Probabilistic Grid Reliability Analysis with Energy Storage Systems (ProGRESS)

Current release version: 1.0

Release date: TBD

Table of Contents

• Introduction
• Key Features of ProGRESS
• Getting Started
• Data Requirements
• Workflow Description
• Contact

Introduction

The Probabilistic Grid Reliability Analysis with Energy Storage Systems (ProGRESS) software tool is a Python-based open-source tool for assessing the resource adequacy of the evolving electric power grid integrated with energy storage systems (ESS). This tool utilizes a Markov Chain Monte Carlo-based stochastic simulation engine to create diverse scenarios that test the limits of the modern power grid consisting of a high volume of ESSs and variable energy resources (VER). State-of-the-art ESS models are incorporated within the Monte Carlo simulation engine. The charge-discharge dynamics of ESSs, along with their evolving state-of-charge (SOC), are captured by the tool. In addition, ESS failures and repair models are also built into the tool, allowing users to analyze the availability of their ESS devices when they are needed most. ProGRESS also offers the capability of handling the uncertainty associated with VERs, such as, wind and solar power generation resources, enabling the user to simulate thousands of different renewable generation scenarios depending on weather conditions. Users are able to build their own grid models, download and utilize historical VER data using APIs, and analyze magnitude, duration, and frequency of expected future outages. ProGRESS allows users to make informed decisions and plan effectively for VER- and ESS-rich future power systems.

Back to Top

Key Features of ProGRESS

Key features of ProGRESS include:

• Emphasis on Energy Storage Systems: ProGRESS is developed for analyzing the resource adequacy of power systems with a special focus on ESS. This tool offers unique features such as integrating failure and repair models of ESS in resource adequacy evaluation while preserving its charge/discharge dynamics and SOC update characteristics. Future updates to the tool will include sizing of ESS for grid reliability applications, economic analysis of using ESS for these purposes, more detailed ESS reliability models, and other ESS-centric features.

• Stochastic Monte Carlo Simulation Engine: At the core of ProGRESS is a Markov Chain Monte Carlo-based engine that allows users to simulate practically unlimited scenarios involving diverse component failures and weather conditions. Each scenario is considered to be a sample of the Monte Carlo simulation and spans 8760 hours (one year). The users can choose as many samples as they want, the choice typically depending on factors such as system size, convergence criteria, and computational resources of the user.

• Historical VER Data: ProGRESS allows users to conveniently download weather data using APIs. Data related to solar weather is downloaded by ProGRESS from NSRDB while wind-related weather data is downloaded from Wind Integration National Dataset Toolkits. ProGRESS then seamlessly converts the weather data to solar and wind power generation data using in-built functions. Users may utilize their own timeseries VRE generation datasets as well.

• VER Uncertainty Handling: Proper handling of the uncertainty associated with VERs is crucial to accurate resource adequacy assessment and ESS sizing for maintaining grid reliability. ProGRESS uses innovative techniques to quantify uncertainty associated with VERs and ensures that these resources are represented appropriately within the simulation. A k-means clustering technique is used to cluster solar power generation while a transition rate matrix method is used for wind power generation.

• Model Flexibility: Users can currently represent their power systems using a transportation or a copper-sheet model. The copper-sheet model runs significantly faster, especially for larger systems, while the transportation model generates more accurate results.

• Modular Structure: The tool is constructed using an Object-Oriented Programming (OOP) structure and a modular design. This approach enables users to easily modify the backend programs to meet their specific requirements.

• User-friendly Graphical User Interface: The interactive Graphical User Interface (GUI) offered by ProGRESS simplifies the process of input data upload, model building, and results interpretation.

• Parallel Programming Capabilities: The backend includes code for parallel programming (using Python's mpi4py library), allowing users with access to high-performance computing resources to run longer simulations with larger systems for more accurate results. Currently, this functionality is not available through the GUI.

Back to Top

Getting started

Prerequisites

• Python (>= 3.9, <3.12) installed on your system
• Git installed on your system

Installing Python

1. Installers can be found at: https://www.python.org/downloads/release/python-3913/
2. Make sure to check the box "Add Python to PATH" at the bottom of the installer prompt.

Installing Git

• Visit git-scm.com to download Git for your operating system.
• Follow the installation instructions provided on the website.

Setting Up a Virtual Environment

1. Open Command Prompt on Windows or Terminal on macOS and Linux.
2. Install virtualenv (if not already installed):

   python -m pip install virtualenv
   

3. Create a virtual environment:

   cd <your_path>
   python -m virtualenv <env_name>
   

   Replace <your_path> with the path to the folder where you want to create the virtual environment.
4. Activate the virtual environment:
   • On Windows:

     cd <your_path>
     .\<env_name>\Scripts\activate
     

   • On macOS/Linux:

     source <env_name>/bin/activate
     

Installing ProGRESS

1. Clone the Repository:

   git clone https://github.com/sandialabs/snl-progress.git

2. Navigate to the snl_progress Directory:

   cd <path_to_snl-progress>

3. Install Dependencies:

   python -m pip install -r requirements.txt
   

Solver Installation

Ensure an optimization solver is installed on your machine. Solvers to consider include:

Open-source Solvers

• GLPK
• Clp
• HiGHs

Commercial Solvers

• Gurobi
• Cplex

Back to Top

Data Requirements

Users must create a Data folder inside the progress directory with the subfolders System, Solar, and Wind to store system, solar, and wind data, respectively. These subfolders must contain the following data files as shown below:

• README.md
• LICENSE
• progress/
  • Data/
    • System/
      • branch.csv
      • bus.csv
      • gen.csv
      • load.csv
      • storage.csv
    • Solar/
      • solar_sites.csv
      • solar_data.xlsx (if solar power generation data is user provided)
    • Wind/
      • w_power_curves.csv
      • wind_sites.csv
      • windspeed_data.csv (if wind speed data is user provided)

The file names should be kept exactly the same as shown above. The column names inside each .csv file should also be left unchanged. File templates with data from the RTS-GMLC system are provided here: Data. A detailed description of the .csv files along with the column names is provided as follows:

System

branch.csv

Column	Description
Branch ID	Unique branch ID
From Bus	From Bus ID
To Bus	To Bus ID
R	Branch resistance p.u.
X	Branch reactance p.u.
B	Branch line charging susceptance p.u.
MTTTR	Mean Time to Repair
MTTF	Mean Time to Failure
TranOutRate	Outage rates of transmission lines

bus.csv

Column	Description
Bus Name	Bus/Zone name
Bus No.	Numeric bus ID

gen.csv

Column	Description
Gen No.	Numeric gen ID
Gen Name	Generator name
Bus No.	Connection bus/zone number
Tech	Technology Type
Max Cap	Maximum capacity of unit
Min Cap	Minimum capacity of unit
FOR	Forced Outage Rate
MTTR	Mean Time to Repair
MTTF	Mean Time to Failure
Cost	Cost of generation

load.csv

Column	Description
datetime	mm/dd/yy hh:mm
day	mm/dd/yy
time	hh:mm:ss
system_wide	Total load in the entire system
Bus_1	Load in bus/zone 1
Bus_2	Load in bus/zone 2
...	Keep adding columns for all buses

Replace Bus_1, Bus_2, ... with the bus names from the bus.csv.

storage.csv

Column	Description
Name	Name of storage unit
Bus	Bus/zone to which it is connected
Pmax	Maximum power rating
Pmin	Minimum power rating
Duration	Duration in hours
max_SOC	Maximum allowed state-of-charge (0 to 1)
min_SOC	Minimum allowed state-of-charge (0 to 1)
Efficiency	Efficiency (0 to 1)
Discharge Cost	Cost of discharge
Charge Cost	Cost of charge
Units	No. of subunits in the ESS
MTTR	Mean Time to Repair
MTTF	Mean Time to Failure

Solar

solar_sites.csv

Column	Description
site_name	Name of the solar site
lat	latitude of the site
long	longitude of the site
MW	Maximum capacity of the plant in MW
tracking	Single/Dual axis tracking (1 or 2)
Zone	Bus/zone in which site is located

Wind

wind_sites.csv

Column	Description
Farm No.	Numeric wind farm ID
Farm Name	Name of wind farm
Zone Name	Name of zone/bus in which wind farm is located
Zone No.	No. of zone/bus in which wind farm is located
Power Class	Wind power class
Latitude	Latitude of the site
Longitude	Longitude of the site
Hub Height	Height of wind turbines
Turbine Rating	Rating of wind turbines in MW

w_power_curves.csv

Column	Description
Start (m/s)	Starting wind speed for this class in (m/s)
End (m/s)	Ending wind speed for this class in (m/s)
Class 1	Conversion ratio from speed to power for Class 1
Class 2	Conversion ratio from speed to power for Class 2
...	Users can add as many wind classes as they want

Back to Top

Workflow Description

Simulations in the ProGRESS tool can be executed in three different ways. Users may choose to A) utilize the Graphical User Interface to run simulations, B) use Python scripts to run simulations on their local computer or a remote server, or C) use Python and bash scripts to run simulations by utilizing parallel processing capabilities of a High Performance Computer. The following sections describe each approach in detail.

A. Instructions for Running Simulations using the Graphical User Interface

Navigate to the directory where ProGRESS is installed and ensure that the virtual environment is activated. Use the following command to launch the tool:

python -m progress

When the application is first launched, users will see the home page:

Step 1. After pressing the Get Started button, users will be prompted to enter API information. Ensure that you have signed up at the NREL Developer Network beforehand using your details and obtained the required api key. You may skip this step if you plan on using your own data.

Once the API information is saved, users can move on to the Solar tab.

Step 2a. Users may upload their own solar power generation data using the format specified in this file or download solar weather data from NSRDB and convert to solar power generation data using the tool. If downloading data, please check for the data availability at the website since the range of years for which data is available is updated periodically. ProGRESS uses pvlib to convert the downloaded solar weather data to solar power generation data.

Step 2b. The next step involves clustering the solar power generation data. A k-means clustering algorithm is utilized to cluster the data into days with similar solar power generation patterns and values. These clusters are later utilized by the MCS to randomly select days based on the month of the year. Users are able to choose the optimum number of clusters by evaluating the performance of different cluster values. For example, if the user inputs 10 in the No. of Clusters to Evaluate field, the tool will evaluate the performance of clusters starting from 2 to 10. The SSE and silhouette scores will be displayed on the GUI once the evaluation is complete and can be used to make informed decision on the optimal number of clusters.

Step 3. The next step involves adding wind data. Users may choose to upload their own wind speed data using the format specified in this file or download the same from Wind Integration National Dataset Toolkits. The windspeed data can then be used to generate a transition rate matrix using the Process Wind Speed Data button. The transition rate matrix will eventually be used by the MCS to estimate the wind power generation for each hour.

Step 4. Once all data has been added, the user can now run the simulation. Guidelines for adjusting the parameters on this page can be found here. Press the Run Simulation button once all the information is entered. The simulation progress will be displayed on the right side of the page.

Step 5. Users can view the results within the application once the simulation is complete.

Results include reliability indices, plots of hourly load curtailment, hourly solar and wind power generation, hourly energy storage state-of-charge, heat maps of outages across different months of the year and hours of the day. The results will also be available in the Results folder. Some example results are shown below.

B. Instructions for Running Simulations using the Command-Line on Local or Remote Computers/Servers

ProGRESS offers the capability to run simulations through script-based execution without the use of the GUI. These scripts can run on local computers or remote servers. Ensure that you have followed the steps outlined in Getting Started for installing the required software and setting up the environment necessary for running the tool. Then navigate to the progress directory and ensure that the virtual environment is activated.

Step 1. Configure the Input File:

Before running the simulation, configure the input.yaml file with the specific simulation parameters. Open the file in a text editor and adjust the parameters according to your requirements. The api_key, email, affiliation, and name fields are required for downloading weather data from the NSRDB and Wind Integration National Dataset Toolkits. Ensure that you have signed up at the NREL Developer Network beforehand using your details and obtained the required api key. Also check the data availability at the websites as the range of years for which data is available is updated periodically. Guidance on setting the simulation parameters are provided as follows:

Parameter	Comments
samples	This is the number of samples that needs to be run for the MCS to converge and depends heavily on the system. Running a small number of samples (e.g., 10-20) might provide a trend of expected outages in the system, although it is recommended that the users run as many samples as required for the MCS to converge for more accurate results. The convergence can be tracked using the Coefficient of Variation (COV) metric plotted in the COV_track.pdf file, which can be found in the Results folder.
sim_hours	The recommended number is 8760 hours or one full year for each sample.
load_factor	Default value is 1. Users may tune this parameter to check how increasing or decreasing the hourly load profile by a constant factor affects system outages.
model	Users can select a Copper Sheet or a Zonal Model. The Copper Sheet model runs faster but the Zonal Model might be more accurate.

Step 2. Download Weather Data:

Run the data_download_process.py file to download the required solar weather and wind speed data using the NREL API.

python data_download_process.py

Running this file will also process the downloaded weather data and convert them into solar and wind power generation data for each site. Users may skip this step if they want to use their own data or have already downloaded and processed the data during a previous run.

Step 3. Run Monte Carlo Simlulation

The final step would be to run the example_simulation.py file.

python example_simulation.py

Running this file executes the MCS for the pre-specified number of samples and generates results that include values of system reliability indices, outage heatmaps, ESS state-of-charge, solar and wind generation plots, and the coefficient of variation. All results will be stored in the Results folder once the simulation is complete. Please refer to Step 8 of the previous section for more details on results.

C. Instructions for Running Simulations on a High Performance Computer (Parallel Computation)

This approach is strongly recommended for users having access to a HPC system and running the tool for analyzing the reliability of large power systems and/or a large number of samples. The example script (example_simulation_multi_proc.py) provided with the tool utilizes Python's mpi4py library to implement parallel computation. The computation time would depend on the number of nodes (and the number of cores in each node) that the simulation is run on.

Ensure that you have followed the steps outlined in Getting Started for installing the required software and setting up the environment necessary for running the tool on the HPC server. Then follow Step 1 and Step 2 from the previous section to configure the input.yaml file and to download the necessary data, respectively. Step 2 may be skipped if users want to use their own data or have already downloaded and processed the data during a previous run. There are two main ways of running simulations on a HPC and utilize parallel computation capabilities: a) Using an interactive node, or b) using a bash file to schedule a job.

a) Using an Interactive Node:

If using a single interactive node, ensure that you are in the progress directory and execute the following:

mpiexec -n x python example_simulation_multi_proc.py

where x is the number of cores (x < total no. of cores in the node ) you want to utilize.

b) Using a bash file for scheduling and running jobs:

An example bash file is provided here. Users can configure this file according to their requirements and typically schedule a job using the following command:

sbatch example_job.bash

The job will run when it reaches the top of the queue.

Back to Top

Sample Case Study

A test case is included with this tool. The test system is the IEEE RTS-GMLC, which is a modernized version of the IEEE RTS-96. A zonal model of the test system is illustrated as follows:

All test system data provided with the tool has been taken from the RTS-GMLC GitHub repository.

Acknowledgment

The ProGRESS tool is developed and maintained by the Energy Storage Analytics Group at Sandia National Laboratories. This material is based upon work supported by the U.S. Department of Energy, Office of Electricity (OE), Energy Storage Division.

Project team:

• Atri Bera
• Andres Lopez
• Yung-Jai Pomeroy
• Cody Newlun
• Tu Nguyen
• Dilip Pandit

Back to Top

Contact

For reporting bugs and other issues, please use the "Issues" feature of this repository. For more information regarding the tool and collaboration opportunities, please contact project developer: Atri Bera ([email protected])