Source code for message_ix_models.model.material.data_methanol

from ast import literal_eval
from typing import TYPE_CHECKING

import pandas as pd
import yaml
from message_ix import make_df

import message_ix_models.util
from message_ix_models.model.material.material_demand import material_demand_calc
from message_ix_models.model.material.util import read_config
from message_ix_models.util import broadcast, same_node

if TYPE_CHECKING:
    from message_ix import Scenario

ssp_mode_map = {
    "SSP1": "CTS core",
    "SSP2": "RTS core",
    "SSP3": "RTS high",
    "SSP4": "CTS high",
    "SSP5": "RTS high",
    "LED": "CTS core",  # TODO: move to even lower projection
}

iea_elasticity_map = {
    "CTS core": (1.2, 0.25),
    "CTS high": (1.3, 0.48),
    "RTS core": (1.25, 0.35),
    "RTS high": (1.4, 0.54),
}


[docs]def gen_data_methanol(scenario: "Scenario") -> dict[str, pd.DataFrame]: """ Generates data for methanol industry model Parameters ---------- scenario: .Scenario """ context = read_config() df_pars = pd.read_excel( message_ix_models.util.package_data_path( "material", "methanol", "methanol_sensitivity_pars.xlsx" ), sheet_name="Sheet1", dtype=object, ) pars = df_pars.set_index("par").to_dict()["value"] if pars["mtbe_scenario"] == "phase-out": pars_dict = pd.read_excel( message_ix_models.util.package_data_path( "material", "methanol", "methanol_techno_economic.xlsx" ), sheet_name=None, dtype=object, ) else: pars_dict = pd.read_excel( message_ix_models.util.package_data_path( "material", "methanol", "methanol_techno_economic_high_demand.xlsx" ), sheet_name=None, dtype=object, ) for i in pars_dict.keys(): pars_dict[i] = unpivot_input_data(pars_dict[i], i) # TODO: only temporary hack to ensure SSP_dev compatibility if "SSP_dev" in scenario.model: file_path = message_ix_models.util.package_data_path( "material", "methanol", "missing_rels.yaml" ) with open(file_path, "r") as file: missing_rels = yaml.safe_load(file) df = pars_dict["relation_activity"] pars_dict["relation_activity"] = df[~df["relation"].isin(missing_rels)] default_gdp_elasticity_2020, default_gdp_elasticity_2030 = iea_elasticity_map[ ssp_mode_map[context["ssp"]] ] df_final = material_demand_calc.gen_demand_petro( scenario, "methanol", default_gdp_elasticity_2020, default_gdp_elasticity_2030 ) df_final["value"] = df_final["value"].apply( lambda x: x * pars["methanol_resid_demand_share"] ) pars_dict["demand"] = df_final return pars_dict
[docs]def broadcast_nodes( df_bc_node: pd.DataFrame, df_final: pd.DataFrame, node_cols: list[str], node_cols_codes: dict[str, pd.Series], i: int, ) -> pd.DataFrame: """ Broadcast nodes that were stored in pivoted row Parameters ---------- df_bc_node: pd.DataFrame df_final: pd.DataFrame node_cols: list[str] node_cols_codes: dict[str, pd.Series] i: int """ if len(node_cols) == 1: if "node_loc" in node_cols: df_bc_node = df_bc_node.pipe( broadcast, node_loc=node_cols_codes["node_loc"] ) if "node_vtg" in node_cols: df_bc_node = df_bc_node.pipe( broadcast, node_vtg=node_cols_codes["node_vtg"] ) if "node_rel" in node_cols: df_bc_node = df_bc_node.pipe( broadcast, node_rel=node_cols_codes["node_rel"] ) if "node" in node_cols: df_bc_node = df_bc_node.pipe(broadcast, node=node_cols_codes["node"]) if "node_share" in node_cols: df_bc_node = df_bc_node.pipe( broadcast, node_share=node_cols_codes["node_share"] ) else: df_bc_node = df_bc_node.pipe(broadcast, node_loc=node_cols_codes["node_loc"]) if len(df_final.loc[i][node_cols].T.unique()) == 1: # df_bc_node["node_rel"] = df_bc_node["node_loc"] df_bc_node = df_bc_node.pipe( same_node ) # not working for node_rel in installed message_ix_models version else: if "node_rel" in list(df_bc_node.columns): df_bc_node = df_bc_node.pipe( broadcast, node_rel=node_cols_codes["node_rel"] ) if "node_origin" in list(df_bc_node.columns): df_bc_node = df_bc_node.pipe( broadcast, node_origin=node_cols_codes["node_origin"] ) if "node_dest" in list(df_bc_node.columns): df_bc_node = df_bc_node.pipe( broadcast, node_dest=node_cols_codes["node_dest"] ) return df_bc_node
[docs]def broadcast_years( df_bc_node: pd.DataFrame, yr_col_out: list[str], yr_cols_codes: dict[str, list[str]], col: str, ) -> pd.DataFrame: """ Broadcast years that were stored in pivoted row Parameters ---------- df_bc_node: pd.DataFrame yr_col_out: list[str] yr_cols_codes: ict[str, list[str]] col: str """ if len(yr_col_out) == 1: yr_list = [i[0] for i in yr_cols_codes[col]] # print(yr_list) if "year_act" in yr_col_out: df_bc_node = df_bc_node.pipe(broadcast, year_act=yr_list) if "year_vtg" in yr_col_out: df_bc_node = df_bc_node.pipe(broadcast, year_vtg=yr_list) if "year_rel" in yr_col_out: df_bc_node = df_bc_node.pipe(broadcast, year_rel=yr_list) if "year" in yr_col_out: df_bc_node = df_bc_node.pipe(broadcast, year=yr_list) df_bc_node[yr_col_out] = df_bc_node[yr_col_out].astype(int) else: if "year_vtg" in yr_col_out: y_v = [str(i) for i in yr_cols_codes[col]] df_bc_node = df_bc_node.pipe(broadcast, year_vtg=y_v) df_bc_node["year_act"] = [ literal_eval(i)[1] for i in df_bc_node["year_vtg"] ] df_bc_node["year_vtg"] = [ literal_eval(i)[0] for i in df_bc_node["year_vtg"] ] if "year_rel" in yr_col_out: if "year_act" in yr_col_out: df_bc_node = df_bc_node.pipe( broadcast, year_act=[i[0] for i in yr_cols_codes[col]] ) df_bc_node["year_rel"] = df_bc_node["year_act"] return df_bc_node
[docs]def unpivot_input_data(df: pd.DataFrame, par_name: str) -> pd.DataFrame: """ Unpivot data that is already contains columns for respective MESSAGEix parameter Parameters ---------- df: pd.DataFrame DataFrame containing parameter data with year and node values pivoted par_name: str name of MESSAGEix parameter """ df_final = df df_final_full = pd.DataFrame() for i in df_final.index: # parse strings of node columns to dictionary node_cols = [i for i in df_final.columns if "node" in i] remove = ["'", "[", "]", " "] node_cols_codes = {} for col in node_cols: node_cols_codes[col] = pd.Series( "".join(x for x in df_final.loc[i][col] if x not in remove).split(",") ) # create dataframe with required columns df_bc_node = make_df(par_name, **df_final.loc[i]) # collect year values from year columns yr_cols_codes = {} yr_col_inp = [i for i in df_final.columns if "year" in i] yr_col_out = [i for i in df_bc_node.columns if "year" in i] df_bc_node[yr_col_inp] = df_final.loc[i][yr_col_inp].values # broadcast in node dimensions for colname in node_cols: df_bc_node[colname] = None df_bc_node = broadcast_nodes( df_bc_node, df_final, node_cols, node_cols_codes, i ) # brodcast in year dimensions for col in yr_col_inp: yr_cols_codes[col] = literal_eval(df_bc_node[col].values[0]) df_bc_node = broadcast_years(df_bc_node, yr_col_out, yr_cols_codes, col) df_bc_node[yr_col_out] = df_bc_node[yr_col_out].astype(int) df_final_full = pd.concat([df_final_full, df_bc_node]) df_final_full = df_final_full.drop_duplicates().reset_index(drop=True) # special treatment for relation_activity dataframes: # relations parametrization should only contain columns where node_rel == node_loc # except for relations acting on a geographic global level # (where node_rel == "R**_GLB") if par_name == "relation_activity": df_final_full = df_final_full.drop( df_final_full[ (df_final_full.node_rel.values != "R12_GLB") & (df_final_full.node_rel.values != df_final_full.node_loc.values) ].index ) return make_df(par_name, **df_final_full)