RATLLE — Risk Assessment Tool for Large Load Induced Events

Update -6/25/2026 - PSS/E UDMs for injecting forced oscillations available now in the UDM folder. Can be used outside the RATLLE framework.

A Python-based script suite for evaluating bulk power system reliability risk from oscillations introduced by large dynamic digital loads (LDDLs), using PSS/E as the simulation engine. Tested with PSS/E v35 and WECC planning cases in v34. Revised - 5/6/2026.

The risk assessment script suite is divided into three modules, designed for pre-screening vulnerable locations, running time-series simulations, and analyzing simulation outputs (Figure 1). The scripts are written in a simplified modular function-style in Python, so that they can be easily understood and modified to suit a utility’s needs.

**To cite this work:**

Biswas, Shuchismita, Antos Varghese, Kaustav Chatterjee, Brett Ross, and Jim Follum. "pnnl/LL-risk-assessment (33525-E)." US Department of Energy (DOE) Software (2026): 68.

S. Biswas, A. C. Varghese, K. Chatterjee, S. Nekkalapu, B. Ross and J. Follum, "Evaluating the Risk to Bulk Power System Reliability from Large Load Induced Oscillations," 2026 IEEE/PES Transmission and Distribution Conference and Exposition (T&D), Chicago, IL, USA, 2026, pp. 1-5, doi: 10.1109/TD48022.2026.11562229.

Figure 1. Script architecture

Script architecture

Module 1: Pre-screening

Planners might consider the risk assessment question from two perspectives:

• How can one quickly identify the most vulnerable locations in their system?
• How would they go about conducting more detailed risk analysis for an individual location?

Step 1: Identify locations with high voltage and angle sensitivity.

• Add fictitious load to user-selected buses (specified voltages in a specified area) without switched shunts and note change in voltage due to change in MW/MVar injection.
• Add fictitious load to user-selected generator buses (>specified generation threshold in a specified area) and note change in bus angle due to change in MW injection.

Step 2: Identify excitable frequencies from a selected location

• Simulate a load impulse at user selected location.
• Analyze ringdown response using an automated FFT+VARPRO pipeline to identify risky mode frequencies excitable from the selected location.

Module 2: Simulation

• Modify existing HV load to represent a data-center load with an oscillation injection block.
• Run PSS/E simulations to inject oscillation from user-selected location and specifiable (shape, frequency, amplitude) combinations.
• Two shapes can be selected - square wave, biperiodic square wave.

Module 3: Analysis

• Analyze simulation outputs to compute risk metrics and identify most-impacted elements.
• Risk is visualized through interactive HTML dashboards as well as summary spreadsheets.
• Thresholds can be set by an utility according to their risk budget.
• Users can also visualize impedance trajectories and RAS activation likelihood for selected elements without having to explicitly configure them in simulation models.

---

Folder structure

Root/
│
├── psse_config.py                  ← Edit once: set your PSS/E install path and version
│
├── Pre_Screening_config.csv        ← Configuration for Steps 1, 2a, 2b, 2c
├── modal_analysis_config.csv       ← Configuration for Steps 2b and 2c
├── simulation_config.csv           ← Configuration for Steps 3a through 8
│
├── PSSE_Cases/                     ← Place your .sav, .dyr, and .raw, .idv files here
│
├── Processing/                     ← Intermediate outputs 
│
├── results/                        ← Simulation outputs and plots 
│
├── Step1_extract_case_info.py
├── Step2a_locational_sensitivity.py
├── Step2b_load_impulse.py
├── Step2c_mode_estimates.py
├── Step3a_simsetup_loadadd.py
├── Step3b_simsetup_monitoredqty.py
├── Step4_runsim.py
├── Step5_analyze_sim.py
├── Step6_metrics_visualization.py
├── Step7a_distance_z3_reach.py
└── Step7b_RAS_check.py

---

First-time setup

Open psse_config.py and set your PSS/E installation details:

PSSE_INSTALL_DIR = r"C:\\Program Files\\PTI\\PSSE35\\PSSBIN"
PSSE_VERSION     = 35

This file is imported by every script that calls PSS/E — you only need to edit it once.

---

Configuration files

All user inputs are driven by three CSV configuration files. Edit these before running any step.

Pre_Screening_config.csv

Used by Steps 1, 2a, 2b, and 2c. One row per case.

Column	Description	Example
case_name	PSS/E .sav file name (without extension)	WECC_2031_HW
dyr_name	Dynamics .dyr file name (without extension)	WECC_2031_HW_dyn
voltage_sensitivity_minKV	Minimum bus voltage (kV) for sensitivity screening	69
voltage_sensitivity_maxKV	Maximum bus voltage (kV) for sensitivity screening	138
area	PSS/E area number to screen (leave blank for all areas)	3
angle_sensitivity_minMW	Minimum MW injection for angle sensitivity calculation	25

modal_analysis_config.csv

Used by Steps 2b and 2c. One row per bus to analyse.

Column	Description	Example
case_name	PSS/E .sav file name (without extension)	WECC_2031_HW
dyr_name	Dynamics .dyr file name (without extension)	WECC_2031_HW_dyn
bus_number	Bus where the load impulse is injected	5003
load_step_MW	Impulse magnitude in MW	50

simulation_config.csv

Used by Steps 3a through 8. One row per simulation scenario.

Column	Description	Example
case_name	PSS/E .sav file name (without extension)	WECC_2031_HW
dyr_name	Dynamics .dyr file name (without extension)	WECC_2031_HW_dyn
bus_number	Bus where the LDDL oscillation is injected	5003
load_id	Load ID at that bus	1
oscillation_shape	Waveform type: square, biperiodic	square
oscillation_frequency	Oscillation frequency in Hz	0.4
oscillation_amplitude	Peak oscillation amplitude in MW	100
'oscillation_frequency_fast'	Faster frequency (Hz) for biperiodic load variation	4

---

Running the scripts

The steps are designed to run in sequence. You can stop after any step if only partial results are needed — for example, stopping after Step 2c gives you mode estimates without running full simulations.

Step 1 — Extract case information

python Step1_extract_case_info.py

Reads the PSS/E case and writes bus, branch, generator, load, and area summary CSVs to Processing/. Run this first for any new case.

---

Step 2a — Voltage and angle sensitivity screening

python Step2a_locational_sensitivity.py

Applies small fictitious injections at each bus in the specified voltage range and computes dV/dP, dV/dQ, and dθ/dP. Use this to identify vulnerable locations in the network.

---

Step 2b — Load impulse simulation

python Step2b_load_impulse.py

Applies a short load impulse at the bus specified in modal_analysis_config.csv and records the ringdown response.

---

Step 2c — Mode estimation

python Step2c_mode_estimates.py

Analyses the ringdown signal from Step 2b to identify excitable oscillatory modes. If a prominent mode is found near a particular frequency, that frequency is a priority candidate for detailed simulation in Steps 3–8.

---

Step 3a — Simulation setup: add LDDL model

python Step3a_simsetup_loadadd.py

Modifies the PSS/E case to represent the LDDL. Moves the existing load to an MV bus behind a step-down transformer, replaces its dynamic model with a CMLD model (NERC LMWG data center parameters), and adds a separate oscillation injection block. Outputs LLmod.sav, LLmod.snp, and a modified .dyr file.

---

Step 3b — Simulation setup: select monitored quantities

python Step3b_simsetup_monitoredqty.py

Uses the case summary from Step 1 to compile the list of buses, generators, loads, and lines to be logged during simulation. Outputs four CSVs to Processing/. Adjust selection criteria in the script if the default channel count is too large for your system.

---

Step 4 — Run simulation

python Step4_runsim.py

Runs the PSS/E dynamic simulation with the oscillation waveform defined in simulation_config.csv. Outputs results/<bus>_sim.out and results/<bus>_sim.csv. Simulation outputs are tagged with a run identifier of the form bus<N>_<freq>Hz_<amp>MW (e.g. bus5003_0.4Hz_100MW) so multiple scenarios can coexist in the results/ folder.

---

Step 5 — Analyse simulation results

python Step5_analyze_sim.py

Reads the simulation CSV and computes swing amplitude, envelope, and thermal loading metrics for generators, lines, buses, and loads. Flags elements that exceed configurable risk thresholds and writes summary and detail violation reports. To adjust the risk thresholds, edit the RISK_THRESHOLDS dictionary near the top of the script.

---

Step 6 — Interactive risk dashboard

python Step6_metrics_visualization.py

Generates a self-contained HTML dashboard (results/risk_visualization_<run\_tag>.html) from the metrics produced by Step 5. The dashboard includes summary risk statistics and time-series plots for the worst elements in each category. Users can interactively change violation thresholds. Example results

---

Step 7a — Zone 3 distance relay check

python Step7a_distance_z3_reach                       # interactive
python Step7a_distance_z3_reach --line 5001-5003-1     # direct

For a selected line, computes the Zone 3 mho relay reach (if not already present in PSSE model) and plots the apparent impedance trajectory from the simulation against the Zone 3 boundary. If RATE_B (or any other branch parameter) is missing from the case data, the script prompts you to enter it manually.

Reach setting philosophy is simple - zone 3 reach is set in a way that the relay doesn't trip for 150% of rate B flow if voltage falls below 0.85 p.u.

Step 7b — RAS trigger check

python Step7b_RAS_check.py                                                              # interactive
python Step8_RAS_check.py --bus 5003 --signal volt --threshold 1.05 --duration 0.5 --direction above
python Step8_RAS_check.py --line 5001-5003-1 --signal P --threshold 300 --duration 0.3 --direction above

Checks whether a user-defined Remedial Action Scheme trigger condition was sustained during the simulation. Only rudimentary rules can be implemented at present. You specify an element, a signal, a threshold, a direction (above/below), and a minimum duration. The script identifies all windows where the condition is met and produces an annotated time-series plot and a per-timestep CSV with a condition flag.

Supported signals:

Element type	Available signals
Bus	Voltage magnitude (pu)
Line	Active power P (MW), Reactive power Q (MVar), Angle difference Δθ from–to (degrees)

---

Test cases

The scripts have been tested with the following publicly available PSS/E cases:

Case	Source	Default scenario
WECC 240-bus	NREL Test Case Repository	1.2 Hz oscillation from bus 6508

---

Reporting issues

Please report bugs, unexpected behaviour, or suggestions to shuchismita.biswas@pnnl.gov.

Known limitations

Simulation script fails if no load present at bus selected to be LDDL oscillation source in the base case.

Number of oscillation cycles to run hard coded as 8 in Step 4.

LDDL MV bus number must be specified if the HV bus number has seven digits.

Visualization improvement opportunities in the HTML dashboard.

Number of step-down transformers hard coded in Step 3a. Can be edited by users.