The secret to low-water-use, high-efficiency concentrating solar power

Many readers have expressed interest in learning more about the water consumption of concentrating solar power and how measures to reduce it might impact system efficiency and cost.  After my recent CSP post, “World’s largest solar power plants with thermal storage to be built in Arizona,” Michael Hogan wrote in the comments (here) about a low-water-consuming cooling system he had experience with.  I asked Hogan, a long-time power industry executive and currently the Power Programme Director for the European Climate Foundation (bio here), to write a longer piece for Climate Progress.  Here is what he put together, with links and figures (click to enlarge).

EXECUTIVE SUMMARY:  If concentrating solar power (“CSP”) is a core climate solution, indirect dry cooling systems (also known as “Heller” systems) will be a crucial enabling technology, since large-scale CSP will be located in desert regions. US power companies have long favored direct dry cooling systems for fossil plants, probably because of the visual impact of Heller systems.  But Heller systems have long experience in certain regions and will probably play an important role in the success of large-scale CSP.  This is due to their higher efficiency, smaller footprints, quieter operation, lower maintenance, higher availability, and more flexible site layout.  Heller systems can reduce water consumption in a CSP plant by 97% with minimal performance impact.  The height of the cooling towers should be less of an issue in remote desert locations, especially since the central tower in power tower facilities will be of comparable height.

Concentrating solar thermal power plants (“CSP”) have been identified a number of times in Climate Progress as a core climate solution due to their almost unique potential to replace coal as the dominant supplier of baseload and/or firm dispatchable capacity to the world’s power grids.  It is said that CSP could represent 3 of the 12-14 wedges in the 450ppm solution — 20-25% of global mitigation potential.  I concur wholeheartedly with that view, and I applaud CP for its efforts to educate readers on the singular challenges of eliminating coal-fired power production at scale.   But if CSP is a core climate solution, dry cooling technologies, and in particular Heller systems, will be a crucial enabler (see note at the end regarding the status of the name “Heller” system).

One of the concerns often cited about CSP is water consumption, particularly because the technology’s reliance on direct normal insolation means that it is most economically located in desert regions.  Because most CSP systems rely on Rankine cycle steam turbine-generators to produce electricity, they face the same requirements as fossil-fired power plants for condensing large volumes of saturated steam back into boiler feedwater. (Parabolic dish systems use Stirling or Brayton engines to produce useful energy, each of which has its own advantages and disadvantages)  Where an abundant and cheap supply of water is available, the most efficient way to accomplish this is by evaporation (or “wet cooling”), which is what produces the large plume of water vapor one often sees rising from power stations.  Convective cooling using ambient air (“dry cooling”) requires higher capital costs and can reduce plant performance, and thus planners of fossil plants have sought to locate them close to adequate supplies of cooling water whenever possible.

In the desert areas where CSP will thrive, the consumption of large amounts of water by conventional wet cooling systems is clearly unsustainable.  Dry cooling alternatives will be required, and CSP will have to demonstrate its commercial viability despite the capital cost and performance penalties this will entail.  Fortunately this is an eminently manageable problem.

[Acronyms: “LEC” = levelized electricity cost; “O&M” = operation & maintenance]

Deutsches Zentrum fur Luft- und Raumfahrt e.V. (“DLR”), a German government research agency, presented a study in 2007 comparing a particular dry cooling technology, the Heller system, with wet cooling for CSP plants in Spain and in the California desert (see figures above).   Water consumption was reduced by 97%, and the performance impact was quite minimal.  Indeed the impact on performance in the higher desert temperatures of California was overwhelmed by the benefits of better annual insolation.  They also noted that the potentially negative impact of high daytime temperatures is mitigated by the use of thermal storage, which uses energy collected during peak daytime insolation to produce electricity when temperatures are considerably lower.  One interesting aspect of the DLR study was their focus on Heller systems over more familiar (at least in the US) direct dry cooling systems, and that is worth a closer examination.

Two basic types of dry cooling systems have long been employed where necessary — “direct” air cooling (usually called an “air-cooled condenser” or “ACC”) and “indirect” air cooling (often referred to as the “Heller system”, after Laszlo Heller, the Hungarian thermodynamics professor who pioneered this approach in the 1950s).  In ACC systems, the saturated steam from the steam turbine exhaust is carried directly to a very large array of A-framed fin-tube bundles, where large mechanical fans force air over the tubes, convectively condensing the steam.

ACC system

In Heller systems, the steam is condensed by spraying water directly into the exhaust flow in a ratio of about 50:1 (called “direct contact jet condensing”), creating a large volume of warm water, some of which is pumped back to the boiler as the working fluid and the rest of which is pumped to bundles of tubes arrayed at the base of a natural-draft hyperbolic cooling tower.  The warm water circulating around the base of the tower and the cooler air at the top of the tower, combined with the tower’s hyperbolic shape, stimulate a powerful updraft that draws ambient air over the tube bundles, thereby convectively cooling the water before it is returned to the condenser.  Both are closed systems.

Heller system [Acronyms: “CW” = cooling water; “DC” = direct contact]

While the Heller system has been widely used elsewhere, there are none in the US.  This is probably because the much lower auxiliary power requirements of Heller systems come with the visual impact of a large hyperbolic cooling tower (typically 150m high and 120m in base diameter), often a difficult sell given that most fossil power stations are located in the vicinity of the populated demand centers they’re intended to serve.  The auxiliary power required to run an ACC system is roughly twice the power required run a Heller system, and the Heller system is considerably quieter, but these have apparently been considered prices worth paying for the lower profile (a typical ACC system can be 40m high), particularly when it was cheap coal-fired power.  Simple lack of familiarity could be another factor in the hidebound world of US power utilities.

The Electric Power Research Institute has kicked off a comparative study of indirect dry cooling (due to be completed in mid 2010), on the theory that it is the most economic dry cooling solution for large-scale thermal applications.  The prospect of large amounts of CSP being built in the world’s deserts calls for a reconsideration of the relative merits of these two approaches, since it would require dry cooling to be deployed in a different application and to a far larger extent than has ever been the case.

Three Bechtel engineers published a paper in 2005 (Digital Object Identifier reference DOI:10.1115/1.1839924) (originally presented at an American Society of Mechanical Engineers conference in 2002) that compared cooling technologies for combined-cycle gas power plants.  They cited the following comparison of installed costs for various cooling systems, including ACC and Heller.

[Acronyms: “WSAC” – wet-surface air condenser]

They also note that the footprint of an ACC system is larger than that required for a Heller system, though specific data is not offered.  Overall system efficiency of a Heller system is in the range of 2% better than an ACC system.  That performance improvement meant one thing in a fossil power plant in the bad old days of cheap dirty power, but when it means 2% less land area covered by solar collectors, and lower auxiliary consumption of much more costly power, it takes on a much greater significance.  The same sources note that since the Heller systems are mechanically far simpler than ACC systems, maintenance is much less of an issue and system availability is significantly greater.  In the remote areas where these plants will be located, and given the large land areas over which they will spread, these are far more significant considerations than they were for compact fossil power plants located close to the populations they served.  Another factor noted in these sources is that an ACC must be located next to the steam turbine it serves, because of the cost of transporting saturated steam over any distance, whereas the Heller system has much more flexibility in where the cooling tower is located.  This will be much more important to CSP, where one can envision clusters of power tower complexes in a given area each with its own steam turbine, than it was with fossil plants.  And finally, the feature that most worked against Heller systems in US fossil plant applications – visual impact – should be far less of an issue in remote desert sites, especially with solar power tower complexes where the central towers will likely be of similar height.

I should note that as a senior executive of the private power company InterGen in the late 1990s I oversaw the deployment of a Heller system on our 2,400 MW gas-fired combined cycle plant in Adapazari, Turkey (see below), which is still the world’s largest installation of an indirect dry cooling system and continues to work extremely well.  I trace my enthusiasm for the technology to that personal experience.

One final note on the term “Heller” system.  A German engineering company, GEA, appears to own the trademark rights to the name “Heller”, which they acquired when the bought EGI, the Hungarian company that commercialized indirect dry cooling systems.  Indirect dry cooling is a generic technical solution that is often referred to as “the Heller system”.  I have no affiliation with GEA.

37 Responses to The secret to low-water-use, high-efficiency concentrating solar power

  1. David B. Benson says:

    Well done!

  2. charger says:

    Good information.

  3. ecostew says:

    I didn’t see a real bottom line on LCA energy efficiency when it comes to elevation and annual average daily temperature.

  4. oxnardprof says:

    Very interesting article. Is there any potential impact from the heat released at 150 m?

    I am wondering about impacts on weather, and possible environmental effects, since heat is being collected over a wide area, concentated and (partially) released in a small area, at an elevation.

  5. dwight says:

    Thanks! That was very informative and makes me optimistic for this to all come together.

  6. EricG says:

    This looks to me like a solar updraft tower. Put a turbine at the top and you can generate electricity from the waste heat.

  7. Jim Eaton says:

    Great information!

    Much of the opposition to siting solar plants in the California desert comes from proposals to build on ecologically sensitive areas and the use of water. Many of the plans call for “old” technology — mirrors that concentrate the reflected sun on a water tank to create steam. And the huge water use comes not mainly from the making of steam, but instead from the frequent washing of the mirrors. See:

  8. paulm says:

    What we need is a global alert level like that for the flu pandemic 1 – 6 which everyone signs up to for global warming.

    It would be up rated may be based on CO2 levels and extreme events such as fires and heatwaves.

    I wonder what the current level would be now taking into consideration the inertia of the system?

  9. Steve Bloom says:

    Thanks for that link, Jim. Note that DiFi is even now slicing and dicing a deal to specify allowable locations for the new plants. That’s clearly better than the default of having developers cut private deals for random sites, but we’ll see what she comes up with. A few questions:

    Re the mirror-cleaning water use, are there some numbers? Also, is it even a problem unless it uses local ground water?

    As I understand it, the advantage of large-scale CSP over roof-top solar PV is that it’s much cheaper (given present technology, although PV has a lot of room to improve) and can produce large amounts of power soon. Given that we do need a lot of power fast to displace fossil, what’s the flaw in this logic?

    Joe, have you posted on this subject? If not, it seems very topical.

  10. Leland Palmer says:

    Wonderful information, very good news.

    There might be a lot of technology that has been marginalized, which might end up being very significant as alternative energy becomes mainstream.

  11. To address some of the questions:

    ecostew – the link at the comment about the 2% efficiency improvement takes you to a slide presentation by a GEA EGI executive that includes a lot of useful information, including performance of various cooling technologies vs ambient dry bulb temperature. Dry bulb temp, seems to be the primary determining factor for performance, I suspect the mass flow impact of changing air pressure with elevation may be a minor contributor.

    oxnardprof – as you point out, the thermal energy exiting the top of the cooling tower is residual solar energy that would have been “collected” by the local ecosystem in some form anyway. A potentially more important concern is the reduction in albedo effect from the capture of solar energy that would otherwise have been reflected, which at the scale we’re talking about might become significant. That can be alleviated by, for instance, siting a given acreage of reflecting mirrors to compensate for the concentrating mirrors. As for the concentrating effect you note of heat from a wide area being released from the top of the tower, I’m not a meteorologist – perhaps someone else can blog on that, but I’ve not been aware of any concerns on that front.

    EricG – nice idea about extracting energy twice from the updraft, once for convective cooling and again for a turbine. Just remember that the efficiency of the cycle depends on the effectiveness of the convective cooling of the circulating water, and if you’re increasing the backpressure on the system by installing an energy extraction device at the top of the tower you’d need to ensure it was extracting more energy than you’re losing from the reduced cooling efficiency. Haven’t seen that calculation.

    Jim Eaton – thanks for the link. The site is obviously one that is concerned, and justifiably so, about the siting of these facilities, and requiring developers to use disturbed desert habitat seems to me a reasonable restriction. But as Joe and many others (including myself) have explained many times to people who think that these problems will go away if you just put enough PV panels on roofs, that solution has no prospect for decades to come, at least, of replacing the role of coal-fired power plants at scale, for reasons I won’t go into here. The amount of water used for mirror cleaning is factored into the DLR data I quoted; the reason the Heller system’s consumption is down 97% and not 100% is because of the mirror cleaning consumption. So 97% of a wet-cooled CSP plant’s consumption would be from cooling, 3% for cleaning and other auxiliary uses. Clearly the cleaning, while still a concern, is a minor one compared to the cooling issue, and that one we can clearly solve.

  12. Steve Bloom – The advantage of CSP is scale and the fact that the solar energy can be stored in thermal form for hours economically, making it a viable replacement for coal-fired power plants. As for price, CSP is a central-station solution and competes with other central station solutions, so it is indeed much cheaper than PV installed on rooftops, but of course it needs to be. PV installed on rooftops competes with the retail price for the power it is replacing, as long as it’s only displacing power that would have been purchased on premises (if it is selling net power back to the grid, it’s competing with a different price point). Since PV energy cannot be stored, you’re really talking apples and oranges in comparing simply the cost of power at the low side of the transformer.

  13. Jeff Green says:

    If a cooler is needed then it seems to me there is a lot of energy there that can be recycled. Plugging in recycled energy I checked out this site

    from there I found

    (Conversion of Low Temperature Waste Heat into Power –The steam-Rankine cycle is the principle method used for producing electric power from high temperature fluid streams. For the conversion of low temperature heat into power, the steam-Rankine cycle may be a possibility, along with other known power cycles, such as the organic-Rankine cycle.)

    organic rankine cycle

    It seems there is a possibility of extracting more energy out of this process before throwing it away in the atmosphere. All that effort to collect energy could be put to better use.

  14. EricG says:

    Michael Hogan – Yes, putting a turbine at the top of the cooling tower would reduce the efficiency of the cooling process, which could be a game stopper. So the engineering would be vital. Perhaps if the tower was designed with waste heat recovery as a goal this concept could be cost effective. Maybe the remote location would allow the tower could be taller or have a larger footprint. The concept is similar to BIPV, where multiple uses allow low efficiency generation to make sense.

  15. Bill Woods says:

    How do these cooling towers compare in size and cost to the sort of hyperbolic towers already in use at coal and nuclear plants?

  16. jcwinnie says:

    Quite informative. Small quibble, and am sure it is a matter of shorthand, still I remember I intially was confused by how the term is used. CSP (Concentrating Solar Power) can apply to either photovolatic or solar thermal.

    I wonder what factors decide when an integrated cycle system is a better investment than super insulated heat storage? Can construction of such large cooling towers be made more environmentally friendly (e.g., carbon footprint of concrete)? Are there easy ways to further increase functionality (e.g., TFPV exterior) of the towers? Do such systems have any impact upon the ability to save tillable areas from encroaching desert?

  17. Bill Woods,

    If you’re talking about coal and nuclear plants that use dry cooling, these are going to be similar in size and cost. As I noted in my post, these towers are already used for fossil fired plants in many parts of the world where cooling water is a problem. We’ve generally assumed in the US that water is plentiful and cheap, which is likely to change, and where we have used them it’s been for fossil plants with smaller thermal loads, like gas-fired combined cycle plants. China has several under construction now at coal sites, since China is already becoming concerned about water conservation in general, not just in areas that are currently constrained – one, for instance, is BaoJi, a 1400 MW coal plant. The tower is 170 m high and 154 m base diameter. There is only one nuclear plant with a dry cooling system, in Russia.

  18. Nick Kong says:

    I haven’t looked closer yet, but many skeptics of CSP do not believe the transmission lines are able to transfer electricity from the SW to the NE, where electricity demand is greater. How do you react to this, Joe?

  19. Leland Palmer says:

    Hi all-

    A potentially more important concern is the reduction in albedo effect from the capture of solar energy that would otherwise have been reflected, which at the scale we’re talking about might become significant. That can be alleviated by, for instance, siting a given acreage of reflecting mirrors to compensate for the concentrating mirrors.

    Had an idea back in the days of the Carter Administration, that white gravel could be spread on the ground under these CSP heliostats or collectors. The white gravel would help balance the albedo effects of the “dark” mirrors or collectors, resulting in an overall albedo balance. White gravel or crushed stone is also cheap, would help keep dust off the optics, and is similar to the natural patina of gravel found in many desert areas.

  20. Leland Palmer says:

    Hmmm… there was an idea for a “wind cyclone” wind power tower, back in the 1980’s, it was in Popular Science, I think. Vanes in a wind power tower channeled wind into a circular cyclone, which then exited out the top of the tower. A small wind turbine in the bottom of the tower harvested energy from this.

    If you’ve already got a tower, maybe you could get a little wind power from it, as a side effect.

    The amount of power generated would be relatively small, I think, but the heat caused updraft might increase the available electricity generated.

    On the other hand, trying to harvest power from this flow might be a bad idea, because it might slow the flow and cause efficiency losses in the cooling system, maybe. There might be trade-offs.

  21. Michael Hogan — Thanks for this most important and informative article. At last we have some small cause to rejoice. The Invanpah CSP project in the Mojave Desert will use no wet cooling, so that’s one obstacle out of the way. Maybe the developers will modify their stupid-looking bulldozing approach.

    Leland Palmer — Good ideas! Of course there is some tradeoff between the power harvested and the blockage of heat rejection due to the harvesting turbine, but a taller tower might compensate. The idea is to focus a thermal plume way up into the great windy heat sink of higher altitudes to force convection up through the tower.

  22. Another thought re the power harvesting turbine in the tall forced convection CSP dry cooling tower: the turbine would organize a vortex into the atmospheric heat sink which would help force convection up the tower, like a vortex expedites the flow of water down a drain.

    Organizing turbulence is overcoming entropy by intelligence and energy. Here’s an interesting fact from Low Life (a history of old New York by Luc Sante 1991): the first control of traffic at intersections in New York was in 1902 and the first one-way streets came in 1910. Organizing traffic flow makes a huge difference, and it is a recent innovation.

  23. Michael,
    Very informative article.
    Well done and much appreciated!

  24. Nick Kong,

    I know you directed your question to Joe, but I’ll offer a response if you’re interested. A single overhead cable of current HVDC technology is known to be capable of transmitting 7000MW over 2000 km with very modest losses (on the order of 6% I believe). That’s enough to go from New Mexico or Southern Colorado, excellent locations for CSP, to the east coast. Transmission should not be a prohibitive technical consideration, though it obviously presents a number of non-trivial non-technical challenges.

  25. Michael,
    A couple questions:

    What are your estimates for total water use for CSP with the Heller system per MWh?

    Also what in your understanding are the variations in geometry possible for the cooling tower in the Heller system?

  26. Michael Hoexter,

    The water consumption by the cooling system itself is effectively zero. I’m not an expert on water requirements for other purposes for a CSP (and jcwinnie is right, CSP can refer either to concentrating PV or concentration thermal – I usually call it concentrating solar thermal for that reason), which is primarily for mirror cleaning, so it will almost certainly depend on site conditions, and the need for washing will evolve with the technology, but I’ve seen a number quoted of about 0.08 m3/MWh. Water consumption for a wet-cooled system would be on the order of 40-50 times that much. As for possible geometries, I’ve seen Heller towers constructed from metal that approximate the natural draft induction characteristics of a hyperbolic tower with two “skirt” tiers above the tube bundles, of progressively steeper pitch and smaller diameter, with the final component being what is essentially a constant-diameter metal chimney above the second tier. There may be other geometries in use, but those are the only two I’ve seen referenced.

  27. Karel says:

    Very informative article!

    How does the size of the Heller tower compare to the ‘evaporative CS’ tower?

    Another advantage of dry cooling is that there are no clouds produced that could shadow the mirrors.

  28. Karel,

    Apologies for neglecting to clarify the acronym “CS” – it stands for “cooling system”. The size of the Heller towers and the “wet” or evaporative towers are roughly comparable. The evaporative towers might be a bit smaller due to the better efficiency of evaporative cooling, but it’s not so much that you would really notice. The natural draft wet cooling towers can be 200 m high and more. Mechanical draft wet cooling towers, which use more auxiliary power, are the ones you typically see at fossil power plants and are very low profile. People in the US tend to associate the natural draft wet towers, like the one shown in the picture, with nuclear plants, which is where they tend to be deployed here because of the very large quantities of thermal energy that must be rejected.

  29. Nick Kong says:


    Thanks for the info- do you think you guys can do a longer piece/update on transmission and where we need to go from there to transmit wind/CSP electricity? I mean, I know what the challenges are but what are some potential solutions, what needs to be done in policy, etc. Also, with HVDC, isn’t there a loss in conversion from DC to AC?

  30. Nick,

    Good idea about a longer post on transmission, but of course that’s up to Climate Progress. There are indeed losses associated with rectification from AC to DC and inversion from DC back to AC; those losses are included in the loss figure I quoted above.

  31. David B. Benson says:

    Here is my understanding of HVDC losses:

    rectifcation/inverter total loss: 1%
    transmission losses: 3% per 1000 km.

    UHVDC probably does a bit better.

  32. Eric M says:

    Leland Palmer –
    Wilmot McCutchen –

    Perhaps something along these lines:

  33. Leland Palmer says:

    In the Las Vegas area, if I recall right, what happens to glass surfaces exposed to the environment is that they cool by radiation at night, collect dew, and the dew makes mud when exposed to dust.

    If this cycle could be broken, the cleaning requirements for the mirrors might be less.

    If, for example, the heliostats were filled with air channels for hot air from waste heat, and the back surfaces of them were used for radiative cooling, pointing at the coolest part of the upper atmosphere at night, this might keep the heliostat mirrors warm at night and keep dew from condensing on them. Very small water channels (like small polyethylene tubing) applied to the back surfaces of the heliostats might be a better way, and would not require heat exchangers.

    And this might also be a way to increase the efficiency of the plant by rejecting some waste heat by radiation. This idea could also be applied to parabolic troughs and non-imaging flat plate fresnel mirrors, of course.

    If it raises the cost of the heliostats or mirrors, though, for CSP, it might not be worth doing, even if it decreases water use for cleaning. The cost of the optics is the majority of the cost of solar power tower CSP, I think.

    This Heller cycle cooling is a real problem solver, IMO! It’s a mature technology, with several installed installations around the world. Good deal, good find, I think.

  34. danl says:

    There was so much thermodyamics in this post. It brought out the engineer in me.Thanks for the analysis, Mr. Hogan.

  35. danl (and everyone else), I was pleased that Joe asked me to post, as I did think this was an important aspect of a core climate solution that was not widely understood. I have very much enjoyed the dialogue, and I am encouraged by some of the ideas that have come out of it. One thought that did occur to me regarding Jim Eaton’s very valid concern about the bulldozing aspect of the Ivanpah project – it is a necessary downside of trough technology that it requires a broad expanse of laser-leveled real estate for the row upon row of long, linear troughs. The power tower technology, which probably offers a greater range of storage strategies, is also not as demanding of the topography – each individual heliostat can be focused on the collector at the top of the tower, to some extent regardless of its position relative to the rest of the heliostats, using relatively inexpensive electronics. That may become another deciding factor in the choice of trough vs. tower.

  36. Nichol says:

    A power-tower system could obviously have one integrated design for cooling tower and the power tower. Is that already being done?

    Also: if the tower is not too noisy, (or hot?) it might be attractive to add appartments with a view to it, on the side not hit by the sun from the mirrors. This would be the sunny side of the tower, so they would need sunshades.

  37. thisi is a very sophesticated topic I like to cooporate with your respective to aschive optimum desgien and cheepest cost for 1KW/hour
    SEnior expert in elec power system EGYPTIAN CONSULTANT ENGINEER
    Mohamad Salah Eldeen Madbouli