Visible to the public LoCal—A Network Architecture for Localized Electrical Energy Reduction, Generation and Sharing


We investigate Information Age approaches for managing society's limited resource: energy, i.e., how pervasive information can change energy production, distribution and use. The world's electric grids are a wonder of last century's physical age, each with vast geographic reach, epitomized by a highly centralized, synchronized, and reliable distribution tree that permits power to be consumed without concern for its source. But changing energy demands, incorporation of renewable sources, and managed load, have pushed this aging marvel to its limit. Taking guidance from the design principles of the cyber age Internet infrastructure, we develop a more scalable, flexible and resilient electric power infrastructure--encouraging efficient use, integrating local generation, and managing demand through awareness of energy availability and use over time. Our insight is to integrate information exchange everywhere that power is transferred.

A LoCal Energy Network is a cyber overlay on energy distribution systems at various physical levels, e.g., machine rooms, buildings, neighborhoods, isolated generation islands and regional grids. Information enables a more efficient scalable energy system with improved resilience and quality of delivered power. Our approach integrates information about energy availability and use, interactive load/supply negotiation protocols, controllable loads and sources, and the allocation of generated energy to its consumption. These yield a system for agile, distributed, and integrated management that buffer energy to reduce peak-to-average energy consumption, moderate infrastructure provisioning, and encourage power-limited design and operation. A key idea is that of "supply-following" loads: i.e., loads that intelligently adapt their work to the availability of energy, accelerating their work rate when energy is plentiful and deferring it when energy is expensive. Loads as diverse as refrigerators and computer datacenters can be made to be "supply-following," providing a new systems approach for exploiting intermittently available energy from renewable sources.

We made significant progress in several areas. In Energy information Infrastructure, we released a new distribution of our sMap layered architecture and software to "wrap" sensors and actuators with a common API. We implemented differential privacy techniques to improve the individual privacy of energy usage measurements. In Systems for Renewable Energy Generation and Use, we designed a high-efficiency DC micro-grid for developing regions, to integrate renewable generation sources into rural regions. In Systems for Energy Generation and Storage, our solar thermal generation and storage prototype, based on a free-piston Stirling engine and non-tracking solar collectors, became operational for generation. We completed an evaluation of how energy storage can assist California meet its future goals for renewable energy penetration. In Energy Efficient Computer Clusters, we developed new scheduling methods to achieve energy proportionality in Soda Hall's high performance computer clusters. We also characterized parallel processing workloads based on map-reduce, and have demonstrated how intelligent scheduling of batch and interactive map-reduce jobs significantly reduces cluster energy. In Buildings as Controllable Supply-following Loads, we deployed energy proportional building-scale facilities for HVAC and lighting in Sutardja Dai Hall, and recently concluded long-term user experimentation with such facilities.

Award ID: 0932209

Creative Commons 2.5