Notes on Frequency Response for FERC Conference on September 23, 2010

Dmitry Kosterev, Steve Ashbaker, and Don Badley

September 21,2010

1. Background At a very basic level, the system frequency is governed by the Newton law of motion

Supply

Demand

1

2H

Frequency

GovernorResponse

D

Event

Frequency-responsiveDemand Response

Figure 1: basic representation of the system frequency governing

“Supply = Demand” means that the system frequency is constant (does not have to be at 60 Hz) “Supply < Demand” means that the system frequency is declining “Supply > Demand” means that the system frequency is rising

1

-5 0 5 10 15 20 25 3059.6

59.7

59.8

59.9

60

60.1

Fre

quen

cy (

Hz)

System Frequency

-5 0 5 10 15 20 25 30-0.03

-0.02

-0.01

0

0.01

Pow

er (

PU

)

Generation and Load Change

Supply = Demand

Supply < Demand

Supply = Demand

Supply > DemandSupply = Demand

LoadGeneration

-5 0 5 10 15 20 25 300

0.01

0.02

0.03

0.04

Pow

er (

PU

)

Total Response

Time (sec)

Total (Load + Generation) Response

Figure 2: illustration of frequency response for a 3% generation loss

2

System factors that affect the nadir frequency: - system inertia, H=4 seconds is often mentioned

o gas-turbine generators 5 to 6.5 seconds o steam-turbine generators 2.5 to 5 seconds o hydro-turbine generators 2.5 to 6 seconds o air-derivative gas turbines 2 – 2.5 seconds o electronically connected generation 0 o fans, propeller type, 0.5 to 1 second o air-conditioner compressor 0.05 to 0.1 seconds

- load damping, D=1 is often mentioned o fan motor D=3 o compressor D=1 o electronic loads D=0

- system size o Eastern Interconnection o Western Interconnection on-peak load as high as 161 GW, off-peak load is

as low as 73 MW o ERCOT

- size of the disturbance event relative to the system size

3

2. Frequency Response Characteristics

Very often, the frequency response discussion focuses on the “amount” of response requirement. “Amount” is only one of the attributes of response, there are other response factors that will have great influence on the nadir frequency. Simple test system to illustrate the point

LOAD, D=1

Baseloaded

Responsive

Tripped \ System size is 100 GW 3 GW of generation tripped All generators have inertia of 4 seconds Load damping D=1 Baseloaded generation does not response to frequency, produces the same MWs Responsive generation has droop setting of 5% and head room of 3GW

4

Case 1: different speed of response of “responsive” units Blue = gas-turbine unit on governor control Red = (fast) hydro-power unit on governor control Green = (ideal) steam-turbine unit on governor control

5

Case 2: different deployment rate Given the same MW amount of head room and speed, the nadir frequency will greatly depend on how the reserves are allocated. On the left side, all the reserves are put on a single unit. On the right, the reserves are spread among three units. With the same droop setting, the frequency drop for the case on the left case will be three times the frequency drop for the case on the right side. Nadir frequency depends on the

15 MW reserve with 10 MW per 0.1 Hz allocaionResereves are fully deployed at 59.85 Hz

0

20

40

60

80

100

120

1 2 3 4 5

Unit

RESERVE

GEN

15 MW reserve with 3.3 MW per 0.1 Hz allocaionResereves are fully deployed at 59.65 Hz

0

20

40

60

80

100

120

1 2 3 4 5

Unit

RESERVE

GEN

The following are inter-related (for a given required frequency response amount)

- Deployment Frequency = the frequency by which the required frequency response amount must be fully deployed

- Deployment rate (MW / 0.1 Hz) = Required frequency response amount divided by the Deployment Frequency

- Responsive Capacity = amount of synchronous capacity that is carrying the frequency responsive reserves

-

6

Simulations are done on the test system above to confirm the importance of the deployment ratet: 3GW of frequency response are distributed among:

- 20 GW of generating capacity (red) - 25 GW of generating capacity (blue) - 30 GW if generating capacity (green)

7

Case 3: Frequency response sustainability. 3GW of frequency response are distributed among the generators with the same governor control:

- Blue = frequency response is sustained - Red = generator has a “slow” load controller returning to MW set-point

8

CONCLUSION The frequency response requirement needs to be characterized by the following properties:

- Response amount (MW) - Deployment frequency (Hz) or Deployment Rate (MW/0.1Hz) - Response time - Sustainability

9

3. Questions related to the primary frequency response. Question 1: What is the frequency control objective?

1. What is the “design” event we are protection for? 2. What is the frequency performance for the “design” event? 3. Does the same performance need to be maintained under all system conditions? 4. How does the historic performance of the interconnection compare to the

frequency control objective? Question 2: Should be the requirement be expressed as Total System Response or as a Controllable Response? Total System Response = “Governor Response” + “Frequency-Responsive Demand

Response” + “Natural Load Response (to Voltage and Frequency)” – “Incremental Transmission Losses”

“Controllable Response” = “Governor Response” + “Frequency-Responsive Demand

Response” For example, WECC studies show that for a 2,800 MW generation outage, the 2,800 MW of total primary response is needed, from which 2,400 MW needs to come from governor response and 400 MW comes from natural load response minus incremental transmission losses. The implications of this decision:

- if the requirement is expressed as “total” response, the requirement needs to be assigned to individual BAs and measured at BA boundaries

- if the requirement is expressed as “resource” response, the requirement needs to be assigned to individual generators

10

4. Transmission system impact on Governor Response

Requirements WECC FRR studies indicated that the following two factors also play a significant role in arresting the system frequency decline:

- voltage sensitivity of loads, as the load voltages can change significantly during the initial disturbance transient

- incremental power losses due to re-distribution of power flows Simulation of 2 Palo Verde (Arizona) outage in Western Interconnection

WILLISTON

EDMONTON

COULEE

MALIN

TESLA

LUGO PALO VERDE

SAN JUAN

COLSTRIP

MONA

MIDPOINT

LANGDON

11

Frequencies across the Western Interconnection

System Frequency

59.75

59.8

59.85

59.9

59.95

60

60.05

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time (sec)

Fre

qu

en

cie

s (

Hz)

F:Edmonton 240

F:Williston 500

F:Colstrip 500

F:Coulee 500

F:Malin 500

F:Tesla 500

F:Lugo 500

F:PaloVerde 500

F:Mona 345

System Frequency

59.75

59.8

59.85

59.9

59.95

60

60.05

-5 0 5 10 15 20 25 30

Time (sec)

Fre

qu

en

cie

s (

Hz)

F:Edmonton 240

F:Williston 500

F:Colstrip 500

F:Coulee 500

F:Malin 500

F:Tesla 500

F:Lugo 500

F:PaloVerde 500

F:Mona 345

12

Generation outage creates a system transient affecting both, bus frequencies and voltages in the system. Voltage changes will be largest in the middle of the system. Embarcadero (downtown San Francisco) bus voltage, frequency and load following a two-unit Palo Verde outage

Voltage Embarcadero

0.995

1

1.005

1.01

1.015

1.02

1.025

1.03

-5 0 5 10 15 20 25 30

Frequency Embarcadero

0.9965

0.997

0.9975

0.998

0.9985

0.999

0.9995

1

1.0005

-5 0 5 10 15 20 25 30

Power Embarcadero

0.98

0.984

0.988

0.992

0.996

1

1.004

-5 0 5 10 15 20 25 30

Time (sec)

13

Change in the total system load during a disturbance: - red (lower) line – change in total load in an interconnection - blue (higher) line – change in total load minus incremental transmission losses

Change in System Load

-1200

-1000

-800

-600

-400

-200

0

200

-5 0 5 10 15 20 25 30

Time (sec)

Po

we

r (M

W)

Delta PGEN=LOAD+LOSS Delta PLOAD

14

5. “Design Event” and Frequency Performance in the Western Interconnection

WECC Reliability Criteria:

Category Events Controlled loss of load permitted?

Minimum frequency

B: N-1 NO 59.6 Hz, 6 cycles C: N-2, N-1-1 YES 59.0 Hz, 6 cycles

WECC largest N-2 is 2,800 MW corresponding to simultaneous outage of 2 Palo Verde generators. There are also several SPS schemes that will trip about 2,800 MW for N-2 transmission line outages under certain conditions. System Operating Limits of many major paths are set by N-2 outages. N-2 outages are stability-limiting, we do not want to be dependent on UFLS operation to keep the interconnection in synch. Generation outages of ~2,800 MW do occur. The consensus is that 2,800 MW is the “design” event for Western Interconnection, for which there will be no Under-Frequency Load Shedding. WECC coordinated UFLS program:

- a block of load at 59.5 Hz with 30 second delay - a large block of load at 59.3 Hz with almost no delay

Our conclusion: The nadir frequency needs to stay above 59.3 Hz (with some margin to account for oscillations) for an instantaneous outage of 2,800 MW of generation. What the settling frequency needs to be? Above 59.5 Hz (with some margin).

UFLS

Time, sec0 30

59.3

59.5

60.0

59.7

15

WECC Generation for 2009

0

20,000

40,000

60,000

80,000

100,000

120,000

140,000

160,000

180,000

0 10 20 30 40 50 60 70 80 90 100

Percent of time

Gen

erat

ion

(M

W)

WECC total generation peak > 150GW, off-peak about ~ 75 GW 2,800 MW generation loss represents less than 2% of system capacity on-peak and 3.7% of system capacity off-peak.

16

6. Historic Performance of Western Interconnection 2,812 MW RAS event on June 17, 2002

System Frequency, July 17 2002 event

59.5

59.6

59.7

59.8

59.9

60

60.1

-10 -5 0 5 10 15 20 25 30 35 4

Time (sec)

Fre

qu

en

cy

(H

z)

0

Grand Coulee frequency (Hz) JohnDay frequency (Hz) Malin frequency (Hz) 2,815 MW RAS event on May 20, 2008

System Frequency, May 20 2008 event

59.5

59.6

59.7

59.8

59.9

60

60.1

-10 -5 0 5 10 15 20 25 30 35 4

Time (sec)

Fre

qu

en

cy

(H

z)

0

ChiefJoseph frequency (Hz) JohnDay frequency (Hz) Malin frequency (Hz)

17

West Wing fault in Arizona on June 14, 2004: - 3,900 MW lost at 0 seconds on plot scale

System Frequency, June 14 2004 event

59.5

59.6

59.7

59.8

59.9

60

60.1

-40 -20 0 20 40 60 80 100

Time (sec)

Fre

qu

en

cy

(H

z)

Coulee500 frequency (Hz) JohnDay500 frequency (Hz) Malin500 frequency (Hz) COI Power

2500

3000

3500

4000

4500

5000

5500

-40 -20 0 20 40 60 80 100

Time (sec)

Po

we

r (M

W)

Captain – Jack – Olinda 500-kV line was out of service during the disturbance

18

7. Primary Frequency Response Requirements in the Western Interconnection

A. If the requirement to be expressed as a “total” response Measured at BA boundaries Requirement at 10 seconds (nadir) Amount: 2,800 MW = design event Deployment frequency = 0.5 Hz (deployment rate = 560 MW / 0.1 Hz) Requirement at 30 seconds (settle) Amount: 2,800 MW = design event Deployment frequency = 0.25 Hz (deployment rate = 1120 MW / 0.1 Hz) B. If the requirement to be expressed as a “controllable resource” response Measured at generators and loads part of the frequency-responsive demand response Requirement at 10 seconds (nadir): Amount: 2,400 MW = generation portion of the total response Deployment frequency = 0.5 Hz (deployment rate = 480 MW / 0.1 Hz) Requirement at 30 seconds (settle): Amount: 2,800 MW Deployment frequency = 0.25 Hz (deployment rate = 1120 MW / 0.1 Hz)

19

8. Simulated typical responses of various types of generators under governor control

9. Comparison30-second plot

- Frequency

-Steam Turbine on Governor Control

-Francis Turbine with Fast Governor

-Francis Turbine with Slow Governor

-Kaplan Turbine with Fast Governor

-Gas Turbine on Governor Control

20

21

Other points

1. While the current focus is on the primary frequency control, the entire duration of frequency response issues must be thought through first to ensure that there is an agreement between the primary, secondary and tertiary frequency control requirements.

2. Frequency response allocation mechanisms must be established. 3. Frequency response measurement can be a major challenge.

a. Likely to require transducers that measure generator active power and frequency at rate of 10 time per second

b. If allocated to BA, will require time-synchronized transducers that measure BA frequency at multiple locations and BA interchange flows

4. How to measure response for multiple sequential unit outages?