Electricitymap-contrib: Seminole Electric Cooperative (Florida)
Seminole Electric Cooperative generates power in central Florida and distributes it to smaller coops shown on this map.
It has a 1.3GW coal plant and an 800MW gas plant plus some solar PV generation.
https://www.seminole-electric.com/facilities/generation/
The solar data (hourly) can be accessed via csv from this chart app.
http://apps.seminole.coop/db/cs/
We can get total generation (hourly, D-1~2 lag) using eiapy from https://www.eia.gov/opendata/qb.php?category=2122629&sdid=EBA.SEC-ALL.NG.H
We can then join these data sets to get a decent breakdown which will lag 1-2 days behind.
systemcatch
All 30 comments
Thanks for the find, good to keep it here.
Let's keep that implementation for a bit later when we'll have a more holistic view of the area!
brunolajoie
on 9 Jan 2019
To be honest I wouldn't expect a better view of the area anytime soon given that there's no ISO and every other generator is for profit which usually means they provide no data.
systemcatch
on 9 Jan 2019
Thanks for the info! I think we could label this issue as "parser buildable" then, right?
brunolajoie
on 10 Jan 2019
Ok I've done that now. One question is if we need to represent this area on map considering it will behind by 1-2 days.
systemcatch
on 10 Jan 2019
If we think this zone breakdown is here to stay, that would make sense to invest a bit of time designing it if we find the geojson to go along, indeed.
Actually, in the end, our map of the US will probably looks like Watttime one's I guess
brunolajoie
on 10 Jan 2019
i zoomed into watttime's map, and it actually looks pretty different from the SEC Map @systemcatch posted above:
snarfed
on 1 Jul 2019
@snarfed huh interesting but not too surprising given how crazy the US grid can be. I'd trust Wattime's map over the rough one on Seminole's website.
If you're writing the parser for this I can assign you to the issue.
systemcatch
on 2 Jul 2019
If you're writing parser for this I can assign you to the issue.
sure, thanks! i have no real GIS experience, so i'm starting with just the parser, but i'll happily at least try to tackle the map and geojson after that.
snarfed
on 3 Jul 2019
We can then join these data sets to get a decent breakdown which will lag 1-2 days behind.
@systemcatch did you have a specific idea for how we'd determine the hourly split between coal and gas? would we just naively subtract solar from the total and then apply the 1.3:.8 GW ratio to the remainder? (seems unlikely.) or report the whole remainder as unknown? or something else?
(apologies if i missed something obvious, i'm new to the domain here.)
snarfed
on 13 Jul 2019
When we only have data for renewable categories we usually throw the rest of production into "unknown" and give it a static carbon intensity based on yearly average. See e.g. https://github.com/tmrowco/electricitymap-contrib/blob/f5934f53bc10354e95da275195e149d860cf2b7e/config/co2eq_parameters.json#L576-L582
It's an imperfect approximation, but it lets regions show off their renewables...
jarek
on 15 Jul 2019
as part of this, i'll need to merge two different sets of production data, solar and gas/coal. i looked for code to reuse, and amusingly, there are at least five different existing merge implementations that work with the fetch_production() output format.
i've included them below. i don't plan to try to unify them as part of the US-SEC PR, but that might be worthwhile at some point in the future.
snarfed
on 17 Jul 2019
I have an internal contact in Seminole if there are questions that might need to be answered. Please @ me if necessary with questions I can relay via email.
brandongalbraith
on 17 Jul 2019
thanks for the offer, @brandongalbraith! much appreciated.
just fyi, the SEC data i've been seeing on EIA has been a bit flaky. sometimes it's only a day or two old, but sometimes it lags by over a week. eg right now it stops on 2019-07-09: https://www.eia.gov/opendata/qb.php?category=2122629&sdid=EBA.SEC-ALL.NG.H
also, i was using the day-level excel file export on http://apps.seminole.coop/db/cs/ , but that also recently started returning empty data. i've switched to the week-level file, which is still fine.
just a heads up on both of those for now.
snarfed
on 18 Jul 2019
i started to tackle the geojson for SEC, but didn't get very far. 🙁
i've read and roughly understand web/README.md. i've run docker-compose run --rm web ./topogen.sh successfully. i get that we use NACIS data, and generate-geometries.js is the starting point that maps our parser zones to NACIS geometries.
...having said that, i have no idea how to go about finding SEC's geometry (ies?), or whether or how i'd edit them, and i only vaguely see where i'd put them in the repo (third_party_maps/?) or how i'd hook them into the pipeline that builds world.json.
apologies, i've never really done any GIS work before. any tips would be appreciated! or if it's easier for you all to do it yourselves, and i should just go hack on more parsers instead, i'm ok with that too. whatever works best.
snarfed
on 21 Jul 2019
Hi @snarfed,
There's a few ways to go here. First, if we can approximate SEC's geometry to some administrative boundary (state, region..), that would be very helpful as those seem to be defined already in our geometries.
Else, we would need to resort to something more complicated by drawing it ourselves.
Would it be a huge mistake to approximate with Florida?
corradio
on 23 Jul 2019
ok, i've ramped up on GIS in general, and our base NACIS map datasets in particular. i've also seen how all of our current USA data is whole states or groups of them. eg ERCOT is not quite all of Texas (afaik), but its geometry is approximated to the whole state.
in this case, though, SEC is maybe just half of Florida. nowhere near as big a fraction as ERCOT in Texas. (see the SEC and WattTime maps above.) so approximating it to all of Florida seems too aggressive? or is it fine? i'm ok either way.
snarfed
on 29 Jul 2019
I'll look into whether SEC's regions map to Florida counties. if so, i could add new geojson based on those, without drawing anything myself.
snarfed
on 29 Jul 2019
Thanks for investigating.
It seems like we have three options here:
1/ We say SEC is Florida (well aware of the approximation we're making)
2/ We approximate with counties (if possible)
3/ We take the geometry of Florida and segment it into SEC and another part.
Let's see what 2 reveals. Thanks Ryan!
Olivier
On Mon, Jul 29, 2019 at 7:54 AM Ryan Barrett notifications@github.com
wrote:
I'll look into whether SEC's regions map to Florida counties. if so, i
could add new geojson based on those, without drawing anything myself.—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/tmrowco/electricitymap-contrib/issues/1713?email_source=notifications&email_token=AAMUIKBCKWAJOSPCUTIUV4DQB2AZVA5CNFSM4GLAMD62YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD27VDRI#issuecomment-515854789,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAMUIKFHK4ZYLNWLTZKCMRLQB2AZVANCNFSM4GLAMD6Q
.
--
Olivier Corradi
Founder, CEO
https://www.linkedin.com/in/oliviercorradi https://twitter.com/corradio
tmrow.com
corradio
on 29 Jul 2019
ok, i have a rough approximation of SEC by Florida counties that only adds ~4KB to world.json. i'll clean up the geo code and add it to this PR.
i'd also like to hide the county boundaries inside the SEC, but i don't know how high priority that will be. we'll see.
snarfed
on 2 Aug 2019
Excellent work @snarfed! You might find https://github.com/tmrowco/electricitymap-contrib/pull/1615 useful as we had to remove county boundaries in that PR.
systemcatch
on 2 Aug 2019
thanks @systemcatch! and hmm. i see your request to merge boundaries there, and the following push and subsequent screenshot without boundaries, but i don't see the code that did it. was it this script? it sounds like that just removed everything above a certain latitude, maybe to split out islands like the Shetlands?
otherwise, the only other issue with the PR (#1902) is that topogen.sh re-downloads US-FL.geojson every time, even though it includes the same if [ ! -e ... ] construct as the NACIS zip downloads, which works fine. i've poked around in the docker image and mockserver container, but haven't figured this out yet. let me know if you have any thoughts. thanks in advance!
snarfed
on 4 Aug 2019
@Kongkille we could really use your help with the geojson here.
systemcatch
on 5 Aug 2019
success! found the right topojson functions: merge/mergeArcs. this example also helped.
snarfed
on 8 Aug 2019
i've added this to generate-geometries.js in e80d5e5. to get it to work, i had to do it at a specific point: after converting to TopoJSON, but before simplifying. so the flow isn't quite as clean as before. apologies. i'm open to alternatives!
otherwise, i think this PR is ready for re-review. thanks again for all your support!
snarfed
on 8 Aug 2019
Hey @snarfed something I noticed is that the US-FL zone over highlights the US-SEC zone. I think if we build Florida from the unused parts of your US-FL.geojson that should stop that happening.
systemcatch
on 13 Aug 2019
@systemcatch agreed! sorry, i noticed that too. i should have fixed it in the original PR. i'll do that soon!
snarfed
on 13 Aug 2019
i've been looking into subtracting US-SEC from US-FL in TopoJSON, but haven't yet found existing code for polygon difference anywhere. closest is this unanswered SO question: https://stackoverflow.com/questions/42684306/d3-js-relative-complement-in-topojson
i may convert back to GeoJSON, use something like https://github.com/mfogel/geojson-clipping , and then convert back to TopoJSON. we'll see.
snarfed
on 28 Aug 2019
If you build FL from the remaining counties (change https://github.com/tmrowco/electricitymap-contrib/blob/master/web/generate-geometries.js#L563 to be like https://github.com/tmrowco/electricitymap-contrib/blob/master/web/generate-geometries.js#L611) that should work I think?
systemcatch
on 29 Aug 2019
hah, true, thanks. that would be the easy way; why would i want to do that? 🤦♂️
i actually got it working the hard way, with polygon-clipping.difference(). i may include that as a commit in the PR, but I'll probably change it to the easy way afterward (in the same PR).
snarfed
on 29 Aug 2019
Closing due to inactivity. Thank you all for your inputs!
corradio
on 6 Jan 2020
Related issues
wojciej
·
3Comments
systemcatch
·
3Comments
ghost
·
5Comments
alixunderplatz
·
3Comments
jiwidi
·
4Comments
Most helpful comment
ok, i have a rough approximation of SEC by Florida counties that only adds ~4KB to
world.json. i'll clean up the geo code and add it to this PR.i'd also like to hide the county boundaries inside the SEC, but i don't know how high priority that will be. we'll see.