(The author is a Reuters market analyst. The views expressed are his own.)
By Gerard Wynn
LONDON, April 26 (Reuters) - Solar power, fast catching up with global wind capacity, presents its own problems for grid operators balancing increasingly variable supply from renewables with the demand fluctuations they were designed to handle.
The challenges of solar power may be less severe than those posed by wind turbines because solar supply peaks are much more predictable, while surges can be briefer.
Wind power variability can be forecast with reasonable accuracy over intra-day timeframes.
Solar power variability meanwhile can be divided between, first, a perfectly predictable element which is the position of the sun, and second, uncertainty according to cloud cover.
A particular problem for grid balancing is posed by extreme short-term variability, or "ramping events", which in the case of solar can be associated with broken cloud cover, and which can occur over shorter periods (in seconds, rather than minutes) than with wind.
Such brief surges can be very localised, however, for example where the shadow of a cloud may be limited to just a part of one solar power plant, meaning balancing the grid across a wider area can dramatically reduce variability.
Meanwhile the fact that maximum solar power generation is known for any time of the day, as defined by clear-sky sunshine and the position of the sun, places a limit on required stand-by or reserve power generation.
A report from Germany's Fraunhofer Institute, published in February, plotted actual daily, weekly and monthly solar and wind power generation in Germany in 2012.
It shows how solar power generation is less variable on both daily and weekly timescales compared with wind. (See Charts 1 and 2)
Smaller daily solar power variation reflects the fact that there is always some daylight over a 24-hour period, notwithstanding peaks and troughs, and the fact that peak solar output is ultimately limited by the power of the clear-sky sun.
That can be of use to grid operators who have a clear, daily trajectory of maximum reserve power that they have to allocate against a loss in solar power.
Solar power is as variable as wind over shorter timescales, but, here too, its characteristics may be easier to handle.
Chart 1: (slides 27-28) goo.gl/YlRPO
Chart 2: (slides 21-22) goo.gl/YlRPO
Chart 3: (page 3) goo.gl/p3dFB
On a daily basis, averaged solar photovoltaic output follows an up-turned "u", rising from sunrise, peaking around noon, and falling back towards sunset.
But actual observations from individual power plants show how output can fall and rise by as much as 75 percent in seconds as generation ramps up and down between full sun and clouds.
Such ramping events are similar to wind power, but briefer, as shown in a report published by the U.S. National Renewable Energy Laboratory (NREL) in January, "Identifying Wind and Solar Ramping Events". (Chart 3)
An NREL report published in 2010 used a modelling approach to assess the grid impact of a hypothetical scenario where solar (photovoltaic and thermal) power generation accounted for 25 percent of total generation in five western U.S. states, far higher than presently.
The total, modelled solar power capacity was 29 gigawatts (GW), and the model produced hourly variation in solar generation of up to 15 GW, with serious implications for the ability of other power generation sources, called net load, to balance the grid.
The researchers acknowledged that they did not fully account for the fact that such a large amount of solar power would occupy a wider geographical area, reducing variability.
Nevertheless, the study, "Impact of High Solar Penetration in the Western Interconnection", illustrated serious impacts from very high levels of solar power.
Those impacts included: greater power exports to surrounding regions; displacement of peaking gas and combined cycle gas turbine (CCGT) generation; and more variable coal generation.
Such modelled phenomena were actually observed in Germany last year, when solar power generation rose 44 percent compared with 2011, reaching 5 percent of gross power generation, and which undercut returns to gas-fired power.
A wider grid area will capture a wider range of weather and so smooth renewable power variability, and it appears that the benefit especially applies to solar power.
The effect was demonstrated by NREL researchers on the small Hawaii island of Lanai, which is vulnerable to variability as an isolated grid, in a study published last year, "PV Ramping in a Distributed Generation Environment".
They showed how combined solar power plants always had less variability, defined as monthly ramps in power generation at the 90th percentile.
The ratio of ramps up or down at individual to aggregated solar plants was always larger than the ratio of their installed capacity, showing how aggregating plants made the ramps smaller than otherwise expected.
They found that the smoothing effect was greater in summer, which they attributed to more broken clouds which increased differences across individual power plants.
NREL recommends a wider grid balancing approach to tackling renewable power variability, for example appearing to downplay the role for energy storage.
Balancing area is defined as the area across which grid authorities have a responsibility to balance power supply and demand, including the use of imports and exports, and could be a region, whole country or countries.
The implication is that greater cooperation between grid operators, including new operating procedures, market rules and expanded transmission capacity, can do much to offset the variability of high levels of solar power. (Reporting by Gerard Wynn; Editing by Anthony Barker)