Crisp, distributed intelligence in critical infrastructure for sustainable power (CRISP)

An overview about EPS, market and ICT architectures and expected performances is reported. A view by CRISP partners of various architectures (Sweden, the Netherlands and France) has been given. The electric system is changing today and the rules are not completely defined. The competition and DG insertion is increasing in most of the countries, step by step. A new concept of 'cell' has been proposed in the document. The proposed separation into sub-networks, the natural area from topology configuration point of view, is extended for all the functions that could be automated in real time in the future. One of the main ideas is that the local agent downloads and updates its part of electric map of the network (specific link with a data base located in a regional operation control room), using local communications and local analysis to make local decision (on changing the configuration or on controlling loads, DG or specific devices). The concept of cell proposed may help to coordinate the different levels of ICT access inside the network, enabling a global parallelisation of the complex and real time work. A crucial key for the future is evaluating the real reliability of a given ICT distribution set.

An overview of the current power markets and hierarchic grid architectures and the role of demand side and supply management therein is given. Further, a framework is presented for the application of new ICTs and agent technology for integrated Supply and Demand Matching in bottom-up high-RES distributed generation environments. Using this framework, four sample scenarios are defined, spanning a bottom-up approach for power distribution networks as well as for bottom-up markets. These results serve as an input for WP2-simulations and WP3-tests in the CRISP-project.

The elementary Electric Power System cell chosen for the experiment is illustrative: electricity and information are distributed and coordinated by the agent called Smart Grid Automation Device (SGAD) or HTFD for the fault diagnosis application. The measured communication streams underline the importance of the data flow rate for the message transmission. The exchanges between the PCs and the various parts of programs are done with interface files: easy re-use and interoperability. The programs may be applied with various kinds of Electric Power System topology and ICT structures. The real time closed loop with Arene has been achieved, and tested with protection system based on automatic re-closing sequences. The application shows that the fault may be isolated correctly in less than one minute, even with low data flow rate. With a 100% cable installation a possible isolation of the faulty section may be achieved in less than 10s. These results reduce drastically the future interruption in the distribution EPS. TCP/IP protocol is a right choice for a cheap future communication box adapted to real time EPS applications. The experimentation shows the high interest of developing a set of communication boxes and programs in a modular way.

A detailed description of the power system is given. The various controls existing in the system are detailed in order to analyse the different aspects of system robustness. This approach is generally expressed in transmission and has a strong link with blackout countermeasures: the major events and defence plans are described and explained. Various levels of DG penetration are investigated for analysing the technical impacts on the networks and the consequences for the robustness. A specific criterion is proposed for having a clear and simple index indicating the real time situation about stability. This indicator may be used in the future in a distributed intelligence way: the real time evaluation of stability is done cell by cell and aggregated at a higher level. In this way a more complex stability study is pointed out with the scalability parameter, taking into account at the same time the local and global constraints. Another specific point from the large DG insertion in the distribution network is the future capacity of intentional islanding. The needs for control and communication are detailed for various power system applications at local scale.

A concept of MV and LV cells is described to take into account the various cases of network configuration and faults that may occur. The future network is expected to be more flexible and variable in its configuration, keeping a relatively fixed topology and a typical radial operation. Various kinds of faults are described and the possible influence of DG units is given. A main scope of the result is focused on the localization of a permanent fault. New kinds of devices, or the performances of existing devices, and new fast fault localization methods are proposed. A great improvement is expected for the existing network with an inherent adaptation to DG penetration. The solution proposed for the future medium voltage distribution network combines the information from the distributed fault passage indicators and the digital protection of the feeders. A dedicated agent managing the cell deals with the given application for the local coordination of detection and network reconfiguration. The existing protection systems take into account some physical events, as the important ratio of fugitive faults. The solution proposed is to meet also the requirement of such reliable existing processes, enabling to maintain a high level of local power availability.

The report gives and overview of security issues related to information protection and ownership related to future virtual utilities. In effect we have to protect and maintain Power Net Operations simultaneously as we protect and maintain Business Models related to value-added energy related services and products. To cope with this complexity we propose a hierarchical coordination model between energy management systems and business management systems. To cope with security issues we propose a security engineering approach aiming at implementing and maintaining robust systems. Specifically, we have addressed issues related to secure execution of non-secure software. Furthermore, we have addressed the use of honey nets to identify the kind and art of threats to a specific system. The task of defining a workable model of assessing security investments in a cost effective way have several inherent challenges. We have identified some of those challenges and also identified a general framework to that end.

Three phasor measurement units of ABB make, type RES 521 were installed in Sydkraft Ölands network. All three terminals are commissioned in beginning of April 2003. In order for PMU to work it is necessary to get the GPS signal. Signal from at least four satellites are required in order to get proper PMU synchronisation. The PMU units itself were installed inside the substation control house. After installation the three units have recorded phasor data time, stamped and synchronized from the three units throughout the summer season 2003, corresponding to the high load and disturbance season. The recording has been analyzed for DG specifics and for load dynamics and will be mined for data in many future studies, due to the data uniqueness. Some of the results are immediately applicable, particularly concerning the management of voltage in a network with high DG insertion, especially AVR (Automated Voltage Regulation) schemes. The Swedish blackout, September 23, was captured by the recordings, which are included in the official disturbance report published by the Swedish National Grid Operator.

Results from the tools developed in the project are summarised and extended. A combination of the partners view on the possible merge of ICT network and EPS in the future electrical architecture is illustrated through case studies. A combination of the partners view on the possible merge of ICT network and EPS in the future electrical architecture is illustrated through case studies. The main scope is the distribution network, which needs strong development in order to become more visible, more controllable. The case studies highlight the expected system structure and the combination of the infrastructures.

Electronic markets are software forms of coordination in large-scale systems; they are especially useful for applications that need to achieve optimal system states in fully distributed ways. Technically, electronic markets are, first of all, constructed out of small, relatively independent pieces of software that are known as agents. Agents act as representatives of various pieces of equipment and/or human users, and carry out tasks on behalf of them: find the optimal state. They do so by communicating and negotiating with other agents, very much analogous to selling and buying on a human auction. The result of the work in this project is a collection of novel market algorithms suitable for power applications. Their description and specification is given in a market algorithms library, and they are practically used and tested in some of the CRISP scenarios, in particular demand-supply matching and intelligent load shedding. However, their application potential goes way beyond that: agent-based electronic market provide distributed mechanisms to achieve optimal control in large systems that combine control-theoretic and cost-benefit microeconomic considerations simultaneously.

The focus of this result is on dependable ICT support of power grid operation. By recasting the three CRISP experiments into three Scenarios in Chapter 2 we claim that we have a good description of benefits and challenges related to future virtual utilities. Among the challenges are securing trustworthy operation from a technical operation side (avoid disturbances such as blackouts) as wee as from a user-centric business point of view (value added power related services). Our investigation on proper means to safeguard operations of future virtual utilities begins with an assessment of lessons learned from recent (2003) big blackouts worldwide. We propose an accident diagnosis and repair model (STAMP++ in) suitable for the complex socio-technical system we envisage for future cell-based virtual utilities. Furthermore, we have developed an experimental platform for security and performance experiments conducted in experiments with IDEA and ECN respectively. The main result are on developments of secure execution environments and a service-oriented architecture supporting dependable cell-based utilities.

The PowerMatcher algoritm developed in the CRISP project has been used in a field test that demonstrates the workings of an Imbalance Reduction System (IRS). Essentially, IRS is a flexible framework for implementing and checking the results of operating strategies for reducing DG-RES related imbalance in the context of external market mechanisms. In the field test is has been shown that the algorithm works in practice (Proof of Feasibility), i.e. a substantial reduction of Imbalance of a portfolio containing a Wind Farm has been achieved (in this case over 40% reduction) while all local installations remained in operation within their operational requirements.

A detailed description of the power system is given. The various controls existing in the system are detailed in order to analyse the different aspects of system robustness. This approach is generally expressed in transmission and has a strong link with blackout countermeasures: the major events and defence plans are described and explained. Various levels of DG penetration are investigated for analysing the technical impacts on the networks and the consequences for the robustness. A specific criterion is proposed for having a clear and simple index indicating the real time situation about stability. This indicator may be used in the future in a distributed intelligence way: the real time evaluation of stability is done cell by cell and aggregated at a higher level. In this way a more complex stability study is pointed out with the scalability parameter, taking into account at the same time the local and global constraints. Another specific point from the large DG insertion in the distribution network is the future capacity of intentional islanding. The needs for control and communication are detailed for various power system applications at local scale.

Comprehensive information is provided on traditional load shedding based on voltage frequency criterion or on voltage magnitude criterion. Thorough information about large power oscillations is presented to underline the complexity of setting proper solutions for reaching a right level of stability in the system. Load shedding is a defence scheme used worldwide and its ideal objective during critical situation is to recover quickly the global balance between production and demand, with dynamical characteristics enabling a correct come back to stable steady-state. The result shows the need for a fundamental change in the associated distribution automation. The terminology used asks for some questions since the local intelligence leads rather to a distributed controlled system, the control being achieved on DG units and on controlled loads. The document takes into account three concepts of technical solution in order to deal with the stability of the system, which is the traditional goal of the existing load shedding: SLR (smooth load relief), CPA (critical preventing action) and DLS (distributed load shedding). Once market-based multi-agent systems are implemented for Supply-Demand matching, market-based approaches for ILS could have a great value, although there should always be a fall-back strategy in terms of immediate manual action.

Deliverables

Share this page

Download