Distributed Grid Control of Flexible Loads and DERs for Optimized Provision of Synthetic Regulating Reserves ARPA-E NODES PROJECT DE-FOA-0001289 University of California-San Diego University of Illinois Urbana-Champaign Typhoon HIL Network Optimized Distributed Energy Systems (NODES) Annual Review Meeting April 26-28, 2017
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Project Summary Objective Integrated control of flexible loads (FLs) and distributed energy resources (DERs) to provide regulation services to bulk power grid Technical development of coordination algorithms, software, and architectures Approach Grid-connected microgrids as entities to support frequency regulation Using existing microgrid controlling infrastructures to incorporate FLs and DERs Benefits arise from efficiently utilizing DERs and FLs instead of acting as constant powers when the microgrid is grid-connected 2 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Project Summary Control Hierarchy for Frequency Regulation Non-profit organization RTO ISO/RTO Aggregators (or DERPs) and resources with power flexibility DERs and FLs in microgrid Aggregators DERs DERs Coordination DERs 3 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Project Summary Control Hierarchy for Frequency Regulation Non-profit organization RTO ISO/RTO Aggregators (or DERPs) and resources with power flexibility DERs and FLs in microgrid Aggregators Challenges Optimally dispatch compensation of area 1 control error to microgrids DERs Cooperation of DERs, FLs in microgrids to 2 DERs track regulation changing every 2-4 seconds Hardware-in-the-loop (HIL) validation Coordination 3 DERs Large-scale simulation to demonstrate 4 positive benefit for the bulk grid 3 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Team - Senior Personnel University of California-San Diego Sonia Mart´ ınez (scalable coordination, distributed opt) Jorge Cort´ es (network control, large-scale systems) Bill Torre (UCSD microgrid, renewable integration) Byron Washom (industrial outreach, tech transition) University of Illinois Urbana-Champaign Alejandro Dom´ ınguez-Garc´ ıa (modeling, control, and optimization of electric power systems) Peter Sauer (power system dynamics and stability, operational reliability, modeling of renewable resources) Typhoon HIL Ivan Celanovic (HIL validation, tech to market) Plus 3 postdocs and 4 PhD students 4 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Team - Senior Personnel University of California-San Diego Sonia Mart´ ınez (scalable coordination, distributed opt) Jorge Cort´ es (network control, large-scale systems) Bill Torre (UCSD microgrid, renewable integration) Byron Washom (industrial outreach, tech transition) University of Illinois Urbana-Champaign Alejandro Dom´ ınguez-Garc´ ıa (modeling, control, and optimization of electric power systems) Peter Sauer (power system dynamics and stability, operational reliability, modeling of renewable resources) Typhoon HIL Ivan Celanovic (HIL validation, tech to market) Plus 3 postdocs and 4 PhD students 4 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Project Progress Task 1: DER and FL abstractions to capture load flexibility Ongoing work (Q4-Q5) M1.1.2: Validation of full-order models M1.1.1: Full-order model of one DER Task 2: Provably-correct coordination algorithms M2.3.1: Distributed DERP-DER ratio consensus M2.4.1: Distributed DERP-DER coordination with dedicated pricing M2.1.1: Distributed RTO-DERP opt. algorithms Task 3: Partial distributed architectures M3.1.1: Test scenarios for convergence rate M3.2.1: Coordination scheme for baseline Task 4: Testing the impact of DERs for control M4.1.2: Initial simulation platform layout for algorithms power distribution system&grid-connected M4.1.1: Test scenario documentation microgrids w/ real&simulated data M4.3.1: Datasets for microgrid emulation Task 5: Technology transition M5.1.2 Tech to market plan M5.1.1 Tech to market plan (1) Novel Capabilities M5.2.1 and M5.2.2: IAB (2) Pathways to adoption 5 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Project Progress ISO/RTO Reduced-order models for inverters 1 (T1.1) Framework for optimal RTO-DERP 2 coordination (T2.1) Aggregators Distributed algorithms for 3 coordination of DERs and FLs (T2.3-4) Robust to uncertainties and communication failures DERs Fast convergence ( ≤ 40 iterations per DERs node for convergence in 2 seconds) Coordination Testing and validation (T4.2) DERs 4 6 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Models for Inverters (T1.1) Motivation Full-order model of inverters for DERs involve inner loops with high complexity ⇒ not suitable for real-time simulation & control Our work Identify full-order model of inverter control Develop reduced-order model for inverters to capture main characteristics Hardware-in-the-loop (HIL) simulation supports accuracy 7 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans RTO-DERP coordination (T2.1) Motivation Current practice in frequency regulation market determines capacity and mileage of energy resources capacity : upper bound on involvement of resource in regulation mileage : total absolute power change during regulation time period Capacity Tracking error Length of lines: Instructed mileage Power Time Time Power output sample Regulation signal set point RTO assigns regulation signal proportionally to procured mileage of each resource (redistributes overshoot power if any, again prop.) Our work Distributed coordination for efficient assignment of regulation signal: optimized cost functions, respecting operational limits 8 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans RTO-DERP coordination (T2.1) Motivation Current practice in frequency regulation market determines capacity and mileage of energy resources capacity : upper bound on involvement of resource in regulation mileage : total absolute power change during regulation time period Packet drops Time (sec) for entering Time (sec) for convergence to 1% band 1% band Chain Ring Ring with edge Chain Ring Ring with edge 0% 0.05 0.05 0.05 0.2 0.15 0.15 1% 0.16 0.15 0.15 0.48 0.3 0.29 5% 0.145 0.14 0.14 0.5 0.32 0.31 10% 0.15 0.15 0.15 0.55 0.35 0.32 RTO assigns regulation signal proportionally to procured mileage of each resource (redistributes overshoot power if any, again prop.) Our work Distributed coordination for efficient assignment of regulation signal: optimized cost functions, respecting operational limits 8 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Distributed algorithms for DERs and FLs (T2.3-4) Motivation: Coordination algorithms to coordinate DERs and FLs inside microgrid to track power reference varying every 2-4 seconds Challenges Nontrivial non-convex optimization, in principle with power flow equations (PFEs) for accurate tracking Tight convergence requirements ( ≤ 2 seconds) Must be robust to communication failures and uncertainty Ideally, want to solve optimal power flow problem every 2 seconds Our work Scheduled-asynchronous algorithm (with PFEs) Ratio-consensus algorithm (with relaxed PFEs) 9 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Distributed algorithms for DERs and FLs (T2.3-4) Scheduled-asynchronous algorithm SDP relaxation exact for power networks of moderate size Distributed using insights from power flow equations, no global clock Converges to optimum with O (1 / n ) rate if topology is bipartite Robust to communication failures and handles load uncertainty 150 Illustration 140 Num. of iterations 130 5 buses with DERs and 9 120 110 buses with uncertain loads 100 0 1 2 3 4 5 6 7 8 9 10 Time (minutes) Tracking a 10 minutes 9000 RegD signal from PJM 8000 7000 Cost Number of iterations per 6000 True Soln. 5000 Dist. Soln. node is 10 4000 0 1 2 3 4 5 6 7 8 9 10 Time (minutes) 10 / 19
Project Summary Team Project Progress Validation Plan Achievements Tech to Market Future Plans Distributed algorithms for DERs and FLs (T2.3-4) Scheduled-asynchronous algorithm SDP relaxation exact for power networks of moderate size Distributed using insights from power flow equations, no global clock Converges to optimum with O (1 / n ) rate if topology is bipartite Robust to communication failures and handles load uncertainty Ratio consensus algorithm Only incorporates resource box constraints – useful in scenarios where power flow constraints not significant inside microgrid Robust to communication failures Faster convergence than scheduled-asynchronous algorithm 10 / 19
Recommend
More recommend
