Simulating a commercial power aggregator at scale: design and lessons learned

Research output: Contribution to conferenceProceedingpeer-review

18 Downloads (Pure)


To evaluate aggregation models in the context of a power system, a software tool (the SmartNet simulator) has been developed to look at the impact of managing Distributed Energy Resources (DERs) on networks' technical operation (e.g. power flows and voltage levels) and simulates wholesale and ancillary services market conditions. This paper focusses on the design and implementation of one of the aggregation models that addresses the Curtailable Generator / Curtailable Load (CGCL) aggregator. The paper outlines the design of such a software aggregator agent and discusses the lessons learned in simulating a more realistic large power grid system. The aggregator is represented as an agent based object orientated model using a financed based buckets system to aggregate bids from up to 300,000 devices across 10 -20,000 power nodes. The concept/implementation can be extended to include more sophisticated bidding strategies and to use multiple perspectives on tranches. Simulation and testing of such a large simulation system was challenging, and we have proved that it is possible to simulate the aggregation and clustering of different types of flexibility into a number of manageable bids in a timely manner.
Original languageEnglish
Number of pages8
Publication statusPublished - 31 Jul 2019
EventSimultech 2019 9th International conference on simulation and Modelling: Methods, Technologies and Applications - e Vienna House Diplomat Prague, Prague, Czech Republic., Prague, Czech Republic
Duration: 29 Jul 201931 Jul 2019
Conference number: 9


ConferenceSimultech 2019 9th International conference on simulation and Modelling
Abbreviated titleSimultech 2019
Country/TerritoryCzech Republic
Internet address


  • agent based modelling
  • aggregation
  • power
  • simulation
  • smart grid


Dive into the research topics of 'Simulating a commercial power aggregator at scale: design and lessons learned'. Together they form a unique fingerprint.

Cite this