Analyzing LFAC Upgrades Across Multiple Time Points

time-series for Variable-Frequency OPF

Project description

The project is based on the Variablefrequency-opf project. In the project, we use the time-series data form the dataset of RTS_GMLC system. (add the link)

flowchart In the project, we basically parse the RTSGMLC data with the default RTSGMLC network. (here I named the output dictionary “hourdata”). Each entry of the “hourdata” is a whole copy of the original RTSGMLC dictionary combined with the time-series data at that specific time point. Then we generate the system of subnetworks (hourmndata) using makemndata. These networks are the data for optimization. We input each time step of hourmndata into multifrequencyopf by a for-loop and store the output in a similar structure as "hourmndata".

After we have the optimization results, the data can be plotted by using the function “multi_plot”, with flexibility on range of data and mode of plot.

Function description & Examples

1. hour_parse:

hour_parse(month_st, day_st, period_st, month_en, day_en, period_en, folder)

This function uses the base network “RTSGMLC.m”, the network file information (subnetworks.csv in 'folder') and the time-series data (in folder) to generate networks for optimization for a single case. This function is used in `houropf` function described below in this project, so it will not shown in the complete example at the end of this chapter.

Arguments:

month_st: the month to start
day_st: the day to start
period_st: the hour to start
month_en: the month that ends parsing
day_en: the month that ends parsing
period_en: the month that ends parsing

(here the periods need to be chosen corresponding to the data set use in the future)

folder: the folder containing the time-sereis data and subnetworks.csv of one single case (For example I can set it to be time_series/enum_net/br1)

2.1. houropf (houropf_modified):

function hour_opf(month_st, day_st, period_st, month_en, day_en, period_en,
      folder:: String,
      subnet_folder:: String,
      output_folder::String,
      obj::String,
      gen_areas = [],
      area_transfer = [],
      plot_results::Bool = true,
      print_results::Bool = false)

This function uses the similar structure of multi-frequency-opf with some additional arguments. Inside the function, we input each time step of hourmndata into multifrequencyopf by a for-loop and store the output in a similar data structure as hourmndata. The output data is then stored into .json file in the specified `outputfolder. The output files are of three types: -bindingconstraints: the cosntriants -optnet: optimal network result of that sub-network case during the specific time period -res_summary`: optimal network result of that sub-network case during the specific time period The names are also specified to subnet numbers and periods. For exmaple,

res_summary_br10_(1.1.1__12.31.24).json

is the res_summary file for branch 10 optimized during the period of 1AM, Jan 1st to 12PM Dec 31st.

hour_opf is in default to run for the whole year period for the convinience when plot using function multi_plot specified below.

hour_opf_modified is of the same format regrading the input and output. It is only for testing purposes.

Arguments:

month_st, day_st, period_st, month_en, day_en, period_en: the time points to specify tim periods for plotting
folder::String:: the directory containing all subnetwork data, subnetworks.csv, interfaces.csv
subne_folder: the number of subnet that is to be optimized (which directs to the sub-folder containing its data)
output_folder: the folder specified to save the output files
plot_results::Bool = true: choice of plot the optimization result (optimal value vs time points) directly after optimization, default false

Below are the similar arguments as multifrequency-opf:

obj::String: the objective function to use, from the following:
- "mincost": minimize generation cost
- "areagen": minimize generation in the areas specified in gen_areas
- "minredispatch": minimize the change in generator dispatch from the initial values defined in the network data
gen_areas: integer array of all areas in which generation should be minimized if obj=="mincost"
area_transfer: two areas with power transfer between them that should be saved and plotted. Results for P, Q, S, and loss are saved for power transfer between the two areas. Must have exactly two elements.
print_results::Bool: if true, print the optimization results

Examples:

hour_opf(1, 1, 1, 1, 1, 3, "./time_series", "br10", "./time_series_out", "mincost", false, false)

This example runs optimization for 3 hours at the beginning of the year. The case is branch 10, the input files are from folder ./time_series and output to ./time_series_out. This is the same for hour_opf_modified

2.2. houropfregularized:

function hour_opf(month_st, day_st, period_st, month_en, day_en, period_en,
      folder:: String,
      subnet_folder:: String,
      output_folder::String,
      obj::String,
      gen_areas = [],
      area_transfer = [],
      plot_results::Bool = true,
      print_results::Bool = false
      regularize_f::Float64=0.0,
      ipopt_max_iter::Int64=20000)

Similar structure as hour_opf, but added 2 more arguments. This function is performing optimization with regularization on the optimal frequency. Larger the differences between the optimal frequency and 60 Hz (the original frequency), larger the penalty. Below are the additional arguments:

regularize_f: the coefficient of regularizer.
ipopt_max_iter: the amount of iterations of the server when solving each time point.

Examples:

hour_opf_regularized(1, 1, 1, 1, 1, 3, "./time_series", "br10", "./time_series_out", "mincost", false, false, 1e0, 20000)

3. multiplot (same for multiplot_new):

function multi_plot(month_st, day_st, period_st, month_en, day_en, period_en, br_arr_raw, folder, output_folder,
      plot_choice::String = "single")

The function uses the period of plotting, the information of networks to plot (the output files from houropf or other data files of similar data structures) and choice of plotting. It is flexible on the types of plots, the range to plot and the data sets to investigate. Separate plot from the process of running the data can help to improve the efficiency. The input folder is the output folder of the `houropf`.

Arguments:

month_st, day_st, period_st, month_en, day_en, period_en: the start and end time for the period of data to plot
br_arr_raw: the array containg the indexex of subnets to plot (in format of Int64). If using the single for plot_choice, the function only reads in the first index of network if the choice is single.
folder: the target folder containg the data files for plot (is the folder containing the output of the hour_opf)
output_folder: the folder specified to save the output files
plot_choice::String = "single": default single,
- "single": some plots of a single upgrade in a time period: optimal result & frequency of a network, improvements compared to the base case
- "compare": comparisons of several upgrades on different aspects: average optimal cost; improvement rates; histogram of improvement rates

Examples:

multi_plot(8, 1, 1, 8, 15, 24, [12 24 41 119], "time_series_out", "time_series_out", "compare")

This plots the comparison of histogram of improvement rates of several upgrades. The result plot is shown as below.

A complete example for the whole process from parse to optimize to plot is:

# Here I run the whole year data for default, then I have all the data and don't need to run it for other periods in the future
hour_opf(1, 1, 1, 12, 31, 24, "./time_series", "br10", "./time_series_out", "mincost", false, false)     

# In this function I set the input folder of multi_plot to be the output folder of the hour_opf.
# The specified time period is 8.1.1 - 8.15.24 and the subnetx number is 10.
multi_plot(8, 1, 1, 8, 15, 24, [10], "time_series_out", "time_series_out", "single")