Of the 12 smart grid projects to win an award as part of the GE Ecomagination Challenge, one in particular caught my attention.
That is GE’s collaboration with Columbia University’s School for Engineering and Applied Science, FedEx Express and Con Edison to take Columbia’s patented Adaptive Stochastic Controller technology for a test run in Manhattan.
Perhaps it’s that Columbia is the only university of the 12 grant winners or a recent conversation I had with a friend about the massive scale of electric vehicle charging infrastructure that is likely to be necessary in the near future, but the blurb about the collaboration in this Sustainable Business article piqued my interest.
The technology, developed by Columbia Engineering’s Roger Anderson and his Smart Grid team, will manage load and power delivery and provide a real time data link between electric vehicle charging stations and Con Edison’s electric distribution management system.
GE will provide $1.1 million in funding as well as expertise and support for the project, which will focus on recharging a fleet of electric delivery vehicles (EDVs) that FedEx will deploy next year.
The stochastic controller technology is described and claimed in Columbia’s U.S. Patent No. 7,395,252 (‘252 Patent), which issued in July 2008. The ‘252 Patent’s decidedly non-energy related title is “Innervated stochastic controller for real time business decision-making support”.
The ‘252 Patent is directed to a controller that optimizes decision-making by training itself using power grid simulations then analyzing grid events and generating planned responses to the events.
The ‘252 Patent describes a Learning System (1400) that includes a reinforcement-learning controller (1002), optional learning matrices (1004) used within a “critic” function (1003) and a model (1006) of a power grid (1600).
The Learning System uses simulation models of the subject power grids to link and analyze specific threat events on the power grid (1600) and generate planned and prioritized responses, while automatically and continuously “learning” during simulation runs.
The Learning System may be configured as a computer-based simulation and training tool that learns “best response scenarios” to these specific events on the grid and can train power control system operators to respond to such events or can act on its own and take automatic control actions.
The Columbia press release explains the technology as follows:
[The] patented Adaptive Stochastic Controller [will] “learn” the energy demands of each truck and coordinate its recharging with Con Edison to make sure the EDVs deliver “on time, every time” at the lowest possible cost while fitting smoothly into Manhattan’s electric-distribution grid. The controller will send commands, such as when to optimally start and stop the charging of both the EDVs and the recharge stations at the delivery depot. The stations will also record and transmit updated information to our complementary Columbia Engineering controller at Con Edison’s Manhattan Electric Control Center to ensure proper grid integration. The Columbia Engineering controller will be able to respond to electric-load-management directives from Con Edison to decrease or increase the current draw from the on-board vehicle inverters and batteries to assure both the stability of the electric grid in the area and the recharge capability of the FedEx Express EDVs.
Some of the other smart grid award winners are OPOWER (energy management systems and software), ClimateWell (efficient appliances), FMC-Tech (intelligent sensor technologies), Soladigm (building efficiency), SustainX (compressed air energy storage), and SynapSense (data center services).