With the cost of photovoltaic devices and wind power dropping dramatically, the economics of some forms of renewable power are becoming increasingly compelling. But these sources of power come with a significant limitation: intermittency. Solar can’t generate power around the clock (and output drops during cloudy days), while wind power can suffer from low output that can last days. There are various ways to work around some lack of production—grid-scale storage and careful matching of supply and demand—but some degree of what’s termed “baseline power” is needed to ensure the stability of the grid.

There are ways to provide this baseline power that don’t involve significant carbon emissions, like nuclear and hydro power. But those come with their own set of issues. So a group of European researchers decided to look into a form of renewable power that hasn’t attracted as much attention: concentrating solar power (CSP), sometimes termed solar thermal power.

CSP involves the use of mirrors to focus sunlight onto a liquid, rapidly bringing it up to extremely high temperatures. The resulting heat can be used immediately to generate electricity, or some fraction of it can be stored and used to drive generators later. Depending on the details of the storage, CSP can typically generate electricity for at least eight hours after the Sun sets, and some plants have managed to produce power around the clock during the summer.

Typically, a CSP plant is optimized for a mixture of generation and storage. But the authors of the new paper note that it’s possible to expand the area where the mirrors are located (called the “solar field”) relative to its power block. This may be less efficient economically, but it allows the plant to start generating more power rapidly at the start of the day and continue to store heat at the same time as generating power.

To determine whether this sort of approach could allow CSP to provide better baseline power, the authors looked at three scenarios: a flat power demand, one based on the European Union (where demand peaks in winter evenings), and one based on California, where demand peaks during summer afternoons. They also looked at three different levels of plant optimization: none at all, optimization of layout and equipment based on economic considerations, and a regional optimization. In this last case, the layout of multiple sites are coordinated in order to provide the best baseline power output.

If this sort of regional coordination can be achieved—and the authors don’t offer any suggestions as to how it could—then as few as 10 sites in southern Europe would be sufficient to provide 70 to 80 percent reliable baseline power at very little added cost. And that, the authors point out, is similar to the reliability of a typical nuclear plant. Looking at other regions of the globe, CSP would also provide similar performance in South Africa, but wouldn’t be as effective in the US and India. The problem in these locations isn’t the lack of good sites, it’s that the weather at the best sites tends to be similar (ie, a cloudy day at one site will often be cloudy at the rest).

The more we’re willing to spend on the CSP plants, the better optimized they can be, and the more reliable the power would be. But this gets into the biggest problem with CSP: it’s expensive. While it was competitive with photovoltaic power a few years ago, the price of PV has plunged, while CSP’s costs have only dropped slowly. We can expect continued declines in price, but it’s likely to remain one of the pricier options.

That situation should sound familiar, because it’s the same problem faced by offshore wind. Although the wind is much more reliable in offshore locations—and thus the power produced there is closer to a baseline quality—the cost of installing wind turbines in the ocean is significantly more expensive. The challenge here is that, once on the grid, the value of power generated by any source is treated equally, with no bonuses attached to the source being renewable or baseline.