CSPonD (CSP On Demand) - CSP heads for the hills

Like existing CSP power-tower developments, the MIT concept uses an array of mirrors, called heliostats, to ­reflect sunlight onto a receiver that powers steam turbines.

But instead of mounting the reflectors on the ground to concentrate sunlight up to a receiver tower, the design would spread the heliostats across a hillside, ­directing the insolation into a roofed vat of molten salt acting as receiver and ­energy storage reservoir.

The concept, called CSPonD (an acronym for CSP on Demand, for its capacity to produce power round the clock), was hatched by MIT researchers Alexander ­Slocum, Charles Forsberg and Jacopo Buongiorno.

Their idea was to find a way to use molten salts, some of which can be heated to more than 1,000°C, as a receiving fluid to overcome the inefficiencies linked to low steam temperatures, thermal ­fatigue and energy flux in conventional CSP power towers.

“They had the initial idea of ­using a heat transfer fluid in a power-tower system as a... receiving fluid, such as you’d find in molten-salt nuclear reactor concepts, but doped with nano­particles to darken the salts and so improve their absorption,” says MIT's Danny Codd, who completed his PhD on the concept and is heading a team aiming to commercialise CSPonD.

The concept has evolved from high-cost “beam-down” designs, where tower-mounted secondary mirrors concentrate sunlight ­focused upwards by heliostats, to a ground-level unit housing an expensive fluoride-salt-based ­receiving fluid.

The system now being advan­ced would be built on a site with a suitable “natural topo­graphy”, with hillside heliostat arrays ­focusing concentra­ted sunlight through a small opening in an insulated domed tank, where it would penetrate four to five ­metres deep into a molten salt mixture.

The heated molten salt would be pumped from the top of the tank through a steam generator to produce electricity, then returned to the bottom to cool. A barrier plate, serving as a physical and thermal barrier between the hot and cold salt layers, would rise and fall within the tank to maintain the volumes needed for continuous operation.

Codd says that depending on local demand, a system built around a 2,500-cubic-metre tank and a “modest” semicircular heliostat field could run a 4MW turbine non-stop powered by seven hours of sunlight and 17 hours of stored energy. The same-sized system could, alternatively, supply a 50MW turbine for three hours and 15 minutes without additional solar input.

“The main advantage of the CSPonD system [compared with traditional power towers] is ­simplicity and robustness,” says Codd.

“Conventional power towers­ are surface-based: they have to limit flux [the total rate of energy emission] on their recei­vers and transience so that they don’t fatigue the [receiver] tubes as the transfer fluid is pumped through. In our... concept­ this is not an issue because the hottest point in our system is our heat transfer fluid [because there are no receiver tubes that could overheat].”

The tube-based receivers in conventional CSP systems also have high reflectivity losses to the ­environment, so they are less ­efficient. Codd claims that “in ours we are containing that within the volume of a cavity-type receiver, so overall losses are reduced”.

Flux limits in a CSP system set the cap on how much wattage of solar energy can stream in ­before a system starts to melt down. Codd points to current power towers that can handle “at most 800 or 900 tonnes” — 800kW per square metre — of flux before ­internal tubing starts to degrade.

“Re-radiation, convection and conduction losses are all reduced as you increase your flux levels,” he says. “For [the CSPonD] the theoretical flux-level limits are up around 10,000 tonnes.”

Levels of flux also affect the choice of heliostats once the 1,500-tonne threshold is crossed, needing higher-precision tracking mirrors and finer optics ­technology.

“No-one is really exploiting the upper limits [of flux levels] where your receiver becomes much more efficient,” says Codd.

“And because we are using a heat-transfer fluid as a receiving medium, the system offers an ­inherent storage capability that you really don’t have with conventional designs, where you are pumping salt through a heat exchanger and then in separate tanks, hot and cold.”

Fleshing out the possibilities of the concept, the MIT team looked at two sun-rich rocky ­areas in the western US using computer modelling.

They calculated that concentrating sunlight with hillside ­arrays into sodium-potassium ­nitrate salt tanks 25 metres in diameter and five metres deep at sites in White Sands, New Mex­ico, and China Lake, California, could allow 20MW installations to be built, with enough storage to generate a day’s power for every ten days of accumulated solar energy.

Casting the net across 10,000sq km in the two areas suggested a potential 40GW using CSPonD.

“Besides looking at overall ­system operation and thermal performance, a heliostat site ­selection code was developed to take into account the solar ­resource of a location and its ­topography,” explains Codd.

“Just in a first appraisal of White Sands and China Lake, ­numerous sites were identified where systems could be located. In total, the two areas had vast solar electric potential.

“The coming years will be a proving ground for next-generation­ CSP. Once the current projects that are under development come on line — even if they are somewhat low-efficiency­ or low-tech, it will ­really open the door to our ­getting the newer technology out there at a quicker pace.”

This ceiling is worth its salt...

Derived from the molten salt baths used in industrial metal treatment, the CSPonD tank is lined with “firebrick” — mortarless insulating refractory brick — set in a carbon-steel shell coated with ceramic-fibre insulation.

A “freeze plane” is formed within the brick that keeps the carbon-steel tank walls from ­being exposed to the corrosive ­effects of the molten salt. Heat losses with this design are estimated at 0.1% per day.

The roof, also made of firebrick, is backside-cooled, so that rising salt vapour condenses on the inner surface, much as frost collects on the evaporator coils within a refrigerator.

During operation of a CSPonD development, the crystalline ­layer of salt will become thicker, increasing the thermal resistance that condenses the vapour, while the ambient heat within the tank — more than 1,000°C — causes the salt crust simultaneously to melt and return to liquid in the pond below. This salt-encrusted­ ceiling is expected to reflect incoming light off the surface of the molten salt.

An alloy divider plate that is ­resistant to corrosion separates the hot and cold salt volumes within the tank, moving up and down “like a raft” as ­energy is collected and depleted.

As the system is powered up and the top section of salt is ­heated, the negatively buoyant divider plate lowers; then, at night or in cloudy conditions, it rises, driven by actuators, to continue to feed the system turbines to produce electricity.

Short bounce opens up new angles Fieldwork has thrown up a new angle on the CSPonD. As well as arraying the heliostats on a south-facing hillside to focus sunlight down onto the molten-salt receiver, modelling has pointed to the possibility of building the receiver in an elevated location, with the array below. An idea first investigated during the 1970s, the “virtual tower” configuration has the potential to create optical efficiencies 10-12% better than in a layout with heliostats set above the receiver. However, insolation directed up into the receiver would need to be reflected down into the molten salt mixture, adding the cost of additional internal mirrors. Massachusetts Institute of Technology (MIT) team leader Danny Codd is keen to explore a variation on this design — dubbed a “short bounce” receiver — that could take advantage of the crystalline build-up that the salt vapour leaves on the ceiling of the tank, to reflect part of the incoming light down into the salt mixture (see story above). “In the high-temperature CSPonD receiver, the salt itself will form a crust that is very reflective, so this could enhance the efficiency of a ‘short bounce’ design, offsetting some of the losses [of indirect input of solar energy],” he notes. The “short bounce” CSPonD design has been put forward for funding through the US SunShot programme, the Department of Energy’s scheme to incubate technologies that have the potential to bring solar production costs down to parity with conventional energy. In Europe, MIT and the Cyprus Research Institute are developing a 100kW demonstration CSPonD unit that would have a four-metre-diameter receiving unit holding 11 tonnes of nitrate salt. It would be able to run continuously on 2.4MWh of daily solar input, with potential for 15 hours of storage. Should the bid for SunShot money come good, a similar-sized unit would be built in southern California. Preliminary capital cost estimates for a 20MW plant with a heliostat field of 345,000 square metres are put at $118m, with operating costs of $1m a year. The levelised cost of energy from this scale of development is $0.08-0.30 per kWh. In the longer run, salt will be critical to commercialising CSPonD. First proof-of-concept designs were based around a tank filled with molten nitrate salts such as have seen use in some concentrating solar power storage systems. It is cheap and available off the shelf, but the chief limitation with this class of salt is that it begins to degrade at about 550°C. Attention is turning to chlorides and carbonates, due to their thermodynamic stability and high operating temperature. Particular interest is being paid to a lithium, sodium and potassium mix that melts down at 430°C and can be heated to up to more than 1,000°C without degrading. These salts are much less expensive than nitrates — which can be bought in industrial bulk for $1 per kilogram — but because they have a higher temperature ceiling, they could make it possible for the development of more efficient systems that use supercritical CO2 instead of steam to drive the turbines.