Bouncing Around in the Sun

I’ve had occasion to think about concentrating solar power again, which has somewhat faded in interest as photovoltaics have rapidly gotten cheaper and more efficient. There has always been something about CSP configurations that has bothered me, something that demands a diagram.

I was lucky enough to have an office in Keck Laboratory and building lore still remembered Caltech’s rooftop solar furnace. This is one of the reasons why I think anyone who believes basic metallurgy is out or reach of decarbonization really hasn’t done the math for themselves yet. This wasn’t even an electrical furnace; this was a reflector in the sun and it reached 3,400C. And I, an intrepid grad student, got to hang out on that same roof and trip over the same rails presumably relocated to that rooftop as the Keck laboratory wasn’t built until 1960 and the solar furnace had been operational since 1922. The details of this amazing apparatus are here, a truly accessible piece of writing from Dr. Pol Duwez.

The key diagram for the lens arrangement is this, and it neatly leads to my big question about Concentrating Solar Power facilities and why they are all “doing it wrong” from my outsider’s perspective.

Fundamentally, you can move the light to wherever you need it and because you are working from the heat flux of the surface of the sun you have plenty of flux to spare (even losing 0.15 at each mirror). So, why not add another mirror to the system to put that point of focus in a place where the thermal mass may be defended through periods of null flux (night, clouds, large bugs)?

Look at this typical CSP configuration, mirrors around a central tower elevated holding the heat target. The tower has prominence to simplify aiming and targeting the mirrors. But really, we have been able to aim mirrors since Archimedes time; that is not the optimization we should be favoring here. Further, the tower is difficult to manage because it has the working components of the power plant and a massive hot object, meaning we have to defocus all mirrors in order to service the facility — ruining our careful configuration at each stoppage or inspection. Better to solve this with another mirror, more easily managed, and redirect the focal point to a target more accessible to servicing and maintaining a heat reservoir.

I’m of course talking about a configuration that looks more like this with a central ring-mirror on the tower that moves up or down into the reflections from the heliostats to bring the light down onto the thermal target underground at the base of the tower. This underground thermal target provides ample space and access for the power generating facilities as well as the potential to close the shaft to conserve and limit heat loss. Put an iris at ground level and you cap the thermal target once the heat flux drops off. This also provides a working environment that is not exposed to the elements if sensitive components must be serviced. The defocusing of the array is accomplished at the tower-ring-mirror, simply raise or lower out of the beams and there is nothing left at the point of intersection to interact with — even if there was that’s not where the focal point is anyway.

The central tower has a larger diameter to facilitate an additional redirection, so maybe some inner ring mirrors could be lost, but the center ring has the smallest reflecting area of the array and the middle of the normal CSP tower arrays already have a significant carve out for the generating facility and operational overhead. Since all that could be underground now one might even get more mirrors in the inner ring rather than fewer. I don’t think that really matters, the 1st order advantage here is conservation of the thermal mass when the sun goes down resulting in greater conversion to power in discharge mode. There are a variety of secondary advantages from reduced servicing danger, navigational hazard, bird threat, materials selection for the tower (doesn’t have to hold up a really hot target), mirror management, etc. The central tower will cast a larger shadow, maybe, well probably. Making it really tall just creates a larger navigational hazard and making it really wide and low just means it is in the way of the sun for much of the day for many of the mirrors. There is probably a relatively simple optimum here depending on site and the transit of the sun, very different in extreme latitudes than towards the equator. With heat to spare during daylight hours these losses seem much less important than the duration of night where no additional input heat can be expected at all. Retaining that hot reservoir for nighttime generation seems more difficult than loosing a little of the excess during peak heat flux (when we’re only operating at 600C in a system we should clear 3,000C against a theoretical limit of 5,600C).

It has always felt to me that folks were just excited to get the thing working in the first place they never went back to make sure it was optimally configured for the stated purpose. By shifting the thermal target underground you could even go nuts and develop a chimney effect over the facility and generate from a turbine within the mirror-ring on the tower — ala solar updraft tower.

Large arrays have mostly moved to photovoltaics due to simplicity of form, growing PV efficiencies, and declining costs. CSP has moved into smaller mirror-contained systems over time that have less need for multiple mirror coordination (parabolic dish and trough), so all this mucking about with extra mirrors and managing the focal point is somewhat moot, but it has always struck me as a missed trick on this development pathway, and it all comes from not really appreciating just how much wasted power and potential there is in a CSP plant operating at even 1,100C. You can go so much hotter with so many fewer resources… and we were doing it 100 years ago.

I suppose heat isn’t everything, but those hard to decarbonize industries point at their 900–1,300C operating temperatures like an insurmountable barrier to progress and… it just isn’t. It is literally child’s play.