Subject: Scheduling of 4Q1997 3.5m Major Engineering Shutdowns

From: elt@astro.Princeton.EDU

Submitted: Fri, 1 Aug 1997 15:16:06 -0400 (EDT)

Message number: 160 (previous: 159, next: 161 up: Index)

The message attached below proposes a new system for scheduling
major engineering shutdowns of the 3.5m.  It was posted to apo35-general
and emailed to institutional schedulers and Users Cmte. members on 2 July
and was discussed by the Users Cmte at its 14 July meeting (see
apo35-general for the minutes of the meeting).  The relevant message
numbers are 151 and 154.

We will now adopt this proposed system for 4Q1997 and perhaps beyond but
will, of course, evaluate how well it works as experience with it

IMPORTANT NOTE:  Some users, and notably the UW 3.5m user community
collectively, felt that it would be desirable to have blocks of
institutional time scheduled which each TAC would allocate with 
2 weeks of advanced notice.  The goal would be to protect or compensate
users from losing time to major shutdowns, particularly users such as
graduate students whose careers may be strongly affected, etc.
The system we will implement will allow for this possibility as follows:
Any institution inclined to do so may request that some fraction (up to
100%) of their time be scheduled to the institution but otherwise left
uncommited.  Such time will be noted on the schedule by the institutional
identifier and the number 00 (e.g., UW00).  The request may indicate that
the 00 time be distributed in any way (blocked or unblocked, distribution
of moon phases, parts of the quarter, etc.) that might be done for a
conventional request.  The 00 request should also be given some specific
place in the institution's priority ranked list of scheduling requests.
In short, it should be submited and will be treated just like any other
scheduling request except that it will not be commited to any specific
observer, instrument, program, etc.  The only restriction on the 00 category
time is that it may not be used for visitor instruments, instrument
commissioning or use of any novel hardware configurations; it may only
be used for standard, routinely supported observing modes.  (The reason, of
course, is that preparation for visitor instruments, etc. requires far
more than 2 weeks notice.)

EVEN MORE IMPORTANT:  It will be the *sole* responsibility of the institutional
TAC and Scheduler in question to inform Gretchen Van Dorn, Bruce Gillespie
and me of the assignment of 00 time to a specific institutional program at
least two weeks in advance.  The program should be fully specified at
this time [PI, observer(s), instrument(s), contact information, write-up
on the standard ASCII template, title, etc.].  Any 00 time not so assigned
will STILL BE CHARGED TO THE INSTITUTION but will be regarded as Director's
Discretionary Time to be used according to my best judgement.  Note that

Ed Turner

++++++++++++++++++++++++++July 2, 1997 Proposal+++++++++++++++++++++++++++++

An extensive set of engineering projects intended to improve the performance
of the APO 3.5m and its instrumentation to levels which are generally
competitive with the best other 4m class telescopes and which are uniquely
powerful in some respects is now well underway.  These include both those
upgrades described in the "Three Year Plan" and remediation of some problems
discovered since its adoption.  A later message from Chris Stubbs will
summarize the results to date.  Fast progress in this effort is being given
very high priority, and the necessary shutdowns of the telescope are being
accepted as an unavoidable cost.  We are continuing to try to optimize the
procedures by which we simultaneously use the telescope for science while
this engineering campaign is in progress.

The difficulties with the schedule for the planned 3Q1997 engineering
shutdown of the 3.5m which were discussed in two preceding messages
are not unique.  In fact, we have encountered quite similar problems with
the three previous major engineering shutdowns: namely, the June shutdown
for installation of the new guider, the April shutdown for installation
of a new support system for the primary, and the Dec-Jan shutdown for
replacement of the enclosure rotation wheels and re-aluminization of the

There are two basic problems:
1) Despite our best efforts, it has proven impossible to accurately predict,
   months in advance, the date on which all necessary preparations for a
   particular shutdown will be completed.  These include receiving parts,
   preliminary work not requiring telescope shutdown and scheduling of
   critical personnel to work on the shutdown.  Often the completion of
   such preparations is affected by matters entirely beyond the control
   of the Observatory and/or which are inherently unpredictable.  In
   practise, since things always take longer than one expects, we have
   never been fully ready at the beginning of a major shutdown.  This
   results in less efficient use of the shutdown time and thus ultimately
   reduces available science time.
2) Despite often heroic efforts by those working on a shutdown, it usually
   has proven impossible to complete all planned engineering tasks
   within the scheduled duration of the shutdown.  This is partly due to
   problem #1, not being fully ready to start.  However, it is also due
   to other factors such as absence of a detailed understanding of the tasks
   involved months in advance, growth in the scope of planned work as time
   passes, discovery of unanticipated problems in the course of the
   shutdown work, need for clear skies for some tasks (typically as the
   telescope is being returned to service) and so forth.  This results in
   both disappointment for users scheduled to use the telescope immediately
   following the shutdown and in enormous stress on those working on the
   engineering tasks as they strive to meet what may well be an unrealistic

Based on experience and such considerations, all those immediately
concerned have concluded that the present system of scheduling extended
shutdowns (EN00 time) is not working in an acceptable way.  We will
therefore try one or more different systems in future quarters.

A specific proposal is described below.  The Users Committee will discuss
it on July 14 and perhaps in subsequent meetings.  Please give your
comments and suggestions to your institutional UC representative and/or
email them to me.

The proposed new system for dealing with EN00 time is as follows:
1) No EN00 time will be explicitly scheduled in each quarter; only EN01
   time will be scheduled specifically (EN01 is typically a single night
   set aside for routine engineering tasks on a regular basis, e.g.,
   pointing models, and usually amounts to about 7% of all available
2) When *all* preparations for a given major engineering task requiring
   extended (more than one night) shutdown of the telescope are *complete*,
   those planning to carry out the work will request a shutdown of some
   specific duration, based on their best estimate of how long it will take.
3) The shutdown will then be scheduled and announced at least, but very
   often no more than, two weeks in the future (2 weeks notice) and will
   simply pre-empt whatever science programs are scheduled for that period.
4) From the Observatory point-of-view, time lost in this way will be like that
   lost to bad weather or equipment problems in that there will be no attempt
   to provide compensating time.  (Of course, institutions may arrange such
   compensation internally if they wish.)  Of course, pre-empted time will be
   accounted as engineering time and not charged to the affected institutions;
   in this respect, it is not like time lost to weather or equipment problems.
5) In scheduling such pre-emptory engineering time, account may be taken of
   programs that are of particularly high priority (as determined by
   institutional TACs) or which would produce data not easily obtainable
   at other times (e.g., occultations, observations coordinated with other
   observatories or spacecraft).  Programs which are believed to merit such
   special protection should be explicitly identified in the proposals and
   institutional scheduling requests.  However, it should be understood that
   no program can be certainly exempted (i.e., no promises) from cancellation
   due to EN00 engineering shutdowns.

This system should solve the first problem described above and somewhat
alleviate the second (which probably cannot be entirely avoided).  Although
it will doubtless produce inconvenience and disappointment for those whose
observing time is canceled with as little as two weeks notice, this should
be viewed in perspective with the fact that observing time is often lost with
much less (even zero) notice to weather and equipment problems.  The overall
gain in total science time available (due to more effective use of shutdown
time) and in telescope/instrument performance (due to the engineering
improvements themselves) should also help to ease the blow.

Many other ways of dealing with this scheduling problem are certainly
possible, and several of them have been proposed and discussed.  However,
the one described above seems to me to be the simplest and most effective.
I look forward to receiving your comments on it.

Ed Turner
APO APO APO APO APO  Apache Point Observatory 3.5m  APO APO APO
APO  This is message 160 in the apo35-general archive. You can find
APO  the archive on
APO  To join/leave the list, send mail to
APO  To post a message, mail it to