Talk:IPRPraha Powerstations Import

From OpenStreetMap Wiki
Jump to navigation Jump to search

Minor tagging comments

This is a great job you've done to import all that substations! I surely come too late here but there are 2 little points about the tagging you use I want to make, not big problems anyway :

  • transformer=* shouldn't be used in combination with building=* areas. Transformers are machines inside substation buildings, a node for each of them inside the building polygon is more suitable.
  • Cabinet substations are missing a man_made=street_cabinet in association with street_cabinet=*.

It depends on how precise the operator's datasets are but I think transformers may be simply removed from building and indoor substations. It's not unusual to find several transformers (with different figures) inside the same building. Let me know what you think Fanfouer (talk) 18:24, 4 April 2019 (UTC)

r00t: Well, ideally that would be true. But except for large substations, where you can see transformers on the photo map, there is no way of telling how and where are transformers inside the building and how many. Most small substation buildings simply have one transformer inside, same for the cabinets. I know all of these stations do contain the transformer(s), so simply adding transformer=* to the building seems to be good enough simplification that preserves this information. For large substations where transformers are mapped separately, I have already done it as you suggested, mapping them separately where they are.

Regarding street cabinets, all cabinets should already have man_made=street_cabinet, example:

The import is done, but I'm currently doing a 2nd pass and doing a cleanup and verifying all tags are correct. If you have seen cabinet without man_made tag, I will probably fix it soon...

Ok for man_made=street_cabinet, that's good.
Regarding transformers, I use to not put transformer=* on substations even if there is one inside I don't know where exactly. There was a proposal last month which had been rejected, about this. Adding a key like transformation=* to say there are transformers, don't know how many, don't know where - is more meaningful than making the whole substation acting like a transformer.
For this big cabinet, I think there are two transformers inside, according T1-T2 label (just guessing). Then 2 nodes may be added inside of it, until I'm wrong. Fanfouer (talk) 19:53, 4 April 2019 (UTC)

r00t: I still think having to add extra nodes for every transformer in a cabinet would just pollute the map. In most cases, same information (number of transformers inside) could be conveyed by a single tag instead of additional nodes. This import have about 3500 power stations, most of them are minor ones with 1-3 transformers... so this would mean adding about 7000 extra nodes that would carry no additional information. And simply placing them "somewhere inside" goes against "what you see on the ground" mapping rule: You can maybe read number of transformers on the label, but you don't know where exactly are they inside unless you can walk into that building or open the cabinet to check or have some very detailed documentation. So appropriate tag would be something like transformer:count noting exactly the information that's visible from outside. Hopefully some better proposal is accepted soon and in that can I can easily re-tag all powerstations with a new scheme. But until then, I'm not going to pollute map with made up data that's not really needed.

It's devices=* which was reviewed for transformers on this proposal, if and only if transformers share the same figures and capabilities.
Mixing "what you see on the ground" and One_feature,_one_OSM_element principles encourage to use tags like transformation=* on the substation and put a different object for transformer devices once data will be available (so not now). I'm not asking for 7000 more objects with no additional information but to distinguish two situations on areas only: "most substations host transformers" (area power=substation + transformation=* (or like)) and "substations are actual transformers" (area power=substation + transformer=*). What is the need to deduce transformers from a substation dataset? If you get a transformer dataset later from power operators (agreed it's not planed nor mandatory), dedicated nodes will support many properties you won't be able to mix with substation attributes (rating, voltage, manufacturer...) Fanfouer (talk) 21:22, 4 April 2019 (UTC)