IEEE Power & Energy Magazine - November/December 2021 - 80

other forms of power generation, with or without additional
regulatory support. In most cases, those electricity markets
comprise several forward stages (e.g., day ahead) and a nearreal-time
balancing stage. Consequently, renewable energy
producers (or portfolio managers) place offers in those forward
electricity markets based on forecasts.
Market participation has been the main motivation for
renewable energy producers to invest in forecasting solutions
over the last few decades. It has also been a strong driver for
innovation in forecasting and for assessing the value of competing
forecast solutions. For instance, besides the problem
of the quantification of reserve requirements (discussed later
in this section), market participation supported the focus on
uncertainty quantification in renewable energy forecasting
and eventual advent of probabilistic forecasting.
Today, emphasis is placed on many problems related to
the market participation of renewable energy producers.
These are all problems of decision making under uncertainty
since the power generation forecasts are not perfectly
accurate. One of the most basic considerations is that the
market-induced penalties for imbalances may not be symmetric,
e.g., the penalty can be higher for underproducing
than overproducing. This incentivizes being strategic when
placing offers in electricity markets.
In addition to such uncertainty-related considerations, one
should be aware of risk-related considerations. Forecast errors
may vary greatly, most often being small but sometimes very
big, and so may market prices and induced penalties. As a
result, revenue losses may become very large, though infrequently.
This has motivated the proposal of risk-aware market
participation strategies, which are similar to those in financial
markets. With the increasing importance of storage in energy
systems, hedging may come from operating additional assets
instead of being based only on market participation strategies.
Forecasting Use Cases for
System Operation and Markets
If the net load is underestimated, there is the potential for
large costs or load curtailments. Conversely, if it is overestimated,
there is a relatively mild increase in operating
expenses. The asymmetric costs of errors are the major
reason operators should (and do) recognize risk in making
decisions. Simple rules of thumb are inadequate to manage
risks when variable renewable energy is providing 30%,
50%, or even more of the supply. Over the past decade,
researchers and vendors have developed powerful tools that
allow operators to explicitly quantify probability distributions
of the net load that reflect current as well as future
weather and consider them in operations decisions.
Probabilistic forecasts of the wind, solar, and consumer
components of net load can be divided into two
basic types:
✔ The first is a marginal probability distribution of a
single variable, such as net load or one of its components,
at a particular time and specific location or
system, conditioned on available weather information.
✔ The second is a multivariate characterization, accounting
for dependencies either over time (such as
the evolution of wind over the day), space (for example,
the diversity of solar resources over a region at a
particular time), or components (accounting for wind,
solar, and load correlations). These characterizations
can be in the form of formulas for joint distributions
or, more commonly, a set of possible scenarios, such
as an ensemble of NWP forecasts.
Some of the simpler ways in which these probabilistic
forecasts can be used include setting reserve requirements,
where the amount of procured reserve is based on an a priori
tolerable risk level-such as covering 95% of occurrences or
Load Forecasts
Renewable
Forecasts
Fuel Forecasts
Long-Term
Planning and
Capacity Markets
Outage Scheduling
Day-Ahead Unit
Commitment
Intraday Reliability
Unit Commitment
Real-Time
Economic Dispatch
Contingency
Analysis
Offer Prices
Offer Quantities
Transmission
Limits
Reserve
Needs/Products
Fuel Limits
figure 2. The evolution of forecasting applications: load, renewable energy, and fuel predictions will be used to develop
other forecasts of increasing needs in the future.
80
ieee power & energy magazine
november/december 2021

IEEE Power & Energy Magazine - November/December 2021

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - November/December 2021