Merge branch '347_curtailment_with_TSA_xarray' into 'develop'

allow calculation of curtailment when using timeseries agrregation and xarray

Closes #347

See merge request iek-3/shared-code/fine!373

fine/IOManagement/dictIO.py

@@ -2,9 +2,27 @@
 
 import fine as fn
 from fine.IOManagement import utilsIO
+from fine.utils import buildFullTimeSeries
 
 
-def exportToDict(esM, useProcessedValues=False):
+def reconstruct_full_timeseries(esM, timeseries, ip):
+    print("Reconstructing timeseries from TSA")
+
+    # switch first index level and column level
+    df = timeseries.copy()
+    df = df.stack().unstack(level=1)
+    df.index.names = [None] * len(df.index.names)
+    full_df = (
+        buildFullTimeSeries(df, esM.periodsOrder[ip], ip=ip, esM=esM, divide=False)
+        .reset_index(level=0, drop=True)
+        .T
+    )
+    full_df.columns = timeseries.columns
+
+    return full_df
+
+
+def exportToDict(esM, useProcessedValues=False, useTSAvalues=False):
     """
     Writes the input arguments of EnergySysteModel and its Components input to a dictionary.
 
@@ -88,6 +106,58 @@ def exportToDict(esM, useProcessedValues=False):
                         compDict[classname][componentname][prop] = getattr(
                             component, prop
                         )
+                # Add aggregatedRate timeseries from TSA
+                if esM.isTimeSeriesDataClustered:
+                    prop_list_full_set = component.__dict__.keys()
+                    for prop in prop_list_full_set:
+                        if (prop != "self") and (prop != "esM"):
+                            if ("aggregated" in prop) and ("Rate" in prop):
+                                timeseries = getattr(component, prop)
+                                # if only one time series was given by user, independent of the number of investment periods, we only save that
+                                original_name = prop.replace("aggregated", "")
+                                original_name = (
+                                    f"{original_name[:1].lower()}{original_name[1:]}"
+                                )
+                                inuptTimeSeries = getattr(component, original_name)
+
+                                if isinstance(inuptTimeSeries, dict):
+                                    compDict[classname][componentname][prop] = {}
+                                    for ip in timeseries.keys():
+                                        ip_name = esM.investmentPeriodNames[ip]
+                                        # get years
+                                        if timeseries[ip] is not None:
+                                            compDict[classname][componentname][prop][
+                                                ip_name
+                                            ] = reconstruct_full_timeseries(
+                                                esM, timeseries[ip], ip=ip
+                                            )
+                                        else:
+                                            compDict[classname][componentname][prop] = (
+                                                None
+                                            )
+                                else:
+                                    ip = 0
+                                    if isinstance(timeseries, dict):
+                                        if (
+                                            timeseries[ip] is not None
+                                        ):  # we only save the first time series since they are all the same (becaue only one time series given by user)
+                                            compDict[classname][componentname][prop] = (
+                                                reconstruct_full_timeseries(
+                                                    esM, timeseries[ip], ip=ip
+                                                )
+                                            )
+                                        else:
+                                            compDict[classname][componentname][prop] = (
+                                                None
+                                            )
+                                    elif timeseries is not None:
+                                        compDict[classname][componentname][prop] = (
+                                            reconstruct_full_timeseries(
+                                                esM, timeseries, ip=ip
+                                            )
+                                        )
+                                    else:
+                                        compDict[classname][componentname][prop] = None
 
     return esmDict, compDict
 
@@ -110,8 +180,20 @@ def importFromDict(esmDict, compDict):
     for classname in compDict:
         # get class
         class_ = getattr(fn, classname)
+        blacklist = [
+            "aggregated"
+        ]  # variable is only needed to save clusterd timeseries data
 
         for comp in compDict[classname]:
+            # get all vars that start with entries in blacklist
+            compBlacklist = [
+                var
+                for var in compDict[classname][comp]
+                if any([var.startswith(bl) for bl in blacklist])
+            ]
+            # remove all vars in blacklist
+            for var in compBlacklist:
+                compDict[classname][comp].pop(var)
             esM.add(class_(esM, **compDict[classname][comp]))
 
     return esM

test/IOManagement/test_xarrayio.py

@@ -118,7 +118,9 @@ def compare_esm_outputs(esm_1: fn.EnergySystemModel, esm_2: fn.energySystemModel
             model_results_1_sorted = model_results_1.sort_index()
             model_results_2_sorted = model_results_2.sort_index()
 
-            assert_frame_equal(model_results_1_sorted, model_results_2_sorted, check_dtype=False)
+            assert_frame_equal(
+                model_results_1_sorted, model_results_2_sorted, check_dtype=False
+            )
 
 
 def test_esm_input_to_dataset_and_back(minimal_test_esM):
@@ -280,3 +282,28 @@ def test_transmission_dims(minimal_test_esM):
     assert operationRateMax.index.name == "time"
 
     esM2.optimize()
+
+
+def test_saving_clustered_timeseries_to_xarray(perfectForesight_test_esM):
+    """Optimize an esM, write it to  netCDF, then load the esM from this file.
+    Compare if both esMs are identical. Inputs are compared with exportToDict,
+    outputs are compared with optimizationSummary.
+    """
+
+    esm_original_pf = deepcopy(perfectForesight_test_esM)
+    esm_original_pf.aggregateTemporally(
+        numberOfTypicalPeriods=1, numberOfTimeStepsPerPeriod=2
+    )
+    esm_original_pf.optimize()
+
+    xrIO.writeEnergySystemModelToNetCDF(
+        esm_original_pf, outputFilePath="test_esM_pf.nc"
+    )
+    esm_datasets = xrIO.writeEnergySystemModelToDatasets(esm_original_pf)
+    assert "ts_aggregatedOperationRateMax" in esm_datasets["Input"]["Source"]["PV"]
+
+    esm_pf_from_netcdf = xrIO.readNetCDFtoEnergySystemModel(filePath="test_esM_pf.nc")
+
+    compare_esm_inputs(esm_original_pf, esm_pf_from_netcdf)
+
+    Path("test_esM_pf.nc").unlink()