1 / 38

Technical Briefing for Lawrence Berkeley National Laboratory 8/31/2012

tm. Technical Briefing for Lawrence Berkeley National Laboratory 8/31/2012. History of ColorPower. MIT CSAIL ZOME Energy Networks BBN Technologies ECCO International ColorPower Center. Challenge.

abla
Download Presentation

Technical Briefing for Lawrence Berkeley National Laboratory 8/31/2012

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. tm Technical Briefing for Lawrence Berkeley National Laboratory 8/31/2012

  2. History of ColorPower • MIT CSAIL • ZOME Energy Networks • BBN Technologies • ECCO International • ColorPower Center

  3. Challenge • Future retail utilities must be able to effectively managedemand fulfillment at the margin • Defer consumption • Pull forward consumption • Cancel uneconomic consumption (in customers’ best interest!) • Problem: Inadequate coordination between the grid and end user devices

  4. Economists vs. Customers Microeconomics View Customer View I do not have a marginal demand for power, I want reliable service I am not a virtual power plant I don’t want price volatility risk or to do laundry at midnight • Customers can be modeled as rational marginal demand functions for a commodity • Customers can be modeled as virtual power plants • Customers need to be sent price signals to modify their behavior

  5. Retail Power Is a Service • Not a hot concert ticket • Not a basket of commodity electrons • Customers prefer to buy power as a service, not a commodity • Just like many other service industries • Congestion pricing is not the typical way service demand peaking issues are solved

  6. Survey of Service Peak Demand Control Paradigms

  7. Technology As Auction Enabler HOW MUCH IS THIS POWER WORTH TO YOU NOW? HOW ABOUT NOW? HOW ABOUT NOW? SORRY YOU WERE JUST OUTBID BY R1CH_P0WRHAWG YOUR POWER HAS MOVED TO A PLACE WHERE IT IS MORE APPRECIATED [POWER OFF] THE GOOD NEWS IS YOU HAVE NOW SAVED 26 OPEN SODA CANS WORTH OF CO2 EMISSIONS

  8. Smart Appliance: Energy Efficient or Trading System? FrigidTrader 3000 Now, with high beta coefficient strategies for those with appetites for high risk/reward energy trading! Unlike cheap knock-offs, this one detects when a seller is trying to walk up the clearing price and refuses to take the bait. Cools your food and integrated supercomputer simultaneously. Icemaker extra.

  9. Retail Price Volatility: Be Careful What You Wish For Smart Appliance/EV Stampedes Price Signal PEANUTS ARE VERY EXPENSIVE RIGHT NOW WIDE-SCALE PRICE VOLATILITY: The Smart Grid GOOD FOR SPECULATORS BAD FOR CONGESTION CONTROL CATASTROPHIC FOR SYSTEM RELIABILITY WE’LL PAY YOU TO TAKE THESE PEANUTS AWAY The Smart Grid

  10. Cooperative Control Systems Natural Self-Organizing Swarm Systems Bacteria Colonies Neural Networks Flocks & Schools Social Insects Human-Designed Self-Organizing Congestion Management Systems Ethernet (CSMA/CD) Internet (TCP/IP) Queue Formation Traffic Signals WiFi (CSMA/CA) ColorPower (AFEM)

  11. Natural Adaptive Feedback Systems • Bees maintain a constant hive temperature, set by instinctive hive “programming” • 35° C to produce foragers • 34° C to produce housekeepers TEMPERATURE SETTING HEAT COOL CREATE BREEZE SUN ADAPTIVE FEEDBACK LOOP CREATE HEAT WIND HEAT COOL

  12. Engineered Self-Organization: The Traffic Light • Increases road capacity by order of magnitude without building more lanes! • Fairly distributes road access to users • When coupled with cooperative drivers, self-organized congestion control system • Increases transportation system utilization: higher efficiency Technical note: Roundabouts are even better than traffic lights!

  13. Introducing ColorPower A Distributed Self-Organizing Stochastic QoS-Oriented Power Load Balancing Protocol

  14. ColorPower Outsources Demand Response to the Appliance Swarms Smart Grid: Coordinates Orderly Power Access For Flexible Appliances & Machines Invisible to Humans ColorPower Appliance Priorities Obey Your Humans’ Preferences Donate Flexibility to Power Grid Humans: Always In Control Privacy Respected Flexible Emergency Not Flexible

  15. A: ColorPower: Self-Identification of Flexible Demand Leaf Button: On=Flexible Green = Price Sensitive Yellow = Reliability Responsive Red = Opt Out Cloud Software With More Options

  16. Measuring Demand Backlog Pool Pump Turn Off Device Tiers Run Now More Demand Resource Tier 500 Solar Resource Tier 499 Export Power Resource Tier 5 Resource Tier 1 Thermostat No Feedback Default Tier Assignment Via Color/Device Type Resource Tier -1 -3 degrees Resource Tier -5 -6 degrees Resource Tier -499 Facility Management Resource Tier -500 Less Demand Stay under 6kW EV Charge Discharge Battery Generator On

  17. Tiered Aggregations Major Inconvenience Emergency Load Dumping Dispatchable Load Resources Emergency Storage Emergency Dispatchable Load Night Run Dishwashers Battery Storage No Inconvenience Misc. Flex - Green Pool Pumps - Green HVACs - Green Sheddable Load Resources Misc. Flex - Yellow Pool Pumps - Yellow HVACs - Yellow Emergency Rationing Major Inconvenience

  18. ColorPower Stochastic Swarm Control New Demand Target -50MW • Groups and Individual Devices Act Randomly—But *Precisely in Aggregate* • Feedback loop recruits resources until demand target satisfied DEMAND CLOUD TIER ADAPTIVE FEEDBACK LOOP DEMAND REPORT: ON: 2034 OFF: 3423 SHED: 1276 NOSHED: 4322 SYSTEM STATE: NEED -3.4% TOTAL DEMAND RELIEF Local Probabilistic Cooperation Calculation Groups and Individual Devices

  19. Locationalx Program Control Programs Locational Participants Emergency Load Dumping Emergency Load Dumping Emergency Load Dumping Emergency Load Dumping Emergency Storage Emergency Storage Emergency Storage Emergency Storage Emergency Dispatchable Load Emergency Dispatchable Load Emergency Dispatchable Load Emergency Dispatchable Load Night Run Dishwashers Night Run Dishwashers Night Run Dishwashers Night Run Dishwashers Battery Storage Battery Storage Battery Storage Battery Storage Misc. Flex - Green Misc. Flex - Green Misc. Flex - Green Misc. Flex - Green Pool Pumps - Green Pool Pumps - Green Pool Pumps - Green Pool Pumps - Green HVACs - Green HVACs - Green HVACs - Green HVACs - Green Misc. Flex - Yellow Misc. Flex - Yellow Misc. Flex - Yellow Misc. Flex - Yellow Pool Pumps - Yellow Pool Pumps - Yellow Pool Pumps - Yellow Pool Pumps - Yellow HVACs - Yellow HVACs - Yellow HVACs - Yellow HVACs - Yellow Emergency Rationing Emergency Rationing Emergency Rationing Emergency Rationing

  20. Demand Signals To Markets Price signals to consumers from markets? ColorPower can send demand signals from consumers to markets.

  21. Flexible Load Shaping STAY WITHIN CAPACITY LIMITS UTILIZE IDLE CAPACITY

  22. More Flexible Than Price Signals Pre-Configured Rules Balance Priorities Across the System Customers First!

  23. ColorPower ™ Algorithm • Challenge: fast, private, robust, non-intrusive • Approach: randomized distributed control • Aggregate flexibility information to shared model • Disseminate control signals • Local decision; coin-flip for fractional color • Weight for availability, over-damped control Control problem: long timeouts on state changes

  24. ColorPower State Transitions • (E)nabled vs. (D)isabled • (R)efractory vs. (F)lexible

  25. ColorPower ™ State Transitions • The evolution of each device is modeled like a modified Markov process • In each round devices in state EF randomly switch off to state DR • Once in DR device waits for certain rounds before transitions to state DF; the waiting time is a fixed number PLUS a uniform random addition to feather the distribution (so not many devices switch states at once) • The other two distributions are complementary

  26. Formal Control Problem For each ColorPower client, set pon, poff for each device group, such that the total enabled power in s(t) tracks g(t)

  27. Formal Control Problem • The control problem is to set the transition probabilities such that the total Enabled Demand tracks the target as closely as possible, subject to the constraints • Device with lower numbered colors are shut off first • If a color has devices that are Enabled and Disabled, then every device is equally likely to be disabled • No device is unfairly burdened by its initial bad luck in becoming Disabled

  28. Constraints • Goal/Forecast tracking: shape power demand • Color priority: respect user preferences • Fairness: no devices are favored • Cycling: don’t keep the same devices off

  29. Controller Design Issues • It is possible that not all constraints can be satisfied; some of them are more important than others • Customer preferences are the most important ones • Goal tracking is the second most important • Least important is the Cycling constraint • The Fairness constraint is the easiest to satisfy (simply the same stochastic algorithm on all clients is executed) • We view the controller as having a “budget” of flexibility to spend with each color offering up to |EF|I of potential reduction in demand

  30. Controller Design Issues • Flexibility builds up as Refractory devices finish their time outs and move to the Flexible state • The controller is formulated as a cascade of priorities of how to spend the “Flexibility budget” indicated by the state s(t) • As the controller considers each constraint in turn, it allocates flexibility to satisfy that constraint (as much as possible) • Then it attempts to satisfy the rest of the constraints with whatever flexibility remains unallocated • Any unallocated flexibility is allowed to accumulate as a reserve improving future controllability

  31. Control Example: Hot Summer Day Flex Emergency Inflexible

  32. Control Example: Emergency Response Flex Emergency Inflexible

  33. Resource Requirements Algorithmic Complexity & Bandwidth Usage ColorPower is computationally trivial—no supercomputers required ColorPower requires an average bandwidth of 100 bytes per second per device. Trivial for modern broadband connections Does require massively parallel datagram traffic

  34. Consumer Privacy • Devices randomly and respond anonymously using local situational info + global system state info • Load reports are anonymous and aggregatable • 2-way anonymous information exchange

  35. Intrinsic Security Hardening Internal Saboteur/ External Hacker • No Infrastructure Targeting • Controllers Cannot Directly Contact Clients • Each Client Randomly Makes Its Own Choices • No Wide-Area Shutdowns • Clients Ignore Unreasonable System State Reports • Controllers Partitioned & Firewalled ColorPower Probabilistic Broadcast ? No Direct Attack Vectors

  36. OpenADR • ColorPower can act as a VEN aggregating small loads • ColorPower can act as a DRAS

  37. Collaboration • ColorPower needs • Research collaboration • Standardization

  38. Thank You • To learn more, see • http://www.colorpower.org • To help bring about this future, please contact • sflorek at colorpower.org

More Related