GridLAB-D: Testing the Smart Grid Before We Build It

Building the Smart Grid in the United States is a huge and expensive endeavor. The U.S. power grid is already considered one of the most complex machines ever built, and converting it to a Smart Grid won’t be easy. It entails overlaying the current infrastructure with hundreds of millions of smart meters numbering, adding a new communication infrastructure, as well as making an untold number of equipment upgrades throughout the electric transmission and distribution system. As we transition to the Smart Grid, how can we be sure it will all work correctly, and how can we quantify the benefits it will deliver?

The answer, or at least part of the answer, lies in a computer model developed at the U.S. Department of Energy's (DOE) Pacific Northwest National Laboratory (PNNL) called GridLAB-D. The sophisticated computer program provides a detailed, simultaneous simulation of the electric grid, including power flow, end-use loads, and market functions and interactions within a power grid. The model allows users to evaluate new technologies and operational strategies, to craft and refine the characteristics of these technologies and strategies, and to predict the results of deploying them.

grid operations graph on computer screen

For instance, when American Electric Power (AEP) was considering the deployment of a new voltage control system on its distribution system, GridLAB-D was used to simulate the system and quantify the expected benefits of its use. After analyzing the results of a field trial, GridLAB-D was able to extrapolate those results to the entire AEP system, allowing AEP to build a business case for the wide-scale deployment of the technology.

Another example of the model’s usefulness is PNNL's examination of Conservation Voltage Reduction (CVR), which aims to save energy by reducing the voltage of distribution-system feeder lines. Feeder lines deliver power from the utility's substations to the customers. PNNL researchers recognized that numerous CVR systems have been deployed in North America, but there was little substantive analytical analysis of their performance. Therefore, they decided to use GridLAB-D to perform detailed simulations of the effect of CVR schemes on 24 prototypical feeder designs. They then were able to extrapolate those results to a national level.

How GridLAB-D Works

So how does GridLAB-D achieve this feat? The model calculates a simultaneous solution to the multiple equations governing the power flow through the system and its interaction with the loads on that system. These equations can be quite complex, because power grids provide both real and reactive power to their loads.

Modeling Real vs. Reactive Power

Real power is the power consumed by resistive loads, like incandescent bulbs, and it relates directly to the energy consumed.

But devices that store and release energy, such as capacitors, or that use coils of wire to induce magnetic fields, such as electrical motors, have the ability to cause increased electrical currents without consuming real power; this is known as reactive power. Effectively, these devices store and release energy over time, making it much more complicated to model the behavior of the entire power grid.

GridLAB-D handles these complications by breaking down the behavior of the grid into discrete chunks of time—as short as one second, or as long as an hour—and then solving all the equations as if the system were operating in steady-state conditions. To be more technical about it, it uses an advanced algorithm to determine the simultaneous state of millions of independent devices, each of which is described by multiple differential equations solved only locally for both state and time. Because of this approach, it is not necessary to integrate all of the device's behaviors into a single set of equations—a task that would be daunting when trying to model millions of devices.

The result is a complete modeling of the power flow throughout the grid, including the grid's interaction with the wide variety of electrical loads that it serves. Market factors and consumer behaviors, such as peak-time pricing and customer's responses to those factors, can also be modeled. GridLAB-D can even model the effect of changing weather in the grid. For instance, cold winter weather causes heating systems to operate more, increasing the load on the grid, while hot summer days spark the use of fans and air conditioners.

GridLAB-D consists of core software for modeling the grid, to which any number of plug-in modules can be added to model specific technologies, market factors, and consumer behaviors. Thus allowing the software to be tailored to different system configurations. The open-source program allows users to understand exactly how the program functions and allows anyone to contribute to the program. To maintain quality assurance, all updates are monitored and controlled by PNNL staff.

The licensing for GridLAB-D is also open source, with no restrictions on uses or applications. This allows vendors to be able to develop plug-in modules to model the behavior of their technologies, whether it be smart appliances or distribution system controls. Vendors can also add or replace components of the program freely and can extract components for commercial use. Vendors can then sell their add-on modules for GridLAB-D and keep their modules proprietary if so desired.

The Past, Present, and Future for GridLAB-D

DOE has been investing in GridLAB-D since fiscal year (FY) 2007, with a total investment of more than $3 million as of FY 2010. This investment brought the computer model from early prototyping, through development and validation, so that it can now be used for preliminary analyses of the grid, such as the CVR study cited above.

TIn FY 2011, PNNL plans to put GridLAB-D to work to make the most out of DOE's investments in Smart Grid demonstration projects through the American Recovery and Reinvestment Act. DOE's Smart Grid Investment Grant program has awarded more than $3.4 billion to 100 Smart Grid projects being conducted throughout the United States, with private funds bringing the total investment to roughly $8.1 billion. To make sure we get the most use out of that investment, PNNL will select a representative sample of the 100 Smart Grid Investment Grant projects to be analyzed.

The analysis of the sample projects will focus on five selected technologies that dominate the impact of the Smart Grid: CVR; demand response; energy storage, including plug-in hybrid vehicles as an energy storage device?; distribution automation; and the integration of renewable energy technologies. As in the CVR analysis cited above, PNNL will apply its analysis to the 24 prototypical feeders found in today's distribution grids. It will then extrapolate the impacts of the Recovery Act projects to the entire nation, giving an estimate of how these Smart Grid technologies could reduce electrical demand and energy use throughout the United States.

These analyses can help guide the near-and long-term future of the Smart Grid. It can provide specific answers to questions currently haunting utilities: "Where is the best place to invest money in Smart Grid technologies?" "What benefits should our company expect?" "How will this benefit our ratepayers?" By providing a quantitative understanding of both the cost and the impact of Smart Grid technologies, GridLAB-D will help direct the future of the electrical grid in the United States.