STRUCTURAL ANALYSIS OF THE MAGELLAN PROJECT EIGHT METER TELESCOPE

Artist's rendering of the Magellan Project 8 Meter Telescope.

The Magellan Project 8 Meter Telescope is one of the number of preliminary design studies recently performed by L&F Industries of Huntington Park, California using Algor's Finite Element Analysis (FEA) System. The primary mirror of this telescope will be approximately three meters larger than that of the well-known Hale Telescope, one of the world's finest large-scale telescopes.

When completed, the telescope will be installed atop a 7000 foot high mountain at Las Campanas, Chile. It will be operated collectively by the Carnegie Institution of Washington, the University of Arizona, and Johns Hopkins University. The system will be used for astronomical research, including investigation of the evolution and space distribution of galaxies, and the study of intergalactic dust clouds and properties of quasars.

Although this project is in its early stages, much has already been learned about the expected performance of the telescope. In fact, the telescope's performance has been improved dramatically thanks in part to FEA on personal computers.

Terminology for the Telescope Structure

For those unfamiliar with large telescopes, the following terminology may prove helpful. The optical support structure (OSS) is, as its name suggests, a structure which supports the focusing mirrors and usually the instruments that detect and analyze the electromagnetic radiation of the object(s) being observed. The primary concave mirror, in this case eight meters or 26 feet in diameter, reflects the light to the so-called prime focus. However, before the radiation reaches a focus it may be intercepted by a convex secondary mirror. This mirror then reflects the light through a hole in the primary mirror where it comes to a more conveniently located focus. At this point, the light is analyzed by the appropriate instrument of choice: photometer, spectrometer, etc. The astronomer can then interpret the results or have them further processed by computer for later analysis.

The second major structure in this system supports the OSS and is called the Azimuth Disk. The OSS rotates on the Azimuth Disk about a horizontal "altitude" axis, while the Azimuth Disk can rotate about a vertical "azimuth" axis. This "Alt-Ax Disk" mount departs significantly from previous designs and has been found to be very efficient. Under computer control, both rotations can then cause the telescope to point or track anywhere in the sky.

The centrifugally cast primary mirror will be the largest made by a process being developed by the Steward Mirror Lab in Arizona. The one-piece borosilicate mirror will have a honeycomb construction and weigh 30,000 pounds.

An Algor model for the Magellan Project 8 Meter Telescope (Diametric view - 17 degrees off zenith).

Analysis

The structural analysis of this system has consisted of static runs to evaluate optical misalignments due to gravity and wind loading, and modal analyses to predict natural frequencies and modeshapes. The "intermediate" level models, although not as detailed as will be used for the final detail design analysis, have sufficient accuracy for the purpose of optimization of the structure. The larger detail design models will then serve only to verify the accuracy of the simplifying assumptions used in the intermediate models. In this way, most of the FEA can be done on a personal computer, with the benefits of a more interactive process.

The hardware used for this work has been an 8 MHz AT clone with a 72 megabyte hard drive required to store some of the larger stiffness matrices. Models have varied widely in size from small local models requiring a few minutes of run time; more detailed models of the complete system requiring a few minutes of run time; and more detailed models of the complete system requiring approximately 30 hours. An example of a complete model required:

• 637 nodes
• 853 plate/shell elements
• 78 beam elements
• 43 megabytes disc space

The final (somewhat larger) detail design models will be run either on a 386 system using Algor's Hypersap, or on a VAX system using the VAX version of Supersap.

The plate and mesh generators included with Algor's FEA System have been invaluable in creating the above models. For example, the primary mirror cell requires 368 plate/shell elements, most of which were created in a very short time using the program RADGEN. These programs - as well as AEdit, PUSH, and Substruct - have been used predominantly to create rather complex, accurate models in a short period of time.

Another of the Algor programs which has been invaluable is Animate. When this work was first begun, it was found that it was so difficult to identify some modeshapes from the scaled deflected shape alone, that they were sometimes being misidentified. That is, what motion is really taking place in a particular mode of vibration? When the modeshape is subsequently animated it is immediately understood. This has also been very valuable for presentation in meetings. A video is made by directly shooting the CRT with a home video camera for later presentation to the customer. A more formal presentation displaying and describing the complete model can also be accomplished by editing the tape and dubbing an audio track onto it.

Mode 1 (lateral translation - 7.34 Hertz).

Using the above-referenced method of optimizing with intermediate level models, the design has progressed dramatically. Two very different mount configurations, as well as detailed structural differences, have been analyzed to improve the vibrational and static properties of the system. It is desirable to increase the natural frequencies so that the structure will be less susceptible to excitation. Weight should be reduced to decrease costs (in most cases) and to improve thermal performance of the structure, since it will be able to thermalize faster to air temperature. For instance, the current design achieves a 7.34 hertz (cycles per second) lowest resonant frequency with a total system rotating weight of 415,000 pounds. An earlier, complete preliminary design achieved only four hertz at 500,000 pounds rotating weight. The new design, which employs a much different approach, is significantly lighter and stiffer than the original.

By comparison, many consider the Palomar 200 inch (five meters) Hale Telescope to be the most successful large telescope in operation anywhere in the world today. If scaled up to an eight meter size in a relatively efficient manner, but keeping the same mount and optical configuration, it would weigh about 3,500,000 pounds, or approximately eight times the weight of the Magellan Project Telescope. Of course, most of this improvement in the Magellan Project Telescope is due to major conceptual changes such as mount type, f-ratio, mirror technology, and the like.

But there is little question that FEA on personal computers allows us to learn more about the performance of a structure, to a greater degree of accuracy, and at lower cost than with previous methods.

Optical misalignments resulting from 15 mph side wind (magnified 150,000 times), 172 micro-inches decenter.

The Magellan project preliminary design study is being performed by L&F Industries, Huntington Park, California. Directing the design is Dr. Al Hiltner, Carnegie Institution of Washington, from the project office in Pasadena, California. A concurrent, cooperative effort is also in progress on the Association of Universities for Research in Astronomy/NOAO 8 Meter Telescope under the direction of Dr. Larry Barr.

Other recent preliminary designs of large telescopes by L&F Industries using Algor's FEA System include the Columbia University Spectroscopic Telescope, and University of Minnesota (consortium) 3.5 Meter Telescope.

Copyright © 1988 Algor, Inc. All rights reserved.