Merge pull request #175 from upb-lea/dct

inductor optimization fixes

Author: gituser789

docs/wordlist

@@ -450,6 +450,14 @@ NbSetpsPerPeriod
 NbSteps
 WaveformType
 powerloss
+StoReluctanceModelInput
+StoReluctanceModelOutput
+StoFemInput
+StoFemOutput
+IoReluctanceModelInput
+IoReluctanceModelOutput
+IoFemInput
+IoFemOutput
 
 # electrostatic
 capacitive

femmt/optimization/io.py

@@ -18,8 +18,8 @@
 import tqdm
 
 # onw libraries
-from femmt.optimization.io_dtos import (InductorOptimizationDTO, InductorOptimizationTargetAndFixedParameters, FemInput,
-                                        FemOutput, ReluctanceModelInput, ReluctanceModelOutput)
+from femmt.optimization.io_dtos import (InductorOptimizationDTO, InductorOptimizationTargetAndFixedParameters, IoFemInput,
+                                        IoFemOutput, IoReluctanceModelInput, IoReluctanceModelOutput)
 import femmt.functions as ff
 import femmt.functions_reluctance as fr
 import femmt.optimization.ito_functions as itof
@@ -208,7 +208,7 @@ def objective(trial: optuna.Trial, config: InductorOptimizationDTO, target_and_f
                     material_mu_r_abs = material_mu_r_abs_value
                     magnet_material_model = target_and_fixed_parameters.magnet_hub_model_list[count]
 
-            reluctance_model_input = ReluctanceModelInput(
+            reluctance_model_input = IoReluctanceModelInput(
                 core_inner_diameter=core_inner_diameter,
                 window_w=window_w,
                 window_h=window_h,
@@ -222,14 +222,15 @@ def objective(trial: optuna.Trial, config: InductorOptimizationDTO, target_and_f
                 magnet_material_model=magnet_material_model,
 
                 temperature=config.temperature,
+                time_extracted_vec=target_and_fixed_parameters.time_extracted_vec,
                 current_extracted_vec=target_and_fixed_parameters.current_extracted_vec,
                 fundamental_frequency=target_and_fixed_parameters.fundamental_frequency,
                 fft_frequency_list=target_and_fixed_parameters.fft_frequency_list,
                 fft_amplitude_list=target_and_fixed_parameters.fft_amplitude_list,
                 fft_phases_list=target_and_fixed_parameters.fft_phases_list
             )
             try:
-                reluctance_output: ReluctanceModelOutput = InductorOptimization.ReluctanceModel.single_reluctance_model_simulation(reluctance_model_input)
+                reluctance_output: IoReluctanceModelOutput = InductorOptimization.ReluctanceModel.single_reluctance_model_simulation(reluctance_model_input)
             except ValueError as e:
                 logger.debug("bot air gap: No fitting air gap length")
                 return float('nan'), float('nan')
@@ -245,13 +246,13 @@ def objective(trial: optuna.Trial, config: InductorOptimizationDTO, target_and_f
             return reluctance_output.volume, reluctance_output.p_loss_total
 
         @staticmethod
-        def single_reluctance_model_simulation(reluctance_input: ReluctanceModelInput) -> ReluctanceModelOutput:
+        def single_reluctance_model_simulation(reluctance_input: IoReluctanceModelInput) -> IoReluctanceModelOutput:
             """Perform a single reluctance model simulation. E.g. during optimization or during a re-simulation of a single operating point.
 
             :param reluctance_input: Input parameters for the reluctance model simulation.
-            :type reluctance_input: ReluctanceModelInput
+            :type reluctance_input: IoReluctanceModelInput
             :return: Output parameters of the reluctance model simulation
-            :rtype: ReluctanceModelOutput
+            :rtype: IoReluctanceModelOutput
             """
             target_total_reluctance = reluctance_input.turns ** 2 / reluctance_input.target_inductance
 
@@ -263,7 +264,11 @@ def single_reluctance_model_simulation(reluctance_input: ReluctanceModelInput) -
 
             r_air_gap_target = target_total_reluctance - r_core
 
-            flux = reluctance_input.turns * reluctance_input.current_extracted_vec / target_total_reluctance
+            # prepare equidistant flux vector for the magnet loss simulation
+            time_interp = np.linspace(reluctance_input.time_extracted_vec[0], reluctance_input.time_extracted_vec[-1], 1024)
+            current_interp = np.interp(time_interp, reluctance_input.time_extracted_vec, reluctance_input.current_extracted_vec)
+
+            flux = reluctance_input.turns * current_interp / target_total_reluctance
             core_cross_section = (reluctance_input.core_inner_diameter / 2) ** 2 * np.pi
             flux_density = flux / core_cross_section
 
@@ -323,7 +328,7 @@ def single_reluctance_model_simulation(reluctance_input: ReluctanceModelInput) -
 
             p_loss = p_winding + p_core
 
-            reluctance_model_output = ReluctanceModelOutput(
+            reluctance_model_output = IoReluctanceModelOutput(
                 p_loss_total=p_loss,
                 volume=volume,
                 area_to_heat_sink=area_to_heat_sink,
@@ -656,7 +661,7 @@ def full_simulation(df_geometry: pd.DataFrame, current_waveform: list, inductor_
                 # instantiate material-specific model
                 magnet_material_model: mh.loss.LossModel = mh.loss.LossModel(material=material_name, team="paderborn")
 
-                reluctance_model_input = ReluctanceModelInput(
+                reluctance_model_input = IoReluctanceModelInput(
                     core_inner_diameter=core_inner_diameter,
                     window_w=window_w,
                     window_h=df_geometry["params_window_h"][index],
@@ -670,14 +675,15 @@ def full_simulation(df_geometry: pd.DataFrame, current_waveform: list, inductor_
                     magnet_material_model=magnet_material_model,
 
                     temperature=local_config.temperature,
+                    time_extracted_vec=target_and_fix_parameters.time_extracted_vec,
                     current_extracted_vec=target_and_fix_parameters.current_extracted_vec,
                     fundamental_frequency=target_and_fix_parameters.fundamental_frequency,
                     fft_frequency_list=target_and_fix_parameters.fft_frequency_list,
                     fft_amplitude_list=target_and_fix_parameters.fft_amplitude_list,
                     fft_phases_list=target_and_fix_parameters.fft_phases_list
                 )
 
-                reluctance_output: ReluctanceModelOutput = InductorOptimization.ReluctanceModel.single_reluctance_model_simulation(reluctance_model_input)
+                reluctance_output: IoReluctanceModelOutput = InductorOptimization.ReluctanceModel.single_reluctance_model_simulation(reluctance_model_input)
 
                 p_total = reluctance_output.p_loss_total
 
@@ -727,7 +733,7 @@ def fem_simulations_from_reluctance_df(reluctance_df: pd.DataFrame, config: Indu
                             core_inner_diameter = reluctance_df["params_core_inner_diameter"][index].item()
                             window_w = reluctance_df["params_window_w"][index].item()
                             window_h = reluctance_df["params_window_h"][index].item()
-                        fem_input = FemInput(
+                        fem_input = IoFemInput(
                             simulation_name=f"case_{reluctance_df['number'][index].item()}",
                             working_directory=fem_working_directory,
                             core_inner_diameter=core_inner_diameter,
@@ -744,15 +750,17 @@ def fem_simulations_from_reluctance_df(reluctance_df: pd.DataFrame, config: Indu
                             fft_frequency_list=target_and_fix_parameters.fft_frequency_list,
                             fft_amplitude_list=target_and_fix_parameters.fft_amplitude_list,
                             fft_phases_list=target_and_fix_parameters.fft_phases_list,
+                            time_vec=target_and_fix_parameters.time_extracted_vec,
+                            current_vec=target_and_fix_parameters.current_extracted_vec
                         )
 
                         # fem simulation here
                         fem_output = InductorOptimization.FemSimulation.single_fem_simulation(fem_input, False)
 
-                        reluctance_df.loc[index, 'fem_inductance'] = fem_output.fem_inductance
-                        reluctance_df.loc[index, 'fem_p_loss_winding'] = fem_output.fem_p_loss_winding
-                        reluctance_df.loc[index, 'fem_eddy_core'] = fem_output.fem_eddy_core
-                        reluctance_df.loc[index, 'fem_core'] = fem_output.fem_core_total
+                        reluctance_df.loc[index, 'fem_inductance'] = fem_output.inductance
+                        reluctance_df.loc[index, 'fem_p_loss_winding'] = fem_output.p_loss_winding
+                        reluctance_df.loc[index, 'fem_eddy_core'] = fem_output.p_core_sine
+                        reluctance_df.loc[index, 'fem_core'] = fem_output.p_core_magnet
 
                         # copy result files to result-file folder
                         source_json_file = os.path.join(
@@ -855,18 +863,18 @@ def fem_vs_reluctance(df: pd.DataFrame, config: InductorOptimizationDTO):
             return df
 
         @staticmethod
-        def single_fem_simulation(fem_input: FemInput, show_visual_outputs: bool = False) -> FemOutput:
+        def single_fem_simulation(fem_input: IoFemInput, show_visual_outputs: bool = False) -> IoFemOutput:
             """
             Perform a single FEM simulation.
 
             For parallel simulations, use different working directories.
 
             :param fem_input: FEM input DTO
-            :type fem_input: FemInput
+            :type fem_input: IoFemInput
             :param show_visual_outputs: True to show visual outputs
             :type show_visual_outputs: bool
             :return: FEM output DTO
-            :rtype: FemOutput
+            :rtype: IoFemOutput
             """
             # 1. chose simulation type
             geo = fmt.MagneticComponent(simulation_type=fmt.SimulationType.FreqDomain, component_type=fmt.ComponentType.Inductor,
@@ -926,16 +934,33 @@ def single_fem_simulation(fem_input: FemInput, show_visual_outputs: bool = False
             current_amplitudes = [[current] for current in fem_input.fft_amplitude_list]
             phases = [[phase] for phase in fem_input.fft_phases_list]
 
+            # prepare core loss simulation
+            peak_current = np.max(fem_input.current_vec)
+            custom_b_wave = 1 / peak_current * fem_input.current_vec
+            time_interp = np.linspace(fem_input.time_vec[0], fem_input.time_vec[-1], 1024)
+            current_interp = np.interp(time_interp, fem_input.time_vec, fem_input.current_vec)
+            normalized_current_interp = 1 / peak_current * current_interp
+
+            # perform single hysteresis loss simulation where a sinusoidal signal with the peak amplitude is used
+            geo.single_simulation(freq=fem_input.fundamental_frequency, current=[peak_current],
+                                  plot_interpolation=False, show_fem_simulation_results=show_visual_outputs)
+
+            # read the flux per mesh cell and transfer it into losses with the help of the magnet model
+            hyst_losses_custom = geo.calc_hystersis_losses_with_MagNet_model_PB_based_on_mesh_results(b_wave=fmt.WaveformType.Custom,
+                                                                                                      custom_b_wave=normalized_current_interp)
+
+            result_dict_hyst = geo.read_log()
+
+            # get the winding losses
             geo.excitation_sweep(frequency_list=fem_input.fft_frequency_list, current_list_list=current_amplitudes,
                                  phi_deg_list_list=phases, show_last_fem_simulation=show_visual_outputs)
-
             result_dict = geo.read_log()
 
-            fem_output = FemOutput(
-                fem_inductance=result_dict['single_sweeps'][0]['winding1']['flux_over_current'][0],
-                fem_p_loss_winding=result_dict['total_losses']['winding1']['total'],
-                fem_eddy_core=result_dict['total_losses']['eddy_core'],
-                fem_core_total=result_dict['total_losses']['core'],
+            fem_output = IoFemOutput(
+                inductance=result_dict['single_sweeps'][0]['winding1']['flux_over_current'][0],
+                p_loss_winding=result_dict['total_losses']['winding1']['total'],
+                p_core_sine=result_dict_hyst['total_losses']['core'],
+                p_core_magnet=hyst_losses_custom,
                 volume=result_dict["misc"]["core_2daxi_total_volume"]
             )
             return fem_output
@@ -1017,7 +1042,7 @@ def full_simulation(df_geometry: pd.DataFrame, current_waveform: list, inductor_
 
             material_mu_r_abs = np.abs(np.array([mu_real_at_f + 1j * mu_imag_at_f]))
 
-            fem_input = FemInput(
+            fem_input = IoFemInput(
                 # general parameters
                 working_directory=working_directory,
                 simulation_name='xx',
@@ -1040,7 +1065,10 @@ def full_simulation(df_geometry: pd.DataFrame, current_waveform: list, inductor_
                 fundamental_frequency=target_and_fix_parameters.fundamental_frequency,
                 fft_frequency_list=target_and_fix_parameters.fft_frequency_list,
                 fft_amplitude_list=target_and_fix_parameters.fft_amplitude_list,
-                fft_phases_list=target_and_fix_parameters.fft_phases_list
+                fft_phases_list=target_and_fix_parameters.fft_phases_list,
+
+                time_vec=target_and_fix_parameters.time_extracted_vec,
+                current_vec=target_and_fix_parameters.current_extracted_vec
             )
 
             fem_output = InductorOptimization.FemSimulation.single_fem_simulation(fem_input, False)
@@ -1051,7 +1079,7 @@ def full_simulation(df_geometry: pd.DataFrame, current_waveform: list, inductor_
             # instantiate material-specific model
             magnet_material_model: mh.loss.LossModel = mh.loss.LossModel(material=material_name, team="paderborn")
 
-            reluctance_model_input = ReluctanceModelInput(
+            reluctance_model_input = IoReluctanceModelInput(
                 core_inner_diameter=core_inner_diameter,
                 window_w=window_w,
                 window_h=df_geometry["params_window_h"][index_number],
@@ -1065,31 +1093,32 @@ def full_simulation(df_geometry: pd.DataFrame, current_waveform: list, inductor_
                 magnet_material_model=magnet_material_model,
 
                 temperature=local_config.temperature,
+                time_extracted_vec=target_and_fix_parameters.time_extracted_vec,
                 current_extracted_vec=target_and_fix_parameters.current_extracted_vec,
                 fundamental_frequency=target_and_fix_parameters.fundamental_frequency,
                 fft_frequency_list=target_and_fix_parameters.fft_frequency_list,
                 fft_amplitude_list=target_and_fix_parameters.fft_amplitude_list,
-                fft_phases_list=target_and_fix_parameters.fft_phases_list
+                fft_phases_list=target_and_fix_parameters.fft_phases_list,
             )
 
-            reluctance_output: ReluctanceModelOutput = InductorOptimization.ReluctanceModel.single_reluctance_model_simulation(reluctance_model_input)
+            reluctance_output: IoReluctanceModelOutput = InductorOptimization.ReluctanceModel.single_reluctance_model_simulation(reluctance_model_input)
 
-            p_core = reluctance_output.p_hyst + fem_output.fem_eddy_core
-            p_total = p_core + fem_output.fem_p_loss_winding
+            p_core = reluctance_output.p_hyst + fem_output.p_core_sine
+            p_total = p_core + fem_output.p_loss_winding
 
             if print_derivations:
                 logger.info(f"Inductance reluctance: {local_config.target_inductance}")
-                logger.info(f"Inductance FEM: {fem_output.fem_inductance}")
+                logger.info(f"Inductance FEM: {fem_output.inductance}")
                 logger.info(f"Inductance derivation: "
-                            f"{(fem_output.fem_inductance - local_config.target_inductance) / local_config.target_inductance * 100} %")
+                            f"{(fem_output.inductance - local_config.target_inductance) / local_config.target_inductance * 100} %")
                 logger.info(f"Volume reluctance: {reluctance_output.volume}")
                 logger.info(f"Volume FEM: {fem_output.volume}")
                 logger.info(f"Volume derivation: {(reluctance_output.volume - fem_output.volume) / reluctance_output.volume * 100} %")
                 logger.info(f"P_winding reluctance: {reluctance_output.p_winding}")
-                logger.info(f"P_winding FEM: {fem_output.fem_p_loss_winding}")
-                logger.info(f"P_winding derivation: {(fem_output.fem_p_loss_winding - reluctance_output.p_winding) / fem_output.fem_p_loss_winding * 100} %")
+                logger.info(f"P_winding FEM: {fem_output.p_loss_winding}")
+                logger.info(f"P_winding derivation: {(fem_output.p_loss_winding - reluctance_output.p_winding) / fem_output.p_loss_winding * 100} %")
                 logger.info(f"P_hyst reluctance: {reluctance_output.p_hyst}")
-                logger.info(f"P_hyst FEM: {fem_output.fem_core_total}")
-                logger.info(f"P_hyst derivation: {(reluctance_output.p_hyst - fem_output.fem_core_total) / reluctance_output.p_hyst * 100} %")
+                logger.info(f"P_hyst FEM: {fem_output.p_core_magnet}")
+                logger.info(f"P_hyst derivation: {(reluctance_output.p_hyst - fem_output.p_core_magnet) / reluctance_output.p_hyst * 100} %")
 
-            return reluctance_output.volume, p_total, reluctance_output.area_to_heat_sink, fem_output.fem_p_loss_winding, p_core
+            return reluctance_output.volume, p_total, reluctance_output.area_to_heat_sink, fem_output.p_loss_winding, p_core

femmt/optimization/io_dtos.py

@@ -68,7 +68,7 @@ class InductorOptimizationTargetAndFixedParameters:
     fft_phases_list: list[float]
 
 @dataclasses.dataclass
-class FemInput:
+class IoFemInput:
     """Input DTO for a FEM simulation within the inductor optimization."""
 
     # general parameters
@@ -95,18 +95,22 @@ class FemInput:
     fft_amplitude_list: list[float]
     fft_phases_list: list[float]
 
+    # magnet loss model
+    time_vec: list[float]
+    current_vec: list[float]
+
 @dataclasses.dataclass
-class FemOutput:
+class IoFemOutput:
     """Output DTO for a FEM simulation within the inductor optimization."""
 
-    fem_inductance: float
-    fem_p_loss_winding: float
-    fem_eddy_core: float
-    fem_core_total: float
+    inductance: float
+    p_loss_winding: float
+    p_core_sine: float
+    p_core_magnet: float
     volume: float
 
 @dataclasses.dataclass
-class ReluctanceModelInput:
+class IoReluctanceModelInput:
     """Input DTO for reluctance model simulation within the inductor optimization."""
 
     target_inductance: float
@@ -122,14 +126,15 @@ class ReluctanceModelInput:
     magnet_material_model: LossModel
 
     temperature: float
+    time_extracted_vec: list
     current_extracted_vec: list
     fundamental_frequency: float
     fft_frequency_list: list[float]
     fft_amplitude_list: list[float]
     fft_phases_list: list[float]
 
 @dataclasses.dataclass
-class ReluctanceModelOutput:
+class IoReluctanceModelOutput:
     """output DTO for reluctance model simulation within the inductor optimization."""
 
     volume: float