New system lambda methods

docs/release_notes.rst

@@ -57,6 +57,8 @@ What's New?
       there were no discharge losses.
 
 *  Add support for Python 3.12.
+*  New :meth:`.DispatchModel.redispatch_lambda` and
+   :meth:`.DispatchModel.historical_lambda` to calculate hourly marginal costs.
 
 Bug Fixes
 ^^^^^^^^^

src/dispatch/model.py

@@ -874,6 +874,30 @@ def _cost(self, profiles: pd.DataFrame) -> dict[str, pd.DataFrame]:
         )
         return {"fuel": fuel_cost, "vom": vom_cost, "startup": start_cost, "fom": fom}
 
+    def redispatch_lambda(self) -> pd.DataFrame:
+        """Return hourly marginal cost (aka system lambda) of redispatch."""
+        return (
+            self.dispatchable_cost[["fuel_per_mwh", "vom_per_mwh"]]
+            .sum(axis=1)
+            .reset_index()
+            .pivot(index="datetime", columns=["plant_id_eia", "generator_id"])
+            .droplevel(0, axis=1)
+            .reindex(index=self.load_profile.index, method="ffill")
+            * (self.redispatch > 0).astype(int)
+        ).max(axis=1)
+
+    def historical_lambda(self) -> pd.DataFrame:
+        """Return hourly marginal cost (aka system lambda) of historic dispatch."""
+        return (
+            self.dispatchable_cost[["fuel_per_mwh", "vom_per_mwh"]]
+            .sum(axis=1)
+            .reset_index()
+            .pivot(index="datetime", columns=["plant_id_eia", "generator_id"])
+            .droplevel(0, axis=1)
+            .reindex(index=self.load_profile.index, method="ffill")
+            * (self.dispatchable_profiles > 0).astype(int)
+        ).max(axis=1)
+
     def grouper(
         self,
         df: pd.DataFrame | dict[str, pd.DataFrame],

tests/model_test.py

@@ -188,6 +188,8 @@ def test_storage_summary(self, mini_dm):
             ("storage_capacity", {}, (), "notna"),
             ("hourly_data_check", {}, (), "notna"),
             ("dc_charge", {}, (), "notna"),
+            ("redispatch_lambda", {}, (), "notna"),
+            ("historical_lambda", {}, (), "notna"),
             pytest.param("full_output", {}, (), "notna", marks=pytest.mark.xfail),
             (
                 "dispatchable_summary",
@@ -220,6 +222,8 @@ def test_outputs_parametric(self, ent_dm, func, args, drop_cols, expected):
         """Test that outputs are not empty or do not have unexpected nans."""
         ind, ent_dm = ent_dm
         df = getattr(ent_dm, func)(**args)
+        if isinstance(df, pd.Series):
+            df = df.to_frame(name=func)
         df = df[[c for c in df if c not in drop_cols]]
         if expected == "notna":
             assert df.notna().all().all()
@@ -836,3 +840,14 @@ def test_file(ent_fresh):
     self.plot_year(2008)
 
     raise AssertionError
+
+
+@pytest.mark.skip(reason="for debugging only")
+def test_weird_adj(test_dir):
+    """Investigation of unexpected hourly load adjustment.
+
+    Potential testing for fossil startup to charge storage.
+    """
+    with DataZip(test_dir / "data/bad_adj.zip") as z:
+        dm = DispatchModel(**z["data"], jit=False)
+    dm()