IEEE Power & Energy Magazine - November/December 2021 - 34

of which change continuously. As illustrated in Figure 4, the
ELCC of solar and storage are both higher when combined
than when they are evaluated separately. These interactive
effects may either be " antagonistic, " where each increment
of solar provides successively lower capacity value, or
" synergistic, " where a portfolio of solar and storage likely
provides more value than the sum of its parts.
Principle 3: Modeling Chronological
Operations Across Many Years is Essential
As our grids rely on VRE and storage for reliability, periods of
risk are no longer isolated to peak load conditions but instead
may be shifting to other periods due to abnormal weather
events or while the sun is setting. The relevant time of analysis
is also a function of the duration of ELRs to ensure that
the energy storage and demand response will be available for
enough hours to get the system through periods of scarce
supply. Chronological stochastic analysis is increasingly
important for modern RA analysis. This process simulates
a full hour-to-hour dispatch of the system's resources, for an
entire year of operation, across many different weather patterns,
load profiles, and random outage draws.
Many years of synchronized hourly weather and load
data are required to understand correlations and interannual
variability between wind and solar generation, outages, and
load. However, analysts and policy makers should be cognizant
of data limitations. Long historical records of correlated,
wind, solar, and load may not be available, and past
observations may no longer be good predictors of future conditions
with a changing climate.
Take the California 2020 rolling blackout events as an
example. California's RA construct and planning reserve
margin is based on the peak gross load, which occurs in
the middle of the day during the summer months. However,
periods of peak risk in California now occur in the late
evening hours as solar resources decline and evening loads
remain high, although lower than midday levels. Figure 5
10
15
20
25
30
35
40
45
50
5
042 68 10 12 14 16 18 20 22 02 46 810121416182022
15 August 2020
14 August 2020
Stage 3 Emergency Demand Net Demand
figure 5. The gross and net load during the 2020 California reliability event. (Data
source: California ISO.)
34
ieee power & energy magazine
Peak Demand
Load Shedding
shows the gross and net load for CAISO for the days when
rolling blackouts occurred. As clearly shown in the figure,
the conventional assumption that peak risk is aligned with
peak load is no longer true. This requires an evaluation of
all hours of the year so that the hours with the highest probability
to be the peak load hours can be identified.
One obvious change would be to focus on peak net load
(load minus renewable output) rather than peak load, but this
too is insufficient. Wind and solar output have probabilities
around their outputs, driven by irregular and complicated
weather. The California event stemmed, at least in part, from
a widespread heatwave that seemed highly improbable based
on historical patterns but may be more likely due to climate
change. More changes are needed to address both potential
conditions and resource availability during these conditions.
Reliability planners should also be careful not to fall into
the trap of using a short historical record of weather data
because it is the only data that is available. Part of reliability
planning is ensuring the system can maintain reliability during
potential-and credible-weather events. The California
heatwave saw some of the highest average temperatures in
the past 35 years, simultaneously spread across most of the
Western region. Similarly, the ERCOT events of February
2021 saw temperatures well below the near-term historical
record for a much longer duration than the 2011 event. Just
because our recent weather data does not include a weather
event does not mean system planners do not have to prepare
for one in the future. More planning should be focused on
identifying potential situations that could occur, and testing
system reliability.
These are examples where the traditional data-driven statistical
modeling has limitations. Future RA analysis should
evaluate potential situations that may not have occurred in
the past but reasonably could occur in the future. Identification
of these high-impact, low probability events can then be
evaluated in isolation to determine if and how they should
be mitigated.
Principle 4: Load
Participation
Fundamentally Changes
the RA Construct
The proliferation of energy storage,
demand response, electric vehicles,
and dynamic rate design bring with
them new options for load flexibility
and should be evaluated
in a similar context as generation
resources, including uncertainty
and availability. Real-time markets,
with a high degree of participation
from price-responsive demand,
may shift the RA planning challenge
away from reliability needs
toward economic considerations,
november/december 2021
GW
IEEE Power & Energy Magazine - November/December 2021

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