NREL's Home Energy Management System—foresee
As a secure home automation system, foresee™ coordinates the operation of connected appliances, home batteries, and rooftop solar, satisfying homeowner values and preferences along with utility grid needs.
NREL's software uses algorithms that learn each home, and its occupants' schedules and patterns, so foresee can predict future energy consumption in homes. Thus, foresee enables highly accurate predictions of comfort needs, energy costs, environmental impacts, and grid service availability by leveraging machine-learning algorithms, advanced data analytics, and physics-based modeling and simulation to derive data-driven appliance models and usage patterns.
Recent results indicate foresee can provide homeowners between 5% and 12% whole-home energy cost savings while delivering a balance of comfort, convenience, and grid benefits. If foresee were used in every U.S. home, at least 5% of residential energy could be saved every day, or more than 1 quadrillion British thermal units (Btu) of primary energy each year, which equates to about $10 billion in energy bill savings. This is roughly equivalent to saving all the energy used by every household in Alaska, Delaware, Hawaii, Idaho, Maine, Montana, New Hampshire, North Dakota, Rhode Island, South Dakota, Vermont, and Wyoming combined.
The foresee software adapts to high and low energy demands and sends/receives energy forecast signals and price signals via two-way communication with the utility. By coordinating energy use and shifting appliances to run at off-peak times, foresee can reduce strain on the power grid and help consumers see lower energy bills by participating in existing demand-response and time-of-use programs.
By monitoring utility prices and weather forecasts, analyzing power consumption, and monitoring connected appliances and systems, foresee schedules the operation of connected systems to best achieve priorities the homeowner/user identifies.
Thanks to these breakthroughs, foresee received a 2018 R&D 100 Award.
Data from foresee-connected appliances are used to develop a predictive model of the building components—from the wall insulation to the air conditioner and water heater and more. Machine-learning algorithms build these models—automatically customizing itself to each home—so foresee can:
- Predict a home's future energy consumption
- Plan how to satisfy occupants' comfort needs
- Reduce energy costs and environmental impacts
- Enable participation in the local utility's incentive programs.
Using receding-horizon model-predictive control to chart a forward-looking path through the day, and then implementing it via minute-to-minute energy decisions, foresee acts on the homeowner's behalf to coordinate the home's connected appliances and systems, including those that are the most energy-intensive:
- Heating/cooling (e.g., thermostat, furnace, and air conditioner)
- Water heater
- Clothes washer
- Clothes dryer
- Large kitchen appliances (e.g., dishwasher and refrigerator)
- Other plug loads and connected electronics
- Pool pumps
- Rooftop photovoltaics (PV)
- Plug-in electric vehicles
- Home batteries.
The foresee system's built-in cybersecurity layer protects all parties, despite the two-way transmission of data between home and utility. Not only are the data cyber-secure to prevent hacking, but a homeowner's data are also protected from the utility—a utility cannot see or access the homeowner's preferences, and the weights derived from the algorithm are secure as foresee learns how users want to run their homes.
This project, including technology development and laboratory experiments, was supported by the Bonneville Power Administration and U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Building Technologies Office. Technical partnerships with Robert Bosch North America, ETAS/ESCRYPT, Colorado State University, and University of Colorado were critical to the project's success.
NREL currently uses foresee in its research in various ways and has provided laboratory demonstrations to more than 10 external organizations.
Battery Size, Selection Methodology
NREL developed a preliminary method to size a home battery by using foresee to optimize how it would be used alongside connected loads in any climate region, residential building, and utility tariff. The team engaged with an advisory board of equipment manufacturers, utilities, academics, and state and nonprofit energy stakeholders to develop a plan to mature this equipment-selection methodology. The early work was presented at the 5th International High-Performance Buildings Conference. Read the conference paper preprint, Economic Sizing of Batteries for the Smart Home. Also, view the webinar presentation. Learn more about this partnership.
The following projects have extended foresee's capabilities to study new opportunities in coordinating building energy use along with the electric power grid.
Distributed Energy Coordination
In partnership with BlockCypher, NREL used blockchain technology to stage a simple, distributed energy marketplace between two simulated home energy systems. Researchers demonstrated this distributed energy marketplace in the Energy System Integration Facility's (ESIF's) Systems Performance Laboratory, where two realistic home energy systems, including foresee, exchanged energy according to a blockchain-based smart contract. Learn more about this project.
Local Voltage Forecasting and Voltage Regulation
Buildings can have a significant effect on the local portions of the electric grid. Especially in places where distributed renewables are being adopted more quickly, high-voltage conditions can occur when many rooftop solar panels produce more power than buildings are simultaneously using. NREL developed a predictive methodology for forecasting high voltages, which allows foresee to shift when appliances use energy to overcome this voltage issue. In utility terms, this service is called "voltage regulation."
NREL is applying this forecasting tool to help the Hawaiian Electric Companies (HECO) understand its options by validating several voltage regulation strategies, making specific use of advanced inverters with voltage support functions, and their integration with controllable end-use devices. This work has included an aggregation study and research into how smart-load shifting can enable higher penetrations of PV.
Clouds passing overhead often create major fluctuations in energy production from PV arrays. To moderate this variability—an effect called "solar firming"—NREL developed a module for foresee to use the connected home appliances and battery to consume any excess energy behind the meter, rather than exporting it to the power system. foresee's performance during this kind of event is shown in the figure below, where foresee reduced power variability and eliminate energy export from the home. This type of solution is expected to enable greater amounts of distributed energy resources on the grid, and improved reliability.
Commercialization and Partnership Opportunities
The foresee software is available for licensing. NREL researchers are interested in partnering with companies that are ready to take foresee to market. Next steps include conducting field experiments/pilots to:
- Further develop the machine-learning methods using data from diverse real buildings and occupants
- Prove that the measured energy savings, grid services, and other benefits are achieved in real buildings.
Stochastic Model Predictive Control for Demand Response in a Home Energy Management System, IEEE Power and Energy Society General Meeting (2018)
User-Preference-Driven Model Predictive Control of Residential Building Loads and Battery Storage for Demand Response, American Control Conference (2017)
Modeling Stationary Lithium-ion Batteries for Optimization and Predictive Control, IEEE Power and Energy Conference (2017)
An Application of the Analytic Hierarchy Process for Prioritizing User Preferences in the Design of a Home Energy Management System, Sustainable Energy, Grids and Networks (2018)
Partners and Sponsors
NREL developed this technology in collaboration with Bosch, ESCRYPT, Colorado State University, and University of Colorado researchers with funding from the Bonneville Power Administration and U.S. Department of Energy Office of Energy Efficiency and Renewable Energy Building Technologies Office.