Skip to Main Content U.S. Department of Energy
Transactional Network

Catalyst - Building Technologies Office (BTO)

Utilizing commercial building controls and operations data in innovative new ways that promote better engagement with building occupants and improve the ability for operations to balance energy and occupant comfort objectives.

Problem statements:

  1. Building users cannot easily visualize operations data. Commercial building users (operators/occupants) need visual data analysis applications that allow them to manipulate and visualize complex and large technical datasets to detect and display patterns, trends, anomalies, and other vital building efficiency information. Visualization tools must be compact, easy to view and interactive across both mobile and desktop platforms.
  2. Scalable methods for analyzing buildings data are needed. Data manipulation methodologies that might combine statistical analysis, pattern recognition, and data visualization using disparate but potentially related data sources to provide innovative insights for building control systems operation do not exist. Such insights might include, for example, the identification of fault conditions, their causes and predicted time periods via correlations between various information sources (e.g., sensors, weather, etc.) Both intuitive and non-intuitive secondary/tertiary causal relationships across multiple parameters must be automatically mined to enable scalable statistical analysis.
  3. Domain-agnostic solutions for managing building data are needed. Available data sources on building controls and operations are growing rapidly but present analytical challenges due to their volume and variety. Open platforms for storing these disparate data sources/formats and translating them into actionable building controls intelligence could support development of a variety of machine learning and data analysis algorithms from a wide community.
  4. Building operations do not account for occupant behaviors. Commercial building operations need ways of identifying, measuring, anticipating and responding to occupant behaviors that affect operational outcomes. Occupant energy behaviors must be represented in a standard way in Building Automation System (BAS) user interfaces, and BAS systems must include capabilities for dynamically detecting occupancy and energy-related actions and assessing occupant comfort alongside building energy demand (possibly using existing data streams from buildings); these data can then be utilized in the optimization of operations. Supporting sensors and control algorithms must contain intrinsic capabilities of machine learning and signal processing to detect changes in building operations (whether facilities or occupant driven) and communicate actionable information to both parties.
  5. Building operators are disconnected from building occupants. Commercial building operators need ways of communicating pending changes in building conditions that may or may not affect the occupants. If operators could provide occupants with visualizations of historical and dynamic thermal, visual, acoustic and air quality metrics alongside energy consumption and/or demand information, it would help manage occupant expectations about interior building conditions and engage them in ongoing trends in the building operation. This capability is particularly useful during load shedding events where the operator wishes to deliver acceptable but detectably reduced services to occupants (thermal, visual, etc.).
  6. Occupants are outside of the building controls loop. Occupants need better ways of providing direct feedback to building operators/control systems and other occupants on their real-time personal comfort and/or productivity. This will require real-time sensing, data analytics and controls capabilities using embedded devices that support both machine to machine and machine to human content/intent communication. Advances can be leveraged in touch/no-touch gesture and voice recognition, mobile platforms, and wearables.
  7. We don’t know enough about occupant behavior over time in buildings. Commercial occupants’ presence/absence and adaptive actions have potentially large effects on energy consumption; yet, few long-term datasets exist to help us understand energy-related occupant actions in these types of commercial settings for the purposes of more efficient building design and operation. One significant challenge is that direct, time resolved data on occupant behavior over time are very time consuming and expensive to collect. This limits collected data to a few disparate case studies that may not be representative of behavior trends across a wide array of commercial building contexts.

One way of getting around the requirement of direct occupant data collection is to generate synthetic occupant behavior data from other, more readily available data streams. For example, an algorithm could be developed that infers likely occupancy and occupant behavior patterns from sparse Building Automation System data streams on building energy use and environmental conditions like temperature and humidity. These data on behavior could then inform real-time operational strategies. Moreover, de-identified behavior data could be shared with a broader community online to assist in the development of control algorithms that better account for the influence of occupant presence and actions on building operation.

Sample Datasets:

A number of sample datasets from commercial buildings can be accessed below. Note that all the information about the datasets is included in the archive, there is not additional information about these datasets. Some of these datasets have multiple buildings and systems data, while others may have data from a single building.

  1. Sample Dataset one; 10 MBs
  2. Sample Dataset two; 23 MBs
  3. Sample Dataset three; 1 MB
  4. Sample Dataset four; 87 MBs
  5. Data for 100 anonymized commercial and industrial building sites

Learn more about VOLTTRON™ Central.

Research Areas


Additional Information