You are on page 1of 1

Contact: Xavier Roboam(xavier.roboam@laplace.univ-tlse.

fr)
Laboratoire Plasma et Conversion d’Energie (LAPLACE)
Unité mixte de Recherche 5213 INPT / UPS / CNRS
BP 7122 - 2 rue Camichel - 31071 TOULOUSE Cedex 7 - France
INPT INPT--LAPLACE Invol vement in MOET project LAPLACE Invol vement in MOET project
WP 3.22: Power conversion
WP 7.24: Quality & Stability studies
3 Electrical power centers including
power transfer equipments
MAPFC test
bench
Inside a DCPFC
New concepts for power management
New power management concepts
Leading towards new architectures
Côté Côté Droit Droit
G G
HVDC D HVDC D
Charge Charge
DD
LLLL
T T11
T T22
TT33
TT44
GG
Côt é Côté Gauche Gauche
Charge Charge
G G
HVDCG HVDC G
GG
Côté Côt é Gauche Gauche
Charge Charge
GG
Côt é Côté Droi t Droit
GG
HVDC G HVDC G HVDC D HVDC D
Charge Charge
DD
DCPFC DCPFC
Temps ( s) Temps (s)
2.78 2. 8 2. 82 2.84 2.86 2. 88 2.9 0
5
10
15
20
25
30
35
40
45
50
Courants fournis par les générateurs (A)
Courants fournis par les générateurs (A)
Droite Droi te
Gauche Gauche
T T11 T T11
Essai expéri mental Essai expéri mental Essai expéri mental Essai expéri mental
Temps ( s) Temps (s)
2.78 2. 8 2. 82 2.84 2.86 2. 88 2.9 -25
-20
-15
-10
-5
0
5
10
15
20
25
Courants sortants du DCPFC (A)
Courants sortants du DCPFC (A)
Droite Droite
Gauche Gauche
T T11 T T11
Essai expéri mental Essai expéri mental Essai expéri mental Essai expéri mental
GG
Côté Côt é Gauche Gauche
Charge Charge
GG
Côt é Côté Droi t Droit
GG
HVDC G HVDC G HVDC D HVDC D
Charge Charge
DD
DCPFC DCPFC
Time Time
PPload
load
Time Time
PPload
load
Temps (s) Temps (s) Courant fourni par le DCPFC pendant la perte (A)
Courant fourni par le DC
PFC pendant la perte (A)
0 0. 05 0. 1 0.15 0.2 0
5
10
15
20
25
30
35
40
Essai expér iment al Essai expér iment al Essai expér iment al Essai expér iment al
Direct Current Power Flow Controller Direct Current Power Flow Controller
Case 1:bidirecti onal Case 1:bidirecti onal
controlled power controlled power
transfer transfer
Case 2: HVDC Case 2: HVDC voltage voltage
regulation regulation
Côté Côté Droit Droit
G G
HVDC D HVDC D
Charge Charge
DD
GG
Côt é Côté Gauche Gauche
Charge Charge
G G
HVDCG HVDC G
TTG1 G1
TTG4 G4
TTG2 G2
TTG5 G5
T T G3 G3
T T G6 G6
T T D1 D1
T T D4 D4
TTD2 D2
TTD5 D5
TTD3 D3
TTD6 D6
MAPFC: Mixed function for Actuation & MAPFC: Mixed function for Actuation &
Power Flow Control Power Flow Control
Twoindependent functions:
Actuator
Power flow controller
Temps (s) Temps (s)
Courant circulant dans la phase 1 (A)
Courant circulant dans la phase 1 (A)
0 0. 01 0.02 0. 03 0.04 0.05 -20
-15
-10
- 5
0
5
10
15
20
Imoyen Imoyen =0 A =0 A
Essai expérimental Essai expérimental Essai expérimental Essai expérimental
Temps (s) Temps (s)
Courant circulant dans la phase 1 (A)
Courant circulant dans la phase 1 (A)
0 0. 01 0.02 0. 03 0.04 0.05 -20
-15
-10
- 5
0
5
10
15
20
Imoyen Imoyen =5 A =5A
Essai expéri mental Essai expéri mental Essai expéri mental Essai expéri mental Temps (s) Temps (s )
Courant circulant dans la phase 1 (A)
Courant circulant dans la phase 1 (A)
0 0.01 0. 02 0.03 0. 04 0.05 -20
-15
-10
- 5
0
5
10
15
20
Imoyen Imoyen ==--5 A 5 A
Essai expérimental Essai expérimental Essai expérimental Essai expérimental
Actuation function only
(
(
(
¸
(

¸
u
+
(
(
(
¸
(

¸

(
(
(
(
(
(
¸
(

¸

o
o +
o +
=
(
(
(
¸
(

¸

u
u
u
0
0
aimant
h i
q i
d i
.
s L 0 0
0 s L sp L
2
3
0
0 0 s L sp L
2
3
h
q
d
“Flux –current” relations in the Park frame
(d,q) axis : Actuationfunction
(h) axis: Power flow control function
Left side Right side
Temps (s) Temps (s)
T
e
n
s
io
n
s
d
u
b
u
s
e
n
d
é
f
a
u
t
(V
)
T
e
n
s
io
n
s
d
u
b
u
s
e
n

d
é
fa
u
t
(
V
)
Essai expérimental Essai expér imental
0 0.05 0.1 0.15 0.2
0
100
200
300
400
500
600
Avec la fonction Avec la fonction
«« No No- -Break Break »»
Sans la fonction Sans la fonction
« « No No- -Break Break »»
Load : 8 kW
Test case: Test case:
This test shows what happened during a
reconfiguration. Without bus control through a
MAPFC (red curve), the bus bar is not able to
maintain the voltage on the bus bar even with a
lowpower load(8 kW). Contrarily, controllingthe
DC bus bar from a MAPFC in case of
reconfiguration allows to maintain this voltage
(blue curve)
The bus capacitor in this case is 4,7mF.
HVDC 1 HVDC 2
HVDC 3
Need to control power flows
Generator Generator
POR POR
The meshed network: a whole HVDC
dedicated test bench
Model building principle Multiple MCUs interaction study
Stability criterion: Routh-Hurwitz Optimization algorithms: an introduction
Generator Generator
Rectifier Rectifier
C Cbus bus
DCbus DCbus
Filter 1 Filter 1 Load 1 Load 1
Filter 2 Filter 2 Load 2 Load 2
Filter 3 Filter 3 Load 3 Load 3
Studied systems: HVDC Networks Studied systems: HVDC Networks
Focus on HVDC Networks with one or several loads.
Load = Drive Load = Drive
PMSM PMSM
Inverter Inverter
Load = Dri ve Load = Dri ve Fi l ter Fi l ter
Inver ter Inver ter
PMSM PMSM
DC Bus DC Bus
Automatic building of network Automatic building of network
models : models : an automated method based
on a Maple package is suggested. It
is a very fast and convenient solution
allowing the model set upin several
steps:
Using intermediate subsystems…
… models of complex architectures
can be automatically set up.
Fromthe transfer function denominator of the whole
systemmodel, stability studies are performed using
the Routh-Hurwitz criterion. It allows to:
know the state of a sized system(stable or unstable)
give conditions on systemparameters leading to
state design(e.g. building stability abaci)
suggest conditions on specification parameters
(i.e. bandwi dth, damping factor…) in order to ensure
whole systemstability
Stability abaci help the
designer in parameters
sizing.
They also allow to study
network behaviour
(e.g. interaction between
loads according to their
states - wire lengths,
equipments number and
powers effects on system
stability…)
A network associating two loads is considered.
Load n°2 filter is sized separately following its
stability abacus.
1: It is sized
stable with goods
stability margins
2: It is sized
around its stability
limit
In both cases, Load
n°1 filter stability is
studied when the two
loads are connected.
Case 1: Stable l oad additi on Case 1: Stable l oad additi on
In this case, found
stability limits are
roughly unchanged,
even if more than
one load are
added on the bus.
Stability brought by
those stable loads
slightly increases
the stable domain.
Case 2: Additi on of a l oad si zed around i ts stabil ity l imi t Case 2: Additi on of a l oad si zed around i ts stabil ity l imi t
In this case, the more load n°2 filter is sized “unstable”, the smaller the stability domain.
This result well
illustrates the possible
interactions between
loads connected on the
same bus, and the
relevance of these
methodologies.
Quality criteria: filtering and damping factor Quality criteria: filtering and damping factor
Stability condition remains insufficient to give a
complete parameters
sizing (stable domain).
Quality criteria are thus
introduced to complete
this sizing:
filtering condition
damping factor

Optimisation algorithms: Optimisation algorithms:
They are introduced in order to increase
the number of handled parameters.
Moreover, they allow to minimize the
energy storage in inductive and capacitive
components through
the convergence
criterion.
Stability and quality criteria are
translated into constraints. Two
kinds of algorithms are used:
An algorithmbased on
gradient convergence;
An crowding based genetic
algorithm(RTS).
DCPFC
Other MOET* partners involved
UPNA
* MOET is a European Project co-funded by the European
Commi ssion within the Sixth Framewor k Programme
Aut omati c desi gn of energy management equati ons by mean of Graphs
theory : seeki ng maxi mal fl ow (¢ = power) wi th mi ni mum cost (energy
effi ci ency)
Desi gn rul es (state of contact ors) based on an expert system approach
Network management
interface
The LAPLACE laboratory test capabilities now
allows HVDC equipment connexion and
validation.
The dSpace-based supervision and
management platform permits an easy
evolutionregarding HVDC equipments.
The complete generation and distribution
chain includes 3 generators (one based on
400-800Hz AC generator) and multiple
programmable loads. Possibilities also exist in
reconfigurationat network level.
New architecture : New architecture :
The idea to use MAPFC
concept on ECS machines
helps in finding a “HVDC
power no-Break” archit-
ecture.
Each machine can use the
homopolar current to
regulate a HVDC bus bar
during reconfiguration.
Without this configuration,
the important load
consumption will lead to an
important voltage drop.
1 11 1
2 22 2 3333
4 44 4
Left EPC Left EPC Left EPC Left EPC Right EPC Right EPC Right EPC Right EPC
Rear EPC Rear EPC Rear EPC Rear EPC
Whol e gener ati on Whol e gener ati on Whol e gener ati on Whol e gener ati on
DCPFC 1 DCPFC 1 DCPFC 1 DCPFC 1
D
C
P
F
C
2
D
C
P
F
C
2
D
C
P
F
C
2
D
C
P
F
C
2 D
C
P
F
C
3
D
C
P
F
C
3
D
C
P
F
C
3
D
C
P
F
C
3
¢(1) =P
gen
G
¢(2) =Pgen
D
¢(3) =Pgen
A
¢(4) =PDCPFC
1
¢(5) =PDCPFC
2
¢(6) =PDCPFC
3
¢(7) =Pload
1
¢(8) =Pload
2
¢(9) =Pload
3
1 2
3
4
5
6
5 55 5
Whol e consumption Whol e consumption Whol e consumption Whol e consumption
7
8 9
The generic description of the
network in graph theory language
allows a complete description of
power transfer s in the network.
As a result, the use of graph theory
algorithms also helps in managing
power references through power
flowcontrollers (DCPFC, MAPFC).
Initial Initial
facts f acts
Domain Domain
rules rules
Facts Facts
Rules Rul es
Inference engine Inference engine
Inference rules Inference rules
Newrules Newrules
New facts Newf acts
In order to avoid problems during contactor
logic writing, a novel approach based on
expert rules definition is implemented in the
supervisionsystem.
With only a simplified set of rules, the
supervision systemis allowed to configure
the network contactors in order to optimize
electrical distribution.
Load model: Equivalent admittance Load model: Equivalent admittance
Drives are expressed as equivalent admittances,
which give their frequency behaviors around an
operating point.
Ydrive expression thus
obtained can be introduced
onthe whole systemmodel.
UPNA