flixopt 2.2.0b0__py3-none-any.whl → 3.0.0__py3-none-any.whl

flixopt/calculation.py

@@ -1,22 +1,24 @@
 """
 This module contains the Calculation functionality for the flixopt framework.
-It is used to calculate a SystemModel for a given FlowSystem through a solver.
+It is used to calculate a FlowSystemModel for a given FlowSystem through a solver.
 There are three different Calculation types:
-    1. FullCalculation: Calculates the SystemModel for the full FlowSystem
-    2. AggregatedCalculation: Calculates the SystemModel for the full FlowSystem, but aggregates the TimeSeriesData.
+    1. FullCalculation: Calculates the FlowSystemModel for the full FlowSystem
+    2. AggregatedCalculation: Calculates the FlowSystemModel for the full FlowSystem, but aggregates the TimeSeriesData.
         This simplifies the mathematical model and usually speeds up the solving process.
-    3. SegmentedCalculation: Solves a SystemModel for each individual Segment of the FlowSystem.
+    3. SegmentedCalculation: Solves a FlowSystemModel for each individual Segment of the FlowSystem.
 """
 
+from __future__ import annotations
+
 import logging
 import math
 import pathlib
 import timeit
 import warnings
-from typing import Any, Dict, List, Optional, Union
+from collections import Counter
+from typing import TYPE_CHECKING, Annotated, Any
 
 import numpy as np
-import pandas as pd
 import yaml
 
 from . import io as fx_io
@@ -24,13 +26,18 @@ from . import utils as utils
 from .aggregation import AggregationModel, AggregationParameters
 from .components import Storage
 from .config import CONFIG
-from .core import Scalar
-from .elements import Component
+from .core import DataConverter, Scalar, TimeSeriesData, drop_constant_arrays
 from .features import InvestmentModel
 from .flow_system import FlowSystem
 from .results import CalculationResults, SegmentedCalculationResults
-from .solvers import _Solver
-from .structure import SystemModel, copy_and_convert_datatypes, get_compact_representation
+
+if TYPE_CHECKING:
+    import pandas as pd
+    import xarray as xr
+
+    from .elements import Component
+    from .solvers import _Solver
+    from .structure import FlowSystemModel
 
 logger = logging.getLogger('flixopt')
 
@@ -38,53 +45,63 @@ logger = logging.getLogger('flixopt')
 class Calculation:
     """
     class for defined way of solving a flow_system optimization
+
+    Args:
+        name: name of calculation
+        flow_system: flow_system which should be calculated
+        folder: folder where results should be saved. If None, then the current working directory is used.
+        normalize_weights: Whether to automatically normalize the weights (periods and scenarios) to sum up to 1 when solving.
+        active_timesteps: Deprecated. Use FlowSystem.sel(time=...) or FlowSystem.isel(time=...) instead.
     """
 
     def __init__(
         self,
         name: str,
         flow_system: FlowSystem,
-        selected_timesteps: Optional[pd.DatetimeIndex] = None,
-        selected_scenarios: Optional[pd.Index] = None,
-        folder: Optional[pathlib.Path] = None,
-        active_timesteps: Optional[pd.DatetimeIndex] = None,
+        active_timesteps: Annotated[
+            pd.DatetimeIndex | None,
+            'DEPRECATED: Use flow_system.sel(time=...) or flow_system.isel(time=...) instead',
+        ] = None,
+        folder: pathlib.Path | None = None,
+        normalize_weights: bool = True,
     ):
-        """
-        Args:
-            name: name of calculation
-            flow_system: flow_system which should be calculated
-            selected_timesteps: timesteps which should be used for calculation. If None, then all timesteps are used.
-            selected_scenarios: scenarios which should be used for calculation. If None, then all scenarios are used.
-            folder: folder where results should be saved. If None, then the current working directory is used.
-            active_timesteps: Deprecated. Use selected_timesteps instead.
-        """
+        self.name = name
+        if flow_system.used_in_calculation:
+            logger.warning(
+                f'This FlowSystem is already used in a calculation:\n{flow_system}\n'
+                f'Creating a copy of the FlowSystem for Calculation "{self.name}".'
+            )
+            flow_system = flow_system.copy()
+
         if active_timesteps is not None:
             warnings.warn(
-                'active_timesteps is deprecated. Use selected_timesteps instead.',
+                "The 'active_timesteps' parameter is deprecated and will be removed in a future version. "
+                'Use flow_system.sel(time=timesteps) or flow_system.isel(time=indices) before passing '
+                'the FlowSystem to the Calculation instead.',
                 DeprecationWarning,
                 stacklevel=2,
             )
-            selected_timesteps = active_timesteps
-        self.name = name
+            flow_system = flow_system.sel(time=active_timesteps)
+        self._active_timesteps = active_timesteps  # deprecated
+        self.normalize_weights = normalize_weights
+
+        flow_system._used_in_calculation = True
+
         self.flow_system = flow_system
-        self.model: Optional[SystemModel] = None
-        self.selected_timesteps = selected_timesteps
-        self.selected_scenarios = selected_scenarios
+        self.model: FlowSystemModel | None = None
 
         self.durations = {'modeling': 0.0, 'solving': 0.0, 'saving': 0.0}
         self.folder = pathlib.Path.cwd() / 'results' if folder is None else pathlib.Path(folder)
-        self.results: Optional[CalculationResults] = None
+        self.results: CalculationResults | None = None
+
+        if self.folder.exists() and not self.folder.is_dir():
+            raise NotADirectoryError(f'Path {self.folder} exists and is not a directory.')
+        self.folder.mkdir(parents=False, exist_ok=True)
 
-        if not self.folder.exists():
-            try:
-                self.folder.mkdir(parents=False)
-            except FileNotFoundError as e:
-                raise FileNotFoundError(
-                    f'Folder {self.folder} and its parent do not exist. Please create them first.'
-                ) from e
+        self._modeled = False
 
     @property
-    def main_results(self) -> Dict[str, Union[Scalar, Dict]]:
+    def main_results(self) -> dict[str, Scalar | dict]:
         from flixopt.features import InvestmentModel
 
         main_results = {
@@ -92,9 +109,9 @@ class Calculation:
             'Penalty': self.model.effects.penalty.total.solution.values,
             'Effects': {
                 f'{effect.label} [{effect.unit}]': {
-                    'operation': effect.model.operation.total.solution.values,
-                    'invest': effect.model.invest.total.solution.values,
-                    'total': effect.model.total.solution.values,
+                    'temporal': effect.submodel.temporal.total.solution.values,
+                    'periodic': effect.submodel.periodic.total.solution.values,
+                    'total': effect.submodel.total.solution.values,
                 }
                 for effect in self.flow_system.effects
             },
@@ -102,39 +119,38 @@ class Calculation:
                 'Invested': {
                     model.label_of_element: model.size.solution
                     for component in self.flow_system.components.values()
-                    for model in component.model.all_sub_models
-                    if isinstance(model, InvestmentModel) and model.size.solution.max() >= CONFIG.modeling.EPSILON
+                    for model in component.submodel.all_submodels
+                    if isinstance(model, InvestmentModel) and model.size.solution.max() >= CONFIG.Modeling.epsilon
                 },
                 'Not invested': {
                     model.label_of_element: model.size.solution
                     for component in self.flow_system.components.values()
-                    for model in component.model.all_sub_models
-                    if isinstance(model, InvestmentModel) and model.size.solution.max() < CONFIG.modeling.EPSILON
+                    for model in component.submodel.all_submodels
+                    if isinstance(model, InvestmentModel) and model.size.solution.max() < CONFIG.Modeling.epsilon
                 },
             },
             'Buses with excess': [
                 {
                     bus.label_full: {
-                        'input': bus.model.excess_input.solution.sum('time'),
-                        'output': bus.model.excess_output.solution.sum('time'),
+                        'input': bus.submodel.excess_input.solution.sum('time'),
+                        'output': bus.submodel.excess_output.solution.sum('time'),
                     }
                 }
                 for bus in self.flow_system.buses.values()
                 if bus.with_excess
                 and (
-                    bus.model.excess_input.solution.sum() > 1e-3
-                    or bus.model.excess_output.solution.sum() > 1e-3
+                    bus.submodel.excess_input.solution.sum() > 1e-3 or bus.submodel.excess_output.solution.sum() > 1e-3
                 )
             ],
         }
 
-        return utils.round_floats(main_results)
+        return utils.round_nested_floats(main_results)
 
     @property
     def summary(self):
         return {
             'Name': self.name,
-            'Number of timesteps': len(self.flow_system.time_series_collection.timesteps),
+            'Number of timesteps': len(self.flow_system.timesteps),
             'Calculation Type': self.__class__.__name__,
             'Constraints': self.model.constraints.ncons,
             'Variables': self.model.variables.nvars,
@@ -146,29 +162,72 @@ class Calculation:
     @property
     def active_timesteps(self) -> pd.DatetimeIndex:
         warnings.warn(
-            'active_timesteps is deprecated. Use selected_timesteps instead.',
+            'active_timesteps is deprecated. Use flow_system.sel(time=...) or flow_system.isel(time=...) instead.',
             DeprecationWarning,
             stacklevel=2,
         )
-        return self.selected_timesteps
+        return self._active_timesteps
+
+    @property
+    def modeled(self) -> bool:
+        return True if self.model is not None else False
 
 
 class FullCalculation(Calculation):
     """
-    class for defined way of solving a flow_system optimization
+    FullCalculation solves the complete optimization problem using all time steps.
+
+    This is the most comprehensive calculation type that considers every time step
+    in the optimization, providing the most accurate but computationally intensive solution.
+
+    Args:
+        name: name of calculation
+        flow_system: flow_system which should be calculated
+        folder: folder where results should be saved. If None, then the current working directory is used.
+        normalize_weights: Whether to automatically normalize the weights (periods and scenarios) to sum up to 1 when solving.
+        active_timesteps: Deprecated. Use FlowSystem.sel(time=...) or FlowSystem.isel(time=...) instead.
     """
 
-    def do_modeling(self) -> SystemModel:
+    def do_modeling(self) -> FullCalculation:
         t_start = timeit.default_timer()
-        self._activate_time_series()
+        self.flow_system.connect_and_transform()
 
-        self.model = self.flow_system.create_model()
+        self.model = self.flow_system.create_model(self.normalize_weights)
         self.model.do_modeling()
 
         self.durations['modeling'] = round(timeit.default_timer() - t_start, 2)
-        return self.model
+        return self
+
+    def fix_sizes(self, ds: xr.Dataset, decimal_rounding: int | None = 5) -> FullCalculation:
+        """Fix the sizes of the calculations to specified values.
+
+        Args:
+            ds: The dataset that contains the variable names mapped to their sizes. If None, the dataset is loaded from the results.
+            decimal_rounding: The number of decimal places to round the sizes to. If no rounding is applied, numerical errors might lead to infeasibility.
+        """
+        if not self.modeled:
+            raise RuntimeError('Model was not created. Call do_modeling() first.')
+        if decimal_rounding is not None:
+            ds = ds.round(decimal_rounding)
+
+        for name, da in ds.data_vars.items():
+            if '|size' not in name:
+                continue
+            if name not in self.model.variables:
+                logger.debug(f'Variable {name} not found in calculation model. Skipping.')
+                continue
+
+            con = self.model.add_constraints(
+                self.model[name] == da,
+                name=f'{name}-fixed',
+            )
+            logger.debug(f'Fixed "{name}":\n{con}')
+
+        return self
 
-    def solve(self, solver: _Solver, log_file: Optional[pathlib.Path] = None, log_main_results: bool = True):
+    def solve(
+        self, solver: _Solver, log_file: pathlib.Path | None = None, log_main_results: bool = True
+    ) -> FullCalculation:
         t_start = timeit.default_timer()
 
         self.model.solve(
@@ -191,11 +250,10 @@ class FullCalculation(Calculation):
 
         # Log the formatted output
         if log_main_results:
-            logger.info(f'{" Main Results ":#^80}')
             logger.info(
-                '\n'
+                f'{" Main Results ":#^80}\n'
                 + yaml.dump(
-                    utils.round_floats(self.main_results),
+                    utils.round_nested_floats(self.main_results),
                     default_flow_style=False,
                     sort_keys=False,
                     allow_unicode=True,
@@ -205,16 +263,30 @@ class FullCalculation(Calculation):
 
         self.results = CalculationResults.from_calculation(self)
 
-    def _activate_time_series(self):
-        self.flow_system.transform_data()
-        self.flow_system.time_series_collection.set_selection(
-            timesteps=self.selected_timesteps, scenarios=self.selected_scenarios
-        )
+        return self
 
 
 class AggregatedCalculation(FullCalculation):
     """
-    class for defined way of solving a flow_system optimization
+    AggregatedCalculation reduces computational complexity by clustering time series into typical periods.
+
+    This calculation approach aggregates time series data using clustering techniques (tsam) to identify
+    representative time periods, significantly reducing computation time while maintaining solution accuracy.
+
+    Note:
+        The quality of the solution depends on the choice of aggregation parameters.
+        The optimal parameters depend on the specific problem and the characteristics of the time series data.
+        For more information, refer to the [tsam documentation](https://tsam.readthedocs.io/en/latest/).
+
+    Args:
+        name: Name of the calculation
+        flow_system: FlowSystem to be optimized
+        aggregation_parameters: Parameters for aggregation. See AggregationParameters class documentation
+        components_to_clusterize: list of Components to perform aggregation on. If None, all components are aggregated.
+            This equalizes variables in the components according to the typical periods computed in the aggregation
+        active_timesteps: DatetimeIndex of timesteps to use for calculation. If None, all timesteps are used
+        folder: Folder where results should be saved. If None, current working directory is used
+        aggregation: contains the aggregation model
     """
 
     def __init__(
@@ -222,47 +294,35 @@ class AggregatedCalculation(FullCalculation):
         name: str,
         flow_system: FlowSystem,
         aggregation_parameters: AggregationParameters,
-        components_to_clusterize: Optional[List[Component]] = None,
-        selected_timesteps: Optional[pd.DatetimeIndex] = None,
-        folder: Optional[pathlib.Path] = None,
+        components_to_clusterize: list[Component] | None = None,
+        active_timesteps: Annotated[
+            pd.DatetimeIndex | None,
+            'DEPRECATED: Use flow_system.sel(time=...) or flow_system.isel(time=...) instead',
+        ] = None,
+        folder: pathlib.Path | None = None,
     ):
-        """
-        Class for Optimizing the `FlowSystem` including:
-            1. Aggregating TimeSeriesData via typical periods using tsam.
-            2. Equalizing variables of typical periods.
-        Args:
-            name: name of calculation
-            flow_system: flow_system which should be calculated
-            aggregation_parameters: Parameters for aggregation. See documentation of AggregationParameters class.
-            components_to_clusterize: List of Components to perform aggregation on. If None, then all components are aggregated.
-                This means, teh variables in the components are equalized to each other, according to the typical periods
-                computed in the DataAggregation
-            selected_timesteps: pd.DatetimeIndex or None
-                list with indices, which should be used for calculation. If None, then all timesteps are used.
-            folder: folder where results should be saved. If None, then the current working directory is used.
-        """
-        if flow_system.time_series_collection.scenarios is not None:
+        if flow_system.scenarios is not None:
             raise ValueError('Aggregation is not supported for scenarios yet. Please use FullCalculation instead.')
-        super().__init__(name, flow_system, selected_timesteps, folder=folder)
+        super().__init__(name, flow_system, active_timesteps, folder=folder)
         self.aggregation_parameters = aggregation_parameters
         self.components_to_clusterize = components_to_clusterize
         self.aggregation = None
 
-    def do_modeling(self) -> SystemModel:
+    def do_modeling(self) -> AggregatedCalculation:
         t_start = timeit.default_timer()
-        self._activate_time_series()
+        self.flow_system.connect_and_transform()
         self._perform_aggregation()
 
         # Model the System
-        self.model = self.flow_system.create_model()
+        self.model = self.flow_system.create_model(self.normalize_weights)
         self.model.do_modeling()
-        # Add Aggregation Model after modeling the rest
+        # Add Aggregation Submodel after modeling the rest
         self.aggregation = AggregationModel(
             self.model, self.aggregation_parameters, self.flow_system, self.aggregation, self.components_to_clusterize
         )
         self.aggregation.do_modeling()
         self.durations['modeling'] = round(timeit.default_timer() - t_start, 2)
-        return self.model
+        return self
 
     def _perform_aggregation(self):
         from .aggregation import Aggregation
@@ -270,41 +330,34 @@ class AggregatedCalculation(FullCalculation):
         t_start_agg = timeit.default_timer()
 
         # Validation
-        dt_min, dt_max = (
-            np.min(self.flow_system.time_series_collection.hours_per_timestep),
-            np.max(self.flow_system.time_series_collection.hours_per_timestep),
-        )
+        dt_min = float(self.flow_system.hours_per_timestep.min().item())
+        dt_max = float(self.flow_system.hours_per_timestep.max().item())
         if not dt_min == dt_max:
             raise ValueError(
                 f'Aggregation failed due to inconsistent time step sizes:'
                 f'delta_t varies from {dt_min} to {dt_max} hours.'
             )
-        steps_per_period = (
-            self.aggregation_parameters.hours_per_period
-            / self.flow_system.time_series_collection.hours_per_timestep.max()
-        )
-        is_integer = (
-            self.aggregation_parameters.hours_per_period
-            % self.flow_system.time_series_collection.hours_per_timestep.max()
-        ).item() == 0
-        if not (steps_per_period.size == 1 and is_integer):
+        ratio = self.aggregation_parameters.hours_per_period / dt_max
+        if not np.isclose(ratio, round(ratio), atol=1e-9):
             raise ValueError(
                 f'The selected {self.aggregation_parameters.hours_per_period=} does not match the time '
-                f'step size of {dt_min} hours). It must be a multiple of {dt_min} hours.'
+                f'step size of {dt_max} hours. It must be an integer multiple of {dt_max} hours.'
             )
 
         logger.info(f'{"":#^80}')
         logger.info(f'{" Aggregating TimeSeries Data ":#^80}')
 
+        ds = self.flow_system.to_dataset()
+
+        temporaly_changing_ds = drop_constant_arrays(ds, dim='time')
+
         # Aggregation - creation of aggregated timeseries:
         self.aggregation = Aggregation(
-            original_data=self.flow_system.time_series_collection.as_dataset(
-                with_extra_timestep=False, with_constants=False
-            ).to_dataframe(),
+            original_data=temporaly_changing_ds.to_dataframe(),
             hours_per_time_step=float(dt_min),
             hours_per_period=self.aggregation_parameters.hours_per_period,
             nr_of_periods=self.aggregation_parameters.nr_of_periods,
-            weights=self.flow_system.time_series_collection.calculate_aggregation_weights(),
+            weights=self.calculate_aggregation_weights(temporaly_changing_ds),
             time_series_for_high_peaks=self.aggregation_parameters.labels_for_high_peaks,
             time_series_for_low_peaks=self.aggregation_parameters.labels_for_low_peaks,
         )
@@ -312,15 +365,155 @@ class AggregatedCalculation(FullCalculation):
         self.aggregation.cluster()
         self.aggregation.plot(show=True, save=self.folder / 'aggregation.html')
         if self.aggregation_parameters.aggregate_data_and_fix_non_binary_vars:
-            for col in self.aggregation.aggregated_data.columns:
-                data = self.aggregation.aggregated_data[col].values
-                if col in self.flow_system.time_series_collection._has_extra_timestep:
-                    data = np.append(data, data[-1])
-                self.flow_system.time_series_collection.update_time_series(col, data)
+            ds = self.flow_system.to_dataset()
+            for name, series in self.aggregation.aggregated_data.items():
+                da = (
+                    DataConverter.to_dataarray(series, self.flow_system.coords)
+                    .rename(name)
+                    .assign_attrs(ds[name].attrs)
+                )
+                if TimeSeriesData.is_timeseries_data(da):
+                    da = TimeSeriesData.from_dataarray(da)
+
+                ds[name] = da
+
+            self.flow_system = FlowSystem.from_dataset(ds)
+        self.flow_system.connect_and_transform()
         self.durations['aggregation'] = round(timeit.default_timer() - t_start_agg, 2)
 
+    @classmethod
+    def calculate_aggregation_weights(cls, ds: xr.Dataset) -> dict[str, float]:
+        """Calculate weights for all datavars in the dataset. Weights are pulled from the attrs of the datavars."""
+
+        groups = [da.attrs['aggregation_group'] for da in ds.data_vars.values() if 'aggregation_group' in da.attrs]
+        group_counts = Counter(groups)
+
+        # Calculate weight for each group (1/count)
+        group_weights = {group: 1 / count for group, count in group_counts.items()}
+
+        weights = {}
+        for name, da in ds.data_vars.items():
+            group_weight = group_weights.get(da.attrs.get('aggregation_group'))
+            if group_weight is not None:
+                weights[name] = group_weight
+            else:
+                weights[name] = da.attrs.get('aggregation_weight', 1)
+
+        if np.all(np.isclose(list(weights.values()), 1, atol=1e-6)):
+            logger.info('All Aggregation weights were set to 1')
+
+        return weights
+
 
 class SegmentedCalculation(Calculation):
+    """Solve large optimization problems by dividing time horizon into (overlapping) segments.
+
+    This class addresses memory and computational limitations of large-scale optimization
+    problems by decomposing the time horizon into smaller overlapping segments that are
+    solved sequentially. Each segment uses final values from the previous segment as
+    initial conditions, ensuring dynamic continuity across the solution.
+
+    Key Concepts:
+        **Temporal Decomposition**: Divides long time horizons into manageable segments
+        **Overlapping Windows**: Segments share timesteps to improve storage dynamics
+        **Value Transfer**: Final states of one segment become initial states of the next
+        **Sequential Solving**: Each segment solved independently but with coupling
+
+    Limitations and Constraints:
+        **Investment Parameters**: InvestParameters are not supported in segmented calculations
+        as investment decisions must be made for the entire time horizon, not per segment.
+
+        **Global Constraints**: Time-horizon-wide constraints (flow_hours_total_min/max,
+        load_factor_min/max) may produce suboptimal results as they cannot be enforced
+        globally across segments.
+
+        **Storage Dynamics**: While overlap helps, storage optimization may be suboptimal
+        compared to full-horizon solutions due to limited foresight in each segment.
+
+    Args:
+        name: Unique identifier for the calculation, used in result files and logging.
+        flow_system: The FlowSystem to optimize, containing all components, flows, and buses.
+        timesteps_per_segment: Number of timesteps in each segment (excluding overlap).
+            Must be > 2 to avoid internal side effects. Larger values provide better
+            optimization at the cost of memory and computation time.
+        overlap_timesteps: Number of additional timesteps added to each segment.
+            Improves storage optimization by providing lookahead. Higher values
+            improve solution quality but increase computational cost.
+        nr_of_previous_values: Number of previous timestep values to transfer between
+            segments for initialization. Typically 1 is sufficient.
+        folder: Directory for saving results. Defaults to current working directory + 'results'.
+
+    Examples:
+        Annual optimization with monthly segments:
+
+        ```python
+        # 8760 hours annual data with monthly segments (730 hours) and 48-hour overlap
+        segmented_calc = SegmentedCalculation(
+            name='annual_energy_system',
+            flow_system=energy_system,
+            timesteps_per_segment=730,  # ~1 month
+            overlap_timesteps=48,  # 2 days overlap
+            folder=Path('results/segmented'),
+        )
+        segmented_calc.do_modeling_and_solve(solver='gurobi')
+        ```
+
+        Weekly optimization with daily overlap:
+
+        ```python
+        # Weekly segments for detailed operational planning
+        weekly_calc = SegmentedCalculation(
+            name='weekly_operations',
+            flow_system=industrial_system,
+            timesteps_per_segment=168,  # 1 week (hourly data)
+            overlap_timesteps=24,  # 1 day overlap
+            nr_of_previous_values=1,
+        )
+        ```
+
+        Large-scale system with minimal overlap:
+
+        ```python
+        # Large system with minimal overlap for computational efficiency
+        large_calc = SegmentedCalculation(
+            name='large_scale_grid',
+            flow_system=grid_system,
+            timesteps_per_segment=100,  # Shorter segments
+            overlap_timesteps=5,  # Minimal overlap
+        )
+        ```
+
+    Design Considerations:
+        **Segment Size**: Balance between solution quality and computational efficiency.
+        Larger segments provide better optimization but require more memory and time.
+
+        **Overlap Duration**: More overlap improves storage dynamics and reduces
+        end-effects but increases computational cost. Typically 5-10% of segment length.
+
+        **Storage Systems**: Systems with large storage components benefit from longer
+        overlaps to capture charge/discharge cycles effectively.
+
+        **Investment Decisions**: Use FullCalculation for problems requiring investment
+        optimization, as SegmentedCalculation cannot handle investment parameters.
+
+    Common Use Cases:
+        - **Annual Planning**: Long-term planning with seasonal variations
+        - **Large Networks**: Spatially or temporally large energy systems
+        - **Memory-Limited Systems**: When full optimization exceeds available memory
+        - **Operational Planning**: Detailed short-term optimization with limited foresight
+        - **Sensitivity Analysis**: Quick approximate solutions for parameter studies
+
+    Performance Tips:
+        - Start with FullCalculation and use this class if memory issues occur
+        - Use longer overlaps for systems with significant storage
+        - Monitor solution quality at segment boundaries for discontinuities
+
+    Warning:
+        The evaluation of the solution is a bit more complex than FullCalculation or AggregatedCalculation
+        due to the overlapping individual solutions.
+
+    """
+
     def __init__(
         self,
         name: str,
@@ -328,47 +521,25 @@ class SegmentedCalculation(Calculation):
         timesteps_per_segment: int,
         overlap_timesteps: int,
         nr_of_previous_values: int = 1,
-        folder: Optional[pathlib.Path] = None,
+        folder: pathlib.Path | None = None,
     ):
-        """
-        Dividing and Modeling the problem in (overlapping) segments.
-        The final values of each Segment are recognized by the following segment, effectively coupling
-        charge_states and flow_rates between segments.
-        Because of this intersection, both modeling and solving is done in one step
-
-        Take care:
-        Parameters like InvestParameters, sum_of_flow_hours and other restrictions over the total time_series
-        don't really work in this Calculation. Lower bounds to such SUMS can lead to weird results.
-        This is NOT yet explicitly checked for...
-
-        Args:
-            name: name of calculation
-            flow_system: flow_system which should be calculated
-            timesteps_per_segment: The number of time_steps per individual segment (without the overlap)
-            overlap_timesteps: The number of time_steps that are added to each individual model. Used for better
-                results of storages)
-            folder: folder where results should be saved. If None, then the current working directory is used.
-        """
         super().__init__(name, flow_system, folder=folder)
         self.timesteps_per_segment = timesteps_per_segment
         self.overlap_timesteps = overlap_timesteps
         self.nr_of_previous_values = nr_of_previous_values
-        self.sub_calculations: List[FullCalculation] = []
-
-        self.all_timesteps = self.flow_system.time_series_collection._full_timesteps
-        self.all_timesteps_extra = self.flow_system.time_series_collection._full_timesteps_extra
+        self.sub_calculations: list[FullCalculation] = []
 
         self.segment_names = [
             f'Segment_{i + 1}' for i in range(math.ceil(len(self.all_timesteps) / self.timesteps_per_segment))
         ]
-        self.selected_timesteps_per_segment = self._calculate_timesteps_of_segment()
+        self._timesteps_per_segment = self._calculate_timesteps_per_segment()
 
         assert timesteps_per_segment > 2, 'The Segment length must be greater 2, due to unwanted internal side effects'
         assert self.timesteps_per_segment_with_overlap <= len(self.all_timesteps), (
             f'{self.timesteps_per_segment_with_overlap=} cant be greater than the total length {len(self.all_timesteps)}'
         )
 
-        self.flow_system._connect_network()  # Connect network to ensure that all FLows know their Component
+        self.flow_system._connect_network()  # Connect network to ensure that all Flows know their Component
         # Storing all original start values
         self._original_start_values = {
             **{flow.label_full: flow.previous_flow_rate for flow in self.flow_system.flows.values()},
@@ -378,106 +549,120 @@ class SegmentedCalculation(Calculation):
                 if isinstance(comp, Storage)
             },
         }
-        self._transfered_start_values: List[Dict[str, Any]] = []
-
-    def do_modeling_and_solve(
-        self, solver: _Solver, log_file: Optional[pathlib.Path] = None, log_main_results: bool = False
-    ):
-        logger.info(f'{"":#^80}')
-        logger.info(f'{" Segmented Solving ":#^80}')
+        self._transfered_start_values: list[dict[str, Any]] = []
 
+    def _create_sub_calculations(self):
         for i, (segment_name, timesteps_of_segment) in enumerate(
-            zip(self.segment_names, self.selected_timesteps_per_segment, strict=False)
+            zip(self.segment_names, self._timesteps_per_segment, strict=True)
         ):
-            if self.sub_calculations:
-                self._transfer_start_values(i)
+            calc = FullCalculation(f'{self.name}-{segment_name}', self.flow_system.sel(time=timesteps_of_segment))
+            calc.flow_system._connect_network()  # Connect to have Correct names of Flows!
 
+            self.sub_calculations.append(calc)
             logger.info(
                 f'{segment_name} [{i + 1:>2}/{len(self.segment_names):<2}] '
                 f'({timesteps_of_segment[0]} -> {timesteps_of_segment[-1]}):'
             )
 
-            calculation = FullCalculation(
-                f'{self.name}-{segment_name}', self.flow_system, selected_timesteps=timesteps_of_segment
+    def do_modeling_and_solve(
+        self, solver: _Solver, log_file: pathlib.Path | None = None, log_main_results: bool = False
+    ) -> SegmentedCalculation:
+        logger.info(f'{"":#^80}')
+        logger.info(f'{" Segmented Solving ":#^80}')
+        self._create_sub_calculations()
+
+        for i, calculation in enumerate(self.sub_calculations):
+            logger.info(
+                f'{self.segment_names[i]} [{i + 1:>2}/{len(self.segment_names):<2}] '
+                f'({calculation.flow_system.timesteps[0]} -> {calculation.flow_system.timesteps[-1]}):'
             )
-            self.sub_calculations.append(calculation)
+
+            if i > 0 and self.nr_of_previous_values > 0:
+                self._transfer_start_values(i)
+
             calculation.do_modeling()
-            invest_elements = [
-                model.label_full
-                for component in self.flow_system.components.values()
-                for model in component.model.all_sub_models
-                if isinstance(model, InvestmentModel)
-            ]
-            if invest_elements:
-                logger.critical(
-                    f'Investments are not supported in Segmented Calculation! '
-                    f'Following InvestmentModels were found: {invest_elements}'
-                )
+
+            # Warn about Investments, but only in fist run
+            if i == 0:
+                invest_elements = [
+                    model.label_full
+                    for component in calculation.flow_system.components.values()
+                    for model in component.submodel.all_submodels
+                    if isinstance(model, InvestmentModel)
+                ]
+                if invest_elements:
+                    logger.critical(
+                        f'Investments are not supported in Segmented Calculation! '
+                        f'Following InvestmentModels were found: {invest_elements}'
+                    )
+
             calculation.solve(
                 solver,
                 log_file=pathlib.Path(log_file) if log_file is not None else self.folder / f'{self.name}.log',
                 log_main_results=log_main_results,
             )
 
-        self._reset_start_values()
-
         for calc in self.sub_calculations:
             for key, value in calc.durations.items():
                 self.durations[key] += value
 
         self.results = SegmentedCalculationResults.from_calculation(self)
 
-    def _transfer_start_values(self, segment_index: int):
+        return self
+
+    def _transfer_start_values(self, i: int):
         """
         This function gets the last values of the previous solved segment and
         inserts them as start values for the next segment
         """
-        timesteps_of_prior_segment = self.selected_timesteps_per_segment[segment_index - 1]
+        timesteps_of_prior_segment = self.sub_calculations[i - 1].flow_system.timesteps_extra
 
-        start = self.selected_timesteps_per_segment[segment_index][0]
+        start = self.sub_calculations[i].flow_system.timesteps[0]
         start_previous_values = timesteps_of_prior_segment[self.timesteps_per_segment - self.nr_of_previous_values]
         end_previous_values = timesteps_of_prior_segment[self.timesteps_per_segment - 1]
 
         logger.debug(
-            f'start of next segment: {start}. indices of previous values: {start_previous_values}:{end_previous_values}'
+            f'Start of next segment: {start}. Indices of previous values: {start_previous_values} -> {end_previous_values}'
         )
+        current_flow_system = self.sub_calculations[i - 1].flow_system
+        next_flow_system = self.sub_calculations[i].flow_system
+
         start_values_of_this_segment = {}
-        for flow in self.flow_system.flows.values():
-            flow.previous_flow_rate = flow.model.flow_rate.solution.sel(
+
+        for current_flow in current_flow_system.flows.values():
+            next_flow = next_flow_system.flows[current_flow.label_full]
+            next_flow.previous_flow_rate = current_flow.submodel.flow_rate.solution.sel(
                 time=slice(start_previous_values, end_previous_values)
             ).values
-            start_values_of_this_segment[flow.label_full] = flow.previous_flow_rate
-        for comp in self.flow_system.components.values():
-            if isinstance(comp, Storage):
-                comp.initial_charge_state = comp.model.charge_state.solution.sel(time=start).item()
-                start_values_of_this_segment[comp.label_full] = comp.initial_charge_state
+            start_values_of_this_segment[current_flow.label_full] = next_flow.previous_flow_rate
 
-        self._transfered_start_values.append(start_values_of_this_segment)
+        for current_comp in current_flow_system.components.values():
+            next_comp = next_flow_system.components[current_comp.label_full]
+            if isinstance(next_comp, Storage):
+                next_comp.initial_charge_state = current_comp.submodel.charge_state.solution.sel(time=start).item()
+                start_values_of_this_segment[current_comp.label_full] = next_comp.initial_charge_state
 
-    def _reset_start_values(self):
-        """This resets the start values of all Elements to its original state"""
-        for flow in self.flow_system.flows.values():
-            flow.previous_flow_rate = self._original_start_values[flow.label_full]
-        for comp in self.flow_system.components.values():
-            if isinstance(comp, Storage):
-                comp.initial_charge_state = self._original_start_values[comp.label_full]
+        self._transfered_start_values.append(start_values_of_this_segment)
 
-    def _calculate_timesteps_of_segment(self) -> List[pd.DatetimeIndex]:
-        selected_timesteps_per_segment = []
+    def _calculate_timesteps_per_segment(self) -> list[pd.DatetimeIndex]:
+        timesteps_per_segment = []
         for i, _ in enumerate(self.segment_names):
             start = self.timesteps_per_segment * i
             end = min(start + self.timesteps_per_segment_with_overlap, len(self.all_timesteps))
-            selected_timesteps_per_segment.append(self.all_timesteps[start:end])
-        return selected_timesteps_per_segment
+            timesteps_per_segment.append(self.all_timesteps[start:end])
+        return timesteps_per_segment
 
     @property
     def timesteps_per_segment_with_overlap(self):
         return self.timesteps_per_segment + self.overlap_timesteps
 
     @property
-    def start_values_of_segments(self) -> Dict[int, Dict[str, Any]]:
+    def start_values_of_segments(self) -> list[dict[str, Any]]:
         """Gives an overview of the start values of all Segments"""
-        return {
-            0: {element.label_full: value for element, value in self._original_start_values.items()},
-            **{i: start_values for i, start_values in enumerate(self._transfered_start_values, 1)},
-        }
+        return [{name: value for name, value in self._original_start_values.items()}] + [
+            start_values for start_values in self._transfered_start_values
+        ]
+
+    @property
+    def all_timesteps(self) -> pd.DatetimeIndex:
+        return self.flow_system.timesteps