Power Electronics for Integration of...

Power Electronics for Integration of Renewables

Ali Mehrizi-Sani Assistant Professor LIPE • EECS • WSU

Based on the work of current and former graduate students: Mehrdad Yazdanian, Chris Stone,

Younes Sangsefidi, Saleh Ziaeinejad, and Hooman Ghaffarzadeh

Microgrid Symposium • Niagara-on-the-Lake, ON • October 2016

2 of 25

3 of 25

Dave Bakken, Professor Anjan Bose, Regents Professor Anamika Dubey, Assistant Professor Adam Hahn, Assistant Professor Carl Hauser, Associate Professor Tosh Kakar, Clinical Associate Professor Chen-Ching Liu, Director of ESIC and Boeing Professor Saeed Lotfifard, Assistant Professor Ali Mehrizi-Sani, Assistant Professor Robert Olsen, Professor Noel Schulz, Professor and First Lady (June 2016) Anurag Srivastava, Associate Professor Mani Venkatasubramanian, Professor

WSU Power Engineering Program

4 of 25

Challenge

Controllers are designed for a prespecified configuration and their performance deteriorates when the host system, which is also part of the plant, varies significantly from what was used for the original design.

Load Disconnected Overshoot: 26%

Settling time: 67 ms

Original System Overshoot: 15%


5 of 25

Background

One of the grand energy challenges is to enable integration of large amounts of renewable energy resources at a competitive cost in the power grid (in the US, 80% by 2050 per NREL). What is missing is a flexible system that accommodates the unique characteristics of renewable resources: – Intermittency – Lack of inertia – Susceptibility to violation of operational limits

Our work addresses the latter: – How can we make sure our units are “tightly” controlled and do

not violate their limits even when the host system changes significantly?

6 of 25

Controller Design: Existing Approaches

Existing approaches to ensure dynamics of the system are handled design controllers based on – Analytical formulation and model-based tuning (Astrom’s work) – Optimization (Gole’s work)

Why not just redesign? – Need updated system models – Need a computational infrastructure to allow redesign – Need access to the internal parameters of the controller – New design will again have limited robustness to topology,

operating point, and system parameters

7 of 25

Proposed Solution

Improving the response by temporarily manipulating the set point without changing the original controller. Features: – Robust to topological changes – Independent of the system model – Requires little information about unit

Secondary Controller

Primary Controller Unit

xsetpoint x(t)Secondary Controller Set Point

Modulation


xsetpoint x(t)

8 of 25

Set Point Modulation

Initial Idea – Choose T1 so that the peak of the response equals the reference – Choose T2 to be the time of this peak

Not Implementable – Faster-than-real-time simulator – Closed-form solution – System parameters

T1 tp t

x(t)

0 T1 T2tp

9 of 25

Finite-State Machine

SPAACE /speɪs/: Set Point Automatic Adjustment with Correction Enabled State Numbering:

Salient Features: – Based on local signals – Independent of model – Robust to changes in parameters

x(t) > xmax

S000

S001

S010

S101

S111

w1

xpred > xmax

Δt < Tmax

Δt > Tmax

Δt > Tmax

w2

Δt < Tmax

w4

violationwait

S100

x(t) < xmax

w3

violation wait

S110

xpred > xmaxx(t) > xmax

w5violation

wait

xpred < xmax

xpred < xmaxx(t) < xmaxwait

10 of 25

Case Study I: Set Point Change

System Response DG2 step change from 0.91 pu to 1.09 pu

DG1 and DG3 unchanged (40% overshoot)

IEEE 34-Bus System Added 3 DG units and a load

Operates in grid-connected mode

800 816

≈

824

828 830 854 852

832

858

888 890

834 860 836 840

842

844

846

848

DG3

DG2

DG1

0 1 2 3 4 50.9

1

1.1

(a)

I (p

u)

0 1 2 3 4 50.9

1

1.1

(b)I

(pu)

Time (ms)

11 of 25

Case Study II: Load Disconnection

System Response Resistive 0.5 pu load disconnected

(15% overshoot)

800 816

≈

824

828 830 854 852

832

858

888 890

834 860 836 840

842

844

846

848

DG3

DG2

DG1

0 1 2 3 4 5 6 7 8 9 10

0.9

1

1.1

I (p

u)

Time (ms)

with SPAACE

IEEE 34-Bus System Added 3 DG units and a load

Operates in grid-connected mode

12 of 25

Case Study III: Unbalanced System

646 645 634633632

611 684 675692671

680652

650

DG1

Test load

0 0.2 0.4 0.6 0.8 10.5

1

1.5

V (p

u)

Time (s)

System Response

Resistive 1 pu load switched off Unstable system to stable system

IEEE 13-Bus Unbalanced System Added a DG unit and a test load

Operates in islanded mode

13 of 25

Experimental Implementation

NI cRIO: SPAACE Algorithm

DC Power Supply

NI cDAQ: Collect Data

Measurement and Control Signals

RSCAD

Real-Time Digital Simulator

14 of 25

Case I: Load Energization (1.2 pu)

Time (s)

V (p

u)

0.900.951.001.051.101.151.20

1.78 1.80 1.82 1.84 1.86 1.88 1.90 1.92

Without SPAACE

With SPAACE

V (p

u)

0.951.001.051.101.15

1.20

Time (s)1.78 1.80 1.82 1.84 1.86 1.88 1.90 1.92

Mp: 5% tsettling: 60 ms


G1

250 MW40 MVAr

93 kV

DER

120 MVA0.9322 pf

lagging

Breaker for load change

test

1 2ZL = 0.014 + j0.064 pu ZT = j0.18 pu ZDER = j0.14 pu

15 of 25

Case II: Step Change in iq

Time (s)

i q (p

u)

2.65 2.70 2.75 2.80 2.85 2.90 2.950.40.60.81.01.2

Without SPAACE

With SPAACE

Time (s)

0.40.60.81.01.2

i q (p

u)

2.65 2.70 2.75 2.80 2.85 2.90 2.95



16 of 25

Variant: Smooth SPAACE

SPAACE is not directly applicable to applications such as drive systems because the step changes introduced in the set point may cause torque pulsation, mechanical fatigue, and stress. A “smooth” variant of SPAACE (SSPAACE) is proposed to modify the set point more gracefully than SPAACE; that is, it introduces a smooth change as opposed to a step change.

17 of 25

SSPAACE with a Hybrid Structure

SSPAACE utilizes a supervisory switching scheme based on observing the set point and the predicted error.

Response with SSPAACE

t

x(t) Response

Response with SPAACE

ProcessController

Lead Compensator Supervisor

+-+

++

-

System

y(t)x'sp(t)

y(t)

xsp

18 of 25

Study System

Setup at Graz University of Technology, Austria

19 of 25

500

600

Spee

d (r

pm)

500

600

Spee

d (r

pm)

500

600

Spee

d (r

pm)

-0.1 0 0.1 0.2 0.3 0.4 0.5

500

600

Spee

d (r

pm)

Time (s)

Set Point

Actual

Adj. Set Point (a)

(b)

(c)

(d)

0 0.2 0.40

50010001500

500

600

Spee

d (r

pm)

500

600

Spee

d (r

pm)

500

600

Spee

d (r

pm)

-0.1 0 0.1 0.2 0.3 0.4 0.5

500

600

Spee

d (r

pm)

Time (s)

Set Point

Actual

Adj. Set Point (a)

(b)

(c)

(d)

0 0.2 0.40

50010001500

Step Change in the Speed Set Point

Simulation Results Step change in ωref from 500 to 600 rpm: a) base case, b) prefilter, c) SPAACE, and d) SSPAACE.

Experimental Results

20 of 25

-0.1 0 0.1 0.2 0.3 0.4 0.5450

500

550

600

650

Spee

d (r

pm)

Time (s)

Base casePrefilter

SPAACESSPAACE

Step Change in the Speed Set Point

21 of 25

450

500

550

Spee

d (r

pm)

450

500

550

Spee

d (r

pm)

450

500

550

Spee

d (r

pm)

0 0.2 0.4 0.6 0.8450

500

550

Spee

d (r

pm)

Time (s)

Set Point

Actual

Adj. Set Point

(a)

(b)

(c)

(d)

450

500

550

Spee

d (r

pm)

450

500

550

Spee

d (r

pm)

450

500

550

Spee

d (r

pm)

0 0.2 0.4 0.6 0.8450

500

550

Spee

d (r

pm)

Time (s)

Set Point

Actual

Adj. Set Point

(a)

(b)

(c)

(d)

External Disturbance: Load Change

Simulation Results Step change in isq from 0 to -20 to 0 A: a) base case: 48 rpm, b) prefilter: 48, c) SPAACE: 42, and d) SSPAACE: 12.

Experimental Results

22 of 25

400

500

600

700

Spee

d (r

pm)

-0.1 0 0.1 0.2 0.3 0.4 0.5400

500

600

700

Spee

d (r

pm)

Time (s)

Set PointActualAdj. Set Point

0 0.2 0.40

50010001500

400

500

600

700

Spee

d (r

pm)

0 0.2 0.4 0.6 0.8 1400

500

600

700

Spee

d (r

pm)

Time (s)

Set Point

ActualAdj. Set Point

0 0.5 10

50010001500

Sensitivity to System Parameters

J is changed to (a) one-third and (b) three times the design value. Step change in ωref from 500 to 600 to 500 rpm.

Simulation Results: 1/3x Base case—Mp: 50%; tsettling: 80 ms SSPAACE—Mp: ~0%; tsettling: 50 ms

Simulation Results: 3x Base case—Mp: 40%; tsettling: 0.4 s SSPAACE—Mp: 10%; tsettling: 0.2 s

23 of 25

Conclusions

By designing the trajectory to reduce overshoots, it is possible for a system to operate closer to its limits. Offline (PSCAD and MATLAB) and real-time (RTDS and Opal-RT) simulation studies as well as experimental results show that S/SPAACE is effective in mitigating transients: – Step change: Mitigating overshoots – Load energization: Eliminating peaks – Load disconnection in an unbalanced system: Stabilizing

oscillatory behavior of voltage

The significance of this work is that it can reduce the need for overdesign and subsequently increase asset utilization.

24 of 25 35 of 24




t

x(t) Response


35 of 24




t

x(t) Response


8 of 46

State of the Art

6 of 46

Microgrid Challenges (3/3)

Example– Effect of large load change on controller performance.

Load DisconnectedOvershoot: 26%


Original SystemOvershoot: 15%


11 of 46

Proposed Solution

Improving the response by temporarily manipulating the set point




Modulation


xsetpoint x(t)

18 of 46

Case Study IV: Unbalanced System

646 645 634633632

611 684 675692671

680652

650

DG1

Test load

0 0.2 0.4 0.6 0.8 10.5

1

1.5

V (p

u)

Time (s)

System ResponseResistive 1 pu load switched off

Unstable system to stable system

29 of 46

Upper Bound of m

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.40.5

m

ζ

f is using IVT described before to ensure f(t) > 0.

31 of 46

Effect on Stability

SPAACE does not make a stable system unstable (but may make an unstable system stable).Sketch of proof (finite number of set point changes):

The response, from the final value theorem, is

37 of 46


NI cRIO:SPAACE Algorithm

DC Power Supply



RSCAD


45 of 46

Applications

Large AC SystemsSegmented Power System

Microgrid ∞S

PCC

Main Grid

DERn

PCn

DER3

PC3

DER2

PC2

DER1

PC1

Emerging Small AC SystemsAll-Electric Ships and Military Systems

ms ctric Machines

6 of 46

Microgrid Challenges (3/3)

Example– Effect of large load change on controller performance.

Load DisconnectedOvershoot: 26%


Original SystemOvershoot: 15%


11 of 46

Proposed Solution

Improving the response by temporarily manipulating the set point




Modulation


xsetpoint x(t)

18 of 46

Case Study IV: Unbalanced System

646 645 634633632

611 684 675692671

680652

650

DG1

Test load

0 0.2 0.4 0.6 0.8 10.5

1

1.5

V (p

u)

Time (s)

System ResponseResistive 1 pu load switched off

Unstable system to stable system

29 of 46

Upper Bound of m

0 0.2 0.4 0.6 0.8 10

0.1

0.2

0.3

0.40.5

m

ζ

f is using IVT described before to ensure f(t) > 0.

31 of 46

Effect on Stability

SPAACE does not make a stable system unstable (but may make an unstable system stable).Sketch of proof (finite number of set point changes):

The response, from the final value theorem, is

45 of 46

Applications

Large AC SystemsSegmented Power System

Microgrid ∞S

PCC

Main Grid

DERn

PCn

DER3

PC3

DER2

PC2

DER1

PC1

Emerging Small AC SystemsAll-Electric Ships and Military Systems

ms ctric Machines

37 of 46


NI cRIO:SPAACE Algorithm

DC Power Supply



RSCAD


8 of 46

State of the Art

35 of 24




t

x(t) Response


25 of 25

Thank You

Power Electronics for Integration of

Renewables

Ali Mehrizi-Sani [email protected]

http://eecs.wsu.edu/~mehrizi

Power Electronics for Integration of...

Documents

Transcript of Power Electronics for Integration of...