EPSRC Grand Challenge

The Autonomic Power System

Can a fully distributed intelligence and control philosophy deliver the future flexible grids required to facilitate the low carbon transition, allowing for the adoption of emerging game changing network technologies and coping with the accompanying increase in uncertainty and complexity?

Industrial partners: Accenture, Agilent, E.On, IBM, KEMA, Mott MacDonald, PB Power, National Grid, SSE

Research themes: power systems, economics, computer science, artificial intelligence, social policy, mathematics, complexity

The Year 2050

The drivers that will shape the 2050 electricity network are numerous: increasing energy prices; increased variability in the availability of generation; reduced system inertia; increased utilisation due to growth of loads such as electric vehicles and heat pumps; electric vehicles as randomly roving loads and energy storage; increased levels of distributed generation; a more diverse range of energy sources contributing to electricity generation; and increased customer participation in system operation.


These changes mean that the energy networks of the future will be far more difficult to manage and design than those of today, for technical, social, environmental and commercial reasons. In order to cater for this complexity, future energy networks must be organised to provide increased flexibility and controllability through the provision of appropriate real time decision-making techniques.

These techniques must coordinate the simultaneous operation of a large number of diverse components and functions, including storage devices, demand side actions, network topology, data management, electricity markets, electric vehicle charging regimes, dynamic ratings systems, distributed generation, network power flow management, fault level management, supply restoration and fuel choice. Additionally, future flexible grids will present many more options for energy trading philosophies and investment decisions. The risks and implications associated with these decisions and the real-time control of the networks will be harder to identify and quantify due to the increased uncertainty and complexity.


[ˌɔːtəˈnɒmɪk] adj.

The Autonomic Power System is a metaphor for moving beyond the current smart grid applications and approaches.

The concept is based on biological autonomic systems that set high-level goals but delegate the decision making on how to achieve them to the lower level intelligence. No centralised control is evident, and behaviour emerges from low-level interactions. This allows highly complex systems to achieve real-time and just-in-time optimisation of operations.

IBM introduced the concept of autonomic computing for large IT infrastructures, where the complexity was beyond that which operational teams could manage. An autonomic computing system was engineered to configure and reconfigure itself under varying (and in the future, even unpredictable) conditions. System configuration or “setup” would occur automatically, as well as dynamically adjusting its configuration to best handle changing environments.


At the heart of the IBM system was “Self*” behaviour, where “*” could be a range of operational functions: self-configuring, self-healing, self-optimising & self-protecting.

This accurately defines the vision for future power systems, and therefore our ambition is to design and prove the concepts, methods and systems that could deliver an autonomic power system.

The autonomic power system is a completely integrated and distributed control philosophy which self-manages and optimises all network operational decisions in real time. To deliver this, fundamental research is required to determine the level of distributed control achievable (or the balance between distributed, centralised, and hierarchical controls) and its impact on investment decisions, resilience, risk, customer participation and control in a trans-national interconnected electricity network.

Research streams

« Swipe left or right to navigate »

Self* Network Operation and Control (SNOC)

The drive towards as fully decentralised network operation and control as is practicable and achievable is being tackled in this stream. In considering the autonomic power system, network operation and control is a key activity as it determines the level of decentralisation that can be achieved and whether an autonomic power system can be designed to ensure stability in a complex and uncertain environment through appropriate real-time control. Research must also determine the methods and algorithms required for the control functions of the autonomic power system.

Multidisciplinary activities are taking place across active network management, distributed intelligence, and artificial intelligence. Activities include:

  • Identification of the most appropriate boundaries for zones of control;
  • The use of a temporal AI planner to solve a temporally extended version of the optimal power flow problem within zones of control;
  • Distributed constraint optimisation (DCOP) and the use of max-sum algorithms within DCOP to provide decentralised network control functions;
  • Methods to autonomously select control algorithms for particular control objectives, network topologies and network states; and
  • The development of a robust methodology for on-line probabilistic assessment and management of a power system dynamic signature.

Research projects

Autonomic Economics: Short-Term (Operational)

The present trading philosophy is centralised with a single objective. This is suitable for a relatively small number of participants and can be implemented by a market operator that clears bids and offers to maximise social welfare centrally, and determines location specific energy prices when necessary. Since one of the goals of the autonomic power system is to permit widespread decentralised energy trading between millions of participants that are not willing to disclose their sensitive information, this current framework is not suitable. Instead of relying on Market Operators, each covering a specific geographical area, the autonomic power system will support the operation of multiple overlapping market operators across interconnected networks, therefore enabling participants with unrestricted choice.

While fully decentralised trading is commonplace for conventional commodities, its application to trading energy over a network of limited capacity will require the development of new economic concepts, including the packaging of commodities and services that are appropriate for the 2050 energy network, new forms of network regulation and new business models for market participants.

The Autonomic Economics Research Stream explores novel frameworks, methodologies, and models for decentralised / localised economic decision making. In terms of operational (short-term) economics, the traditional centralised trading philosophy with a small number of participants and a single objective will not be able to support the vision of widespread market participation of millions of small generators, consumers and prosumers that are not willing to disclose their private information. A framework of fully decentralised trading is required, respecting the limited available network capacity and incorporating business models of different participants. The work carried out so far has produced mechanisms yielding market solutions of proven global optimality without assuming centralised knowledge of any participants’ characteristics based on different approaches, including Lagrangian Relaxation theory, mean-field game theory, agent-based models, self-organisation principles and direct bilateral trading. Furthermore, in order to enhance the profitability of small participants, tools for their optimal allocation across different and potentially conflicting market segments have been developed. These have employed stochastic optimisation and real-option approaches to address the significant uncertainties in market prices and balance rewards against risks related to non-delivery penalties.

Research projects

Autonomic Economics: Long-Term (Investment)

Investment decisions on network reinforcement are currently made centrally using a top-down approach to achieve a single objective. A bottom-up approach must be developed to allow different users of the energy network, each seeking to optimise its own objectives, to support investment programmes that would deliver mutual benefits through a distributed decision making paradigm consistent with an autonomic power system. Therefore, novel tools and methods are required to assess investment options, establish the business case for these options and provide an anticipatory investment framework to tackle the conundrum of whether to build generation or networks first.

This research will lead to scientific advances in distributed decision making; economic theory and governance structures; frontier efficiency analysis; stochastic optimisation; robust optimisation; game theory algorithms; socio-economic based self-organisation; socio-economic based interactions; and uncertainty in economic decision making.

The work carried out so far has modelled distributed decision making in the investment timescale, based on both perfectly competitive and oligopolistic frameworks. In order to enable reaction to new information regarding uncertain parameters, a stochastic network planning approach has been investigated, capturing the option value of smart grid technologies. Moreover, different auction designs for the autonomous development of offshore transmission networks have been developed, and the roles of Independent System Operators (ISO) and Transmission System Operators (TSO) in future systems have been compared.

Research projects

Resilience and Risk Management

The traditional concepts of system security and the deterministic ways of assessing the level of risk will no longer make sense in an autonomic power system and will have to be replaced by probabilistic measures of quality of service and resilience that take into account the information available in real time, the stochastic nature of renewable resources and the capabilities of demand response to deliver corrective control actions. A network design methodology should deliver different levels of “reliability” to different consumers depending on their willingness to pay, a goal that has so far eluded power system designers. While consumers in an autonomic power system may be able to express their preference as to when they might be willing to be disconnected, or to receive reduced quality of service/supply, it is clear that the system should still provide reasonable levels of quality of supply and resilience to major disturbances. Because the mode of operation will differ radically from current practice, new techniques will need to be developed to quantify the risk profile of an autonomic power system and how this profile is affected by the detailed design of the system. A key feature of all models of future energy systems is the uncertainty that arises from not knowing the correct values of the model parameters the fact that any model is only an imperfect representation of the actual physical system. Therefore, uncertainty audits and the creation of statistical emulators are being developed to inform all aspects of the design of the autonomic power system.

Research projects

Active Participation of Consumers

A critical aspect of autonomic power systems is active participation of consumers and small producers at different aggregation levels from local LV networks to the pan EU transmission system.

Active participation of consumers will introduce flexibility of unprecedented proportions that will enable the paradigm change needed for the Autonomic Power System vision. Due to the nature of customer responses, research in this stream will investigate a number of aspects that are critical to fully harnessing and integrating the potential benefits of active demand side response.

First, the social aspects must be investigated to discover underlying behavioural patterns and disaggregate demand to enable analysis of customer preferences. The preferences will then be used to allow for customisation of electricity demand on a large scale, analysis of further effects of improved efficiencies at customers’ premises, and classifying customers to determine the best strategies to engage them.

This Research Stream has activities focusing on customer responses to various attempts to engage them in demand side participation, and how to understand their responses to make them useful for the improved operation of the Autonomic Power System. This is truly interdisciplinary as it brings together expertise from social sciences (evaluating the attitude of customers towards various possibilities to participate in demand programs and as prosumers), as well as computer science and engineering to cover technical solutions regarding scheduling of the demand responses and tools (in the form of serious games) for helping customers to be informed and participate. In addition, engineering research is also looking into the aggregation of the demand so as to evaluate its possible contribution towards network dynamic operation.

Results from this research stream will both influence and be influenced by network operation requirements, as well as market structures and market operation.

Research projects


You can download files from the download area if you have been given a password.


Principal Investigator: Prof Stephen McArthur, University of Strathclyde

Co-investigators: Dr B Chaudhuri (Imperial College London); Dr C Dent (Durham); Prof M Fox (Kings College London); Dr S Hielscher (Sussex); Dr I Kockar (Strathclyde) Dr J Moriarty(Manchester); Prof J Watson (Sussex and UKERC); Prof D Long (Kings College London); , Dr P Mancarella (Manchester); Dr MG Pollitt (Cambridge); Prof P Taylor (Newcastle); Dr J Pitt (Imperial College London); Dr J Mutale (Manchester); Prof j Bialek (Durham); Prof J Milanovic (Manchester); Prof G Strbac (Imperial College London); Prof G Ault (Strathclyde).