GRID integration of Wind energy conversion system

grid integration of wind

1. University a

Grid Integration of Wind Energy
Conversion Systems
ECE 566

Instructor: Dr. Sid Suryanarayanan
Associate Professor, ECE
[email protected]
Week 8 (part A)

Center for Research
and Education in Wind)

BE... !

OS. Suryanarayanan, 2014

2. University 777777777

Practical case study

Voltage Deviation Analysis of
a 50 MW Wind Farm

EPT! NC STATE UNIVERSITY

OS. Suryanarayanan, 2014

3. ee Outline of presentation ©

DAS Het

« BPA’ s Condon wind farm scenario

* The proposed remedy

+ Modeling and simulation based analysis
* Identification of the real cause

* Conclusions

OS. Suryanarayanan, 2014

4. Types of wind power installations ”””

« Fixed speed induction generators (FSIG)
— Squirrel cage induction machines
— Robust and inexpensive
— Directly coupled to grid and draws reactive power

— Fluctuations in wind cause mechanical and
electrical fluctuations (V deviation)

— Not suitable for weak electric grids
« Variable speed induction generators (VSIG)
— Wound rotor induction machines
— Use power electronics
— More expensive and less robust

OS. Suryanarayanan, 2014

5. Ce BPA network G

University

Bonneville Power Administration - federally
owned utility with regions of operation in WA, OR,
ID, and parts of MT

~15000 miles of lines and ~300 substations
Connects to Canada in the North and Los
Angeles in the South thru 500 kV AC lines
Pacific DC intertie between BPA (Celilo) and LA
(Sylmar)

In 2011, BPA network had approximately 30% of
wind power on its system (one of the highest
concentrations in the US)

OS. Suryanarayanan, 2014

6. Colorado Condon Wind Farm onBPA @

™ network
+ 50 MW wind farm on 69 kV u
network of BPA in central
Oregon
+ 83 FSIG wind turbines of 600
kW capacity each
* Owned and operated by

private utility (SeaWest) on
federal electric network

« Reactive support included
« Power factor correction capacitors (10 MVAr) at wind farm
and at nearby substation (5.5 MVAr)

* Cable shunt capacitance (2 MVAr)
» Individual turbine caps (90/180 kVAr)

OS. Suryanarayanan, 2014

7. 3)
Colorado Voltage deviation on PCC at Condon ©

University a

» Yet, voltage deviation on PCC was observed periodically

110 Voltage at Wind Farm [%]

105-

oor A | PT | | TN 1

N BPA’ s nominal operating
voltage range

95, BPA’ s nominal operating
It: level
voltageeve Date [MM/DD/YY]
L L

L L L L L
01/01/04 03/01/04 04/30/04 06/29/04 08/28/04 10/27/04 12/26/04
« Persistent voltage deviation of >9% seen regularly
« Note: x-axis is time (1 year period)

90

OS. Suryanarayanan, 2014

8. e O)
Voltage deviation on PCC at Condon ©

University ENGINEERING

.08 - Voltage at the
.07| Wind farm [pu] 5min SCADA 08/22/2004 - 08/24/2004 |
.06} 2
05)
04}
03)
021
-01 +) Av= 9%
1 || caused
0.99 only in

2222220.

some

0.98 | cases 7]

0.97| al

0.96 À

0.957 2s SCADA 03/05/2006 = 4
-10 Le) 10 20 30 40 50

Real power delivered by wind farm [MW]

Site established as candidate for testing and installing new
technology for mitigating V deviation

OS. Suryanarayanan, 2014

9. Colorado Remedy for voltage deviation on Condon ©

omens wind farm
» Flexible AC Transmission System (FACTS) technology
proposed

+ Static synchronous compensator (STATCOM)
— Shunt device
— Fast acting, reactive power compensator
— Dynamic voltage control

« New technology in power electronics (ETO) based
STATCOM at 10 MVA power rating to be installed at
Condon
— developed by NCSU

+» FSU-CAPS in collaboration with EPRI, BPA, NCSU to test
and characterize novel controller for ETO based STATCOM
— High-fidelity modeling, simulation and hardware-in-the-loop studies
— Also to identify actual source of V deviation

OS. Suryanarayanan, 2014

10. e

University

Hardware-in-the-loop

Real Time Digital Simulator

;

<————~|__A/D

System Data in Simulatio

D/A I

Hardware response

Controller X

AW j
DC Load i

OS. Suryanarayanan, 2014

External Hardware
Universal j
controller H

B Protection relay .

AC/AC power converter i
i (Motor Drive) H

« Requirements
— Detailed models
— Real-time capability
— D/A & A/D conv. w/
amplification

11. Re Steps in modeling effort

BPA system modeling and validation

Individual wind turbine system modeling
and control

Wind farm modeling and control
Validation against SCADA data

Isolate cause of V deviation in wind farm
using high-fidelity models

OS. Suryanarayanan, 2014

12. Klondike wind farm
modeled as
| P and Q injections

+ Dynamic PQ loads

+ T-Lines

+ Xfmrs (w/o tap
changers) Condon wind
F farm modeled
+ C-bank at Fossil as P andQ
+ Breakers injections
(i.e. from

+ HV equivalents: SCADA data)
constant V-source

behind impedance

OS. Suryanarayanan, 2014

13. ©

e _BPA system model validation

« Comparative studies of fault analysis on system model
developed with data from BPA

1 phase-ground fault 3 phase faul

Aspen RTDS % difference | Aspen % difference

(Amps) | (Amps) (Amps)

30099 29173 27308
53205 52003 46672
1117 1115 2716

1942 1961 i 2097

905 898 E 1773

4426 4343 i 7114

OS. Suryanarayanan, 2014

14. Colorado BPA system model validation

University

1.05-—

RE,

Simulated (with capacitors)

Voltage at Condon Wind Farm Substation

ad

N Wu wi, FA À ri

30 40 5 60 70 8 90

-Reactive Power (MVAR)

j |
POND, A
re at et

i

30 40. 50 60 70 80 90
Time (min)

« Condon wind farm as only P and Q injections
Manual switching of capacitors for |V| < 0.95 pu

OS. Suryanarayanan, 2014

15. Coe Modeling characteristics of individual wind E
” turbine

« 600 kW fixed-speed wind turbine

« Modeled with available name plate details

« Induction machine mechanically coupled to a wind turbine
rotor model through a drive train model

* Controls include power electronic soft-starter, pitching, pole
switching, capacitor switching, and grid connection

e eran
Sor Sar e.”

35 pa ame). Let none fe

38 Mos | | Ron a
+ Le

CapactorSark

OS. Suryanarayanan, 2014

16. ee High-fidelity wind farm modeling E)

University 77777777

« Two modeling methods adopted
— Load flow tool based steady state analysis

— Hybrid model with individual strings of turbines for dynamic
analysis

» Developed load flow tool in MATLAB
— All 83 IG wind turbines (input is either torque or wind speed)
— Includes local transformers, cables, and PF correction capacitors
— Models BPA system as lumped impedance

+ Studied voltage magnitude and angle deviations across
the wind farm (to help identifying V-control problem)

« Problems with convergence of the load flow brought to
light suspected voltage collapse problems at the Condon
Loop portion of the BPA system

OS. Suryanarayanan, 2014

17. Colorado Salient characteristics of wind farm ©
University control ENGINEERING

« Turbine goes online in 6 pole mode when wind speed > 3.5 m/s

+ When induction m/c reaches 1100 rpm, soft starter engages
and connects m/c smoothly to grid

« At same time, one local cap bank (90 kVAr) is switched in

« If wind speed > 8m/s for at least 10 minutes, m/c switched to 4
pole operation

— Blades switch out (to remove torque)

— m/c disconnects from grid

— Turbine speeds up by pitching blades back in
— Re-energizing with the gird using soft starter
— 2 local cap banks (180 kVAr) switched in

OS. Suryanarayanan, 2014

18. codo Salient characteristics of wind farm ©
iniversit control ENGINEERING

* Cap banks at substation switched in when |V|
<1 pu

« Cap banks at substation disconnected when
[V| > 1.3 pu

* Caps at Fossil switch in when [VE] < 0.96 pu

« Caps at Fossil disconnected when |VFS| >
1.04 pu

OS. Suryanarayanan, 2014

19. e

V-fluctuations within the wind ©)

University farm ENGINEERING
Wind Farm from WindTurbineTest081406, oCaps No V-profile _
0.9 T T 4 problem within
H i wind farm
ost. — ee
Near end of strings
0.7}.

0.6}

0.5+-+

0.4+-+

0.3;

0.2--

Voltage difference to SS (bus3) [%]

0.1

0!

4:

SS...Substation

i i i i
0.4 05 0.6 0.7 0.8 0.9 1
Torque [pu]

OS. Suryanarayanan, 2014

20. >
ee Comparison between Load Flow Results and ©
University SCADA Data ENGINEERING

Wind Farm P-Q Relationship

Strong
H H evidence that
as = ./A malfunctioning
= = of Condon
> Wind Farm

capacitor
switching
causes the
problem

Leads into
V-Collapse

Voltage [pu]

—— 2 s SCADA Data
0.91-| ——5 min SCADA Data for 3 days starting 22-Aug-2004
—— Only Fossil Capacitors
— With WF Capacitors
—— No Capacitors

T T I I
2 0 0 10 20 30 40 50
Real Power Condon Wind [MW]

MF... Wind Farm (at Substation)
OS. Suryanarayanan, 2014

21. ade 40-turbine model of Condon wind farm and BPA

University

Condon Loop One Line

40 turbines fully modeled,
scaled by 2.075 to account for
the 83 units actually in the field

Nunts !
connected !
through 1
underground y
cabling ı

Tute 1
Model

Turbine!
Model

frame],

Model

OS. Suryanarayanan, 2014

©

22. Colado Validation of hybrid RTDS model ©

University . ENGINEERING
against SCADA Data

1.06

1.04
Z 1.02
El : S More evidence
E y |_| _f — 1.8 that
5 Power ramped by (slowly) increasing \ malhunetioning
° Le iT fl 1
E 0.98 wind speed uniformly Wind Farm
9 capacitor
© 0.87 switching
E causes the
5 0.94! ] T problem
= 2 s SCADA Data

0.92 - -5 min SCADA Data for 3 days starting 22-Aug-2004

—— No Capacitors (RTDS)
0.91-| ——Fossil Capacitors (RTDS)
—— CWF Capacitors (RTDS)
0.88 T T r y 1 1 i t
0 5 10 15 20 25 30 35 40 45 50

Real Power Condon Wind [MW]

OS. Suryanarayanan, 2014

23. BPA Recent Data ©)
tony Proper cap bank switching “

Condon Wind Farm Voltage

Ka
z |
58
ARS
¿a
11 HS | — AAN
120
i
53)
4
; 0
al
Nu
call | |
on
00

7200

2
3

9000

3

3
El
5

When cap switching control at Condon Wind works properly, the voltage stays within BPA
criteria as FSU studies predicted

- As confirmed by BPA via email on 11/22/2006

24. E)

ColoradoContingency scenario in BPA system
om Maupin line down

Wind Farm P-V Relationship With Maupin Connection Open (RTDS)
T

T T T T T

14}

8

o
&

Voltage at CW Substation (pu)

o
©
T

BPA C-banks

I | | i
30 35 40 45 50
os. suis NW) 2014

25. ee Contingency performance of a5 MVA (E)

University

STATCOM ire

Wind Farm P-V Relationship With Maupin Connection Open (RTDS)

T T T T y T
1.1

2 105

=

S

E 1

©

3

on

3 095L O MVAR Caps

a E 3 MVAR Caps | Generic STATCOM

E 6 MVAR Caps model

5 E 10 MVARC:

> = 12.75 V-control loop not
. 15.5 MVAR Caps i 4 optimized

0.85 -| —— Manually Switch Capacitors |---trying to maintain 1.05 pu | Observed steady-state
— Manually Switch Capacitors|---ttying to maintain 1.00 pu
— with 5 MVAR STATCOM | |---With 1.0 pu set point | performance only
08 > — Caps switched manuall
15 20 25 30 35 40 A CAP y

to keep STATCOM VAr

P: (ww)
ower supply close to zero

OS. Suryanarayanan, 2014

26. corsets Conclusions

High-fidelity modeling and simulation of utility
system (BPA’ s), existing wind farm, and generic
STATCOM performed

Actual source of voltage deviation in Condon wind
farm identified as controller of PFC

Possible use of STATCOM during contingency
situations established

Sizing of novel STATCOM aided by high-fidelity
modeling effort

OS. Suryanarayanan, 2014