The demonstration revolves around the DIII-D tokamak operation.
We plan to make the SC03 participants be actual remote participants
to DIII-D operations.
DIII-D officials have agreed to our demonstration plan.
Tokamak may break and not run - what is backup plan?
Tokamak pulse is about a 20 minute cycle. Thus one tokamak
pulse is our demo slot. How does this work? Is it better
to have live AG feed from the control room but fake a faster
pulse cycle?
Alternative scenario is to have a multi-hour demo slot on one day so
we can easily interact with the control room and show people
what it is like to be a remote collaborator. If we only get one
day then we need to tape this, show it on other days, maybe
use it as PR in the future.
Possible backup scenarios
If the tokamak does not operate we can simulate in software data acquisition
and the like to show our software. We lose people in the control room
but we can put people there and fake it.
If there is a network failure such that we can not get to DIII-D then
our options are more limited. In that case, we would need to fake everything
(.e.g MDSplus server, PTDATA server, EFITs, ec. ) from within the SC03 venue.
At first thought this seems like way to much work for a low probability failure.
Between Shot Application
We will run EFIT as our QoS between shot application. This will run on the PPPL
Linux cluster and will duplicate one of our normal EFITs run at DIII-D (EFIT01).
Thus, if we have any trouble, we do not impact tokamak operations.
Data will be read and written to the DIII-D MDSplus server.
Workers in this area include Keahey, Peng, Randerson.
Demo Flow
Starts waiting for a shot....QoS EFIT parameters have been set.
Information about the run can be communicated to audience including
the preparation work done before the day arrives.
Remote particpant sees audio/video of control room (AG node)
Remote particpant sees standard information: where in shot cycle,
logbook, etc (normal DIII-D apps).
As shot happens, remote participant sees real time display
of plasma traces (normal DIII-D apps). Hear's audio of chief operator
counting down time (AG node).
Shot ends, data acquisition starts. Info is displayed as pieces of
data acquisition is completed (normal DIII-D apps).
Kate's QoS EFITs begin and finish
Using ReviewPlus and EFITViewer (IDL aps) remote participant begins
to plot data as it is acquired. Some rudimentary analysis can be done.
Based on analysis, decisions can be fed back into the control room
for the next pulse. Communication via AG node and shared applications.
Based on analysis and comparison to NIMROD simulation more decisions
can be made - uses 2D SCIRun capability.
Working simulation and SCIRun vis into demo
It is a stretch to get a NIMROD simulation and SCIRun vis into this demo.
Any good ideas here? Difficulty is knowing which physics experiment will
be run on our day early enough so NIMROD can be run.
We could use SCIRun to do 2D vis like ReviewPlus just to show the app.
How do we work tiled walls into the demo
Is the AG node view of the DIII-D control room tiled wall enough?
Is there a tiled wall scenario at SC03 show floor that makes sense in our demo?
