By Graham Harris

I first came across the idea of distributed infrastructure systems when we began planning the CSIRO Australia Energy Flagship in 2001-2. The full flowering of these innovations has taken more than a decade to come to fruition and, even now, for reasons I shall discuss, not all aspects of the concept have been implemented in energy networks around the world. Nevertheless, despite impediments, the basic concept is emerging rapidly.

In 2013, on a visit to southern and western Spain, I was intrigued to see the huge wind energy systems as well as the widespread deployment of solar photo-voltaic (PV) systems at the village scale. These were combined with large solar thermal plants coupled to compressed air storage systems. (Ironically, I was on a birding trip and the solar thermal plant occupied a previously much-visited freshwater marshland!)

When solar was combined with hydropower and wind power, Spain generated more than 50 per cent of its power from renewables in the spring of the year I was there. This shows the growing trend towards distributed energy generation systems and what can be done to transform practices and policies.

Andreas Wagner’s idea of distributed robustness in living systems (see Thinking Systems blog #6) involves not just the evolution of redundancy (duplication of parts) but the development of differing multiple sub-systems and alternative pathways so that the overall system is robust to changes in the network structure, deletions of links and differing substrates – feed stocks and generation systems in this context. Such distributed systems exist in robust quasi-stable neutral spaces and can be reconfigured rapidly if change is required.

What we are now beginning to see in distributed energy networks mimics nature in important ways. The original – primitive – energy networks involved distribution of power from centralised base-load power stations. Some redundancy was built in to accommodate peak demands and system failures, but the basic plan was a one-way distribution of power from generators to users. Centralised governance held sway.

What we are now beginning to see are distributed networks where the generation mechanisms are heterogeneous and of various provenances – wind, solar, geothermal, fuel cells, gas turbines, coal, nuclear – some large-scale and capital intensive, some domestic in scale; some supplying base load power from coal, some renewable and ephemeral. The feedstocks vary from solar to wind power to uranium to biomass. The network has to have multiple pathways and distributed control to cope with fluctuations in supply and demand in time and space. New distributed storage technologies are also coming on line – pumped water storage, compressed air, heated salt, batteries – to even out the supply-demand balance.

New battery technologies in the home – e.g. the Tesla PowerWall – and in road vehicles makes it possible not only to take domestic power systems combining solar PV and batteries off the grid altogether, but also to use the charging and storage capabilities of vehicles as backup power supplies at peak times. Electric vehicles spend a lot of time stationary in garages plugged into the grid. Once these vehicles become more common and are linked into the grid in large numbers, then enormous storage capacity will be installed on line. What was essential base load power generation suddenly becomes a back-up to something much more complex, distributed and dynamic.

Power will flow in many directions in the distributed grid and will be routed around many different sources of supply and demand, depending on what the renewable inputs are at any location on any given day, what the pattern of demand is, and what storage capacity is connected where at any given time. There is no central control.

European plans to link EU national grids and for major solar installations to be built in southern Europe and in northern Africa mean that in the future it will be possible to balance supply and demand across entire regions of the globe, evening out fluctuations in weather – wind and solar power – and demand in space and time.

Institutional and contractual arrangements limit our ability to respond to innovation and technological change. The fixed contractual arrangements involving many financial institutions that locked in debt-laden power companies to long-term contracts in the centralized grids are major impediments to change. New decentralised power grids require new agreements.

Every house that installs solar PV is already undermining the current business plans of the largely privatised base-load power and distribution companies.  It is the ability of domestic PV systems and battery storage to even out the daily peaks in demand (by supplying power at times of peak demand when base load power stations make their highest returns); that is one of the reasons why such systems are undermining the traditional energy supply businesses.

Further innovations in power storage, distribution and control will merely smooth the supply and demand curves, and exacerbate this trend. Once again, it is the terms and conditions of existing power supply contracts that are causing problems.

This trend is also reinforced by the steep fall in the cost of solar PV systems, to the point where the cost is now plug-compatible with other generation systems over wide regions, and many new start-up companies are offering services that compete with established businesses. The threat to established businesses is exemplified by the active lobbying that is going on to reduce national renewable energy targets, and by the speedy reduction in feed-in tariffs paid to those who have installed domestic solar PV systems.

It is not only the drop in the cost of solar PV systems in particular and the swift development of new stationary and mobile storage technologies that are game-changers. It is also the computational and communication capacity to manage the robustness of the distributed grid. Despite the ossified legal and financial arrangements currently in place, major changes in business models in the energy generation and distribution systems are afoot. Established companies will either change or will go extinct. Needless to say, there is institutional resistance to change as there are huge sunk costs and system thinking is in short supply. Evidently, those who have invested heavily in recently privatised power stations can only see one way to recoup their outlays.

It is tragic that the Australian renewable energy targets have been reduced rather than lifted. Like Europe and North Africa, we have the climate and the capacity to implement large renewable PV, solar thermal and geothermal power systems, and to better balance supply and demand across space and time. It has been suggested that the electricity network and power generation companies should encourage the switch to electric vehicles so that their business plans could be supported by new sources of demand; after all, there will be a need to charge the electric vehicles, mostly at night. Ironically, electric vehicles could be the saviour of these companies!

What we have in these new energy systems are smart hybrids comprised of both engineered components (the “hardware”) and “soft” living agents (being us, the “wetware”). The nodes of these networks are prosumers in that they can act as both producers and consumers of energy depending on circumstance. The networks are open because components and agents can come and go and they are of heterogeneous natures and origins. There usually is no common goal. Because there is no central controller, these networks have to be designed to comply with constitutive rules (“who gets to play and how”) which deliver distributive justice (“how are resources shared between components? Who gets to do what and when? Who benefits and who pays?”) Innovations in component design and performance as well as user behavior have driven the need for changes to the rules of play.

Further, because the decision-making must done in real time and at speeds exceeding the capability of human agents alone, network users must be supported by automation. There is therefore a key role for inter-dependent computational and human agents in decision-making, and this leads to the concept of computational justice (“how can this be assisted fairly and equitably by computer algorithms?”) a question formulated by Jeremy Pitt and others in UK*.

In theory at least these open, self-organising hybrid networks are truly “smart networks”. These networks raise some of the most challenging problems across many disciplines, from philosophy to computing, from engineering to sociology. In many ways they mimic many natural systems and finding solutions will require multi-disciplinary and international collaboration. In practice, as we have seen, competing interests and legal constraints make agreements on changes to constitutive rules hard to achieve. Other infrastructure systems (e.g. water) are not as far advanced as the energy networks. They are being held back by the slower speed of technological innovation as well as by similar legal and institutional resistance to change.

If they can be effectively implemented, distributed energy systems will be more robust and will sit more lightly on the land. If there is sufficient reinvestment of capital and the rapid development of distributed renewable energy generation and storage, they will help bring about greatly reduced carbon footprints. Can we speedily change business models and take advantage of innovation? Time will tell but, right now, the world around us is changing faster than we can respond. “Old think” and rigid constitutive rules are holding back progress.

To facilitate more rapid change, we will need to quickly rethink our legal, financial and institutional arrangements so as to enhance flexibility. Reform of constitutive rules is one of the most important tasks for the near future. We rapidly need to find new ways to do business together that more closely mimics nature and works with the natural variability.

*See e.g. Jeremy Pitt, Didac Busquets and Régis Riveret  (2013) The pursuit of computational justice in open systems. AI & Society 12/2013; DOI:10.1007/s00146-013-0531-6