Hybrid Time Domain Simulation:

Application to Fault Induced Delayed

Voltage Recovery (FIDVR)

Vijay Vittal
Arizona State University

PSERC Webinar
Jan. 20, 2015
 Presentation background

• The work presented in this webinar is a

part of the work being done in PSERC
project S-58 at ASU
• The PIs on this project are V. Vittal and S.
Meliopoulos
• The graduate student at ASU is Qiuhua
Huang and the collaborator is Dr. John
Undrill

2
 Outline

1. Introduction
2. Part -1: OpenHybridSim: A new hybrid
simulation platform
3. Part -2: Application to FIDVR study
4. Conclusions

3
 Introduction: Background

 An increasing demand for simulating a portion of a

system in detail with a very small time step and capturing
point on wave detail while preserving the effects on the
rest of the system
• Power electronic converter based renewable energy integration
• HV AC/DC systems
• Residential air conditioners, variable frequency drive(VFD) and
other power electronics based loads
• Representing other needed detail at the distribution level

PV Farm VSC-HVDC air conditioner

4
 Introduction: Hybrid simulation
Involves electromagnetic transient (EMT)- electro-mechanical
transient stability(TS) hybrid simulation
Remainder of the system
Region of interest
• Sequence,
• Instantaneous
phasor model
variable model
• TS simulator
• EMT simulator Boundary

Boundary bus i Vi0abc Boundary bus i

Composite Iiabc Ii120
Load Zi + -
Detailed Model
Thévenin External
……

System
Zik

modeled in equivalent Current network

3-phase Vk0abc injections Ik120
PV Zk + -
Converter Vkabc
Boundary bus k
Boundary bus k

Detailed system modeled in an EMT Simulator External system modeled

in a TS Simulator
5
 Introduction: Hybrid simulation

 Key requirements for a successful hybrid simulation

platform
• Architecture : embedded/decoupled
• Interaction: communication + protocol
• Equivalent models of both detailed and external system
• Preparation and initialization of both detailed and external
system
 Synopsis of literature review of EMT-TS hybrid
simulation
• Lab research or proof-of-concept type
• The architecture design is not flexible and the simulators are
limited to run on one computer
• Targeted mainly for three-phase balanced applications
• No publically available tool for hybrid simulation

6
Time scale of transient phenomena

7
 OpenHybridSim:
A new hybrid simulation tool
 Overall design
• Loosely decoupled architecture
• Socket communication
• InterPSS, an open source power system simulator
• A generic interface to an EMT simulator, e.g.,
PSCAD, Maltab/SimPowerSystems

Socket InterPSS Core Engine

EMT
Simulator Communication
Network
I120 Equivalent Helper
EMT (t )
Socket
V abc Socket HybridSim
Component T (t )
Server Manager
8
Socket based communication framework

 TCP/IP socket based communication framework

• Enables decoupling of the EMT and TS simulators
• Supports application environment
- Single computer
- Local area network(LAN)
- Internet
• Socket components in PSCAD and Matlab/SimPowerSystems
are developed for interfacing

Socket component in PSCAD

9
 External system equivalent
for EMT simulation
 Supports both 1-phase and 3-phase equivalents
• 1-phase: mainly for three-phase balanced applications
• 3-phase: any type of fault within the detailed system
• Facilitates simulation of unsymmetrical faults
 3-phase Thévenin equivalent of external system based
on a 3-sequence network model
• The given base case is modeled using 3-sequences
• Details are provided on the next slide

10
 Three-phase Thévenin equivalents
of the external system
• Step 1: Calculate 3-sequence
Norton equivalents
zia +_ VTia
ziab
• Step 2: 3-sequence to 3-phase bus i ziac
transformation for each boundary bus zibc
zib +
_ VTib
120 −1
I Niabc = SI 120
Ni
yi
abc
= Syi S
zic + VTic

……
_
Detailed

……
where S is the transformation matrix, zikabc
is the bus primitive admittance system
zka
zkab
+
_ VTka
• Step 3: 1-phase Norton to zkac
Thévenin for each boundary bus bus k zkbc
zkb +_
VTkb
VTip = I Nip / yip zip = 1 / yip zkc +_
VTkc
where p stands for one of the three phases
Single-phase line Three-phase line

11
Detailed system represented by sequence
current sources in TS simulation
 Need to extract three-sequence, fundamental frequency
currents from instantaneous, waveform values
 Use well-established FFT algorithm
• PSCAD provides an FFT component
 Directly used in the network solution (I=YV) in TS
simulation

FFT Component in PSCAD

12
Three-sequence based TS simulation for
simulating unsymmetrical faults

Positive-sequence network
(1) solution and integration step (1)
I V TS ( t +∆T )
EMT ( t )
x (t ) = f ( x(t ), y ) (1)
(t )
x(t + ∆T ) = x(t ) + x (t ) ∆T
=0 g1 ( x(t + ∆T ), y((1)t + ∆T ) , I EMT
(1)
(t ) )

(2) Negative- and zero-

I EMT ( t ) sequence network solver VTS(2)(t +∆T )
0 = g 2 ( y((2)
t + ∆T ) , I (2)
EMT ( t ) )
I (0) VTS(0)(t +∆T )
EMT ( t )
0 = g 0 ( y((0)
t + ∆T ) , I (0)
EMT ( t ) )

13
 The interaction protocols
between two simulators
2
TS
(a) series 3
5
…
1
4
.. .. ..
…
. . . EMT
T T+ΔT T+2ΔT T+3ΔT

TS
…
(b) parallel
…
... ... ...
EMT
(c) combined T T+ΔT T+2ΔT T+3ΔT

Parallel Protocol Series Protocol Parallel Protocol

TS
… …
… …
... ... ... ... ... ...
EMT
T T+ΔT T+2ΔT T+3ΔT T+4ΔT T+5ΔT T+6ΔT

Fault on Fault cleared

14
Implementation with the series type protocol
 The series protocol
• Use the updated equivalent data for simulation with each simulator
• Performance issue: One simulator has to remain idle when the other
is running the simulation

I (1)
Positive-sequence network
solution and integration step (1)
VTS ( t +∆T )
InterPSS
x(t ) EMT ( t )
x (t ) = f ( x(t ), y((1)t ) )
x(t + ∆T ) = x(t ) + x (t ) ∆T x(t + ∆T )
V 120 Calculate 3-phase
TS (t + ∆T )
socket =0 g1 ( x(t + ∆T ), y((1)t + ∆T ) , I EMT
(1)
(t ) )
socket
server 120
Thévenin server
Negative- and zero- I EMT (t ) equivalent V abc
I (2)
VTS(2)(t +∆T ) T (t + ∆T ) VTS120(t +∆T )
VTS120(t ) EMT ( t ) sequence network solver
0 = g2 ( y (2)
( t + ∆T ) ,I (2)
EMT ( t ) )
(0)
I EMT VTS(0)(t +∆T ) Step (3)
120
(t )
0 = g0 ( y (0)
( t + ∆T ) ,I (0)
)
I EMT ( t )
EMT ( t )
V abc
Step (1) Step (2) Three-sequence TS simulation T (t + ∆T )

Step (4)
120 120
I EMT ( t ) I EMT ( t +∆T )
EMT EMT EMT Step (5) EMT EMT
socket
client step step step …………………………………….. step step
1 2 3 EMT N-1 N
t t + ∆T

15
Implementation with the parallel type protocol

 The parallel interaction protocol

• Steps (d) and (e) run simultaneously
• May cause significant interfacing errors under fast transient
conditions, as the Thévenin equivalent data is one-step delayed
x(t + ∆T )
x(t ) VTS120(t ) Calculate
Three-
socket three-phase socket
server 120 Thévenin sequence TS
VTS120(t ) I EMT ( t ) equivalent
server
simulation
VTS120(t +∆T )
120 Step (b) Step (c) Step (d)
I EMT (t ) Step (a)
V abc
T (t )
120
120 Step (e) I EMT ( t +∆T )
I EMT ( t ) EMT EMT EMT EMT
socket
client
step step ... step step
1 2 N-1 N

t t + ∆T

16
 Automatic protocol switching algorithm
 Automatically switch interaction protocol based on the
system conditions, reflected by the maximum rate of
change of sequence current injections
( p) ( p)
I −I
120 EMT ( i , t ) EMT ( i , t − ∆T )
RI max( max ( )) / ∆T
EMT ( t ) (1)
i p ∈ (1, 2, 0) I
EMT ( i , t − ∆T )

120
1
I EMT ( t ) Maximum Y series
rate of
change
120
RI EMT >ε? N
I 120
EMT (t −∆T ) 120 Y 0
RI EMT N last step Delay
used series? parallel
time step
∆T
Logic of the protocol switching algorithm
17
 Automatic creation and initialization
of the external network
Boundary
bus i
Base case Ii120 bus m

Depth first The remainder

search(DFS) Detailed of the external

……
System system

to identify bus n
Boundary detailed Vk120
Configuration Ik120
system bus k
Boundary

bus i
bus m
1) Set buses and
Ii120
The remainder
branches within the
of the external detailed system out of Power flow
……

system
bus k service
bus n
2) initialize the external
Ik120 Vk120
system

18
 Testing of the developed platform

 Test system
Bus 8
230 kV
Bus2 Bus 7
18 kV 230 kV
0.0085+j0.072
100+j35MVA Bus 9 Bus 3
230 kV 13.8 kV (case 1)
0.0119+j0.1008
G2 G3
0.032+j0.161

0.039+j0.17

192MVA T2 T3 128MVA
XT=0.0625 XT=0.0586
Bus 5 Bus 6
230 kV 230 kV
0.017+j0.092
0.01+j0.085

125+j50MVA
90+j30MVA

Bus 4
230 kV
T1
XT=0.0576
Bus 3
16.5 kV
G1

247.5MVA

The IEEE 9 Bus system Modeling of Bus 5 and the external network
Thévenin equivalents in PSCAD

19
 Case 1: The interaction protocols
 Case 1 setting: 12
PSCAD
- EMT time step = 50 us 10 EMT-TS(series)
EMT-TS(parallel)
-TS time step = 5 ms
EMT-TS(combined)
8

current (kA)
 Protocol control setting:
6
-Threshold ε = 0.004
- Delay = 2 cycles 4

 A three-phase fault at 0.1 s 2

and lasts for 4 cycles
0
 Plots 0.05 0.1 0.15
(a)
0.2 0.25
2
(Top) Positive-sequence 1
0
current injection at bus 5
into the external network -5
120
log(RIEMT(t))
(Bottom) Protocol switching -10
log(0.004)
signal protocol signal
0.05 0.1 0.15 0.2 0.25
Time (s)
(b)
20
 Case 2: Performance of the developed
platform under unsymmetrical faults
 Internal network: Bus 5 and branch Bus5-Bus7
 Two-port three-phase Thévenin equivalent
 Detailed modeling of bus 5 using the WECC composite load model,
with the 1-p air conditioner compressor motor represented by a EMTP
type model developed by Yuan Liu, et al [1]
Bus 8
230 kV Bus 5
Bus2
18 kV
Bus 7
230 kV
100+j35MVA Bus 9 Bus 3
230 kV 13.8 kV
• Single line to 230 kV

230 kV/69kV
G2
0.0085+j0.072 0.0119+j0.1008
G3
ground fault 150 MVA
10%
• starting at 2.0 s
0.032+j0.161

69kV
0.039+j0.17

192MVA T2 T3 128MVA
XT=0.0625 XT=0.0586
• 4 cycles 69/12.47 kV
Bus 5 Bus 6 50 MVA 69/12.47 kV
230 kV 50 MVA
230 kV 8%
0.017+j0.092
0.01+j0.085

8%
125+j50MVA
90+j30MVA 12.47 kV 12.47 kV 12.47 kV

Bus 4 Equivalent
230 kV feeder
T1
XT=0.0576
Bus 3
Composite Composite Composite
16.5 kV
Load Load Load
G1
Model Model Model
247.5MVA
Detailed modeling of Bus 5
[1] Y. Liu, V. Vittal, J. Undrill, and J. H. Eto, "Transient Model of Air-Conditioner Compressor Single Phase Induction Motor," IEEE
Transactions on Power Systems, vol. 28, pp. 4528-4536, 2013.
21
 Case 2: Performance of the developed
platform under unsymmetrical faults
PSCAD EMT-TS(parallel) EMT-TS(combined) protocol signal

200 2
PSCAD
100
Voltage (kV)

EMT-TS(parallel)
1.5
0 EMT-TS(combined)
protocol signal
-100
1
-200
Phase A
-300
1.95 2 2.05 2.1 2.15 2.2 0.5

200

Current (kA)
0
100
Voltage (kV)

0
-0.5
-100 0.7

-200 -1
Phase B 0.6
-300
1.95 2 2.05 2.1 2.15 2.2
-1.5 0.5
200
0.4
100 -2
Voltage (kV)

0 2.014 2.016 2.018

-2.5
1.95 2 2.05 2.1 2.15 2.2
-100
Time (s)
-200
Phase C
-300
1.95 2 2.05 2.1 2.15 2.2
Time (s)
The phase A current injection at
Three-phase voltage of Bus 5
Bus 5 into the external network

22
 Takeaways of part-1

1. The combination of the decoupled architecture and

socket-based communication facilitates both simulators
to be run on either one computer or several computers
to achieve more flexibility and a better performance
2. The proposed combined interaction protocol with the
auto-switching feature helps improve the hybrid
simulation efficiency while a good accuracy is
guaranteed
3. Application of the multi-port three-phase Thévenin
equivalent enables simulating unbalanced faults within
the detailed system, without the constraint of phase
balance at the boundary buses

23
Fault induced delayed voltage recovery (FIDVR)

 Fault
• distribution, sub-
transmission and
transmission
systems
 Delayed voltage
recovery
• several to tens of
seconds
 Root cause of
FIDVR
Voltage profile during a typical FIDVR event [2]
• air conditioner
(A/C) compressor
motor stalling and [2] D. N. Kosterev, A. Meklin, J. Undrill, B. Lesieutre, W. Price, D. Chassin, et al., "Load modeling
in power system studies: WECC progress update," in 2008 IEEE Power and Energy Society General
Meeting - Conversion and Delivery of Electrical Energy in the 21st Century, 2008, pp. 1-8.
prolonged tripping 24
Fault induced delayed voltage recovery (FIDVR)
 Accurate FIDVR studies require
• Detailed A/C modeling
• Detailed network model down to distribution feeder level
 Limitations of the conventional positive-sequence TS
simulator
• Distribution system configurations
• Response of single-phase devices when subjected to
unsymmetrical faults
 FIDVR events are generally localized
• Detailed modeling can be limited to a small portion of a large
power system

FIDVR A large
Area power system

25
 Determination of the boundary
of detailed system
 Use a simple yet generic test case to quantify the voltage dip
threshold at a transmission bus causing A/C motor stalling
 Criterion: A bus is included in the internal system if a single-
phase or three-phase fault at that bus causes a phase-to-neutral
voltage at buses with a large percentage of A/C motor loads to
drop below 0.75 pu.
x = 0.025 x = 0.05 x = 0.1 x = 0.15 x = 0.2
T T T T T

0.5 0.5

Voltage dip magnitude (pu)

Voltage dip magnitude (pu)

0.45 0.45

Equivalent Step-down Equivalent 0.4 0.4

source transformer Bus 2 feeder
115 kV impedance Bus 1 impedance Bus 3
0.35 0.35
1.04 pu 0.005 + j 0.05 X=
T 2.5 − 20% 0.03 + j 0.04
Composite
s Load model
0.3 0.3
115/12.47 kV
0.25 0.25
0.6 0.7 0.8 0.9 1 0.6 0.7 0.8 0.9 1
A/C load percentage A/C load percentage
(a) (b)

Voltage dip magnitude w.r.t. the A/C load

percentage and the transformer impedance: (a)
A/C power = 4.9 kW (b) A/C power = 6.0 kW
26
 Applied the boundary selection criterion
to the WECC system
The WECC system Buses with a large percentage
Transmission of 1-Ф A/C motor load
Buses lines Generators Loads • Bus 24151
15750 13715 3074 7787 • Bus 24138

NORTH STATISTICS OF THE DETAILED SYSTEM MODEL

24156 26105 24097 24042 14005 15061

Total number
238
24086
15021
of buses
sub 15090
sub
500 kV 7
24092
24236
Number of 230 kV 37
24801 EAST
24138 buses of 161 kV 3
sub
28040 different 115 kV 68
15093
sub
24151
sub
voltage levels 92 kV 18
115 kV
500 kV <= 66kV 105
M M M
Load Load sub
sub Sub-system below 500 kV Total Load 11.9 GW
One-line diagram of the study region
27
 Initialization of the detailed system with a
large percentage of load as induction motors
 The built-in initialization function of PSCAD fails to
initialize the detailed system
 A two-step initialization approach
Bus HV
(1) The switch K is turned to the position 0, LV
s
1 0
and the distribution systems and the CLMs
K Static
are energized by the fixed voltage sources equivalent
load
and initialized independently.
(2) After the CLMs are successfully Distribution network
initialized, the switch K is turned to the
position 1 such that distribution systems are
CLM CLM CLM
connected to the transmission system.

NOTE: The magnitude and phase angle of CLM: composite

the fixed voltage source, S, are set based on load model
the power flow solution of the given base
case
28
 Benchmarking hybrid simulation against
conventional transient stability
 Both use constant impedance load 1.05
Bus 24801 positive sequence voltage

 A single line to ground (SLG) fault

Voltage magnitude(pu)
1
is applied on the phase A of bus EMT-TS
InterPSS
24151 at t = 0.2 s 0.95

0.9
Bus 24151 positive sequence voltage
1.1
0.85
1.05 0 0.1 0.2 0.3 0.4 0.5
Voltage magnitude(pu)

Time (s)

1 EMT-TS Bus14931 machine #1 reactive power

Machine reactive power (pu)

0.25
0.95 InterPSS
0.23
0.9 EMT-TS
0.21 InterPSS
0.85

0.8 0.19
0.75
0.17
0.7
0 0.1 0.2 0.3 0.4 0.5 0.15
Time (s) 0 0.1 0.2 0.3 0.4 0.5
Time(s)

29
 FIDVR event triggered by a SLG fault
 Detailed modeling of the region served by Bus 24151
Bus 24151
500 kV
415 +j*75 MVA 415 +j*75 MVA 403 +j*96 MVA 403 +j*96 MVA

0.2536 pu 0.2536 pu 0.2536 pu 0.2536/0.40/0.512 pu

560 MVA 560 MVA 560 MVA 560 MVA
531.615/120 kV 531.615/120 kV 531.615/120 kV

Bus 24160 115 kV Bus 24229 115 kV

22 + j*1.1
46.8 MVAr 0.12 pu
MVA
250 MVA
Bus 25423 115 /12.47 kV
115 kV 0.12 pu
250 MVA
115 /12.47 kV

M M M
f-2 f-3 f-4 f-5 f-6 f-7 f-8
Equivalent Feeder f-1
Equivalent Equivalent Equivalent Equivalent Equivalent Equivalent Equivalent Equivalent
feeder feeder feeder feeder feeder feeder feeder feeder
Model Model Model Model Model Model Model Model

202+j*(-8) MVA 202+j*(-8) MVA 202+j*(-8) MVA 202+j*(-8) MVA 202+j*(-8) MVA 202+j*(-8) MVA 202+j*(-8) MVA 202+j*(-8) MVA

 Feeder 1
Z
A 3
Z
4 feeder 4 feeder
modeled in B
C

two sections M M

A/C
30
 FIDVR event triggered by a SLG fault (cont’)
Terminal voltage of the A/C motors at the 1/4 length point

• 70% 1-Ф A/C motor, 1

phase A

15% 3-Ф induction 0.5 phase B

phase C
0
motor, 15% constant 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Speeds of the A/C motors at the 1/4 length point
1.4 1.5

impedance 1
phase A
0.5 phase B
• Phase A and C were 0 phase C

directly affected by the 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Speeds of the A/C motors at the end of the feeder
1.4 1.5

SLG fault at Bus 24151 1

phase A
0.5 phase B
• A/C motor stalling 0 phase C

propagated to non- 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Sequence voltages of the 115 kV bus# 24160
1.4 1.5

affected phases (phase 1

Positive

B in this case) 0.5 Negative

Zero
0
• Voltages of all three 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
Sequence voltages of the 500 kV bus# 24151
1.4 1.5

phases were depressed 1

Positive
Negative
after 0.95 s 0.5
Zero
0
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5

31
 Effects of load composition on FIDVR
phase A Casephase
1 B phase C
Case 1.5
Load composition
#

Speed (pu)
1
1 75% 1-Ф A/C motor, Case 1
25% constant impedance 0.5

2 75% 1-Ф A/C motor, 0

10% 3-Φ NEMA type B induction 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
1.5
motor, 15% constant impedance

Speed (pu)
1
3 70% 1-Ф A/C motor, Case 2
15% 3-Ф NEMA type B induction 0.5
motor, 15% constant impedance 0
4 60% 1-Ф A/C motor, 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
1.5
15% 3-Ф NEMA type B induction

Speed (pu)
motor, 25% constant impedance 1
Case 3
5 50% 1-Ф A/C motor, 0.5
25% 3-Ф NEMA type B induction 0
motor, 25% constant impedance 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
Speed (pu) 1.5
• Propagation of A/C motor stalling to 1
unfaulted phase is consistent across Case 4
0.5
a substantial range of load 0
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
compositions 1.5
• The impacts of SLG faults close to
Speed (pu)

1
certain regions of the system with Case 5
0.5
high A/C penetration could be more 0
severe than perceived 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
Time (s)
32
 Effects of point-on-wave (POW) on FIDVR
• POW effects on the
Phase A Phase B Phase C
occurrence and
1 1

Speed(pu)

Speed(pu)
evolution of FIDVR (a1) (a2)
0.5 0.5
are apparent,
0 0
based on the 0.5 1 1.5 2 0.5 1 1.5 2
1 1

Voltage(pu)

Voltage(pu)
differences in the (b1)
(b2)
0.5 0.5
response of the
A/C motors of bus 0
0.5 1 1.5 2
0
0.5 1 1.5 2
Positive Seq.Voltage(pu)

Positive Seq.Voltage(pu)
24160 bus 24151
bus 24160 1
bus 24151
bus 24160
1
• The POW when the 0.8
(c1)
0.8
(c2)

fault occurred 0.6 0.6

should be 0.5 1 1.5 2 0.5 1 1.5 2
t(s)
considered for Case 5A, POW = 45 deg
t(s)

Case 5, POW = 0 deg

detailed FIDVR
study
33
 Takeaways of part-2

1. The study shows that a normally cleared,

single line to ground fault at a 500 kV bus
close to the A/C loads can lead to a FIDVR
event. The event begins with A/Cs stalling on
two directly-impacted phases, followed by A/C
stalling propagating to the unimpacted phase.
2. Further, five study cases with quite different
load compositions show similar A/C motor
stalling results.
3. The POW when the fault occurs could have a
significant impact on the response of the A/C
compressor motors.
34
 Conclusions
1. A new EMT-TS hybrid simulation tool is developed,
which features 1) decoupling architecture, 2) generic
interfacing with an EMT simulator, 3) flexible switching
of interaction protocol and 4) support of three-phase
equivalent of external system.
2. The hybrid simulation tool has been applied to study
the FIDVR phenomenon within a region of WECC
system, certain aspects of the evolution of the
phenomenon were uncovered for the first time.
3. The developed tool is not limited to FIDVR study, but
also applicable to studies that require detailed models
and simulation of a part of a large power system, while
preserving the slow dynamics of the rest of the system

35
Thank you!

36