DUBAI // In a sunny field in southern Spain sits a field a mile wide of massive mirrors arrayed like glinting chrysanthemum petals around a slender tower.
This is Gemasolar, a new solar-power plant backed partly by Abu Dhabi's sustainable energy company Masdar. Since it began operations in April, it has achieved a breakthrough: it became the first solar power plant in the world to produce electricity continuously for 24 hours.
Since then, it has repeated the feat more than 30 times.
Earlier this month, the plant in Fuentes de Andalucia, just east of Seville, was inaugurated at a ceremony attended by Emirati and Spanish dignitaries.
The aim is eventually to produce electricity round the clock through the sunny half of the year and for up to 15 hours a day the rest of the time. Each year it should produce 110 gigawatt hours of electricity - enough to power more than 25,000 homes.
Now the challenge is to refine the plant, to maximise its output. "We are still learning on a continuous basis," said Alvaro Lorente, chief executive of Torresol Energy, which has built and is running Gemasolar.
Torresol Energy is a 60-40 joint venture between the Spanish engineering and construction firm Sener and Masdar.
The idea behind Gemasolar is simple enough. It transfers the heat energy from the sun to a new, more useful medium - molten salt - that can store enough for use during the day and overnight.
Designing and building the plant took three years. It consists mainly of 2,650 huge and highly reflective mirrors - heliostats - arranged around a 140m central tower.
These reflectors direct the sun's rays to a "receiver" at the top of the tower. Molten salt is pumped from one storage tank up to the receiver, where it is heated by the rays, and then lowered into a second storage tank. The energy from the hot salt is used to power a traditional steam turbine, which generates electricity.
Previous plants have featured a tower surrounded by mirrors - a system called "concentrated solar power" (CSP) - but not used molten salt. Others have stored energy in molten salt, but transferred it from the sun using oil as an intermediary "transfer fluid".
But oil, an organic compound, decomposes above 400C. Salt, a more stable, inorganic compound, can handle temperatures of up to 550C.
The heliostats generate temperature in that range in the receiver as they track the sun, moving every 20 seconds to reflect as much energy as possible at the tower. They reposition at the command of control centres with which they continuously exchange signals.
Each slightly concave, rectangular reflector spans 120 square metres, the size of a two-bedroom apartment.
Together they can concentrate 1,000 times more solar radiation on to the receiver than it would naturally get. The effect is akin to a magnifying glass focusing radiation on a single point, making it much hotter.
Designing and setting up these heliostats posed one of the biggest challenges. They had to be large to reflect as much radiation as possible, but not so large that they would wobble in the wind or need too much costly steel support.
With so many reflectors to track, the team also had to build a separate control system just to manage the heliostats. The other system runs the rest of the plant.
"It has to manage over 100,000 signals," said Mr Lorente. "Each heliostat has a lot to be controlled - its position, the power it is consuming, the signal coming from the control system to order the heliostat to move or not to move."
Simply arranging the cables needed to connect each heliostat to a control system and a power source proved one of the most difficult steps in figuring out how to build the plant.
"You have thousands of cables all over the place," Mr Lorente said. "If you decide to go for a construction strategy that is not very well organised, you will spend a lot of money and a lot of time.
"You may have to remove some parts of the installation and put them in again."
Designing and building the rest of the plant went smoothly, largely thanks to previous experience.
The team chose the receiver's size based on two years of testing done more than five years ago by the Spanish partner, Sener.
They used the same molten salt mixture - 60 per cent potassium nitrate and 40 per cent sodium nitrate - used in other plants.
And they relied on established industry practices to manage the salt; too hot and it can cause corrosion, too cool and it can solidify.
It is also thicker than oil and thus harder to move around - especially up a 140m tower, said Belen Gallego, the founder of CSP Today, an online publication that tracks CSP developments.
Using molten salts is "an evolutionary step in the right direction ... [but] it is a technical challenge," she said.
To prevent the salt being exposed to the atmosphere and cooling down, the valves and pipes at Gemasolar are clad with heated electric tape. Standard stainless steel equipment minimises the risk of corrosion.
The focus now is operating the plant more efficiently, Mr Lorente said. "We are improving the operation of the plant," he said. "There are not big things in the plant to learn about, but a huge amount of small details."
One area ripe for improvement is the time it takes to turn the plant off and on - currently nearly two hours, time that could be used instead to generate energy.
At start-up, the receiver needs to be warmed up, and then unheated salt needs to be pumped up to the top - which on less sunny days can take as much as an hour and a half. Shutting down, which involves draining the unheated salt back into its storage tank, takes about 20 minutes.
And there is waste, too, in the form of the energy used to operate the plant. This currently drains a tenth of its theoretical total 19.9-megawatt "nameplate capacity". Mr Lorente hopes to cut that "parasitic consumption" by 30 to 40 per cent.
Torresol Energy is also trying to figure out how to predict when equipment is about to fail, especially the most vulnerable parts in direct contact with the hot salt.
It measures all sorts of physical symptoms, such as vibrations and temperature. But the engineers have yet to figure out exactly which signs indicate imminent failure. "The challenge is to interpret the values," Mr Lorente said.
He hopes the lessons learnt at this plant will feed into future ones.
"What is important is to have a continuous pipeline of projects," he said. "If you don't have more projects you cannot apply these improvements."