Final Report Summary - DREAM (Distributed Renewable resources Exploitation in electric grids through Advanced heterarchical Management)
Executive Summary:
The DREAM project launches a novel heterarchical management approach of complex electrical power grids, providing new mechanisms for stable and cost effective integration of distributed renewable energy sources, as well as for enhanced consumer involvement in economic and ecological electricity use.
The principles of autonomous agent-based systems are applied to the control and management of the electricity distribution grid, helping the system to constantly adjust to actual operational conditions and increasing its robustness versus exogenous disturbances. In turn, this allows for greater penetration of intermittent resources and makes the distribution grid more resilient to failures. In order to improve involvement of citizens in the grid management, DREAM implements market validation processes from the end user towards the transmission system delivery point. DREAM includes layers of controls for normal, congested and post-contingency situations according to a traffic light model, using coordination strategies to create ad-hoc federations of agents that flexibly adjust their hierarchical configuration to distribution grid needs.
The system transits smoothly between dynamic control layers depending on local operational conditions, so that responses to disturbances are sized precisely, margins are used parsimoniously and full network flexibility is tapped. The system involves limited data transfers, and promotes extensibility, heterogeneity and easy deployment across Countries with different network architectures and hardware manufacturers.
DREAM demonstrated the economic and technical feasibility of these novel control mechanisms thanks to several real-world small-scale pilots dedicated to different use-cases, and computer simulations have been used to further assess the solutions scalability.
Twenty four years after Rio92, six years after COP15 in Copenhagen and a year after COP21 in Paris, the international community duly notes the need for a new energy paradigm.Such an energy transition would induce the migration from an energy mix driven by conventional and carbonated bulk generation assets to a new paradigm relying on partly distributed renewable assets. As a direct consequence, the amount of conventional capacities naturally available would decrease, while the constraints on the distribution networks would rise.
The principle of aggregation of Distributed Energy Resources (DERs) emerged in previous years, with the objective to propose new solutions to handle such issues.
Project Context and Objectives:
DREAM was developing a novel paradigm for smart grids, capable of responding to current issues and challenges related to renewable integration and opening of the energy markets, and is particularly focused on three main goals in the evolution of energy distribution environment and of its interactions with energy transportation systems and with final users and energy stakeholders:
• Ensuring societal benefits in terms of environmental impact / carbon footprint, in line with the 2020 objectives, and of overall economy of electric systems; this objective is associated with the substantial increase of capabilities to integrate distributed renewable resources, including the intermittent ones like Photovoltaic and Wind, as well as with the reduction of energy losses along the lines and the improvement of efficiency for power plants from reduction of fluctuations in the demand.
• Accommodating smooth and sustainable transition to the new smart grid environment; minimizing drawbacks from the deployment of novel smart grids, through the capability of ensuring economic sustainability of investments, the flexibility in accommodating variable configurations of energy networks, and the scalability to cover millions of smart energy entities;
• Enhancing flexibility and robustness of technical and operational environment for energy network; leveraging the advantages of smart devices in generation, use and storage to improve response capabilities of distribution subnets, dynamic coordination of behaviour to minimize energy flows across subnets, innovative ancillary servicing capabilities and increased resilience to undesired events.
The developed DREAM framework and devices are strengthening power distribution grids robustness and efficiency by developing a novel autonomous, self-learning, heterarchical and agent-based approach to the control and management of the power grid. The project was targeted at the creation and management of several innovative autonomous control mechanisms for energy distribution that should be open, extensible, compatible with heterogeneous hardware components and compliant with existing national energy exchange regulations.
DREAM was developing control mechanisms to allow for the exploitation of all network entities – generators, loads, storage units, tap changers, voltage regulators, controllable switches... – by identifying the relevant behavioural profiles and integrating them into a more general control environment, which accommodates for minimisation of energy unbalances and other technical constraints violations, and thus leads to the following advantages:
• Improved robustness to exogenous perturbations to increase reliability. In addition, the system is including self-defense mechanisms that make it resilient to threats posed to its own integrity, such as faults and equipment failure.
• As a by-product of increased robustness, the system is more tolerant to disturbances such as fluctuating injections of active and reactive power by intermittent generation units; this is allowing accommodating more small-scale renewable resources in the generation mix, allowing for cleaner and more local electric energy production.
• More efficient utilisation of existing and new generation, transmission and distribution equipment, thanks to the reduction of range in operational conditions to be accommodated. It makes it possible to minimize excess capacity expenditures and to operate assets at their maximum efficiency.
The vision described above was achieved by integrating the coarse control of system objectives (such as frequency and transmission system voltages) at TSO level with the local control capabilities of entities constituting, or connected to, the distribution grid. DSOs thus could able to validate flexibilities (that is to say, residual flexibility that remains available after distribution level targets such as volt-var control, congestion management or loss minimization have been achieved satisfactorily) for market purpose use or mechanisms in case of emergency situations.
The control tools are the operational support components of the mechanism. Decision-making follows the principles of autonomous control and of coordinated behaviour, which translate into concrete actions such as generating or consuming active and reactive power, as well as operation of grid devices such as remotely controlled switches and on load tap changers.
The functions are designed for maximum automation while letting the DSO benefit from increased situational awareness with the option to take over the automated control mechanisms from the control centre. From this viewpoint, the DREAM system acts as an intermediary layer (wide area feeder automation) between the DSO and the network that filters out the important information (e.g. outliers detection, contingency anticipation) that request attention from human operators, while ensuring that business-as-usual operations are carried out automatically.
The agent-based heterarchical control addresses the logical and business characteristics of the grid and accommodates for real time, autonomous control of the distributed and heterogeneous energy generation and consumption nodes and hubs. The power grid of the future will be capable of ensuring the efficient integration and coordination of a composite network of smart nodes and hubs, which may be basic individual entities or aggregations which respond to different decision making mechanisms and principles and have various control capacities, but respond to rules and regulations that are enforced through the establishment of physical (e.g. breakers settings) or economic (e.g. non validation of offers) mechanisms. The agent-based system is pursuing efficient integration of existing and new players and technologies and leverage new control mechanisms and decision making flexibility, by discarding the pure hierarchical distribution model; instead, global objectives are pursued by optimising the local behaviour at the level of user nodes, control nodes and so on. Such a heterarchical approach is capable of addressing multi-composite nodes at physical (actual location, type and intrinsic characteristics), logical (behavioural profile, decision rules and controllability of performances) and business (advantage conditions, benefit sharing strategies, incentive policies exploitation) levels.
The heterarchical smart power grid envisioned and then built by the DREAM consortium is a heterarchical, flexible, dynamic and adaptive system. The distributed actors currently display little to no coordination amongst each other – DREAM recognises that decision-making in the grid occurs at the level of these individual actors. This results in highly flexible and dynamic processes for which agent-based models are uniquely suitable. On top of this, the distribution system itself – as well as its relationships with neighbouring entities such as the transmission system – needs to be included in the control loop and provided with adequate support both for its current, hierarchical architecture and control mechanisms, and for future locally control-capability-enabled distribution nodes.
Scientific and Technical Objectives
Detailed and measurable objectives for the DREAM project contribute to the goals described above. They include objectives related to the establishment of best conditions for successful take up and exploitation of project.
Main scientific and technical objectives the DSOs are thus:
• Develop a consistent concept for local energy market portal, compliant with national and international grid codes and networks across European Countries;
• Develop methods and tools for aggregating energy network entities into clusters capable of buying and offering power and ancillary services, for evaluating associated energy performance potential, and for fair and well performing allocation of roles;
• Develop logical methods and support tools to identify dynamic clusters of users to respond to contingent/emergency conditions and to coordinate individual behaviours;
• Define an adequate number of Key performance indicators to measure actual behaviour of new distribution grids mechanisms across trading, normal and contingency operation conditions
• Conduct a sufficient number of tests to ensure a confidence in the order of 90% on viability of concept, computed on the basis of measured values key performance indicators;
• Ensure proper and adequate dissemination of results, such as major stakeholders for new energy distribution grids, public administrations relevant to smart grids, industrial users of novel concepts, plus academy and research centers and the wider public for awareness rising.
Project Results:
The DREAM project is laying the foundations for a novel heterarchical management approach of complex electrical power grids, providing new mechanisms for stable and cost effective integration of distributed renewable energy sources, as well as for enhanced consumer involvement in economic and ecological electricity use.
Applying the principles of autonomous agent-based systems to the control and management of the electricity distribution grid is allowing the system to constantly adjust to current operational conditions and make it robust to exogenous disturbances. In turn, this is allowing a greater penetration of intermittent resources and is making the distribution grid more resilient to failures. DREAM is including several layers of controls for normal, congested and post-contingency situations that are using different coordination strategies ranging from market-based transactions to emergency demand response and create ad-hoc federations of agents that are flexibly adjusting their hierarchy to current needs.
The system is thus transiting smoothly between control layers depending on local operational conditions, so that responses to disturbances are sized precisely, margins are used parsimoniously and full network flexibility is tapped. The system involves only limited data transfers and no centralized control, promoting extensibility, heterogeneity and easy deployment across countries with different network architectures and hardware manufacturers.
The Project has been phased out in three major parts, as seen in Figure 1. The first part related to the mandatory research development was initiated in the “describe and develop...” phase, as described in section 1 of this description. In this phase, several algorithms were developed to cope with defined uses cases by the research institute of the Dream project in interaction with the final users. Secondly, the “combine...” phase was initiated in order to build and integrated the Dream framework. Indeed, the partners were cautious about interactions between distributed algorithms handling various grid constraints and acting on the same parameters. In order to assess and avoid oscillations, the integrated framework, that is described in part 2, was initiated for test and development. Lastly, the “apply...” phase was conducted in order to validate the concept on trials and evaluate the added value of the Dream concept.
1. “describe and develop...” phase
In this first phase, the work was divided in three different parts, belonging to the operation phase the distribution network is facing. This split can be seen as, at first, a distributed validation of business as usual energy and capacity bids or market operations in a normal operation mode, as described in subsection 1.1. The second part is related to a novel mechanism to compute a distributed contingency analysis. In this phase, the network is considered as probable to face emergency operation mode. This analysis, as any operational planning, is referring to the ability for the system to face most probable contingencies (what if?) and validate if the remaining flexibilities, combined with network operation can solve them with regular contracts, as described in subsection 1.2. The third part is related to critical operation mode, when the Distribution System is facing failures. This mode require self-defense reactions of any possible devices (from end user flexibilities to controllable grid components) to solve the critical situations, as presented in subsection 1.3.
1.1 Distributed validation of business as usual energy and capacity bids and market operations
In deliverable 2.1 the scientific barriers to enable such a control were identified and the concepts of the proposed solutions were given.
The Dream project issued some solutions and practical implementations to tackle the main barriers identified to enable a distribution market place at the distribution level. Three use cases have been described that build the frame.
The developed MV Aggregator model, described in the D2.2 facilitates the task of the MV Aggregator in choosing the optimal price levels to be announced to the DERs by transforming, in an equivalent way, the interdependence of the decisions into mathematical expressions defining a problem that can be solved, after applying proper linearization, with commercially available software. In this model, the aggregator entity manages several local entities, producing a bi-level formulation that facilitates the problem resolution. This approach is perfectly matching the heterarchical multi-agent architecture proposed in DREAM, and will directly affect the identified barrier of DSO motivation. The adoption of this kind of solution will help to cope with the three key challenges for a DSO:
• Connecting additional generation from renewables
• Enabling active demand/customer side participation in the market
• Keeping the distribution grid stable and balanced by handling electric power flows in both directions.
• Demonstrate and ease the aggregation of demand through VPP is another relevant option to tackle the prosumer and DSO motivation barriers.
In the second use case presented in the deliverable 2.2 apart from scheduling on a certain time ahead basis, an amount of flexibility, ~25% ramping up and 45% ramping down capabilities from originally planned allocation, is available with a response time that is adequate for offering reserve power services to DSO. However a delicate balance between offering for DSO operations and maintaining economic advantage to cost generated from day ahead forecast must be considered. Forecasting of flexibility response, longevity, impact on future profile as well as monitoring and adaptive learning of boundary conditions is necessary to effectively offer aggregated demand response as reserve power. Next steps are to monitor and forecast the longevity of the offered reserve services to better balance flexibility between DSO and supplier services. Grid co-simulations to evaluate the impact on grid for both cases are advised.
The third use case, introducing then a bid ladder model, additional to PowerMatcher operation, offers a light-weight and fast solution to mobilize demand and generation response in case of grid self-healing and stabilising actions following the DREAM architecture. The mechanism is able to evenly distribute the consequences of a power shortage or abundance to each connected device. All these simulations are also providing examples for use and adoption of DREAM.
Finally, still focusing on the prosumer motivation barriers reduction, it is claimed in the last part of the previous study that investing in DER means of flexibility having reduced marginal cost of activation appears as the most suitable solution to pay-back the flexibility infrastructure by an involvement on the wholesale market. Doing so, the flexibility stays moreover still available for the supply of advanced ancillary services at the benefits of the System Operators.
1.2 novel mechanism to compute a distributed contingency analysis
In deliverable 3.1 the scientific barriers to enable such a control were identified and the concepts of the proposed solutions were given.
Some solutions and practical implementations were developed to enable distributed balancing market place at the distribution level. The new operation and control mechanisms are emphasised in D3.2 and their involved initialization inputs, hardware platforms and implementation reliabilities in practice are also considered. These control mechanisms are demonstrated through three Use Cases, LV / MV voltage control and congestion management, and short-term scheduling.
Based on the cell configuration, LV flexibilities are aggregated and applied to eliminate LV voltage constraint violations and current congestions. This is formulated as an optimal control problem, minimizing flexibility activation cost and subject to network constraints. Heuristic approaches, meta-heuristic approaches, and also deterministic approaches were developed to solve this optimization problem. They are tested on a 12-node LV network and compared in terms of computation speed and control cost. The heuristic method based on flexibility efficiency or the genetic algorithm can reach a good compromise between computation speed and control cost. Considering the potential influence of flexibility forecast on control effects, fuzzy logical algorithm is used to improve the forecast accuracy.
A new aggregation solution to use LV flexibility offers as one particular MV flexibility offer among the others in order to solve voltage deviations and congestions at MV level is developed. It first adjusts the admissible voltage ranges at the MV nodes connected only with LV networks to dispatch LV flexibilities, and then takes the new voltage ranges together with MV flexibilities, OLTC and capacitor banks, as control variables to implement an economical optimization to solve network constraints . The results on a feeder from the Electricité de Strasbourg Réseau partner and a 39-node IEEE network show that the developed algorithms can efficiently coordinate different types of LV/MV flexibilities.
Although the operators try their best to make the forecasts of production and demand as accurate as possible, the deviations between both in real time are commonplace, which are dealt with the last part described in deliverable 3.2 with the help of a decentralized gossip algorithm. The gossips of demand and production are propagated between the neighbouring nodes. Each node decides on its own the provision of generation/demand flexibility depending on the utility function. The weight factors are optimized to obtain the satisfying convergence. The results show that the gossip algorithm can curtail excessive generation following to the evolution of power flow. It is also robust and reliable to communication failures and uncertainties of user behaviors.
1.3 self-defense reactions from the controllable devices at distribution network level
In deliverable 4.1 the scientific barriers to enable such a control were identified and the concepts of the proposed solutions were given.
The identified scientific barriers can be summarized as:
• coordination of strategies in different operation mode for a distributed management system,
• appropriate strategies that fully use this distributed management system,
• levels of DSO control,
• robustness and ICT security.
The deliverable 4.2 presents the solutions and practical implementations to enable distributed direct real time control at the distribution level. The first part of it recaps the barriers and describes the outlines and scopes of the solutions to overcome these barriers. The second part focuses on the specific implementation and first results obtained in simulation, which lead to specific use cases that will be tested on the different DREAM trial sites.
Concerning the barrier that is related to the coordination of strategies in different operation modes, an approach is proposed that is inspired by a traffic light. According to the status of the grid (that leads to different phases, green, amber or red in the traffic light) the DSO agent can use different functionalities and implementations to respond to the situation. This approach also should coordinate the interaction between the markets and the grid operation. Further, the distributed agent structure is taken into account by establishing control mechanisms that supervise the mutual interaction of the agents.
For the distributed agent based system, appropriate strategies have been implemented for contingency management and control of the grid. They use the distributed intelligence of the agents in various forms and specifications, ranging from fully local control within the agents, distributed control, that finds solutions and optimises the grid operation while negotiating between several equal agents, and decentralised optimisation, that takes the decisions decentralised, but only in one agent for one distinct grid area. Each of these strategies aims to optimise the grid operation and solve possible problems that occur while operating the grid, like voltage profile deviations, or overloading of lines.
Another strategy that has been developed can be used for the grid reconfiguration to re-establish the service for as much loads as possible after a fault has occurred, or to minimise the power losses.
The barrier “level of DSO control” describes the interaction between the DSO and the distributed management system. The DSO must be in the position to control and supervise and stop if needed all parts of the automatic execution of the operation management and to be able to guarantee stability and quality of service for its clients and safety for its staff.
A solution is proposed concerning the frequency support from the distribution grid and especially the interconnected distributed energy resources (DER). Up to now, in a grid situation with under frequency the full MV distribution feeders are shed one after the other, by the breakers at the primary substation, according to a pre-defined plan. This classic process can deteriorate the situation, when the distribution grids are no longer “just” loads but incorporate large amounts of decentralised production, whose separating from the main grid has the opposite effect as intended. The proposed solution to this is to coordinate the frequency-controlled reserves over the network at any voltage level to dispose enough reserve and to control only valuable power components for the system stability. A second opportunity is then to propose Demand Response solutions to supply these reserves that used to be usually supplied on the generation side. Last but not least, acting on dispatchable loads enable the frequency droop control closer to the normal frequency and secondly to deploy a reversible process of load reconnection. To do so would enable the delivery of a frequency control product mid-way between the Under Frequency Load Shedding reserves and the Frequency Containment Reserves. The dual part, namely the Over Frequency (distributed) Production/Resource (but coordinated) Shedding/Reconnection is as well targeted, using DER accepting to propose frequency droop control of their process at higher frequency than nominal.
Regarding the barrier of “robustness and ICT security”, the benefits of the architectural properties of distributed agent-based solution to lower these barriers through the persistence package and the communication data protocols are discussed.
2. “combine...” phase
In this second phase, the Dream project was tackling the issue of algorithm aggregation in a single framework.
At first, the consortium built the deliverable D5.1 as first step in establishing the framework in the DREAM project. DREAM aims to build and demonstrate an industry-quality reference solution for the best integration and best use of large amount of DG-RES combined with the flexibility of the Distribution Network. This deliverable created a common vocabulary and object model to be used in next steps in construction of the framework for DG-RES aggregation-level control and coordination, based on commonly available ICT components, standards, and platforms for every actor (DRES owners, grid operators, aggregator, etc...) of the Smart Grids. Clear definitions (terminology) are proposed to clarify the outcomes of DREAM.
It allows a clear understanding of the underlying concepts. The approach used in the document was to adopt international standards (IEC) whenever possible in order to be more compliant as possible with current industrial and research trends. This was mandatory overall for an European research project to be able to contribute to the community speaking “the same language”. The requirements engineering approach is a bottom-up one with a direct validation of any distributed decisions by the stakeholders and especially the DSO with a clear and rigorous methodology.
An entire section is devoted to the business as usual and upcoming operating modes of the network and more especially the distribution network, showing the different evolutions.
The cell concept is presented as the first description of the DREAM Framework. This concept will help to clearly define the heterarchical concept relying on grid components and RTUs managed or in direct connection to the DSO.
The end user role and the commercial roles are described (DER at large sense) and these are essential by they participate actively and must be coordinated with the constraints of the network components. From the use cases, the participation can be seen to focus on energy and balancing markets to prepare (plan) and to support (react) the operation of MV and LV distribution networks in support of the Power System;
Finally, the use cases of DREAM provided, are shown to support the view of an ever evolving hierarchy depending on DN needs. The latter is requested to fully use the agent/autonomic computing ability and avoid the instability between several operating modes.
The integration of components of the DREAM framework and the process of adding the provisions for embedding the ‘upstream’ and ‘downstream’ use cases were further developed in the deliverable D5.2. This means, that from the DREAM internal intelligence perspective, the logics of the use cases described in “describe and develop...” phase is implemented, and in the perspective of operation to the outside world in testing circumstances, the interfaces are defined, specific for the particular field tests (“apply...” phase, see section 3) an inventory of components is given. Furthermore, extensions are described to implement the agents to operate as distributed systems and in the primary process context of consumers, producers or prosumers.
Further, following the UML method used in this “combine...” phase, the detailed UML deployment diagrams are sketched for every laboratory and field test site and the way of interfacing to the algorithmic and the primary process intelligence of the electricity consumers, producers and prosumers. Then, the framework extensions required are explained. The architectural aspects of using protocols for inter-process communication are analysed and the inter-agent communication requirements and protocols are formulated leading to a selection of the most suitable candidates. Subsequently, the required source code configuration and component management tools are described and how these are to be used in a multi-developer, multi-organisation environment. Finally, the produced document delves into the transfer process to the industrial validation phase (3). An implementation of a joint use case, covering grid operation in normal, critical and emergency mode, is as well described.
3. “apply...” phase
In this third phase, the Dream framework is tested, see subsection 3.1 and analyses. Indeed, due to the specificities of the trial sites proposed by the partners, a clear methodology for test analysis was build, see section 3.2.
3.1 Tests
In the merged deliverable 7-9.3 the results from the developed solutions applied in the various pilot sites are presented. The scope of the tests was to demonstrate the economic and technical feasibility of the novel control mechanisms developed in DREAM project. The real-word small-scale pilots that participated in the DREAM demonstration activities tested different use cases depending on both their goals as well as the available infrastructure of the specific test site. The demo sites that participated in DREAM are described next.
The DSO of Strasbourg (Electricité de Strasbourg Réseaux) has as goals to improve the quality of Service and the customer Satisfaction as well as to apply local, real time management of MV&LV grids integrating renewable sources. The infrastructure of the pilot comprises 4 primary substations (63/20 kV), 6 secondary substations (20 kV/410 V) with DREAM advanced RTUs and SDSL boxes, Controllable PV-producer (250 kW peak) & Energy Boxes, fast charging stations for electric vehicles (4x22 kW).
The Hellenic Electricity Distribution Network Operator (HEDNO) has as objectives to reduce the voltage profile variability in LV level and to enable aggregation and provision of local customer flexibility to enable their participation in national markets. The demo sites are a LV seaside camping settlement on the mainland, 1 secondary substation with a DREAM advanced RTU (executing JAVA) and the controllable LV devices in the households that participated in the demos and the simulations in the system of Crete island.
The SEA Aeroporti di Milano aims at optimizing the market operation of the airport. The trial site consists of 2 out of the 8 rings of the Malpensa Airport MV grid, 3 secondary substations (MV/LV), 1 secondary substation equipped with DREAM advanced RTU, and the following flexible devices 7 HVAC systems, 6 sets of apron lighting towers, control rooms at SEA and at the subsidiary SEA Energia.
The DNV-GL facilities had the objective to improve short-term security of supply in LV level and to reduce the dependency from general balance market by using local flexibility. The trial infrastructure is the Opkamer test site in the Netherlands.
The Grenoble INP facilities had the objective to prove the feasibility of distributed frequency support to compete with centralized Frequency Containment Reserve and to support Under Frequency Load Shedding. The lab trial infrastructure is located in Grenoble. Indeed, in order to validate self-defense reaction based on frequency deviation, power amplifier and real time simulations are mandatory to test abnormal operations.
3.2 Proposed analysis methodology and road mapping of Dream concept
The deliverable 6.2 provides the results of an industrial viability analysis on DREAM. A business oriented review of the framework based on the demonstrator implementation.
In order to assess the industrial viability of the DREAM framework, several demonstrations have been carried out during the project. Those demonstrations are in two ways providing meaningful and relevant information for this study. On one hand, they provide tangible results, which can be measured, compared and analysed, to foresee future benefits and costs for DREAM users. On the other hand, the proof of concept demonstration is not far from a real DREAM framework deployment, and real barriers showed during the demonstration phase. The lessons learned and cost of solutions provided are a first instance approach to real costs for implementation and recommended implementations.
From the viability assessment perspective, the chosen roadmap followed a standard methodology to define KPIs based on user and use case goals. The KPIs selected were then paired with the EEGI roadmap. The results found and based on that the KPIs during the demonstration phase needed to be measurable, objective and comparable. Not only comparable within the DREAM project results, but also with other alternatives.
One of the main concerns and as such, goals for every single actor in the electricity industry is to deliver sustainable, economic and secure electricity to the end-user. From DREAM’s project perspective, this can only be done by increased penetration of distributed energy resources (DERs), through an active distribution system management
In order to achieve the best possible optimisation, DREAM proposes to implement the heterarchical coordination of the grid entities using decentralized techniques, dynamic reconfiguration and aggregation. Promoting advantages such as avoidance of the single point-of-failure, reduced computation and communication cost, privacy of the end-user, but most importantly increased scalability and the “Plug-and-Play” capability.
Some tools shall be provided, as a common space to take maximum advantage of the benefits of heterarchical management, since traditional tools are centralized control oriented. DREAM proposal is to create new marketplaces at the distribution level of the power system, with different time scopes, aiming to produce the maximum optimization from the available information granularity.
Contingencies are solved primarily locally based on the concept of traded customer flexibility, and the heterarchical control layer can be smoothly integrated in more traditional and centralized control architectures.
The installation of decentralized agent-based systems (MAS) is not only providing a proven advantage in terms of reliability and scalability, it is also, as required by the efficiency and cost optimization policies, a way to get the most from the installed hardware base. As electronics are becoming cheaper and enhanced RTUs are appearing in the market, not doing so would be an inefficient way to escalate the grid by wasting spare computational capabilities.
In terms of cost and benefit, the heterarchical approach has emerged as an opportunity to take the maximum benefit from existing or future infrastructure, enhance scalability and keep maximum compatibility with existing centralized solutions.
The roadmap for Dream concept deployment is presented in deliverable 6.2. The vision for 2030 sets the required properties for the scalable and deployable coordination and control of the DERs in distribution networks. The Roadmap defines the objectives for three time horizons – the near, mid and long term. For each time horizon, objectives are given. Finally, the needs expressed by the stakeholders and the barriers that are the most probable difficulties and obstacles to be overcome, are analysed. The four categories are: Technical and industrial, Regulatory, Scientific and Societal.
Potential Impact:
Based on the result description model, the DREAM exploitable results have been extensively described and documented in deliverable D10.1. The overall exploitation structure model, industrial viability, road to market, the specific business model and exploitation plans have been developed.
The exploitable results address the different levels of the DREAM contribution to the new paradigm of smart grids for energy distribution, from high level models to real life demonstrated methods and tools.
The common view achieved through the collaborative definition of results and of depositaries as well as users and exploiters offers the foundations on which to build individual and joint strategies for the sustainable and profitable use of project outcomes.
The updated list of results at end of project establishes the conceptual baseline to:
- Improve the focus of post-project activities aimed at increasing the maturity on individual exploitable results
- Support the development of focused awareness on results among identified stakeholders and target users, driving specific communication actions tailored to anticipated impact on different classes.
The assessment of industrial exploitation viability based on estimates for costs/ risks/ benefits in smart grids operations and the identification of a technological roadmap to market allows for reducing the enterprise risks in committing to the full achievement of anticipated impacts from the project.
The sharing of individual use and exploitation intentions at the end of project, as well as of partners’ relevant business strategies through homogeneously structured business models ensures that all partners can effectively interact to plan for actual use and exploitation planning, leveraging knowledge and synergies with other partners and possibly pursuing post-project collaboration schemes with other project beneficiaries.
Finally, the DREAM website, with modern look and feel and based on responsive technology supporting multiple devices, has been regularly maintained with updated contents.
The conferences, papers & others activities related to DREAM are available on http://www.dream-smartgrid.eu/downloads/.
List of Websites:
http://www.dream-smartgrid.eu/
The DREAM project launches a novel heterarchical management approach of complex electrical power grids, providing new mechanisms for stable and cost effective integration of distributed renewable energy sources, as well as for enhanced consumer involvement in economic and ecological electricity use.
The principles of autonomous agent-based systems are applied to the control and management of the electricity distribution grid, helping the system to constantly adjust to actual operational conditions and increasing its robustness versus exogenous disturbances. In turn, this allows for greater penetration of intermittent resources and makes the distribution grid more resilient to failures. In order to improve involvement of citizens in the grid management, DREAM implements market validation processes from the end user towards the transmission system delivery point. DREAM includes layers of controls for normal, congested and post-contingency situations according to a traffic light model, using coordination strategies to create ad-hoc federations of agents that flexibly adjust their hierarchical configuration to distribution grid needs.
The system transits smoothly between dynamic control layers depending on local operational conditions, so that responses to disturbances are sized precisely, margins are used parsimoniously and full network flexibility is tapped. The system involves limited data transfers, and promotes extensibility, heterogeneity and easy deployment across Countries with different network architectures and hardware manufacturers.
DREAM demonstrated the economic and technical feasibility of these novel control mechanisms thanks to several real-world small-scale pilots dedicated to different use-cases, and computer simulations have been used to further assess the solutions scalability.
Twenty four years after Rio92, six years after COP15 in Copenhagen and a year after COP21 in Paris, the international community duly notes the need for a new energy paradigm.Such an energy transition would induce the migration from an energy mix driven by conventional and carbonated bulk generation assets to a new paradigm relying on partly distributed renewable assets. As a direct consequence, the amount of conventional capacities naturally available would decrease, while the constraints on the distribution networks would rise.
The principle of aggregation of Distributed Energy Resources (DERs) emerged in previous years, with the objective to propose new solutions to handle such issues.
Project Context and Objectives:
DREAM was developing a novel paradigm for smart grids, capable of responding to current issues and challenges related to renewable integration and opening of the energy markets, and is particularly focused on three main goals in the evolution of energy distribution environment and of its interactions with energy transportation systems and with final users and energy stakeholders:
• Ensuring societal benefits in terms of environmental impact / carbon footprint, in line with the 2020 objectives, and of overall economy of electric systems; this objective is associated with the substantial increase of capabilities to integrate distributed renewable resources, including the intermittent ones like Photovoltaic and Wind, as well as with the reduction of energy losses along the lines and the improvement of efficiency for power plants from reduction of fluctuations in the demand.
• Accommodating smooth and sustainable transition to the new smart grid environment; minimizing drawbacks from the deployment of novel smart grids, through the capability of ensuring economic sustainability of investments, the flexibility in accommodating variable configurations of energy networks, and the scalability to cover millions of smart energy entities;
• Enhancing flexibility and robustness of technical and operational environment for energy network; leveraging the advantages of smart devices in generation, use and storage to improve response capabilities of distribution subnets, dynamic coordination of behaviour to minimize energy flows across subnets, innovative ancillary servicing capabilities and increased resilience to undesired events.
The developed DREAM framework and devices are strengthening power distribution grids robustness and efficiency by developing a novel autonomous, self-learning, heterarchical and agent-based approach to the control and management of the power grid. The project was targeted at the creation and management of several innovative autonomous control mechanisms for energy distribution that should be open, extensible, compatible with heterogeneous hardware components and compliant with existing national energy exchange regulations.
DREAM was developing control mechanisms to allow for the exploitation of all network entities – generators, loads, storage units, tap changers, voltage regulators, controllable switches... – by identifying the relevant behavioural profiles and integrating them into a more general control environment, which accommodates for minimisation of energy unbalances and other technical constraints violations, and thus leads to the following advantages:
• Improved robustness to exogenous perturbations to increase reliability. In addition, the system is including self-defense mechanisms that make it resilient to threats posed to its own integrity, such as faults and equipment failure.
• As a by-product of increased robustness, the system is more tolerant to disturbances such as fluctuating injections of active and reactive power by intermittent generation units; this is allowing accommodating more small-scale renewable resources in the generation mix, allowing for cleaner and more local electric energy production.
• More efficient utilisation of existing and new generation, transmission and distribution equipment, thanks to the reduction of range in operational conditions to be accommodated. It makes it possible to minimize excess capacity expenditures and to operate assets at their maximum efficiency.
The vision described above was achieved by integrating the coarse control of system objectives (such as frequency and transmission system voltages) at TSO level with the local control capabilities of entities constituting, or connected to, the distribution grid. DSOs thus could able to validate flexibilities (that is to say, residual flexibility that remains available after distribution level targets such as volt-var control, congestion management or loss minimization have been achieved satisfactorily) for market purpose use or mechanisms in case of emergency situations.
The control tools are the operational support components of the mechanism. Decision-making follows the principles of autonomous control and of coordinated behaviour, which translate into concrete actions such as generating or consuming active and reactive power, as well as operation of grid devices such as remotely controlled switches and on load tap changers.
The functions are designed for maximum automation while letting the DSO benefit from increased situational awareness with the option to take over the automated control mechanisms from the control centre. From this viewpoint, the DREAM system acts as an intermediary layer (wide area feeder automation) between the DSO and the network that filters out the important information (e.g. outliers detection, contingency anticipation) that request attention from human operators, while ensuring that business-as-usual operations are carried out automatically.
The agent-based heterarchical control addresses the logical and business characteristics of the grid and accommodates for real time, autonomous control of the distributed and heterogeneous energy generation and consumption nodes and hubs. The power grid of the future will be capable of ensuring the efficient integration and coordination of a composite network of smart nodes and hubs, which may be basic individual entities or aggregations which respond to different decision making mechanisms and principles and have various control capacities, but respond to rules and regulations that are enforced through the establishment of physical (e.g. breakers settings) or economic (e.g. non validation of offers) mechanisms. The agent-based system is pursuing efficient integration of existing and new players and technologies and leverage new control mechanisms and decision making flexibility, by discarding the pure hierarchical distribution model; instead, global objectives are pursued by optimising the local behaviour at the level of user nodes, control nodes and so on. Such a heterarchical approach is capable of addressing multi-composite nodes at physical (actual location, type and intrinsic characteristics), logical (behavioural profile, decision rules and controllability of performances) and business (advantage conditions, benefit sharing strategies, incentive policies exploitation) levels.
The heterarchical smart power grid envisioned and then built by the DREAM consortium is a heterarchical, flexible, dynamic and adaptive system. The distributed actors currently display little to no coordination amongst each other – DREAM recognises that decision-making in the grid occurs at the level of these individual actors. This results in highly flexible and dynamic processes for which agent-based models are uniquely suitable. On top of this, the distribution system itself – as well as its relationships with neighbouring entities such as the transmission system – needs to be included in the control loop and provided with adequate support both for its current, hierarchical architecture and control mechanisms, and for future locally control-capability-enabled distribution nodes.
Scientific and Technical Objectives
Detailed and measurable objectives for the DREAM project contribute to the goals described above. They include objectives related to the establishment of best conditions for successful take up and exploitation of project.
Main scientific and technical objectives the DSOs are thus:
• Develop a consistent concept for local energy market portal, compliant with national and international grid codes and networks across European Countries;
• Develop methods and tools for aggregating energy network entities into clusters capable of buying and offering power and ancillary services, for evaluating associated energy performance potential, and for fair and well performing allocation of roles;
• Develop logical methods and support tools to identify dynamic clusters of users to respond to contingent/emergency conditions and to coordinate individual behaviours;
• Define an adequate number of Key performance indicators to measure actual behaviour of new distribution grids mechanisms across trading, normal and contingency operation conditions
• Conduct a sufficient number of tests to ensure a confidence in the order of 90% on viability of concept, computed on the basis of measured values key performance indicators;
• Ensure proper and adequate dissemination of results, such as major stakeholders for new energy distribution grids, public administrations relevant to smart grids, industrial users of novel concepts, plus academy and research centers and the wider public for awareness rising.
Project Results:
The DREAM project is laying the foundations for a novel heterarchical management approach of complex electrical power grids, providing new mechanisms for stable and cost effective integration of distributed renewable energy sources, as well as for enhanced consumer involvement in economic and ecological electricity use.
Applying the principles of autonomous agent-based systems to the control and management of the electricity distribution grid is allowing the system to constantly adjust to current operational conditions and make it robust to exogenous disturbances. In turn, this is allowing a greater penetration of intermittent resources and is making the distribution grid more resilient to failures. DREAM is including several layers of controls for normal, congested and post-contingency situations that are using different coordination strategies ranging from market-based transactions to emergency demand response and create ad-hoc federations of agents that are flexibly adjusting their hierarchy to current needs.
The system is thus transiting smoothly between control layers depending on local operational conditions, so that responses to disturbances are sized precisely, margins are used parsimoniously and full network flexibility is tapped. The system involves only limited data transfers and no centralized control, promoting extensibility, heterogeneity and easy deployment across countries with different network architectures and hardware manufacturers.
The Project has been phased out in three major parts, as seen in Figure 1. The first part related to the mandatory research development was initiated in the “describe and develop...” phase, as described in section 1 of this description. In this phase, several algorithms were developed to cope with defined uses cases by the research institute of the Dream project in interaction with the final users. Secondly, the “combine...” phase was initiated in order to build and integrated the Dream framework. Indeed, the partners were cautious about interactions between distributed algorithms handling various grid constraints and acting on the same parameters. In order to assess and avoid oscillations, the integrated framework, that is described in part 2, was initiated for test and development. Lastly, the “apply...” phase was conducted in order to validate the concept on trials and evaluate the added value of the Dream concept.
1. “describe and develop...” phase
In this first phase, the work was divided in three different parts, belonging to the operation phase the distribution network is facing. This split can be seen as, at first, a distributed validation of business as usual energy and capacity bids or market operations in a normal operation mode, as described in subsection 1.1. The second part is related to a novel mechanism to compute a distributed contingency analysis. In this phase, the network is considered as probable to face emergency operation mode. This analysis, as any operational planning, is referring to the ability for the system to face most probable contingencies (what if?) and validate if the remaining flexibilities, combined with network operation can solve them with regular contracts, as described in subsection 1.2. The third part is related to critical operation mode, when the Distribution System is facing failures. This mode require self-defense reactions of any possible devices (from end user flexibilities to controllable grid components) to solve the critical situations, as presented in subsection 1.3.
1.1 Distributed validation of business as usual energy and capacity bids and market operations
In deliverable 2.1 the scientific barriers to enable such a control were identified and the concepts of the proposed solutions were given.
The Dream project issued some solutions and practical implementations to tackle the main barriers identified to enable a distribution market place at the distribution level. Three use cases have been described that build the frame.
The developed MV Aggregator model, described in the D2.2 facilitates the task of the MV Aggregator in choosing the optimal price levels to be announced to the DERs by transforming, in an equivalent way, the interdependence of the decisions into mathematical expressions defining a problem that can be solved, after applying proper linearization, with commercially available software. In this model, the aggregator entity manages several local entities, producing a bi-level formulation that facilitates the problem resolution. This approach is perfectly matching the heterarchical multi-agent architecture proposed in DREAM, and will directly affect the identified barrier of DSO motivation. The adoption of this kind of solution will help to cope with the three key challenges for a DSO:
• Connecting additional generation from renewables
• Enabling active demand/customer side participation in the market
• Keeping the distribution grid stable and balanced by handling electric power flows in both directions.
• Demonstrate and ease the aggregation of demand through VPP is another relevant option to tackle the prosumer and DSO motivation barriers.
In the second use case presented in the deliverable 2.2 apart from scheduling on a certain time ahead basis, an amount of flexibility, ~25% ramping up and 45% ramping down capabilities from originally planned allocation, is available with a response time that is adequate for offering reserve power services to DSO. However a delicate balance between offering for DSO operations and maintaining economic advantage to cost generated from day ahead forecast must be considered. Forecasting of flexibility response, longevity, impact on future profile as well as monitoring and adaptive learning of boundary conditions is necessary to effectively offer aggregated demand response as reserve power. Next steps are to monitor and forecast the longevity of the offered reserve services to better balance flexibility between DSO and supplier services. Grid co-simulations to evaluate the impact on grid for both cases are advised.
The third use case, introducing then a bid ladder model, additional to PowerMatcher operation, offers a light-weight and fast solution to mobilize demand and generation response in case of grid self-healing and stabilising actions following the DREAM architecture. The mechanism is able to evenly distribute the consequences of a power shortage or abundance to each connected device. All these simulations are also providing examples for use and adoption of DREAM.
Finally, still focusing on the prosumer motivation barriers reduction, it is claimed in the last part of the previous study that investing in DER means of flexibility having reduced marginal cost of activation appears as the most suitable solution to pay-back the flexibility infrastructure by an involvement on the wholesale market. Doing so, the flexibility stays moreover still available for the supply of advanced ancillary services at the benefits of the System Operators.
1.2 novel mechanism to compute a distributed contingency analysis
In deliverable 3.1 the scientific barriers to enable such a control were identified and the concepts of the proposed solutions were given.
Some solutions and practical implementations were developed to enable distributed balancing market place at the distribution level. The new operation and control mechanisms are emphasised in D3.2 and their involved initialization inputs, hardware platforms and implementation reliabilities in practice are also considered. These control mechanisms are demonstrated through three Use Cases, LV / MV voltage control and congestion management, and short-term scheduling.
Based on the cell configuration, LV flexibilities are aggregated and applied to eliminate LV voltage constraint violations and current congestions. This is formulated as an optimal control problem, minimizing flexibility activation cost and subject to network constraints. Heuristic approaches, meta-heuristic approaches, and also deterministic approaches were developed to solve this optimization problem. They are tested on a 12-node LV network and compared in terms of computation speed and control cost. The heuristic method based on flexibility efficiency or the genetic algorithm can reach a good compromise between computation speed and control cost. Considering the potential influence of flexibility forecast on control effects, fuzzy logical algorithm is used to improve the forecast accuracy.
A new aggregation solution to use LV flexibility offers as one particular MV flexibility offer among the others in order to solve voltage deviations and congestions at MV level is developed. It first adjusts the admissible voltage ranges at the MV nodes connected only with LV networks to dispatch LV flexibilities, and then takes the new voltage ranges together with MV flexibilities, OLTC and capacitor banks, as control variables to implement an economical optimization to solve network constraints . The results on a feeder from the Electricité de Strasbourg Réseau partner and a 39-node IEEE network show that the developed algorithms can efficiently coordinate different types of LV/MV flexibilities.
Although the operators try their best to make the forecasts of production and demand as accurate as possible, the deviations between both in real time are commonplace, which are dealt with the last part described in deliverable 3.2 with the help of a decentralized gossip algorithm. The gossips of demand and production are propagated between the neighbouring nodes. Each node decides on its own the provision of generation/demand flexibility depending on the utility function. The weight factors are optimized to obtain the satisfying convergence. The results show that the gossip algorithm can curtail excessive generation following to the evolution of power flow. It is also robust and reliable to communication failures and uncertainties of user behaviors.
1.3 self-defense reactions from the controllable devices at distribution network level
In deliverable 4.1 the scientific barriers to enable such a control were identified and the concepts of the proposed solutions were given.
The identified scientific barriers can be summarized as:
• coordination of strategies in different operation mode for a distributed management system,
• appropriate strategies that fully use this distributed management system,
• levels of DSO control,
• robustness and ICT security.
The deliverable 4.2 presents the solutions and practical implementations to enable distributed direct real time control at the distribution level. The first part of it recaps the barriers and describes the outlines and scopes of the solutions to overcome these barriers. The second part focuses on the specific implementation and first results obtained in simulation, which lead to specific use cases that will be tested on the different DREAM trial sites.
Concerning the barrier that is related to the coordination of strategies in different operation modes, an approach is proposed that is inspired by a traffic light. According to the status of the grid (that leads to different phases, green, amber or red in the traffic light) the DSO agent can use different functionalities and implementations to respond to the situation. This approach also should coordinate the interaction between the markets and the grid operation. Further, the distributed agent structure is taken into account by establishing control mechanisms that supervise the mutual interaction of the agents.
For the distributed agent based system, appropriate strategies have been implemented for contingency management and control of the grid. They use the distributed intelligence of the agents in various forms and specifications, ranging from fully local control within the agents, distributed control, that finds solutions and optimises the grid operation while negotiating between several equal agents, and decentralised optimisation, that takes the decisions decentralised, but only in one agent for one distinct grid area. Each of these strategies aims to optimise the grid operation and solve possible problems that occur while operating the grid, like voltage profile deviations, or overloading of lines.
Another strategy that has been developed can be used for the grid reconfiguration to re-establish the service for as much loads as possible after a fault has occurred, or to minimise the power losses.
The barrier “level of DSO control” describes the interaction between the DSO and the distributed management system. The DSO must be in the position to control and supervise and stop if needed all parts of the automatic execution of the operation management and to be able to guarantee stability and quality of service for its clients and safety for its staff.
A solution is proposed concerning the frequency support from the distribution grid and especially the interconnected distributed energy resources (DER). Up to now, in a grid situation with under frequency the full MV distribution feeders are shed one after the other, by the breakers at the primary substation, according to a pre-defined plan. This classic process can deteriorate the situation, when the distribution grids are no longer “just” loads but incorporate large amounts of decentralised production, whose separating from the main grid has the opposite effect as intended. The proposed solution to this is to coordinate the frequency-controlled reserves over the network at any voltage level to dispose enough reserve and to control only valuable power components for the system stability. A second opportunity is then to propose Demand Response solutions to supply these reserves that used to be usually supplied on the generation side. Last but not least, acting on dispatchable loads enable the frequency droop control closer to the normal frequency and secondly to deploy a reversible process of load reconnection. To do so would enable the delivery of a frequency control product mid-way between the Under Frequency Load Shedding reserves and the Frequency Containment Reserves. The dual part, namely the Over Frequency (distributed) Production/Resource (but coordinated) Shedding/Reconnection is as well targeted, using DER accepting to propose frequency droop control of their process at higher frequency than nominal.
Regarding the barrier of “robustness and ICT security”, the benefits of the architectural properties of distributed agent-based solution to lower these barriers through the persistence package and the communication data protocols are discussed.
2. “combine...” phase
In this second phase, the Dream project was tackling the issue of algorithm aggregation in a single framework.
At first, the consortium built the deliverable D5.1 as first step in establishing the framework in the DREAM project. DREAM aims to build and demonstrate an industry-quality reference solution for the best integration and best use of large amount of DG-RES combined with the flexibility of the Distribution Network. This deliverable created a common vocabulary and object model to be used in next steps in construction of the framework for DG-RES aggregation-level control and coordination, based on commonly available ICT components, standards, and platforms for every actor (DRES owners, grid operators, aggregator, etc...) of the Smart Grids. Clear definitions (terminology) are proposed to clarify the outcomes of DREAM.
It allows a clear understanding of the underlying concepts. The approach used in the document was to adopt international standards (IEC) whenever possible in order to be more compliant as possible with current industrial and research trends. This was mandatory overall for an European research project to be able to contribute to the community speaking “the same language”. The requirements engineering approach is a bottom-up one with a direct validation of any distributed decisions by the stakeholders and especially the DSO with a clear and rigorous methodology.
An entire section is devoted to the business as usual and upcoming operating modes of the network and more especially the distribution network, showing the different evolutions.
The cell concept is presented as the first description of the DREAM Framework. This concept will help to clearly define the heterarchical concept relying on grid components and RTUs managed or in direct connection to the DSO.
The end user role and the commercial roles are described (DER at large sense) and these are essential by they participate actively and must be coordinated with the constraints of the network components. From the use cases, the participation can be seen to focus on energy and balancing markets to prepare (plan) and to support (react) the operation of MV and LV distribution networks in support of the Power System;
Finally, the use cases of DREAM provided, are shown to support the view of an ever evolving hierarchy depending on DN needs. The latter is requested to fully use the agent/autonomic computing ability and avoid the instability between several operating modes.
The integration of components of the DREAM framework and the process of adding the provisions for embedding the ‘upstream’ and ‘downstream’ use cases were further developed in the deliverable D5.2. This means, that from the DREAM internal intelligence perspective, the logics of the use cases described in “describe and develop...” phase is implemented, and in the perspective of operation to the outside world in testing circumstances, the interfaces are defined, specific for the particular field tests (“apply...” phase, see section 3) an inventory of components is given. Furthermore, extensions are described to implement the agents to operate as distributed systems and in the primary process context of consumers, producers or prosumers.
Further, following the UML method used in this “combine...” phase, the detailed UML deployment diagrams are sketched for every laboratory and field test site and the way of interfacing to the algorithmic and the primary process intelligence of the electricity consumers, producers and prosumers. Then, the framework extensions required are explained. The architectural aspects of using protocols for inter-process communication are analysed and the inter-agent communication requirements and protocols are formulated leading to a selection of the most suitable candidates. Subsequently, the required source code configuration and component management tools are described and how these are to be used in a multi-developer, multi-organisation environment. Finally, the produced document delves into the transfer process to the industrial validation phase (3). An implementation of a joint use case, covering grid operation in normal, critical and emergency mode, is as well described.
3. “apply...” phase
In this third phase, the Dream framework is tested, see subsection 3.1 and analyses. Indeed, due to the specificities of the trial sites proposed by the partners, a clear methodology for test analysis was build, see section 3.2.
3.1 Tests
In the merged deliverable 7-9.3 the results from the developed solutions applied in the various pilot sites are presented. The scope of the tests was to demonstrate the economic and technical feasibility of the novel control mechanisms developed in DREAM project. The real-word small-scale pilots that participated in the DREAM demonstration activities tested different use cases depending on both their goals as well as the available infrastructure of the specific test site. The demo sites that participated in DREAM are described next.
The DSO of Strasbourg (Electricité de Strasbourg Réseaux) has as goals to improve the quality of Service and the customer Satisfaction as well as to apply local, real time management of MV&LV grids integrating renewable sources. The infrastructure of the pilot comprises 4 primary substations (63/20 kV), 6 secondary substations (20 kV/410 V) with DREAM advanced RTUs and SDSL boxes, Controllable PV-producer (250 kW peak) & Energy Boxes, fast charging stations for electric vehicles (4x22 kW).
The Hellenic Electricity Distribution Network Operator (HEDNO) has as objectives to reduce the voltage profile variability in LV level and to enable aggregation and provision of local customer flexibility to enable their participation in national markets. The demo sites are a LV seaside camping settlement on the mainland, 1 secondary substation with a DREAM advanced RTU (executing JAVA) and the controllable LV devices in the households that participated in the demos and the simulations in the system of Crete island.
The SEA Aeroporti di Milano aims at optimizing the market operation of the airport. The trial site consists of 2 out of the 8 rings of the Malpensa Airport MV grid, 3 secondary substations (MV/LV), 1 secondary substation equipped with DREAM advanced RTU, and the following flexible devices 7 HVAC systems, 6 sets of apron lighting towers, control rooms at SEA and at the subsidiary SEA Energia.
The DNV-GL facilities had the objective to improve short-term security of supply in LV level and to reduce the dependency from general balance market by using local flexibility. The trial infrastructure is the Opkamer test site in the Netherlands.
The Grenoble INP facilities had the objective to prove the feasibility of distributed frequency support to compete with centralized Frequency Containment Reserve and to support Under Frequency Load Shedding. The lab trial infrastructure is located in Grenoble. Indeed, in order to validate self-defense reaction based on frequency deviation, power amplifier and real time simulations are mandatory to test abnormal operations.
3.2 Proposed analysis methodology and road mapping of Dream concept
The deliverable 6.2 provides the results of an industrial viability analysis on DREAM. A business oriented review of the framework based on the demonstrator implementation.
In order to assess the industrial viability of the DREAM framework, several demonstrations have been carried out during the project. Those demonstrations are in two ways providing meaningful and relevant information for this study. On one hand, they provide tangible results, which can be measured, compared and analysed, to foresee future benefits and costs for DREAM users. On the other hand, the proof of concept demonstration is not far from a real DREAM framework deployment, and real barriers showed during the demonstration phase. The lessons learned and cost of solutions provided are a first instance approach to real costs for implementation and recommended implementations.
From the viability assessment perspective, the chosen roadmap followed a standard methodology to define KPIs based on user and use case goals. The KPIs selected were then paired with the EEGI roadmap. The results found and based on that the KPIs during the demonstration phase needed to be measurable, objective and comparable. Not only comparable within the DREAM project results, but also with other alternatives.
One of the main concerns and as such, goals for every single actor in the electricity industry is to deliver sustainable, economic and secure electricity to the end-user. From DREAM’s project perspective, this can only be done by increased penetration of distributed energy resources (DERs), through an active distribution system management
In order to achieve the best possible optimisation, DREAM proposes to implement the heterarchical coordination of the grid entities using decentralized techniques, dynamic reconfiguration and aggregation. Promoting advantages such as avoidance of the single point-of-failure, reduced computation and communication cost, privacy of the end-user, but most importantly increased scalability and the “Plug-and-Play” capability.
Some tools shall be provided, as a common space to take maximum advantage of the benefits of heterarchical management, since traditional tools are centralized control oriented. DREAM proposal is to create new marketplaces at the distribution level of the power system, with different time scopes, aiming to produce the maximum optimization from the available information granularity.
Contingencies are solved primarily locally based on the concept of traded customer flexibility, and the heterarchical control layer can be smoothly integrated in more traditional and centralized control architectures.
The installation of decentralized agent-based systems (MAS) is not only providing a proven advantage in terms of reliability and scalability, it is also, as required by the efficiency and cost optimization policies, a way to get the most from the installed hardware base. As electronics are becoming cheaper and enhanced RTUs are appearing in the market, not doing so would be an inefficient way to escalate the grid by wasting spare computational capabilities.
In terms of cost and benefit, the heterarchical approach has emerged as an opportunity to take the maximum benefit from existing or future infrastructure, enhance scalability and keep maximum compatibility with existing centralized solutions.
The roadmap for Dream concept deployment is presented in deliverable 6.2. The vision for 2030 sets the required properties for the scalable and deployable coordination and control of the DERs in distribution networks. The Roadmap defines the objectives for three time horizons – the near, mid and long term. For each time horizon, objectives are given. Finally, the needs expressed by the stakeholders and the barriers that are the most probable difficulties and obstacles to be overcome, are analysed. The four categories are: Technical and industrial, Regulatory, Scientific and Societal.
Potential Impact:
Based on the result description model, the DREAM exploitable results have been extensively described and documented in deliverable D10.1. The overall exploitation structure model, industrial viability, road to market, the specific business model and exploitation plans have been developed.
The exploitable results address the different levels of the DREAM contribution to the new paradigm of smart grids for energy distribution, from high level models to real life demonstrated methods and tools.
The common view achieved through the collaborative definition of results and of depositaries as well as users and exploiters offers the foundations on which to build individual and joint strategies for the sustainable and profitable use of project outcomes.
The updated list of results at end of project establishes the conceptual baseline to:
- Improve the focus of post-project activities aimed at increasing the maturity on individual exploitable results
- Support the development of focused awareness on results among identified stakeholders and target users, driving specific communication actions tailored to anticipated impact on different classes.
The assessment of industrial exploitation viability based on estimates for costs/ risks/ benefits in smart grids operations and the identification of a technological roadmap to market allows for reducing the enterprise risks in committing to the full achievement of anticipated impacts from the project.
The sharing of individual use and exploitation intentions at the end of project, as well as of partners’ relevant business strategies through homogeneously structured business models ensures that all partners can effectively interact to plan for actual use and exploitation planning, leveraging knowledge and synergies with other partners and possibly pursuing post-project collaboration schemes with other project beneficiaries.
Finally, the DREAM website, with modern look and feel and based on responsive technology supporting multiple devices, has been regularly maintained with updated contents.
The conferences, papers & others activities related to DREAM are available on http://www.dream-smartgrid.eu/downloads/.
List of Websites:
http://www.dream-smartgrid.eu/