Skip to content

couplers.remind

REMIND coupling backends: GDX and IAMC (.mif).

Two Coupler subclasses implementing the source-specific hooks (build_regional_demand and extract_cost_parameters) for REMIND output:

  • RemindGdxCoupler — GDX backend (reads load_sector/cost symbols directly).
  • RemindIamcCoupler — IAMC .mif backend (derives demand via T&D efficiency + AC residual; reads per-parameter variable-sets, converts FOM to %/capex, derives nuclear fuel cost/efficiency from mass-basis REMIND variables).

All other builders (build_co2_prices, discount_rates, downscale_country_demand) are inherited from Coupler unchanged — they work for both backends via spec-shape dispatch.

RemindGdxCoupler

Bases: Coupler

Coupler specialised for REMIND GDX output.

Implements: - build_regional_demand: reads load_sector GDX symbol (TWa→MWh via spec). - extract_cost_parameters: reads all cost symbols from GDX, applies REMIND-GDX tech-facts (tnrs/peur nuclear mass-basis → USD/MWh_el, storage $/MWh label, GAMS-dropped zeros filled).

Source code in src/iampypsa/couplers/remind.py 
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
class RemindGdxCoupler(Coupler):
    """Coupler specialised for REMIND GDX output.

    Implements:
    - ``build_regional_demand``: reads ``load_sector`` GDX symbol (TWa→MWh via spec).
    - ``extract_cost_parameters``: reads all cost symbols from GDX, applies REMIND-GDX
      tech-facts (tnrs/peur nuclear mass-basis → USD/MWh_el, storage $/MWh label,
      GAMS-dropped zeros filled).
    """

    def build_regional_demand(self) -> pd.DataFrame:
        """Read REMIND regional sectoral demand as tidy ``[year, region, sector, value]`` (MWh/yr).

        Reads ``demand_fe_sectors`` (``p32_load_sector``), with TWa→MWh conversion applied by
        ``load_frame``, and restricts to the configured REMIND regions. All available years are
        returned; the year filter to planning horizons happens in ``downscale_country_demand``.
        """
        raw = load_frame(self.loader, self.symbols["demand_fe_sectors"])
        raw["year"] = raw["year"].astype(int)
        return convert_loads(raw, regions=self.model_regions, unit_factor=1.0)

    def extract_cost_parameters(self, year: int) -> pd.DataFrame:
        """Extract REMIND GDX cost parameters as long
        ``[region, technology, parameter, value, unit]``.

        Unit conversions are config-declared (applied by ``load_frame``/``load_set``). The
        REMIND-GDX-specific tech-facts encoded here are:
        - ``tnrs`` (nuclear) efficiency is mass-basis (TWa_elec/Mt_Ur); combined with ``peur``'s
          $/g_U fuel price into a true USD/MWh_el cost + 1.0 p.u. efficiency (see
          ``_nuclear_fuel_cost``).
        - Storage techs (``h2stor``, ``btstor``) share the $/MW capex factor but are
          relabelled $/MWh.
        - GDX/GAMS drops explicit zeros, so entries missing for a modeled technology are
          true zeros (filled via ``_fill_missing_with_zero``).
        """
        year_str = str(year)
        load = lambda name: load_frame(self.loader, self.symbols[name])  # noqa: E731

        # Investment: T$/TW->$/MW applied in load_frame; storage techs relabelled $/MWh.
        costs = load("cost_investment")
        costs = costs.loc[costs["year"].astype(str) == year_str].copy()
        costs = annotate_cost_rows(costs, parameter="investment", unit="USD/MW")
        costs.loc[costs["technology"].isin(["h2stor", "btstor"]), "unit"] = "USD/MWh"

        # tech_data: mixed-unit set (lifetime/FOM/VOM) - split + converted per YAML schema.
        techd = load_set(self.loader, self.symbols["tech_data"])

        # CO2 intensity: Gt_C/TWa->t_CO2/MWh applied in load_frame; carrier filter here.
        co2i = load("emission_factor")
        co2i = co2i.loc[
            (co2i["to_carrier"] == "seel")
            & (co2i["emission_type"] == "co2")
            & (co2i["year"].astype(str) == year_str)
        ].copy()
        co2i = annotate_cost_rows(co2i, parameter="CO2 intensity", unit="t_CO2/MWh_th")

        # GDX/GAMS drops explicit zeros, so missing entries for modeled technologies are true zeros.
        modeled_techs = costs[["region", "technology"]].drop_duplicates()
        co2i = self._fill_missing_with_zero(co2i, modeled_techs, "CO2 intensity", "t_CO2/MWh_th")
        vom = techd[techd["parameter"] == "VOM"]
        vom_filled = self._fill_missing_with_zero(vom, modeled_techs, "VOM", "$/MWh")
        techd = pd.concat([techd[techd["parameter"] != "VOM"], vom_filled], ignore_index=True)

        # Efficiency
        dataeta = load("efficiency_data")
        dataeta = dataeta.loc[dataeta["year"].astype(str) == year_str]
        eta = load("efficiency_conv")
        eta = eta.loc[eta["year"].astype(str) == year_str]
        keys = set(zip(dataeta["region"], dataeta["technology"]))
        fallback = eta[
            ~pd.MultiIndex.from_arrays([eta["region"], eta["technology"]]).isin(keys)
        ]
        eff = annotate_cost_rows(pd.concat([dataeta, fallback]), parameter="efficiency", unit="p.u.")
        # GDX nuclear efficiency is raw TWa_elec/Mt_Ur (mass basis, not thermal %). Convert to
        # MWh/g_U via the TWa->MWh, Mt->g factor (HOURS_PER_YEAR/1e6 = 8.76e9/1e12).
        eff.loc[eff["technology"] == "tnrs", "value"] *= HOURS_PER_YEAR / 1e6
        eff.loc[eff["technology"] == "tnrs", "unit"] = "MWh/g_U"

        # Fuel: T$/TWa->$/MWh for all except peur (already $/g_U in GDX).
        fuel = load("fuel_price")
        fuel = fuel.loc[fuel["year"].astype(str) == year_str].copy()
        fuel.loc[fuel["technology"] != "peur", "value"] *= unit_factor("T$/TWa", "$/MWh")
        fuel = annotate_cost_rows(fuel, parameter="fuel", unit="USD/MWh_th")
        fuel.loc[fuel["technology"] == "peur", "unit"] = "USD/g_U"

        eff, fuel = self._nuclear_fuel_cost(eff, fuel)

        df = pd.concat([costs, techd, co2i, eff, fuel])[
            ["region", "technology", "parameter", "value", "unit"]
        ]
        # Output boundary: raw REMIND tokens -> canonical vocabulary
        names = self.symbols.get("technology_names", {})
        df = rename_technologies(df, names)
        df = broadcast_fuel_prices(df, self._tech_fuel_map_from_pe2se())
        df = df[df["technology"].isin(set(names.values()))]
        return df[df["region"].isin(self.model_regions)].reset_index(drop=True)

    def _tech_fuel_map_from_pe2se(self) -> dict[str, str]:
        """Build the canonical ``technology -> fuel`` map from the GDX ``pe2se`` set.
        """
        names = self.symbols.get("technology_names", {})
        pe2se = load_frame(self.loader, self.symbols["pe2se"])
        seel = pe2se[pe2se["all_enty_1"] == "seel"]
        return {
            names.get(row["all_te_2"], row["all_te_2"]): names[row["all_enty_0"]]
            for _, row in seel.iterrows()
            if row["all_enty_0"] in names  # priced PE carrier (has a technology_names entry)
        }

    @staticmethod
    def _fill_missing_with_zero(
        sparse: pd.DataFrame, modeled_techs: pd.DataFrame, parameter: str, unit: str,
    ) -> pd.DataFrame:
        """Reindex against modeled technologies (from investment data), filling gaps with 0 —
        GDX/GAMS drops explicit zeros, so a missing entry there is a true zero, not a gap."""
        merged = modeled_techs.merge(
            sparse[["region", "technology", "value"]], on=["region", "technology"], how="left",
        )
        merged["value"] = merged["value"].fillna(0.0)
        merged["parameter"] = parameter
        merged["unit"] = unit
        return merged

    @staticmethod
    def _nuclear_fuel_cost(eff: pd.DataFrame, fuel: pd.DataFrame) -> tuple[pd.DataFrame, pd.DataFrame]:
        """Correct tnrs/peur's rows in-place: mass-basis MWh/g_U, USD/g_U -> USD/MWh_el fuel cost
        + 1.0 p.u. efficiency.
        """
        eff = eff.copy()
        fuel = fuel.copy()

        tnrs_eta = eff.loc[eff["technology"] == "tnrs", ["region", "value"]].set_index("region")["value"]
        peur_price = fuel.loc[fuel["technology"] == "peur", ["region", "value"]].set_index("region")["value"]
        usd_per_mwh_el = peur_price / tnrs_eta  # g_U mass unit cancels

        is_tnrs = eff["technology"] == "tnrs"
        eff.loc[is_tnrs, "value"] = 1.0
        eff.loc[is_tnrs, "unit"] = "p.u."

        is_peur = fuel["technology"] == "peur"
        # Categorical region dtype survives .map() as Categorical floats; cast to plain float
        # before assigning into the float64 column, or .loc hits a pandas AssertionError.
        mapped = fuel.loc[is_peur, "region"].map(usd_per_mwh_el).astype(float)
        fuel.loc[is_peur, "value"] = mapped
        fuel.loc[is_peur, "unit"] = "USD/MWh_el"

        return eff, fuel

build_regional_demand()

Read REMIND regional sectoral demand as tidy [year, region, sector, value] (MWh/yr).

Reads demand_fe_sectors (p32_load_sector), with TWa→MWh conversion applied by load_frame, and restricts to the configured REMIND regions. All available years are returned; the year filter to planning horizons happens in downscale_country_demand.

Source code in src/iampypsa/couplers/remind.py 
45
46
47
48
49
50
51
52
53
54
def build_regional_demand(self) -> pd.DataFrame:
    """Read REMIND regional sectoral demand as tidy ``[year, region, sector, value]`` (MWh/yr).

    Reads ``demand_fe_sectors`` (``p32_load_sector``), with TWa→MWh conversion applied by
    ``load_frame``, and restricts to the configured REMIND regions. All available years are
    returned; the year filter to planning horizons happens in ``downscale_country_demand``.
    """
    raw = load_frame(self.loader, self.symbols["demand_fe_sectors"])
    raw["year"] = raw["year"].astype(int)
    return convert_loads(raw, regions=self.model_regions, unit_factor=1.0)

extract_cost_parameters(year)

Extract REMIND GDX cost parameters as long [region, technology, parameter, value, unit].

Unit conversions are config-declared (applied by load_frame/load_set). The REMIND-GDX-specific tech-facts encoded here are: - tnrs (nuclear) efficiency is mass-basis (TWa_elec/Mt_Ur); combined with peur's $/g_U fuel price into a true USD/MWh_el cost + 1.0 p.u. efficiency (see _nuclear_fuel_cost). - Storage techs (h2stor, btstor) share the $/MW capex factor but are relabelled $/MWh. - GDX/GAMS drops explicit zeros, so entries missing for a modeled technology are true zeros (filled via _fill_missing_with_zero).

Source code in src/iampypsa/couplers/remind.py 
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
def extract_cost_parameters(self, year: int) -> pd.DataFrame:
    """Extract REMIND GDX cost parameters as long
    ``[region, technology, parameter, value, unit]``.

    Unit conversions are config-declared (applied by ``load_frame``/``load_set``). The
    REMIND-GDX-specific tech-facts encoded here are:
    - ``tnrs`` (nuclear) efficiency is mass-basis (TWa_elec/Mt_Ur); combined with ``peur``'s
      $/g_U fuel price into a true USD/MWh_el cost + 1.0 p.u. efficiency (see
      ``_nuclear_fuel_cost``).
    - Storage techs (``h2stor``, ``btstor``) share the $/MW capex factor but are
      relabelled $/MWh.
    - GDX/GAMS drops explicit zeros, so entries missing for a modeled technology are
      true zeros (filled via ``_fill_missing_with_zero``).
    """
    year_str = str(year)
    load = lambda name: load_frame(self.loader, self.symbols[name])  # noqa: E731

    # Investment: T$/TW->$/MW applied in load_frame; storage techs relabelled $/MWh.
    costs = load("cost_investment")
    costs = costs.loc[costs["year"].astype(str) == year_str].copy()
    costs = annotate_cost_rows(costs, parameter="investment", unit="USD/MW")
    costs.loc[costs["technology"].isin(["h2stor", "btstor"]), "unit"] = "USD/MWh"

    # tech_data: mixed-unit set (lifetime/FOM/VOM) - split + converted per YAML schema.
    techd = load_set(self.loader, self.symbols["tech_data"])

    # CO2 intensity: Gt_C/TWa->t_CO2/MWh applied in load_frame; carrier filter here.
    co2i = load("emission_factor")
    co2i = co2i.loc[
        (co2i["to_carrier"] == "seel")
        & (co2i["emission_type"] == "co2")
        & (co2i["year"].astype(str) == year_str)
    ].copy()
    co2i = annotate_cost_rows(co2i, parameter="CO2 intensity", unit="t_CO2/MWh_th")

    # GDX/GAMS drops explicit zeros, so missing entries for modeled technologies are true zeros.
    modeled_techs = costs[["region", "technology"]].drop_duplicates()
    co2i = self._fill_missing_with_zero(co2i, modeled_techs, "CO2 intensity", "t_CO2/MWh_th")
    vom = techd[techd["parameter"] == "VOM"]
    vom_filled = self._fill_missing_with_zero(vom, modeled_techs, "VOM", "$/MWh")
    techd = pd.concat([techd[techd["parameter"] != "VOM"], vom_filled], ignore_index=True)

    # Efficiency
    dataeta = load("efficiency_data")
    dataeta = dataeta.loc[dataeta["year"].astype(str) == year_str]
    eta = load("efficiency_conv")
    eta = eta.loc[eta["year"].astype(str) == year_str]
    keys = set(zip(dataeta["region"], dataeta["technology"]))
    fallback = eta[
        ~pd.MultiIndex.from_arrays([eta["region"], eta["technology"]]).isin(keys)
    ]
    eff = annotate_cost_rows(pd.concat([dataeta, fallback]), parameter="efficiency", unit="p.u.")
    # GDX nuclear efficiency is raw TWa_elec/Mt_Ur (mass basis, not thermal %). Convert to
    # MWh/g_U via the TWa->MWh, Mt->g factor (HOURS_PER_YEAR/1e6 = 8.76e9/1e12).
    eff.loc[eff["technology"] == "tnrs", "value"] *= HOURS_PER_YEAR / 1e6
    eff.loc[eff["technology"] == "tnrs", "unit"] = "MWh/g_U"

    # Fuel: T$/TWa->$/MWh for all except peur (already $/g_U in GDX).
    fuel = load("fuel_price")
    fuel = fuel.loc[fuel["year"].astype(str) == year_str].copy()
    fuel.loc[fuel["technology"] != "peur", "value"] *= unit_factor("T$/TWa", "$/MWh")
    fuel = annotate_cost_rows(fuel, parameter="fuel", unit="USD/MWh_th")
    fuel.loc[fuel["technology"] == "peur", "unit"] = "USD/g_U"

    eff, fuel = self._nuclear_fuel_cost(eff, fuel)

    df = pd.concat([costs, techd, co2i, eff, fuel])[
        ["region", "technology", "parameter", "value", "unit"]
    ]
    # Output boundary: raw REMIND tokens -> canonical vocabulary
    names = self.symbols.get("technology_names", {})
    df = rename_technologies(df, names)
    df = broadcast_fuel_prices(df, self._tech_fuel_map_from_pe2se())
    df = df[df["technology"].isin(set(names.values()))]
    return df[df["region"].isin(self.model_regions)].reset_index(drop=True)

RemindIamcCoupler

Bases: Coupler

Coupler specialised for REMIND IAMC .mif output.

Implements: - build_regional_demand: FE sector rebasing via derived η_td + AC residual. - extract_cost_parameters: loads per-parameter variable-sets, computes FOM%, derives nuclear fuel cost/efficiency from mass-basis REMIND variables.

Source code in src/iampypsa/couplers/remind.py 
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
class RemindIamcCoupler(Coupler):
    """Coupler specialised for REMIND IAMC ``.mif`` output.

    Implements:
    - ``build_regional_demand``: FE sector rebasing via derived η_td + AC residual.
    - ``extract_cost_parameters``: loads per-parameter variable-sets, computes FOM%,
      derives nuclear fuel cost/efficiency from mass-basis REMIND variables.
    """

    def build_regional_demand(self) -> pd.DataFrame:
        """Derive regional sectoral electricity demand from IAMC mif variables (MWh/yr).

        Variable names and sector token labels come from the symbol config.
        The algorithm applied on top is REMIND-specific:

        1. η_td = (SE − Losses) / SE  (derived T&D efficiency, replaces GDX pm_eta_conv).
          2. Electricity FE sectors are rebased to the SE level: FE_sector_MWh / η_td.
              ``demand_h2`` is treated as a hydrogen-demand quantity and is not rebased.
        3. Electrolysis electricity demand (MWh_el) =
               (SE|Hydrogen|Electricity − SE|Input|Hydrogen|Electricity) / η_elec
           Both SE variables are in EJ H2; the difference is the net H2 from electricity
           destined for final-energy demand (not cycling back via fuel cells). Dividing
           by η_elec converts to the electricity consumed to produce it.
           This matches REMIND's p32_load_sector("elh2") on the GDX path exactly.
        4. AC = (SE − Losses) − Σ(rebased FE sectors) − electrolysis  (residual).
           Negative values are clamped to 0 with a warning.

        Returns ``[year, region, sector, value, unit]`` matching the GDX path.
        """
        fe_df = load_spec(self.loader, self.symbols["demand_fe_sectors"])
        fe_df = fe_df[fe_df["region"].isin(set(self.model_regions))]

        se_var      = self.symbols["demand_se_electricity"]["symbol"]
        losses_var  = self.symbols["demand_transmission_losses"]["symbol"]
        h2_prod_var = self.symbols["demand_h2_from_electricity"]["symbol"]
        h2_turb_var = self.symbols["demand_h2_to_turbines"]["symbol"]
        eta_var     = self.symbols["demand_electrolysis_efficiency"]["symbol"]

        scalar_df = read_iamc(
            self.loader.source,
            variables=[se_var, losses_var, h2_prod_var, h2_turb_var, eta_var],
        )
        scalar_df = scalar_df[scalar_df["region"].isin(set(self.model_regions))]

        rows = []
        for (region, year), grp in scalar_df.groupby(["region", "year"]):
            get = self._group_getter(grp)

            se = get(se_var)
            losses = get(losses_var)
            eta_td = self._td_efficiency(se, losses)

            fe_slice = fe_df[(fe_df["region"] == region) & (fe_df["year"] == year)]
            fe_rows, rebased_sum_mwh, has_se_h2 = self._convert_fe_sectors_to_se(
                fe_slice, eta_td, region, year
            )
            rows.extend(fe_rows)

            h2_demand_mwh = self._net_h2_demand_mwh(
                get(h2_prod_var), get(h2_turb_var), region, year
            )
            if not has_se_h2:
                rows.append(
                    {
                        "year": year,
                        "region": region,
                        "sector": "demand_h2",
                        "value": h2_demand_mwh,
                        "unit": "MWh_H2",
                    }
                )

            eta_elec = get(eta_var) / 100.0  # mif reports in %
            elec_h2_mwh = h2_demand_mwh/eta_elec

            rows.append({
                "year": year, "region": region, "sector": "electrolysis",
                "value": elec_h2_mwh, "unit": "MWh",
            })

            ac_mwh = self._ac_residual(se, losses, rebased_sum_mwh, elec_h2_mwh, region, year)
            rows.append({
                "year": year, "region": region, "sector": "AC",
                "value": ac_mwh, "unit": "MWh",
            })

        return (
            pd.DataFrame(rows)
            .sort_values(["year", "region", "sector"])
            .reset_index(drop=True)
        )

    # -- build_regional_demand helpers --------------------------------------

    @staticmethod
    def _group_getter(grp: pd.DataFrame):
        """Return a ``get(variable) -> float`` summing that variable's values in ``grp``."""
        def get(v: str) -> float:
            vals = grp.loc[grp["variable"] == v, "value"]
            return float(vals.sum()) if not vals.empty else 0.0
        return get

    @staticmethod
    def _td_efficiency(se: float, losses: float) -> float:
        """Derived T&D efficiency η_td = (SE − Losses) / SE (1.0 when SE is non-positive)."""
        return (se - losses) / se if se > 0 else 1.0

    @staticmethod
    def _convert_fe_sectors_to_se(
        fe_slice: pd.DataFrame, eta_td: float, region: str, year: int
    ) -> tuple[list[dict], float, bool]:
        """Return FE rows, rebased-electricity sum, and whether FE provided ``demand_h2``.

        Electricity FE sectors are rebased by ``η_td`` and contribute to the AC residual
        subtraction. ``demand_h2`` is treated as hydrogen demand and passed through
        unchanged, i.e. it is not rebased and does not enter the AC residual sum.
        """
        rows = []
        rebased_sum_mwh = 0.0
        has_se_h2 = False
        for _, fe_row in fe_slice.iterrows():
            sector = str(fe_row["sector"])
            if sector == "demand_h2":
                has_se_h2 = True
                rows.append(
                    {
                        "year": year,
                        "region": region,
                        "sector": sector,
                        "value": fe_row["value"],
                        "unit": "MWh_H2",
                    }
                )
                continue

            se_val_mwh = fe_row["value"] / eta_td if eta_td > 0 else fe_row["value"]
            rebased_sum_mwh += se_val_mwh
            rows.append({
                "year": year, "region": region, "sector": sector,
                "value": se_val_mwh, "unit": "MWh",
            })
        return rows, rebased_sum_mwh, has_se_h2

    @staticmethod
    def _net_h2_demand_mwh(h2_prod_ej: float, h2_turb_ej: float, region: str, year: int) -> float:
        """Net hydrogen demand (MWh_H2) from electricity-route hydrogen balances.
        Seasonal storage is a pypsa-decision for elec supply and not a load -> discarded

        Computes ``(SE|Hydrogen|Electricity - SE|Input|Hydrogen|Electricity)`` and
        converts EJ/yr to MWh/yr. Negative values are clamped to zero.
        """
        net_h2_mwh = (h2_prod_ej - h2_turb_ej) * _EJ_TO_MWH
        if net_h2_mwh < 0:
            logger.warning(
                "Negative net H2 demand for region=%s year=%s: %.4f MWh - clamped to 0.",
                region,
                year,
                net_h2_mwh,
            )
            return 0.0
        return net_h2_mwh

    @staticmethod
    def _ac_residual(
        se: float, losses: float, rebased_sum_mwh: float, elec_h2_mwh: float,
        region: str, year: int,
    ) -> float:
        """AC residual: (SE − Losses) − Σ(rebased FE) − electrolysis, in MWh (clamped ≥ 0)."""
        ac_mwh = (se - losses) * _EJ_TO_MWH - rebased_sum_mwh - elec_h2_mwh
        if ac_mwh < 0:
            logger.warning(
                "Negative AC residual for region=%s year=%s: %.4f MWh — clamped to 0.",
                region, year, ac_mwh,
            )
            ac_mwh = 0.0
        return ac_mwh

    def extract_cost_parameters(self, year: int) -> pd.DataFrame:
        """Extract REMIND mif cost parameters as ``[region, technology, parameter, value, unit]``.

        Reads six per-parameter variable-sets declared in the IAMC symbol config
        (cost_investment, tech_lifetime, cost_omf, cost_omv, efficiency, fuel_price,
        emission_factor), queries to ``year``, and:

        - Computes FOM%/yr = absolute FOM (USD/MW/yr) / capex (USD/MW) × 100, because the
          mif reports absolute FOM whereas PyPSA-Eur uses percent-of-capex.
        - Derives nuclear's fuel cost/efficiency from mass-basis price/conversion-factor
          variables (see ``_nuclear_fuel_cost``) — not a fallback, real per-region REMIND data.
        - Reads real per-tech CO2 intensity (t_CO2/MWh_th); biomass techs without a mif
          variable fall back to 0.0 (carbon-neutral), declared in the symbol config.

        Battery cost tokens are intentionally omitted (their mif values are full-system
        costs per kW_power and are not comparable to PyPSA-Eur's separate inverter+storage
        parametrisation; those techs fall back to the PyPSA-Eur baseline).
        """
        y = str(year)
        currency_factor: float = self.config.get("currency_factor", 1.0)

        def load(name: str) -> pd.DataFrame:
            df = load_variable_set(self.loader, self.symbols[name])
            return df[df["year"].astype(str) == y].copy()

        # --- investment (USD/MW) ---
        capex = annotate_cost_rows(load("cost_investment"), parameter="investment", unit="USD/MW", factor=currency_factor)

        # --- lifetime (yr) ---
        lifetime = annotate_cost_rows(load("tech_lifetime"), parameter="lifetime", unit="yr")

        # --- FOM: compute %/yr = absolute / capex × 100 ---
        fom_abs = load("cost_omf")
        if currency_factor != 1.0:
            fom_abs["value"] *= currency_factor
        fom_pct = self._compute_fom_pct(capex, fom_abs)

        # --- VOM (USD/MWh) ---
        vom = annotate_cost_rows(load("cost_omv"), parameter="VOM", unit="USD/MWh", factor=currency_factor)

        # --- efficiency (p.u.) ---
        eff = annotate_cost_rows(load("efficiency"), parameter="efficiency", unit="p.u.")

        # --- fuel price (USD/MWh_th) ---
        fuel = annotate_cost_rows(load("fuel_price"), parameter="fuel", unit="USD/MWh_th", factor=currency_factor)

        # --- nuclear: fuel cost (USD/MWh_el) + efficiency (1.0 p.u.), computed from the
        # uranium mass-basis price/conversion-factor variables (mass unit cancels in the
        # ratio); see _nuclear_fuel_cost.
        nuclear_fuel, nuclear_eff = self._nuclear_fuel_cost(year)
        if currency_factor != 1.0:
            nuclear_fuel["value"] *= currency_factor
        fuel = pd.concat([fuel, nuclear_fuel], ignore_index=True)
        eff = pd.concat([eff, nuclear_eff], ignore_index=True)

        # --- CO2 intensity (t_CO2/MWh_th) — biomass techs fall back to 0.0 (carbon-neutral) ---
        co2i = annotate_cost_rows(load("emission_factor"), parameter="CO2 intensity", unit="t_CO2/MWh_th")

        # --- assemble ---
        frames = [capex, lifetime, fom_pct, vom, eff, fuel, co2i]
        keep = ["region", "technology", "parameter", "value", "unit"]
        df = pd.concat(
            [f[keep] for f in frames if set(keep).issubset(f.columns)],
            ignore_index=True,
        )
        # Output boundary (mirrors RemindGdxCoupler): rename is a no-op here — mif labels are
        # already canonical — then per-fuel price rows become one `fuel` row per technology.
        df = rename_technologies(df, self.symbols.get("technology_names"))
        df = broadcast_fuel_prices(df, self.symbols.get("tech_fuel_map"))
        return df[df["region"].isin(set(self.model_regions))].reset_index(drop=True)

    # -- extract_cost_parameters helpers ------------------------------------

    @staticmethod
    def _compute_fom_pct(capex: pd.DataFrame, fom_abs: pd.DataFrame) -> pd.DataFrame:
        """FOM %/yr = absolute FOM (USD/MW/yr) / capex (USD/MW) × 100, joined on tech/region/year."""
        fom_pct = capex[["year", "region", "technology", "value"]].merge(
            fom_abs[["year", "region", "technology", "value"]],
            on=["year", "region", "technology"],
            suffixes=("_cap", "_fom"),
        )
        fom_pct["value"] = fom_pct["value_fom"] / fom_pct["value_cap"] * 100
        fom_pct["parameter"] = "FOM"
        fom_pct["unit"] = "%/yr"
        return fom_pct

    def _nuclear_fuel_cost(self, year: int) -> tuple[pd.DataFrame, pd.DataFrame]:
        """Compute nuclear's fuel cost (USD/MWh_el) and efficiency (1.0 p.u.) for ``year``.

        REMIND reports nuclear on a uranium-mass basis (price ÷ conversion factor cancels the
        mass unit); efficiency is reported as a genuine 1.0 p.u. so downstream consumers
        (marginal_cost = fuel / efficiency, Generator.efficiency) stay consistent.
        """
        y = str(year)
        conversion = load_frame(self.loader, self.symbols["nuclear_conversion_factor"])
        price = load_frame(self.loader, self.symbols["nuclear_price"])
        conversion = conversion[conversion["year"].astype(str) == y]
        price = price[price["year"].astype(str) == y]

        merged = price.merge(
            conversion[["region", "value"]].rename(columns={"value": "conversion_factor"}),
            on="region",
        )
        # kg_Ur mass unit cancels; the remaining GJ->MWh mismatch needs unit_factor.
        merged["value"] = (
            merged["value"] / merged["conversion_factor"] * unit_factor("US$2017/GJ", "USD/MWh")
        )
        merged["technology"] = "uranium"
        fuel = annotate_cost_rows(merged, parameter="fuel", unit="USD/MWh_el")
        # Tag the computed fuel price with the canonical fuel name; broadcast_fuel_prices
        # then folds it into nuclear's own `fuel` row via the config's tech_fuel_map
        # (nuclear: uranium), like every other fuel.
        fuel = fuel[["year", "region", "technology", "parameter", "value", "unit"]]

        eff = fuel[["year", "region"]].copy()
        eff["technology"] = "nuclear"
        eff = annotate_cost_rows(eff, parameter="efficiency", unit="p.u.")
        eff["value"] = 1.0

        return fuel, eff

build_regional_demand()

Derive regional sectoral electricity demand from IAMC mif variables (MWh/yr).

Variable names and sector token labels come from the symbol config. The algorithm applied on top is REMIND-specific:

  1. η_td = (SE − Losses) / SE (derived T&D efficiency, replaces GDX pm_eta_conv).
  2. Electricity FE sectors are rebased to the SE level: FE_sector_MWh / η_td. demand_h2 is treated as a hydrogen-demand quantity and is not rebased.
  3. Electrolysis electricity demand (MWh_el) = (SE|Hydrogen|Electricity − SE|Input|Hydrogen|Electricity) / η_elec Both SE variables are in EJ H2; the difference is the net H2 from electricity destined for final-energy demand (not cycling back via fuel cells). Dividing by η_elec converts to the electricity consumed to produce it. This matches REMIND's p32_load_sector("elh2") on the GDX path exactly.
  4. AC = (SE − Losses) − Σ(rebased FE sectors) − electrolysis (residual). Negative values are clamped to 0 with a warning.

Returns [year, region, sector, value, unit] matching the GDX path.

Source code in src/iampypsa/couplers/remind.py 
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
def build_regional_demand(self) -> pd.DataFrame:
    """Derive regional sectoral electricity demand from IAMC mif variables (MWh/yr).

    Variable names and sector token labels come from the symbol config.
    The algorithm applied on top is REMIND-specific:

    1. η_td = (SE − Losses) / SE  (derived T&D efficiency, replaces GDX pm_eta_conv).
      2. Electricity FE sectors are rebased to the SE level: FE_sector_MWh / η_td.
          ``demand_h2`` is treated as a hydrogen-demand quantity and is not rebased.
    3. Electrolysis electricity demand (MWh_el) =
           (SE|Hydrogen|Electricity − SE|Input|Hydrogen|Electricity) / η_elec
       Both SE variables are in EJ H2; the difference is the net H2 from electricity
       destined for final-energy demand (not cycling back via fuel cells). Dividing
       by η_elec converts to the electricity consumed to produce it.
       This matches REMIND's p32_load_sector("elh2") on the GDX path exactly.
    4. AC = (SE − Losses) − Σ(rebased FE sectors) − electrolysis  (residual).
       Negative values are clamped to 0 with a warning.

    Returns ``[year, region, sector, value, unit]`` matching the GDX path.
    """
    fe_df = load_spec(self.loader, self.symbols["demand_fe_sectors"])
    fe_df = fe_df[fe_df["region"].isin(set(self.model_regions))]

    se_var      = self.symbols["demand_se_electricity"]["symbol"]
    losses_var  = self.symbols["demand_transmission_losses"]["symbol"]
    h2_prod_var = self.symbols["demand_h2_from_electricity"]["symbol"]
    h2_turb_var = self.symbols["demand_h2_to_turbines"]["symbol"]
    eta_var     = self.symbols["demand_electrolysis_efficiency"]["symbol"]

    scalar_df = read_iamc(
        self.loader.source,
        variables=[se_var, losses_var, h2_prod_var, h2_turb_var, eta_var],
    )
    scalar_df = scalar_df[scalar_df["region"].isin(set(self.model_regions))]

    rows = []
    for (region, year), grp in scalar_df.groupby(["region", "year"]):
        get = self._group_getter(grp)

        se = get(se_var)
        losses = get(losses_var)
        eta_td = self._td_efficiency(se, losses)

        fe_slice = fe_df[(fe_df["region"] == region) & (fe_df["year"] == year)]
        fe_rows, rebased_sum_mwh, has_se_h2 = self._convert_fe_sectors_to_se(
            fe_slice, eta_td, region, year
        )
        rows.extend(fe_rows)

        h2_demand_mwh = self._net_h2_demand_mwh(
            get(h2_prod_var), get(h2_turb_var), region, year
        )
        if not has_se_h2:
            rows.append(
                {
                    "year": year,
                    "region": region,
                    "sector": "demand_h2",
                    "value": h2_demand_mwh,
                    "unit": "MWh_H2",
                }
            )

        eta_elec = get(eta_var) / 100.0  # mif reports in %
        elec_h2_mwh = h2_demand_mwh/eta_elec

        rows.append({
            "year": year, "region": region, "sector": "electrolysis",
            "value": elec_h2_mwh, "unit": "MWh",
        })

        ac_mwh = self._ac_residual(se, losses, rebased_sum_mwh, elec_h2_mwh, region, year)
        rows.append({
            "year": year, "region": region, "sector": "AC",
            "value": ac_mwh, "unit": "MWh",
        })

    return (
        pd.DataFrame(rows)
        .sort_values(["year", "region", "sector"])
        .reset_index(drop=True)
    )

extract_cost_parameters(year)

Extract REMIND mif cost parameters as [region, technology, parameter, value, unit].

Reads six per-parameter variable-sets declared in the IAMC symbol config (cost_investment, tech_lifetime, cost_omf, cost_omv, efficiency, fuel_price, emission_factor), queries to year, and:

  • Computes FOM%/yr = absolute FOM (USD/MW/yr) / capex (USD/MW) × 100, because the mif reports absolute FOM whereas PyPSA-Eur uses percent-of-capex.
  • Derives nuclear's fuel cost/efficiency from mass-basis price/conversion-factor variables (see _nuclear_fuel_cost) — not a fallback, real per-region REMIND data.
  • Reads real per-tech CO2 intensity (t_CO2/MWh_th); biomass techs without a mif variable fall back to 0.0 (carbon-neutral), declared in the symbol config.

Battery cost tokens are intentionally omitted (their mif values are full-system costs per kW_power and are not comparable to PyPSA-Eur's separate inverter+storage parametrisation; those techs fall back to the PyPSA-Eur baseline).

Source code in src/iampypsa/couplers/remind.py 
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
def extract_cost_parameters(self, year: int) -> pd.DataFrame:
    """Extract REMIND mif cost parameters as ``[region, technology, parameter, value, unit]``.

    Reads six per-parameter variable-sets declared in the IAMC symbol config
    (cost_investment, tech_lifetime, cost_omf, cost_omv, efficiency, fuel_price,
    emission_factor), queries to ``year``, and:

    - Computes FOM%/yr = absolute FOM (USD/MW/yr) / capex (USD/MW) × 100, because the
      mif reports absolute FOM whereas PyPSA-Eur uses percent-of-capex.
    - Derives nuclear's fuel cost/efficiency from mass-basis price/conversion-factor
      variables (see ``_nuclear_fuel_cost``) — not a fallback, real per-region REMIND data.
    - Reads real per-tech CO2 intensity (t_CO2/MWh_th); biomass techs without a mif
      variable fall back to 0.0 (carbon-neutral), declared in the symbol config.

    Battery cost tokens are intentionally omitted (their mif values are full-system
    costs per kW_power and are not comparable to PyPSA-Eur's separate inverter+storage
    parametrisation; those techs fall back to the PyPSA-Eur baseline).
    """
    y = str(year)
    currency_factor: float = self.config.get("currency_factor", 1.0)

    def load(name: str) -> pd.DataFrame:
        df = load_variable_set(self.loader, self.symbols[name])
        return df[df["year"].astype(str) == y].copy()

    # --- investment (USD/MW) ---
    capex = annotate_cost_rows(load("cost_investment"), parameter="investment", unit="USD/MW", factor=currency_factor)

    # --- lifetime (yr) ---
    lifetime = annotate_cost_rows(load("tech_lifetime"), parameter="lifetime", unit="yr")

    # --- FOM: compute %/yr = absolute / capex × 100 ---
    fom_abs = load("cost_omf")
    if currency_factor != 1.0:
        fom_abs["value"] *= currency_factor
    fom_pct = self._compute_fom_pct(capex, fom_abs)

    # --- VOM (USD/MWh) ---
    vom = annotate_cost_rows(load("cost_omv"), parameter="VOM", unit="USD/MWh", factor=currency_factor)

    # --- efficiency (p.u.) ---
    eff = annotate_cost_rows(load("efficiency"), parameter="efficiency", unit="p.u.")

    # --- fuel price (USD/MWh_th) ---
    fuel = annotate_cost_rows(load("fuel_price"), parameter="fuel", unit="USD/MWh_th", factor=currency_factor)

    # --- nuclear: fuel cost (USD/MWh_el) + efficiency (1.0 p.u.), computed from the
    # uranium mass-basis price/conversion-factor variables (mass unit cancels in the
    # ratio); see _nuclear_fuel_cost.
    nuclear_fuel, nuclear_eff = self._nuclear_fuel_cost(year)
    if currency_factor != 1.0:
        nuclear_fuel["value"] *= currency_factor
    fuel = pd.concat([fuel, nuclear_fuel], ignore_index=True)
    eff = pd.concat([eff, nuclear_eff], ignore_index=True)

    # --- CO2 intensity (t_CO2/MWh_th) — biomass techs fall back to 0.0 (carbon-neutral) ---
    co2i = annotate_cost_rows(load("emission_factor"), parameter="CO2 intensity", unit="t_CO2/MWh_th")

    # --- assemble ---
    frames = [capex, lifetime, fom_pct, vom, eff, fuel, co2i]
    keep = ["region", "technology", "parameter", "value", "unit"]
    df = pd.concat(
        [f[keep] for f in frames if set(keep).issubset(f.columns)],
        ignore_index=True,
    )
    # Output boundary (mirrors RemindGdxCoupler): rename is a no-op here — mif labels are
    # already canonical — then per-fuel price rows become one `fuel` row per technology.
    df = rename_technologies(df, self.symbols.get("technology_names"))
    df = broadcast_fuel_prices(df, self.symbols.get("tech_fuel_map"))
    return df[df["region"].isin(set(self.model_regions))].reset_index(drop=True)

read_region_map(source='country', target='model_region', file_path=None, flatten=False)

Read the REMIND region↔country mapping as {source: [target, ...]}.

Reads the ;-separated mapping CSV (columns RegionCode/CountryCode), converts ISO3 country codes to ISO2, and adds Kosovo (XK → NES). Pass source/target as "model_region" or "country" to select the groupby direction.

Source code in src/iampypsa/couplers/remind.py 
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
def read_region_map(
    source="country",
    target="model_region",
    file_path: str | PathLike | None = None,
    flatten: bool = False,
) -> dict:
    """Read the REMIND region↔country mapping as ``{source: [target, ...]}``.

    Reads the ``;``-separated mapping CSV (columns ``RegionCode``/``CountryCode``), converts
    ISO3 country codes to ISO2, and adds Kosovo (XK → NES). Pass ``source``/``target`` as
    ``"model_region"`` or ``"country"`` to select the groupby direction.
    """
    if file_path is None:
        base_path = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
        file_path = os.path.join(base_path, "data/remind", "regions.csv")
    region_mapping = pd.read_csv(file_path, sep=";").rename(columns={"RegionCode": "model_region"})
    region_mapping["country"] = coco.convert(names=region_mapping["CountryCode"], to="ISO2")
    region_mapping = region_mapping[["country", "model_region"]]

    # Kosovo: PyPSA-Eur uses "XK" (not recognised by country_converter); part of NES.
    region_mapping = pd.concat(
        [region_mapping, pd.DataFrame({"country": ["XK"], "model_region": ["NES"]})]
    ).reset_index(drop=True)

    grouped = region_mapping.groupby(source)[target].apply("unique").apply(list)
    if flatten:
        grouped = grouped.apply(lambda x: x[0])
    return grouped.to_dict()