum_{n,m=1}^\infty \frac{b_c m\pi}{b_c^2+(m\pi)^2}(A_{nm}+B_{nm})\sin(\frac{n\pi x}{L})\cos(\frac{m\pi y}{l})$

$R = \sum_{n,m=1}^\infty \frac{b_c m\pi}{b_c^2+(m\pi)^2}(A_{nm}+B_{nm})e^{\frac{-b_c y}{l}}$

$S = \sum_{n,m=1}^\infty(A_{nm}+B_{nm})\sin(\frac{n\pi x}{L})\sin(\frac{m\pi y}{l})$

Here ,$\gamma^2 = (\frac{n\pi}{L})^2 + (\frac{m\pi}{l})^2$ and $c,\delta=(T_{ci}-T_a)$ are constants.

Obviously, there would be another relation that would be a linear equation in $A$ and $B$ to finally deduce the values of $A_{nm}$ and $B_{nm}$ (from the $z=w$ BC) separately.My question pertains to the fact as to how should I handle the $e^{\frac{-b_c y}{l}}$ here.


I can multiply both sides of the equation $\mathrm{(1)}$ with $\int_0^L\sin(\frac{k\pi x}{L})$ and $\int_0^l\sin(\frac{j\pi y}{l})$ and then use the principle of orthogonality to finally arrive at

For some $n=k$ and $m=j$, $\mathrm{(1)}$ becomes

$$ \frac{1}{p_c}\frac{L}{2}\frac{l}{2}\gamma(A_{kj}-B_{kj})=\delta e^{\frac{-b_c y}{l}}\int_0^L\sin(\frac{k\pi x}{L})\mathrm{d}x\int_0^l\sin(\frac{j\pi y}{l})\mathrm{d}y+\underbrace{\frac{b_c^2}{b_c^2+(j\pi)^2}(A_{kj}+B_{kj})\frac{L}{2}\frac{l}{2}}_{P} - \overbrace{0}^{Q} + \underbrace{\sum_{n,m=1}^\infty \frac{b_c m\pi}{b_c^2+(m\pi)^2}(A_{nm}+B_{nm})e^{\frac{-b_c y}{l}}\int_0^L\sin(\frac{k\pi x}{L})\mathrm{d}x\int_0^l\sin(\frac{j\pi y}{l})\mathrm{d}y}_{R} - \overbrace{(A_{kj}+B_{kj})\frac{L}{2}\frac{l}{2}}^{S} $$

Fourier expansion of $e^{\frac{-b_c y}{l}}$ in the interval $y \in [0,l]$

I expanded $e^{\frac{-b_c y}{l}}$

$$e^{\frac{-b_c y}{l}} = \frac{(1-e^{-b_c})}{b_c} + \sum_{r=1}^{\infty}\bigg[\frac{2b_c(1-e^{-b_c})}{(b_c)^2 + (2r\pi)^2}\cos\bigg(\frac{2r\pi y}{l}\bigg) + \frac{4r\pi(1-e^{-b_c})}{(b_c)^2 + (2r\pi)^2}\sin\bigg(\frac{2r\pi y}{l}\bigg)\bigg]$$

I guess, I need to substitute this in $(1)$ for $e^{\frac{-b_c y}{l}}$

Any suggestions or comments would be immensley helpful and really appreciated.

If you look at $(1)$, you can see how if the expansion of $e^{\frac{-b_c y}{l}}$ is substituted the $R$ term would go to a triple summation which makes it way complicated. Can there be some alternative way to handle these calculations in Mathematica. I do know that the PDE solvers could give a result using discretization schemes. I was hoping if I could use Mathematica to help me in the final step of finding these Fourier coefficients.


Your Answer

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

Browse other questions tagged or ask your own question.