Merge pull request #840 from openego/feature/#797-move_function_assig…

…n_busID

Move and merge the two assigne_gas_bus_id functions

CHANGELOG.rst

@@ -379,6 +379,8 @@ Changed
   `#787 <https://github.com/openego/eGon-data/issues/787>`_
 * Update pypsa-eur-sec fork and store national demand time series
   `#402 <https://github.com/openego/eGon-data/issues/402>`_
+* Move and merge the two assign_gas_bus_id functions to a central place
+  `#797 <https://github.com/openego/eGon-data/issues/797>`_
 * Add coordinates to non AC buses abroad in eGon100RE
   `#803 <https://github.com/openego/eGon-data/issues/803>`_
 

src/egon/data/datasets/ch4_prod.py

@@ -232,44 +232,6 @@ def load_biogas_generators(scn_name):
     return biogas_generators_list
 
 
-def assign_bus_id(dataframe, scn_name, carrier):
-    """Assigns bus_ids (for H2 buses) to points (contained in a dataframe) according to location
-
-    Parameters
-    ----------
-    dataframe : pandas.DataFrame
-        DataFrame cointaining points
-    scn_name : str
-        Name of the scenario
-    carrier : str
-        Name of the carrier
-
-    Returns
-    -------
-    res : pandas.DataFrame
-        Dataframe including bus_id
-    """
-
-    voronoi = db.select_geodataframe(
-        f"""
-        SELECT bus_id, geom FROM grid.egon_gas_voronoi
-        WHERE scn_name = '{scn_name}' AND carrier = '{carrier}';
-        """,
-        epsg=4326,
-    )
-
-    res = gpd.sjoin(dataframe, voronoi)
-    res["bus"] = res["bus_id"]
-    res = res.drop(columns=["index_right"])
-
-    # Assert that all power plants have a bus_id
-    assert (
-        res.bus.notnull().all()
-    ), f"Some points are not attached to a {carrier} bus."
-
-    return res
-
-
 def import_gas_generators(scn_name="eGon2035"):
     """Insert list of gas production units in database
 
@@ -306,7 +268,9 @@ def import_gas_generators(scn_name="eGon2035"):
     CH4_generators_list = CH4_generators_list.assign(**c)
 
     # Match to associated CH4 bus
-    CH4_generators_list = assign_bus_id(CH4_generators_list, scn_name, "CH4")
+    CH4_generators_list = db.assign_gas_bus_id(
+        CH4_generators_list, scn_name, "CH4"
+    )
 
     # Remove useless columns
     CH4_generators_list = CH4_generators_list.drop(columns=["geom", "bus_id"])

src/egon/data/datasets/ch4_storages.py

@@ -13,7 +13,6 @@
 from egon.data import config, db
 from egon.data.config import settings
 from egon.data.datasets import Dataset
-from egon.data.datasets.ch4_prod import assign_bus_id
 from egon.data.datasets.gas_grid import (
     ch4_nodes_number_G,
     define_gas_nodes_list,
@@ -125,7 +124,9 @@ def import_installed_ch4_storages(scn_name):
 
     # Match to associated gas bus
     Gas_storages_list = Gas_storages_list.reset_index(drop=True)
-    Gas_storages_list = assign_bus_id(Gas_storages_list, scn_name, "CH4")
+    Gas_storages_list = db.assign_gas_bus_id(
+        Gas_storages_list, scn_name, "CH4"
+    )
 
     # Add missing columns
     c = {"scn_name": scn_name, "carrier": "CH4"}

src/egon/data/datasets/chp/match_nep.py

@@ -10,7 +10,6 @@
 from egon.data.datasets.chp.small_chp import assign_use_case
 from egon.data.datasets.power_plants import (
     assign_bus_id,
-    assign_gas_bus_id,
     assign_voltage_level,
     filter_mastr_geometry,
     select_target,
@@ -503,7 +502,9 @@ def insert_large_chp(sources, target, EgonChp):
     ).bus_id
 
     # Assign gas bus_id
-    insert_chp["gas_bus_id"] = assign_gas_bus_id(insert_chp_c).gas_bus_id
+    insert_chp["gas_bus_id"] = db.assign_gas_bus_id(
+        insert_chp_c, "eGon2035", "CH4"
+    ).bus
 
     insert_chp = assign_use_case(insert_chp, sources)
 

src/egon/data/datasets/chp/small_chp.py

@@ -8,7 +8,6 @@
 from egon.data import config, db
 from egon.data.datasets.power_plants import (
     assign_bus_id,
-    assign_gas_bus_id,
     filter_mastr_geometry,
     select_target,
 )
@@ -87,7 +86,9 @@ def existing_chp_smaller_10mw(sources, MaStR_konv, EgonChp):
 
         # Assign gas bus_id
         mastr_chp_c = mastr_chp.copy()
-        mastr_chp["gas_bus_id"] = assign_gas_bus_id(mastr_chp_c).gas_bus_id
+        mastr_chp["gas_bus_id"] = db.assign_gas_bus_id(
+            mastr_chp_c, "eGon2035", "CH4"
+        ).bus
 
         # Assign bus_id
         mastr_chp["bus_id"] = assign_bus_id(
@@ -175,9 +176,9 @@ def extension_to_areas(
 
             selected_areas = selected_areas.to_crs(4326)
             # Assign gas bus_id
-            selected_areas["gas_bus_id"] = assign_gas_bus_id(
-                selected_areas.copy()
-            ).gas_bus_id
+            selected_areas["gas_bus_id"] = db.assign_gas_bus_id(
+                selected_areas.copy(), "eGon2035", "CH4"
+            ).bus
 
             # Select randomly one area from the list of possible areas
             # weighted by the share of demand

src/egon/data/datasets/hydrogen_etrago/h2_grid.py

@@ -5,7 +5,6 @@
 import geopandas as gpd
 
 from egon.data import db
-from egon.data.datasets.ch4_prod import assign_bus_id
 from egon.data.datasets.etrago_setup import link_geom_from_buses
 from egon.data.datasets.scenario_parameters import get_sector_parameters
 
@@ -74,7 +73,7 @@ def insert_h2_pipelines():
     new_pipelines.set_crs(epsg=4326, inplace=True)
 
     # find bus in H2_grid voronoi
-    new_pipelines = assign_bus_id(new_pipelines, "eGon2035", "H2_grid")
+    new_pipelines = db.assign_gas_bus_id(new_pipelines, "eGon2035", "H2_grid")
     new_pipelines = new_pipelines.rename(columns={"bus_id": "bus1"}).drop(
         columns=["bus"]
     )

src/egon/data/datasets/industrial_gas_demand.py

@@ -14,7 +14,6 @@
 from egon.data import db
 from egon.data.config import settings
 from egon.data.datasets import Dataset
-from egon.data.datasets.ch4_prod import assign_bus_id
 from egon.data.datasets.etrago_helpers import (
     finalize_bus_insertion,
     initialise_bus_insertion,
@@ -169,7 +168,7 @@ def read_industrial_CH4_demand(scn_name="eGon2035"):
     industrial_loads_list = read_industrial_demand(scn_name, carrier)
 
     # Match to associated gas bus
-    industrial_loads_list = assign_bus_id(
+    industrial_loads_list = db.assign_gas_bus_id(
         industrial_loads_list, scn_name, carrier
     )
 
@@ -201,10 +200,10 @@ def read_industrial_H2_demand(scn_name="eGon2035"):
     industrial_loads_list = read_industrial_demand(scn_name, "H2")
     industrial_loads_list_copy = industrial_loads_list.copy()
     # Match to associated gas bus
-    industrial_loads_list = assign_bus_id(
+    industrial_loads_list = db.assign_gas_bus_id(
         industrial_loads_list, scn_name, "H2_grid"
     )
-    industrial_loads_saltcavern = assign_bus_id(
+    industrial_loads_saltcavern = db.assign_gas_bus_id(
         industrial_loads_list_copy, scn_name, "H2_saltcavern"
     )
 

src/egon/data/datasets/power_plants/__init__.py

@@ -537,39 +537,6 @@ def assign_bus_id(power_plants, cfg):
     return power_plants
 
 
-def assign_gas_bus_id(power_plants):
-    """Assigns gas_bus_ids to power plants according to location
-
-    Parameters
-    ----------
-    power_plants : pandas.DataFrame
-        Power plants (including voltage level)
-
-    Returns
-    -------
-    power_plants : pandas.DataFrame
-        Power plants (including voltage level) and gas_bus_id
-
-    """
-
-    gas_voronoi = db.select_geodataframe(
-        """
-        SELECT * FROM grid.egon_gas_voronoi WHERE scn_name = 'eGon2035' AND
-        carrier = 'CH4'
-        """,
-        epsg=4326,
-    )
-
-    res = gpd.sjoin(power_plants, gas_voronoi)
-    res["gas_bus_id"] = res["bus_id"]
-
-    # Assert that all power plants have a gas_bus_id
-    assert res.gas_bus_id.notnull().all(), f"""Some power plants are
-    not attached to a gas bus: {res[res.gas_bus_id.isnull()]}"""
-
-    return res
-
-
 def insert_hydro_biomass():
     """Insert hydro and biomass power plants in database
 

src/egon/data/datasets/storages/pumped_hydro.py

@@ -2,26 +2,25 @@
 The module containing code allocating pumped hydro plants based on
 data from MaStR and NEP.
 """
-
-import pandas as pd
+
+from geopy.geocoders import Nominatim
+from sqlalchemy.orm import sessionmaker
 import geopandas as gpd
-import egon.data.config
-from egon.data import db, config
+import pandas as pd
+
+from egon.data import db
+from egon.data.datasets.chp.match_nep import match_nep_chp
+from egon.data.datasets.chp.small_chp import assign_use_case
 from egon.data.datasets.power_plants import (
-    assign_voltage_level,
     assign_bus_id,
-    assign_gas_bus_id,
+    assign_voltage_level,
     filter_mastr_geometry,
     select_target,
 )
-from egon.data.datasets.chp.match_nep import match_nep_chp
-from egon.data.datasets.chp.small_chp import assign_use_case
-from geopy.geocoders import Nominatim
-from sqlalchemy.orm import sessionmaker
 
 
 def select_nep_pumped_hydro():
-    """ Select pumped hydro plants from NEP power plants list
+    """Select pumped hydro plants from NEP power plants list
 
 
     Returns
@@ -58,7 +57,7 @@ def select_nep_pumped_hydro():
 
 
 def select_mastr_pumped_hydro():
-    """ Select pumped hydro plants from MaStR
+    """Select pumped hydro plants from MaStR
 
 
     Returns
@@ -140,7 +139,7 @@ def match_storage_units(
     consider_carrier=True,
     consider_capacity=True,
 ):
-    """ Match storage_units (in this case only pumped hydro) from MaStR
+    """Match storage_units (in this case only pumped hydro) from MaStR
     to list of power plants from NEP
 
     Parameters
@@ -255,7 +254,7 @@ def match_storage_units(
 
 
 def get_location(unmatched):
-    """ Gets a geolocation for units which couldn't be matched using MaStR data.
+    """Gets a geolocation for units which couldn't be matched using MaStR data.
     Uses geolocator and the city name from NEP data to create longitude and
     latitude for a list of unmatched units.
 

src/egon/data/db.py

@@ -3,7 +3,7 @@
 import functools
 import time
 
-from psycopg2.errors import UniqueViolation, DeadlockDetected
+from psycopg2.errors import DeadlockDetected, UniqueViolation
 from sqlalchemy import create_engine, text
 from sqlalchemy.exc import IntegrityError, OperationalError
 from sqlalchemy.orm import sessionmaker
@@ -163,7 +163,7 @@ def wrapped(*xs, **ks):
 
 
 def select_dataframe(sql, index_col=None, warning=True):
-    """ Select data from local database as pandas.DataFrame
+    """Select data from local database as pandas.DataFrame
 
     Parameters
     ----------
@@ -351,3 +351,41 @@ def commit(*args, **kwargs):
         return ret
 
     return commit
+
+
+def assign_gas_bus_id(dataframe, scn_name, carrier):
+    """Assigns bus_ids to points (contained in a dataframe) according to location
+
+    Parameters
+    ----------
+    dataframe : pandas.DataFrame
+        DataFrame cointaining points
+    scn_name : str
+        Name of the scenario
+    carrier : str
+        Name of the carrier
+
+    Returns
+    -------
+    res : pandas.DataFrame
+        Dataframe including bus_id
+    """
+
+    voronoi = select_geodataframe(
+        f"""
+        SELECT bus_id, geom FROM grid.egon_gas_voronoi
+        WHERE scn_name = '{scn_name}' AND carrier = '{carrier}';
+        """,
+        epsg=4326,
+    )
+
+    res = gpd.sjoin(dataframe, voronoi)
+    res["bus"] = res["bus_id"]
+    res = res.drop(columns=["index_right"])
+
+    # Assert that all power plants have a bus_id
+    assert (
+        res.bus.notnull().all()
+    ), f"Some points are not attached to a {carrier} bus."
+
+    return res