Load Modeling in WECC

DOE-NERC FIDVR Workshop September 2009 Washington DC

Dmitry Kosterev(*) and John Undrill (**) (*) Bonneville Power Administration (**) John Undrill LLC

Load Modeling Timescale of Interest

Time

cycle minute

hour

day

year

10 years

Our primary interest is the dynamic behavior of loads in cycle to minute time frame, not projections of future demand.

second

Load Model Development

History Of Load Modeling in WECC • 1990’s – Constant current real, constant impedance

reactive models connected to a transmission bus • 1990’s – IEEE Task Force recommends dynamic load

modeling, however the recommendation does not get traction in the industry

• 1996 – Model validation study for July 2 and August 10 system outages: – Need for motor load modeling to represent oscillations and

voltage decline • 2000 – 2001 – WECC “Interim” Load Model:

– 20% of load is represented with induction motors – Tuned to match inter-area oscillations for August 10 1996 and

August 4, 2000 oscillation events – Recognized model limitations and the need to continue…

o

o

o o o o

o

o

o o o o

Simulations – instantaneous voltage recovery

This is what we thought would happen using “interim” load model…

Motivation for Better Load Modeling

30 seconds

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

1.10

0 5 10 15 20 25 30Seconds

Volta

ge (p

u)

… and this is what actually happened

Reality – 30-second voltage recovery, 12 seconds below 80%

Motivation for Better Load Modeling

30 seconds

100%

75%

History Of FIDVR Load Modeling • Late 1980’s – Southern California Edison observed events

of delayed voltage recovery attributed to stalling of residential air-conditioners

• 1994 – Florida Power published an IEEE paper on the similar experiences of delayed voltage recovery following transmission faults

• 1997 – SCE model validation study of Lugo event: – Need to represent a distribution equivalent – Need to have special models for air-conditioning load

• 2005–2008 – Similar observations are made by Southern Company in their analysis of FIDVR events in Atlanta area

Looking Forward • Loads will play much more influential role in power system

stability • Resistive-type loads are phasing out (incandescent lights

and resistive heating) – Energy inefficient but “grid-friendly”

• Electronic loads, VFDs, AC compressors, heat pumps, CFLs are moving in – Energy efficient but have undesirable characteristics from

standpoint of grid dynamic stability – Increasing penetration of residential air-conditioners

WECC Load Modeling Task Force

• WECC LMTF was formed in 2002

• Component-based approach – Understand load behavior over the wide range

of disturbances

• Bottom-up model development

• Top-down model validation

Questions • How to capture the variety of electrical end-

uses in the model ? • How to represent electrical distance between

the transmission bus and a end-user buses with a distribution equivalent ?

• How to represent seasonal / daily / geographical variations in load composition ?

• How to validate load model performance ?

Modeling Electrical End-Use

LBNL Electricity Use In California Study

0

1

2

3

4

5

6

7

8

Res. - AirConditioning

Com'l. - AirConditioning

Com'l. -InteriorLighting

Com'l. - Other Res. -Miscellaneous

Res. -Refrigerator

Com'l. -Ventilation

Res. -Cooking

Res. - Dryer Com'l. -Refrigeration

0

5

10

15

20

25

30

35

Peak Demand

Annual Consumption

Source: CEC Demand Analysis Office

Peak Demand (GW) Consumption (TWh)

Residential AC

Commercial AC

Lighting

Refrigeration Ventilation

Residential Single-Phase Air-Conditioners

A/C Compressor Motors are Prone to Stall

7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 120

50

100

150

200

250Voltage (Volts)

7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 120

50

100

Current (Amps)

Time (sec)

7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 120

5

10

15

20Active Power (kW)

7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 120

5

10

15

20

Time (sec)

Reactive Power (kVAR)

Dip to 55% for 3 cycles

Stall Thermal Trip

To find out WHY, SCE, BPA and EPRI tested more than 30 AC units

(a) A/C Compressor Motors are non-symmetric

R

S

C Run Capacitor

Thermal Relay

Auxiliary Winding Main

Winding

1 phase supply, 2 windings

capacitor-run motor

(a) A/C Compressor Motors are non-symmetric

Non-symmetry becomes more pronounced as the supply voltage is declining The forward-rotating torque component is counter-acted by backward-rotating component at lower voltages

0 10 20 30 40 50 60 700

50

100

150

200

250Voltage

0 10 20 30 40 50 60 700

20

40

60Current

0 10 20 30 40 50 60 700

1000

2000

3000

4000

5000Compressor Power

TotalPositive Seq.Negative Seq.

TotalMainAuxiliary

(a) Single-phase motors produce much less electrical torque than a comparable 3-phase motor

at lower voltages and slower speeds

0 50 100 150 200 250 300 350 4000

2

4

6

8

10

12

14

16

18

rotor speed

torq

ue

Critical speed = 30% of rated

Critical voltage near 60%

Torque/speed and load torque curves Plot prepared by Dr. Bernard Lesieutre, LBNL (now with University of Wisconsin)

Electrical torque for 1-phase motor at various voltages Electrical torque for 3-phase motor at nominal voltage Mechanical load torque

(b) Compressor Motors Inertia is Very Low

310 mm

75 mm

E.g. 3.5-ton compressor motor: Weight: 4.6 kg

H = 0.03 – 0.05 seconds

(c) Compressor Load Torque

360 720

Torq

ue

Rotor Position

It is very possible that the motor stalls at the next compression cycle

Types of Compressors • There are two types of compressors – reciprocating and

scroll – Reciprocating compressors represent majority of the installed

capacity – Scroll units will be majority of new units moving forward

• Scroll compressors have more favorable characteristics: – Less prone to stall (lower voltage, longer sag) – Can re-accelerate when the voltage recovers above 90% – Pressure seems to equalize faster – However, can restart and run backwards for hours

0 10 20 30 40 50 60 70 800

50

100

Cur

rent

(Am

ps)

Current

0 10 20 30 40 50 60 70 8050

100

150

200

Pre

ssur

e (P

si)

Compressor Pressure

Time (sec)

What do the motors do after they stall ?

stall thermal trip

thermal trip

thermal re-close

0 5 10 15 20 25 30 35 40 45 50 550

20

40

60

80Current

Cur

rent

[pu]

0 5 10 15 20 25 30 35 40 45 50 550

50

100

150

200Compressor Pressure

Pre

ssur

e [p

si]

Time [sec]

What do the motors do after they stall ?

stall thermal trip

successful restart

Compressor Motor Steady-State Loading

80 85 90 95 100 105 110 1152.6

2.8

3

3.2

3.4

3.6P

ower

(kW

)

Ambient Temperature (F)80 85 90 95 100 105 110 115

0.56

0.58

0.6

0.62

0.64

0.66

Sta

ll V

olta

ge (p

er u

nit)

* Compressor motors have high power factor ~0.97

How to Model Single-Phase AC Motors ?

• We were unable to fit the observed behavior into a three-phase motor model structure

• Several Modeling Approaches are considered, prototyped and tested: – Phasor Model (MOTORC) – Differential equations, used as a

benchmark

– Performance Model – empiric model based on the test results

– Hybrid model – three-phase model for running motor, short-circuit impedance when stalled (original SCE approach)

Performance Model Static Empirical Model

0 0.2 0.4 0.6 0.8 10

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5Real Power

Rea

l Pow

er (p

er u

nit)

Voltage (per unit)

RUNSTALL

STALL

0 0.2 0.4 0.6 0.8 10

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5Reactive Power

Rea

ctiv

e P

ower

(per

uni

t)

Voltage (per unit)

RUN

STALL

STALL

RUN STATE - Polynomial STALL STATE - Impedance Switch from RUN to STALL when voltage drops below Stall Voltage

Single-Phase Motor Model • 1-phase model includes:

– Performance model to represent a single-phase motor model

– Thermal relay model

– Under-voltage relay model

– Contractor model

• Model is validated against actual tests

• Model is tested for numeric stability

• Implemented in PSS®E (ACMTBL) and PSLF (LD1PAC)

Composite Load Model Structure

Composite Load Model Structure

Electronic

M

M

M 115-kV 230-kV

Static

M UVLS

UFLS

1-phase AC

3-phase compressor

Fan

Pump

Load Model Data

WECC Composite Load Model

Electronic

M

Load Model Composition Data

M

M 115-kV 230-kV

Static

Load Component Model Data

Distribution Equivalent Data

UVLS and UFLS Data

M

Load Model Data

• Load model data records include: – Distribution equivalent model data

• PG&E developed methodology, proven to work, feel very confident

– Model data for load model components (e.g. motor inertia, driven load, electrical data, etc)

• Getting test data, develop better understanding of end-use characteristics

• Motor protection and control remains least known

– Fractions of total load assigned to each load model component

– UFLS and UVLS data

Load Composition • The most challenging part of model data

• Sources of data: – LBNL Reports on Electricity Use in California

– California Energy Commission: 2006 Commercial End-Use Survey

– SCADA load profiles

– BPA end-use monitoring data

• PNNL-BPA Load Composition Model

California Commercial End-Use Survey • 15 climate zones in California • Four seasons • Typical, Hot, Cold, Weekend • 24-hour data

0 5 10 15 20 250

2

4

6

8

10

12

14x 10

5

FCZ10 Season:Summer Day:Typical

AirCompMotorsProcessMiscOfficeEquipIntLightExtLightRefrigCookingWaterHeatVentCoolingHeating

Data is available on CEC web-site

< 1%4%< 1%5%3%

25%

< 1%8%

3%< 1% 10%

41%

< 1%

FCZ10 Season:Summer Day:Typical Hour:

16

AirCompMotorsProcessMiscOfficeEquipIntLightExtLightRefrigCookingWaterHeatVentCoolingHeating

Load Composition Data • PNNL-BPA Load Composition Model • WECC LMTF provided regional defaults

– winter, summer, and shoulder seasons – 6:00, 9:00, 15:00, 18:00 hours – Multiple climate zones in WECC – Composition for residential, commercial and mixed load

types

• Utilities are encouraged to provide bus-specific load composition information

Load Model Studies

Load Model Studies

• Event validation studies, initial focus is on on-peak summer conditions: – Local FIDVR events in Southern California

– Interconnection-wide disturbances

• System impact studies (in progress) – Large generation outages in WECC

– 3-phase faults on major inter-tie

– 3-phase faults in large load centers

Event Validation Studies

Load Model Studies – Initial Observations • What is important for FIDVR studies:

– Load Composition Data • Bus-specific information is recommended

– Motor control and protection • Sensitivity studies are required

– With respect to “reasonable” variations of “key parameters”

– Scenario planning • Generators and SVC may trip because of depressed voltages • Plants due to internal protection and control issues • Generator protection operation (OEP during low voltages,

Volts/Hertz during high voltages) • Line protection operation

FIDVR Scenario Planning

Time

Bus Voltage

Sensitivity to load composition

Sensitivity to motor protection Line protection Generator and SVC tripping

Consequential plant trips Volts Per Hertz

Load Model Implementation in WECC • Load Model Approval:

– Composite Load Model structure (WE ARE HERE) – Load model data – Validation and system impact studies

• Process for Load Model Building

– Tools for managing load model data

• WECC Voltage Dip Criteria

Closing Remarks • Composite load model is capable of reproducing in

principle the phenomenon of delayed voltage recovery that is known to occur

• Composite load model is adequate to evaluate FIDVR exposure risks when used appropriately

• Composite load model is not intended to provide “accurate” representation of details of a particular feeder response

• Composite load model is a tool to help make an engineering judgment when appropriate scenario planning is used

Thank You