Merge pull request #218 from Breakthrough-Energy/daniel/hifld_create_…

…transformers

feat: add function to create transformers within substations

prereise/gather/griddata/hifld/data_process/tests/test_transmission.py

@@ -5,6 +5,7 @@
 from prereise.gather.griddata.hifld.data_process.transmission import (
     augment_line_voltages,
     create_buses,
+    create_transformers,
 )
 
 
@@ -100,3 +101,18 @@ def test_create_buses():
     expected_return["baseKV"] = expected_return["baseKV"].astype(float)
     bus = create_buses(lines)
     assert_frame_equal(bus, expected_return)
+
+
+def test_create_transformers():
+    bus = pd.DataFrame(
+        {
+            "sub_id": [1, 1, 2, 2, 3, 3, 3, 4],
+            "baseKV": [69, 345, 69, 115, 115, 230, 345, 230],
+        },
+        dtype="float",
+    )
+    expected_transformers = pd.DataFrame(
+        {"from_bus_id": [0, 2, 4, 5], "to_bus_id": [1, 3, 6, 6]}
+    )
+    transformers = create_transformers(bus)
+    assert_frame_equal(transformers, expected_transformers)

prereise/gather/griddata/hifld/data_process/transmission.py

@@ -303,6 +303,25 @@ def create_buses(lines):
     return buses
 
 
+def create_transformers(bus):
+    """Add transformers between buses within the same substation. The assumed topology
+    is that the highest-voltage bus in each substation is connected via a tansformer to
+    every other voltage.
+
+    :param pandas.DataFrame bus: columns 'sub_id' and 'baseKV'
+    :return: (*pandas.DataFrame*) -- each row is one transformer, columns are
+        ["from_bus_id", "to_bus_id"].
+    """
+    bus_pairs = [
+        (b, volt_series.idxmax())
+        for sub, volt_series in bus.groupby("sub_id")["baseKV"]
+        for b in volt_series.sort_values().index[:-1]
+        if len(volt_series) > 1
+    ]
+
+    return pd.DataFrame(bus_pairs, columns=["from_bus_id", "to_bus_id"])
+
+
 def build_transmission():
     """Main user-facing entry point."""
     # Load input data