Electricity Networks

Smart Energy Systems Han La Poutré CWI, Amsterdam Centrum Wiskunde & Informatica

&  Utrecht University

Future Scenarios:  Smart Energy Systems (SES)

Smart Grids

ICT Techniques for  Supply/Demand Matching • Some (first) approaches and solutions exist – E.g., first market mechanisms (Powermatcher), first clustering  (microgrids), smart powermanagement (factories), ...

• A variety of settings exist – Depending on areas, possible business roles, regulation, etc. – Different solutions needed (as above)

• And: – We want (to keep our) stable electricity supply!

• How to approach this? Questions in  – Electrical engineering • Network stability, energy efficiency, controlability, supply/demand matching

– Economics and business science • Who does what (services, control)

– Law • Regulation

– Computer science!!

ICT Techniques for  Supply/Demand Matching • Questions in Computer science: High complexity • E.g.: Efficient solutions – Can the ICT system react fast on sudden changes – Can I indeed get all my desired power when I bid? What if not?

• E.g.: Stable solutions – How do I know that using my washing machine this afternoon, with sunny weather, indeed gives low prices? – How do I know that when the washing/drying machine starts, prices remain low for the next hours? Any guarantee?

• E.g.: Robust solutions – Will prices remain low when just one large consumer suddenly  may start using power? – What happens if everybody uses the same software with similar  decisions?

– (Market) mechanisms that yield sufficient certainty?

ICT Techniques for  Supply/Demand Matching • Questions in Computer science: High complexity • Large amounts of data and decisions (computations) – Millions of households in NL; smart meters in network – Tbytes data per day; millions of decisions per day – Aim for decentralized ICT solutions

• Efficient ICT solutions – Efficient (market/coordination) mechanisms for supply/demand  matching – Fast response and computation

• Stable ICT solutions – Low uncertainty (of market prices): predictability

• Scalable ICT solutions – For hundreds to thousands to millions of actors

• Robust solutions – Heterogeneous environments: large and small actors, .. – Reliability of ICT performance

– Typical Computer Science issues • Current research

ICT Techniques for  Supply/Demand Matching • Coordination and control – Remote control • At homes, by electricity companies – Electric devices (freezer, e‐vehicle, heater)

– Actors respond via market mechanisms • Dynamic pricing • Auctions (several types) • Other market types – Negotiation

– Other forms of organisation • Contracting (day/night, advanced, ...)  • Clustering (smart neighbourhoods)

Non‐technical requirement for SES

Some Important Problems in SES

Smart Grids:  Supply/Demand Matching

General Problem Domain • Decentralized decision making  • Economic and environmental optimization • Decentralized logistics

• Local decision makers • • • •

Limited information Adaptive to dynamic environments Repeated decisions Learning from past

Areas • Application and modeling  areas – Decentralized logistics • Cargo transportation 

– Patient Logistics • Hospitals 

– Energy markets • Decentralized supply and demand

Economic Paradigms in ICT

ICT / Agent Techniques • Software agent • Independent intelligent  software module – Can interact autonomously with  other agents

• For automatic actions on  behalf of a party  – E.g., its owner

• Multiagent system – Many independent software  agents;  ICT system – Coordinating together

• Conceptual use

ICT Techniques

• Coordination between agents – Market mechanisms • Auctions  – Continuous, double auctions, first price, second price, ascending bid, …

• Negotiation – Bilateral, multilateral

• Dynamic pricing – Central dynamic price determination

– Efficient and robust market mechanisms • For ICT solutions: bidding agents • Scalable; heterogenous; low uncertainty; robust

– Computable solutions • Strategies of agents • Computation of “winners” (allocations) in auctions

– Other requirements • E.g. as before

• Important field in computer science

• Adaptive, learning agents – Learning from the past • Data, actions, situations • Forecasting 

• Computational intelligence techniques – E.g. neural networks, genetic algorithms,  hybrid heuristics

Power Generators • Power generators, plannable – Large power plants • Gas, oil, coal, nuclear, ..

– Small (green) power generators • Biomass, batteries (e‐vehicles), μCHPs (μWKKs), warehouse  CHPs, ..

– Partly “plannable”: wind parks (weather forecasts), solar  cell fields,.. • Delivery or storage

Reduction of Uncertainty for Power Generators • Power generators generate (as in large scale  power supply) – Default power • Amount of power determined one day ahead for a specific time slot (half hour)

– Balancing power • Additional mount of power determined “just” before a spefic time slot (half hour)

• When to deliver how much power, for which  price? 

Reduction of Uncertainty for Power Generators • Actors (agents) submit bids to generate power at a  certain price – Default power: 1 day ahead • How much power to generate, in the folowing day at a specific  time slot (half hour)

– Reserve power • Maximal additional amount of power to be delivered “just” before a spefic time slot (half hour) (“optional power”)

– 2 independent auctions per time slot • Default power and reserve power • 24 hours ahead

Reduction of Uncertainty for Power Generators • Actors (agents) submit 2 bids to generate power at a  certain price – 2 independent auctions per time slot • Default power and reserve power

– Uncertainty about combined allocations • Problematic for planning  • Problematic for revenue

– Often default and reserve power generation are related – How to combine these?

Reduction of Uncertainty for Power Generators • Actors (agents) submit 2 bids to  generate power at a certain  price – How to combine these? – Ratio between default and  reserve power • For power generators • For markets (by e.g. network operators) • Chosen, fixed • Combined bid

Power Consumers • Several similar features; e.g. – Power consumption/storage devices • E‐vehicle, freezer, washing machine running at night; or combination

– Prescheduled power consumption • Couple of hours in advance

– Possible additional power consumption, if requested • • • •

Freezer: extra freezing E‐vehicle: extra charging Turn on washing machine When access power is available (uncertain green power supply)

– How to deal with  • Uncertainty of reserving power for consumption at specific prices • Unexpected deviations in power availability • Combine both: interconnected consumption

Combined market mechanism • Combined market mechanism with ratio • Advantages – Precise: generators submit only one non‐linear bid – More plannable: generators choose a ratio  between default and reserve power – More certain revenues

• More information: Nicolas Höning  – Poster session

Comprehensibility of  Agent Systems in SES

Comprehensibility of  Agent Systems in SES

Some Important Problems in SES

Computer Science Research on SES

Conclusion

• Emerging field: Computer Science research on SES – Last couple of years • After EE and economics

– International results • Conferences, projects, networks

– Organisations  • • • • •

NWO/STW SES program EU FP7 Agentschap.nl EIT ICT Labs institute EIT InnoEnergy institute

• Required for SES

• Smart Energy Systems – Novel ICT required – Requirements from • EE, society, regulations, economy,  business

– Several basic solutions exist – More advanced solutions underway • Computer science research – Scalability, stability, heterogenous, ..

– Contact: [email protected]