2016 Smart Grid R&D Program

Peer Review Meeting

(Fast Response to Major Outages in Secondary

Distribution Networks)

(Chen-Ching Liu)

(Washington State University)

(August 17, 2016)

December 2008

Fast Response to Major Outages in Secondary Distribution Networks

Objectives & Outcomes

Life-cycle Funding Summary

($K)

Prior to

FY 16

FY16,

authorized

FY17,

requested

Out-year(s)

150 280 160 160

Technical Scope

(Note: The life-cycle funding table above should include all FY funds received and to be requested, from the project beginning year to the project ending year)

To enhance the resilience of distribution systems with

respect to extreme events, microgrids and distributed

generators can be utilized to serve critical loads

connected at distribution feeders. Service restoration

strategies have been proposed for both radial distribution

systems and secondary networks. The proposed method

has been applied to the Pullman-Washington State

University Distribution System. A field test is planned and

expected by the end of 2016.

• Small distributed generators have limited ability to absorb

shocks and maintain system stability. Generation

resources within microgrids can be limited and hard to

support after an extreme event. Formulation and

algorithm considering dynamic and generation-resource

constraints are proposed. The proposed method has

been evaluated with the PNNL test system and Pullman

distribution system.

• Specific technical issues associated with secondary

network service restoration have been identified. A new

method has been proposed to address these issues.

Transient simulations are performed with PSCAD for

demonstration.2

December 2008

Progress and Results Since Last Review Meeting

• Microgrids for service restoration to critical load in a resilient (radial) distribution system (FY15)

• What has been done: problem formulation, algorithm, case study, application to Pullman system, IEEE journal paper

• Pullman-WSU distribution system field test (FY16)

• What has been done: partners, test plan

• What will be done: implementation, final report

• DG-critical load restoration in a secondary network distribution system (FY16)

• What has been done: technical issues, problem formulation, algorithm

• What will be done: case study, IEEE paper

3

December 2008

Microgrids for Service Restoration to Critical Loadin a Resilient Distribution System

• Resilience: “..ability to prepare for and adapt to changing conditions and withstand and recover rapidly from disruptions..”*

• When a major outage occurs, microgrids can be controlled to provide an efficient service restoration strategy to restore critical loads in a distribution system and improve the resilience.

• Challenges:

• Distributed generators (DGs) in a microgrid have relatively smallcapacity. They have limited ability to absorb shocks and maintain stability

• Scarcity of generation resources: Fuelsfor generators, e.g., diesel and naturalgas, electric energy in storage devicesare limited and hard to support after an extreme event

4

Restoration schemes considering

DERs and Microgrids

Microgrid

* Office of the Press Secretary of the White House, Presidential Policy Directive

21 – Critical Infrastructure Security and Resilience [Online]. Available:

http://www.whitehouse.gov/the-press-office/2013/02/12/presidential-policy-

directive-critical-infrastructure-security-and-resil

FY15

December 2008

Methodology for Critical Load Restorationin a Radial Distribution System

• Problem Formulation

• Objective: maximizing the cumulative service time of microgrids to loads on the distribution feeders weighted by their priority.

• Dynamic, generation-resource, operational, and topological constraints are considered.

• Algorithm

• By introducing the concepts of restoration tree and load group, restoration of critical loads is transformed into a maximum coverage problem, which is a linear integer program (LIP). The restoration paths and actions are determined for critical loads by solving the LIP.

• Case Study

• The proposed method has been tested on the PNNL 1069-node test system.

• Application

• A strategy using WSU generators to restoration critical loads in the Pullman distribution system, i.e., Pullman Regional Hospital and City Hall, is obtained by applying the proposed method.

• GridLAB-D dynamic simulations are performed for PNNL test system and Pullman-WSU distribution system.

5

FY15

* Y. Xu, C. C. Liu, K. P. Schneider, F. K. Tuffner, and D. T. Ton, “Microgrids for Service Restoration to Critical Load in a Resilient

Distribution System,” Accepted for IEEE Trans. Smart Grid.

December 2008

Pullman-WSU Distribution System Field Test

• Purposes:

• Demonstrate the feasibility of using DGs within microgrids to serve critical loads after a major outage.

• Validate simulations conducted with GridLAB-D. Once validated, we can explore other operational strategies via simulation.

• Partners: PNNL, WSU Facilities Services, SEL, Avista, Enercon, PCE, and Cummins. Feasibility study contract signed with SEL as the lead.

• Progress and Timeline:

• A detailed field test plan has been proposed. A report by SEL is to be completed in August. A draft has been submitted on Aug. 3.

• Proposal to WSU and Avista management for approval of actual field test.

• Field test will be performed by end of November, 2016, if approved.

• A final report including the test plan, results, and analysis will be submitted to DOE by end of December, 2016.

6

FY16

December 2008

SEL Performed Detailed Study on Field Test Plan

7

G3

Natural

Gas

G2

Diesel

EBG3 G2

EP-2 EB-11 EB-M

WSU Loads

T-B

4.16/13.2 kV

SPU122

SPU-XFRM_1

13.2/115 kV

SPU121

Y – Δ

Y – Δ

P1411

Feeder

SPU122

Test Sequence

• De-energize feeder SPU122 from 115kV side of transformer SPU-XFRM_1. Open breakers on the path from

WSU generators to SPU substation, including breakers SPU122, P1411, EB-M, and EB-11.

• Close breaker EP-2. Close G3 breaker. Synchronize G2 with G3. Close EB-11. Energize some WSU loads by

closing EB-10, EB-5, EB-8, EB-3, EB-7, EB-2, EB-1, and EB-4, loading the generators to 1.8MW.

• Energize transformer T-B, feeder SPU122, and transformer SPU-XFRM_1 by closing breakers EB-M, P1411,

and SPU122, respectively.

• Disconnect WSU generators, re-energize feeder SPU122 and WSU loads from Avista side.

FY16

December 2008

Outline of Report from SEL

8

• Description of the test system

• Settings that need to be applied before field test• Settings of protective relays, measurement device upgrade and

adjustment, modifications of Enercon’s and PCE’s SCADA systems

• Test procedures: a complete list of operations included

• Risk analysis• Equipment damage prevention, including electrical protective and

monitoring systems analysis, generation protection system analysis,

breaker EB-M, EB-11, and EP-2 protection system analysis, and in-rush

and load pickup current evaluation (see figure below)

FY16

Data extracted from SEL-300G relay protecting WSU generator

December 2008

Result of Feasibility Study

9

• The proposed field test plan has

been validated by SEL and

partners subject to temporary

modification of relay settings

and automatic functions.

• The target date for the field test

is between September 15 and

October 31.

• The proposed field test plan will

require final approval of WSU

administration and Avista.

FY16

December 2008

Critical Load Restoration in Network Secondaries

10

DG2

DG1

DS

CL2 CL1

CL3

Rooftop

Solar Battery

Backup

DG3

Standalone

• In FY16, we are extending our work from radial systems to Secondary Networks

• Secondary Networks are widely used in metropolis downtown and central business districts.

• Our Goal: Improve resilience of secondary networks using service restoration.

• Progress:

• Identify technical issues and formulate critical load restoration problem (completed)

• Design algorithm for optimal or near-optimal solution (on-going)

• Submit an IEEE conference or journal paper by the end of December, 2016.

CL: Critical Load

* ABB Power Systems Inc., Electrical Transmission and Distribution Reference Book, 4th ed., “Chapter 21: Primary and secondary network distribution systems,” pp. 689-718.

FY16

December 2008

Technical Issues

Load Division among DGs– Depend on locations and control schemes of DGs, as well as

impedances of secondary mains, transformers, and filters

– Ideally, DG ’s share of power is proportional to its capacity

Voltage Regulation– Maintain secondary voltage by DG control

– Avoid inverse power flow through network protector

Cycling of Network Protectors– Undesirable tripping and closing of a network protector,

referred to as cycling of the protector, may occur when there are DGs on the primary feeder and secondary network

Synchronization– Network protector does NOT measure frequency

In-Rush– Transients can occur when line sections, secondary mains, and

transformers are energized

– In-rush may lead to violation of DG capacity limits, deviation of system voltage/frequency, and tripping of protective devices 11

DG1

DG2

FY16

December 2008

Problem Formulation

Objective: maxσ𝑖 𝑐𝑖𝑡𝑖– Maximize cumulative service time to critical loads weighted by their

priority

– Load division among DGs will affect the value of objective function

Constraints– Issues related to network protectors (NPs): operation of NPs,

synchronization

– Dynamic constraints: stability, transient voltages and currents (including in-rush), steady-state and transient frequency

– Limited generation resources: amount of diesel/natural gas, state-of-charge (SOC) of batteries

– Operational constraints: unbalanced three-phase power flow, limits on steady-state voltages and currents, and limits on capacity

12

FY16

December 2008

Algorithm: DG-Critical Load Restoration Strategy

• Assumptions

• A central control system is available for switching operations, DG control, and optimization

• DGs are connected at the primary feeders and secondary network, which can be used to serve critical loads at secondary network

• NPs do not have the ability to synchronize two dynamic systems

• Procedure to determine service restoration strategies

• Step 1: Only one primary feeder with DGs can be used for service restoration. The one with maximum total capacity (kW) of DGs will be selected.

• Step 2: Optimal combination of DGs and critical loads in secondary network can be determined by solving a Linear Integer Program (LIP), where only DG capacity and generation resources constraints are considered.

• Step 3: Determine a sequence in which selected DGs and critical loads are connected to the network. Simulations of the dynamic and operational constrains are performed. Rank DGs by its capacity and critical loads by its priority. Greedy search is used to determine the sequence for DGs to pick up critical loads.

• If a feasible sequence is found to connect all selected DGs and critical loads to the network, solution is found.

• Otherwise, remove the critical load that cannot be restored and go to step 2.

13

FY16

December 2008

An Example of Service Restoration toCritical Loads in a Secondary Network

14

DG4 will start serving CL1 at

the moment when utility

power is unavailable

DG1 is used to pick up CL2

– Close CB1, NP1, and SW1

– Evaluate the effect of in-rush

DG3 is used to restore CL3

– Close CB3 and SW2

– Load evenly divided between

DG1 and DG3

Connect CL1 to the network

and reallocate load among

DGs

– Close CB6

– Shift 100 kW load from DG4 to

DG1 and DG3

Critical load restoration is

completed

DG1

DG2

DG3

CL1, 200 kW

Backup Gen

200kW, 5-h fuel

CB1

CB2

CB3

CB6

NP2 NP1

Feeder 2 Feeder 1

CL2

200 kW

SW1

SW2

CL3

100 kW

DG4

150kW, 8-h fuel

250kW, 10-h fuel

250kW,

10-h fuelSecondary Mains

CB4CB5

100

200

0

200

Outage

Duration:

10 h

1

2

36

4

5

FY16

December 2008

An Example with Transient Simulations

15

DG1:

installed at

primary

feeder

DG2: installed at

secondary network

Restore Critical Load 1

1 DG1 energizes primary feeder and network

transformers. Key issue: in-rush current.

2 2

Network Protectors close.

Secondary Network

energized by DG1.

2 2

5 Restore Critical

Load 2

6 Restore Critical

Load 3

3

4

4 Connect DG2 to

secondary network. Key

issue: synchronization.

DGs are modeled as synchronous generators. AWoodward diesel governor model (DEGOV1) is used to maintain frequency. Asimplified exciter system model (SEXS) is used to regulate the generator’s terminal voltage.

FY16

December 2008

Transient Simulation Results

• In-rush Current when energize the primary feeder and network transformers.

16

• Synchronization between DG2 and the secondary network.

—— DG2 ——DG1

Adjust rotor speed of DG2

Connect DG2 to

the network

Rotor Speed (p.u.) Voltages at the terminals of

Breaker CB2, Phase A (kV)

Connect DG2 to

the network

Current of DG1 (kA)

FY16

December 2008

Collaborations and Technology Transfer

• Collaborations

• PNNL: WSU has been working with PNNL on distribution system service restoration using microgrids and distributed generators for three years. GridLAB-D (with unbalanced three-phase power flow and dynamic simulation capabilities), developed by PNNL, is a powerful tool supporting our research. Several joint papers are published.

• Avista: Avista has been supporting our research by providing model and data of Pullman distribution system and in-kind service for field test.

• Field Test Partners: WSU Facilities, SEL, Avista, Enercon, PCE, Cummins

• Software Development

• Spanning Tree Algorithm for Distribution System Restoration has been integrated into GridLAB-D Version 3.2

• MATLAB software developed for Critical Load Restoration with Microgrids

• MATLAB software developed for Optimal Placement of Remote-Controlled Switches in a distribution system

• MATLAB software developed for Distribution System Reliability Assessment Considering Service Restoration

• Field Test Experience

• A detailed field test plan using WSU generators to energize a Avista feeder and substation transformers has been developed. 17

December 2008

Publications

• Publications after last review meeting:

• Y. Xu, C. C. Liu, K. P. Schneider, F. K. Tuffner, and D. T. Ton, “Microgrids for Service Restoration to Critical Load in a Resilient Distribution System,” Accepted for IEEE Trans. Smart Grid.

• Y. Xu, C. C. Liu, K. P. Schneider, D. T. Ton, “Placement of Remote-Controlled Switches to Enhance Distribution System Restoration Capability,” IEEE Trans. Power Systems, March 2016.

• Y. Xu, C. C. Liu, K. P. Schneider, and D. T. Ton, “Toward a Resilient Distribution Systems,” IEEE PES General Meeting, July 2015.

• Y. Xu, C. C. Liu, H. Gao, “Reliability Analysis of Distribution Systems Considering Service Restoration,” IEEE PES ISGT, Feb. 2015.

• F. D’Agostino, F. Silvestro, Y. Xu, C. C. Liu, K. P. Schneider, and D. T. Ton, “Reliability Assessment of Distribution Systems Incorporating Feeder Restoration Actions,” Power Systems Computation Conference (PSCC), June 2016.

• K. P. Schneider, F. K. Tuffner, M. A. Elizondo, C. C. Liu, Y. Xu, and D. T. Ton, “Evaluating the Feasibility to Use Microgrids as a Resiliency Resources,” Accepted for IEEE Trans. Smart Grid.

• Publications before last review meeting:

• J. Li, X. Y. Ma, C. C. Liu, K. P. Schneider, “Distribution System Restoration with Microgrid Using Spanning Tree Search, “ IEEE Trans. Power Systems, Nov. 2014, pp. 3021-3029.

18

December 2008

Lessons Learned

19

• What Worked Well

• By working with PNNL closely, we are able to simulate switching actions for Pullman-

WSU distribution system using GridLAB-D.

• Performing a field test is much more complicated than performing simulations. Many

practical issues need to be considered, i.e., settings of protective relays, adjustment

of control and management systems, etc. By working with industrial partners, we

have come up with a practical field test plan, which cannot be achieved by

simulations only.

• What Could be Improved

• Avista installed a 1-MW battery in Pullman (Turner 116/117 Feeder) in April, 2015. In

our study of the Pullman-WSU case, the battery has not been taken into account.

• Transient simulations with PSCAD/EMTDC® only serves as a demonstration.

However, it is time-consuming and impractical to use PSCAD for simulation of a

large network. GridLAB-D, which has dynamic simulation capability, will be used for

case study and work with optimization algorithm.

• In our study, communication failures have not been considered.

December 2008

New: Interfacing with DMS in Testbed

TCP/IP

e-terradistributionTM

GE Grid Solutions

Interface Module

DMS

Interface

CSV

Files

System Topologyand DPF Results

Research Applications

(e.g., Spanning Tree)

DMS

Smart City Testbed ... ...

... ...

Data Acquisition and Restoration Actions

e-terra

browser

20

December 2008

Pullman DMS Model

*Pullman data and system model provided by AvistaGE eterradistribution model

1 MW Battery

(Avista)

72 kW Solar Array (WSU)

• Pullman Turner 117 Feeder

21

December 2008

Spanning Tree Restoration Actions

IDs of devices in DMS:

• TUR117_395-2425532_68

• TUR117_395-2425534_69

• TUR117_395-2425536_70

22

December 2008

Plan for FY17

• Resiliency in a complex distribution system environment with microgrids, renewables, and storage

• Method to handle optimal utilization of energy in an extreme condition

• Reconfiguration for service restoration of distribution systems with non-radial topology and multiple sources

• Steady-state and dynamic performance (GridLAB-D simulation)

• Inverters as control devices in a complex system condition

• Development of use cases through industry collaboration

• Validating the proposed method with the WSU DMS testbed with the full model of Pullman distribution feeders

23

December 2008

Contact Information

24

Chen-Ching Liu, Ph.D.

Boeing Distinguished Professor of Electrical Engineering

Washington State University

School of Electrical Engineering and Computer Science

PO Box 642752

Pullman, WA 99164 USA

Tel: 509-335-1150

liu@eecs.wsu.edu

December 2008

Back-up Slides

25

The back-up slides contain details of our work for FY15

December 2008

Problem Formulation:Critical Load Restoration in a Radial Distribution System

• Objective: Maximize cumulative service time to critical loads weighted by their priority

• Constraints:

• Dynamic constraints• Stability and limits on steady-state frequency• Limits on transient frequency• Limits on terminal voltages and currents of DGs

• Generation-resource constraints• Limits on the amount of energy a microgrid can provide to

external critical loads• Operational Constraints• Unbalanced three-phase power flow• Limits on steady-state bus voltages and line currents• Limits on steady-state output power of DGs

• Topological Constraints• Maintain a radial network structure

26

FY15

December 2008

Algorithm to Determine Restoration Strategy

• A four-step graph-theory-based heuristic is proposed:

• Step 1: Feasible restoration paths microgrids to critical loads are identified

• Step 2: Load groups are formed. A load group is a subset of load zones that can be restored as a group by a microgrid through their restoration paths

• Step 3: Formulate and solve a maximum coverage problem

• Step 4: Determine restorative actions

27

Identify Feasible

Restoration Paths

Form Load Groups

Solving a Maximum Coverage Problem

Z1 Z2 Z3 Z4 Z5 Z6

Z8Z7

S2 S3 S4 S5 S6

S10S9

S7 S8

DG

M

S1

S13

Z9

Source 1

Restoration paths

starting from Source 1

Z1 Z2 Z3 Z4 Z5 Z6

Z8Z7

S2 S3 S4 S5 S6

S10S9

S7 S8

DG

M

S1

S13

Z9

Source Load Zones

1 Z3, Z2, Z1 (CL1)

1 Z3, Z7 (CL2)

2 Z8, Z9 (CL3)

Examples of load groups

Restoration

Strategy

CL1

CL2 CL3

CL1

CL2 CL3

FY15

December 2008

Case Study: the PNNL Test System

28

• Utility power unavailable

• 7 faults

• 4 microgrids

• 5 critical loads

• Restoration path identified (green paths)

• Switching operations determined

FY15

December 2008

Case Study: the PNNL Test System (Cont’d)

29

Microgrid 1 restores

critical loads CL1,

CL2, and CL3 in five

switching operations.

Microgrid 3 restores

critical loads CL4 in one

switching operation.

• Transient Frequency of Microgrids 1 and 3

FY15

December 2008

Application: Using WSU Generators to Serve Critical Loads in the Pullman Distribution System

• A diesel and two natural-gas generators on WSU campus are used to serve the Pullman Regional Hospital and Pullman City Hall.

30

Generation Resources• Natural Gas: Pipeline from British Columbia, Canada,

with a back up• Diesel: 250,000-gallon fuel tank; 10,000 gallons per

delivery, 8 deliveries per year• Worst Case: Two pipelines are damaged, a full tank of

diesel can serve WSU critical loads for 5-7 days.

Frequency

DG Terminal Voltages

FY15

December 2008

A Computational Tool for Reliability Assessment Considering Service Restoration

• Evaluate impact of service restoration strategies and distribution automation

• Determine optimal switching sequence to minimize target reliability index

31

Distribution System

Restoration (DSR)

Program

Reliability

Analysis

Program

Switching

Operations

Distribution

System

Information

Set Target

IndexSwitching

Sequence

Values of

Reliability Indices

Reliability Indices considered in the program include:

• SAIDI, SAIFI, CAIDI, ASAI, ASIFI, and ASIDI

FY15

December 2008

Reliability Assessment Considering Service Restoration: the Proposed Methodology

32

• A set of scenarios

• States

• Transition arches

• Levels

• Optimal path

• SAIDI per event

• Calculate Indices

POST FAULT STATE

FINAL RESTORATION STATE

Find optimal

switching sequence

Identify shortest

path

The contribution to reliability indices are

calculated once the shortest path is identified.

FY15