by
Subhash Joshi T G
Email: subhashj@cdac.in
1 of 63
SUMMARY OF PRESENTATION
Unit vectors and its significance
Basics of transformations
Session 1: Unit Vectors For 3 Balanced/Unbalanced Grid
Selection of reference variable for control
Selection of output parameter for PIcontroller
Design of PIcontroller constants
System bandwidth and method to select bandwidth
Unit vector under unbalanced grid condition
Test results
A Simple method for unit vector construction for balanced grid
Session 2: Unit Vectors For 1 Grid
Constructing two unit magnitude 900 displaced components
Mitigating the effect of grid frequency variation
Approximation method
Rigorous method
Test results 2 of 63
Unit Vectors
What is unit vector ?
Two unity magnitude fundamental
sinusoidal quantities, which are displaced
by 900 from each other
One of the unit vector is in phase with grid
voltage
This should be free from harmonics
Phase angle error from grid voltage should be minimum
Significance of unit vector ?
STATCOM is an independent voltage source
Unit vector helps to synchronize STATCOM voltage and Grid
voltage
Also unit vector is used to extract the active and reactive power
component separately
It is also used to separate the individual harmonics in case of
active power filters 3 of 63
Transformations β axis
XB
3 to  transformations θ
X X R X Y X B
1 120o α  axis
120o XR
2
X
3
X Y X B XY
2 β axis
 to DQ transformations Q axis
X d U1 U 2 X Xres
X
Xβ
X D axis
q 2 U U 1
θ ρ
U1 and U2 unity magnitude Xα α axis
components, where U2 lag U1 by 900
For grid connected system U1 is
aligned along Rphase grid voltage
4 of 63
SESSION 1
5 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Let the grid voltages are,
VR
VR Vg sin g t V
VY 3 
VY Vg sin( g t 120) VB
V
VB Vg sin( g t 120)
Objective,
U1 sin g t
U 2 cos g t
Apply three phase to two phase () transformation,
1 3
V VR (VY VB ) Vg sin g t
2 2
3 3
V (VY VB ) Vg cos g t 6 of 63
2 2
Unit vector for Balanced/Unbalanced
Grid condition
If U1 and U2 are known then V and V can be transformed to
DQ axis VR  V DQ Vd
Assume U1 and U2 are known, then V
Y 3 
Vd U1 U 2 V VB V Vq
V V
q 2 U U 1
U1 U2
Let U1 is not synchronized to VR and its frequency is g’, then,
3 3
Vd sin g t cos g t 2 g
V sin t
g V
2 g cos(( g g )t )
V cos t sin t 3
3
q g g Vg cos g t Vg sin(( g g )t )
2 2
When U1 gets synchronized with VR, then g=g’ and,
Vd 3 Vg
V 2
q 0 7 of 63
Unit vector for Balanced/Unbalanced
Grid condition
If Vq = 0 is ensured, then U1 will be synchronized to VR
To ensure Vq = 0, feedback with simple PIcontroller can be
used
VR  V DQ Vd
VY e(t) ki
3  + kp
VB V Vq s
U1 U2 0
8 of 63
Unit vector for Balanced/Unbalanced
Grid condition
What should be the output parameter of PIcontroller ?
PIcontroller performs well when the selected output parameter magnitude
swing is minimal
Our objective is to obtain U1 and U2, but they are sinusoidally varying
quantities and results in large swing in magnitude
The angle of U1 and U2 are also will be varying with time in large extend
Instead the frequency of U1 and U2 can be thought of selecting as output
parameter of PIcontroller
U1 sin t , U 2 cos t
VR  V DQ Vd
VY 3  +
e(t)
kp
ki
VB V Vq s
U1 U2 0
9 of 63
Unit vector for Balanced/Unbalanced
Grid condition
What should be the output parameter of PIcontroller ?
If the grid frequency variation is minimal, then
PIcontroller performance can be further improved by selecting the output
of PIcontroller as ∆ (variation in grid frequency) instead of absolute grid
frequency,
VR  V DQ Vd
VY e(t) ki ∆ 1 t
3  + kp +
VB V Vq s s
U1 U2 0 ref
U1 Lookup
ref is the nominal grid frequency U2 Table
10 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Design of PIcontroller constants ?
At time t=0, let grid and unit vectors (U1 and U2) frequency be g
At time t=0+, grid frequency changed from g to g and unit vector
frequency remains at g
3 to  transformation of grid voltage is,
3 3
V Vg sin g t , V Vg cos g t
2 2
 to DQ transformation gives,
3 3
Vd sin g t cos g t 2 g
V sin t
g V
2 g cos(( g g )t )
V cos t sin t 3 3
q g g V cos t Vg sin(( g g )t )
2 2
g g
11 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Design of PIcontroller constants ?
Assuming grid frequency variation is minimal, then
sin(( g g )t ) ( g g )t 3
V cos(( )t )
Vd 2 g g g
V 3
Hence Vq is, q Vg sin(( g g )t )
2
3
Vq Vg ( g g )t
2
VR(’g)
 V(’g) DQ Vd
VY(’g)
e(t) ki ∆ 1 gt
VB(’g)
3   kp +
V(’g) Vq + s s
0 ref
e(t) ki ∆ 1 gt
 kp +
Vq + s s
0 ref
g t
3 +
Vg 
2 14 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Design of PIcontroller constants ?
The block diagram can be further reduced by removing the quantities not
taking part in transients
ref is constant and no effect on transients
Once grid voltage is changed from g to g , g
can be assumed to be constant
Simplified block diagram is,
3 sk p ki 1 Vqref=0 ki 1
G( s) H ( s) Vg +
 kp
2 s s s s
s
1
3
Vg
3 ki k p 2
G ( s ) H ( s ) Vg 2
2
s
k i
16 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Design of PIcontroller constants ?
Find asymptotic bode plot of the system,
s
For zero: 1
Corner frequency, z ki k p 3 ki k p
Magnitude plot: +20dB/decade G ( s ) H ( s ) Vg 2
2
Phase plot: +450/decade
For poles:
s
k i
Corner frequency, p ki
Magnitude plot: 40dB/decade
Phase plot: 1800
17 of 63
Unit vector for Balanced/Unbalanced
Grid condition
s
Design of PIcontroller constants ? 1
3 ki k p
Closed loop system will be stable, G ( s ) H ( s ) Vg 2
2
If the open loop gain crosses 0dB (unity s
k
gain) with 20dB/decade and ensuring i
18 of 63
Unit vector for Balanced/Unbalanced
Grid condition
s
Design of PIcontroller constants ? 1
3 ki k p
z = p implies, G ( s ) H ( s ) Vg 2
2
s
k
ki i
ki
kp
k p ki
ki
One more component present
in the open loop transfer
function is,
3 V
2 g
This will not affect the phase
plot, but gain cross over
frequency will be pushed further
away increasing the bandwidth
of the system
19 of 63
Unit vector for Balanced/Unbalanced
Grid condition
k p ki
Find system bandwidth ?
Gain = 0dB
Defined as the frequency at
which the closedloop 20dB/dec
40dB/dec
magnitude is equal to 3 dB g
For phase of 1200, the open
loop bandwidth and closed loop 20dB/dec
bandwidth are found to be ki
closer
Since gain at ki is 0dB, the
gain cross over frequency of
open loop system is,
g ki
1350
20 of 63
Unit vector for Balanced/Unbalanced
Grid condition
k p ki
Find system bandwidth ?
Gain = 0dB
The modified crossover
frequency considering 3/2Vg is, 20dB/dec
40dB/dec
Gain at g ki is, g
3
20 log Vg 20dB/dec
2 ki
3
0 20 log Vg
2 20
1350
log g log ki
21 of 63
Unit vector for Balanced/Unbalanced
Grid condition
k p ki
Find system bandwidth ?
Gain = 0dB
The modified crossover
frequency considering 3/2Vg is, 20dB/dec
40dB/dec
3 g
g Vg k i
2 20dB/dec
ki
Closed loop bandwidth (appx.)
is,
3
BW Vg k p
2
Once the bandwidth is fixed
proportional constant (kp) can
be found
1350
22 of 63
Unit vector for Balanced/Unbalanced
Grid condition
k p ki
How to fix the bandwidth ?
Bandwidth is decided by the harmonics present in Vd and Vq components
as well as the response time requirement in transient conditions
For a balanced three wire system the minimum harmonics expected on
Vd and Vq are 300Hz (transformation of 5th and 7th harmonics)
Here bandwidth of 30Hz to 60Hz will be sufficient for proper
attenuation of harmonics in grid voltages
In Summary:
Based on the application and transient requirement fix the bandwidth
If grid voltage peak is known, proportional constant (kp) can be found
BW
kp
3 V
2 g
Once kp is known integral constant ki can be computed
ki k p
2
23 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Unit vector under unbalanced grid condition ?
Under unbalanced grid conditions, the grid voltage contains fundamental
positive sequence as well as fundamental negative sequence components
Let us construct the unit vector such that it is synchronized with
fundamental positive sequence component
As earlier fundamental positive sequence present in the grid voltage when
transformed to DQ reference frame reflected as d.c. component and
fundamental negative sequence present in the grid voltage reflected as
100Hz component
Since we required only d.c. component in Vq, 100Hz component can be
easily removed by using simple low pass filter of corner frequency around
10Hz
24 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Unit vector under unbalanced grid condition ?
C is the corner frequency of low pass filter
Low pass filter will not affect the information of fundamental positive
sequence, as it is a d.c. quantity
VR
 V DQ Vd
VY s e(t) ki ∆ 1 t
VB
3 
s C
+ kp +
V Vq s s
0 ref
U1 U2 U1 Lookup
U2 Table
25 of 63
Unit vector for Balanced/Unbalanced
Grid condition
Test results
Transient performance analysis carried out by simulation using PSIM
simulation package
3 Grid
VR VY VB
400
200
200
400
Unit vectors
cosrow sinrow
0.5
0.5
1
Unit vectors
cosRow sinRow
1
0.5
0
0.5
1
t
theta
400
300
200
100
0
0.36 0.38 0.4 0.42 0.44 0.46
Time (s)
Unit vector
V t Vg sin g t , V t Vg cos g t
3 3
2 2
The output of Low Pass Filter (LPF) is given by,
V s V s
s s
V s , V s
s c s c
VR V s V’
 s C
VY
3 s
VB s C V’
V
29 of 63
A Simple method for
Unit vector in Balanced Grid condition
Output of LPF in time domain is,
3 V 3 V
V t 2 sin g t , V t 2 cosg t
g c g c
g 2 c 2 g 2 c 2
g
where, tan
c
V (t ) and V (t ) are always 90 displaced
Unit vectors
0
X1 U1t sin g t
X1 / Y1
Y1
VR V s V’(t)
 ZCD S&H ABS
s C Trigger
VY
3 s Trigger
VB s C V’(t) ZCD S&H ABS
V
Y2
U 2 (t ) cos g t
X2 / Y2
X2
31 of 63
A Simple method for
Unit vector in Balanced Grid condition
Let us perform the following operations,
F1 (t ) V (t ) V (t )
3 Vg c
2 sin g t
c g2 2
F2 (t ) V (t ) V (t )
3 Vg c
2 cos g t
c g2 2
c g
where, tan
c g
Phase angle () is minimum when c=250
For c=250, is zero when grid frequency is 50Hz 32 of 63
A Simple method for
Unit vector in Balanced Grid condition
Complete block diagram of unit vector construction
where, c=250
X1 U1 t sin g t
X1 / Y1
V s V’ Y1
VR + ZCD S&H
 s C  ABS
Trigger
VY
3 Trigger
s
VB ++ ZCD S&H ABS
V s C V’ Y2
U 2 (t ) cos g t
X2 / Y2
X2
This method will introduce a small phase angle error when grid
frequency varies
2
Each harmonics is reduced by 2 when compared to
fundamental 1 h
33 of 63
A Simple method for
Unit vector in Balanced Grid condition
Test results
Grid voltage
Unit vector
34 of 63
SESSION 2
35 of 63
Overview of the presentation
Constructing two unit magnitude 900 displaced components
Mitigating the effect of grid frequency variation
Approximation method
Rigorous method
36 of 63
Overview of the presentation
Constructing two unit magnitude 900 displaced components
Mitigating the effect of grid frequency variation
Approximation method
Rigorous method
37 of 63
Innovative method for constructing
unit vector
Let Rphase grid voltage be, VR = Vg sin(St)
Let the above voltage is passed through a LPF of corner
frequency, c
Using Laplace analysis
Vg C S 1 C S s
FR(s) 2 2
2 2
2 2
S C s C S s S s S +

c
VR FR
The steady state output, FR(t) is given by s
Vg C S
FR(t)steady sin S t and tan
S 2 C 2 C
Transient term of FR(t) is given by,
V gcs ct
FR(t)transient= e
2 2
c s
38 of 63
constructing unit vector (contd.)
Vg C
Let the FR(t) is again pass through a FR(t)steady sin S t
S C
2 2
+

c +

c
VR FR’
s FR s
Using Laplace analysis
FR s VgC S
2 1
1
2
2 1
s
2
2 1
2
2 2 2
S
2
C
2
C
s C
S C
s C 2
s S
C 2 2 C S
s S
2
cos S t
result be FR’’(t) S
2
C
2
VR = Vg sin(St)
+

c +

c
VR FR’
s FR s

+ FR’’
FR’
s FR s
C
2 2
S

+ FR’’
Vg C S
FR t sin S t
S
2
C
2
FR’(t) and FR’’(t) are always 900
displaced irrespective of grid
frequency and corner frequency of
LPF
At the zero crossing of FR’’(t), FR’(t)
will be in peak and vice versa
Unit magnitude of FR’(t) and FR’’(t)
can be obtained by dividing the term
by its own magnitude 41 of 63
constructing unit vector (contd.)
VR +

c +

c FR’ X F1= cos(St+)
X/Y
s FR s Y
ZCD S&H ABS
Trigger
Trigger
ZCD S&H ABS

Y
+
X/Y
FR’’ X F2 =sin(St+)
42 of 63
constructing unit vector (contd.)
Inference from the above result (contd.) VR = Vg sin(St)
F1(t) = cos(St+)
If corner frequency of LPF (c) is set F2(t) = sin(St+)
equal to grid frequency (s), i.e. c = s : S
2 2
sin
C
phase shift = 0 S C
2 2
2C S
F2(t) is in phase with grid voltage cos
S 2 C 2
and F1(t) is lagging the grid voltage by 900
Fix the corner frequency of LPF (c) is equal to s = 250
rad/sec, where grid frequency fs = 50Hz
Let Grid frequency varies a maximum of 10% (45Hz to
55Hz)
Phase shift, = 60 for 10% of grid frequency variation
Loose the proper synchronization
43 of 63
constructing unit vector (contd.)
Phase error with variation in grid frequency
1ph without comp
7
6
6.0
5
3
phase error
0
10 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10
1
2
3
4
5
6
7
% ge of grid frequency variation
44 of 63
Overview of the presentation
Constructing two unit magnitude 900 displaced components
Mitigating the effect of grid frequency variation
Approximation method
Rigorous method
45 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation
Give a variation for the grid frequency s in the peak
of FR’’(t) V = V sin( t)R g S
Substitute c= s and solving will give
F t
V
sin t
g C S
C
R 2 2 S
1
S
Vg
s
F
' '
R s s 2 2
c s 1
1
s 2 s
Vg
2
F '' 1
if is negligible compare to 1
R s s 2 2 s
c s
Peak of FR’’(t), F’’R(peak), is more or less independent of
grid frequency variation
F’’R(peak) Vg/2 for c = s
46 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
The following trignometric relation can be V = V sin( t) R g S
S
2 2
cos S t Vg cos sin S t Vg sin Vg cos S t sin
C
S C
2 2
2C S
Vgcos = 2F’’R(peak) cos
S 2 C 2
Vg C S
FR t sin S t
Solve the following mathematical relation
S 2 C 2
V gcs
2 FR( peak ) Vg
2Vg C
2
Vg Vg
S
2 2
F ''
R peak
2 2
C c s
C
2 2 2 2
S S C F’’R(peak) Vg/2 for c = s
VgC
2
2 FR( peak ) Vg Vg sin FR t
S
2
C
2
cos t S
V gc 2
Vg sin 2 FR( peak ) 2 FR( peak ) F'
R peak
c 2 s 2
47 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Rewriting the relation V = V sin( t)
V gc 2
R g S
sin S t Vg cos cos S t Vg sin Vg sin S t Rpeak c2 s2
F '
V gcs
F ''
R peak
2 2
Substituting Vg C S
c s
FR t sin S t
Vgcos = 2F’’R(peak)
S 2 C 2
Vg sin 2 FR( peak ) 2 FR( peak ) VgC
2
FR t
S
2
C
2
cos t
S
Vg sin S t U1 sin S t (2 FR( peak ) ) cos S t 2 FR( peak ) 2 FR( peak )
cosst
U 2FR '' (t ) 2FR ' (t ) 2 F
1 R( peak )
2VgC S C S
2 2
C S
U1 sin S t tan
S
2
C
2
and
S
48 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Rewriting the relation V =V R g sin(St)
V gc 2
F'
cos S t Vg cos sin S t Vg sin Vg cos S t R peak
c 2 s 2
V gcs
F ''
R peak
2 2
Substituting Vg C S
c s
FR t sin S t
Vgcos = 2F’’R(peak)
S 2 C 2
Vg sin 2 FR( peak ) 2 FR( peak )
VgC
2
FR t
S
2
C
2
cos t
S
Vg cos S t U 2 cos S t (2 FR( peak ) ) sin S t 2 FR( peak ) 2 FR( peak )
2VgC S C S
2 2
C S
U2 cos S t tan
S
2
C
2
and
S
49 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Comparing steady state term of U1(t) and U2(t) V = V R g sin(St)
2VgC S C S
2 2
U1 sin S t
S
2
C
2
2VgC S C S
2 2
U2 cos S t
S
2
C
2
Inference from the above result
U1(t) and U2(t) are 900 displaced
At the zero crossing of U1(t), U2(t) will be in peak and vice
versa
Unit magnitude of U1(t) and U2(t) can be obtained by
dividing the term by its own magnitude
U1(t) is phase shifted from grid voltage by angle ()
50 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
VR = Vg sin(St)
Inference from the above result (contd.) C S
tan
Fix the corner frequency of LPF (c) is equal S
to s = 250 rad/sec, where grid frequency fs sin
S
2 2
C
S C
2 2
= 50Hz
2C S
Let Grid frequency varies a maximum of cos
S 2 C 2
10% (45Hz to 55Hz)
Give a variation and substitute c= s in
the phase angle ()
tan tan
tan
1 tan tan
2
1
2 S
2
3
tan S S
if 2 1
and is negligible compare to 1
C S
s
2
s
1
S 51 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Phase error with variation in grid frequency
Phase shift, () = 0.320 for 10% grid frequency variation
1ph with comp (appx)
1
phase error
0
10 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10
0.32
1
2
3
52 of 63
% ge of grid frequency variation
constructing unit vector (contd.)
FR’(peak)
VR +

c +

c FR’(t)
F1(t)
s FR s X
Y + U1(t)
ZCD S&H ABS X/Y 
+
Trigger
Trigger
+
ZCD S&H ABS X/Y 
+
 Y
X
U2(t)
+
FR’’(t)
FR’’(peak) F2(t)
cosst
U1(t)
X U 2FR '' (t ) 2FR ' (t ) 2 F
Y cos 1 R( peak )
ZCD S&H ABS X/Y
U 2 2 FR (t ) 2FR( peak ) cos S t 2FR( peak ) sin S t
Trigger
Trigger F1(t) = cos(St+)
ZCD S&H ABS X/Y F2(t) = sin(St+)
U2(t) Y
X
sin
53 of 63
Overview of the presentation
Constructing two unit magnitude 900 displaced components
Mitigating the effect of grid frequency variation
Approximation method
Rigorous method
54 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
VR = Vg sin(St)
The approximation made in the earlier S
2 2
sin
C
derivation is: S C
2 2
S
2 2
Vg C S
2 2
Vg sin
S
S C
FR peak Vg C 2
2 2
2
S C
2
55 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Rewriting the relation VR = Vg sin(St)
V gc 2
sin S t Vg cos cos S t Vg sin Vg sin S t
F'
R peak
c 2 s 2
V gcs
cos S t Vg cos sin S t Vg sin Vg cos S t F ''
R peak
2 2
c s
Substituting
U 2FR '' (t ) 2FR ' (t ) 2 F cos st
1 R ( peak )
FR( peak ) FR( peak )
2 2 S
U1 Vg sin S t 2 FRt
FR( peak )
S
FR( peak ) FR( peak )
U 2 Vg cos S t 2 FR( peak ) cos S t
2 2
sin t
FR( peak )
S
56 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Phase error with variation in grid frequency
Phase shift, () = 00 for any grid frequency
1ph with comp (rigorous)
1
phase error
0
10 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10
0
1
2
3 57 of 63
% ge of grid frequency variation
constructing unit vector (contd.)
 FR’(peak)
VR +

c + c FR’(t) X F1(t)
X/Y + U1(t)
s FR s +
Y1
ZCD S&H ABS
Trigger Y12
+ X

Y22 X Y
Trigger
ZCD S&H +
ABS Y2 2
 X/Y  U2(t)
+ X F2(t)
2
FR’’(t)
FR’’(peak)
U1(t) X
U1 Vg sin S t 2 FRt
FR( peak ) FR( peak )
2 2
cos t
Y cos
FR( peak )
S
ZCD S&H ABS X/Y
Trigger
FR( peak ) FR( peak )
U 2 Vg cos S t 2 FR( peak ) cos S t
2 2
sin t
FR( peak )
S
Trigger
ZCD S&H ABS X/Y F1(t) = cos(St+)
U2(t) Y
X
sin
F2(t) = sin(St+) 58 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Test results
Grid frequency (simulated using function generator)
is varied at t=t1 from 50Hz to 45Hz
Grid voltage and unit vector without Grid voltage and unit vector with
compensation for grid frequency variation compensation for grid frequency variation
59 of 63
constructing unit vector (contd.)
Mitigating the effect of grid frequency variation (contd.)
Test results
Steady state waveform at 25Hz and 75Hz with
approximation method
Grid voltage and unit vector without Grid voltage and unit vector with
compensation for grid frequency variation compensation for grid frequency variation
61 of 63
Conclusions
Single grid voltage is considered for the construction of unit
vector
Two unity magnitude fundamental sinusoidal quantities,
which are displaced by 900 from each other
One of the unit vector is in phase with grid voltage
irrespective of the grid frequency variation
62 of 63
Thank you
63 of 63