Recovery Logistics Co-Optimization of Power Systems against Natural - - PowerPoint PPT Presentation

Recovery Logistics Co-Optimization of Power Systems against Natural Disasters

Shunbo LEI1 A joint work with Dr. Yunhe Hou1,

• Dr. Chen Chen2 & Dr. Yupeng Li3

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
• r weeks [GridWise Alliance, NERC, etc.])
• For Hong Kong
• A coastal city threatened by typhoon, etc.

Observed Outages to the Bulk Electric System (Source: Energy Information Administration) News clipping from Macau

• n 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 enclave.

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.

• 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,
• perate manual switches, etc.
• Other flexibilities: distributed generations, demand response, etc.

Problem statement: “disaster recovery logistics”

• Typical restoration process
• Especially effective for single-fault outage scenarios

Problem statement: “disaster recovery logistics”

2 3 4 5 1 6 7 8 2 3 4 5 1 6 7 8 2 3 4 5 1 6 7 8 2 3 4 5 1 6 7 8 2 3 4 5 1 6 7 8 2 3 4 5 1 6 7 8 (a) (b) (c) (d) (e) (f) Closed branch Open branch Load Breaker Closed switch Open switch

(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

• 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

Problem statement: “disaster recovery logistics”

RC dispatch MPS dispatch DS restoration Co-optimization need coordination better improves improves improves interdependence interdependence Disaster recovery logistics

Route repair crews and mobile power sources in the transportation network, schedule them in the distribution system, and

• perate the distribution

system, in an efficient and coordinated manner, for electric service restoration.

• Resilience level as a function of time with respect to an event
• Survivability: R at [te, tr] (especially Rpe at [tpe, tr]) → Resourcefulness
• Recovery: R at [tr, tpir] → Optimal dynamic dispatch of resources

(including: repair crews, mobile power sources, and the power grid)

A two-stage framework

• 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
• Two stages: pre-positioning & dynamic dispatch of resources

Natural disaster strikes Pre-positioning Dynamic disaptch

Timeline

Weather forecasting and monitoring Distribution system and road network damage assessments

Week/days ahead Days/hours ahead Minutes/hours/days afterwards

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

PI 1 PI 2 PI 3 PI 4

To form a microgrid, it needs a power supply, and some topology control. Mobile power sources: supply power Repair crews and distribution switches: topologize the distribution network Pre-positioning of mobile power sources: enhance resourcefulness, so as to enhance survivability. Coordinated with: proactive network reconfiguration, so as to attain a state less impacted/stressed by the event. 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
• Survivability enhancement: left shaded area
• Recovery enhancement: right shaded area

te tpe tr tpr tir tpir Time R

R0 Rpe

pr

R' Rpr

pe

R'

R1: Solely conventional restoration R2: Coordinated w/ dispatch of MPSs & RCs

t0 tpir

'

Resilience level By the way: In this work, we measure the resilience level by the weighted sum of supplied loads:

• Pre-positioning of mobile power sources

Models, algorithms & case studies

Outermost level Middle level Inner- most level

• Two-stage robust optimization

Pre-position mobile power sources Radiality constraints (proactive network reconfiguration) Maximum number of damages (uncertainty budgets) Real/reactive power balance Real/reactive power capacities of mobile power sources Load pick-up Power flow limits on branches DistFlow model Voltage limits

• Pre-positioning of mobile power sources
• Column-and-constraint generation algorithm

Models, algorithms & case studies

Compact form: Master problem (MP): Sub-problem (SP):

• Pre-positioning of mobile power sources
• Case I: IEEE 33-node test system

Models, algorithms & case studies

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.

• Pre-positioning of mobile power sources
• Case II: IEEE 123-node test system

Models, algorithms & case studies

(EV fleets: two 150kW/150kWh electric buses *2 MESSs: 500kW/776kWh *2 MEGs: 800kW/600kVar *2) Again, enhanced survivability, especially when coordinated with proactive network reconfiguration. Statistics of 10,000 Monte Carlo simulations Connecting nodes of mobile power sources in pre- positioning, & network topology

• Coordinated dynamic dispatch of available resources

(repair crews, mobile power sources & the power grid)

• Routing and scheduling of repair crews

Models, algorithms & case studies

Visit at most 1 damaged component at each time Scheduling Routing Required repair time of damaged components Remain intact once repaired A damaged component: repaired by only 1 crew Resource capacity of repair crews Travel time of repair crews between 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

• Coordinated dynamic dispatch of available resources

(repair crews, mobile power sources & the power grid)

• Routing and scheduling of mobile power sources

Models, algorithms & case studies

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

• 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

Models, algorithms & case studies

(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 More adaptive Distribution system operational constraints

• Coordinated dynamic dispatch of available resources
• The co-optimization objective
• Coupling between the distribution system & repair crews
• Coupling between the distribution system & mobile power sources

Models, algorithms & case studies

max weighted sum of restored loads min sum of travels of repair crews & mobile power sources A damaged component is operable only if it is repaired by one of the repair crews. Power outputs of distribution nodes depend on the connection of mobile power sources.

• Coordinated dynamic dispatch of available resources
• Linearization technique: McCormick envelopes
• Pre-assign a minimal set of repair tasks to repair crews
• Reduce the number of candidate nodes for connecting mobile power

sources

Models, algorithms & case studies

• Coordinated dynamic dispatch of available resources
• Case I: IEEE 33-node test system

Models, algorithms & case studies

Dynamically form microgrids (powered by mobile power sources, reconfigured by switching actions, extended by repairing actions.) Gradually return to the normal state

• Coordinated dynamic dispatch of available resources
• Case II: IEEE 123-node test system

Models, algorithms & case studies

• Conclusions
• Resources pre-positioning → Resourcefulness → Survivability
• Coordination/co-optimization of resources dispatch → Recovery
• Microgrid formation → Survivability & Recovery
• Future research
• A more generalized/practical/applicable framework for distribution

system disaster recovery logistics (drone/crew dispatch for damage assessments, etc.)

• Hardening of the system, to enhance damage “prevention”
• Routing & scheduling of private-owned EVs for electric service

restoration (in a distributed/game-based manner, etc.)

• Technical issues in microgrid formation (control, etc.)

Conclusions & future research

Thank you!