add steps, we almost always simply regenerated the
model from the procedure using the translator.
Underlying all was the ASPEN planner-scheduler. No
adaptation-specific code was used for Orbital Express
… all of the adaptation used the out-of-the-box capabilities of ASPEN. To plan any given day the following steps were followed:
1. Load all of the known opportunities (AFSCN visibilities, TDRSS visibilities, and procedures that have
already been preplanned by hand or from previous
runs of ASPEN).
2. Instantiate as many alternative plans as necessary
to accommodate the uncertain parameters. In practice, long-term plans had several alternatives but
short-term plans had only one.
3. Schedule all activities using an earliest-deadline-first
ordering, also known as forward dispatch scheduling.
This results in realigning activities to the known availabilities and initial expansion into the supporting subactivities.
4. Iteratively repair the plan to address any plan flaws
(for example, unexpanded decompositions, open temporal associations between activities) and constraint
violations (such as resource oversubscription, state
violations, and timing violations).
Each of these steps was automatically performed
by the ASPEN system under the direction of the SRP.
The planning process for any given day began weeks
in advance. A plan was built from knowledge of the
existing contacts available and an ASPEN-generated
and edited model of what the procedure was to do
and how the contacts should lay out across time (
The AFSCN contacts were reserved up to a limit
and occasionally with elevated priorities specifically
for the unmated scenarios. TDRSS support was originally also scheduled in the long-range planning time
frame for all scenarios; however, cost constraints and
changes to the plan in the short term dictated the
need for a policy change.
It was determined more efficient to schedule
TDRSS at the daily planning time, except in the case
of unmated scenarios, where the timing and the
more definite guarantee of contacts was crucial.
Although the essential replanning generally
occurred at the daily planning time, variations on
the long-range planning occurred from several fac-
tors. First, our launch delay created the need to
replan all existing long-range plans to shift both
AFSCN and TDRSS requests. Second, changes to mod-
els occurred often during the long-range process, due
to many factors, including updated knowledge of
timing, procedure step removals and additions, and
general modifications to procedure step times or
requirements. Third, occasionally, maintenance
requirements or site operating hours were learned
postdelivery of the long-range planning products and
a replan was necessary. Finally, other factors that
required replanning the long-range products were
often late enough in the plan time line that a new
“midrange” plan was created. This usually was done
a few days outside of the daily planning. Figure 6
depicts the planning flow.
In the morning of daily planning, the SRP would
receive the list of contacts lost to other spacecraft and
any suggested additions to replace these losses, and
he or she would also receive the most up-to-date list
of TDRSS availabilities. The contact losses would need
to be evaluated against the procedure objectives of
the day to determine whether they could still be met.
The ASPEN model of the procedure could be adjust-
ed as needed to reflect any operations updates, and
the ASPEN activity could be moved around through-
out the day to accommodate the contact require-
In the nominal case, the planning process would
call for the use of the long-range plan and simply
update necessary timing information to create the
daily plan. However, daily planning was based on
many variable factors culminating in a need for both
simple updating of the plan and completely replanning the long-range plan: ( 1) The visibilities of contacts with the position of the spacecraft drifts slightly per day and must be updated in the short term to
make most efficient use of the AFSCN communication times. Even one minute of contact coverage loss
was, at times, considered valuable. ( 2) The daily
deconfliction process can mean a loss of several contacts based on any number of reasons (site-specific
issues, other satellite conflicts). Losses may require a
shift of the procedure to perform the requested objectives. Also, losses are often accompanied by gains,
and replanning can be based on such new additions
to the plan. ( 3) Scoping of the day’s long-range plan
may change due to both anomalies and new direction from operations. Updating the existing plan at
the daily planning time was often required for previously unknown amounts of needed coverage or for
real-time failures of contacts pushing into the next
day. ( 4) TDRSS support was originally requested in
advance for all long-range planning, but as cost
became an issue for unused contacts, the requests for
TDRSS became part of the daily planning process.
This was a major addition to the update of the long-range plan. ( 5) Dealing with the sometimes unpredictable conditions of space and limited mission
time, a number of unforeseen events could cause the
need to update the long-range plan.