Controller

Basic Controller

In the following the basic class Controller of pandapower is listed, which should serve as a platform for user-defined control implementations.

class Controller(net, in_service=True, order=0, level=0, index=None, recycle=False, drop_same_existing_ctrl=False, initial_run=True, overwrite=False, matching_params=None, **kwargs)

Base-Class of all controllable elements within a network.

add_controller_to_net(net, in_service, initial_run, order, level, index, recycle, drop_same_existing_ctrl, overwrite, **kwargs)

adds the controller to net[‘controller’] dataframe.

INPUT:

in_service (bool) - in service status

order (int) - order

index (int) - index

recycle (bool) - if controller needs a new bbm (ppc, Ybus…) or if it can be used with prestored values. This is mostly needed for time series calculations

control_step(net)

If the is_converged method returns false, the control_step will be called. In other words: if the controller did not converge yet, this method should implement actions that promote convergence e.g. adapting actuating variables and writing them back to the data structure.

finalize_control(net)

Some controller require extended finalization. This method is being called at the end of a loadflow. It is a separate method from restore_init_state because it is possible that control finalization does not only restore the init state but also something in addition to that, that would require the results in net

finalize_step(net, time)

Note

This method is ONLY being called during time-series simulation!

After each time step, this method is being called to clean things up or similar. The OutputWriter is a class specifically designed to store results of the loadflow. If the ControlHandler.output_writer got an instance of this class, it will be called before the finalize step.

initialize_control(net)

Some controller require extended initialization in respect to the current state of the net (or their view of it). This method is being called after an initial loadflow but BEFORE any control strategies are being applied.

This method may be interesting if you are aiming for a global controller or if it has to be aware of its initial state.

is_converged(net)

This method calculated whether or not the controller converged. This is where any target values are being calculated and compared to the actual measurements. Returns convergence of the controller.

repair_control(net)

Some controllers can cause net to not converge. In this case, they can implement a method to try and catch the load flow error by altering some values in net, for example load scaling. This method is being called in the except block in run_control. Either implement this in a controller that is likely to cause the error, or define a special “load flow police” controller for your use case

restore_init_state(net)

Some controllers manipulate values in net and then restore them back to initial values, e.g. DistributedSlack. This method should be used for such a purpose because it is executed in the except block of run_control to make sure that the net condition is restored even if load flow calculation doesn’t converge

set_active(net, in_service)

Sets the controller in or out of service

set_recycle(net)

Checks the recyclability of this controller and changes the recyclability of the control handler if necessary. With this a faster time series calculation can be achieved since not everything must be recalculated.

Beware: Setting recycle wrong can mess up your results. Set it to False in init if in doubt!

time_step(net, time)

It is the first call in each time step, thus suited for things like reading profiles or prepare the controller for the next control step.

Note

This method is ONLY being called during time-series simulation!

ConstControl

As already mentioned at the beginning of the chapter “Controller Simulation”, the ConstControl controller is intended for use in time series simulation. This is used to read the data from a DataSource and write it to a network.

class ConstControl(net, element, variable, element_index, profile_name=None, data_source=None, scale_factor=1.0, in_service=True, recycle=True, order=-1, level=-1, drop_same_existing_ctrl=False, matching_params=None, initial_run=False, **kwargs)

Class representing a generic time series controller for a specified element and variable. Control strategy: “No Control” -> updates values of specified elements according to timeseries input data. If ConstControl is used without timeseries input data, it will reset the controlled values to the initial values, preserving the initial net state. The timeseries values are written to net during time_step before the initial powerflow run and before other controllers’ control_step. It is possible to set attributes of objects that are contained in a net table, e.g. attributes of other controllers. This can be helpful e.g. if a voltage setpoint of a transformer tap changer depends on the time step. An attribute of an object in the “object” column of a table (e.g. net.controller[“object”] -> net.controller.object.at[0, “vm_set_pu”] can be set if the attribute is specified as “object.attribute” (e.g. “object.vm_set_pu”).

INPUT:

net (attrdict) - The net in which the controller resides

element - element table (‘sgen’, ‘load’ etc.)

variable - variable (‘p_mw’, ‘q_mvar’, ‘vm_pu’, ‘tap_pos’ etc.)

element_index (int[]) - IDs of the controlled elements

data_source (obj) - The data source that provides profile data

profile_name (str[]) - The profile names of the elements in the data source

OPTIONAL:

scale_factor (real, 1.0) - Scaling factor for time series input values

in_service (bool, True) - Indicates if the controller is currently in_service

recycle (bool, True) - Re-use of internal-data in a time series loop.

drop_same_existing_ctrl (bool, False) - Indicates if already existing controllers of the same type and with the same matching parameters (e.g. at same element) should be dropped

Note

If multiple elements are represented with one controller, the data source must have integer columns. At the moment, only the DFData format is tested for the multiple const control.

control_step(net)

Set applied to True, which means that the values set in time_step have been included in the load flow calculation.

is_converged(net)

Actual implementation of the convergence criteria: If controller is applied, it can stop

set_recycle(net)

Checks the recyclability of this controller and changes the recyclability of the control handler if necessary. With this a faster time series calculation can be achieved since not everything must be recalculated.

Beware: Setting recycle wrong can mess up your results. Set it to False in init if in doubt!

time_step(net, time)

Get the values of the element from data source Write to pandapower net by calling write_to_net() If ConstControl is used without a data_source, it will reset the controlled values to the initial values, preserving the initial net state.