Modern commercial buildings have thousands of networked sensors (commonly using BACnet – Building Automation and Control Network). Typical sensors include room (zone) temperatures, humidity, air quality, light level, fan speeds, damper positions, etc. A centralized building management system processes this data to determine control set points for various building equipment. Many researchers (myself included) believe this data is valuable for much more than simple building control. However, this data is typically stored in proprietary systems with no methods for open access.
The first goal of this project is to expose the building systems in Academic West using a well-defined, easy to use interface (e.g., RESTful web service). The team, with the help of facilities, will identify all of the building systems and what sense and control points they offer. Most likely these will come from several different manufacturers and be completely inconsistent (and possibly undocumented). The team will develop a mapping of physical devices onto a set of user friendly generic devices. For example, every HVAC system has at least one fan. It might report RPM or just ON/OFF status. It might also allow you to set the speed or not. The system should cope with this while still providing all of the native functionality of the end-device. This process will be strongly tied to what the team discovers when they map out all of the building systems.
The second phase of this project is to use the interfaces developed to log and visualize the activity of building systems. This should include real-time status and the ability to examine historical time-series data. Data will be stored in a database of some kind, possibly in the cloud (AWS or similar). Part of the system design should be to consider scalability issues using the datarate of the sensors. The system should support device hierarchy and creating virtual devices backed by several physical devices. For example, a high-level air handling unit (AHU) is composed of a supply fan, return fan, coils, and possibly several dampers or valves. By combining several AHUs, we could consider a whole building’s HVAC system with a single high level virtual device. Visualization and control should be possible at any level (end device or virtual device). Ideally the system should be easy to install into a new building through robust network probing and monitoring to identify what devices and capabilities exist on the network (eg plug-and-play auto-discovery).
Interface design / abstraction
Able to hack low-level serial communication protocols
Points of Contact
- Prof. Alan Marchiori, Dept. of Computer Science, Bucknell University