Subject: Re: Minutes of the November User's Committee Meeting

From: waddell@astro.washington.edu

Submitted: Thu, 7 Dec 1995 11:20:10 -0800

Message number: 29 (previous: 28, next: 30 up: Index)

I would like to submit a number of, hopefully clarifying, comments related
to topics covered in the last User's Committee Meeting (based on the
November minutes).

Before we jump into the subject of achievable bottom line image quality
from the 3.5, I thought many in the community would find it useful to see
the original design budget for image quality, as well as the performance
numbers for some of the optical elements for which we have real
measurements.  (Although it would be desirable to rewrite this table with
measured values, practicality and resource conservation ($) will require us
to make "lumped" tests and separate some elements analytically.)

The first colummn in the following chart ('84 Budget) details the expected
image size "contributions" for the relevant sub assemblies.  For optics
figuring entries, the values selected represented reasonable tolerances to
be met by present day professional fabricators of optics.

In the "8/94" column, the numbers are the same as in the original image
size budget, with the exception that shop measurements of the
"as-delivered" primary and secondary figure errors have been folded in.
The effects of the excellent quality of the primary and the unfinished
secondary are included.

In the "Present" column Lick Observatory measurements, made after Kodak ion
figured our secondary, show a much better bottom line.  However, the
improvement was insufficient to come close to our original goals.

The last column indicates the expected performance regime we would be in if
we return to our original fabrication specs on the secondary mirror figure.


3.5m -  no wind, zenith pointing FWHM image size (visible lambda) estimates
(does NOT include site seeing)

Contributing source             '84     8/94   Present       '97 M2
                                Budget

Primary mirror (M1) figuring    0.23    0.14    0.14            0.14
M1 support design               0.10    0.10    0.10            0.10
M1 support implementation       0.03    0.30    0.03            0.03
M1 thermal loading              0.07    0.07    0.07            0.07
*Other optics figuring          0.08    1.55    0.58            0.08
Other optics support            0.07    0.70    0.07            0.07
Local seeing                    0.13    0.13    0.13            0.13
Control system                  0.07    0.07    0.07            0.07
Focus                           0.03    0.30    0.03            0.03
Collimation                     0.07    0.07    0.07            0.07

        Subtotal                0.33    1.63    0.64            0.27



As budgeted in '84, 10 percentile site seeing, estimated to be 0.60 arcsec
was expected to provide images in the neighborhood of 0.68 arcsec FWHM.
Adding this amplitude of site seeing to the "Present" column provides
images with 0.87 arcsec, which is about right for what we see today; the
secondary has an overwhelming impact.  It is clear that we are missing out,
entirely, at times of great seeing and while working at longer wavelengths
where r0 is larger.

It is important to stress here that no earlier design decisions were taken
that would limit us to numbers higher than these.  The goal has always been
to design the 3.5m to be able to take advantage of times of excellent image
quality.





From the November meeting notes:
>>Best seeing ever seen on this telescope is 0.9"; budget for telescope
>>contribution to this is 0.5". Hartmann test shows that half of the
>>error budget is taken up in optics, the other half is due to "other":
>>mechanical jitter, collimation, and enclosure and site seeing. There
>>are strong suspicions that the optics problems are due to the
>>secondary. Practically speaking, we are pretty sure that it is indeed
>>the secondary. To clinch it, one can look whether particularly deviant
>>rays from the Hartmann test come from particularly rough areas on the
>>secondary.  Doing Hartmann testing on the primary alone will require a
>>big project putting a prime focus instrument on the telescope. Another
>>approach would be to rotate the secondary 120 degrees. We also don't
>>have a hard price for replacement of the secondary (we cannot get a
>>serious bid until we have a firm budgetary committment; a chicken and
>>egg problem). Board will look at this at their meeting on November 20,
>>but they won't have all the necessary information to make a firm
>>decision.





>>
>>Seeing of DIS is systematically higher than for DSC, GRIM, which
>>presumably is due to the undersampling of the DIS.
>>
Also recall that Rnaught is larger for GRIM (goes as lambda^6/5ths)






The brief discussion related to the enclosure wheel cracks described the
absence of "a smoking gun" cause of failure.  For those with a strong
interest in the subject, the following investigation summaries from APO and
UW technical staff members may help mitigate some of the mystery. Note that
analysis indicates that maximum stress within the wheels occurs at a depth
of 0.13" while the original wheel design called for hardening to 0.06".  In
supplement to the following, Jon Davis, at the site, has mentioned that
tests indicate that the wheels may not have been hardened to the design
spec. of RC44 and may have been a soft as RC32-34.



>From: Jon Davis, Bruce Gillespie, Charlie Hull, Mark Klaene and Walter Siegmund
>
>Description:
>
>About March of 1992, cracks were noticed in some of the enclosure rotation
>wheels. Four wheels, 30" in diameter and 3.5" wide, support the 200,000 lb
>enclosure. The wheels are machined from 1040 steel plate. The specification
>calls for flame hardening to Rockwell C 44-48, 1/16" deep. When first
>noticed, a single crack extended circumferentially part way around the
>wheel. The crack grew until it went all the way around the wheel and in
>some cases parallel cracks appeared. The initial crack grew in width so
>that the worst crack is about 0.020" wide.
>
>Ultra sound measurements were made of the crack depth about 1 year ago.
>These indicated that the deepest crack was on the left-rear wheel (facing
>the shutter from the inside). The crack was about 0.5" deep. Measurements
>taken about 1 month ago indicated that this crack has grown to 1.0" deep.
>Also on the left-rear wheel, two parallel cracks are in the process of
>merging. At the conclusion of this process, we can expect a piece of the
>wheel to spall off the wheel leaving a 0.4" x 1.5" pit.
>
>The right-rear wheel has no cracks apparent under visual or ultra sound
>inspection. The front wheels have cracks that are intermediate in extent.
>
>Discussion of failure:
>
>The compressive stress in the contact region between a cylinder and a plane
>can be much larger than the nominal compressive strength of the materials
>(see Shigley, Mechanical Engineering Design). This is because the
>3-dimensional or triaxial stress field, to first order, acts to increase
>the density of the material and this does not lead to failure. It is the
>second order shear stresses that lead to failure if they approach the yield
>strength in shear of the material. For the wheels, the maximum shear stress
>occurs about 0.130 inches below the surface.
>
>The flame hardening process begins at one point at the circumference and
>proceeds around the wheel. The process always leaves a region of softer
>material where it terminates.
>
>We find the following scenario plausible. A crack initiated beneath the
>surface at the depth of maximum yield stress (most likely in the soft spot
>left by the flame hardening process) and propagated to the surface. Once
>there, it propagated around the wheel and deepened. Other cracks initiated
>at the soft spot subsequently and began to propagate also.
>
>We can expect the propagation of the cracks to continue. The number of
>cycles until failure decreases exponentially as stress increases. If the
>dominant stress causing crack growth is currently the lateral loads on the
>wheel due to slight misalignment  of the wheel with the track (so-called
>tracking error), we can expect the stress to rise slowly as the crack
>deepens and growth will accelerate. If, however, crack growth is due to
>contact stress, growth may slow as the crack propagates away from the
>contact region. The empirical evidence from the ultra sound measurements
>suggests that crack growth is accelerating.
>
>Amelioration:
>
>The most immediate failure is likely to be the spalling of the left-rear
>wheel. The fragment may be left on the track where it might be run over by
>the same or a different wheel. This would brinell the track at that
>location. Consequently, the track wipers should be removed from the
>left-rear wheel and strong magnets placed near the wheel and above the
>track so that any fragments are captured. Also, the wheel should be checked
>daily and by the observer whenever he or she visits the intermediate level.
>
>The contact region of the wheel is separated from the intermediate level by
>8" of steel plate. We believe that the chance of a fragment being ejected
>by a failure into the intermediate level walkway is negligible.
>
>Crack propagation should continue to be slow rather than catastrophic. Even
>if the crack propagates to the hub before the enclosure is shut down, the
>wheel will still be captured by the 8" steel housings on either side of the
>wheel. Even if the wheel were to disappear, the building would be supported
>by 1.5"x3" track wiper supports that span across the track between the 8"
>steel housings. These clear the track by about 1/2". The building will
>continue to be restrained lateral by the cam followers which will be
>unaffected by the failure of the wheel.
>
>Recommendations:
>
>The replacement of the left-rear wheel should begin as soon as a
>satisfactory solution is identified. This is a very high priority task
>since either a large spall or crack deeping beyond about 2" will result in
>the shut down of the building rotation system to avoid more extensive
>damage.
>
>APO engineers, with the assistance of the engineers at L&F Industries and
>other consultants, should propose a solution to the problem. In parallel,
>plans for  removing handling the left-rear wheel should be developed. The
>wheel weighs 817 lbs and the wheel housings weigh about 1500 lbs each so
>careful planning is essential. Addition equipment should be acquired as
>necessary.
>
>Once the left-rear wheel has been replaced, it may be possible to
>remanufacture it (see below) so that it serves as a spare. Crack growth in
>the other wheels should be monitored and they should be replaced and
>remanufactured as their condition warrants.
>
>A specific approach that may be fruitful is to replace the wheel with 4340
>steel alloy that is hardened in an oven to a depth of 1/4 inch, i.e., below
>the depth of maximum shear stress. This process would avoid leaving a soft
>spot in the circumference of the wheel. Either the same wheel design could
>be used (with the change in material) or a 4340 forged tire could be welded
>to a 1040 steel wheel. The later approach is particularly applicable to the
>remanufacture of the other existing wheels.


APO APO APO APO APO  Apache Point Observatory 3.5m  APO APO APO
APO
APO  This is message 29 in the apo35-general archive. You can find
APO  the archive on http://astro.princeton.edu:82/apo35-general/INDEX.html
APO  To join/leave the list, send mail to apo35-request@astro.princeton.edu
APO  To post a message, mail it to apo35-general@astro.princeton.edu
APO
APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO APO