fixing breaking world bank country names and multiple small ones

rules/supply_curves.smk

@@ -35,13 +35,32 @@ def _product_uses_renewables(product):
     return True
 
 
+def _demand_levels_for_product(product):
+    """Demand levels to sweep when solving LCoX for a product.
+
+    Renewable-fed routes (e.g. hbi/dri) get costlier at scale as cheaper
+    resource classes are exhausted, so the full sweep is needed to trace
+    the curve. Grid-connected routes (e.g. steel/eaf) draw on grid
+    electricity at a fixed, uncapped marginal cost in this stage, so their
+    LCoX is flat with respect to volume (verified: identical across all
+    demand levels for every region). Solving all levels for those products
+    is redundant; one point at the top of the range is enough and keeps
+    capacity headroom for the trade model.
+    """
+    demand_levels = config.get("steel_demand_levels")
+    if _product_uses_renewables(product):
+        return demand_levels
+    return [max(demand_levels)]
+
+
 def _all_supply_curve_targets():
     targets = []
+    cost_year = config["trade_chains"].get("cost_year", 2050)
     for region in config["regions"]:
         wacc = config["trade_chains"].get("wacc", "uniform")
         for product in SUPPLY_CURVE_PRODUCTS:
             targets.append(
-                f"resources/supply_curves/cost_year~2050/wacc~{wacc}/{region}_marginal_cost_{product}.csv"
+                f"resources/supply_curves/cost_year~{cost_year}/wacc~{wacc}/{region}_marginal_cost_{product}.csv"
             )
     return targets
 
@@ -161,12 +180,12 @@ if config["enable"].get("run_supply_curve", True):
                 scenario=config.get("supply_curve", {}).get(
                     "default_scenario", "allocated_share"
                 ),
-                product_demand_mt=config.get("steel_demand_levels"),
+                product_demand_mt=_demand_levels_for_product(wildcards.product),
             ),
             lco_unreserved=lambda wildcards: (
                 expand(
                     f"resources/lco-{wildcards.product}/cost_year~{wildcards.cost_year}/wacc~{wildcards.wacc}/{wildcards.region}_unreserved/results_{{product_demand_mt}}.csv",
-                    product_demand_mt=config.get("steel_demand_levels"),
+                    product_demand_mt=_demand_levels_for_product(wildcards.product),
                 )
                 if config.get("supply_curve", {}).get("generate_unreserved", False)
                 and _product_uses_renewables(wildcards.product)

workflow/notebooks/prepare-wacc.ipynb

@@ -8,7 +8,6 @@
    "outputs": [],
    "source": [
     "import pandas as pd\n",
-    "import pycountry\n",
     "import wbdata"
    ]
   },
@@ -97,50 +96,20 @@
     "regions = config['regions']"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "a532a77c",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "country_name_corrections = {\n",
-    "    \"Democratic Republic of the Congo\": \"Congo, The Democratic Republic of the\",\n",
-    "    \"Republic of the Congo\": \"Republic of the Congo\",\n",
-    "    \"Kosovo\": \"Republic of Kosovo\",  # pycountry not supported\n",
-    "    \"Russia\": \"Russian Federation\",\n",
-    "    \"Turkey\": \"Türkiye\",\n",
-    "    \"Venezuela\": \"Venezuela, Bolivarian Republic of\",\n",
-    "    \"Tanzania\": \"United Republic of Tanzania\",\n",
-    "    \"Bolivia\": \"Plurinational State of Bolivia\",\n",
-    "    \"Vietnam\": \"Viet Nam\",\n",
-    "    \"South Korea\": \"Korea, Republic of\",\n",
-    "    \"North Korea\": \"Korea, Democratic People's Republic of\",\n",
-    "    \"Taiwan\": \"Taiwan, Province of China\",\n",
-    "    \"Laos\": \"Lao People's Democratic Republic\",\n",
-    "    \"Brunei\": \"Brunei Darussalam\",\n",
-    "    \"Equatorial French Guiana\": \"French Guiana\",\n",
-    "    \"Syria\": \"Syrian Arab Republic\",   \n",
-    "    \"Palestine\": \"Palestine, State of\",\n",
-    "    \"Moldova\": \"Republic of Moldova\",\n",
-    "    \"North Korea\": \"Korea, Democratic People's Republic of\",\n",
-    "}"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": null,
    "id": "7accdafc",
    "metadata": {},
    "outputs": [],
    "source": [
-    "# Build a mapping from country name to ISO-3 code (still needed for GDP World Bank lookup)\n",
-    "country_name_to_iso = {}\n",
-    "for country in pycountry.countries:\n",
-    "    country_name_to_iso[country.name] = country.alpha_3\n",
-    "    # Add common names\n",
-    "    if hasattr(country, 'official_name'):\n",
-    "        country_name_to_iso[country.official_name] = country.alpha_3\n",
+    "# Build a mapping from World Bank's own country name to ISO-3 code.\n",
+    "# Using World Bank's own metadata (rather than pycountry) avoids name-matching\n",
+    "# mismatches for countries whose WB display name differs from pycountry's\n",
+    "# (e.g. WB \"Turkiye\" vs pycountry \"Türkiye\", WB \"Korea, Rep.\" vs pycountry\n",
+    "# \"Korea, Republic of\"), which previously caused those countries' GDP rows to be\n",
+    "# dropped and, for single-country regions, produced NaN GDP-weighted WACC.\n",
+    "country_name_to_iso = {c[\"name\"]: c[\"id\"] for c in wbdata.get_countries()}\n",
     "\n",
     "# Build a mapping from ISO-3 code to region (config regions now use ISO-3 codes directly)\n",
     "iso_to_region = {}\n",

workflow/scripts/calculate_lcox.py

@@ -408,15 +408,15 @@ def solve_network(network, config):
         logger.error(f"Solver exception: {e}")
         raise
 
-    return network
+    return network, status_ok
 
 
 # ============================================================================
 # RESULTS EXTRACTION
 # ============================================================================
 
 
-def extract_lcox(network, product, demands):
+def extract_lcox(network, product, demands, status_ok=True):
     """Extract LCOX from an optimized network.
 
     Parameters
@@ -428,6 +428,13 @@ def extract_lcox(network, product, demands):
     demands : dict
         Must include key `'product_demand_mt'` (float, Mt/year) used to
         compute annual production and per-unit LCOX.
+    status_ok : bool
+        Whether the solver reported a genuinely optimal termination status.
+        A non-optimal solve (e.g. numerical trouble during the barrier
+        method) can leave a stale but finite `network.objective` behind even
+        though the result isn't trustworthy; checking only `None`/`NaN` lets
+        that garbage value through as if it were a valid LCOX. Requiring
+        `status_ok` closes that gap.
 
     Returns
     -------
@@ -455,6 +462,9 @@ def extract_lcox(network, product, demands):
     )
 
     try:
+        if not status_ok:
+            raise ValueError("Optimization did not report an optimal status")
+
         obj_value = network.objective
         if obj_value is None or np.isnan(obj_value):
             raise ValueError("Optimization failed to return valid objective")
@@ -598,16 +608,19 @@ def extract_lcox(network, product, demands):
     if snakemake.config.get("debug_network_inspection", False):
         inspect_network(network, product)  # Debug inspection
     try:
-        solve_network(network, snakemake.config)
+        network, status_ok = solve_network(network, snakemake.config)
         optimization_status = (
             "optimal"
-            if network.objective is not None and not np.isnan(network.objective)
+            if status_ok
+            and network.objective is not None
+            and not np.isnan(network.objective)
             else "infeasible"
         )
     except Exception as e:
         logger.warning(
             f"Solver error for product demand {product_demand_mt} Mt/year: {e}"
         )
+        status_ok = False
         optimization_status = "error"
 
     if optimization_status != "optimal":
@@ -620,6 +633,7 @@ def extract_lcox(network, product, demands):
     results_df = extract_lcox(
         network=network,
         product=product,
+        status_ok=status_ok,
         demands=scaled_demands,
     )