SEVENTH FRAMEWORK PROGRAMME
THEME [ENERGY.2011.7.3-1]
[Innovative tools for the future coordinated and stable
operation of the pan-European electricity transmission
system]
Project Deliverable
Deliverable D 7.4
“Workshop results on solutions
for maximising power transits”
Project acronym: UMBRELLA
Project full title:
Toolbox for Common Forecasting, Risk
assessment, and Operational Optimisation
in
Grid
Security
Cooperations
of
Transmission System Operators (TSOs)
Grant agreement no.:
282775-2
Research Project UMBRELLA
Page 2
Contents
1 INTRODUCTION ... 3
2 ORGANIZATION AND STRUCTURE OF THE WORKSHOPS ... 4
2.1 Workshop on Optimized Solutions ... 4
2.2 Workshop on System State Modelling and Toolbox Design ... 5
3 SUMMARY OF THE WORKSHOP PRESENTATIONS ... 7
3.1 Workshop on Optimized Solutions ... 7
3.1.1 WP3.1 Presentation and discussion ... 7
3.1.2 WP3.2 Presentation and discussion ... 7
3.1.3 WP3.3 Presentation and discussion ... 8
3.1.4 WP4 Presentation and discussion ... 8
3.2 Workshop on System State Modelling and Toolbox Design ... 9
3.2.1 Overview ... 9
3.2.2 WP2: Modelling uncertainties relevant for the operation of the European transmission grid ... 9
3.2.3 WP3: Optimization algorithms for transmission system operation... 10
3.2.4 WP4: Risk-based Security Assessment incorporation Forecast Uncertainty and Cascading events ... 11
3.2.5 Toolbox Requirements based on TSO Demands and Testing Environment ... 11
Research Project UMBRELLA
Page 3
1
Introduction
UMBRELLA is developing an innovative toolbox to support the decentralised grid security approach of TSOs, giving the opportunity to increase cooperation when facing the increased complexity in system's operation. A decentralised network security analysis with everyone "on board" looking at the same results and evaluating solutions in a coordinated and optimised way, increases the efficiency of the network operation. Furthermore, umbrella methodologies gives a step forward in the evaluation of uncertainties and their impact in different operational timeframes, the introduction of risk-based assessment and optimisation of remedial actions. This toolbox to be used in different operational timeframes includes:
Modelling and simulation of uncertainties due to market activities, renewable energy sources on different time scales (RES forecast) and outages
Optimisation algorithms of remedial actions in reaction to simulated risks on different time scales according to total costs and transmission capacities
Development of risk based assessment concepts for anticipated system states with and without corrective actions
In the present deliverable, the results of the workshop ‘UMBRELLA Workshop on Optimized Solutions’ and ‘UMBRELLA Workshop on System State Modelling and Toolbox Design’ are summarized. The workshops were both hosted at ENTSO-E’s premises on October 22, 2013 and October 23rd, 2014 with a total number of 15 and 33registered participants respectively, including experts from industry, academia and regulatory bodies. This deliverable is structured in the following way: Chapter 1 introduces the framework for an introduction of the UMBRELLA project, Chapter 2 presents the organization and structure of both workshops, Chapter 3 presents summaries of the workshop presentations and in the Appendix the presentation slides of the workshop of October 23rd 2014 are
displayed. The slides of the workshop of October 22nd 2013 are not included since an
updated version is also published within the previous Deliverable 7.3.
The presentations and related documents are available for download at www.e-UMBRELLA.eu, as approved by all presenters.
Research Project UMBRELLA
Page 4
2
Organization and structure of the workshops
First the agenda and structure of the workshop on Optimized Solutions of October 22nd
2013 will be discussed. Secondly the workshop on System State Modelling and Toolbox Design of October 23rd 2014 will be discussed.
2.1
Workshop on Optimized Solutions
The first workshop was held at ENTSO-E’s premises in Brussels on October 22nd, 2013.
Most of the stakeholders were invited by email.
The agenda of the workshop consisted of the following parts:
10:00 Workshop opening
10:15 Overview of WP3(Optimization algorithms) 10:30 Presentation Task 3.1
Proactive EOPF for anticipated Critical System States
12:00 Lunch Break
13:00 Presentation Task 3.2
Short Term and Real Time EOPF
14:30 Lunch Break
14:45 Presentation Task 3.3
Advanced methods for uncertainty accounting in (enhanced) optimal
power flow
15:15 General Discussion 15:45 Coffee Break
16:00 Presentation WP4 (Risk based assessment) 17:00 Workshop Closing
Research Project UMBRELLA
Page 5
2.2
Workshop on System State Modelling and Toolbox Design
The last workshop was held at ENTSO-E’s premises in Brussels on October 23rd , 2014.
Most of the stakeholders were invited by email. Moreover, the consortium published a newsletter in the middle of August 2014. This newsletter contained information about the workshop and about the way to subscribe.
The agenda of the workshop consisted of the following parts: 10:00 Workshop opening
10:30 Introduction
Welcome speech by ENTSO-E
Introduction to the UMBRELLA Project (work packages 1 & 7)
11:00 Part A: Results on optimized solutions for maximizing power transits, system state modeling and toolbox functionalities
Forecasting (work package 2):
Modeling uncertainties relevant for the operation of the European transmission grid
Optimization (work package 3):
Optimization algorithms for transmission system operation
12:30 Lunch
13:30 Part A Continued
Risk‐based assessment concepts (work package 4):
Risk‐based Security Assessment Incorporating Cascading Events and Forecast Uncertainty
14:15 Coffee Break
14:30 Part B: Prototyping and toolbox design
Synthesis and prototyping (work package 5)
Demonstration and testing (work package 6)
16:15 Conclusion and summary 17:00 End of workshop
Research Project UMBRELLA
Page 6
Figure 1: UMBRELLA Workshop October 23rd 2014
Research Project UMBRELLA
Page 7
3
Summary of the workshop presentations
The content of both workshops will be discussed. Again, first the workshop on Optimized Solutions of October 22nd 2013 and second the workshop on System State Modelling and
Toolbox Design of October 23rd 2014 will be discussed.
3.1
Workshop on Optimized Solutions
3.1.1
WP3.1 Presentation and discussion
The presentation on Work Package (WP) 3.1 includes a short description of the requirements on the optimization algorithms aligned in a meeting in Aachen in January 2013, an explanation of the developed optimization algorithms and some exemplary results.
The availability and activation time of remedial measures is discussed, since it is an important part of the prototype parameterization. The optimization algorithm itself is capable to deal with the corrective application of remedials, if they are available on a short term. Operational practice allows a limited number of topology switching measures. This requirement is met by the step-by-step selection of topology modifications starting from an initial network state and selecting only one modification at a time. The properties of the chosen heuristic are discussed in terms of optimality. Further enhancement of the optimization tool, especially in terms of power plant startups, will be made in WP 3.3.
3.1.2
WP3.2 Presentation and discussion
An overview of the used methods is given for Short Term and Real Time EOPF: modelling of measures (dispatch, HVDC lines, shunt elements, transformers with adjustable taps), handling N-1 security and Topology optimization.
There has been some discussion on the estimation of contingency situation. The estimated overloads caused by different line-failures are put in the objective function in a linear or quadratic way, since different lines may have different importance with regard to failure. Therefore the presented model includes the possibility to weigh the lines according to their importance. The idea of putting overloads as a penalty into the objective function relates to the work being done in WP4.3.
WP3.1 and WP3.2 have in common that they are both using topology-optimization, but different methods are used. The method presented by WP3.1 is a heuristic approach intended to reduce overloadings before starting the optimization process. The heuristic is used to effectively limit the search space, but suggested topology modifications are used only, if they lead to improvement. The method presented by WP3.2 serves to reduce costs and congestions after a feasible solution is found. The different topology-optimization
Research Project UMBRELLA
Page 8
methods are not confined to the two respective work package parts. The two work package parts give the possibility to compare, for small instances, the obtained solutions to optimal solution found by brute force, say using MINLP.
Finally, some work on extending Matpower to automatically optimize transformer tap position is presented. For variable transformers, there is currently no extension for Matpower. It would be a very useful addition for the broader community. The extension will be made available when it’s ready.
3.1.3
WP3.3 Presentation and discussion
The presentation contains the requirements of WP3.3 according to the description of work and the basic concept on how those requirements could be considered in the implementation. In addition, possible probabilistic criteria as alternative for conventional (N-1)-criterion are presented. This could be the TSO experience or an input of WP4.
The advantages of considering uncertainties are discussed. TSOs see the calculation of startup necessities for power plants as key benefit. It is difficult for TSOs to define a maximum risk level for each line due to regulatory reasons. The probabilistic criterion should be an outcome of WP4.
3.1.4
WP4 Presentation and discussion
WP4 presented some ideas on how to model severity (i.e. as technical violations or overall system risk). One severity function model currently being developed within WP4is reflecting risks (e.g. cascading) or costs (e.g. of remedial measures) that are not explicitly accounted for in the OPF. The physical implications of this severity function and how it used as an interface between the other work packages.
Finally, better estimates of probability of outages (i.e. sensitivity to errors in the estimation) and available data were discussed.
Research Project UMBRELLA
Page 9
3.2
Workshop on System State Modelling and Toolbox Design
3.2.1
Overview
UMBRELLA is developing an innovative toolbox to support the decentralised grid security approach of TSOs, giving the opportunity to increase cooperation when facing the increased complexity in system’s operation. A decentralised network security analysis with everyone “on board” looking at the same results and evaluating solutions in a coordinated and optimised way, increases the efficiency of the network operation. Furthermore, umbrella methodologies gives a step forward in the evaluation of uncertainties and their impact in different operational time-frames, the introduction of risk-based assessment and optimisation of remedial actions. This toolbox to be used in different operational time-frames includes: Modelling and simulation of uncertainties due to market activities, renewable energy sources on different time scales (RES forecast) and outages [Work Package (WP) 2]. Optimisation algorithms of remedial actions in reaction to simulated risks on different time scales according to total costs and transmission capacities [WP3]. Development of risk based assessment concepts for anticipated system states with and without corrective actions [WP4]. To perform the presented work, Umbrella has the following project structure: All this will be integrated into one toolbox [WP5]. Finally, the functionality of this toolbox will be tested and demonstrated [WP6].
3.2.2
WP2: Modelling uncertainties relevant for the operation of the
European transmission grid
WP2 is concerned with the assessment and description of uncertainties that influence the grid operation notably. Thereby, the uncertainties are described in a way that allows for a straightforward integration of the additional information into the TSO’s operational planning. Therefore, so-called system state parameters, which characterise crucial information about the whole system in a feasible way for operational processes, have to be derived and subsequently forecasted.
The forecast uncertainties of renewable energy infeed and load are modelled at each grid node with non-parametric kernel density approaches. The spatial interdependence of each factor is described with a copula. The uncertainty of short-term trading is modelled by a merit order model, which takes the aforementioned uncertainties as input. This allows for running a Monte-Carlo simulation that returns the distributions of load and infeed at each grid node. With subsequent load-flow calculations, the distributions of system state parameters (line loading, voltage etc.) can be computed. By filtering critical systems states and their respective forecast conditions, the relationship can be used to forecast critical system states.
The methods for describing the uncertainties have been fully developed and tested. Currently, the developed methods for forecasting critical system states are tested.
Research Project UMBRELLA
Page 10
3.2.3
WP3: Optimization algorithms for transmission system operation
The aim of the UMBRELLA Project is to assist transmission grid operators to ensure secure transmission grid operation. Therefore, optimization algorithms have been developed, which are capable of determining remedial measures to maintain system security at minimal costs. In order to provide system operators with adequate action recommendations, it is of special significance, that these algorithms take into account all available remedial measures as well as all relevant contingencies. In particular, relatively new technologies with great influence on power flows like HVDC connections operated in parallel to the asynchronous grid and phase shifting transformers are considered among conventional measures like topology modifications and redispatch.
The new options in transmission grid operation resulting from an increased controllability of load flow control devices in contingency situations are optimized as corrective remedial measures and provide significantly increased flexibility to transmission system operators in stressed grid situations. However, uncertainties resulting from error-prone feed-ins of renewable energy sources can lead to deviations from anticipated system states.
A reasonable way to deal with this issue is to postpone activation decisions of remedial actions as long as possible and react if critical deviations from anticipated system states actually occur. Therefore, a short term optimization algorithm has been developed, which takes into account measures available on a very short term as well as real-time constraints. Nevertheless, it is crucial for system security to avoid unmanageable system situations. To deal with this issue, probabilistic optimization algorithms have been developed, which are capable of directly incorporating uncertainties within the operational planning process. This way, especially optimal power plant start-up decisions, which have to be performed up to 24h before the actual operation, can be taken.
Figure 3: General overview of Work Package 3
Network model including load/feed-in
Network related remedials
Redispatch potential
© SD 2013
Network use cases
0:302:30 4:30 6:308:3010:3012:3014:3016:3018:3020:3022:300:30 D D+ 1 Contingency simulation Estimation of relevant contingencies Results Contingency list Outage A Branch i: Node j: Outage B Branch k: Outage …
Optimization of transmission system operation
Possible switching states Continuous optimization Topology optimization Remedial measure utilization Umbrella Optimization Tool
Secure network state Uncertainties
Research Project UMBRELLA
Page 11
3.2.4
WP4: Risk-based Security Assessment incorporation Forecast
Uncertainty and Cascading events
The topic of WP4 is risk-based security assessment, which encompasses both the probability and the severity of events. Since the beginning of the UMBRELLA project, we have developed and tested different models for power system operational risk and different representations of uncertainty related to fluctuating in-feeds from RES. The result is a set of methods which can be used for a risk-based security assessment, including both risk-based, probabilistic optimal power flow (OPF) formulations and methods for probabilistic evaluation of the risk from cascading events.
The last deliverable of Workpackage 4 is “D4.3 Methods for optimization of power transits” and has two parts. First, the previously developed risk-based methods have been extended to handle state-of-the-art technological means such as HVDC and PSTs. This allows for an assessment how HVDC and PSTs can be used to reduce risk and handle uncertainty, thus allowing for an optimization of the power transfer capacities. Second, the developed methods are used to assess how different market designs and cooperation rules influence the risk in system operation.
3.2.5
Toolbox Requirements based on TSO Demands and Testing
Environment
3.2.5.1 WP5: Synthesis and Prototyping
The review of the concepts developed for forecasting, optimization and risk-based security assessment is completed and led to the selection of those functionalities and modules which will be included in the toolbox design and the development of the toolbox prototype. The synthesis of the selected modules with the expectations and requirements formulated by TSOs is finished as well.
Figure 4: General overview of Work Package 5
Prototype
Development
Toolbox
Synthesis
Usage
Concept
Visualisation
Concept
Toolbox Design
Research Project UMBRELLA
Page 12
The conclusion of this work forms the basis for the developing of a toolbox design and of the toolbox prototype which will be used for demonstration purposes and tested by TSOs. First steps on the development of the usage and visualization concepts have been carried out. The work on both concepts started with an analysis of the scientific state-of-the-art and the approaches currently applied by TSOs. Based on the results of the analysis, the two concepts will be developed as part of the toolbox design and integrated in the prototype. The feedback provided from the testing of the prototype will show if these concepts fulfill the requirements by TSOs in terms of the integration of the toolbox in the existing processes and of the plain interpretation of the results obtained by operating the toolbox.
The discussion on the setup of the toolbox prototype concluded in a centralized approach i.e. the toolbox prototype will be installed on a central server.
Figure 5: Schematic overview of centralized toolbox
The TSOs executing the demonstration and testing will have access to the central server and be able to run the toolbox prototype. The major advantages of such an approach is the fact that a bidirectional communication is required between the central server and the users/testers of the toolbox prototype only and no data exchange between the different TSOs is required. Furthermore the implementation of experimental software in secured TSO-IT-systems is minimized. Nevertheless a decentralized use of the toolbox is possible by using an appropriate IT-framework.
3.2.5.2 WP6: Demonstration and testing
Since the last newsletter two major steps have been taken forward within the demonstration and testing part of the project.
Toolbox-Prototype
Research Project UMBRELLA
Page 13
Step one was the finalization of the “testbook”. Three test cases have been chosen which are each described with a general synopsis by all TSO. Furthermore, every case features detailed descriptions by each TSO concerned, comprising the encountered grid situation, the countermeasures taken and the expectation of the TSO regarding the toolbox results. The test case “Cold snap February 2012” has also been harmonized with the FP7-project iTesla. The other test cases cover more aspects - e.g. high wind infeed in the northern part of Germany and different seasonal current limits on lines.
Figure 6: Average minimum temperatures in February 2012 & 2013 to illustrate the cold snap period graphically. (source: www.wetteronline.de)
Step two was the first testing workshop. Due to the mixed approach of a physical meeting with the possibility to participate via web conference, all involved TSOs could participate in the test. Although, in this stage of the prototype only the deterministic functions of the toolbox are available, the optimization possibilities and the short computation time were really impressive. Further guided online test sessions and two physical meetings shall be held to roll out the additional functions of the toolbox. Within these sessions, users will get more acquainted with the toolbox and feedback for the improvement of the toolbox prototype will be collected.
Acknowledgement
The UMBRELLA consortium would like to thank ENTSO-E for hosting both workshops and the organizational work for the very good cooperation.
Research Project UMBRELLA
Page 14
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 11
The Umbrella project
Workshop on System State Modelling
hop on S
and
System State Mod
on S
d Toolbox Design,
23rd
and
and
d
d oolbox De
To
To
oolbox Desig
d of October 2014, Brussels (
n,
n
esig
esig
s (
(
(
s (
(
(
entso
so
so-
o
o
o
o
o
o
o
o
o
o
o
o
--
-
-e
e
e premises)
Agenda - Morning
10:00 Welcome with coffee
10:30 Introduction
a) Welcome speech by ENTSO-E
b) Description of the UMBRELLA project (Work Packages 1&7)
11:00 Part A: Results on optimised solutions for maximising power
transits, system state modelling and toolbox functionalities
a) Forecasting (Work Package 2)
Modelling uncertainties relevant for the operation of the European transmission grid
b) Optimization (Work Package 3)
Optimization Algorithms for Transmission System Operation
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 3
Agenda - Afternoon
13:30 Continuation of Part A
a) Risk-based Assessment Concepts (Work Package 4)
Risk-based Security Assessment Incorporating Cascading Events and Forecast Uncertainty
14:15 Coffee break
14:30 PART B: Prototyping and Toolbox Design
a) Synthesis and prototyping (Work Package 5)
b) Demonstration and Testing (Work Package 6)
16:15 Conclusion and summary
17:00 End of workshop
Helmut He He He He He He He He HeHe ut Paeschke (TenneT ke (TenneTeT TSO O GmbH), Wulf Engl (Englgl-glglgl----Energie on behalf alf TenneT TSO GmbHmbHH), Laura TSO TSO Gm eT eT TS ura ura Ramírez Gm rez
rez-bH), Wulf Engl (Englglglglgl----Energie on behalEnEn alf TennTe bH GmbH Gm z z z z z z z z z z z z z z z z
z----Elizondo (Delft University of Technology)
Workshop on System State Modelling and Toolbox Design
Project Overview
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 5
Agenda
I.
Welcome
II.
Workshop Agenda & Introduction to the Panel
III. Motivation of the UMBRELLA Project
IV. Project Objectives
V.
Project structure and Participants
VI. Demonstrator and KPIs
VII. Dissemination
Conferences, Workshops, Newsletter
o
Uncertainties in transmission network operation due to increase of
ointermittent renewable energy sources (RES) and
o
volumes of market-based cross border flows and related physical flows
o
Maximization of transportation possibilities
o
New interconnections and devices for load flow control
o
Zonal structure of the European
energy market with legal responsibilities
of TSOs imposes increasingly complex
requirements to the TSOs’ cooperation
concerning quality and accuracy
à Development of coordinated grid security tools
taking into account all technological measures
for flexible power system operation
Motivation of the UMBRELLA Project
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 7
Agenda
I.
Welcome
II.
Workshop Agenda & Introduction to the Panel
III. Motivation of the UMBRELLA Project
IV. Project Objectives
V.
Project structure and Participants
VI. Demonstrator and KPIs
VII. Dissemination
Conferences, Workshops, Newsletter
“Toolbox for Common Forecasting, Risk Assessment, and
Operational Optimisation in Grid Security Cooperations of TSOs”
•
Develop an innovative toolbox to support the decentralised grid
security approach of TSOs
•
This toolbox shall include:
§ Simulation of uncertainties (WP 2 Forecast) due to market activities
and renewables on different time scales
§ Optimisation (WP 3) of corrective actions in reaction to simulated risks
on different time scales according to total costs and transmission
capacities
§ Development of risk based assessment concepts (WP 4) for
anticipated system states with and without corrective actions (WP 4)
•
Demonstrate the enhancement of existing and running procedures
by utilisation of the developed toolbox (WP 5 & WP 6)
•
Provide a scientifically sound basis to support common TSO
decisions
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 9
Agenda
I.
Welcome
II.
Workshop Agenda & Introduction to the Panel
III. Motivation of the UMBRELLA Project
IV. Project Objectives
V.
Project structure and Participants
VI. Demonstrator and KPIs
VII. Dissemination
Conferences, Workshops, Newsletter
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 11
Project structure
Risksksk--based Risksksk basedbaba Assessment OptimizationForecasting Synthesis andSynthesis and
Prototyping
PM
M
Project Management (WP1)
PMB
PMB
Project
Project
Management
Manage
Board
WP2
WP3
WP4
WP5
Demonstration Demonstration and TestingWP6
DisseminationWP7
Project Consortium
U Duisburgburgg--Essen RWTH Aachen ETH Zurich Amprion TransnetBW TU Delft TenneT TSO Germany
PMT
T
Project Management Team
Project Mana
P
(PM, WG
Mana
G
G
-agement T
na
ana
G
G--
Leaders)
Participating TSOs: TenneTeTTSO Participat Pa O GmbH, ating TSOs icipat bH, bH, Amprion Os: TSOs nnGmbHbHbHbH, , ČEPSPSPS, , a.sa.sa.ss.., Tenn Elektro Tenn tro tro -eT eT TSO TSO nneT Tenn o o--Slovenija O ija ija, GmbH Gm Gm Gm a a, d.o.o bH o.o o.o, Amprion Am n Gm Am bH bH, Am o o o, TransnetBW GmbH Gm BW BW BW, bH bH, bH bH GmbH BW BW, BW PSE
ČEPSPSPS, a.a.a.sa.ss
, ČEČEČE
E
E Operator
s., or orS.A.,
Elektrtroo SlovSl
swissgrid ov Slov d d dag ovenij ov ag ag ag ag, a a, d.o.d.d. ij ija enij ag ag, TenneT , Tr o. o.oo eT eTTSO ansnet Tran Tr Tr O O O B.V. ansnet V. V., etBWBWBW, BW PSPS et snet V. V., Austrian Operator Op PSE PSE Op an an anPower or or erator wer wer Grid A., or A.S.A. id idAG
Participating Universities and research institutes:
Participating Universities Delft University of es and resea sities of of Technology esea gy gy, rch institutes earc esea gy gy, gyETH Zurich, Delft University o Graz University of Technology Te ty o of of ofTechnology logy gy gy, ETH Zurich, ET ET gy gy gy gy gy gy gy, gy gy gy gy gy gy gy logygy gy gy, gy RWTH Aachen, az University of University of TechnoloTe y ofof ty tyDuisburg nolo rg rg -logygygygygygygygygygygygygygy, logy nolo rg rg--Essen , RW sen sen, RWTH Aac RWTH RW RW RW n n, FGH TH Aac H H e.V he ache Aac e.V e.V e.V e.V.
Agenda
I.
Welcome
II.
Workshop Agenda & Introduction to the Panel
III. Motivation of the UMBRELLA Project
IV. Project Objectives
V.
Project structure and Participants
VI. Demonstrator and KPIs
VII. Dissemination
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 13
The five innovation clusters of TSOs
Cluster Name Functional
Objective Full names of Functional Objectives
C1 Grid architecture
T1 Definition of scenarios for pan-European network expansion T2 Planning methodology for future pan-European transmission system
T14 Towards increasing public acceptance of transmission infrastructure
C2 Power technologies
T3 Demonstration of power technology to increase network flexibility and operation means T4 Demonstration of novel network architectures
T5 Interfaces for large-scale demonstration of renewable integration
C3 Network operation
T6 Innovative tools and methods to observe and control the pan-European network T7 Innovative tools and methods for coordinated operation with stability margin evaluation T8 Improved training tools and methods to ensure better coordination at theregional and pan- European levels
T9 Innovative tools and approaches for pan-European network reliability assessment
C4 Market designs
T10 Advanced pan-European market tools for ancillary services and balancing, including active demand management
T11 Advanced tools for capacity allocation and congestion management
T12 Tools and market mechanisms for ensuring system adequacy and efficiency in electric systems integrating very large amounts of RES generation
C5 Asset management
T15 Developing approaches to determine and to maximize the lifetime of critical power components for existing and future networks
T16 Development and validation of tools which optimize asset maintenance at the system level, based on quantitative cost/benefit analysis
T17 Demonstrations of new asset management approaches at EU level
Source: Methodological Guide on EEGI KPIs
Demonstrator and KPIs
Source: Methodological Guide on EEGI KPIs
Project KPIs
Overarching KPIs
Progress of EEGI activities to overarching goal
Specific KPIs
Link with Clusters and Functional Objectives. Measure impact of group of R&I activities Defined on a project by project basis Pr ov
EEGI
Roadmap
Cluster of
Projects
ect Is De projProjects
o
A KPI should be reasonable, understandable, meaningful
o
The overall number of KPIs for a project should be reasonable
o
A KPI usually is calculated by comparing the “business as usual (BAU)”
case versus the “research and innovation (R&I)” case
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 15
Agenda
I.
Welcome
II.
Workshop Agenda & Introduction to the Panel
III. Motivation of the UMBRELLA Project
IV. Project Objectives
V.
Project structure and Participants
VI. Demonstrator and KPIs
VII. Dissemination
Conferences, Workshops, Newsletter
Second Newsletter (2014/1)
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 17
Posters
Conferences
o
CIGRÉ 2014 @ Paris, France
o
Innogrid2020+ @ Brussels, Belgium
o
Energycon @Dubrovnik, Croatia
o
EU Energy week, @ Brussels, Belgium
EEGI Grid+ Event
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 19
Thank you very much for your attention!
www.e-umbrella.eu
This research work has been carried out within the scope of the project UMBRELLA, supported under the 7th Framework Programme of the European Union, grant agreement 282775.Questions?
Comments?
Dr. ir. L.M. Ramirez Elizondo
Assistant Professor DC Systems & Storage
TU Delft Mekelweg 4 2628 CD Delft Telephone: +31 15 27 81848 Mailto: L.M.RamirezElizondo@tudelft.nl Website: www.ewi.tudelft.nl/en/the-faculty/departments/electrical-sustainable-energy/ Dr. Wulf A. Engl
Consulting, Interim and Project management
Engineering consultant Engl-Energie Thingstr. 24 D-82041 Oberhaching Telephone: +49 (89) 905470 - 90 Telefax: +49 (89) 905470 - 88 Mailto: info@engl-energie.de wulf.engl@tennet.eu Website: www.engl-energie.de
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 11
Christoph Weber (University of Duisburg-Essen)
Workshop on System State Modelling and Toolbox Design
Forecasting
23rd of October 2014, Brussels
Agenda “Forecasting” (WP 2)
I.
Background and overview
II.
Challenges
III.
Scientific Approach
IV. Results
V.
Outlook
VI. Conclusions
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 3
WP 2: Background and Overview
Main objective:
•
Developing a set of methods describing the key uncertainties
relevant for operating the transmission grid
Considered uncertainties:
•
Feed-ins from renewable energy sources
•
Load
•
Intraday trading
•
Power plant outages
WP 2: Operational planning today
•
DACF: TSOs create point forecasts of certain system parameters
(load, infeeds, …)
ܺ
•
Snapshots: Measurements of system parameters
ܺ
differ from the
forecasts à forecast error:
ܧ
ହൌ
ܺ
ହെ
ܺ
ହݐ
ܺ
௧ͳ
Ͳ
ʹ
͵
Ͷ
ͷ
ܺ
ହܺ
ହܧ
ହൌ
ܺ
ହെ
ܺ
ହ Point forecast: Measurement: Forecast error:UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 5
WP 2: Forecast error
•
Possible deviations from the point forecast can be used for
operational planning
•
Deviations are described by the random variable
ܧ
௧•
What is the probability of a certain deviation? à WP2
ݐ
ܺ
௧െͶ
െͷ
െ͵
െʹ
െͳ
Ͳ
ܺ
ଵ=
ܺ
ଵെ ܧ
ଵ Probability of the forecast error?ͳ
WP 2: Forecast error
•
WP2 does not make nor improve any point forecasts
•
WP2 develops methods that describe the deviation from the point forecast
and their probability à focus on forecast error/ forecast uncertainty
•
How to describe forecast uncertainty à probability density function (pdf)
ݐ
ܺ
௧െͶ
െͷ
െ͵
െʹ
െͳ
Ͳ
݂
ܺ
ଵൌ ݂
ܺ
ଵെ ܧ
ଵFocus on:
݂ ܧ
ଵͳ
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 7
WP 2: Time dependence
•
Uncertainty increases with an increasing time horizon à each
look-ahead time step should have its own forecast error pdf
•
݂
ଵܧ
ǡ ݂
ଶܧ
, ….,
݂
்ܧ
ݐ
ܺ
௧െͶ
െͷ
െ͵
െʹ
െͳ
Ͳ
ͳ
ʹ
͵
WP 2: Forecast condition
•
Different forecast conditions (
point forecast
ܺ
) lead to different forecast
error pdf
à pdf must be conditional on the point forecast à conditional
pdf (cpdf)
•
݂
ଵܧ
ȁ
ܺ
ଵǡ ݂
ଶܧ
ȁ
ܺ
ଶ, ….,
݂
்ܧ
ȁ
ܺ
்ݐ
ܺ
௧െͶ
െͷ
െ͵
െʹ
െͳ
Ͳ
ͳ
ʹ
͵
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 9
WP 2: Spatial interdependence
•
Point forecasts and uncertainty forecasts are only usable at a grid
node level à cpdf for each grid node ݊ ൌ ͳǡ ǥ ǡ ܰ: ݂
௧ሺܧ
ሻ
•
݂
ଵܧ
ȁ ܺ
ଵǡ ݂
ଶܧ
ȁ ܺ
ଶ, ….,
݂
்ܧ
ȁ ܺ
்•
݂
ଵܧ
ȁ ܺ
ଵǡ ݂
ଶܧ
ȁ ܺ
ଶ, ….,
݂
்ܧ
ȁ ܺ
்•
…
•
݂
ଵࡺܧ
ȁ ܺ
ଵࡺǡ ݂
ଶࡺܧ
ȁ ܺ
ଶࡺ, ….,
݂
்ࡺܧ
ȁ ܺ
்ࡺ•
Forecast errors at different grid nodes are correlated
•
E.g.: if the forecast error at
݊ ൌ ͳ is high, the forecast error at ݊ ൌ
ʹ is also high
•
Linear correlation coefficients are limited to linear relationships à
use of copulas
WP 2: Copulas
•
A copula defines the functional relationship between a joint
distribution function,
ܨሺݔ
ଵǡ ǥ ǡ ݔ
ேሻ, and its marginals ܨ ݔ
ଵǡ ǥ ǡ ܨሺݔ
ேሻ
•
ܨ ݔ
ଵǡ ǥ ǡ ݔ
ேൌ ܥሺܨ ݔ
ଵǡ ǥ ǡ ܨሺݔ
ேሻሻ or as a density function
f
ݔ
ଵǡ ǥ ǡ ݔ
ேൌ ܿሺ݂ ݔ
ଵǡ ǥ ǡ ݂ሺݔ
ேሻሻ
•
Thus, this allows to separate the modelling of the marginal
distributions and the relationship between them
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 11
WP 2: Main benefits for
operational planning
•
Likely contingencies could be overlooked if a point forecast is used merely
– However, e.g. power plants need a certain startup timeà Optimization necessary(WP3)
•
HILP events might lead to cascading events
– However, risk assessment necessary (WP 4)ݐ
ܺ
௧െͶ
െͷ
െ͵
െʹ
െͳ
Ͳ
ͳ
ʹ
͵
Redispatch
necessary?
High impact, low probability (HILP) event?
Agenda
I.
Background and overview
II.
Challenges
III.
Scientific Approach
IV. Results
V.
Outlook
VI. Conclusions
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 13
WP 2: Challenges
Main challenge:
•
Most of the processes are more or less easy to predict individually
•
However, TSOs can only observe an aggregation of them
Solution:
(1) Disaggregation and individual simulation
(2) Incorporation of interdependences
(3) Reaggregation
Transmission grid Distribution grid Observable Non-observableWP 2: Challenges
(1) Disaggregation and individual simulation of :
•
RES infeed
•
Residual load and underlying
conventional generation
•
Power plant outages
(2) Incorporation of interdependences
•
RES infeed
↔
Intraday market uncertainty
•
…
(3) Reaggregation to system parameters
than can be used in operational planning:
•
ܲ
ௗ(
includes RES infeed
,
residual load
)
•
ܲ
ௗ(includes intraday market, power plant outages)
aint
t
y
Transmission grid
Distribution grid
Observable Non-observable
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 15
Agenda
I.
Background and overview
II.
Challenges
III.
Scientific Approach
IV. Results
V.
Outlook
VI. Conclusions
WP 2: Uncertainty in renewable power
forecasts
Objective
•
Identify appropriate possibilities for describing these uncertainties in
a way that grid operators may make use of the information most
efficiently
Methodology
•
Estimation of forecast uncertainty for each grid node
– Conditional probability density function (cpdf) of the forecast error
conditional on the deterministic forecast
– Different estimation for each grid node and look-ahead time using
non-parametric approaches (Nadaraya-Watson estimator)
•
Modelling of spatial interdependences
– Copula models (foregoing goodness-of-fit tests)
•
Simulation of forecast uncertainty
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 17
WP 2: Uncertainty in load forecasts
Objective
•
Identification of suitable descriptions of the associated uncertainties
in view of providing a consistent overall system state description
Methodology
•
Estimation of forecast uncertainty for each grid node
– Different estimation for each grid node and look-ahead time using
non-parametric approaches (kernel density estimation)
1. If underlying RES infeeds: residual load is model
2. If “pure” load node: vertical grid load (same as in DACF) is modelled
•
Modelling of spatial interdependences
– Copula models (foregoing goodness-of-fit tests)
•
Simulation of forecast uncertainty
– cpdf + point forecast + copula à Monte-Carlo-Simulation
WP 2: Uncertainty in power plant outages
Objective
• Identification of suitable descriptions of the associated
uncertainties in view of providing a consistent overall system
state description
Methodology
• Time to fail is modelled for each power plant type using an
exponential distribution
• Time to repairs is not modelled, because it is assumed that
power plant operators notify TSOs in advance
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 19
WP 2: Uncertainty in short-term trading
Objective
• Detailed analysis of intraday renominations and a subsequent
proposal for statistical description
Methodology
• Merit order model: running power plants will be active in the
intraday market depending on their marginal costs
• Uncertainties on the load side can be used to anticipate
intraday trades
• Remaining uncertainties (fuel costs, irrational behaviour,
provision of spinning reserves) are represented by a sampled
merit order
WP 2: Uncertainty in short-term trading
Price
Quantity E(PRES)
E(Load) Input: Uncertainty
of RES and load
Output: Change in power
production at each grid node and, thus, change of trades
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 21
WP 2: Deriving forecast
distributions for the system state
Objective
•
Development of methods to derive information about future system
states and their occurrence probability based on previously
discussed forecast distributions
•
Provide information about the system state to system dispatch
Methodology
•
Based upon the modelled uncertainties a Monte-Carlo-Simulation is
run
– Requires the reaggregation of uncertainties to vertical grid load and
infeed at each node
•
For each Monte-Carlo run a load flow calculation is carried out
•
System state parameters (voltage, line loading etc.) can be
computed
WP 2: Deriving forecast
distributions for the system state
Load flow Calculations
Load flow Calculations
Load flow Calculations
Load flow Calculations
Distributions of load and RES infeed
Incorporation of specifics, in particular related to
intraday trading
Load flow Calculations
Distributions of system states
P and Q (load)
• Pump storages
• Cross-border flows
• …
Monte-Carlo-Simulation
Distributions of intraday trades
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 23
WP 2: Deriving forecast
distributions for the system state
Objective
•
Quantify the probability of occurrence for specific combinations of
uncertain events that cause critical system states in real-time
Methodology
•
Running a full Monte-Carlo-Simulation is too time-consuming in
real-time
•
Thus different approaches, that link certain forecast conditions with
the occurrence of critical system states, are currently tested:
– Artificial intelligence approaches
– Selective approach
•
A continuous improvement and training with new data will be
handled offline
WP 2: Deriving forecast
distributions for the system state
Load flow Calculations
Load flow Calculations
Load flow Calculations
Load flow Calculations
Distributions of load and RES infeed
Incorporation of specifics, in particular related to
intraday trading
Load flow Calculations
Distributions for system states
P and Q (load)
• Pump storages
• Cross-border flows
• …
Monte-Carlo-Simulation
Distributions of intraday trades
P and Q (load+injection)
Forecast conditions
• Level of expected wind infeed • Hour of the day • …Forecast of
critical
system
states
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 25
Agenda
I.
Background and overview
II.
Challenges
III.
Scientific Approach
IV. Results
V.
Outlook
VI. Conclusions
Ͳ
Deviation from
the point forecast More wind infeed
that expected
Less wind infeed that expected
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 27
Simulation of wind power uncertainty
Ͳ
Deviation from the point forecast
Rather strong interdependence à
deviations into one direction occur often
simultaneously
Simulation of solar power uncertainty
Ͳ
Deviation from the point forecast
Local weather effects (e.g. clouds) have an higher
impact à weaker interdependence
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 29
Simulation of solar power uncertainty
Ͳ
Deviation from the point forecast
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 33
Agenda
I.
Background and overview
II.
Challenges
III.
Scientific Approach
IV. Results
V.
Outlook
VI. Conclusions
Outlook
• System is currently implemented for 9 TSOs
• Further testing with data from 2012
• Integration of the forecasting module into the
Umbrella toolbox
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 35
Functional Link
Probabilistic Forecasting
Critical System States
Optimisation Framework
System States with Remedial Measures
Security Margins for Circuit Capacities Risk Assessment of Cascading Events Uncertainty of Circuit Flows Severity Function
External input data
External input data External input data
External input data
Critical System States
Optimisation Framework
System States with Remedial Measures
Security Margins for Circuit Capacities Risk Assessment of Cascading Events Uncertainty of Circuit Flows Severity Function
External input data External input data
External input data External input data External input data
External input data
WP2
Agenda
I.
Background and overview
II.
Challenges
III.
Scientific Approach
IV. Results
V.
Outlook
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 37
WP 2: Conclusions
• The Umbrella toolbox goes beyond today’s practice by
regarding the uncertainty of factors relevant for
operational planning such as wind infeed or intraday
trades
• In order to assess their uncertainty, nowadays’
processes have to be disaggregated, modelled
individually and finally reaggregated while regarding
relevant interdependences
• By using additional information (such as spatial
correlations, forecast conditions and time dependences)
for describing the uncertainties, the state space is limited
what allows for a computation within a practical period of
time
Thank you very much for your attention!
Questions?
Comments?
www.e-umbrella.eu
This research work has been carried out within the scope of the project UMBRELLA, supported under the 7th Framework Programme of the European Union, grant agreement 282775.UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 11
Jonas Eickmann, Tobias van Leeuwen, Andreas Moormann
Institute for Power Systems and Power Economics, RWTH Aachen University
Workshop on System State Modelling and Toolbox Design
Optimization
23rd of October 2014, Brussels
Agenda
“Optimization” (WP 3)
I.
Background and Overview
II.
Challenges
III.
Scientific Approach
IV. Results
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 3
o
Growing complexity and importance of transmission system operation
è
Development of expert system to provide optimized guidance
(n-0)-secure
Overall Objective
Objective
o
Reaching a (n-1)-secure grid state
Selected Approach
è
Development of an Optimization Framework
to support TSO in operational planning and
grid operation
o
Considering all available remedial
measures
o
Accounting for uncertainties in operational
planning process to provide sufficient
security margins and controllability
oReacting to deviations with real-time
applicability
(n-1)-secure
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 5
Probabilistic load -flow
Critical System States
Optimization Framework
System States with Remedial Measures
Security Margins for Circuit Capacities Risk Assessment of Cascading Events Uncertainty of Circuit Flows Severity Function
External input data
External input data External input data
External input data
Probabilistic load -flow
Critical System States
Security Margins for Circuit Capacities Risk Assessment of Cascading Events Uncertainty of Circuit Flows Severity Function np
External input data External input data
np
External input data External input data
np
External input data External input data
Overview Umbrella Project
WP3
Agenda
I.
Background and Overview
II.
Challenges
III.
Scientific Approach
IV. Results
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 7
Challenges
System Size
o
Large number of elements in European transmission grid
o
Cross border flows couple neighbouring transmission networks
(n-1)-Criterion
o
Transmission grid operation needs to cope with contingencies
o
Optimization problem size grows linear with number of relevant outages
Discrete Decisions
o
Several remedial actions require binary variables
o
Topology modifications have strong interdependency with other measures
Uncertainties
o
Future system state development in planning stage unknown
o
Accounting for a range of possible upcoming states required
Agenda
I.
Background and Overview
II.
Challenges
III.
Scientific Approach
IV. Results
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 10
Optimization for operational planning
Optimization Algorithms
Network model including load/feed-in
Network related remedials
Redispatch potential
Network state Network state
© SD 2013
Network use cases
0:30 2:30 4:30 6:30 8:3010:3012:3014:3016:3018:3020:3022:30 0:30 D D+ 1 Contingency simulation Estimation of relevant contingencies Results Contingency list Outage A Branch i: ܲǡ ܳ Node j: ܷǡ ȣ Outage B Branch k: ܲǡ ܳ Outage …
Optimization of transmission system operation Possible switching states Continuous optimization Topology optimization Remedial measure utilization Umbrella Optimization Framework
Secure network state
Optimization for real time application
System State MoSSystem State Mo
Uncertainty
dellinin and Toing aning aninin Mode Modede nty accounting Tool ng in of O of Design 23rdof Desi De box x De olbobo in in optimization Uncertainties Uncertainties
Risk based security assessmentReal time restrictions
Optimization Objectives
o
Selection of appropriate objective function key for representation of
Security Considerations
o
Minimization of constraint violations
o
Minimization of Risk
Economic Effects
o
Minimization of redispatch costs
o
Minimization of grid losses
Regulatory Considerations
o
Minimization of redispatch volume / market impact
è
Multi objective optimization with appropriate weighting of different
objectives required
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 12
Remedial Actions
Topology Modification
•
Switching lines on/off
•
Adjustment of busbar configuration
Transformer Tap Changing
•
Longitudinal & quadrature
Shunt Elements
•
Capacitors & reactors
HVDC Connections
•
Market based & TSO operated
•
Point-to-point & radial structures
Redispatch
•
Conventional generation units (incl. startups)
•
Curtailment of RES
•
Load shedding
Key Features
Decoupling of topology optimization and continuous optimization
o
Optimization problem tractable for large scale power systems
Deterministic optimization
o
Consideration of temporary admissible transmission loading (TATL)
after the outage (curative measures)
o
Real-time applicability
Probabilistic optimization
o
Minimization of probability of constraint violations and risk
o
Accounting for short term controllability while determining power plant
startups
Time coupling optimization of subsequent use cases
UMBRELLA Workshop on System State Modelling and Toolbox Design, 23rdof October 2014, Brussels 14
Umbrella Optimization Framework
è
Flexible adjustment for different time horizons and requirements possible
Je d er N N F iso li ert Je d er N N F iso li ert
Network use cases
0:30 2:30 4:30 6:30 8:30 10:30 12:30 14:30 16:30 18:30 20:30 22:30 0:30
D D+1
Topologieschaltmaßnahmen TopologieschaltmaßnahmenRobust Switching States
time-coupling, continuous relaxed, mixed-integer optimization Contingency simulation Load flow Solve problem time-co continuous Build problem Results
Load Flow Losses Remedials Costs
N e tw or k u se cases
Selection of topologyof topology ൎ scenarios scenarios scenarios
• (n-1)-constraints
• Voltages
• Currents
Topology Optimization (1/2)
o
Estimation of relevant congestions
o
For case without outage
oFor all contingencies
o
Heuristic preselection of relevant
topological measures (TM)
o
Congestion based selection of
relevant TM based on fast load flow
approximation
o
Normal operation
oContingency situations
o
Full continuous optimization
procedure for remaining measures
to estimate best topology
modification in current state
108 103 102 10 1 Initialization Identification of congestions Heuristic selection Substations Topological measures
Congestion driven selection Load flow approximation
(n-0) case (n-1) cases
Redispatch driven selection Continuous optimization