GitHub

Data Structure

Input Data

Case

PowerModelsADA uses a dictionary of dictionaries to store the case data and the subproblem data. The case data dictionary is similar to the one in PowerModels with area assignment for each bus. The buses in the data dictionary must contain an area key with more than one distinct area ID. The subproblem data is similar to the case data with additional information. The area data contains the area-specific data and the distributed algorithm parameters.

To load a data file, we use parse_file function as follow:

case_path = "test/data/case14.m"
data = parse_file(case_path)

Partitioning

To check the areas ID in a data dictionary, use get_areas_id(data) to get all areas' IDs in data. If the data dictionary doesn't contain more than one area, we can partition the system manually using assign_area! function. An example of partition file is shown in partition example.

PowerModelsADA.assign_area! — Function

assign area to the system data using a dictionary with (bus => area) integers pairs

source

assign area to the system data using a CVS file with buses and area ID

source

assign area to the system data using a vector with (bus => area) pairs

source

assign area to the system data using a matrix with [bus, area] columnsor rows

source

Before running the distributed algorithm, PowerModelsADA internally decomposes the original system into subsystems using decompose_system function. the function decouples the tie-lines between two areas by introducing dummy buses and virtual generators at the tie-lines' ends.

Output Data

The output of the distributed algorithms is stored in a dictionary. The dictionary's keys are the areas ID, and the dictionary's values are the areas data dictionary with the results stored in solution dictionary.

Saving Iterations Data

To save a specific data during the distributed algorithm (e.g., store the "shared_variable" dictionary each iteration), use the option save_data::Vector{String}=[] in the solve function and add the key of the data (e.g., save_data=["shared variable"]). The output of the solve function will contain a dictionary with a key called "previous_solution" that contains vectors of the selected stored data ordered by the iteration number.

Generation Cost

To calculate the objective function of the central algorithm use calc_dist_gen_cost.

To compare the distributed algorithm objective function value with the central OPF, use compare_solution to get the absolute value of the relative error.