@sallyrobson I have a couple of things I'd love to discuss at our Thursday meeting. Storing them here to document. No need to look at them until Thursday.
Output Folders
Right now, in E4ST.jl, the output_path is provided in the config file, and all of the outputs are written to that path. When the config file is loaded, E4ST checks to see if there is anything already in the folder. If it does contain anything, it will move the contents of that folder to another folder named <output_path>_backup_<timestamp>. That sometimes causes some problems, like when the io for the E4ST log or the solver log is still open, or didn't get properly closed/flushed from the memory. This happens quite frequently unless you are very careful to close things
Another way to do this would be to still specify the same output_path in the config file, but when the config file is loaded, create a new timestamped file folder inside of the original output_path, i.e. <output_path>/yyyymmdd_HHMMSSsss. Then possibly rename config[:output_path] to the newly created path, and make config[:base_output_path] equal to the original.
Results Data Structures
Objective Scaling
we decided to scale it, then scale it right back so that the user only has to do things in dollars. Scaling can happen later on.
Ray: If voltage angle differences are large, combining co-located buses could make a big difference
Ray: Are they also directly connected, i.e. via transformer? There could be flow limits on the transformer that would be eliminated by combining the buses.
Dan: We could do a test of difference in solution between running with and without combining co-located buses. This testing method could be a way of testing out voltage angle difference threshold as well.
Ray: simulating to get phase angle differences could be better with wide ranges of scenarios/operating points.
Ray: we could find equivalent reactances (even if not directly connected) to find equivalent reactance, then threshold on the reactance. Large reactance (small susceptance) will give a small phase angle difference given the same amount of power flow. Could combine buses given that their equivalent reactance is large, above a threshold.
@sallyrobson I have a couple of things I'd love to discuss at our Thursday meeting. Storing them here to document. No need to look at them until Thursday.
Output Folders
output_path
is provided in the config file, and all of the outputs are written to that path. When the config file is loaded, E4ST checks to see if there is anything already in the folder. If it does contain anything, it will move the contents of that folder to another folder named<output_path>_backup_<timestamp>
. That sometimes causes some problems, like when theio
for the E4ST log or the solver log is still open, or didn't get properly closed/flushed from the memory. This happens quite frequently unless you are very careful to close thingsoutput_path
in the config file, but when the config file is loaded, create a new timestamped file folder inside of the originaloutput_path
, i.e.<output_path>/yyyymmdd_HHMMSSsss
. Then possibly renameconfig[:output_path]
to the newly created path, and makeconfig[:base_output_path]
equal to the original.Results Data Structures
Objective Scaling
we decided to scale it, then scale it right back so that the user
only
has to do things in dollars. Scaling can happen later on.Power balance expression