Wind Power Variability in The Grid: Regulation & Load Following

You might also like

Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 24

Wind Power Variability in the Grid:

Regulation & Load Following

J. McCalley
Outline
1. AGC
2. AGC and wind
3. Control performance standards (CPS)
4. Effect of AGC on CPS

2
Two Area System

BA 1 P12
BA 2

e0
Stiffness coefficient: T 
X

And if the two areas are operating with only primary


control, then their “slow” dynamics are represented by
the following block diagram.

3
Two area system with primary control
dynamics only
1
R1

-
+ 1 ΔPV1(s) 1 ΔPm1(s) + 1
Σ Σ
1+sTG,1 Δω1(s)
ΔPref,1(s) 1+sTT,1 - - M1s+D1
ΔPtie(s)
T1(s) ΔPNL1(s) G1(s)
+
T
s Σ
ΔPNL2(s)
-
- ΔPtie(s)
+
+ ΔPV2(s) 1 ΔPm2(s) + 1
1 Σ
Σ Δω2(s)
ΔPref,2(s) 1+sTG,2
1+sTT,2 M2s+D2
-
T2(s) G2(s)
1
R2
See http://home.eng.iastate.edu/~jdm/ee553/AGC1.pdf
4
State equations for this system
 1 1 1
PV 1 (t )   PV 1 (t )  1 (t )  Pref 1 (t )
TG1 TG1 R1 TG1
 1 1
Pm1 (t )  PV 1 (t )  Pm1 (t )
TT 1 TT 1
 1 D 1 1
1 (t )  Pm1 (t )  1 1 (t )  Ptie (t )  PNL1 (t ) Form of equations is
M1 M1 M1 M1
the same, except for
 1 1 1 sign of ΔPtie term in 3rd
PV 2 (t )   PV 2 (t )  2 (t )  Pref 2 (t )
TG 2 TG 2 R2 TG 2 equation of each set.
 1 1
Pm 2 (t )  PV 2 (t )  Pm 2 (t )
TT 2 TT 2
 1 D 1 1
2 (t )  Pm 2 (t )  2 2 (t )  Ptie (t )  PNL 2 (t )
M2 M2 M2 M2

Ptie (t )  T1 (t )  T2 (t )

5
Will this work?
For a load change in area 1, we desire:
∆ Pm1∞=∆PNL1 Each BA compensates for
∆Pm2∞=0 its own load change.
∆ω∞=0
∆Ptie∞=0

Steady-state values

The system of the previous two slides consists of just


primary control, and as we have seen will distribute the
generation imbalance to all units, leaving a non-zero
steady-state frequency error. Thus, ΔPm2∞≠0; Δω∞≠0,
and ΔPtie∞≠0. We need an additional control loop. 6
Introduce Area Control Error
ACE1=-B1∆ω-∆Ptie,
ACE2=-B2∆ω+∆Ptie
PNL1

The additional loop is an


integral control loop, which
provides the ability to zero the
steady-state error of the
system output (frequency) in PNL2
response to a unit step
disturbance.

See http://home.eng.iastate.edu/~jdm/ee553/AGC1.pdf
7
State equations for this system
 1 1 1
PV 1 (t )   PV 1 (t )  1 (t )  Pref 1 (t )
TG1 TG1 R1 TG1
 1 1
Pm1 (t )  PV 1 (t )  Pm1 (t )
TT 1 TT 1
 1 D 1 1
1 (t )  Pm1 (t )  1 1 (t )  Ptie (t )  PL1 (t )
M1 M1 M1 M1

Pref 1 (t )   KB11 (t )  KPtie
 1 1 1
PV 2 (t )   PV 2 (t )  2 (t )  Pref 2 (t )
TG 2 TG 2 R2 TG 2
 1 1
Pm 2 (t )  PV 2 (t )  Pm 2 (t )
TT 2 TT 2
 1 D 1 1
2 (t )  Pm 2 (t )  2 2 (t )  Ptie (t )  PL 2 (t )
M2 M2 M2 M2

Pref 2 (t )   KB2 2 (t )  KPtie

Ptie (t )  T1 (t )  T2 (t )

8
AGC and participation factors
ACE, being a measure of
how much the total system
generation needs to
change, is allocated to the
various units that comprise
the balancing area via
participation factors.
The participation factors
are obtained by linearizing
the economic (market)
dispatch about the last
base point solution (see
Wood & Wollenberg,
section 3.8).

Base point calculation is performed


by the real-time market every 5 mins.

See http://home.eng.iastate.edu/~jdm/ee553/AGC1.pdf
9
Summary of power balance control levels
No. Control Name Time frame Control objectives Function
Power balance and
Transient frequency
1 Inertial response 0-5 secs transient frequency dip
control
minimization

Primary control,
Power balance and
Transient frequency We are
2 1-20 secs transient frequency
governor
recovery
control addressing the
system’s ability
Secondary 4 secs to 3 Power balance and steady- to control
3 Regulation
control, AGC mins state frequency
steady-state
Power balance and Load following and frequency.
4 Real-time market Every 5 mins
economic-dispatch reserve provision Why consider
the real-time
Day-ahead Power balance and Unit commitment and market?
5 Every day
market economic-unit commitment reserve provision

The real-time market has a secondary influence on the system’s ability to


control steady-state frequency because it computes base points based on a net
load forecast. The accuracy of this forecast determines how much the units will
be moved by AGC and as a result, how much frequency variability is present.
10
So let’s take a look at how the real time market uses a net load forecast.
Base point calculation via real-time market
ADS: automatic dispatch system
Focus on interval 2, { t+5, t+10}. DOT: dispatch operating target

For interval 2, a short-term net


load forecast is made 7.5 min
before interval 2 begins, at t-2.5,
and generation set points are
computed accordingly.

At t+2.5, which is 2.5 minutes


before interval 2 begins, the units
start to move.

The units are ramped at a rate


which provides that they reach the
desired base point at t+7.5 min, Key point: The base point is
which is 2.5 min after the interval computed from a net load
begins. forecast. There is error in this
forecast, which typically
increases as wind penetration
Source: Y. Makarov, C. Loutan, J. Ma, and P. de Mello, “Operational increases. This error contributes
impacts of wind generation on California power systems,” IEEE Trans on
Power Systems, Vol. 24, No. 2, May 2009. to frequency deviation. 11
Wind farm participation in AGC

PNL1
Most windfarms do not
participate in AGC today.

However, windfarms do
affect the net load seen by PNL2
AGC, as indicated here.

12
Control performance standards
Control Performance Standards CPS1 and CPS2 evolved from earlier metrics and
were enacted by NERC in 1997 to evaluate a balancing area’s frequency control
performance in normal interconnected power system operations.

The motivation underlying CPS is to ensure a targeted long term frequency control
performance of the entire interconnection.

CPS measures each balancing area’s frequency control performance in achieving


control objectives.

N.Jaleeli and ,L.VanSlyck, “Control performance standards and procedures for interconnected operation,” Electric Power
Research Institute, Dublin, Ohio, Tech.Rep. TR-107813, Apr.1997.
N.Jaleeli and L.S.Vanslyk, “NERC’s new control performance standards. IEEE Trans. Power Syst.,” vol.14, pp.1092-1099,
Aug.1999.
13
Control performance standards
CPS1 CPS2

NERC Standard BAL-001-0.1a — “Real power balancing control performance,” http://www.nerc.com/files/BAL-001-0_1a.pdf.


14
CPS1:a measure of a balancing area’s long
term (12 mo) frequency performance.
• Control objective - bound excursions of 1-minute average
frequency error over 12 months in the interconnection.
• Measures control performance by comparing how well a
balancing area’s ACE performs in conjunction with the
frequency error of the interconnection.
ACE  ( Ptie ,a  Ptie , s ) | B | F
Ref: M. Terbrueggen, “Control Performance Standards” 2002
    
Ptie
Average ACE, ΔF over 1 min to compute:
ACE1min
CP1min   F1min
10 | B |
10B to give units of Hz.

ΔF is an interconnection measure. ΔPtie is a balancing area measure.


When ΔF<0, the interconnection needs generation, so desire BA to make
ΔPtie large  ACE>0 (helping). If ACE<0, it means BA is undergenerating  “hurting.”

So we want to see CP negative, large in mag.


15
CPS1:a measure of a balancing area’s long
term (12 mo) frequency performance.
• Control objective - bound excursions of 1-minute average
frequency error over 12 months in the interconnection.
• Measures control performance by comparing how well a
balancing area’s ACE performs in conjunction with the
frequency error of the interconnection.
Ref: M. Terbrueggen, “Control Performance Standards” 2002
ACE  ( Ptie ,a  Ptie , s ) | B | F
Average ACE, ΔF over 1 min to compute:
ACE1min
CP1min   F1min
10 | B |
Average CP 1min (CP1min )12 Month
over 12 mo to CF 
compute: (1 ) 2
ε1 =target bound for 12 month of 1min avg freq error. e.g., 0.018Hz in EI, 0.228Hz
in WECC, 0.020 Hz for ERCOT. Must be squared to normalize Hz2 in numerator.

CPS1  (2  CF ) 100  100%


16
CPS1:a measure of a balancing area’s long
term (12 mo) frequency performance.
• Control objective - bound excursions of 1-minute average
frequency error over 12 months in the interconnection.
• Measures control performance by comparing how well a
balancing area’s ACE performs in conjunction with the
frequency error of the interconnection.
Ref: M. Terbrueggen, “Control Performance Standards” 2002
ACE  ( Ptie ,a  Ptie , s ) | B | F
Average ACE, ΔF over 1 min to compute:
ACE1min
CP1min   F1min
10 | B |
Average CP 1min (CP1min )12 Month
over 12 mo to CF 
compute: (1 ) 2 Problem: balancing area can grossly
over- or under-generate (as long as
it is opposite frequency error) and
get very good CPS1, yet impact its
CPS1  (2  CF ) 100  100% neighbors with excessive flows
(large ACEPtie,a>>Ptie,s).
17
CPS2: measure of a balancing area’s ACE
over all 10-minute periods in a month.
• Control objective – limit ACE variations & bound
unscheduled power flows between balancing areas.
• Developed to address “problem” of previous slide.

Requirement: |ACE10min |< L10  1.6510  10 Bi   10 Bs 


CPS2=100%-(Percent of 10 min periods in violation)>90%

• L10 is max value within which ACE10min must be controlled


• BS=sum of B values for all control areas.
• ε10 =target bound for 12 mo RMS of10-min avg freq error:
e.g., 0.0057Hz in EI, 0.0073 for the WI and ERCOT.
• In 2003, the 10Bs were ~ -5692 mw/0.1hz for EI, -1825
mw/0.1hz for WEEC, -920 mw/0.1Hz for ERCOT. 18
Simulation System
•Two Area System (Area A and Area B)
Wind power is assumed in area A
•Each area consists of 10 conventional units, with inertia and with
speed governing
• Based points are computed from net load forecast made 7.5 min
ahead, with an assumed error of 1% for load and 4.5% for wind.
•Wind penetration levels- 6%, 10%, 13%, 17%, 21%, 25% (Pw/Pnw)
are considered (by capacity).
• Wind is assumed to displace conventional units
• Actual sec-by-sec p.u. value of load and of wind power data from
one wind farm is used.
Con
units Con
Wind
A B units
units
C. Wang and J. McCalley, “Impact of Wind Power on Control Performance Standards,” under review, IEEE Trans on Pwr Sys.
19
2 Area Simulation System
reg
PNL1 (t )

ACE  ( Ptie ,a  Ptie, s ) | B | F

reg
PNL 2 (t )

C. Wang and J. McCalley, “Impact of Wind Power on Control Performance Standards,” under review, IEEE Trans on Pwr Sys.
Inputs for 2 Area Simulation System

Area 1 input Area 2 input


2 (t )  [ PNL 2 (t )]  [ PG 2 , RTED (t )]
reg
1 (t )  [ PNL1 (t )]  [ PG1, RTED (t )]
reg
PNL PNL

  PL1 (t )  Pw1 (t )  PL1, fcst (t )  Pw1, fcst (t )  
  PL 2 (t )  PL 2, fcst (t ) 
  PL1 (t )  PL1, fcst (t )    P
w1 (t )  Pw1, fcst (t )    PL 2 (t )  PL 2, fcst (t ) 
 P (t )  P (t )
reg
L1
reg
w1
 PLreg
2 (t )

The sec-by-sec generation levels in each case


(PG1,RTED and PG2,RTED) are determined by linearly
interpolating between their respective 5 minute
load and wind forecasts.
C. Wang and J. McCalley, “Impact of Wind Power on Control Performance Standards,” under review, IEEE Trans on Pwr Sys.
1
Study results
0.9

0.8 Case A
Case A: Area 1, Area 2
Normalized CPS1 Score

Case B have same size.


0.7
Case B: Area 1
0.6
Normalized CPS1 unchanged. Area 2 load
0.5 and gen scaled up by 10.
0.4

0.3

0.2
0% 6.67% 10.14% 13.71% 17.37% 21.12% 24.94%
Wind Energy Penetration Level in Area 1

0.9
Conclusions: Case A
Normalized CPS2 Score in Area 1

0.8 Case B
1.CPS1 and CPS2 deteriorates with 0.7
increasing wind penetration. Normalized CPS2
0.6
2.The effect is larger for “smaller”
0.5
interconnections.
0.4

0.3

0.2
0% 6.67% 10.14% 13.71% 17.37% 21.12% 24.94%
Wind Energy Penetration Level in Area 1
C. Wang and J. McCalley, “Impact of Wind Power on Control Performance Standards,” under review, IEEE Trans on Pwr Sys. 22
Study results
Measures to improve CPS1, CPS2:
•M1: Increase primary frequency control capability in Area 1
•M2: Increase the forecast accuracy of wind power
•M3: Control wind power output to be no more than a band around forecasted value
•M4: Combining control areas.
CPS1 AND CPS2 SCORE WITH DIFFERENT MEASURES AT 25% WIND POWER
ENERGY PENETRATION LEVEL IN AREA1, CASE A
Measures CPS1 Improvement CPS2 Improvement
over Original over Original
CPS1 CPS2
M1 52.93% 31.74% 59.71% 3.75%
M2 * 65.58% 63.21% 71.22% 23.75%
M2 ** 92.90% 131.21% 100% 75.00%
M3 60.76% 51.22% 66.19% 15.00%
M4 73.84% 83.78% - -
In M2*, NRMSE of wind power forecast is assumed to be 3%; In M2**, NRMSE of wind power forecast is assumed to be 0%.

CPS1 AND CPS2 SCORE WITH DIFFERENT MEASURES AT 25% WIND POWER
ENERGY PENETRATION LEVEL IN AREA 1, CASE B
Measures CPS1 Improvement CPS2 Improvement
over Original over Original
CPS1 CPS2
M1 91.50% 4.61% 96.52% 2.21%
M2 * 93.78% 7.21% 98.61% 4.41%
M2 ** 96.56% 10.40% 100% 5.88%
M3 91.16% 4.22% 97.92% 3.68%
M4 99.07% 13.25% - -
In M2*, NRMSE of wind power forecast is assumed to be 3%; In M2**, NRMSE of wind power forecast is assumed to be 0%.

C. Wang and J. McCalley, “Impact of Wind Power on Control Performance Standards,” under review, IEEE Trans on Pwr Sys. 23
Solutions to variability & uncertainty
1. Do nothing: fossil-plants provide reg & LF (and die ).
2. Improve forecasts (M2)
3. Increase control of the wind generation
a. Control wind to band around forecasted value (M3)
b. Provide wind with primary control
• Reg down (4%/sec), but spills wind following the control
• Reg up, but spills wind continuously
c. Limit wind generation ramp rates
• Limit of increasing ramp is easy to do
• Limit of decreasing ramp is harder, but good forecasting can warn
of impending decrease and plant can begin decreasing in advance
4. Increase non-wind MW ramping capability during periods of
expected high variability using one or more of the below (M1):
a. Conventional generation %/min $/mbtu $/kw LCOE,$/mwhr
b. Load control Coal 1-5 2.27 2450 64
c. Storage Nuclear 1-5 0.70 3820 73
d. Expand control areas NGCC 5-10 5.05 984 80
5. Combine control areas (M4) CT 20 5.05 685 95
4 Diesel
24 40 13.81

You might also like