3

Here is the problem posted:

problem1

Now here is my solution for a)

problem2

I knew to discard the $e^{ky}$ and $\sin(kx)$ due to $T\to 20$ when $y\to\infty$ and $T= 20$ at $x = 0$.

Which leaves me with $\cos(kx)*e^{-ky} = T$.

Now I have two main problems.

1) I don't know how to solve for $k$.

If I plug in $(0,0)$ I get $cos(0*k) = 20$

2) I don't know if I'm understanding the problem correctly, because if I plug in $x = 0$ to the bottom edge equation, I get $T = 20$ which equals the given left edge temperature.

But if I plug in $x = 30$ I get $T = 320$, not 20, which should be the right edge.

I really need help on this.

dustin
  • 8,241

1 Answers1

2

I suggest you start making $t=T-20$ your new temperature function; recall that the Laplace operator is linear, and if $T$ is a solution so is $t$ and vice versa. The advantage of this new temperature is that it makes 3 out of 4 boundary conditions homogeneous, making the problem easier to handle.

More precisely, if you set $t=T-20$ then $t(x=0)=t(x=30)=t(x,y=\infty)=0$. Now let $t=f(x)g(y)$, separation of variables yields: $$f=-k^2f_{xx} \text{ and } g=k^2g_{yy}.$$

Making $f(x)=A\sin{kx}+B\cos{kx}$ and using the boundary conditions at $x=0$ and $x=30$ we get $B=0$ and $k=\frac{n\pi}{30}, n\in\{1,2,3,\dots\}$. The solutions for $f$ have the form: $$f_n=A_n \sin{\left(\frac{n\pi}{30}x\right)}$$ Proceeding in an analogous way for $g$ we get $g(y)=Ce^{ky}+De^{-ky}$ and from $t\rightarrow 0$ as $y\rightarrow \infty$ we deduce $C=0$, so, taking into account the previously calculated values for $k$: $$g_n=D_n e^{-\frac{n\pi}{30}y}.$$ Now note that each $t_n=f_n g_n$ is a solution, and so is their sum; the solution can be therefore written as:

$$t(x,y)=\sum_{n=1}^\infty A_n \sin{\left(\frac{n\pi}{30}x\right)}\exp{\left(-\frac{n\pi}{30}y\right)}.$$

Where I have renamed the constants. We have only to determine the coefficients $A_n$, we use the condition at $y=0$ for this. From $t(x,y=0)=10x$ we obtain $$10x=\sum_{n=1}^\infty A_n \sin{\left(\frac{n\pi}{30}x\right)}.$$

Multiplying by $\sin{\left( \frac{m\pi}{30}x\right)}$, changing variables, integrating and using the orthogonality relations between sines we finally arrive at $$A_n=10\frac{2}{\pi}\frac{\pi}{30}\int_0^{30}x \sin{\left( \frac{n\pi}{30}x \right)} \operatorname d\!x .$$

This is the Fourier series of your lower boundary conditions (I suggest you check this calculation as I might have made mistakes).

Once you calculate the coefficients you can go back to $T$ by summing the baseline 20 to $t$, the constant term of the Fourier series for the solution..

alonso s
  • 1,151
  • Ok doing this and going through the process, I end up with 10x = sin(kx). When I plug in x = 0 I end up with 0 = sin(0*x) which is a trivial solution, can I plug in x = 1/10 and have 1 = sin(kx)? – SwiftySwift Nov 22 '14 at 22:07
  • @Wimpy35 I edited my original answer, please check it. – alonso s Nov 22 '14 at 23:46
  • Would this be the correct solution? I'm a bit confused by your last statement. solution – SwiftySwift Nov 23 '14 at 20:31
  • Also, I tried using Mathematica to visualize the problem also, does this make sense? mathematica – SwiftySwift Nov 23 '14 at 20:48
  • Looks reasonable, what is the kink on the lower right corner? – alonso s Nov 23 '14 at 21:41
  • I actually have no idea, that is what I am trying to understand. It is correct for me to change from the t solution to the T solution simply by adding 20 at the end, correct? – SwiftySwift Nov 23 '14 at 22:05
  • Looks reasonable enough everywhere but the lower right corner but that's to be expected; these problems are always somewhat artificial on the boundary conditions, you got the right side at 20 degrees, but the rightmost end of the lower border nears 320. On a side note, I'm just curious if the general procedure was clear; you had the correct idea but were forgetting superposition. More often than not you won't be able to fit your solution to a functional boundary condition unless you use your solution functions as a basis. – alonso s Nov 23 '14 at 22:41
  • I think I understand everything up until the point where you go back from the solution for t to the solution for T.

    could you explain that a bit more, and if I did it correctly?

    – SwiftySwift Nov 23 '14 at 23:50
  • 1
    It is exactly what you did: $T=t+20$. Note that $T$ satisfies all boundary conditions. – alonso s Nov 24 '14 at 01:19