Add updated aluminum capacity calibration

Author: macflo8

message_ix_models/data/material/aluminum/aluminum_techno_economic.xlsx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e203bd3b535e8494767e6301fa87d11eb3d8d19de2139f27108a04bcd569f10c
-size 121450
+oid sha256:2144ce3f352e6bca67ed46467dd656615abdfb8bdba8548862e32bea119f4448
+size 58799

message_ix_models/data/material/aluminum/archived/aluminum_techno_economic.xlsx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:087e3fad92e1c9fcf292ea606e3d2ccf510e773bad478a9a650a2e48661267ca
+size 121353

message_ix_models/data/material/aluminum/smelters-with 2022 projection.xls

@@ -15,7 +15,13 @@
 
 from .data_util import read_rel, read_timeseries
 from .material_demand import material_demand_calc
-from .util import combine_df_dictionaries, get_ssp_from_context, read_config
+from .util import (
+    add_R12_column,
+    combine_df_dictionaries,
+    get_pycountry_iso,
+    get_ssp_from_context,
+    read_config,
+)
 
 
 def read_data_aluminum(
@@ -733,3 +739,94 @@ def gen_data_alu_trade(scenario: message_ix.Scenario) -> dict[str, pd.DataFrame]
     results["bound_activity_lo"].append(bound_act_net_export_chn_2025)
 
     return {par_name: pd.concat(dfs) for par_name, dfs in results.items()}
+
+
+def gen_hist_new_cap():
+    df_cap = pd.read_excel(
+        package_data_path("material", "aluminum", "smelters-with 2022 projection.xls"),
+        sheet_name="Sheet1",
+        skipfooter=23,
+    ).rename(columns={"Unnamed: 0": "Country", "Unnamed: 1": "Region"})
+    df_cap.Technology = df_cap.Technology.fillna("unknown")
+    df_cap = df_cap[~df_cap[1995].isna()]
+    df_cap.Country = df_cap["Country"].ffill()
+    df_cap["ISO"] = df_cap.Country.apply(
+        lambda c: get_pycountry_iso(
+            c,
+            {
+                "Surinam": "SUR",
+                "Trinidad": "TTO",
+                "Quatar": "QAT",
+                "Turkey": "TUR",
+                "UAE": "ARE",
+                "Gernamy": "DEU",
+                "Azerbaydzhan": "AZE",
+                "Russia": "RUS",
+                "Tadzhikistan": "TJK",
+                "UK": "GBR",
+                "Total": "World",
+                "Bosnia": "BIH",
+            },
+        )
+    )
+    df_cap = add_R12_column(
+        df_cap, file_path=package_data_path("node", "R12.yaml"), iso_column="ISO"
+    )
+
+    # generate historical_new_capacity for soderberg
+    df_cap_ss = df_cap[
+        (df_cap.Technology.str.contains("SS") & ~(df_cap.Technology.str.contains("PB")))
+        & ~(df_cap.Technology.str.contains("HAL"))
+        & ~(df_cap.Technology.str.contains("P"))
+    ]
+    df_cap_ss_r12 = df_cap_ss.groupby("R12").sum(numeric_only=True)
+    sample = df_cap_ss_r12[df_cap_ss_r12[df_cap_ss_r12.columns[-1]] != 0][
+        [i for i in range(1995, 2020, 5)] + [2019]
+    ]
+    hist_new_cap_ss = compute_differences(sample, 1995) / 10**6
+    hist_new_cap_ss = hist_new_cap_ss.rename(columns={2019: 2020})
+    hist_new_cap_ss = (
+        hist_new_cap_ss.reset_index()
+        .melt(id_vars="R12", var_name="year_vtg")
+        .assign(unit="Mt", technology="soderberg_aluminum")
+        .rename(columns={"R12": "node_loc"})
+    )
+
+    # generate historical_new_capacity for prebake
+    df_cap_pb = df_cap.loc[df_cap.index.difference(df_cap_ss.index)]
+    df_cap_pb_r12 = df_cap_pb.groupby("R12").sum(numeric_only=True)
+    sample = df_cap_pb_r12[[i for i in range(1995, 2020, 5)] + [2019]]
+    hist_new_cap_pb = compute_differences(sample, 1995) / 10**6
+    hist_new_cap_pb = hist_new_cap_pb.rename(columns={2019: 2020})
+    hist_new_cap_pb = (
+        hist_new_cap_pb.reset_index()
+        .melt(id_vars="R12", var_name="year_vtg")
+        .assign(unit="Mt", technology="prebake_aluminum")
+        .rename(columns={"R12": "node_loc"})
+    )
+
+    return {"historical_new_capacity": pd.concat([hist_new_cap_ss, hist_new_cap_pb])}
+
+
+def compute_differences(df, ref_col):
+    # Initialize a DataFrame to store differences
+    differences = df[ref_col].to_frame()
+
+    # Start with the reference column
+    ref_values = df[ref_col].copy()
+
+    for col in df.columns:
+        if col == ref_col:
+            continue  # Skip the reference column
+
+        # Compute differences
+        diff = df[col] - ref_values
+        diff[diff <= 0] = 0  # Keep only positive differences
+
+        # Store differences
+        differences[col] = diff
+
+        # Update the reference column where the current column is greater
+        ref_values = ref_values.where(df[col] <= ref_values, df[col])
+
+    return differences

message_ix_models/model/material/data_aluminum.py

@@ -5,6 +5,7 @@
 import message_ix
 import openpyxl as pxl
 import pandas as pd
+import pycountry
 import yaml
 from scipy.optimize import curve_fit
 
@@ -394,3 +395,36 @@ def path_fallback(context_or_regions: Union[Context, str], *parts) -> Path:
             return c
 
     raise FileNotFoundError(candidates)
+
+
+def get_pycountry_iso(row, mis_dict):
+    try:
+        row = pycountry.countries.lookup(row).alpha_3
+    except LookupError:
+        try:
+            row = mis_dict[row]
+        except KeyError:
+            print(f"{row} is not mapped to an ISO")
+            row = None
+    return row
+
+
+def get_r12_reg(df, r12_map_inv, col_name):
+    try:
+        df = r12_map_inv[df[col_name]]
+    except KeyError:
+        df = None
+    return df
+
+
+def add_R12_column(df, file_path, iso_column="COUNTRY"):
+    # Replace 'your_file_path.yaml' with the path to your actual YAML file
+    # file_path = private_data_path("node", "R12_SSP_V1.yaml")
+    yaml_data = read_yaml_file(file_path)
+    yaml_data.pop("World")
+
+    r12_map = {k: v["child"] for k, v in yaml_data.items()}
+    r12_map_inv = {k: v[0] for k, v in invert_dictionary(r12_map).items()}
+
+    df["R12"] = df.apply(lambda x: get_r12_reg(x, r12_map_inv, iso_column), axis=1)
+    return df