Jump to content


Photo

My method for the control of the boundary layer

This topic has been archived. This means that you cannot reply to this topic.
21 replies to this topic

#1 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 30 August 2009 - 12:28 PM

Nothing groundbreaking new. Here it is.
http://autocostrutto...ng-mirrors.html

#2 Luigi

Luigi

    Fly Me to the Moon

  • *****
  • Posts: 5320
  • Joined: 03 Jul 2007

Posted 30 August 2009 - 12:49 PM

Nice and neat, though I suspect the smooth laminar flow above the mirror would not be as effective as turbulent air directed at and across the surface as is more commonly done.

#3 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 31 August 2009 - 03:21 AM

Effective for what?

For cooling the mirror sure. Turbulent flow causes greater heat exchange rate. However greater heat flux combined with turbulence is bad for seeing... but the situation improves quickly as the mirror cools... but a quickly cooled mirror may deform ( http://visualsky.blo...o-specchio.html http://visualsky.blo...o-specchio.html http://visualsky.blo...re-piccolo.html ).

However the goal should be to control the boundary layer and keep it as uniform as possible (thus laminar) so that no wavefront error is generated. At least this is the aim of my solution.

#4 Luigi

Luigi

    Fly Me to the Moon

  • *****
  • Posts: 5320
  • Joined: 03 Jul 2007

Posted 31 August 2009 - 06:48 AM

>>>Effective for what?<<<

I meant effective for removing the boundary layer. But yes, also more effective for cooling the substrate.

#5 Mike I. Jones

Mike I. Jones

    Gemini

  • *****
  • Posts: 3259
  • Joined: 02 Jul 2006

Posted 31 August 2009 - 08:19 AM

Your concept appears to employ flow entrainment to draw the BL radially away from the mirror. The higher your mass flow rate, the deeper the entrainment effect reaches out onto the mirror aperture. But there is a turbulent aero-optical lensing effect around the lip of your diaphragm you may need to monitor. The flow may be creating a weak unsteady toroidal aero-optic lens around the mirror edge.

Optimization variables would be the mass flow rate, the vertical gap height between the diaphragm and mirror edge, and the shape contour of the diaphragm (like a radius rather than a sharp edge to slow down and distribute flow acceleration).

Which CFD program are you using? Your plot looks like RANS analysis rather than averaged LES. Do a plot of density, and set the plotting contours to + and - 1% of free stream. The index of refraction of air depends only on wavelength and density per the Gladstone-Dale formula:

n(density,wavelength) = 1 + kGD(wavelength) x density

where kGD(wavelength) = 0.113 + 0.00092/wavelength^2

Wavelength is in microns and density is in slugs/ft3. Convert to MKS as needed.

The formula works quite well, as we have repeatedly tested its validity in wind tunnel interferometry. If you see significant density color contouring at 1-2% either side of free stream, you're going to have aero-optical effects strong enough to bother astronomical seeing. And these are edge effects, with the highest influence on the imagery.

If you can, calculate OPD through the 3D density field. At these low density gradients you don't have to refract; just interpolate the density along ray segments between interpolated mesh points, and multiply the segment lengths by the local air index given by Gladstone-Dale to give the OPD of the individual segments. Then sum them up, repeat for a ray grid over a pupil, then subtract the length of the central ray, giving the OPD map over the pupil.

Cool to see CFD being used for this application.
Mike

#6 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4140
  • Joined: 03 Jun 2005

Posted 31 August 2009 - 02:07 PM

I'll just pass on what Mike said (!) but ask this question - what mechanism is operating to adjust the flow closer to the center of the mirror? Although I was planning to include a boundary layer turbulent scrubbing system into my new dob design, I could also easily implement this scrubbing method, as it has a structural ring element with (in the current version) 1/4" clearance all around the mirror edge. Actually, I could do both and test the results. :question:

Whatever the mechanism, I'm intrigued by your results.

Best,
Mark

#7 Mason Dixon

Mason Dixon

    Vostok 1

  • -----
  • Posts: 164
  • Joined: 29 Jun 2007

Posted 31 August 2009 - 11:53 PM

Nice work, is it possible to take a video through the eyepiece showing you turning the system on and off?

#8 Bryan Greer

Bryan Greer

    Vendor

  • -----
  • Vendors
  • Posts: 187
  • Joined: 14 Jun 2003

Posted 01 September 2009 - 11:03 AM

Hi Mauro,

For what it's worth, I like this, and is similar to what I've done on a smaller Newtonian (6"). It works well, and I've come to believe this is the best strategy for smaller scopes in closed tubes.

I have not personally implemented it for a larger telescope like yours, so I'm interested to see how well it works. You can directly observe the boundary layer yourself at the eyepiece with the modified star test I described in the May 2004 S&T, so you can conclusively determine if it helps by watching the boundary layer shadows with the fan on and off.

Sincerely,
Bryan Greer

#9 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 01 September 2009 - 04:40 PM

Hi (all),

I use COMSOL http://www.comsol.com/. The CFD analysis was done just to check the flow. It was done on a slightly different geometry (a lot earlier that I implemented it). I checked the turbulence intensity near the edge(s) (I used the k-epsilon turbulence model). Of course it is possible to refine the analysis and find, for example, the OPD. Thanks for the suggestion.
It is possible to better guide the flow by using profiles on mirror edge and diaphragm but my feeling is that probably a torus with inner radius equal to the diaphragm could be enough. As for what concerns the flow rate, I suspect the more the better (apart for vibrations an power in the field) unless the flow becomes turbulent. The distance of the diaphragm should probably be rather large so, that the annulus area is similar to the mirror area (and flow speed remains rather constant in modulus, only change direction). This arrangement should allow to extend better towards the mirror centre the effect (but the centre is not so critical).

I have three fans and I can see the turbulent cells. I can see the effect on the turbulent cells. When the mirror is hot they are not killed completely. If I turn the the system off I see that the cells grow and become ugly (an ugly spider web with strong spikes). Turning the system on they are reduced but not disappear completely (not with the two smaller fans, I have yet to try the 220 V 40 W one).

However, the flow gradually cools the mirror and eventually they disappear. So the system allows to use the scope during the cooling period with good performance an when the mirror temperature is close enough to the air the system achieve complete effectiveness. Note that the mirror does not need to be exactly at air temperature. I have yet to find the tolerance though. I have the feeling that thick mirrors can now chase the air temperature close enough to maintain good views during the whole night.

I will make a few experiment with different fans and position of the diaphragm and then perhaps refine the CFD analysis.

I do not think this method is perfect but I have yet to see one which is satisfactory. I used aggressive cooling methods too, but pyrex mirrors deform: I could see spherical aberration to appear, and even astigmatism for flows that have a preferential direction across the mirror.

The diaphragm in front of the mirror is a key element (without it the system is a lot less effective).

#10 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4140
  • Joined: 03 Jun 2005

Posted 01 September 2009 - 05:34 PM

For laminar flow the velocity (flow rate) shouldn't be that big of issue. Simply turning it over every 30s or so ought to do the job - my intuition here is that the boundary layer stretches out from the edges and thins out, resulting in the improved performance. That boundary layer is being fed by the temp differential of the mirror to the air, so it's no surprise it doesn't go away until the mirror cools.

Interesting!

Best,
Mark

#11 Bryan Greer

Bryan Greer

    Vendor

  • -----
  • Vendors
  • Posts: 187
  • Joined: 14 Jun 2003

Posted 01 September 2009 - 07:53 PM

Note that the mirror does not need to be exactly at air temperature. I have yet to find the tolerance though.


Based on rainbow schlieren testing I did about ten years ago on a 6" Pyrex blank, if the mirror temperature was within about 2 degrees Celsius of ambient, I rarely measured a p-v wavefront error that exceeded 1/10th wave. It was also hard to even perturb the weak boundary layer into any form other than laminar. To me, this is good enough.

I have the feeling that thick mirrors can now chase the air temperature close enough to maintain good views during the whole night.


For typical after-sunset temperature gradients around here (Ohio, USA), mirrors up to roughly 1-3/4" thick can track the falling temperature close enough with only a rear fan. Thicker mirrors need airflow over both surfaces. When the mirrors get extremely thick, you will never achieve a low enough delta T even with forced convection on all surfaces. In this case, you just rely on scrambling the front boundary layer as much as possible. The effect on the wavefront is to preserve high spatial frequencies (fine detail) at the expense of low spatial frequencies (veiling glare). This is a good trade off when doing high-resolution imaging, as veiling glare can easily be processed out.

Note that my results above won't apply if you live in a region that experiences different temperature gradients. The late Jeff Medkeff reported his 10" mirrors never settling down, and he lived in Arizona where they have large temperature differences from day to night. If you live along a coastline or a small island, you probably have a small rate of temperature drop, and your thermal problems will be less severe.

Bryan Greer

#12 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 02 September 2009 - 04:10 AM

For typical after-sunset temperature gradients around here (Ohio, USA), mirrors up to roughly 1-3/4" thick can track the falling temperature close enough with only a rear fan. Thicker mirrors need airflow over both surfaces.


My mirror is 2" thick. I note that, besides thinning the boundary layer, the flow on the back of the mirror produces additional cooling effect. Actually I could either evaluate the heat flow rate in the CFD or measure mirror temperature. But I have already seen that the mirror cools rather quickly.

In this case, you just rely on scrambling the front boundary layer as much as possible. The effect on the wavefront is to preserve high spatial frequencies (fine detail) at the expense of low spatial frequencies (veiling glare). This is a good trade off when doing high-resolution imaging, as veiling glare can easily be processed out.


I think yo have pointed out a difference with the method of blowing air onto the mirror. Turbulent flow makes exactly what yous say: preserves fine details but produces glare (until the mirror is cooled) due the the very tiny edges of the turbulent flow which resembles micro roughness. For visual use i think i prefer not having glare (low macro contrast).

Note that my results above won't apply if you live in a region that experiences different temperature gradients. The late Jeff Medkeff reported his 10" mirrors never settling down, and he lived in Arizona where they have large temperature differences from day to night. If you live along a coastline or a small island, you probably have a small rate of temperature drop, and your thermal problems will be less severe.


I often noticed that the temperature difference increase with time even at home. However, my use of the telescope is mostly in the week end, when I drive two hours to reach ark mountain skyes. In this case the mirror may leave home at ~15°C (the garage where it is stored) and be used a couple of hours later at -10-20°C. Imagine the thermal problems.
I have recently used it at ~10°C starting from 30°C at home and it did work fine. One of the mountain site is so elevated that the seeing is rather good (0.5-1"). Two weeks ago I had one of the finest views of Jupiter.

#13 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4140
  • Joined: 03 Jun 2005

Posted 04 September 2009 - 07:40 PM

I'm getting thoroughly interested in this idea, I believe I can fit the entire collector/exhaust portion so that it has negligible depth beneath my (open) mirror cell, and make it removable as well. Did you try or model different heights to the diaphragm?

I should point out the diaphragm will serve as the primary mirror baffle in a system like this, and that's a good thing.

This is in conjunction with a thin (14.7x0.8", 18:1 aspect ratio) fused silica mirror, and aimed at maximum optimization of performance, if that helps.

Best,
Mark

#14 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 05 September 2009 - 03:50 AM

Only 0.8" thick?!? I think such a thin mirror is very fast to cool down.

I did not try different heights and the two small fans that I tried do not show appreciable differences.
If I point low at the horizon (~20°-30°) and I stop the fan, intrafocal view show the growth of the cells and that they move toward the high side of the mirror (where they really become ugly). The bottom side of the mirror remains rather clean. I think that as the convective cells forms on the low side they shift upwards.
When I turn the fan on I seen that the "strong" cells on the top are gradually absorbed and the top and bottom look very similar, with some remnant of the convective cells.
I have to stress that this is at the beginning, but eventually the mirror cools. As the mirror cools the situation improves in all the surface.

I think that the height of the diaphragm is related to the effect toward the centre of the mirror. Increasing the height produces a more uniform effect along the radius. In practice in may case the diaphragm is about 5 cm above the mirror (40 cm diameter). This means that the area of the mirror is 1257 cm^2 and the area of the annulus is 1257 cm^2 !! With 80 cfm fan I should get an average air speed of 0.3 m/s both in front of the mirror (air descending on the mirror) and radially exiting the mirror.

Changing the height of the diaphragm affects the speed of the air in the annulus, not the speed of the air descending on the mirror (which is the air that "squeezes" the boundary layer.

Reducing the heigh produces a higher annulus speed, but probably the centre of the mirror will suffer somewhat. I do not see differences between centre and edge now, but probably my system could be a bit more effective if the diaphragm were closer to the mirror.

Another important point is the flow rate. The bottom of my mirror box is chocked (there is a not so large hole). However it is clear that greater flow rate should be beneficial. In you case you can either put more more fans in parallel (e.g. 3 fans) and/or bigger ones. A solution could be 4 fans: on in the centre and three around. Use the four fans in the initial phase and turn to the central one alone when the mirror is cooled.

If you already have surface scrubbing fans it would be interested to have a comparison.

#15 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 20 November 2009 - 11:23 AM

Hello,

I have made a few Computational Fluid Dynamics simulations. The pictures refer to different locations of the diaphargm (no diaphragm, close, very close and tight).

The domain is the inside of the mirror box (half for symmetry). The dent in the bottom left part is the mirror (600 mm diameter, 60 mm thick). The symmetry axis is on the left.
Above is the diaphragm (the dent from the right) and the mirror box top (which seats the lid: the design is similar to the Obssession box).
Further above is the air column inside the shroud.

You can see the streamlines, and the air speed coded with colours. Typical speed is about 0.1 m/s (for 0.04 m^3/s fan).

The "cleaning" effect is evident. Only the centre of the mirror remains unswept. On the other hand, air speed near mirror edge can be really high, especially if the diaphragm is close to the mirror. In this case it could cool much too fast and produce TDE.

Apparently a good choice for the distance between diaphragm and mirror could be 0.2-0.5 the radius of the mirror.

Attached Files



#16 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4140
  • Joined: 03 Jun 2005

Posted 20 November 2009 - 11:48 AM

Excellent! Although with fused silica there'll be no effects due to uneven air speed.

Could you run that with the diaphragm higher up? The flow lines take quite a while to show effects from the close-in one.

Best,
Mark

#17 DHurst

DHurst

    Vostok 1

  • -----
  • Posts: 153
  • Joined: 10 Mar 2006

Posted 20 November 2009 - 11:58 AM

In a tube, a rear fan also creates a vortex which scrubs and lifts much of the boundry layer away too.

At least it does in my 10" f/5.

#18 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 20 November 2009 - 12:12 PM

My (future) mirror is SUPRAX (and scope is going to be like this http://lnx.costruzio...menti/dob01.JPG , with the above mods), so it is sensitive to the TDE caused by uneven cooling.
I can place the diaphragm higher, although, in facts, the opening for the Obsession-like lid is in fact exactly that. I can try a position in the middle.
Another thing that I would like to do is evaluate the heat exchange coefficient and the mirror deformation.

Anyway, it seems that a diaphragm should be placed quite high (half radius or little less) and that the main factor deciding the "cleaning" effect is the fan flow rate.
However it seems to me that even 5 cm per second could be sufficient to prevent excessive growth of the boundary layer (say keep it a few uniform centimeters thick, at least until the mirror cools).
The back of the mirror is strongly cooled.

#19 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4140
  • Joined: 03 Jun 2005

Posted 20 November 2009 - 04:17 PM

I would try it with variable fan speeds in practice. It may be that a quite small fan would suffice, looking at the angle of the flow you're showing.

I'll have a lot less clearance around the mirror, though. Any ideas how that would affect the performance?

Best,
Mark

#20 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 21 November 2009 - 06:26 AM

The clearance you see is an "average" clearance taking into account the square cross section of the mirror box and the round section of the mirror. The model is 2D axis symmetric. There is more clearance on the box diagonal and less in the direction parallel to the sides.
However I have also made a 3D simulation and there seems to be little 3D effects. I, will post it.
Ypu may compute the average speed in the narrows by dividing the flow rate of the fan by the area between the mirror box and the mirror. It will. probably be much less than the exahust speed and therefore I do not expect drawbacks.

I now use a very small (4cmx4cm) fan 12V 0.12A which reduces (but not eliminates) the boundary layer (when the mirror is warm).
A variable speed fan could be an option.

#21 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 22 February 2010 - 03:40 PM

Addition. I just changed the scope. Now I have a 24" Here two details of the implementation (this scope is a Kriege&Berry style).

Just a short explanation of how it works. I tested the system only one night last week (the other nights were cloudy).
I first tried with smoke and I saw the smoke approaching the mirror, turning toward the mirror edge and disappearing to the sides! Just like it should be.

I pointed immediately Mars and it was a orange blob at the very beginning. I checked the situation of the boundary layer on a star (just go inside focus and you see a boiling hexagonal pattern of the convective cells). However as the fan was turned on the scope could hold 330x n Mars. In a few minutes 420x and in half an hour 600x. The view was getting clearer and clearer... (low contrast details like aetheria were already visible and well contrasted) unfortunately the clouds arrived.

Attached Files



#22 Mauro Da Lio

Mauro Da Lio

    Viking 1

  • -----
  • Posts: 571
  • Joined: 12 Sep 2004

Posted 22 February 2010 - 03:43 PM

mirror box rear

Attached Files








Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics