You are on page 1of 7

Electric Power Systems Research 81 (2011) 967–973

Electric Power Systems Research 81 (2011) 967–973 Contents lists available at ScienceDirect Electric Power Systems Researchcommon practice to re present all gen- Corresponding author. E-mail addresses: stijn.cole@esat.kuleuven.be , stijn.cole@gmail.com (S. Cole). 0378-7796/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi: 10.1016/j.epsr.2010.11.032 erators by the classic generator model, and only a few generators of interest by detailed models of the generators and their con- trols. Standard models exist for numerous power system equipment, such as excitation systems [1] , and steam and hydro turbines [2] . However, to the authors’ best knowledge, no standard mod- els for VSC HVDC have been proposed in literature. We believe there is a need for generic, standard models for VSC HVDC sys- tems. VSC HVDC has an increasingly wide range of application in the power system [3] . Hence, it is clearly an area of inter- est and receives a lot of attention. Quite some research papers are dedicated to the subject of modelling VSC HVDC systems [4–7] . However, most of these models are not generic, e.g. the model in [6] is derived explicitly for twelve pulse converters [4,7] are only valid for systems using Pulse Width Modulation (PWM). This paper proposes two standard VSC HVDC dynamic mod- els for power system stability studies, that are generic, i.e. valid regardless of power electronics topologies. The phasor modelling approach is used in this paper, as is customary in transient stabil- ity studies. First, a full model will be constructed: the converter equations, equations of the DC circuit and coupling equations are derived in Section 2 . In Section 3 , the controller equations are appended to the system of equations to get the complete sys- tem of equations for the full model. Next, a reduced order model is derived in a mathematically rigorous way in Section 4 . Lastly, simulations validate and compare the proposed models (Section 5 ). " id="pdf-obj-0-6" src="pdf-obj-0-6.jpg">

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Electric Power Systems Research 81 (2011) 967–973 Contents lists available at ScienceDirect Electric Power Systems Researchcommon practice to re present all gen- Corresponding author. E-mail addresses: stijn.cole@esat.kuleuven.be , stijn.cole@gmail.com (S. Cole). 0378-7796/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi: 10.1016/j.epsr.2010.11.032 erators by the classic generator model, and only a few generators of interest by detailed models of the generators and their con- trols. Standard models exist for numerous power system equipment, such as excitation systems [1] , and steam and hydro turbines [2] . However, to the authors’ best knowledge, no standard mod- els for VSC HVDC have been proposed in literature. We believe there is a need for generic, standard models for VSC HVDC sys- tems. VSC HVDC has an increasingly wide range of application in the power system [3] . Hence, it is clearly an area of inter- est and receives a lot of attention. Quite some research papers are dedicated to the subject of modelling VSC HVDC systems [4–7] . However, most of these models are not generic, e.g. the model in [6] is derived explicitly for twelve pulse converters [4,7] are only valid for systems using Pulse Width Modulation (PWM). This paper proposes two standard VSC HVDC dynamic mod- els for power system stability studies, that are generic, i.e. valid regardless of power electronics topologies. The phasor modelling approach is used in this paper, as is customary in transient stabil- ity studies. First, a full model will be constructed: the converter equations, equations of the DC circuit and coupling equations are derived in Section 2 . In Section 3 , the controller equations are appended to the system of equations to get the complete sys- tem of equations for the full model. Next, a reduced order model is derived in a mathematically rigorous way in Section 4 . Lastly, simulations validate and compare the proposed models (Section 5 ). " id="pdf-obj-0-17" src="pdf-obj-0-17.jpg">

A proposal for standard VSC HVDC dynamic models in power system stability studies

S. Cole a, , R. Belmans a,b

a Katholieke Universiteit Leuven, ESAT/ELECTA, Kasteelpark Arenberg 10, 3001 Heverlee, Belgium b Elia, Keizerslaan 20, 1000 Brussel, Belgium

article

info

Article history:

Received 4 January 2010

Received in revised form 24 October 2010 Accepted 30 November 2010

Available online 3 January 2011

HVDC converters HVDC transmission HVDC transmission control Power system modelling Power system simulation

abstract

In this paper, two standard VSC HVDC dynamic models are proposed. The full system model, consisting of the converter and its controllers, DC circuit equations, and coupling equations, is derived mathematically. Special attention is given to the filter and phase-locked loop (PLL), often neglected in VSC HVC modelling. A reduced order model is then derived from the full model by neglecting the smallest time constants, resulting in a reduced set of differential equations that can be integrated with a larger time step. The

models are implemented in MatDyn, a Matlab based stability program. Simulations validate the proposed models.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Power system stability studies require phasor models of the var- ious equipments present in the modern power system, such as generators and their controls, FACTS devices, and loads. In most power system software, the user can select models from a stan- dard models library or specify his own, user-defined models. Using standard models for power system equipment has certain advan- tages. Data exchange between utilities, and the transition to new software packages proves much more convenient if standard mod- els are used. Furthermore, standard models can be used for a wide range of studies. In the planning stage, when the details of the equipment are not yet known, standard models can be used. In order to obtain acceptable results, the generic models should be quite detailed. We call such models ‘full models’. If the equip- ment is located far away from the part of the system under study, the equipment can be modelled with less detail; we refer to such models as ‘reduced’ models. Only for a detailed study of a part of the system where the equipment plays a prominent role, standard models should not be used. A model that very closely mimics the behaviour of the actual system is needed. Such a ‘detailed model’ is usually delivered by the manufacturer. An example of this approach are stability studies in the early days of computer simulation of power systems when it was common practice to represent all gen-

Corresponding author. E-mail addresses: stijn.cole@esat.kuleuven.be, stijn.cole@gmail.com (S. Cole).

0378-7796/$ – see front matter © 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.epsr.2010.11.032

erators by the classic generator model, and only a few generators of interest by detailed models of the generators and their con- trols. Standard models exist for numerous power system equipment, such as excitation systems [1], and steam and hydro turbines [2]. However, to the authors’ best knowledge, no standard mod- els for VSC HVDC have been proposed in literature. We believe there is a need for generic, standard models for VSC HVDC sys- tems. VSC HVDC has an increasingly wide range of application in the power system [3]. Hence, it is clearly an area of inter- est and receives a lot of attention. Quite some research papers are dedicated to the subject of modelling VSC HVDC systems [4–7]. However, most of these models are not generic, e.g. the model in [6] is derived explicitly for twelve pulse converters [4,7] are only valid for systems using Pulse Width Modulation (PWM). This paper proposes two standard VSC HVDC dynamic mod- els for power system stability studies, that are generic, i.e. valid regardless of power electronics topologies. The phasor modelling approach is used in this paper, as is customary in transient stabil- ity studies. First, a full model will be constructed: the converter equations, equations of the DC circuit and coupling equations are derived in Section 2. In Section 3, the controller equations are appended to the system of equations to get the complete sys- tem of equations for the full model. Next, a reduced order model is derived in a mathematically rigorous way in Section 4. Lastly, simulations validate and compare the proposed models (Section

5).

Q

  • 968 S. Cole, R. Belmans / Electric Power Systems Research 81 (2011) 967–973

U f i i tr pr X X tr R tr pr R pr U s
U f
i
i
tr
pr
X
X
tr
R tr
pr
R pr
U s
C f
U c
Fig. 1. One-line diagram of the AC Circuit with filter. U s is the system voltage, U c
the converter voltage.
  • 2. VSC HVDC model

2.1.

Converter

 

equations of this circuit,

 
 

di

pr

+ R pr i pr ,

u c u f =

L pr

dt

di

tr

+ R tr i tr ,

u f u s = L tr

dt

i pr = i tr + C f

du f

 

dt

,

di

d

pr

dt

di

q

= −

L

R pr

pr

d

i pr + ωi

q

pr +

1

L

pr

(u d u

c

d

f

pr

dt

di

d

tr

dt

= −

R pr

L

pr

q

i pr ωi

d

pr +

1

L

pr

= −

R tr

L

i tr d + ωi q

tr +

1

L

(u

(u q u

c

q

f

d

f

u

d

s

),

 

tr

tr

di

q

tr

dt

R

= −

L

tr

i tr q ωi d

tr +

1

L

(u

q

f

u

q

s

),

tr

tr

du

d

 

f

dt

d

= −ωu

f

+

1

C

f

(i

d

pr i d

tr ),

du

q

 

f

dt

= ωu

q

f

+

1

C

f

(i

pr i q

q

tr ).

2.2.

Phase-locked loop

 

VSC HVDC converters are connected to the system through a

phase reactor and a transformer. A filter is connected between the

phase reactor and the transformer (Fig. 1). It was shown in [8], that

the filter behaves as a pure capacitor at system frequency. The basic

(1)

(2)

(3)

(4a)

(4b)

(5a)

(5b)

(6a)

(6b)

are transformed to a rotating reference frame:

),

),

The angle ωt is provided by the phase-locked loop (PLL), arbi-

trarily assumed here to align system voltage with the q-axis.

In general, a PLL is “a circuit synchronizing an output signal (gen-

erated by an oscillator) with a reference or input signal in frequency

as well as in phase.” [9, p. 1]. It is a control system that acts on

the phase difference between the reference signal and the output,

such that the phase of the output is locked to the phase of the ref-

erence [9, p. 1]. All PLLs consist of three basic components: a phase

detector (PD), a voltage controlled oscillator (VCO), and a loop fil-

ter (LF). The phase detector compares the phase of the input or

reference signal with the phase of the output signal, produced by

the voltage controlled oscillator. The voltage controlled oscillator

produces an oscillating signal with a frequency determined by the

output signal of the loop filter. The loop filter removes noise and

high frequency signals, and is responsible for the control of the PLL.

Here the circuit of Fig. 2 is used. It is a type 2 loop, most prevalent

+ θ i θ ο 1+sa K vco K pd sb s − Fig. 2. PLL
+
θ i
θ ο
1+sa
K vco
K pd
sb
s
Fig. 2. PLL implementation.
0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0.15 0.16 0.17 0.18 0.19
0.1
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0.15
0.16
0.17
0.18
0.19
0.2
0.21
0.22
0.23
0.24
0.25

Time [s]

Fig. 3. Converter reactive power without PLL (close up).

in phase-locked loops [10, p. 16]. The differential equations of the

system are:

x˙ = bK pd ( i o ),

(7)

˙ o = K vco (bK pd ( i o ) + ax).

(8)

The design of the PLL can have a large influence on the dynamic

behaviour of the system. Therefore, it is now common practice

to include a detailed PLL model in electromagnetic programs. A

problem that can occur is that during or immediately after abnor-

mal system conditions such as faults, the PLL is not able to lock,

and consequently does not produce a correct phase signal. Another

problem is the operation in weak AC grids: the angle settling time

after disturbances is slow, while the current loop is fast. To cor-

rectly study this kind of behaviour and interactions, a phasor model

is not sufficient, as it can not capture the behaviour of the PLL in

abnormal system conditions and weak AC grids. A detailed electro-

magnetic model needs to be used. In phasor models it is therefore

less common to include the PLL. However, it has been shown in

some contributions that the PLL has an influence on stability. A PLL

introduces a small delay which is here in the order of 1 ms. In [11],

it has been shown that the PLL delay can cause undesired power

exchange. To investigate the influence of the PLL on power system

stability, we perform simulations with and without PLL, using mod-

ified versions of MATPOWER [12] and MatDyn [13]. At t = 0.2, a large

load with P and Q component is switched in. Fig. 3 shows a close up

around t = 0.2 of the converter’s reactive power output when no PLL

is included in the model. If a PLL is modelled, its time delay causes

the reactive power output to drop first (Fig. 4), in accordance with

the results obtained in [11]. When looking at overal reactive power

output (Figs. 5 and 6), it can be seen that the difference is very small

and does not significantly impact the other state variables, and thus

can be safely neglected in power system stability studies.

2.3. Filter

Many models proposed in literature do not take into account

the filter bus, e.g.: [4–6,14]. The one-line diagram is then simplified

Q

Q

Q

S. Cole, R. Belmans / Electric Power Systems Research 81 (2011) 967–973

969

0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0.15 0.16 0.17 0.18 0.19
0.1
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0.15
0.16
0.17
0.18
0.19
0.2
0.21
0.22
0.23
0.24
0.25

Time [s]

Fig. 4. Converter reactive power with PLL (close up).

0.8 0.6 0.4 0.2 0 0 1 2 3 4 5
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5

Time [s]

Fig. 5. Converter reactive power without PLL.

0.8 0.6 0.4 0.2 0 0 1 2 3 4 5
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
1 C=0 C=0.015 0 C=0.020 C=0.025 −1 −2 −3 0 0.05 0.1 0.15 0.2 U d
1
C=0
C=0.015
0
C=0.020
C=0.025
−1
−2
−3
0
0.05
0.1
0.15
0.2
U d [pu]

Time [s]

Fig. 8. d-Axis converter voltage for different values of filter capacitance in p.u.

5 C=0 C=0.015 4 C=0.020 C=0.025 3 2 1 0 0.05 0.1 0.15 0.2 U q
5
C=0
C=0.015
4
C=0.020
C=0.025
3
2
1
0
0.05
0.1
0.15
0.2
U q [pu]

Time [s]

Fig. 9. q-Axis converter voltage for different values of filter capacitance in p.u.

1.2 C=0 C=0.015 1 C=0.020 C=0.025 0.8 0.6 0.4 0.2 0 0.05 0.1 0.15 0.2 P
1.2
C=0
C=0.015
1
C=0.020
C=0.025
0.8
0.6
0.4
0.2
0
0.05
0.1
0.15
0.2
P [pu]

Time [s]

Fig. 10. Converter active power for different values of filter capacitance in p.u.

Time [s]

Fig. 6. Converter reactive power with PLL. i X eq R eq U s U c
Fig. 6. Converter reactive power with PLL.
i
X
eq
R eq
U s
U c

Fig. 7. One-line diagram of the AC circuit without filter.

1.6 C=0 1.4 C=0.015 C=0.020 1.2 C=0.025 1 0.8 0.6 0.4 0.2 0 0.05 0.1 0.15
1.6
C=0
1.4
C=0.015
C=0.020
1.2
C=0.025
1
0.8
0.6
0.4
0.2
0
0.05
0.1
0.15
0.2
Q [pu]

(Fig. 7), and Eqs. (4a)–(6b) reduce to:

Time [s]

di

d

dt

di

q

dt

= −

R eq

L

eq

i d + ωi q +

= −

R

eq

L

eq

i q ωi d +

Fig. 11. Converter reactive power for different values of filter capacitance in p.u.

1

L

eq

(u d u

c

1

L

eq

(u q u

c

d

),

(9a)

 

s

 

2.4.

DC circuit modelling

q

).

(9b)

 

s

 

The DC circuit of a VSC HVDC is represented in Fig. 12[15]. The

In Figs. 8 and 9 the response of converter voltage to a step in

d-axis reference current is shown for different values of filter capac-

itance. A slightly higher or lower value of the filter capacitance

has a large influence on the magnitude of the converter voltage.

The effect on active and reactive power output is less pronounced

(Figs. 10 and 11).

basic differential equations are:

C

dc

du dc 1

dt

C

dc

du dc 2

dt

= i dc 1 i cc ,

(10)

= i dc 2 + i cc ,

(11)

  • 970 S. Cole, R. Belmans / Electric Power Systems Research 81 (2011) 967–973

R L dc dc i u dc1 cc u dc2 C C dc dc i i
R
L
dc
dc
i
u dc1
cc
u dc2
C
C
dc
dc
i
i
dc1
dc2
C
C
dc
dc
i
cc
R
L
dc
dc

Fig. 12. DC circuit.

  • L dc

    • di cc

dt

= u dc 1 u dc 2 R dc i cc .

(12)

These equations can represent overhead lines, underground

cables or back-to-back installations by an appropriate selection of

parameters.

  • 2.5. Coupling equations

The AC and DC circuit equations have to be coupled. As men-

tioned in Section 1, some models proposed in the literature

explicitly rely on PWM to derive the model. In that case, the mod-

ulation index can be used to provide a relation between AC and DC

side. Here we aim to develop a generic model, also valid for VSC

HVDC systems that do not use PWM. AC and DC side are related

by the active power balance, that allow to calculate the q-axis ref-

erence current of the slack converter, 1 and the DC currents of the

other converters:

q

  • i =

pr

ref

n

2i dc n · u dc n i

d

pr n

· u d

c

n

u q

c

n

,

  • i =

dc i

u d i · i

c

d

pr i + u c q

i

·

i

q

pr i

2u dc i

,

i n 1.

(13)

(14)

The factor two in the denominator is present because a bipolar

configuration is assumed. For amonopolar configuration, this factor

needs to be removed.

  • 3. Control systems Control strategies is one of the most popular research topics in

the field of VSC HVDC. A variety of controllers have been proposed

in literature: internal model control in [16], H controllers are pro-

posed in [16,17], a control strategy based on Lyapunov functions in

[18], and optimal coordinated control of VSC and AC line in par-

allel [19]. While it is commendable to try to improve VSC HVDC

systems’ performance by innovative control strategies, these new

control strategies are often very specific and therefore not very suit-

able for a generic standard model. Furthermore, more conservative

control strategies are used in real applications. The control strategy

that is most used is vector control [17], which will be adopted here

too, in conjunction with PI controllers.

1 We call the converter that controls DC voltage the ‘slack converter’ because it compensates for the losses in the DC system.

3.1.

Current controllers

The converter equations with filter, (4a)–(6b), or without filter,

(9a) and (9b), are controlled using vector control. The reference

voltage in d and q components are expressed as:

u d

c

ref

= u

d

s

+ u d

c

+ K p1

+ K i1

s

(i

d

pr ref

i

d

pr ),

u q

c

ref

= u

q

s

+ u q + K p1

c

+ K i1 (i

s

q

pr ref

i

q

pr ),

with

u d =

c

q

ωL pr i pr ,

d

u q = −ωL pr i pr ,

c

(15a)

(15b)

(16)

(17)

the cross-coupling terms, that compensate for the cross coupling

between the two control loops.

We introduce two additional state variables, M d and M q , such

that:

dM d

dt

=

K i1 i

d

pr +

K i1 i

d

pr ref

,

dM q

dt

=

K i1 i

q

pr +

K i1 i

q

pr ref

.

(18a)

(18b)

The reference voltage can now be expressed as an algebraic

equation:

u d

c

ref

= u

d

s

+ u d + K p1 (i

c

d

pr ref

i

d

pr ) + M d ,

u q

c

ref

= u

q

s

+ u q + K p1 (i

c

q

pr ref

i

q

pr ) + M q .

(19a)

(19b)

The actual value of the voltage lags the reference due to the

time-lag of the converter. The relation between the actual value

and the reference value can be represented by a time delay with

time constant T [20]:

du d

c

dt

= −

1

T

u d +

c

1

T

u d ref ,

c

du q

c

dt

= −

1

T

u q +

c

1

T

u q ref .

c

(20a)

(20b)

Combining the time-lag equations and the equations of the cur-

rent controllers yields:

du d

c

dt

= −

K p1

T

i

d

pr

ω L pr

T

i

q

pr

1

T

u d +

c

1

T

M d + K p1 i

T

d

pr ref

+

1

T

u

d

s ,

(21a)

du q

c

dt

= −

K p1

T

i

pr + ω L pr

q

T

i

d

pr

1

T

u q +

c

1

T

M q + K p1 i

T

q

pr ref +

  • 1 q

T

u

s .

(21b)

3.2.

Outer controllers

3.2.1.

Reactive power control and voltage control

Every converter can control its reactive power injection. When

the system voltage is aligned with the q-axis, the reactive power Q

can be calculated as:

Q = u

q

s

d

i pr .

(22)

d

The d-axis current setpoint, i pr ref , is thus calculated from the

reactive power setpoint Q ref . A combination of an open loop and a

PI controller is used, leading to the equation [21]:

i

d

pr ref =

Q ref

u

q

s

+ K p2 + K i2 (Q ref Q ).

s

(23)

S. Cole, R. Belmans / Electric Power Systems Research 81 (2011) 967–973

971

We finally introduce state variable N d , resulting in an additional

differential equation for the PI controller,

dN

d

dt

= K i2 (Q ref u

q

s

d

i pr ),

(24)

such that substitution of (22) and (24) in (23) leads to an algebraic

expression for the d-axis current reference:

i

d

pr ref =

Q ref

u

q

s

+ N d + K p2 (Q ref u

q

s

d

i pr ),

which can be substituted in (21a).

(25)

Instead of reactive power, AC system voltage can be controlled.

The d-axis current reference can be calculated using a PI controller:

i

pr ref = K p3 + K i3

d

s

(U s ref u

q

s

).

  • 3.2.2. Active power control

(26)

The q-axis reference current can be calculated in a similar way

as the d-axis reference current. It must be noted that only n 1

converters of a general n converter VSC HVDC system can set the

active power reference.

i

q

pr ref =

P ref

u

q

s

+ K p4 + K i4

s

(P ref u

q

s

q

i pr ).

  • 3.2.3. DC voltage control

(27)

The remaining q-axis reference current can be obtained from

the DC voltage control equation of the slack converter:

i

dc

ref

n

=

K p5 + K i5 · (u dcref u dc n ).

s

(28)

The dynamics of the DC circuit are approximated by a time con-

stant T dc :

di

dc n

dt

= −

1

T dc

i dc n +

1

T dc i dc

ref

n

.

(29)

All equations are now derived. The full model consists of the AC

side equations with filters ((4)–(6)) or without (9), DC circuit equa-

tions ((10)–(12)), coupling equations ((13) and (14)), and current

(() and outer controllers ((24), (25) and (27)(29)). It is explained in

[15] how these equations can be generalized to account for multi-

terminal VSC HVDC systems.

  • 4. Simplified models In Section 1 it was explained that reduced order, or simplified,

models can be useful in power system analysis. In this section, a

set of new models is now derived that are simplifications of the full

model. A correct way of deriving simplified models is eliminating

very small time constants. This is equivalent to making the assump-

tion that very fast dynamics are infinitely fast compared to slower

phenomena. The full model has five time constants. For the AC side:

T , the time constant of the power electronics, and the time con-

stant associated with the phase reactor. For the DC circuit: the time

constant T dc , and the time constants related to L dc , and C dc . Each of

the time constants are analysed now.

AC current dynamics

By removing the phase reactor from the AC circuit (L pr = 0), the

AC current responds instantaneously to variations in the voltage

difference between the AC system and the converter. The dif-

ferential equations describing the current dynamics (4a)–(6b) or

(9a) and (9b) disappear. This simplification is only of theoretical

interest, as the phase reactor is the most important element of

the converters AC side.

AC voltage dynamics

The AC voltage dynamics can be neglected formally by assum-

ing that the voltage is equal to the voltage reference, or T = 0.

The equations describing the voltage dynamics (21a) and (21b)

disappear.

DC line current (i cc ) dynamics

By removing the inductor from the DC circuit, L dc is set to zero

and the dynamics of the DC line currents disappear from the

model.

DC voltage dynamics

By removing the capacitors from the DC circuit, C dc is set to zero

and the dynamics of the DC voltage disappear from the model.

This is also of theoretical interest only, as the DC capacitors are

the most important element in the DC circuit.

DC current (i dc ) dynamics

Lastly, the assumption T dc = 0, removes the DC current dynam-

ics. It assumes that the DC current required by the DC voltage

controller is provided by the converter without delay.

The five simplifications can be implemented separately or can

be combined with each other. In total, 31 simplified models can

be constructed, corresponding to 2 5 1 = 31 possible combinations.

The advantage of the simplifications is a reduction of the number

of differential equations. Furthermore, less data is needed which

can be hard to collect. Even more advantageous is the possibility

to speed up calculation significantly by using a larger integration

step size, if the overal systems dominant time constant is removed.

An intelligent way to simplify the full model is thus to remove one

or more of the smallest time constants. The difficulty in proposing

a standard simplified model lies in the fact that the smallest time

constants are dependent on the parameters, which can vary con-

siderably between different systems. The following generalities can

nevertheless be agreed upon:

the DC capacitors is the dominant element in the DC circuit and

its corresponding time constant should be maintained;

the phase reactor is the dominant element in the converter AC

side and its corresponding time constant should be maintained.

The three time constants that can be removed, are: T , T dc , and

the time constant related to L pr . The remaining dynamics are those

associated with the outer controllers, the phase reactor and the DC

capacitors. Compared to the full model, the order of the model is

reduced with 2n + m + 1, for a n converter system with m DC lines.

Furthermore, a larger step size can be selected.

5. Simulations

The step response of two models are compared. The first model

is a full model, without filter or PLL. The second model is a reduced

order model that neglects time constants T , T eq , and the time con-

stant related to L pr , as explained in Section 4. The VSC HVDC system

is connected between two infinite buses. The response to a step in

DC voltage reference is shown in Figs. 13–16. All simulations are

performed in a modified version of MatDyn [13,22]. For this simula-

tions operating limits are not taken into account. It can be observed

in all figures that the full model responds satisfactorily to the step

input in DC voltage. The reduced order model, while not as accu-

rate as the full model, reproduces the overal dynamic phenomena.

q

Especially for the variables U

c

and I cc , shown in Figs. 14 and 16,

it can be observed that the initial response of the reduced order

model is more abrupt than for the full order model. This conforms

to the mathematical derivation of the simplified models: the time

constants associated to these variables are eliminated. Mathemat-

ically, this means that the variables can change instantaneously. In

Fig. 15, it can be clearly seen that the current dynamics are elimi-

  • 972 S. Cole, R. Belmans / Electric Power Systems Research 81 (2011) 967–973

0.6 full reduced 0.5 0.4 0.3 0.2 0.1 0 −0.1 −0.2 0 0.2 0.4 0.6 0.8
0.6
full
reduced
0.5
0.4
0.3
0.2
0.1
0
−0.1
−0.2
0
0.2
0.4
0.6
0.8
1
U c d [pu]

Time [s]

Fig. 13. Converter d-axis voltage.

1.4 full reduced 1.3 1.2 1.1 1 0.9 0.8 0.7 0 0.2 0.4 0.6 0.8 1
1.4
full
reduced
1.3
1.2
1.1
1
0.9
0.8
0.7
0
0.2
0.4
0.6
0.8
1
U c q [pu]

Time [s]

Fig. 14. Converter q-axis voltage.

1.14 full reduced 1.12 1.1 1.08 1.06 1.04 1.02 1 0.98 0 0.2 0.4 0.6 0.8
1.14
full
reduced
1.12
1.1
1.08
1.06
1.04
1.02
1
0.98
0
0.2
0.4
0.6
0.8
1
U dc [pu]

Time [s]

Fig. 15. DC voltage.

0.2 full reduced 0.15 0.1 0.05 0 −0.05 −0.1 −0.15 −0.2 0 0.2 0.4 0.6 0.8
0.2
full
reduced
0.15
0.1
0.05
0
−0.05
−0.1
−0.15
−0.2
0
0.2
0.4
0.6
0.8
1
I cc [pu]

Time [s]

Fig. 16. DC current.

nated from the reduced order model: the remaining dynamics are

those associated with the charging of the DC capacitors.

6. Conclusion

Two standard dynamic models for VSC HVDC systems have been

proposed in this paper. The full model is constructed by combin-

ing AC side equations, DC system equations, control systems and

coupling equations. The influence of the AC filter and PLL, often

neglected, has been investigated. It is verified by simulations that

the influence of PLL on stability can be safely neglected. Neglect-

ing the filter bus results in a small error. A reduced model was

derived mathematically by neglecting three small time constants.

The advantages are a reduced set of differential equations, and the

possibility to increase the integration step size. Simulation of a

change in DC voltage reference has been simulated. It showed that

the full model responds satisfactorily and that the reduced model

simulates the lower order dynamics.

References

[1] IEEE Committee Report, Excitation system models for power system stability studies, IEEE Trans. Power apparatus and systems 100 (2) (1981) 494–509,

doi:10.1109/TPAS.1981.316906.

[2] IEEE Committee Report, Dynamic models for steam and hydro turbines in

power system studies, IEEE Trans. power apparatus and systems 92 (6) (1973)

[3]

1904–1915, doi:10.1109/TPAS.1973.293570.

S. Cole, R. Belmans, Transmission of bulk power: the history and applications of

voltage-source converter high-voltage direct current systems, IEEE Ind. Elec- tronics Magazine 3 (3), (2009) 19–24.

[4] C. Zheng, X. Zhou, R. Li, Dynamic modeling and transient simulation for VSC based HVDC in multi-machine system, in: Proc. International Conference on Power System Technology PowerCon 2006, 2006, pp. 1–7.

[5] S. Cole, R. Belmans, Modelling of VSC HVDC using coupled current injectors, in: Proc. IEEE Power and Energy Society General Meeting – Conversion and Delivery of Electrical Energy in the 21st Century, 2008, pp. 1–8. [6] K.R. Padiyar, N. Prabhu, Modelling control design and analysis of VSC based HVDC transmission systems, in: Proc. International Conference on Power System Technology PowerCon 2004, vol. 1, 2004, pp. 774–779,

doi:10.1109/ICPST.2004.1460096.

[7] O.A. Giddani, G.P. Adam, O. Anaya-Lara, K.L. Lo, Grid integration of a large off- shore wind farm using VSC-HVDC transmission system in parallel with AC submarine cable, in: Proceedings of the 44th International Universities Power Engineering Conference, University of Strathclyde, 2009. [8] C. Du, E. Agneholm, G. Olsson, Comparison of different frequency controllers

for a VSC-HVDC supplied system, IEEE Trans. Power Delivery 23 (4) (2008) 2224–2232, doi:10.1109/TPWRD.2008.921130. [9] R.E. Best, Phase Locked Loops, McGraw-Hill Professional, 2007. [10] F.M. Gardner, Phaselock Techniques, Wiley-Interscience, 2005. [11] A.H. Norouzi, A.M. Sharaf, Two control schemes to enhance the dynamic per- formance of the STATCOM and SSSC, IEEE Trans. Power Delivery 20 (1) (2005) 435–442, doi:10.1109/TPWRD.2004.839725(410) 2.

S. Cole, R. Belmans / Electric Power Systems Research 81 (2011) 967–973 973

[12] Matpower website, http://www.pserc.cornell.edu/matpower/. [13] MatDyn website, http://www.esat.kuleuven.be/electa/teaching/matdyn/. [14] J.M. Mauricio, A.G. Exposito, Modeling and control of an HVDC-VSC transmission system, in: Proc. IEEE/PES Transmission. Distribution Confer- ence and Exposition: Latin America TDC’06, 2006, pp. 1–6, doi:10.1109/

TDCLA.2006.311385.

[15] S. Cole, R. Belmans, Generalized dynamic VSC MTDC model for power sys- tem stability studies, IEEE Trans. Power Systems 25 (3) (2010) 1655–1662,

[16]

doi:10.1109/TPWRS.2010.2040846.

H. Liang, G. Li, P. Li, M. Yin, Analysis and design of H1 controller in VSC HVDC sys- tems, in: IEEE/PES Transmission and Distribution Conference and Exhibition:

Asia and Pacific, 2005, p. 6.

[17] L. Zhang, H.-P. Nee, Multivariable feedback design of VSC-HVDC connected weak AC systems, in: PowerTech 2009, 2009, pp. 1–8.

[18] G. Li, G. Ma, C. Zhao, G. Li, Research of nonlinear control strategy for VSC-HVDC system based on Lyapunov stability theory, in: Proc. 3rd International Confer- ence on Electric Utility Deregulation and Restructuring and Power Technologies DRPT 2008, 2008, pp. 2187–2191, doi:10.1109/DRPT.2008.4523773. [19] C. Mao, Z. Hu, J. Lu, D. Chang, S. Fan, Application of an optimal coordinated control strategy to VSC HVDC, in: Proc. IEEE PES Power Systems Confer- ence and Exposition PSCE’06, 2006, pp. 2141–2145, doi:10.1109/PSCE.2006.

296275.

[20] G. Terörde, Electrical Drives and Control Techniques, Acco, 2004. [21] C. Du, VSC-HVDC for industrial power systems, Ph.D. Thesis, Chalmers Univer- sity of Technology, 2007. [22] S. Cole, R. Belmans, MatDyn, a new Matlab based toolbox for power system dynamic simulation, IEEE Trans. Power Syst., in press.