You are on page 1of 25

Experience from introducing a

Remedial Action Optimisation at APG –


features and operational experience
Benoît BLETTERIE
PowerFactory TSO-User group meeting
03.05.2023
Agenda
1. Background and motivation
2. Requirements from operational
planning / features
3. Solution architecture and GUI
4. Operational experience
5. Conclusion

© Austrian Power Grid Österreich braucht Strom.


Agenda
1. Background and motivation
2. Requirements from operational
planning / features
3. Solution architecture and GUI
4. Operational experience
5. Conclusion

© Austrian Power Grid Österreich braucht Strom.


Austria on the way to 100% renewable generation

2050
Wind: 24 GW (2)
PV: 32 GW (2)

x3
2030
Wind: 9 GW (1)
PV: 12 GW (1)

2018 x5
Wind: 3,1 GW
PV: 1,4 GW
Wind
PV

2008 2018 2030 2050


(1) Source: „Stromzukunft Österreich 2030 – Analyse der Erfordernisse und Konsequenzen eines ambitionierten Ausbaus erneuerbarer Energien“; TU Wien; Mai 2017
(2) Source: „ ÖSTERREICHS WEG RICHTUNG 100 % ERNEUERBARE. EINE ANALYSE VON 2030 MIT AUSBLICK 2050. “; Energy Brainpool; Jänner 2019
4
Redispatch costs strongly increasing since 2013
237 days with redispatch (2022)
2021: 442 M€ | 2022: 736 M€ (total costs)
800
736

700

600

500
442
M€

400
346
319
300
243
203 204
200 157

100
13 22 23
2 1
0
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023

5
Main motivations for introducing a Redispatch
optimization tool
• Optimisation to limit
redispatch costs
• Reduce workload of op. planning
operators during DACF / IDCF Data from 2022

 Introduction of a self-developed
tool in 2018
 Go-Live in October 2022
 Development of a Proof of
Concept in 2019
(Unit Commitment @ PowerFactory)
 Development of a professional
RAO tool in 2020
© Austrian Power Grid 6
Agenda
1. Background and motivation
2. Requirements from operational
planning / features
3. Solution architecture and GUI
4. Operational experience
5. Conclusion

© Austrian Power Grid Österreich braucht Strom.


Requirements from operational planning perspective
Easy to use, flexible, fast 

• Feasible results: considering the time coupling of the optimization problem (minimum on/off time,
lead time, ...)
• Execution as automated as possible
• Input data automatically gathered, processed, validated / corrected
• User interaction: operator able to modify data
(e.g. cancel planned disconnections, correct gen. availability)
• Visualisation of input and output data and reporting - configurable plots
(e.g. generation profile, n-1 loading, redispatch volume, …)
• Flexible use by operator (fulfil about 15 use cases)
- Repeat calculations with modified parameters (e.g. modified redispatch, max. acceptable n-1 loading)
- Consider scenarios (e.g. scenario with lack of wind generation)

• Stable and fast enough to be used in operational processes (DACF and IDCF)
© Austrian Power Grid 8
Features and implementation aspects (1/2)
• Merge of (up to) 24 UCTE models into one project as a prerequisite to the optimization
(time coupling)
• Use of Virtual Power Plants (VPP) for generators in the 110kV network not modelled (GSK)
[not relevant anymore since the 110 kV network is now modelled]
• Thermal rating considered
• Remedial actions:
• Fully part of the optimisation:
- PSTs and transformers
- Redispatch
• Not part of the optimisation (decision based on comparison / operator experience):
- Cancelling of planned outages if they can be postponed
- Special switching states (e.g. 2 nodes operation): can be freely defined and considered
(hourly activation) in the calculation
• Constrains modelled as soft to ensure a solution is always found
© Austrian Power Grid 9
Features and implementation aspects (2/2)
• Verification after modification: calculated remedial actions can be modified and a new calculation
is performed (verification)
• Individual configuration of the maximal loading (safety margin) or selection between standard and
extended margins
• Project structure in PowerFactory (operation scenarios, network variations, ...) allows efficient
ex-post analyses (e.g. switch between before / after Redispatch, activation / deactivation of special
switching states...)

• Possibility to calculate only selected time stamps (without time coupling) for a faster first result
• Scenarios to take into account specific uncertainties to come in the near future

© Austrian Power Grid 10


Agenda
1. Background and motivation
2. Requirements from operational
planning / features
3. Solution architecture and GUI
4. Operational experience
5. Conclusion

© Austrian Power Grid Österreich braucht Strom.


Architecture overview
• Tool developed as a collaboration APG, FAW
(platform incl. GUI, data and process
management) and DIgSILENT (automation,
computing core)
• Communication between platform/client and
PowerFactory through webservice and
PowerFactory Connector Service
• Numerous scripts written by DIgSILENT in to
• “pre-process and import” the UCTE files • PowerFactory used as engine mode
(LF, CA, UC)
• prepare and trigger the calculations
• The operator does not directly interact with
• prepare and export the results PowerFactory
Parallelisation used when possible • PowerFactory used for ex-post analyses
(e.g. contingency analysis after remedial action (trouble-shooting), comprehensive analyses
optimisation) (e.g. analysis of planning criteria, critical
conditions and special remedial actions)
© Austrian Power Grid 12
Workflow-based process execution
• Main Workflows
• DACF / IDCF
• Auto / Manual calculation
• With default values / with additional safety margins

• Main blocks
• parametrization
• versioning
• sanity check
• data correction
• data modification (generators, PST/transformers)
• calculation
• report generation
• visualisation

Klassifizierung: INTERN 15
GUI – User Selection
Selection of:
• Business day and calculation time
• Process (DACF / IDCF)
• Model version
• Time range
• Contingency list to be used
(with e.g. exceptional contingencies)
• ...
• Special switching states to be applied

© Austrian Power Grid 16


Presentation of the results - loading (before / after RAO)

© Austrian Power Grid 17


Presentation of the results -PST and transformers
(relative tap position and change)

© Austrian Power Grid 18


Presentation of the results – Redispatch (aggregated)

© Austrian Power Grid 19


Agenda
1. Background and motivation
2. Requirements from operational
planning / features
3. Solution architecture and GUI
4. Operational experience
5. Conclusion

© Austrian Power Grid Österreich braucht Strom.


Managing forecast uncertainty

• Deviations between forecasted and real time n-1 loading can exceed 30%!
• Passed approach: keep remedial actions which can be quickly activated (e.g. PST taps)
in reserve.
However, keeping e.g. ±10 from ± 30 PST taps for uncertainties does not guarantee that they can be
effectively used...
• Main uncertainty factors:
• Model error (grid element status (on/off),
topological error, PST tap position...) can be avoided / limited
• Short-term trade (mainly pump storage hydro)
Explicitly
(dedicated
• Forecast of renewable generation (Wind, PV) scenario)
• Load forecast (incl. geographical distribution) shall be taken into account
• Regulating power Implicitly
(safety
• ...
margin)
© Austrian Power Grid 21
Using larger safety margins?
• Study performed for about 20 interesting days (complex situation with large Redispatch amount)
• with default settings (e.g. max. loading of 97%) and default PST range (e.g. ±5 taps)
• with extended margins (e.g. max. loading of 80%) and extended PST range (all ±30 taps)

Despite what we know (and don’t know


from uncertainties),
 Results as expected with some
learnings:
• Normal safety margin needed due to
calculation uncertainty (when starting from e.g.
more than 150% loading) and other effects
• When using the full PST range (for the extended
safety margins), flows totally reverse and some
new overloads can appear (constraint filtering)

© Austrian Power Grid 22


Managing forecast uncertainty – example for wind
power (explicitly considered = with dedicated scenario)
 Currently under study, not yet systematically in use

Example of criterion (not yet finalised):


>> if Pforecast > 1500 MW, a shortage of 750 MW should
be considered (95% confidence)
© Austrian Power Grid 23
Lessons learnt
• The quality of the input data is very important.
• Network models: a few UCTE-specific issues appeared and have been solved (most of the not
relevant (e.g. convergence problems in specific islands...))
• Inconsistencies between “offline” and “daily” data has to be managed. For example:
 Schedule > Pmax
 daily data provided by power plant operators not according to the specifications
(e.g. in order to determine possible “must-run” conditions)
 min on/off time not respected due to tests operation or due to differences between
contractual and technical capabilities
 ...
• The execution time is for some days (in which a decision must be quickly taken) too large:
• total execution time 30-45 min (unit commitment calculation time about 15 min)
• process optimisation under work to reduce calculation time (parallelisation used where possible)
• study on how valuable results from a DC calculation can be start soon
• Overall, good success rate of the tool and in particular of PowerFactory
© Austrian Power Grid 24
Agenda
1. Background and motivation
2. Requirements from operational
planning / features
3. Solution architecture and GUI
4. Operational experience
5. Conclusion

© Austrian Power Grid Österreich braucht Strom.


Conclusion

• Time consuming preparation work in the development project and parallel run:
• data preparation and validation – high data quality standard required for
operational use!
• parametrization, testing and validation
• Training of operators is important to ensure that the tool is well understood and to
ensure that it can fully support operations (fast, reliable, ...)
• RAO is a complex process since many operational aspect have to be considered
(cancelling planned disconnections, special switching states, safety margins...).
The final decision is still taken by operators, who report significant deviations from
the results
• Ensuring a fast execution is of prime importance – work in progress
• Valuable experience for APG in the context of the introduction of an international
process (ROSC)

© Austrian Power Grid 26


Thank you for your attention!
Dipl.-Ing. Dr. techn. Benoît Bletterie
Power Grid Control – Main Control Center
Operational Planning
Austrian Power Grid AG
T: +43 (0) 50 320 - 53291
M: +43 (0) 664 883 42 832
E: benoit.bletterie@apg.at
www.apg.at

© Austrian Power Grid 27

You might also like