with Environment-Friendly and Resilient Operation Strategies LEI - - PowerPoint PPT Presentation

Enhancing Power Grid Sustainability with Environment-Friendly and Resilient Operation Strategies

LEI Shunbo

• Dept. of Electrical & Electronic Eng.

The University of Hong Kong

April 25, 2018

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

• Power grid sustainability
• Challenges
• Effective integration of renewable energy sources (RESs)

(transmission-level & distribution-level)

• Emission control (air pollutants, etc.)
• Resilience against natural disasters

Background and motivation

Sustainable power grids Environmental sustainability Economic sustainability Social sustainability Environment-friendly

• peration

Resilient operation RES integration Emission issue …… ……

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

• Air pollution problem
• One of the worst environmental health risks
• PM2.5: deeply penetrate into human lungs and blood stream
• Millions premature deaths, trillions economic cost (UNEP, WHO)
• Responsibility of the power system
• Major consumer of fossil fuels
• A high portion of air pollutants emissions (coal-fired plants, etc.)

Robust UC considering air pollutant dispersion

• Under-utilized environmental benefits of wind powers
• Uncertainty and variability

(spinning reserve requirement, etc.)

• Only limiting total emissions

(spatial distribution of air pollutants should be accounted for)

• Research gap

Robust UC considering air pollutant dispersion

Gap: Generation scheduling that explicitly considers wind powers’ environmental benefits to reduce ground-level air pollutant concentration (GLAPC) at load centers.

• C. Wang, Z. Lu, and Y. Qiao, “A Consideration of the Wind Power Benefits in Day-Ahead Scheduling of Wind-

Coal Intensive Power Systems,” IEEE Trans. Power Syst., vol. 28, no. 1, pp. 236-245, Feb. 2013.

• E. Denny and M. O'Malley, “Wind generation, power system operation, and emissions reduction,” IEEE Trans.

Power Syst., vol. 21, no. 1, pp. 341-347, Feb. 2006. …

• Air pollutant dispersion models
• Emission rate function
• Plume and puff models

Robust UC considering air pollutant dispersion

2

( )

b b it i it i it i

E c P b P a     

x y z W i n d d i r e c t i

• n

Δh x y z W i n d d i r e c t i

• n

(a) (b)

H

r

2 2 2 2 2 2 2

1 exp 2 2

y it ijt y z x y x z x

d E H C w I I d I d I d                    

Considers meteorological conditions and the system’s geographical distribution, to estimate GLAPC due to emissions of generations.

• Problem formulation
• Objective function
• Cost due to air pollutant dispersion (APDC)

Consider population density & background pollution Minimize people’s exposure to air pollutants

Robust UC considering air pollutant dispersion

min( ) CC BDC APDC  

1 1

( )

G

N T i it i it t i

CC su SU sd SD

 

   



2 1 1

[ ( ) ]

G

N T b b i it i it i it t i

BDC P P O

 

       



2 1 1 1

[ ( ) ]

G D

N N T b b j ij i it i it i it t j i

APDC K c P b P a O

  

        



2 j j j

s NP BP     Commitment cost: start-up/shut-down of generators Base-case dispatch cost: real power outputs of generators

• Problem formulation
• Constraints

Robust UC considering air pollutant dispersion

Uncertainty set of wind power outputs. Robust feasibility check: ensure adequate spinning reserve, to avoid environmentally unfriendly operations of more polluting coal-fired units in real time.

• Solution algorithm
• Compact matrix form
• Robust feasibility check
• Iterative algorithm

Robust UC considering air pollutant dispersion

1) Set the number of iterations K=0. Choose a tolerance δ(>0) for the robust feasibility check. 2) Solve the MP to update the current optimal solution (x*, y1

*, y2,1 *,…, y2,K *).

3) K=K+1. Solve the SP with xK = x*, to obtain uK. 4) If objective_of_SP ≤ δ, return (x*, y1

*) and terminate. Otherwise, go to step 2).

A mixed-integer quadratic programing problem with a robust feasibility check

• Case studies
• IEEE 14-bus system

Robust UC considering air pollutant dispersion

2 4 6 8 10 12 50 60 70 AUCC (k$) PGLAPC (µg/m3) 28 28.5 29 PGLAPC without wind AGLAPC without wind AGLAPC (µg/m3) PGLAPC with wind AGLAPC with wind

UC cost k$ Total emission Ton Peak GLAPC µg/m3 Average GLAPC µg/m3 Total exposure ·105µg/m3 Case 1 224.58 100.25 70.32 29.37 150.92 Case 2 200.19 93.43 70.32 29.22 150.42 Case 3 213.38 65.00 64.79 28.29 148.09 Case 4 209.79 67.90 56.07 28.10 145.81 Bus No. Back-ground PM2.5 µg/m3 BPj NPj Peak GLAPC µg/m3 Average GLAPC µg/m3 Case No. Case No. 1 2 3 4 1 2 3 4 4 22 3 9203 22.00 22.00 22.00 22.00 20.00 22.00 22.00 22.00 5 50 5 1463 70.32 70.32 64.79 56.07 59.97 59.15 56.04 52.53 7 22 3 320 22.00 22.00 22.00 22.00 22.00 22.00 22.00 22.00 9 22 3 5680 22.00 22.00 22.00 22.00 22.00 22.00 22.00 22.00 10 30 4 1733 30.02 30.02 30.02 30.01 30.01 30.01 30.00 30.00 11 36 5 674 53.28 53.28 53.28 53.28 46.14 46.01 46.13 46.75 12 23 3 1174 32.36 32.36 31.99 31.46 28.60 28.34 27.71 27.61 13 18 3 2600 21.02 21.02 20.97 20.90 20.57 20.37 19.73 19.97 14 13 2 2869 13.13 13.13 13.13 13.13 13.06 13.06 13.02 13.06

Case 1: w/o wind power, w/o APDC Case 2: w/ wind power, w/o APDC Case 3: w/ wind power, w/o APDC, w/ total emissions limit Case 4: w/ wind power, w/ APDC

 Emission issues need to be explicitly considered, so as to utilize wind power’s benefits in

air pollution control.

 Only limiting total emissions is less cost effective.

• Case studies
• Guangdong Grid system

Again: 1) need to explicitly consider emission issues; and 2)

• nly limiting total emission is not

effective. Observation : wind power makes a system more cost effective in and more capable of air pollution control.

Robust UC considering air pollutant dispersion

UC cost ·107 $ Total emission Ton Peak GLAPC µg/m3 Average GLAPC µg/m3 Total exposure ·108µg/m3 Case 1 1.6254 26640.70 74.36 58.88 133.59 Case 2 1.4596 24183.33 74.36 55.43 124.21 Case 3 1.5290 11500.00 74.36 33.60 66.54 Case 4 1.5158 19228.48 24.36 14.52 16.93

23.0 29.5 36.0 42.5 49.0 55.5 62.0 68.5 74.0 1 2 3 4 5 6 7 8 9 1013.0 19.5 26.0 32.5 39.0 45.5 52.0 58.5 PGLAPC without wind AGLAPC without wind PGLAPC with wind AGLAPC with wind AUCC (105$) PGLAPC (µg/m3) AGLAPC (µg/m3)

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

• What is DSDR?
• Reconfiguration

From a radial topology to another radial topology, by opening/closing switches.

• DSDR: dynamically reconfigure the network over time
• Applications
• Power loss reduction, operation cost minimization
• Electric service restoration, reliability enhancement
• Supply capacity improvement
• Renewable distributed generation (DG) integration

Distribution system dynamic reconfiguration

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 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

• System requirement
• DSDR relies on the

deployment of remote- controlled switches (RCSs), which are supported by the distribution automation system.

• Research gap
• Papers on RCS allocation/placement/installation/planning do not

consider RCSs’ effect in facilitating DG integration by DSDR.

• Papers applying DSDR do not account for the allocation of RCSs

(simply assume all/some of the switches to be remote-controlled).

Distribution system dynamic reconfiguration

Gap: Optimal RCS allocation to facilitate DG integration.

Control center Substation

Communica- tion server Feeder terminal unit Remote-controlled switch Feeder breaker Remote terminal unit Communication network Workstation Database and server

• Problem statement
• Identify a limited number of critical switches that optimally enable

intro-day DSDR to minimize DG curtailments

• Optimization model formulation: mixed-integer linear programming
• Uncertainty consideration: robust optimization (worst-case

consideration)

• Solution algorithm: nested column & constraint generation algorithm

based on strong duality

Distribution system dynamic reconfiguration

Optimization model formulation Single/multiple objectives ANM strategies coordination Solution algorithm Hybrid Benders/C&CG Strong duality v.s. KKT Uncertainty consideration Stochastic characteristics Time-series issues …… …… ……

• Problem formulation
• Deterministic formulation of the critical switch identification problem

Distribution system dynamic reconfiguration

Obj.: minimize DG curtailments Number of selected critical switches

Switch-type-dependent operation constraints

Intro-day DSDR constraints

• Problem formulation
• Robust formulation of the critical switch identification problem

Innermost min: seek the optimal intro-day DSDR strategy to achieve MDGC, using selected switches, with given loads and DG outputs. Mid-level max: obtain loads and DG outputs that maximize MDGC. Outermost min: find critical switches that minimize the maximum MDGC.

Distribution system dynamic reconfiguration

Uncertainty set of loads Uncertainty set of DG outputs Modified obj., oriented by the worst case

• Solution algorithm: nested C&CG algorithm
• Inner C&CG

Distribution system dynamic reconfiguration

• Solution algorithm: nested C&CG algorithm
• Outer C&CG

Distribution system dynamic reconfiguration

• Case studies
• Robust model out-performs deterministic model
• Several critical switches adequately enable DSDR to integrate DG
• Marginal benefit of adding more RCSs decreases

Distribution system dynamic reconfiguration

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 DG DG DG DG DG DG

20 40 60 80 100 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 % Time Period Wind Solar 8.58 6.36 5.32 5.21 4.98 4.53 4.53 8.58 6.88 5.82 5.55 5.71 5.03 4.53 4 5 6 7 8 9 10 2 4 6 8 10 37 W-MDGC (MWh) Value of Π𝑆𝐷𝑇 Robust Model Deterministic Model

Robust Model Deterministic Model 2 12-13*, 14-15 12-13*, 16-17 4 6-26*, 27-28*, 12-13, 16-17 6-26*, 27-28*, 9-15, 17-18 6 6-26*, 7-8*, 16-17*, 27-28*, 12-13, 14-15 6-26*, 7-8*, 16-17*, 27-28*, 9-15, 20-21 8 20-21*, 27-28*, 2-3, 6-26, 12-13, 14-15, 16-17, 21-8 20-21*, 27-28*, 5-6, 6-7, 9-15, 17-18, 26-27, 29-30 10 6-26*, 8-9*, 9-15*, 12-13*, 27-28*, 2-19, 14-15, 16-17, 20-21, 21-8 6-26*, 8-9*, 9-15*, 12-13*, 27-28*, 2-3, 7-8, 11-12, 17-18, 24-25

• Case studies
• Only several switching

actions in intro-day DSDR

• Only 1 (at most 2) pairs of

switching actions in each reconfiguration

Distribution system dynamic reconfiguration

Open Switches 2 9-15, 11-12, 14-15, 20-21, 27-28 4 6-26, 9-15, 11-12, 16-17, 20-21 6 7-8, 9-15, 11-12, 20-21, 27-28 8 2-3, 6-26, 9-15, 11-12, 16-17 10 6-26, 8-21, 9-15, 11-12, 20-21 Time Period Switching Actions 2 Hour 4 Open: 12-13. Close: 14-15. Hour 14 Open: 14-15. Close: 12-13. 4 Hour 4 Open: 12-13. Close: 16-17. Hour 6 Open: 16-17. Close: 12-13. Hour 22 Open: 27-28. Close: 6-26. Hour 23 Open: 6-26. Close:27-28 6 Hour 14 Open: 6-26, 14-15. Close: 7-8, 27-28. Hour 15 Open: 7-8, 27-28. Close: 6-26, 14-15. Hour 18 Open: 16-17. Close: 7-8. Hour 21 Open: 6-26. Close: 27-28. Hour 22 Open: 7-8, 27-28. Close: 6-26, 16-17. 8 Hour 4 Open: 21-8, 27-28. Close: 6-26, 16-17. Hour 6 Open: 14-15. Close: 2-3. Hour 14 Open: 6-26. Close: 27-28. Hour 15 Open: 16-17. Close: 14-15. Hour 17 Open: 20-21. Close: 21-8. Hour 21 Open: 2-3. Close: 20-21. 10 Hour 5 Open: 12-13. Close: 20-21. Hour 14 Open: 14-15. Close: 12-13. Hour 15 Open: 12-13, 16-17. Close: 14-15, 9-15. Hour 16 Open: 8-9. Close: 16-17. Hour 17 Open: 16-17. Close: 12-13. Hour 21 Open: 20-21. Close: 8-9. Hour 23 Open: 9-15. Close: 16-17.

A small number of critical switches, and also a small number of switching actions of them, are adequate to enhance DG integration greatly.

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

Background

• Extreme weather events: the single leading cause of power
• utages
• Complete or partial wide-area blackouts

Observed Outages to the Bulk Electric System (Source: Energy Information Administration) U.S. 2014 Billion-Dollar Weather and Climate Disasters (Source: National Oceanic and Atmospheric Administration)

Background

Event Date Region/Division/State Customers Affected Superstorm Sandy October 2012 Northeast 8,100,000 Derecho July 2012 Middle Atlantic 4,200,000 Early season snow October 2011 New England 3,000,000 Tropical Storm Irene August 2011 Middle Atlantic 3,200,000 Wildfires July 2012 California, Colorado 2,000,000 Windstorm November 2011 Southern California 400,000 2011-2012 Natural Disasters and Reported Customers Affected by Power Outages

The GridWise Alliance, “Improving electric grid reliability and resilience: Lessons learned from Superstorm Sandy and other extreme events,” 2013 [Online].

• Significant electric service disruption

» Numerous affected consumers

Background

• Significant electric service disruption

» Prolonged outage duration

3 2

[2] NERC, “Hurricane Sandy Event Analysis Report,” 2014 [Online].

Figure 1: Damaged high voltage transmission line Figure 2: Damaged substation equipment Figure 3: Damaged distribution system equipment Figure 4: Crews clearing miles of road to access and repair a line

1 4

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

• Distribution system restoration
• (a) fault occurrence, (b) post-fault state, (f) post-restoration state
• (c)-(f) fault location, fault isolation, and service restoration
• Followed by repairing of damaged component(s), etc.
• A process with multiple times of reconfigurations, involving

multiple switching actions of manual/remote-controlled switches

RCS allocation enabling prompt restoration

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

• Outage management system
• Objective: prompt electric service restoration when faults happen
• One of the solutions: increase the number of RCSs in distribution

systems

• Challenges: high cost of RCSs, vast extent of distribution systems,

etc.

RCS allocation enabling prompt restoration

Optimal RCS allocation (to improve reliability/resilience of distribution systems): optimizing the number and locations of RCSs to be installed/upgraded, to enable prompt restoration.

• Research gap
• Different heuristics algorithms
• Mixed-integer linear programming (MILP)
• This work: mixed-integer convex programming (MICP) & MILP

Global optimum & more efficient computation

RCS allocation enabling prompt restoration

C.-S. Chen, C.-H Lin, H.-J. Chuang, C.-S. Li, M.-Y. Huang, and C.-W. Huang, “Optimal placement of line switches for distribution automation systems using immune algorithm,” IEEE Trans. Power Syst., vol. 21,

• no. 3, pp. 1209-1217, Aug. 2006.
• H. Falaghi, M. Haghifam, and C. Singh, “Ant colony optimization-based method for placement of

sectionalizing switches in distribution networks using a fuzzy multiobjective approach,” IEEE Trans. Power Del., vol. 24, no. 1, pp. 268–276, Jan. 2009.

• A. Moradi, and M. Fotuhi-Firuzabad, “Optimal switch placement in distribution systems using trinary

particle swarm optimization algorithm,” IEEE Trans. Power Del., vol. 23, no. 1, pp. 271-279, Jan. 2008.

• G. Levitin, S. Mazal-Tov, and D. Elmakis, “Optimal sectionalizer allocation in electric distribution systems

by genetic algorithm,” Elect. Power Syst. Res., vol. 31, no. 2, pp. 97–102, Nov. 1994. ……

• A. Abiri-Jahromi, M. Fotuhi-Firuzabad, M. Parvania, and M. Mosleh, “Optimized sectionalizing switch

placement strategy in distribution systems,” IEEE Trans. Power Del., vol. 27, no. 1, pp. 362-370, Jan. 2012.

• O. K. Siirto, A. Safdarian, M. Lehtonen, and M. Fotuhi-Firuzabad, “Optimal distribution network

automation considering earth fault events,” IEEE Trans. Smart Grid, vol. 6, no. 2, pp.1010-1018, Mar. 2015.

• Problem formulation
• RCS-restoration model

A MICP to generate restoration strategies utilizing only RCSs. Feasible set and optimal solution(s): function of RCS allocation decisions.

RCS allocation enabling prompt restoration

Obj.: max restored loads & min switching actions Only allow RCSs to be utilized Reconfiguration constraints

• Problem formulation
• RCS allocation models for different objectives

RCS allocation enabling prompt restoration

Obj.: max reduction of customer interruption cost, & min maintenance and operation cost. Loads that can be restored: constrained by the number and locations of RCSs. Obj.: max reduction of system average interruption duration index, & min M&OC. Number of customers that can be restored: constrained by the number and locations of RCSs. Loads that can be restored: constrained by the number and locations of RCSs. Restorable loads using RCSs: equal to restorable loads using all switches (manual/remote-controlled). Obj.: min the number of RCSs to be installed.

• Problem transformation: large-scale MICP to small-scale MILP
• Computational complexity analysis

Define the set of feasible restoration strategies, which is infinite.

• Practical candidate restoration strategies, which is finite

Features: larger amount of loads cannot be restored by the same set of switching actions; do not involve redundant switching actions.

RCS allocation enabling prompt restoration

• Problem transformation: large-scale MICP to small-scale MILP
• Equivalent re-formulation
• Linearization & simplification (also equivalent)

Integer variables: from to Computational complexity: from to

RCS allocation enabling prompt restoration

Nonlinear New integer variables

• Case studies
• Test systems
• Number of PCRSs

424 in total (compared to 2^37 possible combinations of switching actions)

• CIC-oriented RCS allocation

A small number of RCSs can reduce distribution system CIC effectively. The solution of this method is robust to the changes of CIC parameters.

RCS allocation enabling prompt restoration

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 1 2 8 9 15 56 57 58 59 60 50 54 117 118 119 121 52 53 55 116 36 51 33 49 70 114 115 122 47 48 71 113 35 34 28 46 109 111 112 123 27 45 69 110 30 29 43 108 106 44 68 105 107 26 25 38 42 103 39 41 67 104 24 23 77 37 62 72 64 63 65 73 21 20 74 66 75 76 7 19 99 18 17 101 97 5 95 93 94 88 87 90 6 102 98 14 16 100 96 4 92 89 86 91 61 78 79 80 81 82 83 84 85 31 32 3 12 11 10 13 22 120 40

Fault location 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Number of PCRSs 2 83 74 11 9 8 4 5 5 4 2 2 1 1 Fault location 17 19 20 21 23 24 26 27 28 29 30 31 32

• Number of PCRSs

1 10 2 1 15 1 36 33 31 14 39 29 1

• CIC changes Number of RCSs RCS cost ($/yr)

RCIC ($/yr) Optimal RCS locations 25% 4 2467.31 7257.55 8-9, 12-22, 18-33, 25-29 50% 5 3084.14 15571.38 8-9, 12-22, 18-33, 25-29, 30-31 100% 5 3084.14 31142.77 8-9, 12-22, 18-33, 25-29, 30-31 200% 5 3084.14 62285.53 8-9, 12-22, 18-33, 25-29, 30-31 400% 7 4317.80 124727.83 6-7, 8-21, 11-12, 12-22, 18-33, 25-29, 31-32

• Case studies
• SAIDI-oriented RCS allocation

 With the number of RCSs increasing, upgrading

• ne more switch to RCS brings a decreasing

marginal benefit in terms of SAIDI reduction.

• Restorable-load-oriented RCS allocation, & comparisons
• Summary

Different objectives lead to quite different optimal RCS allocation schemes. With a specific objective, a proper model should be developed and applied (or, compare results of different models before finalize the decision).

RCS allocation enabling prompt restoration

5 10 15 20 25 30 35 40 45 50 45 40 35 30 25 20 RCS cost (102$/yr) RSAIDI (min/yr)

Allocation model Allocation scheme RCIC minus RCS cost ($/yr) RSAIDI (min/yr) Restoration capability CIC-oriented 8-9, 12-22, 18-33, 25-29, 30-31 28058.63 46.89 24/27 SAIDI-oriented 6-7, 8-21, 18-33, 25-29 27351.96 45.96 18/27 RL-oriented 9-10, 9-15, 12-22, 14-15, 18-33, 25-29, 28-29, 30-31 26249.58 47.14 27/27

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

Mobile emergency generator dispatch

• Mobile emergency generator (MEG) features

FEMA, “Mitigation assessment team report: Hurricane Sandy in New Jersey and New York,” 2013 [Online].

A MEG supplying power for a data center after Hurricane Sandy (Manhattan, NY)

A MEG is a truck-mounted emergency response generator. Merits: mobility, large capacity (up to several MW) “Distributed generation technologies such as microgrids and mobile generators can enhance the resilience of electric infrastructure serving critical loads.”

The GridWise Alliance, “Improving electric grid reliability and resilience: Lessons learned from Superstorm Sandy and

• ther extreme events,” 2013 [Online].

• Research gap
• MEGs are currently under-utilized.
• Investigation of MEG dispatch is necessary.

Case of Hurricane Sandy D. Barrett, “Few big FEMA generators humming,” 2012.

Before Sandy struck (Oct. 29) 400 industrial-size generators prepared. 3 days afterwards (Nov. 2) Only 50 of them working, while millions still without electricity. Connecticut: 1/51 MEGs working, 300,000 consumers without power. New York State: 7/13 MEGs working.

Mobile emergency generator dispatch

• C. Yin, J. Wang, and T.-C. Ma, "System design of mobile emergency power plant based on fuel cells," in Proc.

2011 Int. Conf. on Electr. Inf. and Control Eng., Apr. 2011, pp. 5419-5422.

• F. T. Dai, "Risks of network protection for mobile generator applications," in Proc. IET 9th Int. Conf. on
• Develop. in Power Syst. Protection, Mar. 2008, pp. 681-686.
• C. R. Nightingale， "The design of mobile engine driven generating sets and their role in the British

telecommunications network," in Proc. 5th Int. Telecom. Energy Conf., Oct. 1983, pp. 144-150.

• L. Zhou, M. Fan, and Z. Zhang, "A study on the optimal allocation of emergency power supplies in urban

electric network," in Proc. 20th Int. Conf. and Exhib. on Electr. Distrib., Jun. 2009, pp. 1-4.

• J. Shang, X. Sheng, J. Zhang, and W. Zhao, "The optimized allocation of mobile emergency generator based
• n the loads importance," in Proc. 2009 Asia-Pacific Power and Energy Eng. Conf., Mar. 2009, pp. 1-4.

• Problem statement
• What is MEG dispatch problem?
• Requirements of MEG dispatch

Rational matching between MEGs and loads: to reduce outage scale, to fully utilize limited MEG resources, and to deal with load priorities. Fast response: to reduce outage duration, to take advantage of mobility of MEGs, and to timely restore highly critical loads.

[Online] www.duquesnelight.com

Mobile emergency generator dispatch

To dispatch MEGs effectively and timely in distribution systems to restore critical loads, thus reducing both outage scale and outage duration, as resilient response to natural disasters.

• Problem statement
• Features of the MEG dispatch problem

Both isolated areas and intact areas have to be considered. Optimal decisions are partly determined by road network conditions. Optimal decisions can only be made after the natural disaster strike: highly sensitive to system conditions (damages, outages, etc.),

and difficult to forecast system conditions accurately (statistical/simulation-based models).

Mobile emergency generator dispatch

S 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 29 21 22 23 24 25 26 27

Modified 27-node system from: S. Civanlar, J. J. Grainger, H. Yin, and S.

• S. H. Lee, “Distribution feeder reconfiguration for loss reduction,” IEEE
• Trans. Power Del., vol. 3, no. 3, pp. 1217-1223, Jul. 1988.

• Proposed MEG dispatch method
• Microgrids formation with MEGs

To dispatch MEGs as distributed generators in distribution systems to restore critical loads by forming multiple microgrids.

(Pre-installed microgrids V.S. dynamically formed microgrids) (Pre-installed distributed generators V.S. dynamically dispatched MEGs)

• Two-stage dispatch: pre-position and real-time allocation

Mobile emergency generator dispatch

Major decisions MEGs allocation: allocate each MEG to which node. Microgrids formation:

• pen/close each line, pick up or not each load.

Natural disaster struck

Pre-position Real-time allocation Timeline

Weather/conditions forecast and monitor

Distribution system & road network damages monitor & assessment Week/days

• ahead

Days/day/hours

• ahead

Minutes/hours/day/days afterwards

• Proposed MEG dispatch method
• Two-stage dispatch: pre-position and real-time allocation

Mobile emergency generator dispatch

S1 S2 S3 S4 100 102 106 108 112 114 150 116 118 120 122 110 124 126 156 154 152 132 188 190 158 164 130 128 148 146 144 142 134 160 136 140 162 138 299 275 225 201 202 203 227 213 212 205 242 244 228 229 209 230 231 232 208 233 234 237 238 211 241 240 210 236 235 206 220 204 214 218 207 224 222 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 S1 Substations

Distribution nodes Candidate distribution nodes for MEG connection DS1 DS2 DS3 DS4 Normally closed lines Normally closed lines

S1 S2 S3 S4 100 102 106 108 112 114 150 116 118 120 122 110 124 126 156 154 152 132 188 190 158 164 130 128 148 146 144 142 134 160 136 140 162 138 299 275 225 201 202 203 227 213 212 205 242 244 228 229 209 230 231 232 208 233 234 237 238 211 241 240 210 236 235 206 220 204 214 218 207 224 222 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 S1 Substations

Distribution nodes Candidate distribution nodes for MEG connection DS1 DS2 DS3 DS4 Normally closed lines Normally closed lines

Geographic information of road network & distribution systems Distribution systems

Boundaries

• f DSs

Candidate distribution nodes for MEG connection Staging locations

DS1 DS4 DS2 DS3

Road edges Intersections

S1 S2 S3 116 132 136 203 204 234 306 311 403 413 419

• Pre-position formulation: scenario-based stochastic MILP
• Scenarios: uncertain power & road networks damages
• Optimize expected performance of future real-time allocation
• Minimize outage duration considering priorities and sizes of loads

Mobile emergency generator dispatch

• Algorithms
• Vehicle routing: Dijkstra’s algorithm
• Pre-position: scenario-based decomposition

Improve computational efficiency & solution optimality, by taking advantage of the block-diagonal structure:

( lg( )) O E V V  

Mobile emergency generator dispatch

min{ ( , ): , }

n n n n n

f  



x y x Λ y Ω min{ ( , ): , , }

n n n n n n n n n n

f   

 

x y x Λ y Ω A x h

• R. T. Rockafellar and R. J.-B. Wets, “Scenarios and policy aggregation in optimization under uncertainty,” Math. of Oper. Res., vol. 16,
• no. 1, pp. 119-147, 1991.
• S. Ahmed, “A scenario decomposition algorithm for 0-1 stochastic programs,” Oper. Res. Lett., vol. 41, no. 6, pp. 565-569, 2013

• Case studies
• Microgrids formation with MEGs is an effective operation

strategy to fully utilize MEG resources (rational matching).

Mobile emergency generator dispatch

Boundaries

• f DSs

Candidate distribution nodes for MEG connection Staging locations

DS1 DS4 DS2 DS3

Road edges Intersections

S1 S2 S3 116 132 136 203 204 234 306 311 403 413 419

S1 S2 S3 S4 100 102 106 108 112 114 150 118 120 122 110 124 126 156 154 152 132 188 190 158 164 130 128 148 146 144 142 134 160 140 162 138 299 275 225 201 202 203 227 213 212 205 242 244 228 229 209 230 231 232 208 233 234 237 238 211 241 240 210 236 235 206 220 214 218 207 224 222 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 401 402 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 S1 Substations

Distribution nodes Candidate distribution nodes for MEG connection DS1 DS2 DS3 DS4 Normally closed lines Normally closed lines

MG 204 MG 413 MG 403 403 204 136 116

Minimize sum of (load_priorities*load_size*outage_time) Minimize sum of unserved loads MEG Capacity utilization rate Times of CUR≥85% Capacity utilization rate Times of CUR≥85% 1 73.03 % 221/500 74.58 % 228/500 2 64.13 % 97/500 80.87 % 252/500 3 75.93 % 242/500 87.28 % 323/500 4 63.65 % 149/500 71.36 % 221/500 5 60.41 % 126/500 72.14 % 179/500 Average 67.43 % 167/500 77.25 % 240.6/500

• Case studies
• Optimal pre-position of MEGs can help to reduce outage

duration of critical loads.

• Road network conditions needs to be explicitly considered in

MEG dispatch if shorter outage duration is desired.

• Two-stage dispatch of pre-position and real-time allocation is

an effective framework for MEG dispatch.

Mobile emergency generator dispatch

—— Average amount of loads restored by MEGs Average travel time to assigned locations Non-optimal pre-position w/ traffic consideration 3866.8 kW 28.26 minutes Optimal pre-position w/ traffic consideration 3910.9 kW 21.58 minutes w/o traffic consideration 4082.5 kW 35.05 minutes

Outline

• Background and motivation
• Environment-friendly operation strategies
• Robust unit commitment considering air pollutant

dispersion

• Distribution system dynamic reconfiguration
• Resilient operation strategies
• Remote-controlled switch allocation enabling prompt

restoration

• Mobile emergency generator dispatch
• Co-optimize service restoration with dispatch of repair

crews and mobile power sources

Co-optimize restoration with RC and MPS dispatch

• Disaster recovery logistics
• Distribution system: network reconfiguration, microgrid formation, etc.
• Routing and scheduling of repair crews
• Routing and scheduling of mobile power sources

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

Co-optimize restoration with RC and MPS dispatch

• Routing and scheduling of repair crews

The issues of transportation-power networks coupling and their different timescales are resolved in a simpler manner compared with the common traveling salesman problem formulation.

Co-optimize restoration with RC and MPS dispatch

• Routing and scheduling of mobile power sources

Co-optimize restoration with RC and MPS dispatch

• Distribution system network reconfiguration

Removing l ≥ 0 edges from a spanning tree leads to a spanning

• forest. (Subgraphs of a spanning

tree are also spanning trees, thus forming a spanning forest.)

Co-optimize restoration with RC and MPS dispatch

• The co-optimization objective
• Coupling between service restoration & repair crew dispatch
• Coupling between service restoration & mobile power source

dispatch

Co-optimize restoration with RC and MPS dispatch

• Solution method
• Pre-assign a minimal set of repair tasks to repair crews
• Reduce the number of candidate nodes for mobile power source

connection

Co-optimize restoration with RC and MPS dispatch

• Case studies

The proposed method is especially effective in coordinating RC dispatch and MPS dispatch to restore loads by dynamically forming microgrids in the DS. The microgrids are powered by MPSs, and reconfigured and extended by switch actions

• f the DS and repair actions of RCs.

Co-optimize restoration with RC and MPS dispatch

• Case studies

Thank you!