California’s renewable portfolio standard, which requires utilities to source 33 percent of their power from renewables by 2020, will test how such a high share can be integrated into the grid cost-effectively and without compromising reliability.

“Integrating large quantities of variable energy resources (predominantly wind and photovoltaic solar) into the North American bulk power system requires significant changes to electricity system planning and operations,” said a joint report by the California Independent System Operator (CAISO) and the North American Electric Reliability Corporation (NERC).

“Operationally, an increase in wind and solar resources continues to challenge operators with the inherent swings, o ramps, in power output,” noted the report on “Maintaining bulk power system reliability while integrating variable energy resources: CAISO approach”, issued on Wednesday.

“In certain areas where large concentrations of wind resources have been added, system planners are accommodating added variability by increasing the amount of available regulating reserves and potentially carrying additional operating reserves.”

But maintaining higher reserves will add to the cost of electricity as conventional power plants must be operated below their efficient capacity or held in a high state of readiness.

CAISO’s efforts drew an endorsement from NERC as a case study for other grid managers in North America. By extension they will be useful for countries in Europe, such as Britain and Germany, which are connecting growing amounts of wind and solar to their bulk power systems.

Following CAISO’s example, NERC expects the challenge to be met through a combination of market pricing, technology enhancements, and new reliability requirements for wind and solar generators to force them to meet standards that are already common for fossil fuel, hydro and nuclear generators.

Conventional fossil fuel, hydro and nuclear generators contribute to a number of aspects of grid reliability.

Conventional power plants have a highly predictable operating performance and well-understood reliability characteristics, and grid operators have accumulated many years of experience with how they behave.

In California, conventional generators receive dispatch instructions from the automatic grid control system every four seconds and must respond fully to any instruction to ramp up or ramp down their production within five minutes.

In response to excess supply or demand, the turbines at coal- and gas-fired power stations, hydro plants and nuclear generating stations, speed up or slow down very slightly.

By instructing generators to increase or reduce the flow of gas, steam or water to the turbines by opening or closing the throttles, grid managers can offset small imbalances in seconds, and maintain grid frequency at a steady level.

But enormous amounts of energy are stored as inertia in these giant rotating turbines, which helps smooth short-term imbalances between power supply and demand across the network.

Conventional plants are also required to provide extra reactive power to the network in response to any drop in the voltage, and must be able to “ride through” a large short-term change in frequency or voltage without shutting down to maintain grid stability rather than risk a cascading failure.

Power from variable energy resources cannot be forecast and dispatched the same way. Output from variable energy resources, especially solar photovoltaic farms, can change drastically within minutes, particularly because of cloud cover.

While geographic diversity can smooth some of the variability, the net impact can still be large in sections of the grid.

Photovoltaic generators have no inertia at all, and wind farms have virtually none unless they are specifically designed to provide it.

“With large amounts of certain types of variable energy resources, particularly solar PV, a potential decrease in system inertia and frequency response could result in reliability impacts on the bulk power system,” according to CAISO and NERC.

“Compared to conventional generation, variable energy resources are less effective in providing the system with sufficient inertia to arrest frequency decline. Variable energy resources may not create adequate ... response to stabilize the system following the loss of a large generator.

“While variable energy resources are able to displace energy requirements, and capacity to a lesser extent, other essential reliability services traditionally provided by conventional generation must also be replaced,” the study concluded.

If variable energy resources are harder to predict and control, other sources of generation and demand must become more flexible to maintain power quality and grid reliability.

Other generators will have to respond to the sudden ramp up or ramp down of wind and solar by changing their own output quickly.

But problems occur when conventional generators are operating near full capacity (when there is little room to ramp up further) or at very low capacity (when more output reductions become hard to achieve without compromising their efficient operation).

To offset the increased variability of wind and solar, conventional generators, especially small gas units, will have to run below full capacity, to provide greater flexibility for ramping.

Running generators part-loaded will reduce their efficiency and increase operating costs. “CAISO may require subeconomic operation to meet frequency response obligations,” the study acknowledges.

There will also be “more frequent dispatch instructions and the starting and stopping of flexible, gas-fired generators, which will potentially incur more wear and tear”, according to NERC.

“Increased wear and tear can reduce (average) time to failure on generation components and potentially lead to increased forced outage rates.”

Lower capacity utilization, coupled with lower prices in the wholesale electricity market, could leave gas-fired generators with inadequate revenue to cover their costs and make a sufficient return for investors, as it has in Germany.

In response, CAISO has introduced rules that require wind and solar generators to turn down their output or even disconnect from the network in response to an instruction from grid controllers in both normal and emergency conditions.

Since 2010, CAISO has required new wind and solar generators with more than 20 megawatts of capacity to provide reactive power capability and maintain voltage controls through inverters or other reactive power devices.

The biggest problem remains forecasting short-term changes in solar and wind output, which forces grid operators to keep larger amounts of conventional capacity part-loaded.

“Wind and solar energy production is dependent on localized weather information. The electrical grid has evolved to the point at which operators need to know what the weather will be within a one-square-mile area in the next two hours or less,” according to NERC.

On one day in February 2008, the Texas electricity grid operator, ERCOT, was forced to curtail power to some customers after wind farm output dropped more quickly than expected.

Since 2004, CAISO has received hourly energy forecasts for wind farms from an external forecaster. But the forecasts “are heavily weighted on the current weather condition”: In other words, they tend to predict more of the same and are not good at identifying sudden changes.

Solar forecasting is even less well-developed because forecasts have to be able to predict cloud cover.

ERCOT now employs a wind-ramp tool developed by AWS Truewind to help plan. The large-ramp alert tool warns grid operators up to six hours ahead of large and rapid increases or decreases in wind output.

The forecast includes the likely size and duration of the wind ramp, as well as the probability that it will start in any particular time interval. Similar systems are being developed for solar forecasting.

Forecasts are crucial. If forecast accuracy and confidence levels can be improved, grid operators will be able to identify periods of risk or uncertainty and take mitigating action, while reducing (expensive) reserves at other times.

n John Kemp is a Reuters market analyst. The views expressed are his own.