Numerical Integration of Blackbody Radiation Function


MPD78
12-17-2009, 09:19 AM
Hello all,

In the field of heat transfer in the mode of radiation there is an integral that can be used find the fraction of the total emission from a blackbody at a certain wave length.

The equation is: (please forgive the notation, my LaTex knowledge is minimal.)

Integrand = (E(lambda of a blackbody) / sigma*T^5 ))

F(0 to lambda) = Integrand d(lambda *T)

The limits of integration are

Lower = 0
Upper = lambda*T

Additional information:

lambda = wavelength (microns)
T = temperature of the blackbody (K)
sigma = Stefan-Boltzman constant = 5.670e-8 (W/m^2-K^4)

E is a function of wavelength and Temperature

Here is the euqation for E:

E(lambda of blackbody, T) = C1 / lambda^5*(exp(C2/lambda*T) - 1)

C1 = 2*PI*h*c^2 = 3.742e8 (W*microns^4 / m^2)
C2 = (h*c/k) = 1.439e4 (microns*K)

From Siegel's book "Thermal Radiation Heat Transfer" the integrand is only a function of the wavelength temperature product.

So basically the integral is set equal to a function f(lambda*T).

A table of the results are given. For example,

f(lambda*T) = 2200 (microns*K) the value of F(0 to lambda) = 0.100888

My question is this;

I am assuming that this integral was solved numerically, but how was that performed? I tried to used qsimp and qtrap but didn't obtain the correct results.

Any help would be greatly appreciated.

See attachment for a much "easier" read of the equations (crudely) described above.

EDIT: I did a quick search of Google books and the newest edition is there with as a limited preview. The solution methodology used to evaluate the function is descibed on pages 21 - 22.

Here is a link to page 21.

http://books.google.com/books?id=O389yQ0-fecC&lpg=PP1&pg=PA21#v=onepage&q=&f=false

END EDIT:

Thanks
Matt

jaje
12-24-2009, 02:08 AM
The integral you need to compute simplifies to

http://i50.tinypic.com/23rai5c.gif

where http://i46.tinypic.com/sdo0w9.gif ; there is an infinite series given in the reference you gave which you can use for computations (or from which you can derive a nice Pade approximant); but if you insist on integration, you might want to look into the double-exponential quadrature routines used in NR.

Jan M.

MPD78
12-24-2009, 08:14 AM
Thanks for your input.

... there is an infinite series given in the reference you gave which you can use for computations ...

Yes, I have been using the series solution. It converges in a short number of iterations.

Thanks
Matt