What Is GridLAB-D?
The software targets the analysis and design of power distribution systems and can also be used to study storage and hybrid system design. Its core coordinates the state of millions of independent devices at once, each described by its own differential equations. Because loads are not collapsed into reduced-order aggregate models, effects such as many EVs charging behind one transformer come out of the simulation instead of being assumed.
PNNL developed GridLAB-D for the US Department of Energy with Office of Electricity funding and continues it with industry and academic partners. It was used in PNNL's analyses for DOE's Smart Grid Investment Grant program and serves as the distribution simulator inside PNNL's GridAPPS-D and TESP platforms. Do not confuse it with HiPAS GridLAB-D, a separate SLAC-led fork now hosted by LF Energy as Arras Energy.
GridLAB-D is a free distribution grid simulator from Pacific Northwest National Laboratory (PNNL). It models a feeder down to individual houses, appliances, EV chargers, PV inverters and batteries, then steps the whole system through time to show how they affect voltages, peak load and reliability.
Utility distribution planners, government planners and grid researchers use it to test rooftop PV, storage, EV charging or a demand-response program on a feeder before it is rolled out. It is not a solar design tool: there is no module layout, shading analysis, yield report or customer proposal, and no graphical interface, so installers and PV designers will be better served elsewhere.
Pros and Cons
Pros
- Free and open source: you can see exactly how each model works and change the code when a study needs something new.
- Device-level load models show how individual homes respond to weather, prices or control signals, which aggregate load shapes cannot capture.
- Developed by PNNL with DOE funding and still active: version 6.0 was tagged in October 2026.
- Tests demand-response and transactive schemes against real feeder limits such as voltage and transformer loading (built-in market up to 5.3, HELICS co-simulation in 6.0).
- One run can combine quasi-static and transient mode, for example an hour of normal operation with a fault in the middle.
- Ready-made feeder models to start from: IEEE test feeders, PNNL's taxonomy feeders and 12 Pacific Gas and Electric prototypical circuits.
Cons
- Steep learning curve: models are hand-written text, results need power systems expertise to interpret, and custom device models require C++.
- Command-line only: there is no built-in graphical interface like those in commercial alternatives.
- Documentation has moved between sites (SourceForge, a wiki, now Read the Docs), and several pages are still marked incomplete, so newcomers struggle to find answers.
- Version in transition: 5.3.0 (January 2025) is the latest release with installers; 6.0 (October 2026) has none on GitHub yet and drops the built-in market and MySQL modules.
- Distribution only: no transmission studies, and transient mode models electromechanical dynamics, so it is not a substitute for an EMT tool.
Features
Advanced Power Flow Analysis
Solves unbalanced three-phase power flow on radial and meshed feeders with forward-backward sweep (the default) or Newton-Raphson solvers, down to split-phase service at individual homes.
Multi-Domain Modeling
Combines power flow, end-use loads and weather in one simulation and can feed in price signals, so you can see how a heat wave or a retail price change shows up in feeder loading and voltage.
Time-Series Simulation
Runs quasi-static time series by default for studies of hours to years and switches to transient mode (deltamode) for sub-second events such as faults, with steps down to 100 microseconds. Transient models cover electromechanical dynamics, not electromagnetic transients (EMT).
Agent-Based Device Modeling
Each house uses an Equivalent Thermal Parameters (ETP) model with its own HVAC, water heater, lighting, plug and EV charger loads, alongside grid devices such as transformers, regulators and capacitor banks.
Distributed Energy Resource (DER) Integration
Includes solar, wind turbine, battery storage, diesel generator and inverter models; the dynamic inverter (inverter_dyn) supports grid-forming and grid-following control for microgrid studies.
Market and Pricing Simulation
Releases up to 5.3 include a double-auction market object, so devices in the model can set their supply or demand from real-time price signals. The 6.0 pre-release (October 2026) no longer builds the market module; transactive studies there run through HELICS co-simulation, as in PNNL's TESP.
Extensible Module Architecture
A small core loads plug-in modules such as powerflow, residential, generators, climate, tape and reliability, so developers can add model types without changing the engine.
GridLAB-D Pricing
GridLAB-D is released by Battelle Memorial Institute under a BSD-style license, so it is free to download, use, modify and redistribute, including commercially.
Free Plan: Installers, source code and example models are free on GitHub.
Paid Plans: There are no paid subscription tiers or commercial licensing fees for the core software.
Usage-Based Fees: There are no usage-based fees.
Enterprise and Custom Pricing: None. The real cost is engineering time to build, calibrate and script models, in-house or through a consultant; there is no vendor support contract to buy.
Integrations
GridLAB-D connects to other tools through text files, a Python API and HELICS co-simulation rather than ready-made connectors.
- Coding and Development: The source code is hosted on GitHub, where users report issues and ask questions in GitHub Discussions.
- Modeling Environments: The third-party CPS-VO GridLAB-D Design Studio, built on WebGME, edits models visually but uses GridLAB-D 4.0 (2017).
- Data Processing: The climate module reads TMY2/TMY3 and CSV weather files, player objects feed CSV time series such as prices or schedules into object properties, and recorders write results to CSV.
- API and Extensibility: Python bindings (the gridlabd package on PyPI) let scripts load a model, step simulation time and read or write object properties; new device models are written as C++ modules.
- High-Performance Computing (HPC): The C++ engine is built for very large models, and PNNL lists national-level simulations of smart grid benefits among its applications.
How to Set Up GridLAB-D
- Open the GitHub releases page: 5.3.0 (January 2025) is the latest version with installers; 6.0 (October 2026) has none yet, but its Python bindings are on PyPI (pip install gridlabd).
- Run the installer for your system (Windows .exe, macOS .sh, Linux .deb or .rpm) or compile the source code yourself.
- Make sure the PATH and GLPATH environment variables include the GridLAB-D bin, lib and etc folders so the core can find its modules.
- Download or write a test model (.glm file), such as the IEEE 13 Node Feeder, to check the installation.
- Launch the simulation from the command line (for example, gridlabd model.glm); GridLAB-D has no built-in graphical interface.
How to Use GridLAB-D
Day-to-day work means writing and editing .glm (GridLAB-D model) files, plain-text descriptions of a distribution system. You usually start from an existing feeder model or define the network yourself: lines, transformers and voltage regulators. Then you attach end-use “objects”, such as houses with their own HVAC settings and water heaters, plus any PV, batteries or EV chargers under study.
Next you add input data, such as weather for the climate object and time series for prices or schedules, plus recorder, collector or histogram objects for the variables you want to track: voltages, energy use, cost. The run writes them to output files over the simulated period. The analysis happens afterwards in your own scripts or a spreadsheet, for example to check whether a neighborhood battery delivers the expected benefit or causes voltage swings.
What You Can Manage with GridLAB-D
- Power Distribution Models: Text-based feeder models from the substation down to individual customer meters.
- Smart Grid Cost-Benefit Analyses: Energy, peak demand and cost results that feed the business case for a new technology; the financial analysis itself is done outside the tool.
- Load Growth Projections: Multi-year runs showing when rising demand, for example from EV adoption, overloads existing transformers and lines.
- Reliability Metrics: The reliability module calculates 12 IEEE 1366-2003 distribution indices, including SAIFI, SAIDI, CAIDI and MAIFI, from simulated outage events.
- Transactive Energy Programs: Simulation of demand-response and real-time pricing strategies.
- DER Impact Studies: How high PV and wind penetration affects feeder voltages and, in transient mode, frequency.
FAQs
What does GridLAB-D do?
GridLAB-D runs time-series simulations of a distribution feeder: unbalanced three-phase power flow coupled with models of the houses, appliances, PV systems, batteries and EV chargers connected to it, driven by weather and price inputs. Typical outputs are node voltages, equipment loading, energy use, cost and reliability indices.
Who is GridLAB-D best for?
It suits utility distribution planners, grid researchers and government agencies who need customer-level behavior, not just feeder power flow, and who are comfortable writing text models and scripts. It is a poor fit for PV installers and designers, for anyone who needs a graphical interface, and for teams that require vendor support.
Is GridLAB-D free?
Yes. GridLAB-D is released under a BSD-style license from Battelle, so you can use, modify and redistribute it in academic and commercial work without license or subscription fees.
What are the main limitations of GridLAB-D?
The main barrier is skill: GridLAB-D is a command-line program with no built-in graphical interface, and models are written as text files. Help comes from developers and other users on GitHub Discussions rather than a vendor help desk, and parts of the documentation are still incomplete. It also covers distribution feeders only, and its transient mode models electromechanical dynamics, not electromagnetic transients.
What are the best alternatives to GridLAB-D?
Common alternatives are the commercial suites CYME and ETAP, which come with graphical interfaces and vendor support, and the open-source OpenDSS. All of them handle distribution power flow; GridLAB-D stands out when a study depends on detailed end-use load models, such as thermostat-driven houses and EV chargers, or on price-responsive demand.




