Introduction

The past semester I conducted research with Utah State University’s Center for Anticipatory Intelligence, examining the resilience of the Texas natural gas industry. Alongside interviews with five subject matter experts from Baker Tilly, ZGlobal, Vistra Corp, and Pacific Northwest National Lab, we utilized ExaGO to simulate the impact of cyberattacks on Texas critical infrastructure. This blog post will focus on the technical side of our simulations, rather than our overall analysis and findings.

To get started, we’ll need to install some software:

We will also need to download sythetic grid data to manipulate. We’ll use an example based on the Texas Interconnection. A big thank you to the researchers at Texas A&M for their work.

After the the downloaded is finished, unzip it. I like to place the files in their own folder so I don’t clutter my home directory.

unzip ACTIVSg2000.zip -d ~/ACTIVSg2000

Prepare the Data

With MATLAB and MATPOWER installed, open the ACTIVSg2000 directory. We’ll use a simple MATLAB script to randomly turn off a percentage of the total natural gas power plants, simulating a sophisticated cyberattack on a significant natural gas pipeline or multiple gas-fired power plants.

mpc = case_ACTIVSg2000;
ng_indices = find(strcmp(mpc.genfuel, 'ng'));

% Calculate the number of generators to turn off
num_to_turn_off = round(length(ng_indices) * 0.5); 

% Randomly select indices of generators to turn off
turn_off_indices = ng_indices(randperm(length(ng_indices), num_to_turn_off));

% Set status of selected generators to 0 (offline)
mpc.gen(turn_off_indices, 8) = 0;

savecase('my_new_texas_case.m', mpc);

Run that script, and now you should have your scenario saved as my_new_texas_case.m

Run ExaGO

A brief explanation of the command we’re about to run. opflow finds the optimal AC power flow in the overall system, using some really sophisticated math that is way above my pay grade. We include the flag -opflow_include_loadloss_variables because with half of the natual gas plants turned off, we are necessarily going to have to shut off power to (i.e. brownout) over burdened areas. If we don’t include this flag, opflow will fail because the problem is impossible to solve without shedding load. -netfile and -gicfile provide information on grid infrastructure and geospatial location, respectively. Finally, we save the ouput in JSON format to be used in the visualization later.

source spack/share/spack/setup-env.sh
cd ~/ExaGO
opflow -opflow_include_loadloss_variables -netfile ~/ACTIVSg2000/my_new_texas_case.m -gicfile ~/ACTIVSg2000/ACTIVSg2000_GIC_data.gic -save_output -opflow_output_format JSON

After that’s finished running, you should have a new file named opflowout.json in the ExaGO directory.

Visualize the results

It’s time to get to the fun part, visualizing our results!

One slightly annoying bug we’ll need to deal with first, there will be an extra comma in opflowout.json. I’m not sure why this is the case and I’ve opened a bug report. Let’s edit opflowout.json real quick.

vim opflowout.json
# Shift+G to jump to the end of the file
# Delete the extra comma
# :wq to save and exit
cp opflowout.json ~/ExaGO/viz/data

The remaining instructions are basically copy+pasted from the ExaGO visualization and Node.js documentation.

# installs NVM (Node Version Manager)
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash

# download and install Node.js
nvm install 16.13.0

One last step, open a new terminal window:

cd ~/ExaGO/viz
npm install --legacy-peer-deps
python3 geninputfile.py opflowout.json
npm start

Hooray! Your visualization should be available at http://localhost:8080.

I encourage you to mess around and have fun with the MATLAB files and ExaGO, turning off different power plants and testing out different grids. There are a number of other datasets available from Texas A&M researchers, available here. Although this project has made me sleep a little bit more uneasy at night knowing how vulnerable our critical infrastructure is, it’s been very fun to geek out and learn about the confluence of national security, operational technology, and my home state. Thanks for reading!