Re: Natural Convection in a Vertical Column

Russ

I don't know that CFD won't be a valid tool for your case, BUT:

  1. it sounds like there would be a substantial uncertainty due to
the arrangements of cables within the conduits: a tortuous path 
perhaps, change at various points along the length. Not a situation
amenable to CFD approaches if so, since there'd be so many options
each requiring a different model/mesh.

  2.  for most complex closed cavities, it is a tough problem for CFD 
in part because those codes have the largest uncertainty in the heat
transfer coefficients, and in part because you have to start from
various initial conditions because there can be more than one stable
answer, at least analytically.

So CFD might be valid, but still might not be worth the bother.

What is going on at the ends (top and bottom)? That will have a lot of
influence on the flow field. If they're completely open, it is a
chimney effect and you can work with 1D methods, which are quick to
solve and allow you to test uncertainties.  Or perhaps there is a
horizontal section of the conduit at either end?

Any test or empirical data at all? You can always calibrate a model to
that, however sparse, to better trust any predictions of untested
cases.

Regards,

Brent


On 9 Oct 2003 07:13:55 -0700, [EMAIL PROTECTED] (Russ) wrote:

>I am currently analyzing a problem involving a long vertical column
>with a relatively small conduit along the length column containing a
>group of power cables. There is a significant amount of heat
>generation in the cable due to a large current. I'm primarily
>interested in the temperature of the conductors and surrounding
>insulation.
>
>It is fairly straightforward to take a control volume and perform a
>simple conservation of energy calculation to obtain component
>temperatures. This assumes that the temperature of the surrounding air
>is determined exclusively by that of the components.
>
>In reality, since there is heat generation along the entire cable,
>natural convection will carry the heat upward and result in a higher
>air temperature at the top of the column. My problem is to quantify
>what "higher" is. I can do a parametric study to examine the component
>temperatures as the surrounding air temperature increases but I need
>some qualitative means of determining where we're really at along this
>parametric curve. The simplified analysis yields temperatures that are
>close to the specified limit for the insulation, even a modest
>increase in temperature could potentially push the insulation over the
>limit.
>
>I've been considering a CFD analysis of the entire system but I have
>some concerns about the viability of CFD for natural convection
>modeling.  Also, CFD modeling, even for this relatively simple
>geometry may be a massive undertaking. I have done some work in CFD
>for forced flow without any temperature considerations but I'm afraid
>that introducing this additional level of complexity will make CFD
>unreasonable for this application. Any thoughts on that?
>
>I'm looking for any suggestions or reference information that you may
>have. Although CFD is an option I'm considering I'd be interested in
>any potential analytical approach.
>
>Thank you,
>Russ D.

-----------------------------------------------
Brent Cullimore, [EMAIL PROTECTED]
C&R Technologies,  www.crtech.com
Thermal/fluid Software and Consulting

Hot engineering ... Cool software (R)
-----------------------------------------------