Distributed Solutions for Grid Control

GCEP WorkshopNov 1-2, 2007

Dr. Deepak Divan School of Electrical Engineering

Georgia Institute of Technology, Atlanta, GA, USA

777 Atlantic Drive deepak.divan@ece.gatech.eduAtlanta, GA

Confidential

Intelligent Power Infrastructure Consortium – Research Thrusts

Sensors, Communications,

Neural Nets, Monitoring

Power Systems

Engineering

PowerElectronics,Drives, Controls

Power line sensornetsWide area monitoringDistributed power line sensing

Power line communicationsNeural nets for system I/D and control

Wireless sensor networksSensorless motor fault detectionUnscheduled down-time identification

Neural network controls for PEComponent failure predictionPower and analog ICs

Active filtersMotor DrivesCurrent surge limitersHigh power inverters & dc/dc converters

D-FACTS devicesCurrent limiting conductorsFault current limitersPower quality and reliability

Wide area controlWide area protection

Power system operation and state estimation Protective relayingBulk power system reliability

Congestion managementSystem control and stability with FACTSDistributed energy sources

UTILITY APPLNS INDUSTRIAL APPLNS

Confidential

Confidential

Axioms for Smart-Grid Design• Achieving a sustainable energy infrastructure will

require an efficient, flexible, dynamic, fault tolerant and reliable electricity delivery network.

• Unlike other modern networks, the power grid lacks distributed intelligence and automation, resulting in poor reliability and asset utilization

• A complete replacement of the electricity infrastructure seems unjustified as there is no performance metric, e.g. ‘bandwidth’, that promises orders of magnitude improvementThis leads to several fundamental tenets for Smart Grid design.

• Changes in infrastructure will have to be incrementally built to enhance and augment the legacy infrastructure

• Solutions cannot degrade the intrinsic reliability of the system – high system reliability through redundancy and fail-safe modes.

• Resulting grid needs to be self-healing, exhibiting graceful degradation as communications and vital elements are lost

Confidential

The Future is Massively Distributed!

• Transformation of communications and computing sectors has occurred because of the use of massively distributed solutions, e.g.

– Super-computers to Cluster-computers– Satellite communications to Cellular communications– Point to point communications to the Internet– Multiple individual sensors to Sensor networks or Sensornets

• Standard, redundant, mass-manufactured parts in smart networks result in system performance and reliability gains while realizing lower cost.

• The new networks are robust and resilient and reorganize themselves as units fail.

• Systems continue to operate at basic sub-optimal levels as communication links fail.

• System exhibits graceful degradation and provides visibility to operator even as the main system fails.

• The sustainable energy infrastructure is predicated upon distributed sources and distributed smart loads, but the power delivery infrastructure is designed to operate with single point hierarchical control – is this the weak link?

Confidential

I-Grid – A Distributed Grid Monitoring Solution

$250 per sensor

ISP

ISPWAN

Provider

DA

TA IN

GE

STO

R

DYNAMIC

REAL-TIME

DATABASE

STATIC

DATABASE

WEB

ACCESS

FLEXIBLE

NOTIFI-

CATION

ENGINE

OTHER

DATABASES DATA

EXPORT

INTERNET

OR

INTRANET

WIRELESS

CONNECT

I-GRID PLATFORMREMOTE

MONITORS

REMOTE

USERS

WEB BROWSER

NOTIFICATION

UTILITY STAFF

& CUSTOMERS

END-USER

REPORTS

INTERNET

Courtesy: Soft Switching Technologies

Confidential

Wide Area Monitoring – August 2003 Blackout

Frequency data recorded by I-Grid on August 14, 2003

Data was available within minutes of the blackout. Cost was under $200 per monitored point.

Confidential

Power Line Sensornets

• Low-cost couplers that perform sensing, power scavenging and communications functions

• Operate as a self-organizing sensor network that performs distributed condition monitoring of the entire length of the power line

• Sensing of dynamic line capacity, imminent vegetation contact, failing insulators, security risks

• Improve system reliability and asset utilization

• Sensors that perform sensing, power scavenging and communications functions exist today – the cost is still too high

• Moving towards self-organizing sensor networks that perform distributed condition monitoring of the entire length of the power line

• Target functions include sensing of dynamic line capacity, power theft, imminent vegetation contact, failing insulators, security risks

• Objective: Improved system reliability and asset utilization

Courtesy: Protura

Confidential

Distributed Monitoring

• Distributed monitoring of the grid is rapidly becoming technically feasible. • Self-organizing ad-hoc sensor networks offer a new approach for realizing a

resilient and cost-effective distributed monitoring system• Distributed monitoring can enable several value streams – improved system

reliability, lower power theft, reduced downtime, improved asset utilization, and improved visibility/response to security threats

• Sensor and communication costs (including installation and operating costs) need to be extremely low

• Data latency, compaction and visualization are critical issues that need to be addressed.

• Having data on parameters such as the dynamic thermal capacity of a line can improve system utilization, but do we have the ‘muscle’ to realize the desired control for all lines in the system?

Confidential

Major Challenges for Power Delivery

• Existing infrastructure is aging and underutilized • New lines are expensive and difficult to get approved

(ROW & NIMBY)• As new generation and load is added, portions of the

grid are getting overloaded• Power flow in individual lines cannot be controlled.

This results in sub-optimal grid operation and poses significant operational challenges for utilities –

• Danger of load shedding and poor reliability under high-load and under contingency conditions

• Fault currents that may exceed safe levels• Lines that exceed thermal ratings, even as

neighboring lines are under-utilized• Conservative operation because of lack of visibility,

leading to significant under-utilization of assets• Inability to cope with dynamics and intermittencies

that are characteristic of a sustainable system.

Solutions are needed for improving the controllability and utilization of the power grid

Confidential

Typical Power Delivery Systems

Radial system

• Radial System– Easy to control power flow and node

voltage with conventional solutions such as tap changers, capacitors and phase shifting transformers

– Suffers from low system reliability under contingency conditions

• Meshed System or Network – Highest reliability due to redundancy of

transmission lines– Poor line utilization because current

flow in parallel lines cannot be controlled

– Difficult to satisfy voltage and line loading constraints simultaneously, even when grid state data is available.

Meshed system

Confidential

Meshed Distribution Grid – NYC

Confidential

Improving Grid Reliability and Utilization

• Reliable power is the paramount mission for a utility. • Cost and system-utilization can be more effectively controlled in radial

systems, whereas reliability of meshed networks is substantially higher. • However, it is difficult to control where current flows in a networked (meshed)

power system - results in loop flows, congestion and poor system utilization.

• In such networks, the first line that gets thermally limited constrains the power transfer capacity of the entire network.

• The real situation is worse, as (N-1) and (N-2) contingency conditions can result in overloaded lines and a cascading failure.

• Networks are often used in dense urban areas, such as New York, where reliability needs override asset utilization issues.

j16 Ω

j24 Ω

138kV∠0°

138kV∠7.75°675A

450A

Confidential

Other Issues Associated With Lack of Grid Control

• Static and dynamic balance of line currents under dynamic load, source and network conditions is not possible as there aren’t sufficient ‘control’ handles.

• Even if controllers could be deployed at every node, finding the optimal operating points would be difficult for a meshed network.

• Insufficient transmission capacity creates islanded networks that require higher level of generation reserves to ensure system reliability.

• Transmission constraints can limit ability of end-users to access low-cost generation, leading to higher costs.

• Building a new line can eliminate congestion, but will reduce line utilization and cost of electricity, making it harder to recover the investment!

• Utility operations are based on the assumption that it is not cost effective to control power flow along individual lines.

• Realizing dynamic control of power flows on the grid can have a profound impact on grid reliability, asset utilization and sustainability.

Confidential

Loading Levels for the IEEE 39 Bus System

•Poor line utilization, average of 59%

•(N-1) contingency reduces system capacity from 1904 MW to 1450 MW, utilization to 47%.

Confidential

Achieving Power Flow Control

LXVVP δsin21

12 =

• Active power flow can be controlled by changing the impedance of the line or the angle of bus voltages

• Conventional Solutions– Phase Shifting transformers: Expensive, dynamic

control is difficult– Shunt compensation: Provides voltage support

but does not have the ability to control real power flow

• Alternative Solution– Series compensation: Offers the strongest control

capability by dynamically changing the reactive line impedance Kalyan Sen Xian Tutorial

Confidential

Shunt vs. Series VARs for Power Flow Control

Simple two bus example with fixed voltage sources compares the power flow control using shunt or series VAR compensation

07.86.79 −∠006.79 ∠

0 10 20 30 40 50 60 70 8025

30

35

40

45

50

55

60

MWARS

MW

S T

hrou

gh L

ine

1 Series Compensation

Shunt Compensation

• Series VARs have a much stronger influence on the real power flow

• For series VARs, 8 MVAR increases power by 25 MW

• For shunt VARs, 80 MVAR increases power by 5 MW

• Decoupled control is needed –• Shunt VARs to control voltage • Series VARs to control power flow

Confidential

Existing Solutions for Increasing Grid Capacity

• Traditional solutions, such as new lines, are expensive and subject to siting and ROW delays. New lines also make system utilization worse.

• Shunt VAR compensation provides voltage support but has weak impact on controlling power flows in the system.

• Power flow control requires Series VAR Compensation – not widely used.

• Technology solutions such as Flexible AC Transmission Systems (FACTS) are well known and have been commercially available for >15 years.

Type Parameter FACTS Device

Type A – Series/Shunt P and Q UPFC

Type B - Series P TCSC, SSSC

Type C - Shunt Q SVC, Statcon

Confidential

Typical FACTS Configurations

Shunt Shunt

Series Series - Shunt

Confidential

Typical FACTS System

Series Series - ShuntEstablished in 1998, 160 MVA unit. This was the first UPFC installation in the World

Confidential

Experience With FACTS Devices

• High total cost of ownership. (The 200 MVA Marcy Convertible Static Compensator (CSC) cost $54 million.)

• High fault currents (60,000 Amps) and high voltage levels (1000 kV) make implementation of series FACTS systems difficult and expensive.

• Utilities are claiming reliability levels of ~96-98% for FACTS devices, significantly lower than the 99.999% reliability that is typical for the utility system itself.

• Highly customized and complex design of FACTS devices requires skilled work force in the field, normally not within a utility’s core competency.

• As a result, even though the technology is mature, significant market penetration has not occurred. There is an opportunity for a new paradigm in grid power flow control

STATCOM

Confidential

Distributed Solution for Series VARs – Smart Wires

• Distributed control of line impedances offers a new approach for controlling power flow in networked systems, allowing higher reliability & utilization

• Line impedance control is accomplished using a large number of ‘clamp-on’‘couplers’ (modules) that float on the line, electrically and mechanically.

• The modules use magnetic induction to increase or decrease line impedance, to ‘push’ current away from, or to ‘pull’ current into a line.

• Modules are ‘standardized’ and made in high volume using easily available low-cost components.

• Reliability is obtained through redundancy, where failure of individual modules has no impact on system operation.

• Smart Wires can be implemented using ‘passive’ or ‘active’ modules.

couplers

Confidential

Distributed Series Impedance – Passive Smart Wires

Power Line

Transformer

S1

Xc

XM

XL

R2

S

Inje

cted

Impe

danc

e

Indu

ctiv

eC

apac

itive

Controller

• Power flow control achieved through change of line impedance. • Injects series capacitance or inductance to change line impedance as needed.• A large number of modules realize gradual change in line impedance and increase

reliability. • A coupling device in the form of a Single Turn Transformer (STT) is used to inject

the required impedance.• System control and communications used to set level of injection.

Confidential

Distributed Series Reactance – Passive Smart Wires

I0 Ithermal

Line Current (ILine)

N Xm

Control

Power Line

Transformer

S1

XM

Power Supply

SM

• Simplest implementation of DSI, with inductive impedance injection (Current Limiting Conductor or CLiC) – functions as a current limiting system

• As current in a line approaches the thermal limit, CLiC modules incrementally turn on, diverting current to other under-utilized lines

• Increase in line impedance can be realized by injecting a pre-tuned value of magnetizing inductance of the STT

• Each module is triggered at a predefined set point to reflect a gradual increase in line impedance

• No communication required and the devices operate autonomously

Confidential

Can Smart Wires Work in Practical Systems?

138 kV 345 kV 765 kV Thermal Line Capacity 184 MVA 1195 MVA 6625 MVA Current carrying capacity 770 A 2000 A 5000 A # of conductors/ diameter (inches) 1/1.0 2/1.2 4/1.45 Reactance ohms/mile 0.79 0.60 0.54 Reactive voltage drop/mile 608 V 1200 V 2700 V 1% Compensation/mile 6.1 V 12 V 27 V Smart Wire kVA- 1% Comp/mile 14 kVA 72 kVA 400 kVA Total 10 kVA modules/mile/1% 1.4 7.2 40

Number of modules for 10% control on a 30 mile 138 kV line at 770A: 420 modules or 4.6 modules/phase/mile.

Confidential

Simulation Results – Passive Smart WiresIncrease in ATC by as much as 75%

0 0.5 1 1.5 2 2.50

0.25

0.5

0.75

1

Time (sec)

Line

Cur

rent

(KA)

Profile of Line 2 Current

DSR Active

548.7 A

750 A

624.2A

GeneratorTaken Off

Simplified Four Bus System

Contingency Condition: Generator Outage 20 40 60 80 100 120 140 160

20

40

60

80

100

120

140

160

Load 2 (MW)

Load

1 (M

W)

Increase in Transfer Capacity

Line 2 Overload

Line 5 Overload

Line 2,4 Overload

Line 2,5 Overload

Line 1,2,5 Overload

MVARMWPI

∆∆

=Network Performance Index:

Perf

orm

ance

Fac

tor

Inje

cted

MVA

Rs

Confidential

Prototype Passive Smart Wires Module Specifications

• Electrical– Operating : 161 KV / 1,000 A– Injected impedance= 10 mΩ per module– ACSR Conductor: Drake (795 Kcmil)

• R= 0.128Ω, X= 0.4Ω per mile– Fault level: 50,000 A

• Mechanical– Target weight per module: 120 lb (critical design parameter)– Packaging to avoid corona discharge, and other mechanical, thermal and

environmental issues

Confidential

CLiC Prototype

Confidential

Active Smart Wires

• Distributed Static Series Compensator (DSSC)– Active solution employing a synchronous voltage source inverter– Each module rated for 5 KVA (capable of injecting ± 4.6 V @ 1000 A)– Communication interface is required to realize the bi-directional control

Power supply

Main transformer

Filter

PWMInverter

DC Capacitor

V

ControlsCommunication

Module

Current feedbackLine Current

Confidential

Active Smart Wires Prototype – Experimental Waveforms

-600-500-400-300-200-100

0100200300400500600

-0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02

Time (s)

Cur

rent

(A) o

r V

olta

ge (V

)

I_total/2 I_1 (DSSC) I_2 (uncontrolled) V_DSSC*100

-800

-600

-400

-200

0

200

400

600

800

-0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02

Time (s)

Cur

rent

(A) o

r V

olta

ge (V

)

I_total/2 I_1 (DSSC) I_2 (uncontrolled) V_DSSC*100

Line2 Line1

(a)(b) (c)

-800

-600

-400

-200

0

200

400

600

800

-0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02

Time (s)

Cur

rent

(A) o

r V

olta

ge (V

)

I_total/2 I_1 (DSSC) I_2 (uncontrolled) V_DSSC*100

Line1-DSSCLine2

Injected Voltage x100 Min.

Module Rating: 1500 Amps @ 4.6 volts

Confidential

Smart Wires – Operational Benefits

• Ability to increase or decrease steady state line current autonomously or under system controller command

• Ability to respond autonomously in case of fast transients or communications channel failure

• Ability to monitor actual conductor temperature and manually or automatically limit current as a function of conductor temperature

• High system reliability due to massive redundancy, single unit failure has negligible impact on system performance

• Zero footprint solution• Robust and rugged under typical fault conditions• Can be used with conventional or advanced conductors• Mass produced modules can be stocked on the shelf and repaired in factory

– does not require skilled staff on site• Easy and rapid installation (may be possible on live line)

Confidential

Reliability/Economic Benefits

•System is resilient and operates as a self-organizing and self-healing power grid even in the face of unplanned contingencies

•Enhance network reliability and operation under (N-X) contingencies by automatically routing current from overloaded lines to lines with available capacity – reduces possibility of cascading failures

•Distributed scalable solution allows strategic, targeted and incremental deployment of modules for maximum budget flexibility.• Relieve network congestion at specific points as required.

• Defer investments in new power lines while enhancing system capacity

• Enable energy contracts between low-cost generators and interested end-users

using existing lines that have available capacity (merchant transmission)

•Zero footprint solution. Reduce or defer access to new ROW.

•Mass produced modules. High system reliability and availability. Rapid and

incremental deployment. Reprogrammable to meet changing needs

Confidential

Increase in Network Utilization

Network Performance With CLiC

Line

2_3

Line

6_5

Line

6_7

Line

9_39

Line

10_1

3

Line

12_1

1

Line

13_1

4

Line

19_1

6

Line

22_2

1

Line

23_2

4

Line

25_2

6

Line

26_2

7

Line

29_2

6

Line

29_2

8

Power Lines

Line

Cur

rent

s (%

Ther

mal

Lim

it)

0

20

40

60

80

100

Line currents with CLiCLine currents without CLiC

G1

G8

G10

230

1

G2

G3

G9

G4G5

G6

G7

39

9

8 7

5

4

3

18

37

25

17

26 28 29

38

24

27

15

14

12

13

1011

32 34

20

19

21 2235

23

36

16

6

31

22.76 MVAR

19.52 MVAR

18.77 MVAR

14.75 MVAR12.64 MVAR

3.52 MVAR

12.86 MVAR

12.04 MVAR

9.15 MVAR

• Increase in Transfer Capacity from 1904 MWs to 2542 MWs (59% to 93.3%)

(congested corridors and the required MVARs are shown by red lines)

• System simulations are used to establish worst case loading under normal and contingency conditions to compute number of DSR modules.

0.5 1 1.5 2 2.50

0.2

0.4

0.6

0.8

1

Time (s)

Line

Cur

rent

(KA)

Current Profile With CLiC Modules

Current Without CLiC Modules

GeneratorTaken Off

Current With CLiC Modules

940 A

643 A

940 A

CLiC Active

Current in Line 22-23 drops from 940A to 643 A

with CLiCs following outage of generator 7

Confidential

Increase in System Reliability and Capacity

• Baseline transfer capacity 1904 MWs• Under (N-1) contingency, capacity is limited to 1469 MWs• With (N-1) contingency, DSR’s can increase system capacity to 2300 MW• High PI until capacity reaches ~2000 MW (most cost effective regime)

Congestion relief with 2400 modules rated at 24 MVARs increases capacity by 530 MW220 KW of increased capacity per module

G1

G8

G10

230

1

G2

G3

G9

G4G5

G6

G7

39

9

8 7

5

4

3

18

37

25

17

26 28 29

38

24

27

15

14

12

13

1011

32 34

20

19

21 2235

23

36

16

6

31

51.53 MVAR

26.59 MVAR

3.84 MVAR12.78 MVAR

7.88 MVAR

6.26 MVAR

13.1 MVAR

Perf

orm

ance

Fac

tor

Inje

cted

MVA

RS

Confidential

Investment Cost on IEEE 39 Bus

1400 1600 1800 2000 2200 2400 2600 28000

0.5

1

1.5

2

2.5x 10

7

Load (MWs)

Inve

stm

ent C

ost (

$)

Line 22-21

Line 2-3

Line 6-5Line29-28

Line 19-16

Investment Cost withonly DSI technology

Investment Cost w ith acombination of DSI technologyand building additional lines

X

A

B'

A'

• Cost of laying additional lines: $500,000/mile• Target cost of DSR unit: $100/KVA

– Cost of redeployment: $30/KVA

• ATC of the system is limited to 1904 MWs• DSR modules can relieve network congestion

at a lower cost for the first 400 MWs– Initial performance index is ~7.4

• DSR continues to be attractive for the next 300 MWs

– Building additional lines can be suspended for the first 700 MWs

• A combination of the two schemes can provide a much lower investment cost

• As a line is built DSR modules can be redeployed to realize high performance index

Confidential

Solution Cost Under Contingencies

• (N-1) Line contingency is simulated here. (Line 19_16 is taken out. Worst case)

• Network congestion occurs at a load of 1469 MW, as opposed to 1904 MW with no contingencies

• CLiC modules provide a very attractive investment for the first 530 MW of load growth.

• If 2.5% load growth is assumed, the investment in the new line can be deferred for 13-15 years, even when contingencies are factored in. Reduces uncertainty and risk in building a new line.

• Interest on deferred cost of new line will pay for CLiC modules!

• Time for line permitting and building, and time value of money, indicate that the total cost of ownership for the DSR module solution may be significantly lower cost than for new line construction, even when the ‘first’ cost is the same.

1400 1600 1800 2000 2200 2400 2600 2800 30000

0.5

1

1.5

2

2.5x 107

Load (MWs)In

vest

men

t Cos

t ($)

Line 22-21

Line 21-16

Line 22-21

Line 2-3

Line 6-5Line 29-28

Confidential

Economic Impact of CLiC Modules

• In an open market, if Gen A had lower cost, he would have direct incentive to build new transmission capacity, to be paid for by windfall profits and increased sales.

• Under Locational Margin Pricing, neither Gen A nor the Transmission Operator get share of windfall profits, eliminating incentive for investment, and reducing competitive pressure on Gen B.

• If increasing marginal costs are factored in, Consumer A actually pays a higher price if new transmission capacity is built. Further, Consumer B gets no relief as he still pays the Gen B pricing. As a result, consumers do not benefit directly from transmission investment.

• Congestion costs are returned in the form of FTRs, insuring against pricing spikes.

• Societal benefit from building new transmission capacity is clear, but the existing market rules reduce the ROI and the benefit that can be derived from new capacity.

Confidential

Smart Wires in a Smart Grid to Enable Sustainability

•It is proposed that the use of distributed solutions based on low-power power electronics can allow utilities to move towards dynamically controllable meshed grids, significantly enhancing grid reliability, capacity and utilization. This enables-

• Improved coordination between sources and loads

• Improved dynamic coordination between regions

• Reduction in dynamic capacity reserve for generators

•Can be applied at the transmission, sub-transmission and distribution levels.

•Can be layered onto the existing infrastructure as desired, and will not degrade the inherent reliability of the existing system.

•Makes the grid self-healing, automatically maintaining safe operating levels even in the face of contingencies.

•Can significantly enhance grid capacity and utilization without building new lines.

•Redundancy and ability to operate with local data provide high system reliability and availability•Provides solutions that are low-cost and can be implemented in a gradual manner

as resources and budgets permit