IEEE Power & Energy Magazine - July/August 2021 - 45

Optimal distributed coordination and
appropriate rewards may not be enough:
prosumers need to be willing to participate.
OPF selects the most competitive set of bids to match supply
and demand while ensuring that the transmission network
remains within its limits. In addition to settling on the dispatch
of all participants, the calculation determines the marginal
price of power in each region used to pay generators and retailers.
Differences in prices among regions reflect constraints in
the network connecting the regions when the constraints prevent
the free flow of power and equalization in prices.
The same concepts can be transferred down to the distribution
system, where prosumer DERs take the place of large
generators, and prices can be calculated per distribution network
node instead of regionally. These high-fidelity prices are
then known as LMPs. However, there are many practical considerations.
A simple extension of existing wholesale market
structures to include the distribution system will not be able to
cope with the increasing complexity of managing DERs:
✔ Participant scale: The wholesale NEM is cleared every
5 min, and it currently manages on the order of
hundreds of schedulable generating units. In contrast,
Australia already has 2.5 million passively managed
small-scale PV systems and faces the potential for
large uptake in other DERs, such as battery systems
and electric vehicles. The existing centralized design
will not be able to directly coordinate such an influx
of active participants.
✔ Distribution network modeling: The NEM market
process considers an approximate, linearized model
of the transmission network. In total, the distribution
network is several orders of magnitude larger and requires
a more accurate unbalanced three-phase model
to capture network voltages and flows.
✔ Statefulness of DERs: The NEM market clearance is
relatively myopic, considering only a single 5-min
interval at a time in its clearance. Similarly, the market
bidding structure makes it difficult to accurately
represent participants' requirements over time. This
will become increasingly problematic for stateful
DERs, such as batteries and electric vehicles, which
may need several hours' notice to charge or discharge
in anticipation of an event or price fluctuations.
For these reasons, we need to look to alternative
approaches to solve the OPF problem for our distribution systems
in a DER-heavy future, solutions that can handle the
increases to scale and complexity while still optimizing the
outcomes for the system. One promising approach is to introduce
localized, distribution-level markets that coordinate
with wholesale markets rather than being directly integrated
july/august 2021
into them. These distribution markets still need to solve
a challenging OPF problem, but they can focus on smaller
chunks of the network-say, one distribution feeder at a time.
The approach we developed and demonstrated in this
project explores a key feature of a distribution-level market:
how to optimize the use of participating DERs for network
support. The approach, known as NAC, is a distributed algorithm
that allows prosumers to iteratively negotiate (bid)
their requirements to supply or consume power. NAC converges
to the LMPs that ensure the most efficient utilization
of prosumer and network resources and that their respective
constraints are satisfied. The EMS automates this iterative
negotiation over a forward horizon instead of a single time
step so that better long-term decisions are made about the
battery state of charge.
By distributing the computation, we can leverage the
computing resources of the EMS so that the multiperiod
OPF can be solved within 5 min, enabling near-real-time
dispatch of DER resources, which is in line with the existing
wholesale NEM, despite the increase in modeling complexity.
Figure 4 shows the key steps of the NAC negotiation,
which is based on the alternating direction method of multipliers
algorithm.
Using this approach, we break the multiperiod OPF into
two simpler subproblems, one that manages the network
constraints solved in the cloud NAC server (stage 2) and one
that manages the customer DER constraints and preferences
solved by the EMS (stage 3). The three stages reflected in
Figure 4 consist of the following:
1) Network data as well as measurements of the customer
connection point power, network recloser readings,
and local weather are collected by the NAC cloud
3
2
1
figure 4. The NAC algorithm involves negotiation between
residential EMSs and the NAC cloud server.
ieee power & energy magazine
45
IEEE Power & Energy Magazine - July/August 2021

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - July/August 2021
