Fix bug preventing add_energy_storage_database to run through

doc/release_notes.md

@@ -26,6 +26,8 @@
 
 * Correct the nuclear fuel price (`nuclear`, `uranium`) using Lazard's LCOE+ assumption (<https://github.com/PyPSA/technology-data/pull/291>)
 
+* Fixed bug preventing add_energy_storage_database to run through (https://github.com/PyPSA/technology-data/pull/286).
+
 ## [v0.15.0](https://github.com/PyPSA/technology-data/releases/tag/v0.15.0) (9th June 2026)
 
 **Breaking changes**

scripts/compile_cost_assumptions.py

@@ -3809,19 +3809,29 @@ def add_energy_storage_database(
         engine="calamine",
     )
     df = df.drop(columns=["ref_size_MW", "EP_ratio_h"])
-    df = df.fillna(df.dtypes.replace({"float64": 0.0, "O": "NULL"}))
+    # Fill by column kind, not dtype name: text columns are "object" in pandas 2 but "str" in pandas 3
+    df = df.fillna({col: 0.0 for col in df.select_dtypes("number").columns}).fillna(
+        "NULL"
+    )
     df.loc[:, "unit"] = df.unit.str.replace("NULL", "per unit")
+    df.loc[(df["parameter"] == "efficiency") & (df["unit"].isna()), "unit"] = "per unit"
 
     # b) Change data to PyPSA format (aggregation of components, units, currency, etc.)
     df = clean_up_units(df, "value")  # base clean up
 
     # rewrite technology to be charger, store, discharger, bidirectional-charger
     df.loc[:, "carrier"] = df.carrier.str.replace("NULL", "")
-    df.loc[:, "carrier"] = df["carrier"].apply(lambda x: x.split("-"))
-    carrier_list_len = df["carrier"].apply(lambda x: len(x))
-    carrier_str_len = df["carrier"].apply(lambda x: len(x[0]))
-    carrier_first_item = df["carrier"].apply(lambda x: x[0])
-    carrier_last_item = df["carrier"].apply(lambda x: x[-1])
+    df["carrier"] = (
+        df["carrier"]
+        .fillna("")
+        .astype(str)
+        .str.strip()
+        .apply(lambda x: x.split("-") if x else [])
+    )
+    carrier_list_len = df["carrier"].apply(len)
+    carrier_str_len = df["carrier"].apply(lambda x: len(x[0]) if len(x) > 0 else 0)
+    carrier_first_item = df["carrier"].apply(lambda x: x[0] if len(x) > 0 else "")
+    carrier_last_item = df["carrier"].apply(lambda x: x[-1] if len(x) > 0 else "")
     bicharger_filter = carrier_list_len == 3
     charger_filter = (carrier_list_len == 2) & (carrier_first_item == "elec")
     discharger_filter = (carrier_list_len == 2) & (carrier_last_item == "elec")
@@ -3942,7 +3952,8 @@ def add_energy_storage_database(
                     or tech_name == "Pumped-Heat-store"
                 ):
                     x1 = pd.concat(
-                        [x, pd.DataFrame(other_segments_points)], ignore_index=True
+                        [x.reset_index(drop=True), pd.Series(other_segments_points)],
+                        ignore_index=True,
                     )
                     y1 = y
                     factor = 5
@@ -3955,7 +3966,7 @@ def add_energy_storage_database(
                             number_of_terms=i + 1,
                         )
                         y1 = pd.concat(
-                            [y1, pd.DataFrame([cost_at_year])], ignore_index=True
+                            [y1, pd.Series([cost_at_year])], ignore_index=True
                         )
                     f = interpolate.interp1d(
                         x1.squeeze(),
@@ -3965,7 +3976,8 @@ def add_energy_storage_database(
                     )
                 elif tech_name == "Hydrogen-charger":
                     x2 = pd.concat(
-                        [x, pd.DataFrame(other_segments_points)], ignore_index=True
+                        [x.reset_index(drop=True), pd.Series(other_segments_points)],
+                        ignore_index=True,
                     )
                     y2 = y
                     factor = 6.5
@@ -3976,7 +3988,7 @@ def add_energy_storage_database(
                             number_of_terms=i + 1,
                         )
                         y2 = pd.concat(
-                            [y2, pd.DataFrame([cost_at_year])], ignore_index=True
+                            [y2, pd.Series([cost_at_year])], ignore_index=True
                         )
                     f = interpolate.interp1d(
                         x2.squeeze(),
@@ -3986,7 +3998,8 @@ def add_energy_storage_database(
                     )
                 else:
                     x3 = pd.concat(
-                        [x, pd.DataFrame(other_segments_points)], ignore_index=True
+                        [x.reset_index(drop=True), pd.Series(other_segments_points)],
+                        ignore_index=True,
                     )
                     y3 = y
                     factor = 2
@@ -3997,7 +4010,7 @@ def add_energy_storage_database(
                             number_of_terms=i + 1,
                         )
                         y3 = pd.concat(
-                            [y3, pd.DataFrame([cost_at_year])], ignore_index=True
+                            [y3, pd.Series([cost_at_year])], ignore_index=True
                         )
                     f = interpolate.interp1d(
                         x3.squeeze(),
@@ -4043,7 +4056,10 @@ def add_energy_storage_database(
                 df = pd.concat([df, df_new], ignore_index=True)
 
     # d) Combine metadata and add to cost database
+    df["source"] = df["source"].fillna("").astype(str)
+    df["reference"] = df["reference"].fillna("").astype(str)
     df.loc[:, "source"] = df["source"] + ", " + df["reference"]
+    df["note"] = df["note"].fillna("NULL").astype(str)
     for i in df.index:
         df.loc[i, "further description"] = str(
             {