Challenges Arise
The landscape of energy demand and generation forecasting is barely recognizable when compared to just a decade ago. Resources used to be mostly dispatchable, and demand curves followed a familiar pattern. Today, though, grid operators must consider things like non-dispatchable, utility-scale renewables; retirement of traditional generation; behind-the-meter solar generation; demand response; energy storage; and the impact of extreme weather; all of which increase the complexity and uncertainty of forecasts. Meanwhile, the high stakes and costly repercussions of missing a forecast remain.
Fighting Uncertainty
Southwest Power Pool (SPP) formed its Uncertainty Response Team (URT) in 2018 to address growing system threats. The URT continuously assesses three factors: load, renewable generation availability, and generation outages. Weather conditions play a key role in each. Load is highly correlated with temperature, which drives consumers’ heating and cooling needs; renewable generation at SPP is mainly determined by wind; and extreme winter weather contributes significantly to renewable and traditional generation outages.
The URT creates a probabilistic forecast by analyzing weather and renewable power projections that factor in past errors given a set of weather-related variables for a particular time period. The result is a forecast that compares net load and resource availability, as illustrated in the graph below.
Figure 1. Net Load vs. Resource Availability on July 21, 2023, in SPP.
Net load (colored lines) vs. capacity (grey background) time series at confidence intervals ranging from 50% (light blue) to 99.5% (orange).
The capacity vs. uncertainty chart highlights periods of time during which net load is somewhat more likely exceed available generation capacity. During these periods available capacity can be adjusted to include reliability units and reserve requirements. When a confidence band exceeds all availability, it indicates a chance that SPP could have to issue controlled service interruptions, or “shed load,” unless it’s able to import energy from other regions.
Taking Action
These probabilistic forecasts enable the URT to make recommendations to ensure reliability under SPP’s multi-day reliability assessment. This may include bringing on longer-lead-time resources, rescheduling planned outages away from a concerning time interval, and recommending system advisories. Since new generation cannot be created out of thin air, SPP makes sure its generation is optimized to minimize the chances it may need to declare an energy emergency alert or shed load.
Ensemble Forecasting
As the frequency of extreme weather events increases and forecasts, in turn, become more sensitive to the impacts of weather, system reliability will increasingly depend on grid operators’ and analysts’ ability to accurately identify and respond to high-risk periods. Global numerical-prediction weather models produce ensemble products that can be used as part of a probabilistic forecast. Operationally, ensembles can be used in several forecasting aspects, including wind power generation. Looking at an ensemble forecast plume can give an idea of intervals with above average or below average forecast confidence and possible forecast ranges.
Fig 2. Ensemble Wind Power Scenario Time Series Forecast
Outlook
The complexity of generation and demand forecasting is certain to increase over time, as renewable energy penetration continues to grow and as improving battery technology drives rapid expansion of storage. Sole utilization of deterministic forecasting in the power industry is becoming obsolete. Developing subject matter experts to utilize probabilistic forecasting to assess risk becomes critical in maintaining a balance between an economically efficient and a reliable grid.
Jeff Baskin
Meteorologist/Technical Analyst
Southwest Power Pool
Garrett Crowson
System Operations, Manager Uncertainty Response
Southwest Power Pool
Leave a Reply