40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
 COMSOL elog Not logged in
 Sun Jun 22 23:44:56 2014, Sam Moore, Optics, , Going through Heinert's 'TR noise of cylindrical test masses' paper Mon Jun 23 11:40:17 2014, Sam Moore, Optics, , Going through Heinert's 'TR noise of cylindrical test masses' paper Mon Jun 23 16:10:17 2014, Matt A., Optics, , Going through Heinert's 'TR noise of cylindrical test masses' paper
Message ID: 84     Entry time: Mon Jun 23 11:40:17 2014     In reply to: 83
 Author: Sam Moore Type: Optics Category: Subject: Going through Heinert's 'TR noise of cylindrical test masses' paper

 Quote: At this point, my are goals are to 1) convert the time-dependent heat equation into stationary, complex form, 2) use the Levin approach to calculate the TR noise given this stationary PDE, and 3) verify the results in COMSOL  I have looked at the Heinert paper and converted the time-dependent partial differential Heat equation to a stationary, complex one.  I did this by assuming a temperature distribution of the form T(\vec{r},t) = T_0 ( \vec{r} ) e^{i \omega t }, where  \omega is the frequency of oscillation of the input heat field: q( \vec{r} , t) = q( \vec{r} ) e^{i \omega t}.  Hence, I am assuming the steady-state solution of the temperature field.  From this ansatz, the PDE becomes C_p T(\vec{r}) + \frac{ i \kappa}{ \omega }  \nabla^2 T ( \vec{r} ) = q ( \vec{ r} ).   This complex conversion already seems to have been done in the paper, since the stationary solution of the form T(r, z) = \sum_{n = 0}^{\infty} \sum_{m = 0}^{ \infty} T_{n m} J_0 (k_n r) \cos ( \ell_m z) is assumed. k_n and \ell_m are obtained from the adiabatic boundary conditions.  Moreover, if I plug this form for T(r, z) into the stationary PDE, I get Heinert's eq. 12. ( although I'm not quite sure how to take the second derivative of a zeroth-order bessel function).   I've started to try out the cylindrical geometry in COMSOL, using the structural mechanics and heat transfer modules.  At this point, I'm not quite sure how to specify the PDE that I'm interested in, nor am I sure about how to specify the boundary conditions.

I carried out the calculations for Heinert's simple model in greater detail and was able to obtain equation 12 (with the exception of a plus sign instead of a minus sign in the denominator; I guess the Fourier transform method gives a minus sign.  In the end, that doesn't matter, since the modulus square yields the same result in the denominator.)  I obtained the same spectral density S_z (\omega), with the exception of a missing factor of R^2.  I don't know where this R^2 went, or why it is that the spectral density is associated with the extrinsic variable z (instead of r, say).  It's supposed to be a "homogeneous [noise] readout along the z direction."

ELOG V3.1.3-