Re: [ATM] Thermal mirror deformation

["Arjan te Marvelde" <arjan.te.marvelde@hetnet.nl>]

This is a bit longish reply to some issues that were raised in the
discussion about my PDE model:

Tony Gondola:
"The paper talks about much more then just cooling a mirror faster with
fans.
  I have no argument with that. The main point of the paper is how the
mirror
  deforms when fan cooled on one side. When you start working at that level
of
  detail, just how and where the air moves over the full surface of the
mirror
  becomes important....."

AtM: This is not entirely true, although in the model mentioned in the paper
there is a fan on the back only, the point it tries to make is that when
radiation is taken into account temperature equilibrium is never reached.
The really BIG surprise (at least to me) was that the resulting heatflow
causes a considerable temperature gradient.It is this temperature diference
between front and back that corresponds with Wolfgang Rohrs' article.

Tony Gondola:
"The thing that bothers me is when ATMs take a vary preliminary result and
  run with it without really understanding the limits of the modeling. I've
  already seen glimmers of that in some of the responses to the paper in
other
  groups."

AtM: Which groups? I think that the paper presents an invitation to explore
rather than a set of facts.

Tony Gondola:
"I'm sure it is based on the limits of the model."

AtM: Of course it is, but the specific mirror that reaches 0.2K as
temperature difference happens to be about the same as the one in Wolfgangs'
example.

Tony Gondola:
"I can believe the basic cooling times but I'm not sure about the rest of
the
  conclusions. The biggest problems I see are:
  Assuming that the front surface is only cooled by radiation and natural
  convection."

AtM: This is only one case, representing many telescope implementations.
Fans on both sides and no fans are the other cases mentioned.

Tony Gondola:
"Absense of side effects"

AtM: That's right. The side effect is simply modeled as the mirror edge
having a different cooling efficiency (I have set it to a value between
front and back). I have tried with a raised cosine profile for blower
efficiency, which makes things worse at best. Other things to try are
non-flat profile for radiation loss, and to integrate the actual deformation
in the model. This is what I would like people to do: play with the model
and see which parameters have effect and which have not. What would you
propose to try out?

Mike Lockwood:
"As I read it (assuming I read it right), a model was proposed that
  indicated that cooling a mirror on one side would cause the mirror to
  bend."

AtM: Yes when cooling is read as radiative cooling.

"While true for the chosen "ideal" temperature distribution in
  the disk, that distribution, though convenient mathematically, is
  probably not realistic, because it does not take into account the
  cooling effect of the edge of the mirror."

AtM: While the graphs show the temperatures on the optical axis only, the
reached equilibrium also shows a radial gradient due to the simple edge
modeling menytioned above. This is however much smaller in most cases and
will in reality be worse. A lot will depend on the actual construction of
the telescope (as mentioned in my paper) but this is precisely what we try
to analyze!

Mike Lockwood:
"The curve of the mirror caused by the "ideal" temperature distribution
  is spherical, and could easily be negated by simply refocusing.  The
  more complex shape of the mirror caused by a more realistic
  temperature distribution (including edge cooling) is NOT spherical in
  nature, and likely cannot be negated by refocusing.  Therefore, this
  is a problem optically."

AtM: Yes, somebody on the list mentioned an Italian analysis of a blower
cooling only the center of the mirror back, resulting in horrible
non-spherical deformation. Conclusion could be to try to restrict residual
temperture gradients to axial direction as much as possible? If you don't
use fans it might be worthwhile optimizing the scope construction for this,
e.g.by taking care that the amount of sky seen by any point on the mirror is
the same (so either put the mirror on the bottom of a pipe or with a 180deg
view on the air).

In any case, thank for all the comments. Some of the issues I was already
planning to assess further, when time permits.
Note that the paper was started because someone on a Dutch mailinglist could
not get good startest results, while bench testing was OK. He suggested
permanent deformation to compensate for the radiative cooling!
Few people could believe that the radiation effect could possibly be so
large that it would become visible. The quick FlexPDE model I whipped up
suggests it can be...

Arjan te Marvelde
www.udjat.nl








_______________________________________________
ATM mailing list http://www.atmlist.net/