Recovery Logistics Co-Optimization of Power Systems against Natural Disasters Shunbo LEI 1 A joint work with Dr. Yunhe Hou 1 , Dr. Chen Chen 2 & Dr. Yupeng Li 3 1 Dept. Electrical & Electronic Eng., The University of Hong Kong 2 Energy Systems Division, Argonne National Laboratory (U.S.) 3 Dept. Computer Science, The University of Hong Kong Nov. 7, 2018
Outline Background & motivation Problem statement: “disaster recovery logistics” A two-stage framework Models, algorithms & case studies Conclusions & future research
Background & motivation Significant power outages Outage scale: numerous affected consumers ( millions of people) Outage duration: prolonged electric service disruption (even for days or weeks [ GridWise Alliance, NERC, etc. ]) For Hong Kong A coastal city threatened by typhoon , etc. News clipping from Macau on 23 Aug 2017. Five people died, at least 153 were injured and two were still missing in Macau on Wednesday night, while the city also endured a power shutdown for several hours, after Typhoon Hato battered the former Portuguese Observed Outages to the Bulk Electric System enclave. (Source: Energy Information Administration)
Background & motivation The concept of “ resilience ” Different definitions Same essence: guarantee electricity supply against low-probability but high-impact natural disasters, extreme weather events, etc. Three elements of resilience [ EPRI ] Prevention : the application of engineering designs and advanced technologies that harden the distribution system to limit damage Recovery : the use of tools and techniques to quickly restore service to as many affected customers as practical Survivability : the use of innovative technologies to aid consumers, communities, and institutions in continuing some level of normal function without complete access to the grid Electric Power Research Institute of U.S., “Enhancing Distribution Resiliency: Opportunities for Applying Innovative Technologies,” 2013.
Problem statement: “disaster recovery logistics” Outage management of distribution systems Why distribution systems? → account for 70% power interruption A clarification: rapid restoration also enhances survivability Restoration decisions Switching actions : network topology, load pick-up, etc. Routing and scheduling of crews : repair damaged components, operate manual switches, etc. Other flexibilities: distributed generations, demand response, etc.
Problem statement: “disaster recovery logistics” Typical restoration process 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 (a) (b) 1 2 3 4 5 6 7 1 2 3 4 5 6 7 8 8 (c) (d) 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 (e) (f) Closed branch Open branch Closed switch Open switch Breaker Load (a) fault occurrence (b) post-fault state (c)-(f) fault location fault isolation service restoration (f) post-restoration state After that: repair damaged component & return to normal state Especially effective for single-fault outage scenarios
Problem statement: “disaster recovery logistics” Response/restoration against extreme weather events More damages More resources (more repair crews, and transportable/mobile power sources for isolated outage areas, etc.) More interdependent/coupled system (mobile power sources deliver power via power grid and also transportation network; etc.) Other issues, e.g., more time periods, different time-scales Route repair crews and interdependence interdependence DS restoration mobile power sources in improves improves the transportation network, better improves schedule them in the Co-optimization distribution system, and operate the distribution system, in an efficient and need coordination RC dispatch MPS dispatch coordinated manner, for electric service restoration. Disaster recovery logistics
A two-stage framework Resilience level as a function of time with respect to an event Survivability : R at [ t e , t r ] (especially R pe at [ t pe , t r ]) → Resourcefulness Recovery : R at [ t r , t pir ] → Optimal dynamic dispatch of resources (including: repair crews, mobile power sources, and the power grid) M. Panteli, and P. Mancarella, “The Grid: Stronger, Bigger, Smarter?— Presenting a Conceptual Framework of Power System Resilience,” IEEE Power and Energy Magazine, vol. 13, no. 3, pp. 58 -66, May/Jun., 2015.
A two-stage framework An important strategy: microgrid formation 23 24 25 29 30 31 32 33 To form a microgrid, it needs a power supply , and some topology control . PI 2 PI 3 26 27 28 5 6 PI 4 Mobile power sources: supply power 1 2 3 4 7 8 9 10 11 15 18 14 Repair crews and distribution switches: PI 1 19 12 13 16 17 topologize the distribution network 20 21 22 Two stages: pre-positioning & dynamic dispatch of resources Pre-positioning of mobile power Distribution system and road sources: enhance resourcefulness , network damage assessments so as to enhance survivability . Weather forecasting and monitoring Coordinated with: proactive network Natural disaster strikes Dynamic disaptch Pre-positioning reconfiguration , so as to attain a state Timeline less impacted/stressed by the event. Week/days Days/hours Minutes/hours/days ahead ahead afterwards Dynamic dispatch of all resources: dynamically form microgrids , which are powered by mobile power sources, and topologized by repair actions of repair crews and switching actions of distribution network; and, at the same time, gradually return to normal state .
A two-stage framework Enhanced survivability and recovery Resilience R level R 1 : Solely conventional restoration By the way: R 2 : Coordinated w/ dispatch of MPSs & RCs R 0 In this work, we measure the resilience level by the R ' pr weighted sum of supplied R pr loads : R ' pe R pe Time t 0 t e t pe t r t pr t ir t pir t pir ' Survivability enhancement: left shaded area Recovery enhancement: right shaded area
Models, algorithms & case studies Pre-positioning of mobile power sources Two-stage robust optimization Pre-position mobile power sources Outermost level Radiality constraints (proactive network reconfiguration) Maximum number of damages Middle level ( uncertainty budgets ) Real/reactive power balance Real/reactive power capacities of mobile power sources Inner- Load pick-up most Power flow limits on branches level DistFlow model Voltage limits
Models, algorithms & case studies Pre-positioning of mobile power sources Column-and-constraint generation algorithm Compact form: Master problem (MP): Sub-problem (SP):
Models, algorithms & case studies Pre-positioning of mobile power sources Case I: IEEE 33-node test system MESS stations: truck-mounted mobile emergency generators (MEGs) and mobile energy storage systems (MESSs) Charging stations: medium-duty electric vehicles (EVs) Connecting nodes of mobile power sources in pre-positioning, & network topology (EV fleet: two 150kW/150kWh electric buses MESS: 500kW/776kWh MEG: 800kW/600kVar) Statistics of 10,000 Monte Carlo simulations Enhanced survivability, especially when coordinated with proactive network reconfiguration.
Models, algorithms & case studies Pre-positioning of mobile power sources Case II: IEEE 123-node test system Statistics of 10,000 Monte Carlo simulations Connecting nodes of mobile power sources in pre- (EV fleets: two 150kW/150kWh electric buses *2 positioning, & network topology MESSs: 500kW/776kWh *2 MEGs: 800kW/600kVar *2) Again, enhanced survivability, especially when coordinated with proactive network reconfiguration.
Models, algorithms & case studies Coordinated dynamic dispatch of available resources (repair crews, mobile power sources & the power grid) Routing and scheduling of repair crews Visit at most 1 damaged component at each time Required repair time of damaged components Scheduling Remain intact once repaired A damaged component: repaired by only 1 crew Resource capacity of repair crews Travel time of repair crews between Routing damaged components Formulation of routing behaviors: Proposed formulation V.S. Traveling salesman problem based formulation 1. Straightforward & simple (structurally equivalent to min up/down in UC ) 2. Resolve coupling of transportation-power networks & their different timescales 3. More flexible and adaptive
Models, algorithms & case studies Coordinated dynamic dispatch of available resources (repair crews, mobile power sources & the power grid) Routing and scheduling of mobile power sources Routing & travel scheduling Power scheduling Real/reactive power outputs of MEGs Charging/discharging behaviors of MESSs Reactive power outputs of MESSs State-of-charge variations and limits of MESSs
Models, algorithms & case studies Coordinated dynamic dispatch of available resources (repair crews, mobile power sources & the power grid) Dynamic network reconfiguration (microgrid formation) and load pick-up of the distribution system (a) Spanning tree (common distribution network reconfiguration) (b) Spanning forest (microgrid formation) Removing edges from a spanning tree leads to a spanning forest: Simpler More straightforward Distribution system operational constraints More adaptive
Recommend
More recommend
