Merge pull request #89 from staadecker/improv_hydro

pesap · staadecker · commit e2f053f9b2eb · 2023-01-29T14:38:55.000-06:00
Improve hydro module
diff --git a/REAM Model Changelog.md b/REAM Model Changelog.md
@@ -16,4 +16,5 @@ Changes are listed from oldest (first line) to newest (last line of table).
 | #56 | June 2021 | Convert 2020 predetermined build years to 2019 in `get_inputs.py` to avoid conflicts with 2020 period. |
 | #57 | June 2021 | Specify predetermined storage energy capacity in inputs (previously left unspecified). |
 | #68 | June 2021 | Change financial params to 2018 dollars & 5% interest rate. Start using terrain multipliers (which now include the economic multiplier). |
-| #72 | July 2021 | Drop build and O&M costs of existing transmission lines. |
+| #72 | July 2021 | Drop build and O&M costs of existing transmission lines. |
+| #89 | August 2021 | Change hydro module average flow constraint to a monthly constraint rather than per timeseries and change it to a <= rather than ==. |
diff --git a/switch_model/generators/extensions/hydro_simple.py b/switch_model/generators/extensions/hydro_simple.py
@@ -23,18 +23,26 @@
 
 INPUT FILE INFORMATION
 
-The single file hydro_timeseries.csv needs to contain
-entries for each dispatchable hydro project. The set of hydro projects
-is derived from this file, and this file should cover all time periods
-in which the hydro plant can operate.
-
-Run-of-River hydro projects should not be included in this file; RoR
-hydro is treated like any other variable renewable resource, and
-expects data in variable_capacity_factors.csv.
-
-hydro_timeseries.csv
-    hydro_generation_project, timeseries, hydro_min_flow_mw,
-    hydro_avg_flow_mw
+    The file hydro_timeseries.csv needs to contain
+    entries for each dispatchable hydro project. The set of hydro projects
+    is derived from this file, and this file should cover all time periods
+    in which the hydro plant can operate.
+
+    Run-of-River hydro projects should not be included in this file; RoR
+    hydro is treated like any other variable renewable resource, and
+    expects data in variable_capacity_factors.csv.
+
+    hydro_timeseries.csv
+        hydro_generation_project, hydro_timeseries, hydro_min_flow_mw,
+        hydro_avg_flow_mw
+
+    The file hydro_timepoints.csv is an optional mapping of timepoints
+    to a hydro timeseries. Hydro timeseries are different from the SWITCH
+    timeseries (timeseries.csv) as this allows hydro constraints to be
+    specified over a different time period.
+
+    hydro_timepoints.csv (optional)
+        timepoint_id,tp_to_hts
 """
 from __future__ import division
 
@@ -43,6 +51,7 @@
 # switch_model.hydro.simple, and the advanced components into
 # switch_model.hydro.water_network. That should set a good example
 # for other people who want to do other custom handling of hydro.
+import os.path
 
 from pyomo.environ import *
 
@@ -58,66 +67,95 @@
 
 def define_components(mod):
     """
-
     HYDRO_GENS is the set of dispatchable hydro projects. This is a subet
     of GENERATION_PROJECTS, and is determined by the inputs file hydro_timeseries.csv.
     Members of this set can be called either g, or hydro_g.
 
-    HYDRO_GEN_TS is the set of Hydro projects and timeseries for which
+    HYDRO_TS is the set of hydro timeseries over which the average flow constraint is defined.
+    These hydro timeseries are different from the timeseries used in the reset of SWITCH
+    and are defined by the input file hydro_timepoints.csv. If hydro_timepoints.csv doesn't exist,
+    the default is for the timeseries to be the same as the SWITCH timeseries from timeseries.csv.
+    Members of this set can be abbreviated as hts.
+
+    HYDRO_GEN_TS is the set of Hydro projects and hydro timeseries for which
     minimum and average flow are specified. Members of this set can be
-    abbreviated as (project, timeseries) or (g, ts).
+    abbreviated as (project, hydro_timeseries) or (g, hts).
 
     HYDRO_GEN_TPS is the set of Hydro projects and available
     dispatch points. This is a filtered version of GEN_TPS that
     only includes hydro projects.
 
-    hydro_min_flow_mw[(g, ts) in HYDRO_GEN_TS] is a parameter that
+    tp_to_hts[tp in TIMEPOINTS] is a parameter that returns the hydro timeseries
+    for a given timepoint. It is defined in hydro_timepoints.csv and if unspecified
+    it defaults to be equal to tp_ts.
+
+    hydro_min_flow_mw[(g, hts) in HYDRO_GEN_TS] is a parameter that
     determines minimum flow levels, specified in units of MW dispatch.
 
-    hydro_avg_flow_mw[(g, ts) in HYDRO_GEN_TS] is a parameter that
+    hydro_avg_flow_mw[(g, hts) in HYDRO_GEN_TS] is a parameter that
     determines average flow levels, specified in units of MW dispatch.
 
     Enforce_Hydro_Min_Flow[(g, t) in HYDRO_GEN_TPS] is a
     constraint that enforces minimum flow levels for each timepoint.
 
-    Enforce_Hydro_Avg_Flow[(g, ts) in HYDRO_GEN_TS] is a constraint
-    that enforces average flow levels across each timeseries.
-
+    Enforce_Hydro_Avg_Flow[(g, hts) in HYDRO_GEN_TS] is a constraint
+    that enforces average flow levels across each hydro timeseries.
     """
+    mod.tp_to_hts = Param(
+        mod.TIMEPOINTS,
+        input_file="hydro_timepoints.csv",
+        default=lambda m, tp: m.tp_ts[tp],
+        doc="Mapping of timepoints to a hydro series.",
+        within=Any,
+    )
+
+    mod.HYDRO_TS = Set(
+        dimen=1,
+        ordered=False,
+        initialize=lambda m: set(m.tp_to_hts[tp] for tp in m.TIMEPOINTS),
+        doc="Set of hydro timeseries as defined in the mapping.",
+    )
+
+    mod.TPS_IN_HTS = Set(
+        mod.HYDRO_TS,
+        within=mod.TIMEPOINTS,
+        ordered=False,
+        initialize=lambda m, hts: set(t for t in m.TIMEPOINTS if m.tp_to_hts[t] == hts),
+        doc="Set of timepoints in each hydro timeseries",
+    )
 
     mod.HYDRO_GEN_TS_RAW = Set(
         dimen=2,
         input_file="hydro_timeseries.csv",
         input_optional=True,
-        validate=lambda m, g, ts: (g in m.GENERATION_PROJECTS) & (ts in m.TIMESERIES),
+        validate=lambda m, g, hts: (g in m.GENERATION_PROJECTS) & (hts in m.HYDRO_TS),
     )
 
     mod.HYDRO_GENS = Set(
         dimen=1,
         ordered=False,
-        initialize=lambda m: set(g for (g, ts) in m.HYDRO_GEN_TS_RAW),
+        initialize=lambda m: set(g for (g, hts) in m.HYDRO_GEN_TS_RAW),
         doc="Dispatchable hydro projects",
     )
-    mod.HYDRO_GEN_TS = Set(
-        dimen=2,
-        initialize=lambda m: set(
-            (g, m.tp_ts[tp]) for g in m.HYDRO_GENS for tp in m.TPS_FOR_GEN[g]
-        ),
-    )
+
     mod.HYDRO_GEN_TPS = Set(
         initialize=mod.GEN_TPS, filter=lambda m, g, t: g in m.HYDRO_GENS
     )
 
+    mod.HYDRO_GEN_TS = Set(
+        dimen=2,
+        initialize=lambda m: set((g, m.tp_to_hts[tp]) for (g, tp) in m.HYDRO_GEN_TPS),
+    )
+
     # Validate that a timeseries data is specified for every hydro generator /
     # timeseries that we need. Extra data points (ex: outside of planning
     # horizon or beyond a plant's lifetime) can safely be ignored to make it
     # easier to create input files.
     mod.have_minimal_hydro_params = BuildCheck(
-        mod.HYDRO_GEN_TS, rule=lambda m, g, ts: (g, ts) in m.HYDRO_GEN_TS_RAW
+        mod.HYDRO_GEN_TS, rule=lambda m, g, hts: (g, hts) in m.HYDRO_GEN_TS_RAW
     )
 
-    # To do: Add validation check that timeseries data are specified for every
-    # valid timepoint.
+    # Todo: Add validation check that timeseries data are specified for every valid timepoint.
 
     mod.hydro_min_flow_mw = Param(
         mod.HYDRO_GEN_TS_RAW,
@@ -127,16 +165,13 @@ def define_components(mod):
     )
     mod.Enforce_Hydro_Min_Flow = Constraint(
         mod.HYDRO_GEN_TPS,
-        rule=lambda m, g, t: (
-            m.DispatchGen[g, t] >= m.hydro_min_flow_mw[g, m.tp_ts[t]]
-        ),
+        rule=lambda m, g, t: Constraint.Skip
+        if m.hydro_min_flow_mw[g, m.tp_to_hts[t]] == 0
+        else m.DispatchGen[g, t] >= m.hydro_min_flow_mw[g, m.tp_to_hts[t]],
     )
 
     mod.hydro_avg_flow_mw = Param(
-        mod.HYDRO_GEN_TS_RAW,
-        within=NonNegativeReals,
-        input_file="hydro_timeseries.csv",
-        default=0.0,
+        mod.HYDRO_GEN_TS_RAW, within=NonNegativeReals, input_file="hydro_timeseries.csv"
     )
 
     # We use a scaling factor to improve the numerical properties
@@ -146,12 +181,11 @@ def define_components(mod):
     enforce_hydro_avg_flow_scaling_factor = 1e1
     mod.Enforce_Hydro_Avg_Flow = Constraint(
         mod.HYDRO_GEN_TS,
-        rule=lambda m, g, ts: (
-            enforce_hydro_avg_flow_scaling_factor
-            * sum(m.DispatchGen[g, t] for t in m.TPS_IN_TS[ts])
-            / m.ts_num_tps[ts]
-            == m.hydro_avg_flow_mw[g, ts] * enforce_hydro_avg_flow_scaling_factor
-        ),
+        rule=lambda m, g, hts: enforce_hydro_avg_flow_scaling_factor *
+        # Compute the weighted average of the dispatch
+        sum(m.DispatchGen[g, t] * m.tp_weight[t] for t in m.TPS_IN_HTS[hts])
+        / sum(m.tp_weight[tp] for tp in m.TPS_IN_HTS[hts])
+        <= m.hydro_avg_flow_mw[g, hts] * enforce_hydro_avg_flow_scaling_factor,
     )
 
     mod.min_data_check("hydro_min_flow_mw", "hydro_avg_flow_mw")
diff --git a/switch_model/solve.py b/switch_model/solve.py
@@ -108,6 +108,8 @@ def debug(type, value, tb):
         # Patch pyomo if needed, to allow reconstruction of expressions.
         # This must be done before the model is constructed.
         patch_pyomo()
+        patch_to_allow_loading(Set)
+        patch_to_allow_loading(Param)
 
         # Define the model
         model = create_model(modules, args=args)
diff --git a/switch_model/wecc/get_inputs/get_inputs.py b/switch_model/wecc/get_inputs/get_inputs.py
@@ -643,27 +643,52 @@ def query_db(full_config, skip_cf):
     # zone + watershed. Eventually, we may rethink this derating, but it is a reasonable
     # approximation for a large hydro fleet where plant outages are individual random events.
     # Negative flows are replaced by 0.
+    write_csv_from_query(
+        db_cursor,
+        "hydro_timepoints",
+        ["timepoint_id", "tp_to_hts"],
+        f"""
+        SELECT 
+            tp.raw_timepoint_id AS timepoint_id, 
+            p.label || '_M' || date_part('month', timestamp_utc) AS tp_to_hts
+        FROM switch.sampled_timepoint AS tp
+            JOIN switch.period AS p USING(period_id, study_timeframe_id)
+        WHERE time_sample_id = {time_sample_id}
+            AND study_timeframe_id = {study_timeframe_id}
+        ORDER BY 1;
+        """
+    )
+
     write_csv_from_query(
         db_cursor,
         "hydro_timeseries",
         ["hydro_project", "timeseries", "hydro_min_flow_mw", "hydro_avg_flow_mw"],
         f"""
-        select generation_plant_id as hydro_project,
-            {timeseries_id_select},
+        SELECT 
+            generation_plant_id AS hydro_project,
+            hts.hydro_timeseries,
             CASE
                 WHEN hydro_min_flow_mw <= 0 THEN 0
                 ELSE least(hydro_min_flow_mw, capacity_limit_mw * (1-forced_outage_rate)) END,
             CASE
                 WHEN hydro_avg_flow_mw <= 0 THEN 0
                 ELSE least(hydro_avg_flow_mw, capacity_limit_mw * (1-forced_outage_rate)) END
-            as hydro_avg_flow_mw
-        from hydro_historical_monthly_capacity_factors
-            join sampled_timeseries on(month = date_part('month', first_timepoint_utc) and year = date_part('year', first_timepoint_utc))
-            join generation_plant using (generation_plant_id)
+            AS hydro_avg_flow_mw
+        FROM (
+            SELECT DISTINCT
+                date_part('month', tp.timestamp_utc) as month,
+                date_part('year', tp.timestamp_utc) as year, 
+                p.label || '_M' || date_part('month', timestamp_utc) AS hydro_timeseries
+            FROM switch.sampled_timepoint AS tp
+                JOIN switch.period AS p USING(period_id, study_timeframe_id)
+            WHERE time_sample_id = {time_sample_id}
+                AND study_timeframe_id = {study_timeframe_id}
+        ) AS hts
+            JOIN switch.hydro_historical_monthly_capacity_factors USING(month, year)
+            JOIN switch.generation_plant USING(generation_plant_id)
             JOIN temp_generation_plant_ids USING(generation_plant_id)
-        where hydro_simple_scenario_id={hydro_simple_scenario_id}
-            and time_sample_id = {time_sample_id}
-        order by 1;
+        WHERE hydro_simple_scenario_id={hydro_simple_scenario_id}
+        ORDER BY 1;
         """,
     )
 
