New paper: Decentralized Microgrid Controls

Solar panels, electric vehicles, smart appliances, and storage systems are rewriting the electricity production landscape. When used intelligently, these Distributed Energy Resources (DERs) can reduce cost, improve reliability, and help fight global warming by integrating more renewable energy into the electricity grid. However, if left uncontrolled these DERs can create additional stresses and costs for the electricity grid- as solar production surges when the sun rises, or returning commuters plug in their EVs and drain the power system. Traditional models of centralized control can’t scale to handle the millions of electric vehicles and smart appliances that will be plugging in over the next 5-10 years… forcing grid operators to look for new tools.

To date, proposals for integrating more DERs fall into three categories:

  • Restructuring tariffs: The current approach, this takes years to iterate, typically doesn’t reflect local and time-changing constraints, and inevitably breeds conflicts.
  • Creating decentralized ‘transactive energy’ markets: These are popular amongst economists, DER manufacturers, and aggregators, but they don’t account for network constraints and so are typically a non-starter with utilities.
  • Running a decentralized algorithm in which the utility directly controls DERs: These ignore mechanisms for compensating DER manufacturers, aggravating trust issues between utilities and customers or DER manufacturers.

A solution would need to scale to handle large populations of DERs, address these trust issues, provide a transparent way of pricing DER services, and respect network constraints to make sure that reliable power is delivered at minimal cost.

A new research paper that I’ve released with Jon Mather proposes a system for accomplishing these goals, by outlining a method for decentralized optimization that minimizes energy cost by controlling DERs in a local distribution network.  We use a model for energy flow to ensure that power quality and line constraints are met across the grid, while computing local energy costs so that DERs can be compensated for their services.  We consider 4 types of DERs:

  • Rooftop solar panels which produce ‘must-take’ power.
  • Deferable loads such as smart dryers or dishwashers, which must complete their cycle by a time specified by the user.
  • Shapeable loads such as electric vehicles, whose power consumption can be adjusted over the course of their operation.
  • Battery systems, which can charge when extra power is available, and discharge when power is needed by other appliances.

In our model, each of the DERs computes the solution to a local problem based on its own constraints, and then provides an updated price that it’s willing to pay to the rest of the network.  Communication between nodes doesn’t reveal the consumers’ constraints (an important tool for rpeserving privacy), but allows the system to come to an optimal solution across the entire network.

An important extension in our work is the use of a blockchain for running the coordination step between the DERs. This allows us to guarantee that schedule is fair for all participants, and that nobody can cheat or distort prices -not even the system operator. While prior research on grid control algorithms doesn’t address these trust and incentive issues, our work takes advantage of the power of blockchains for enabling coordination between non-trusting entities, an important step forwards for addressing the concerns of both utilities and DER manufacturers.

Unlike prior transactive energy projects such as the Brooklyn Microgrid Project and Powerledger, we provide a method for integrating the distribution network into computing the system dispatch- an important criteria to making this decentralized approach acceptable to utilities and regulators.

 

 

 

We’re looking at extensions to integrate more types of DERs, allow for coordination between different distribution feeders, and remove the reliance on smart meters as trustworthy reporters of power consumption. We’d love to get your feedback as we move this work forwards- leave a comment below!

Leave a Reply

Your email address will not be published.