Error 503

Transcription

Error 503
NEW SOLUTIONS FOR THE AUTOMATION OF LARGE INDUSTRIAL DISTRIBUTION NETWORKS
Luc HOSSENLOPP
ALSTOM T&D Protection and Control
Avenue de Figuières 34 975 LATTES Cedex FRANCE
[email protected]
Large industrial companies are now increasingly
developing their own generation facilities as a
consequence of the general deregulation process in the
electrical industry. The example of an oil and gas
refinery is one of the most extreme, where the refined
oil (or gas) is used to feed some boilers in order to
produce the energy needed to perform the primary
refinement task. The next step consists in optimising
the exchanges with the main (national) electricity grid
depending on the cost of energy produced by the
internal generators and the utility, see figure 1.
Selling Oil & Gas
Feeding the
loads (motors, ..)
ELECTRICITY
DISTRIBUTION
Powering
the site
Selling
electricity
OIL & GAS
PROCESS
Powering the site
POWER
MANAGEMENT
Burning
Oil & Gas
ELECTRICITY
GENERATION
Optimising energy
production & use
EXTERNAL
GRID
Figure 1: Electricity management in an oil & gas process
This leads to an evolution of the site automation: the
process control system is assisted by an electrical
power management system in order to cope with all
electricity related tasks.
The process control system focuses on the optimisation
of the primary industry objective, for instance oil
refining.
The electrical power management system copes with
the complete power generation and distribution needs.
It provides the unification of traditionally distinct
automation and analysis islands:
- Generator protection and control.
- Substation. protection and control.
- Motors and drives protection and control.
- Industrial network management.
- Power optimisation with the external grid.
- Power quality and network fault analysis.
The aim of this paper is to illustrate the key features of
such new solutions in the area of electrical power
management system.
The proposed solution relies on a series of simple
concepts that bring tangible benefits to the user:
- integration of different automation domains,
- distributed processing,
- extensive use of simulation capabilities.
Various technological innovations contributes to the
added value offered by the system:
- Fast and fine load shedding algorithms.
- Co-ordinated
frequency/active
power
and
voltage/reactive power regulations.
- Condition monitoring of transformers and
switching devices.
- Power quality monitoring devices.
- Economical optimisations.
The integration does not simply mean the
interconnection of different devices but added value
through better automation, improved analysis features
and easier maintenance. Fast and fine load shedding
avoids the oversizing of generation capacity (up to a
factor 2) for instance. Accurate calculation of energy
cost for the manufacturing of a given product is another
example.
This modern automation technology can be generalised
to several dispersed generation plants in the future. The
mix of automation and economical functions will
become more and more important for asset managers.
Further enhancements will progressively introduce
features traditionally reserved to utility control centres
as the industrial networks become more complex. This
will include features such as fault isolation and
reconfiguration, state estimator, etc.
NOUVELLES SOLUTIONS POUR L’AUTOMATISATION DES GRANDS RESEAUX DE DISTRIBUTION
INDUSTRIELS
Luc HOSSENLOPP
ALSTOM T&D Protection and Control
Avenue de Figuières 34 975 LATTES Cedex FRANCE
[email protected]
Les grandes entreprises industrielles développent
désormais de plus en plus leurs propres moyens de
production d’énergie, en conséquence du processus de
dérégulation dans l’industrie électrique. L’exemple
d’une raffinerie pétrolière est un cas extrême, où le
pétrole raffiné sert à alimenter les chaudières
produisant l’énergie nécessaire au raffinage lui-même.
L’étape suivante consiste à optimiser les échanges avec
le réseau électrique externe (éventuellement national),
en fonction des coûts de l’énergie produite par les
générateurs internes et de ceux du distributeur (voir
figure 1).
Selling Oil & Gas
Feeding the
loads (motors, ..)
ELECTRICITY
DISTRIBUTION
Powering
the site
Selling
electricity
OIL & GAS
PROCESS
Powering the site
POWER
MANAGEMENT
Burning
Oil & Gas
ELECTRICITY
GENERATION
Optimising energy
production & use
EXTERNAL
GRID
Figure 1: Gestion de l’énergie dans un contexte de pétrochimie
Ceci conduit à une évolution de l’automatisation du
site: le contrôle de process est assisté par un système de
gestion de la puissance, en charge de toutes les tâches
liées à l’électricité.
Le contrôle de process optimise la production primaire
de l’industrie, par exemple le raffinement de pétrole.
Le système de contrôle de puissance s’occupe de
l’ensemble des besoins de génération et distribution de
puissance. Il permet de réunir des îlots
d’automatisation et d’analyse traditionnellement
séparés:
- Protection et contrôle des générateurs.
- Protection et contrôle des postes électriques.
- Protection et contrôle des moteurs.
- Gestion du réseau industriel.
- Optimisation des échanges avec le réseau externe.
-
Analyse de la qualité de l’énergie et des défauts
électriques.
L’objectif de cet article est d’illustrer les points clés de
cette nouvelle solution de système de gestion de la
puissance.
La solution proposée s’appuie sur un ensemble de
principes simples qui apportent des bénéfices tangibles
à l’utilisateur:
- intégration de domaines d’automatisation distincts,
- traitements distribués,
- utilisation poussées des possibilités de simulation.
Plusieurs innovations technologiques contribuent à la
valeur ajoutée apportée par le système:
- Des algorithmes de délestages rapides et fins.
- La régulation coordonnée en fréquence/puissance
active et tension/puissance réactive.
- La surveillance d’état des transformateurs et
disjoncteurs.
- Les outils de mesures de la qualité de l’énergie.
- Les optimisations économiques.
L’intégration
ne
signifie
pas
seulement
l’interconnexion des différents équipements mais une
valeur ajoutée résultant d’une meilleure automatisation,
d’analyses plus pertinentes et de maintenance plus
facile. Ainsi le délestage rapide et fin permet d’éviter le
surdimensionnement des capacités de génération
(jusqu’à un facteur 2). Le calcul précis des coûts de
l‘énergie utilisée pour la fabrication d’un produit est un
autre exemple.
Cette technologie d’automatisation peut être
généralisée dans le futur. La cohabitation de fonctions
d’automatismes et financières deviendra de plus en
plus importante pour les gestionnaires dans le futur.
Les évolutions suivantes consisteront à introduire
progressivement des caractéristiques
réservées
traditionnellement aux centres de téléconduite
régionaux, en fonction de la complexité des réseaux
industriels. Cela intégrera par exemple des possibilités
de détection et reconfiguration du réseau, d’estimation
d’état, etc.
NEW SOLUTIONS FOR THE AUTOMATION OF LARGE INDUSTRIAL DISTRIBUTION NETWORKS
Luc HOSSENLOPP
ALSTOM T&D Protection and Control
Avenue de Figuières 34 975 LATTES Cedex FRANCE
[email protected]
Abstract
Deregulation of electrical sector is changing the
landscape of the distribution systems: large industries
do not only rely on the utility grid but produce
electricity for their own and even sell the remaining
energy to the grid. One example is the petro-chemical
refinery where it may prove to be more economical to
transport energy rather than fuel. This new situation
leads to the introduction of specific automation
compared to the traditional approach. Combining this
with constraints for higher functional integration and
use of modern communication technology brings a new
generation of automation systems. The architecture,
features and benefits of the system performing these
functions are described in this paper.
1
INTRODUCTION
Large industrial companies are now increasingly
developing their own generation facilities as a
consequence of the general deregulation process in the
electrical industry. The example of an oil and gas
refinery is one of the most extreme, where the refined
oil (or gas) is used to feed some boilers in order to
produce the energy needed to perform the primary
refinement task. The next step consists in optimising
the exchanges with the main (national) electricity grid
depending on the cost of energy produced by the
internal generators and the utility, see figure 1.
This leads to an evolution of the site automation: the
process control system is assisted by an electrical
power management system in order to cope with all
electricity related tasks.
The process control system focuses on the optimisation
of the primary industry objective, for instance oil
refining.
The electrical power management system copes with
the complete power generation and distribution needs.
It provides the unification of traditionally distinct
automation and analysis islands:
- Generator protection and control.
- Substation. protection and control.
- Motors and drives protection and control.
- Industrial network management.
- Power optimisation with the external grid.
- Power quality and network fault analysis.
The aim of this paper is to illustrate the key features of
such new solutions in the area of electrical power
management system.
2
The proposed solution relies on a series of simple
concepts that bring tangible benefits to the user:
- integration of different automation domains,
- distributed processing,
- extensive use of simulation capabilities.
2.1
Selling Oil & Gas
Feeding the
loads (motors, ..)
ELECTRICITY
DISTRIBUTION
Powering
the site
Selling
electricity
OIL & GAS
PROCESS
Powering the site
POWER
MANAGEMENT
Burning
Oil & Gas
ELECTRICITY
GENERATION
Optimising energy
production & use
EXTERNAL
GRID
Figure 1: Electricity management in an oil & gas process
CONCEPTS
Integration of different automation domains
The integration of Substation Control System (SCS),
Motor Control Centre (MCC) and Energy Management
System (EMS) enable to:
- Efficiently share the information. Avoiding a
duplication of information reduces the costs
(sensors, wiring, etc.). It also leads to better timely
decision through advanced automation (see further
fast load shedding example), on-line adaptive
setting and expert analysis.
- Cross-fertilise the best practices of each domain
such as 1 ms time tagging sequence of events for
better post mortem analysis, IEC 1131-3
automation features, simulation capabilities, high
electro-magnetic compatibility (EMC), etc.
- Simplify the user training and operation through a
unification of the human machine interfaces and
automation computers. This also reduces the
associated maintenance costs.
2.2
-
Extend the life of network components such as
transformers and circuit breakers using newly
developed condition monitoring technologies,
Optimise the use of generators on a medium term
basis vs. external grid electricity price and
associated contracts.
Help optimising the network components and
protection setting using off-line network analysis.
Permit on-line real time test of worst case scenario,
for example “what if this generator fails ?” in order
to improve the user confidence.
-
Distributed system
The power management system is distributed in the
factory, i.e. some PLC (Programmable Logic
Controllers) are located in the substations, some are
dedicated to drives and other to generators. They are
interconnected through high speed communication
networks in order to permit fast automation between
PLCs.
-
Time horizon
The distribution reduces the wiring: as an example a 23
km of 7 electrical cores cable was replaced by a 1 km
of 2 fibre optic cable.
PLC are directly integrated with the switchboards: this
eliminates specific cubicles and provides easy
installation at site since only a fibre optic connection is
needed (“pluggable switchgear”).
Off-line
Network analysis
1 day
Unit commitment planning
1s
On-line worst case simulation
50 ms
Fast load shedding
Figure 3: Simulation time horizons
The coupling of the generators with the grid is an
example of fast distributed automation: the PLC
located at the grid connection point will send order to
all generators controllers (based on the phase
differences measured between grid and generator
voltages) in order to synchronise them – see figure 2.
PLC
Grid voltage vs.
Internal voltage
comparison
500 m
Lower/raise requests
PLC
Generator control
(U/Q, f/P)
Figure 2: Example of distributed automation
Conversely when the industrial network becomes
islanded due to a grid fault for instance the generator
mode shall be changed in order to elect an isosynchronous master while the other generators will
keep their droop mode. This correspond to a fast
transfer of information between the incoming feeder
(that has detected the grid fault) and the generators.
2.3
Extensive use of simulations
The combination of powerful PLCs together with
various automation domain expertise and integration
enable a new generation of simulation facilities. They
are extensively used at different time horizons (see
figure 3) in order to:
- Decrease the reaction time as requested by new
automation applications.
3
3.1
EXAMPLES
OF
INNOVATIONS
TECHNOLOGICAL
Traditional load shedding scheme
Compared to conventional distribution networks large
industry distribution networks differ by the size and
generation capacities, therefore sensitivity to
miscellaneous disturbances. In case of a remote site the
external grid can be weak or even none existent: the
reaction (tripping) time for load shedding in case of a
loss of a generator shall be less than 200 ms in order to
keep a satisfactory system performance.
Conventional
under-frequency
load
shedding
algorithms (see figure 4) are not satisfactory since they
proceed by successive approximations: if shedding one
load (or a series of loads) is not sufficient then after a
predefined time-out and considering another frequency
threshold a new load shedding will occur. This is
incompatible with the above constraint.
f (Hz)
fN
Disturbance
f L1
f L2
P(MW)
t (s)
Load priority 1
New load
balance
Load priority 2
t (s)
Figure 4: conventional under-frequency load shedding
3.2
Fast load shedding in case of a generator loss
The loss of a generator is extremely critical for the
application. The fast and fine load shedding algorithm
is based on the following principles:
- The amount of load to be shed is calculated per
generator lost based on the actual power delivered
by the generator. This simulation permits a single
step to restore the supply/load balance instead of a
series of steps (with the traditional underfrequency approach) – refer to figure 5.
- The list of loads to be shed (per generator) is
dynamically defined by calculating the number of
loads of the lowest priority that will correspond to
such amount.
- The load priorities can be modified dynamically in
order to better follow the evolution of the
environment.
3.3
3.3.1
Other load shedding features
Underfrequency load shedding revisited
In order to cover other types of disturbances a
sophisticated version of the under-frequency load
shedding algorithm has been developed.
The principle is to shed a pre-defined load amount
instead of a fixed list of feeders if the frequency
reaches a low threshold in order to reach the balance in
one step.
Like for the previous algorithm the list of feeders is
calculated by summing the loads of the lowest priority
that will correspond to such amount. These values are
calculated per island and are loaded into each computer
managing feeders (see figure 7).
f (Hz)
Disturbance
fN
Power to be
shed
(examples)
Time delay
Feeder 1
Feeder 2
Feeder 3
…..
Feeder N
t (s)
P(MW)
Load shedding
(single step)
New load
balance
t (s)
Figure 5: Fast and fine load shedding
The detail algorithm is slightly more complex since:
- It shall take into account the possible islanded
subnetworks of the installation. The algorithm is
instantiated for each island based on the network
topology.
- It shall take into account the spinning reserve
available and needed in each island.
- It is implemented in a distributed control system.
An algorithm has been developed in order to cope
with the performances and have a good security:
each computer receives periodically the list of
load to shed in the case of the loss of each
generator (see figure 6). Only the generator loss
information needs then to be transmitted in order
to shed quickly the relevant loads. Once a
computer receives notification of generator loss it
will decide locally which feeder to shed.
Feeders
by
priority order
Feeder 1
Feeder 2
Feeder 3
…..
Feeder N
Loss of
G1
x
Loss of
G2
x
Loss of
S3
x
x
x
x
Figure 6: Example of a dynamically pre-set table in each distributed
computer in case of a generator loss
Group 1
f<f1
Group 2
f<f2
Group 3
f<f3
6MW
10MW
20MW
100ms
x
x
300ms
500ms
x
x
Figure 7: Example of a dynamically pre-set table in each distributed
computer for advanced under-frequency load shedding
3.3.2
Momentary load shedding
A momentary overload shedding is using the spinning
reserve of a sub-network reaching an on line
configurable value. When the power demand goes over
this threshold, the load management system activates
an alarm and after a time delay, begins to shed the
feeders.
The time delay has to be sufficient for an operator to
have an action before automatic load shedding. The
feeders are tripped by priority order (priorities are the
same as for the loss of generator), with another time
delay between each trip, until the power demand goes
back under the threshold.
3.4
Regulation facilities
Regulation of voltage and active power on one hand,
and of frequency and reactive power on the other hand,
clearly illustrate the benefits of merging various
automation features. Figure 8 synthesises the
relationships between these electrical network
parameters and the primary equipment.
Transformer
Capacitor
Primary
Secondary
Tertiary
Frequency
Active
power,
Busbar
voltage
Busbar
voltage
Power
factor,
busbar
voltage
Reactive
power
Exchanges with
external grid
Other algorithms based on temperature and load
current are used to calculate the ageing of the
transformer and decide the acceptance of temporary
overload.
1000
Stufung von Stufe 2 nach 3
Stufung von Stufe 6 nach 7
W
Reactive
power
600
Reactive power
import
reduction from
an external grid
Leistung
Device/
Control type
Generator
400
200
0
Figure 8: Hierarchy of regulation functions managed by the electrical
power management system
A regulation function is split between the different
primary equipment dispersed in the factory, therefore a
strong co-ordination is needed including priorities
between the equipment able to perform the function.
For instance in an islanded network the generators will
be in droop mode: this will automatically counteract
output power swings caused by frequency variations. If
the frequency falls below a minimum then an order will
be sent to the generators: they will increase their output
powers and shift their droop curves (see figure 9).
Frequency
New droop
curve
f nom
fmin
P1
P2
0
1
2
3
4
s
6
Zeit
Figure 10: Transformer power consumption of motor drive
For HV circuit breaker the monitoring of SF6 density
through 4-20 mA sensors instead of all or nothing
relays enable to predict when a refill will be necessary.
Analysis of the travel curve of the circuit breaker by
comparing it to a fingerprint contributes to an early
detection of a (future) failure.
3.6
Power quality
Power quality devices are used to:
- Monitor contractual interfaces – either when
importing or exporting energy.
- Correlate power quality disruption with process
trouble.
- Monitor a set of parameters linked to the
technology of the generating device (flicker,
unbalance, inter-harmonics) – see figure 11.
- Help decide the use of corrective device by
monitoring disturbing loads/harmonic pollution
together with network simulation facilities.
Power
Figure 9: Generators characteristic in droop mode
A co-ordination of the generators is needed to
minimise power exchanges between busbars in order to
improve the power stability. This is achieved through
the fast data exchanges data (new power setting)
between devices managing the generators controllers.
3.5
Condition monitoring
Condition monitoring is a series of new functions
available to provide preventive maintenance of
HV/MV equipment (such as transformers or circuit
breakers) and extend their life duration.
Figure 10 illustrates the analysis of the power
consumption of the motor driving the tap changer. The
sequence is divided into elementary sub-sequences
corresponding to individual mechanical steps and
enabling the early detection of a failure.
Figure 11: typical power quality follow up
3.7
Economical optimisations
The optimisation of energy cost is calculated on an
horizon of 1 day to 1 week. It takes into account:
- various start-up and generation costs (fuels,
temperature, …),
- most economical start up and shutdown times,
- spinning reserve requirements,
4
4.1
system load forecast,
contractual condition for energy transactions with
the external grid.
SOLUTION OVERVIEW
System architecture
4.2
Communication with the process control
system
The process control system and the electrical power
management system are communicating through an
open interface (typically Ethernet OPC). The
communication with the loads (such as motors) and
generator controllers is typically done by MODBUS.
The system architecture is designed in three layers with
increasing horizon processing (see figure 12):
- The IED (Intelligent Electronic Device) provides
reflex automation for a single generator
(controller) or circuit breaker (protection). It also
performs advanced monitoring functions for power
quality or condition monitoring.
- The PLCs (Programmable Logic Controller) coordinate the actions of various IEDs in order to
provide fast distributed automation. For instance
load shedding total time of less than 100ms.
- The HMI (Human Machine Interface) presents
information to the users such as simulation,
training or analysis. It also processes medium term
regulation, for instance unit commitment.
Process control system
OPC
Electrical power
management system
MODBUS
Grid
Generators
Loads
Figure 13: Environment of the Electrical power management system
HMI level:
Simulation and
medium term
regulation
Unit commitment
Disturbance analysis
PLC level:
Fast automation
between equipment
Load shedding
Parallel
transfo/generator
t
IED level:
Reflex automation on
single equipment
Protection
Generator controller
Figure 12: Functional layers
The IEDs and PLC are located close to the sensors and
actuators – typically in the switchboards. Therefore a
high resistance against elecro-magnetic influences
(such as 4 kV fast transient burst) is a must.
The communication with IEDs uses slow legacy
protocols (<64 kbps) such as MODBUS while
communication between PLCs and the HMI benefits
from high speed deterministic and redundant network
(> 3 Mbps). Communication between HMIs is based
on Ethernet (10/100 Mbps).
5
CONCLUSION
The solution presented in this paper offers to the user a
very high integration of all functions related to
electrical power management: generation, distribution,
regulation, simulations and analysis.
The integration does not simply mean the
interconnection of different devices but added value
through better automation, improved analysis features
and easier maintenance. Fast and fine load shedding
avoids the oversizing of generation capacity (up to a
factor 2) for instance. Accurate calculation of energy
cost for the manufacturing of a given product is another
example.
This modern automation technology can be generalised
to several dispersed generation plants in the future. The
mix of automation and economical functions will
become more and more important for asset managers.
Further enhancements will progressively introduce
features traditionally reserved to utility control centres
as the industrial networks become more complex. This
will include features such as fault isolation and
reconfiguration, state estimator, etc.
BIOGRAPHY
Luc HOSSENLOPP was born in Grenoble, France, on
January 7, 1964. He graduated from ENSTA (Ecole
Nationale de Techniques Avancées) and joined
ALSTOM in 1987. He is currently Control System
Marketing Director. He is member of IEC and CIGRE
committees.