EPCC Workshop Bedford Springs, 3 June 2013 Smart Grid with Intelligent Periphery (Smart GRIP) Felix Wu Distinguished Visiting Professor in Clean Energy and Environment University of Hong Kong Professor Emeritus University of California, Berkeley 1
Grid of the Future Renewable and Distributed resources Smart Grid Plug-in EV Demand response Distribution system becomes more like transmission system Generation becomes stochastic Demand becomes controllable 2
Advanced Applications Server System Server Dispatchers With Workstations Failover Logic Alternate System Server Alternate Communications Gateway Communications Gateway Dual Redundant System Network AMI meets EMS Generation Transmission Substation Distribution Consumers EMS EMS Control Center Bridge to Corporate LAN Communications Device Power Station Encoder / Decoder State Monitor Control Driver Measuring SCADA Remote Terminal Device AMI 3
EMS Control 4
Market Operation A typical (wholesale) market consists of: 1. Day ahead 2. Hour ahead (Real time) 3. Ancillary services i. Load following ii. Frequency regulation Day-ahead 10 am DAM 75 Minutes ahead HAM 7.5 Minutes ahead ASM Real-time New Entrants Mostly distributed Different rules Storage EV Renewables Demand response 5
6 Source: www.cisco.com/web/strategy/docs/energy/control_architecture.pdf
Solution T.E. DyLiacco, Control of Power Systems via the Multi-level Concept, PhD Thesis, Case Western Reserve University, 1968. Chapter II: A Control Structure for Power System Operation Section 2.1: Decomposition of power systems networks into areas 7
Idea of an Area Nathan Cohn, Control of Generation and Power Flow on Interconnected Systems, John Wiley & Sons, 1966. Each area is expected to adjust its own generation to absorb its own load changes. Tie-line power flows between areas are scheduled and maintained. Otherwise, each area will define its own objectives. 8
Grid with Intelligent Periphery (GRIP) Structure the distribution system into areas With necessary modifications Clusters
Architecture of Smart GRIP Hierarchical Clustering System operator Substation Aggregator computation control information Clusters computation control information computation control information Layered Architecture Application Service Consumer Device 10
Ensuring Cluster Net Power Balance Risk-limiting dispatch» Sequential dispatch of (deterministic) net supply to follow (stochastic) net demand Electric-spring embedded smart load» Device to make local load to follow remaining supply-demand fluctuation in real-time Limit enforcement» Device to control supply/demand when supply/demand deviates from scheduled amount 11
Redefine Supply and Demand First to separate controllable/ non-controllable supply and controllable/ non-controllable demand. SD( t) W( t) DC( t) D( t) Controllable supply Non-controllable supply (wind, solar, ) Controllable demand Non-controllable demand SD( t) DC( t) St () d( t) D( t) W( t) Net supply Net demand The net supply is deterministic and controllable. The net demand is stochastic and non-controllable, RLD dispatches net supply (deterministic) to follow net demand (stochastic) 12
Risk Supply-demand balance at the operating time S( t) d( t) The net demand d (t) is stochastic has a probability distribution St () The risk of not meeting the requirement P{ d( t) S( t) and d( t) S( t) } RLD guarantees supply demand balance at the time of operation within the risk level specified by the user 13
Stochastic Sequential Decision xk ( 0) sk ( 1) + Y 1 Stage 1 xk ( 1) 1 2 Select that minimize the total cost + s( k, k ) Y 2 Stage 2 x( k1, k2) + s( k, k, k ) 1 2 3 x( k1, k2, k3) Y 3 Stage 3 s( k ) s( k, k ) s( k, k, k ) 1 1 2 1 2 3 Subject to the condition that the risk of supply demand imbalance is acceptable (i.e., risk-limiting) The sequential decisions are based on up-to-date info provided by the smart grid sensor networks Flexibility in future options of corrective actions are incorporated in the selection of optimal decisions 14 St () xk ( ) 1 0 sk ( ) s( k, k ) 1 2 s( k, k, k ) 1 2 3
The Optimal Solution The solution is surprisingly simple and is characterizes by two thresholds that have close-form solutions. s( k1,... k j ) s ( k,... k ) [ x( k,... k ) ] if x( k,... k ) 1 j 1 j 1 j 1 j 1 j 1 j j 1 j 1 j s ( k1,... k j ) [ j x( k1,... k j 1)] if x( k1,... k j 1) j s( k,... k ) 0 if x( k,... k ) x( k,... ) 1 k j 1 j j 15
Smart Load with Electric Spring Design a smart load to follow generation Idea of a spring Extended Compressed Neutral (a-1) Neutral position (b-1) Mechanical push (upward force) (c-1) Mechanical pull (downward force) v s = v o = v s_ref v s = v s_ref v s = v s_ref + v a = 0 + v a + v a Z 1 v o = v s_ref Z 1 v o < v s_ref Z 1 v o > v s_ref (a-2) Neutral position (b-2) Voltage boosting function (c-2) Voltage reduction function 16
Realization Current-controlled voltage source vs vs va = g(q) = f(ic) C ic va i c Current controller v s_ref - + va i c Current controller v s_ref - + vo Z1 (a) (b) (c) Series reactive compensator as electric spring Series Reactive Compensator as Electric Spring Vs PWM Power Inverter - + Vs ref Other Load v o Controllable Load S.Y.R. Hui, C.K. Lee and F.F. Wu, Power control circuit and method for stabilizing a Power Supply, PCT patent application 61/389,489, filed on 4 October 2010 17
Experimental Setup Power line electrically powered by dynamically changing wind power (Line voltage v S may vary from the nominal value) v G R X L X v S R G L G G Source imped anc e Wind Power Simulator (10kVA) Power cable v a Electric spring + i o Current Controller Vs_ref e.g. 220V Vs - + S Critical electric loads Z 2 v o Non - critical load under control Z 1 90KVA Inverter V G 2km V Renew 1km V S Smart Load Network Box: Load: P G Q G S SC =36KVA E G =430 Z G =5.1 X G /R G =10 R G =0.51Ω L G =16.3mH 2km: R=0.1Ω, X=2.4mH 1km: R=0.1Ω, X=1.22mH R 1 : 50.5Ω, R 2 : 53Ω Network Box Network Box V a ES P Renew Q Renew P 1 P 2 10KVA Renewable Energy Source Simulator V o Non-critical Load (R 1 ) Q ES Critical Load (R 2 ) 18
Pre-recorded intermittent power profile Test Results Repeated intermittent power profile 230 200 Volt (V) 210 190 170 150 Critical Load Non-Critical Load Electrical Spring 0 180 360 540 720 900 1080 1260 1440 Time (s) 150 100 50 0 ES Volt (V) 1030 1500 Power (W) 930 830 730 630 Critical Load Power Smart Load Power Smart Load Reactive Power 0 180 360 540 720 900 1080 1260 1440 Time (s) 1000 500 0-500 Q (Var) 19
Summary We have proposed a Smart Grid with Intelligent Periphery (Smart GRIP). Empowering the periphery» Management of uncertainty is placed close to its source, i.e., the periphery» Responsibility of maintaining reliability is distributed to the periphery» Control intelligence is distributed to the periphery The building blocks of Smart GRIP are clusters. Clusters are hierarchically aggregated: from smart homes, aggregators, substations to system operators. All clusters have the same control structure with different functions that are served by a layered architecture. 20
Summary (cont.) Clusters are, like control areas, responsible for its own net power balance, which is achieved through» Risk-limiting dispatch of net supply (deterministic) to follow net demand (stochastic)» Smart load embedded with electric springs to smooth out fluctuations» Schedule enforcement 21
GRIP architecture Reference» D. Bakken, A. Bose, K. Mani Chandy, P. Khargonekar, A. Kuh, S. Low, S. von Meier, K. Poolla, P.P. Varaiya, F. Wu, GRIP: Grid with Intelligent Periphery: A Control Architecture for Grid2050, 2011 CDC. Risk-limiting dispatch» R. Rajagopal, E. Bitar, P. Varaiya, F. Wu, Risk-Limiting Dispatch for Integrating Renewable Power, Intl J. Electrical Power & Energy Systems, No. 4: 2013, pp. 615-628 Smart load embedded with electric springs» S.Y.R. Hui, C.K. Lee and F. Wu, Electric Springs A New Smart Grid Technology, IEEE Transactions on Smart Grid, Vol.3, No.3, Sept. 2012, pp: 1552-1561 22
Center for Electrical Energy System The University of Hong Kong http://www.eee.hku.hk/~cees 23