1. Limited time only! Sign up for a free 30min personal tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Homework Help: Derivation of Heat Conduction in Spherical Co-Ordinates

  1. Feb 20, 2012 #1
    I have two questions. I believe I have solved the first question and would like confirmation of this answer; the second question I'm a little bit lost on so any help there would be greatly appreciated!

    I am working on a problem set in which I must derive the equation for heat conduction in spherical co-ordinates. I have completed part a, in which I derived the heat flux equation:

    [itex]\frac{dq_{r}}{dr} + \frac{2}{r}q_{r} - \rho H = 0[/itex]

    I have used Fourier's Law (to rewrite this equation for teperature), which states

    [itex]\textbf{q}[/itex] = -k[itex]\nabla T[/itex], where [itex]\textbf{q}[/itex] is the heat flux and it is a vector (I couldn't find a vector symbol, so it is simply bolded) and [itex]\nabla [/itex] is the gradient.

    Using the gradiant for speherical co-ordinates, and considering only changes in the radial direction (so that [itex] \frac{\delta T}{\delta \phi}[/itex] and [itex]\frac{\delta T}{\delta \theta}[/itex] are 0) we can write:

    [itex]\nabla T = \frac{dT}{dr} \widehat{r}[/itex]

    So we can write:

    [itex]\textbf{q} = q_{r} = -k \nabla T = -k \frac{dT}{dr} \widehat{r}[/itex]

    [itex]\frac{dq_{r}}{dr} + \frac{2}{r}q_{r} - \rho H = 0[/itex]

    [itex]\frac{d}{dr} ( -k \frac{dT}{dr}) - \frac{2k}{r} \frac{dT}{dr} - \rho H = 0[/itex]

    Assuming k is a constant:

    [itex]-k \frac{d^{2}T}{dr^{2}} - \frac{2k}{r} \frac{dT}{dr} - \rho H = 0[/itex]

    [itex] (\frac{d^{2}T}{dr^{2}} + \frac{2}{r} \frac{dT}{dr}) + \frac{\rho H}{k} = 0[/itex]

    So I have written this as an equation for temperature instead of heat flux. Is this correct?


    Next I am asked to solve the above equation for T(r) subject to the boundary conditions T(R) = [itex]T_{s}[/itex] (where R is the radius of the planet) and T(0) must be finite.

    [itex] (\frac{d^{2}T}{dr^{2}} + \frac{2}{r} \frac{dT}{dr}) + \frac{\rho H}{k} = 0[/itex]

    Using a hint that is provided I can write:

    [itex] \frac{1}{r^{2}} \frac{d}{dr} (r^{2} \frac{dT}{dr}) = - \frac{\rho H}{k} [/itex]

    Now, I'm not too sure how to simplify this. My E&M textbook says that the radial component of the Laplacian in spherical co-ordinates is equal to what I have on the left side of that equation so I could write (because we are assuming, as above that [itex] \frac{\delta T}{\delta \phi}[/itex] and [itex]\frac{\delta T}{\delta \theta}[/itex] are 0):

    [itex] \nabla^{2}T + \frac{\rho H}{k} = 0 [/itex]

    What should I do next? Is there any way for me to simplify this further? Should I have not written it in terms of the Laplacian? How can I solve for an expression T(r)?

    Thanks in advance! I appreciate any insight you can give me as to what the next step might be! :)
  2. jcsd
  3. Feb 21, 2012 #2
    So I've been thinking about this a bit more and I know that this expression must be correct because it agrees with the expression derived in class using Cartesian co-ordinates.

    However, I need an expression for T in terms of r and this is where I get stuck.

    [itex] \nabla^{2}T + \frac{\rho H}{k} = 0 [/itex]

    [itex] \frac{1}{r^{2}} \frac{d}{dr} (r^{2} \frac{dT}{dr}) = - \frac{\rho H}{k} [/itex]

    [itex] \frac{d}{dr} (r^{2} \frac{dT}{dr}) = - \frac{\rho Hr^{2}}{k} [/itex]

    We can then integrate both sides with respect to r:

    [itex] r^{2} \frac{dT}{dr} = - \frac{\rho Hr^{3}}{3k} + c_{1}[/itex]

    [itex] \frac{dT}{dr} = - \frac{\rho Hr}{3k} + c_{1}[/itex]

    Integrating again we can write:

    [itex] T = - \frac{\rho Hr^{2}}{6k} + c_{1}r + c_{2}[/itex]

    Is this correct? I'm concerned that I did something incorrect while integrating the first time; I feel like the 6 shouldn't be in the denominator. In Cartesian co-ordinates we have a 2 in the denominator. Where am I going wrong?

    I believe there must be a 2 and not a 6 because if r is solely in the y direction (x and z = 0) the equation should reduce to what we found in the Cartesian case.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook