How intermittent renewables impact CalISO
“Pretty challenging times,” said Clyde Loutan as he began his presentation on how the CalISO works today and what changes probably will be necessary to accommodate the state’s aggressive 33% RPS by 2020. An intermediate step: 20% of California’s kilowatt-hours must come from renewable resources by 2012—a goal most other RPS states have established for 2020. California policy also calls for emissions of greenhouse gases to be at 1990 levels by 2020.
The CalISO controls about 80% of the state’s load, which was about 50.2 GW during the peak year of 2006.
Specific operational challenges faced by CalISO over the next 10 years include the following:
Loutan had three slides to illustrate how intermittent renewables impact grid operations. Fig 7 was developed from wind production data for
April 2009. Each line tracks wind generation for one day that
month and shows how difficult it is to predict with accuracy the production of wind resources in the day-ahead and hour-ahead timeframes.
The variability of wind and solar are presented together in Fig 8, actual data for last June 24 (a cloudy day) recorded for a 150-MW wind farm and a 24-MW solar PV field. “How would you balance these resources in real time?” Loutan asked the workshop participants.
Next, he s
howed the group how dramatically the dispatch of conventional resources would change with significant contributions of solar and wind power (Fig 9). Before intermittent renewables, generation resources would be dispatched to follow load, the top curve. But when solar and wind are added to the mix, dispatch of conventional resources would have to follow the red curve.
The CalISO has roughly 60,000 MW of capacity at its disposal today, including 5000 MW of dynamic schedule. Ramps up and down will be one of the biggest challenges for conventional assetsgoing forward, Loutan told the group.
Fig 10 shows current ramp rates for the ISO’s generating units.
Note that relatively few assets can ramp at 20 MW/min or more, and most of those are hydro. The grid expert stated that on a typical summer weekday, between 8 and 10 am, load can increase by about 4000 MW an hour.
Today, the Ca
lISO can meet this hourly ramp requirement, mitigate unexpected intra-hour variability, and comply with control performan
ce standards with a combined ramp rate of 60 to 100/MW/min. However, to meet the 33% RPS, technical studies show ramp rates may triple, which is not possible for the ISO’s conventional generation as configured today. Loutan thinks the need for flexible conventional generation going forward cannot be overstated.
Perhaps the most valuable part of the presentation for many in the room who had spent their careers managing generation assets was Loutan’s description of how the CalISO works. This was important so all could grasp the challenge grid operators would face if too much intermittent renewables capacity were added before existing infrastructure was upgraded or replaced to accommodate the wind and solar generation.
Fig 11, note that grid operators begin lining up available generating assets a day before they are needed by issuing a “day-ahead schedule” for each hour of the next operating day to meet expected hourly demand (blue). The day-ahead market closes at 10 am the day prior to the operating day.
A mixture of extremely long-start units (those requiring more than 18 hours to start), long-start (between five and 18 hours), medium-start (between two and five hours), short-start (less than two hours), and fast-start (less than 10 minutes) resources are lined up in “economic order” to serve load at lowest cost to consumers.
requirements are adjusted continually based on forecast revisions. The green line shows the hour-ahead adjustment needed based on the revised hour-ahead demand forecast. The short brown arrow between the two horizontal line segments represents capacity that, in this case, must be added to meet the load expected.
Currently, the CalISO interchange schedules and self-scheduled generation are changed from one hour to the next over a 20-min ramp period beginning 10 minutes before the hour ends.
On a more
granular scale, very five minutes, the CalISO economically dispatches its generation fleet to follow the expected load five minutes ahead of time. In the diagram, observe that “load following,” or the five-minute dispatch, is the difference between the hour-ahead curve and the red line defining current requirements every five minutes. Asset flexibility is especially important for load following.
To illustrate how challenging load following might become under an aggressive RPS, Loutan offered this example: A fast-moving cloud bank can knock out a 500-MW PV field within five to eight minutes. It’s unlikely such an anomaly could be completely accommodated with generation resources available to the system because of ramp constraints.
Additional support through dynamic schedules from a neighboring balance area, load reduction (via a demand-side management solution), and storage devi
ces might all be necessary to balance the system with a high penetration of renewables generation.
The black curve at the top of the chart represents the actual load demand, which also corresponds to the total generation requirement. Regulation, defined here as the difference between generation and load in real time (so-called “imbalance”), is not dispatched through the CalISO’s market software.
Rather, it is dispatched through the CalISO’s energy management system every four seconds to correct for deviations in system frequency and deviations from interchange schedules with neighboring balancing authorities.
With the ISO primer putting everyone in the room on the same page, so to speak, Loutan summarized the results of a study the CalISO conducted to identify the operational requirements and resource options needed to operate its grid reliably under the 2012 20% RPS and the 2020 33% RPS. Another objective of the study was to provide information required by other stakeholders—including state agencies, market participants, etc—for decision-making.
Loutan began by identifying the renewable portfolios assumed for the study (Table 1). Then he revealed the expected increase in regulation and load-following capacity requirements (within the hour) to meet those portfolios (Table 2). Note that in 2006, load was the only significant vari
able; for 2012 the variables primarily are load and wind, so capacity requirements increase significantly; in 2020 the variables will be load, wind, and solar, requiring a doubling of the 2012 requirements. Detailed study results were presented for the summer; however, spring, fall, and winter requirements are available at www.integrating-renewables.org.
Stand back for a moment and think about the variability that might have to be accommodated by the grid for the cases defined in the tables. Example: What if 10,000 MW of wind is forecast in the day-ahead timeframe and a substantial amount does not materialize during the operating hour? This scenario can create significant operational challenges—such as the ability to commit resources with the ramping flexibility to meet real-time variability.
As part of the study, CalISO investigated how operation of the combined cycles on its system would be impacted by the additional solar and wind generation assumed for the 2012 case compared to the baseline 2006 case. Results (annual basis) are presented in Table 3.
Loutan summarized the operational impacts of intermittent renewables, as defined in the study, this way:
Two additional considerations for the CalISO, the grid expert added, are these:
CalISO is actively pursuing operational and market enhancements to support renewables integration, Loutan said. The list of valued operational enhancements includes the following:
Market enhancements valued:
Summing up, Loutan segregated the resources required for renewables integration into three buckets: generation, storage, and demand response. Characteristics of the generation portfolio would include quick-start units, fast-ramp capability, wide operating range (especially the ability to back way down in load without exceeding emissions limits), and regulation capability.
Storage assets should enable the balancing authority to shift energy from off-peak to on-peak, mitigate over-generation, and provide voltage support and regulation. Demand response should be capable of frequency correction, provide rapid response to gaps in wind energy production, respond quickly to ISO dispatches, and be able to distinguish between loads that are price sensitive and those that are not.