Pump Up the Storage

If we adopt solar and wind as major components of our energy infrastructure as we are weaned from fossil fuels, we have to solve the energy storage problem in a big way. An earlier post demonstrated that we do not likely possess enough materials in the world to simply build giant lead-acid (or nickel-based or lithium-based) batteries to do the job. Comments frequently pointed to pumped hydro storage as a far more sensible answer. Indeed, pumped storage is currently the dominant—and nearly only—grid-scale storage solution out there. Here, we will take a peek at pumped hydro and evaluate what it can do for us.

Gravitational Storage Basics

When you lift an object, you must supply a force to counter gravity (the weight of the object) and apply this force over the height through which you lift the object. The weight of an object—and therefore the force applied to lift it—is its mass times the acceleration due to gravity (application of Newton’s F = ma; in this case, mg, where g is the gravitational acceleration, or about 10 m/s²). Work is defined as force times distance, so lifting an object of mass m a height h results in an energy (work) investment of mgh. This is called gravitational potential energy.

It is called a potential energy because it is possible to put the invested energy on a shelf—literally, in fact—to be accessed later. A dropped brick that had previously been given gravitational potential energy can do useful work, like driving a nail into a piece of wood (huge force times small distance = same work). The stored energy does not degrade one iota over time: in that sense it represents perfect long-term storage.

The idea for pumped hydro storage is that we can pump a mass of water up into a reservoir (shelf), and later retrieve this energy at will—barring evaporative loss. Pumps and turbines (often implemented as the same physical unit, actually) can be something like 90% efficient, so the round-trip storage comes at only modest cost.

Raccoon Mountain pumped storage concept.

The main problem with gravitational storage is that it is incredibly weak compared to chemical, compressed air, or flywheel techniques (see the post on home energy storage options). For example, to get the amount of energy stored in a single AA battery, we would have to lift 100 kg (220 lb) 10 m (33 ft) to match it. To match the energy contained in a gallon of gasoline, we would have to lift 13 tons of water (3500 gallons) one kilometer high (3,280 feet). It is clear that the energy density of gravitational storage is severely disadvantaged.

What we lack in energy density, we make up in volume. Lakes of water behind dams, for instance represent substantial storage.

Flow Power

When water is let out from the bottom of a dam, it carries energy as if it had been “shelved” at the surface of the lake behind the dam. How does water at the bottom “know” how high the lake surface is? Pressure—which is proportional to the weight of water overhead. So let’s take a cubic meter of water, at a mass of 1000 kg, and send it through the turbine. The mgh energy in the cube of water for a 100 m high dam is (1000 kg)(10 m/s²)(100 m) = 106 J, or one megajoule.

If this 100 m high dam only has one cubic meter per second flowing through, it would produce 1 MJ/sec, or 1 MW. I am ignoring the roughly 90% efficiency of hydroelectric turbines to keep numbers tidy and approximate. More typically, flow rates are measured in the 1000 m³/s range, so that our 100 m dam would produce 1 GW at this scale.

So the recipe is simple for understanding a hydroelectric dam: multiply the height of water behind the dam (in meters) by ten-thousand times the flow rate in cubic meters per second to get the power in Watts.

We Need How Much Storage?

The U.S. has a power diet of about 3×1012 W, or 3 TW. Two-thirds of this feeds heat engines (power plants, cars, etc.), at an average efficiency of 30%, delivering 0.6 TW of useful work in the bargain. The other 1 TW is direct heat (lots of this in industrial process heat), and electricity from nuclear and hydro sources. Imagining that we replace our heat engines with direct electricity and electrified transport, we need something like 2 TW of total power, accounting for some inefficiency. If you’re happier with half this, fine—a factor of two will not qualitatively change the giant scale of the problem.

The next question is: how long do we need our storage to last? In the Nation Sized Battery post, I argued that we need 7 days of storage for it to be invisible to the end-user. That is, if Americans insist on not changing any of their habits, and having zero storage-crunch outages on a decade timescale (read about the total shutdown of San Diego in a recent county-wide power outage), then 7 days is probably not far from the mark. I got flak for this choice, but I use it again here because A) it is not all that unreasonable, B) it allows side-by-side comparison to the national battery calculation, and C) you’ll see it does not make or break the case: even one day of storage is super-hard. Divide all my scale numbers by 7 if you wish that I had used one day of storage, for instance.

Note that 7 days of storage does not literally mean that we are prepared to experience 7 days with zero input from the renewable infrastructure. Operating at 30% of the break-even amount over a period of 10 days also leaves the system with a 7-day energy deficit, for instance. This circumstance is not too difficult to imagine: a cloudy winter week over the southwest while the wind speed over the country is half its average value (means eight times less power) over the same period.

So 2 TW for 7 days means 336 billion kWh of storage capability.

First-Blush Pumped Hydro

What scale would this amount of storage require if we did a pumped-hydro scheme? One immediate scale reference is to note that we have 78 GW of installed hydroelectric power in the U.S., amounting to 4% of the target 2 TW demand. Our traditional hydro capacity could not be scaled up by even a factor of two—since the premier river sites have been plucked already.

What about potential pumped hydro installations: not on current rivers, but in the mountains where we could wall off a high valley and fill it with water?

I say mountains because we need a significant height differential for pumped storage to make much sense. We won’t see pumped storage in the plains. The horizontal distance must also be minimized, so we need sharp relief—meaning mountains.

To first approximation, we can imagine mountains as lumps. They have pointy tops that point up. They are distinctly not very bowl-like. Upside-down bowls, maybe. They do, however, often produce hollows (“hollers” in some parts) ringed by arms/ridges of the mountain. Walling off the opening to the hollow allows us to fill this useless void with water. The pikas and marmots can just learn how to swim!  We also need another equal-volume body of water below, to catch the water in the storage cycle.

I can’t say that I’ve studied the topography of our lands to see how many places are amenable to these grand-scale engineering marvels.  I may be oblivious to the widespread existence of natural bowls perched on the edges of cliffs.  Whatever the case, the 22 GW of pumped storage we do have at present presumably picked the primo spots.  Instead of fussing over topographical maps, I am using the simple “hollow” model informed by my time in the mountains and staring at relief maps.

In any case, let’s not allow these details to prevent us from doing some math! Let’s say our average candidate hollow allows a 500 m high wall (1650 ft) on one end, and another few-hundred-meter wall lower down for the lower reservoir (the hollow is wider here—maybe even a vale by now—so the same volume is accommodated by less depth and more area).

Simple model for filling a hollow with water to height, h.

My model for the hollow will have a V-shaped profile, with sides at a 20% slope and the hollow floor running up at a 10% slope. Thus the 500-m-high dam wall is 5 km across at the top, and the lake extends 5 km back in a triangle. This geometry produces a reservoir 2 cubic kilometers in volume. Considering the tapering shape, the stored gravitational potential energy is 2 billion kWh. We just need to build 170 of these things. Never-mind the fact that we have never built a wall of such proportions. Or the fact that the largest pumped storage facility to date stores 0.034 billion kWh—60 times less capacity.

But let’s continue to play the game: If we indeed demanded 2 TW of power from about 170 pumped-hydro stations, we’re talking 12 GW of production capability each. This is significantly larger than the biggest hydroelectric installation in the U.S. (Grand Coulee, at 6.8 GW). Times 170.

Perhaps I was too ambitious in starting with a 500 m dam height. A greater number of smaller reservoirs would allow more sensible power stations and perhaps avoid turning the seven wonders of the world into the 177 wonders of the world (with lots of redundancy).

The energy stored in the walled-off-hollow scales like the reservoir height to the fourth power! So if we drop to 250 m height (still impressive to me, being taller than Hoover Dam), we need 16 times as many installations (over 2,500), each with 600 MW capacity. For scale, we currently have 24 hydroelectric installations in the U.S. rated at > 600 MW capacity.

Hoover Dam: 221 m high; 2.0 GW power; 2.5 million cubic meters of concrete.

I think at this point, you can see why quibbling about the need for 1 TW vs. 2 TW or requiring 2 days of storage vs. 7 days is not going break the logjam of a hard problem. Even accomplishing 1% of the requirement I have laid out would be super-impressive.

All That Concrete!

These dam walls will require a lot of concrete. A survey of dam construction suggests that the base thickness is approximately 65–90% the height of the dam. Picking 75% and tapering to a cusp, our foregoing geometry requires a concrete volume 25% larger than h³, where h is the dam height. For our 250 m set of dams, we need 19 million cubic meters of concrete apiece. Each dam then contains as much concrete as exists in the Three Gorges and Grand Coulee dams combined! And this is the “small” version of our dams. And we need over 2,500 of them. I’m just sayin’.

At an energy cost of 2.5 GJ per ton of concrete, and a density of 2.4 tons per cubic meter, we end up needing 32 billion kWh of energy per dam, and 90 trillion kWh total. This over 250 times the amount of energy impounded by the dams, and represents three years of the total energy appetite of the U.S. today.

Note that I’m totally ignoring requirements for the lower reservoir.

Ample Room for Water Skiing

I’m keen now to understand what this looks like relative to our landscape. How much area will all these lakes take?

In the 500 m dam-height model, the area of the upper reservoir is 12.5 square kilometers. Times 170 reservoirs is 2125 square kilometers. In the 250 m model, we have 3 square kilometers per reservoir, or 8500 km² for the whole set. So the total necessary area scales like the inverse square of the characteristic dam height.

We also need to add the area for the lower reservoir. Since the terrain is likely less sloped lower down, let’s assume that the lower reservoir surface area is twice as big as the upper reservoir, so now we have about 25,000 km² in new lake area (both reservoirs are not full at once, but this land is no place to build a mall).

We get an area equivalent to 160 km on a side. This is the same area as Lake Erie (and more than its volume). Add another Great Lake’s worth of space to the map. No trivial affair. I haven’t asked yet where we get the water for this endeavor. Good thing water shortages are of no concern on this planet.

It is worth also comparing to the area of a photovoltaic system providing the 2 TW of average power. Such performance would require 10 TW of installed capacity (accounting for day/night, sun angle, weather). At 15% efficiency and 1 kW/m² of incident peak solar energy, we need about 65,000 square kilometers of panel—roughly comparable scales. Keep in mind that the water area is based on over 2,500 gigantic 250 m dams, each taller than Hoover Dam, and containing 8 times as much concrete. For smaller, more realistic projects, the area of water could easily exceed the solar panel area. Converting land to pumped storage carries far greater environmental impact than converting to a solar farm, so that storage concerns dominate. Wind takes substantially more land (about 50 times) than solar, so the pumped storage lakes would not rival the area dedicated to wind farms.

Variations and Scalings

We’ve relied on loads of assumptions in our exploration of the potential for pumped storage. It is easy to lose track of the choices and the impacts they have. Is the 20% slope on the sides important? How do things scale with the dam height?

In a general analysis, it works out that the number of dams needed is proportional to the total energy storage required times the side-slope of the hollow (in %, e.g.) times the slope of the hollow floor divided by the height of the dam to the fourth power. But interestingly, the total volume (and therefore energy) required for concrete only depends on the hollow floor slope divided by the height of the dam.

The result is that one 500 m dam replaces 16 250 m dams, while taking only half the total amount of concrete. Scaling therefore favors the big projects over the dinky. Of course the number of acceptable sites for the mega-projects may be too slim, while the pressure to find 16 times as many lesser sites is no walk in the park.

Total lake area scales as the inverse of side slope and the inverse square of the dam height. So, naturally, broader shallower lakes will be more evident from space. Total water volume needed just follows the inverse height of the dams.

Of course any real implementation would have a wide variety of dam heights in the set. I treat them all as the same to establish baseline numbers. Strict averages do not work due to scalings that are not linear, but this at least gives us an idea. An analysis where I allowed a distribution of dam heights would just waste my time and yours.

A common trick is to build a large feed-tube from the bottom of the upper dam to a turbine/pump located far below. This will not be easy to accomplish everywhere, but an additional 500 m drop improves the 250 m dam by a factor of 3.6, and a 500 m dam by a factor of 2.3. Doing this reduces the number of such projects needed by a similar factor (still large numbers). But don’t get too excited by this option: we still need a place to put the lower reservoir. If you give up too much height, you run out of natural walls and vertical relief, demanding a very large flooded area to catch the water.

Comparison to Real Examples

Ludington pumped storage: 110 meters; 1.87 GW; 15 hours; 27 million kWh.

Enough fooling around. Let’s compare this fantasyland to something real. We have 22 GW worth of pumped storage in the U.S., which is about 1% of my 2 TW goal. But they tend to be sprinters rather than marathon runners (typically about 12 hour run-time at capacity), so the actual storage falls short of what we need by a factor of 1500 or so. Think we only need one day of storage? Still a factor of 200 off.

The largest pumped hydro installation in the U.S. (in terms of energy, not power) is at Raccoon Mountain, in Tennessee. I owe much of my air-conditioned comfort as a kid to this facility. Sitting atop a mountain, the reservoir unloads to the Tennessee River 300 m below (technically Nickajack reservoir). The installed capacity is 1.532 GW, implying a flow rate of 575 m³/s. The upper reservoir provides an unusually long 22 hours of service, so that the volume of useful water is 45×106 m³, and the energy storage is 34 million kWh. The surface area of the lake is 2.16 square kilometers, resulting in an average depth of 21 m. The (earthen) dam is 70 m high and 1800 m long, from which I calculate a dam volume of about 106 m³—about half that of Hoover Dam.

Raccoon Mountain: 302 m; 1.53 GW; 22 hours; 34 million kWh.

What can these real numbers tell me about my simplified geometry and the guesses that went in? The main difference is that the Raccoon Mountain geometry has much gentler slopes: something like 3–5% up the “hollow,” and about 8% up the sides. We would need 10,000 Raccoon Mountains to meet my baseline energy capacity—although we could scale back on power per unit. This becomes 50,000 if you can’t use the trick of dumping to a reservoir far below. For 10,000 replicas of Raccoon Mountain, the total lake area (including the area of the lake below) is about three times the size of Lake Erie (Lake Superior-size). The dam volume is about one-fifth what we had before, becoming comparable to the extent that the deep-drop trick is not employed. The total volume of water sequestered is comparable for the two cases (because this is just mgh, and our baseline had h = 250 m, while Raccoon Mountain uses h = 300 m).

Re-purposing the Hydroelectric Infrastructure

If at any point in this development you thought, “wait a minute: why build all these giant dams in the mountains when we’ve got large lakes and dams already, with water already delivered to the doorstep?!” then you are not alone: I wondered the same thing.

The first note is that our installed hydroelectric capacity in the U.S. is 78 GW; a factor of 25 short of the necessary full-scale capacity.

The next note is that water flow is not always available to realize the capacity of the installed power. For instance, the U.S. hydroelectric plants produce about 270 billion kWh each year, which is only 40% what would be delivered if all dams ran at 100% capacity all year round. For example, Hoover Dam has an annual production of 4.2 billion kWh, which is 23% of what the 2.08 GW installed capacity could churn out in a year. Even the mighty Columbia fluctuates enough that the Grand Coulee dam only realizes 35% of its capacity.

These points are relevant because in order to achieve the necessary 2 TW power output, we need to multiply the hydroelectric capacity flow by a factor of 25, or a factor of 60 greater than the average flow. We might predict a few erosion problems here and there.

Let’s Do It Anyway!

Let’s not be wimps. Let’s just beef up our hydroelectric capacity at the developed sites and ask whether we have enough energy storage behind the dams. One way to look at this is to figure out how much power would be generated if all lakes impounded behind hydroelectric plants dropped by one meter over a 24 hour period. Computing this for each dam based on each lake’s surface area yields a total of 170 GW of power. We need more than this. Our demand for electricity alone in this country averages 450 GW, and of course we’re shooting for about four times this to cover all our energy demands.

The upshot is that getting sufficient energy out of the current infrastructure would require draining each reservoir by a little more than 10 meters per day. But as the lakes drain, the surface area shrinks, so that my ten meter estimate is too low. Additionally, many dams will tap out once we get beyond the 10 meter range, and the fact that the energy delivered drops as the height of water drops reduces the capacity further. Using the volume reported behind each dam, I find that draining all reservoirs over a 7-day period delivers a power of 500 GW. Of course dams are often serial along a river, so we get to re-use water along the way. This will give us a factor of several, and put us close to our need.

But let’s not forget that our scheme here involves emptying all the lakes and rivers of water, and at a rate far in excess of what the channels are accustomed to carrying. It’s an extreme maneuver.

Drain the Great Lakes

While we’re having “fun,” let’s see what we could get out of the Great Lakes. The upper four lakes are all at essentially the same elevation (6 meter drop from Superior to Erie), while there is a 99 m drop between Erie and Ontario. We call this Niagra Falls, although only half the drop is developed across the falls proper.

If we drained one meter from every upper lake, we would get 54 billion kWh of energy: about a sixth of the target capacity. If performed over seven days, the flow would be 375,000 cubic meters per second, or 125 times the normal flow over the falls. Now I’d pay to see that! But I would first want to visit every town along the St. Lawrence River one last time.

If we tried to trap the water in Lake Ontario so-as to spare those downstream of the wrath, its level would rise 12 meters (39 feet). Watch out Toronto & Rochester!

The pipe delivering this water to the turbines would have to be over 125 meters in diameter (or 160 tubes each 10 m in diameter) to limit the velocity of the water through the pipes/turbines to below freeway speeds! What fun.

Am I Insane?

Why do I always do this: pick a challenge and show how ridiculous it is to solve the problem by a monolithic approach? Maybe I’m the one being ridiculous!

This tendency is a reflection of my quest to understand how we might face the tremendous energy challenges ahead. The first step is always to assess the potential of a solution relative to the full-scale demand. If it wipes the floor with an excess capacity, then great: it is inarguably a no-brainer go-to solution. If it comes up short, that’s very informative too.

Yes, a diverse portfolio of a half-dozen inadequate solutions may be able to add to an adequate solution. But a half-dozen woefully inadequate solutions cannot pull off the same stunt. So far, my quest keeps turning up the woefully inadequate type. The scale of fossil fuel replacement is so daunting that we very quickly get into trouble when putting numbers to proposed solutions.

A common reaction to the Nation Sized Battery post—especially on the Oil Drum Forum—was that I was being silly by considering a full-scale lead-acid battery, and that pumped storage was such a more obvious solution to the problem. It was not obvious to me, but I had not yet done the math. The fact that just one of the “small” dams considered here has as much concrete as the Three Gorges and Grand Coulee dams combined is humbling. I would be impressed if we made one. I would be astounded if we made 25. And this just gets us to 1% of our need (or 7% if you still bristle at a 7-day battery).

It is clear enough that pumped storage exists and works quite well in certain locations. But demonstration does not imply scalability, and scaling the existing installations did not deliver a radically different answer (in fact, demanding more installations). The enormous scale I calculate means simple factors of two or even ten here and there do not change the overall flavor of the conclusion.

Let’s be clear that I am not making any claim that large scale storage at the level we need is impossible. But it’s far more daunting than almost anyone realizes. It’s not a matter of “just” building up when the time comes. We could easily find ourselves ill-prepared and suffering insufficient energy supplies, intermittency, and a long, slow economic slide because we collectively did not anticipate the scale of the challenges ahead.

Acknowledgment: Thomas Tu contributed research on hydroelectric installations, consolidating capacity, height, and capacity factors for dams, along with surface areas and volumes of impounded lakes.

117 thoughts on “Pump Up the Storage

  1. There is a proposal on the table in Germany to lift a large piece of stone instead of water. The claim:

    “The key advantage of this approach lies in the extraordinarily large amount of energy that could be stored and the relatively small investments compared to a similar hydro storage.”
    Like your article, they tak about storing days of the nation’s electricity demand. It would also need less space than water pumped storage.


    They plan to get funding to build a pilot.

    • If that worked, would we re-enter the stone age? This is so crazy I don’t know what to say. Density gains you only a factor of three or so. And stones don’t flow very well, so its hard to get hundreds of meters of height like you can with water. I didn’t think they smoked much marijuana in Germany…

      • The specific plan (click on the link) is to use an underground water resevoir to lift a vast stone piston. The energy is in the potential energy of the stone, which provides hydraulic pressure on the water. You get a multiplier from the density, as you point out, and also might find it a lot easier to find suitable sites.

        • Stone just isn’t that big a help and as tmurphy rightly pointed out, it’s a solid whose size and shape is generally fixed. Water can transport itself great distances vertically or horizontally and it’s quite malleable. Self-malleable.

          It also occured to me that to have a solid – solidity aside – give you an order of magnitude density improvement or more, your choices for starting points are e.g. silver, lead, and thorium. If that’s a little too rich, toxic, and/or radioactive for your blood, you can trade down all the way to iron at 7.9g/cm^3, but you’d have to scale up the height or the amount to compensate.

          I’m concerned about what it will take to implement the storage it would take to even out the generation from wind, solar, and other intermittents, even if the storage is co-located with those facilites. Economies of scale suggest that you wouldn’t want to.

          • This “piston” idea also completely omits the problem to produce a high pressure resistant hydraulic-grade seal around the stone slab. Stone doesn’t float, so if the piston doesn’t seal perfectly, you end up pushing a ring of water upwards. The paper only states “the piston is sealed”, but nothing about how they want to achieve that in such a large scale.

  2. Once again, a great analysis! I was reasonably convinced already, by my own even rougher numbers that this was a no-go, but your explaination bought it home very vividly.

    I guess the next topic should/could be doing the math on large scale compressed air/gas storage. Hey!, maybe we could use all that sequested CO2 and kill two birds with one stone! (actually, could we? hmmm…)


    • See my response to J Anthony below, which addresses this concept. Glad you got the 70,000 number too. David MacKay is a hero to me. His book is one of the few links in my blogroll.

      • There are a few good micro hydro electric generators that work with low head. Might it not be better to generate when needed rather than store. Barring drought this might make more sense.

  3. I guess i question the total size, requirements based on your 500m assumption. if u assume something like 1500m or so u drastically alter the issue As the height scale is X^4 that minimizes our requirement by a factor of 81.

    Heck if you used something like Flagstaf, AZ =>Phoenix, AZ thats about a 1800m.

    Correct me if Im wrong, but via your design of 500m u said that there was a 340 cubic km of water requirement. (2 cubic km per each of the 170), divided by 81 gives us ~4.2 cubic km of water. if u assume we build this as a dug lake height is 1km in depth your talking about a space of only 2 by 2 km.

    If u build the plant at the bottom edge of this new “lake” it seems “reasonable” we could dig this out. now I’m ignoring what u do with the waste water/inflow water, but it seems reasonable that its not a always needed and u COULD just create an outflow river and pump in sea water from the gulf of California. I’m not saying that’s an efficient measure electricity wise to build up the energy, but as far as actual storage it seems like a doable thing.

    • You can’t scale volume the same way: the factor of 81 would apply to the number of dams, but the volume would just be reduced by a factor of three.

      I shied away from the 500 m mega project, you ran the other way!

      For 1500 m projects, you would need two; the volume of water would be 55 cubic km each (bigger than any reservoir in the U.S.), and the concrete needed for each dam would be 4 cubic km (about 400 times as much as in Three Gorges). Now I know why I ran toward smaller projects: I have a tendency to follow the slope from fantasy to practicality.

  4. Great post. Very educational applied physics. Pity our politicians can’t do math

    Would you be interested in doing the numbers for solar powered satellites?


    Obvious problems are the sun not shining at night, collecting and converting the energy in space using mirrors or solar panels, microwave atmospheric absorption and losses from the rectenna collection.


    • Uh, a main point of SPS is getting around “sun doesn’t shine at night”, because with the right orbit there is no night. Also no weather.

      While I’m commenting:
      “Wind takes substantially more land (about 50 times) than solar, so the pumped storage lakes would not rival the area dedicated to wind farms.”

      Though OTOH wind is far lower impact on the land it uses than anything else; you can still use the land as farmland or preserve. It’s not shading the land like solar or drowning it like water. And some of the ‘land’ is off-shore ocean. (And how much land is used presumably depends on the wind speeds.)

      • Wind does kill birds, a lot of birds, this is by count under the rotors at every location where wind turbines have been installed. This makes it unsuitable for many locations and shows that it is not risk free or low impact.

        • David MacKay makes the point that domestic cats take a far, far, far greater toll. Now if windmills were also cute!

    • I think the fantastic cost rules anything like this out. The advantages in being in space are energy generation factors of perhaps 2-5 at most I guess. But the cost of getting it into space is many many orders of magnitude greater (and you’ve got to maintain it). It just isn’t going to happen. Someone else can do the math if they like.

      • Well put. Anything involving a lot of mass in space is going to kill the economics. And getting the energy down to Earth is anything but easy—despite what enamored proponents might say.

        But solar from space isn’t the subject of this post, so I will not continue this thread—though I will add that Chris Nelder offered his link on the topic.

        • Space based energy systems will never work out due to the energy cost of lofting them into orbit, it does not scale. I believe that it took 10% of the production of the US for 8 years to put a few men on the moon. The rocket that did that was the most powerful energy converter that mankind has ever produced and used out side of the bomb.

  5. Tom,

    Brilliant post. I once calculated that I’d need a chunk of concrete the size of my two-car detached garage, lifted to a height of 130 feet to supply reliable power to my house from the wind and solar resources available. Imagine a neighborhood full of those! (however, you could use the same tower for the turbine and to hoist the counterweight…)

    [edit: counterweight just means weight]

  6. Energy storage is of course needed because the most promising alternative energy sources aren’t on all the time, and don’t have their peak production in phase with peak demand. Electricity generation is of course the most useful goal of energy storage, but I wonder how much demand could be shifted to other storage. The specific example I have in mind is for air conditioning and refrigeration: could we make ice during peak generation and melt it during peak need?

  7. The Dutch were considering a scheme (can’t find a good link) where they built pumped storage at sea. Basically you encircle a huge area of sea and then pump the water out when you have excess electricity and then let it back in again when you need power. If you were lucky it also became a source of tidal energy. Less planning issues (perhaps, what about international treaties?)than land based pumped storage and your volume scales well with wall circumference, but a massive initial investment. Still they thought they had the experience from building dykes.
    I’d be interested to see some number on that idea…

      • actually, you could “just” anchor a huge ballon to the bottom of the sea, and compress air into it – against the pressure of the water, so it would even be gravity storage in a way!
        Having a soft, submerged container would lessen some of the problems with waves shaking your structure apart.

        Problem is, compressing air is less efficient than pumping water.

        • When you compress air, it heats up. In an underwater balloon, it will cool off. That represents a sizable loss of energy.

          • This is a generic issue with compressed air, and whether underwater or not, the compressed air will heat to above ambient and then have to cool back down to ambient. Sure, it’s a little worse if the storage is a bit colder then the ambient air, but this is not a huge loss for the sorts of temperatures we’re talking.

            Also, in any compressed solution, charging slowly enough prevents heat-up-loss from being a serious loss.

    • I heard about that. I think they concluded that the costs would be too high, but I’m not sure. They called the idea “energy island”.

  8. Tom,

    Here’s another “magic bullet” I keep hearing about — CAES.

    Compressed Air Energy Storage has some problems for sure – heat buildup being one of them. But both heat and pressure can be used for energy. In an earlier post, you briefly talked about household level compressed air storage, and it had some advantages. I’m sure I’m not the only one who’d be interested in seeing your thoughts air pressure being used for grid scale storage.

    In the scenarios I’ve seen, a man-made underground cavern is used to store the air pressure, plus surplus heat in a pebble bed (in some designs, temps go over 1000 degrees F.) The air pressure can be used to drive a turbine or a modified steam engine (as an example), This will sound counterintuitive, but the heat has to be used to keep the turbine/engine from freezing up as the air expands. Some existing CAES systems use natural gas to heat the expanding air.

    I hear that the hard part of building the cavern is just drilling through the cap rock. After that, water jets can hollow-out a large volume cheaply and quickly. But I’m no geologist.

    What’s your opinion. Is CAES just more noise or something to start investing in?

  9. I enjoyed your article on Peak Oil and pumped hydro storage. If the US is to survive the looming end of cheap energy then we need to localize as much as possible. The most local you can get is your own home.

    First – Build super-insulated, tight homes that use much less energy.

    Second – Install solar, wind, or whatever is next at the home site to supply its energy needs and then some.

    Third – Store excess supply. Here is where I think pumped hydro may work in many areas. I have a 300 foot elevation difference on my rural property. If I have a 1 Kw excess capacity in my solar array (and five hours of sunlight) I can use 5 KwH to pump water to a tank at the top. When I need that potential energy I can let the water drive a micro-hydro generator at the bottom. If it’s only 80% efficient I’ll get back 4KwH at night. Owners without a handy water head could use a grid-tied system (or compressed air or something else) to store their excess.

    By having homeowners energy self-sufficient, brownouts and blackouts become an historical footnote. Local homeowner energy production is clean and non-polluting. Large enough systems could power personal electric transportation. Add local farmer’s markets and industries and you’ve just about made the transition to a post-oil world.


    • I agree mostly although instead of seeking energy-sufficiency on the level of individual homes, it makes more sense at the community level. An energy-efficient community with a small CHP (combined heat and power) plant, fueled with sustainable biomass, and an array of wind turbines and perhaps solar power installations. You don’t need to achieve 100% sufficiency, the important point is that load on the the far-distance grid can be reduced substantially, perhaps by an order of magnitude (don’t forget that this starts with efficiency and conservation). The remaining problems will be a lot easier to solve.

      Installing more hydro reservoirs is emphatically not the answer. The environmental and social cost of large scale dams is steep. There are other small-scale solutions to the storage problem. Some cities store electricity from night wind in the form of ice underground and use it to cool buildings during the day. There are also some approaches to storing solar heat (see http://www.slideshare.net/amenning/solar-energy-9778330). A single comprehensive solution, such as explored in the blog post, is simply unrealistic but that is really no reason for despair. A sustainable energy infrastructure will have to be diverse, scalable and decentralized.

      Hydrogen of course has some promise as a storage medium. Maybe that will be the topic of the next post?

  10. Excellent article as always, and a little depressing. Need to brush up on my camping/backpacking/survival skills.

    If in fact we need “65,000 square kilometers of PV panel” has anyone (winking at you) ever measured the area of southern exposure roof tops on home across America? PV Panels are nice, but I was looking at something the other day or month that was a PV shingle. Hence you would cover the entire roof with this PV shingle (like other shingles but with pv properties). I believe it was CIGS like and had lower efficiency % but higher area coverage and cheaper price (maybe).

    Of course this would mean that all homes would have them and then be tied onto the grid with sell back and such. Hard in the cities where whole buildings are in the shade most of the time. We will still need all the hydro we got and some PV farms in the desert and wind farms in the midwest.

    • That was the promise of thin film PV. It has not worked out so far but it may be possible with three layer films.

  11. Good analysis, proving the grid storage problem has not been solved.

    Checkout GravityPower.net they are boring holes for a pumped storage solution with a smaller footprint.

    • I throw in some analysis of that, as well as a second angle on our host’s problem, here.

      It’s a clever idea, but scaling up to the 7 days our host wants would take $5 trillion just for the material cost of pure concrete pistons, never mind manufacturing or drilling costs (for 1-2 kilometer shafts), or the iron — 20x more expensive — in their designs. Really, we’re talking tens of trillions.

      BTW, the water in their system is just a hydraulic, the real storage element is a 200 meter tall piston or iron and concrete.

      My idea is to ignore mountains and simply build artificial Great Lakes propped up by rammed earth. I’m not sure of the cost, but I have lower bounds on the scale needed, about 100-1000x the mass of the coal we burn every year, never mind the engineering of 100m tall (or 1000 m tall) enclosures.

      • An interesting fact is that earth weighs around 2,000,000 pounds per acre per foot. So it takes a lot of energy to move it around and ram it, more if you have to dug it out prior to the moving and ramming.

  12. I wonder what would be the energy cost of draining rocks instead of building dams. If we can’t build the walls for our reservoir, let’s make a big hole in the ground ;-) Note that I’ve no idea how deep we could go before it would get too hot.

  13. Also, I wonder if we really DO need so much storage. If solution could be global, we wouldn’t have to care about PVs not working during night hours — the other half of this planet will be in the sun at the same time we sleep. How much power do we lose when transmitting it to another hemisphere and could something be done here to improve efficiency? Or are we doomed :-(

    • gshegosh,

      A global energy grid has been proposed – google for Project Genesis. This article by Stuart Staniford has a quick summary of the idea, mentioning losses, costs and so on:-


      It’s a long piece, so just search within the page for “Project Genesis” if you don’t want to read the rest.
      My take-away: a global grid is technically and financially feasible (just).

  14. Nuclear power.

    End of story.

    We already generate something close to 50% of our power with nuclear in some areas of the country. Scaling up by doubling or tripling the number of reactors is quite doable in a engineering sense. The waste problem can be drastically reduced if we resumed reprocessing. Left over waste can be sunk into mile deep bore holes or dropped into the marianas trench. In any event, storing the waste has to be easier than building enough hydro storage and solar capacity to cover half the country.

    Wind and solar are great. I am having 4.3 Kw installed on my roof. But to seriously run an energy economy the scale of ours without fossil fuels, there is no option other than nuclear. Each post I have read here on different schemes for trying to make solar or wind work just reinforces this fact.

    • Such certainty. Nuclear is a good stable baseline, but can’t follow demand fluctuations and does not represent a storage solution for the renewables that we will no doubt continue to incorporate. Nuclear can offset some of the storage need (factor of several?)—and that’s useful. But nuclear isn’t all that.

      • Nuclear is not normally used for to load follow, but that is largely an economic decision. It is cheaper to run them 24 hours a day. French do ramp the power of their NPPs up and down and that is the reason why their capacity factors are lower than in US, for example. (They have to since most of their electricity is from NPPs.) There are some technical issues involved, but some reactor designs have less of a problem with those than others. In principle, there is also no reason why you couldn’t use the power NPPs produce at night to produce some useful liquid fuels for example.

    • Fissionable material is also a finite non-renewable resource. Nuclear power plants are already consuming more uranium/year then what can be mined. http://en.wikipedia.org/wiki/Uranium_depletion
      It has the same problem as fossil fuels in that we will run out of it eventually. Leaving a waste product far more scarier than CO2…

      Although some do claim there is enough uranium to be found in the ocean to outlast the lifetime of the sun, but no one is extracting uranium from ocean water at this point, so I guess it’s all science-fiction for now.

    • I was flooded with comments about nuclear, which would send the discussion about pumped hydro way off track. I’ll have to dedicated a post to nuclear sometime soon.

    • Nuclear as in thorium reactors as proposed by Hansen and others is a good source for energy, a lot of energy, safely. They can run on our current nuclear wastes plus thorium which is rather plentiful and their wastes products do not last anywhere as long. Plus twin byproducts of them are that they can produce Rare Earth elements as bi products and do not produce bomb elements. To bad reactors have such a bad name as these are very different from what we are using at the moment. True as in all reactors they provide base load power very well but not peak load easily. Their cores do sour the same as the reactors we currently use so peak power is difficult but for a system it is not an issue that can not be managed.

  15. Regardless of storage form, electrochemical or pumped hydro, if the “national battery” is that which can be quickly charged and discharged (depleted and refilled) then perhaps it should be sized for, say, one day at 2 TW, since as Dr Murphy states the battery/hydro source is really a “sprinter”. Then, for the rare or even freakish outages requiring backup for the additional six days, fall back to the less efficient hydrogen cycle using electrolysis and fuel cell (or gas turbine or other combustion engine), or even fossil fuels. In these rare events the low efficiency or emissions from H2/CNG matter little. In this case the split for the “battery” is 48 billion kWh, and 288 billion kWh sourced by H2/CNG. Using 40% efficient turbines would require 256 billion liters of liquid H2 (10 MJ/liter) standing by, less for fuel cells. By comparison the US Strategic Petroleum Reserve holds 115 billion liters.

    NREL seems to have been exploring hydrogen back up scheme for solar and wind for some time.
    Some do-it-yourselfers have also gone with a solar powered battery-hydrogen hybrid storage backup scheme, using the H2 to run them through the winter completely off the grid *and* power an H2 converted car.

    • It is the physical nature of managing hydrogen that stops any system in its tracks that attempts to use it. The efficiency or the amount of use does not matter, Hydrogen’s physical properties make it far to difficult to handle in any scale that we are talking about. It is a non starter in any application that does not require a high graveametric energy density. This limits it to rockets and related uses.

  16. There is a firm in San Francisco working on pulling buoys to the bottom of the bay to store power. I suspect that even if their approach works, you couldn’t scale it up due to the cost of materials, though you do get a little boost from tides. However, this approach could be used for storage at some island locations that couldn’t be part of a larger grid and don’t have the land area for pumped water storage.

  17. The point with the rock lifting concept is: The capacity grows with the 4th-power, the cost with the 2nd-power of the radius. Resulting in a price proportional to 1/r², this means, you can build infinit cheap storage as low as 1$/kWh!

    • Careful not to get carried away with talk of infinite cheap storage. Scaling arguments can be taken only so far, even when they are accurate.

  18. While I’m not saying the concept of a “nation-sized battery” can be transposed anywhere else, there is already one such “device” working pretty well as we speak. Hydro-Québec’s 26 large reservoirs can store up to 170 TWh of energy (one year’s worth of electricity). Volumes of water and land required are massive (figures for the 5 largest reservoirs: http://www.hydroquebec.com/learning/hydroelectricite/gestion-eau.html ) but the system it could accommodate more must-run renewables if needed on top of the 37 GW hydro capacity already deployed (~1 GW deployed + 2.5 GW scheduled for completion before 2015).

  19. I followed up on the hydraulically-supported falling weight storage system, and found what seems like a more practical version than the German rock: Gravity Power Systems LLC:
    with 51 min video:
    proposes hydraulically supported falling weights in 10m diameter vertical shafts 2000m deep, to provide 150 MW for 4h storage time; groups of 8 shafts would give 1.2 GW for 4h on 3 acres of real estate. Building time of 3y (per 150 MW unit?) is estimated.
    The company is just prototyping now, proposing commercial systems in 2014. Hoped-for cost $1/watt ($1000/kW) for 4h storage (all this is on the video linked above). Advantages over pumped storage are claimed, a primary one being real estate.
    How would this compare with the Tom Murphy analysis?

    • Is it a cute idea? Sure. Can it scale? Let’s see. Based on the information at the provided link the larger system (10 meter diameter tubes) provide 600 MWh per tube (I confirm this is the ballpark expected). To compare to the pumped storage evaluated here, we need 560,000 tubes, 2 km deep, 10 m diameter. This is the same amount of tubular excavation as a subway tunnel that would criss-cross the U.S. about 250 times, or go three times the distance to the Moon. Wowie.

      We need a water volume of 88 cubic km, which is 8 times smaller than our 250 meter dams require, or 5 times smaller than the 10,000 Raccoon Mountain installations. The land area is small-ish, at 700 square kilometers.

      Each weight is 210,000 tons, which I assume to be half concrete and half iron ore by mass. The weights then need 25 billion cubic meters of concrete (half what we needed for our 250 m dams, twice what we need for 10,000 Raccoon Mountains). But they also need 60 billion tons of iron ore. This amount of iron ore (25% is iron) would build 250,000 steel-heavy Empire State buildings. The U.S. estimated resources (12% of world total) is only twice this in ore, and half this in actual iron content. The proven reserve base is only one-fourth of what is needed.

      I like the idea, actually, but does anyone still think this is serious, or another thing we won’t see scale up? Again, accomplishing even 1% of the task would be like the most amazing thing ever: like a two-tunnel subway tube from NYC to LA.

      For cost,the 8-tube set would seem to cost a bit over $1 billion, if we can trust the estimate. This works out to $250 per kWh, which is about twice the cost of lead acid. There are problems scaling lead acid too, but it’s presently cheaper.

      • Tom – salient analysis, as (I’m afraid) we’ve all come to expect. I work in the power plant development industry (solar and wind, 20-150 MW per location). There’s a force your analyses avoid: the permitting authority. For those not acquainted, building power plants requires governmental permission, which is a long, and expensive path for even the most benign of energy projects. I can’t wait until this group tells a county permitting authority that they want to drill 6000 ft holes in the ground and fill them with water. Forget construction, getting a permit for the first project will require years of environmental, cultural and geologic studies in California!

      • I still think it’s an interesting alternative for places where pumped storage isn’t available due to lack of mountains.

        I also think that you are not being quite fair to the hydro storage technologies: It is fairly obvious even without the detailed math, none of them will be a magic bullet against a nation-wide 7-day renewable outage. Pumped hydro (and by extension this gravity power module thing) make a lot of sense for short-term storage with fast response time though. So if we’re talking about 4hours of storage instead of a week, this can actually make a difference. And that is exactly what the GPM people seem to be targeting. I think their main advantage is actually ease and speed of construction: You figure it out once and then you just build a bunch of them. Pumped hydro on the other hand is always individually planned large-scale dams and tunnels and as such, takes much more time and resources to build.

        As far as I can tell at this point, long-term storage will have to be chemical, i.e. making hydrogen and then burning it, or something of the kind. At the price of much lower efficiency but with less required land/space/capital.

        [parting comment redacted]

  20. How about this idea from Isentropic Ltd.

    Use a heat pump to transfer heat between two insulated tanks of gravel. Then use the heat pump in reverse as a heat engine to recover the heat and generate electricity. Cooling 15 tonnes of granite gravel down to -150°C and heating another 15 tonnes up to 500°C is sufficient to store 1MWh of electricity. Round trip efficiency of 72% to 80%. And before anyone suggests it, no, it does not break the Carnot limit. As a heat pump, it transfers more heat energy than it uses electicity to do so by a factor given by the coefficient of performance. In an ideal fully reversible system this factor is exactly what is required for the heat engine to regenerate all the input electricity operating at ideal thermal efficiency. This indeed is what reverability means.

    Isentropic believe they can come fairly close to reversibility with a reciprocating compressor/expander using a Brayton cycle.

    There is plenty of gravel in the world, it occupies a fraction of the space of a pumped storage system and it can be put anywhere.

  21. [shortened by moderator; full comment visible here]

    So much for Wind and PV.

    It’s time to break that mould of assumption that PV & Wind are THE renewable options. They ain’t. A custom-made super-grid cuts, but doesn’t remove, the need for a weeks-worth of reserve power capacity, which pumped storage evidently cannot get near providing, even in the mountainous USA. (Consider Holland too).

    Geothermal, Biomass, Solar-thermal-salt and Ocean Wave offer power on demand. (Varies slowly with the latter option, but effectively flows without interruption). That diversity of power-on-demand sources directly avoids the scale of pumped storage that over-reliance on PV and Wind would demand.

    IF pumped storage could meet the reserve capacity needs of a 10% reliance on PV & Wind (big IF) that implies the need of 90% being met by the four scaleable baseload sustainable energies above (with minor contributions from Micro-hydro, etc). At present, due to the mal-priorities of enviros, entrepreneurs and officialdom alike, these baseload energies’ combined deployments/year are perhaps only 1/100th of the two over-hyped intermittents, when they would need to out-deploy those two options by at least nine to one.

    Thus to quantify the relative priorities in new energy supply it could be said that the next unit of baseload sustainable energy will be at least 900 times as important as the next unit of Wind or PV.

    • The simple reason for solar and wind emphasis is the math. These are the only options that come close to being able to reach full-scale. That gets people excited. And the technology is relatively cheap (wind competitive with fossil, solar getting there fast).

      I’ll soon evaluate a variety of other ideas. For now, I will refrain from launching an off-topic debate on the pumped storage page…

      • Tom – thanks for your response – and also for your many fine articles that have been both informative and a pleasure to read.

        Some queries regarding people getting excited over wind and PV having potential to supply the full power requirement, and these options’ nearing cost parity with fossil supply -

        I’d understood that both geothermal and solar-thermal-salt potentially share both those attributes, while your analysis has shown the requisite pumped storage reserve capacity for Wind and PV to be utterly unnaffordable, thus negating any prospect of their true competitiveness at scale. Even as part of a rational future portfolio strategy their competitiveness with sustainable baseload options has to be in question.

        I look forward to your evaluation of the other energy supply options, and would hope that, this being a critically global issue, you’d consider options’ merits including their applicability to the resources of simple developing societies that are currently increasing their fossil fuel dependency.



  22. My “solution” to the storage problem is to advocate changing the grid from being demand-based to being supply-based.

    Currently the grid is demand-based in the sense that users demand power from the grid, at any time and in any amount, and the generating infrastructure does whatever is necessary to meet that demand.

    With a supply-based grid the generating infrastructure, ideally including much solar and wind, supplies whatever it can and we modulate demand to meet supply. The obvious way to modulate demand is to vary the price. When it’s sunny and windy then power is cheap. When it’s dark and calm then power, which at that point has to come from hydro-electric, is expensive. How expensive? Expensive enough to bring down demand. This would require a lot of changes to mindset and metering, but mindset is going to have to change anyway and many places already have smart-meters installed.

    I can already hear people howling that this isn’t acceptable because, ah, um, because it’s different than the way things are now. Well bad new: there is no doubt that in the near future things are going to be very different than they are now; the only question is about the nature of the difference.

    We have a supply-based grid now because fossil-fuel powered generating station worked well with that model. As we move to generating stations powered by renewables then moving to a new grid model makes sense. And if you personally need power any time regardless of weather, there are always lead-acid batteries.

    Tom: if there is any math to be done on this, I’d love to see you do it.

    • That’s just an indirect way of dealing with the real problem, which is demand. If we’re going to let the power grid send that message, why not start talking directly about it and formulating shared plans for transition?

      The answer, of course, is that the overclass won’t permit it. Their wealth and power is premised on treating energy demand as a non-issue.

      • It is not a question of overclass, but the fact that a big chunk of our modern high industrial productivity comes from the 24/24 use of costly productive assets. If these assets are forced to function intermittently, the loss of wealth due to intermittency will be extremely significant, and it is not only the overclass that will suffer from it.

        • I don’t disagree about the widespread nature of the costs and sacrifices of backing down from cornucopian industrialism. My point is different: Why are we not permitted to discuss this issue?

          • I’m all for discussing this issue, but this is not the ideal forum (math/physics/estimation focus). If anyone knows of a place where the discussion focuses on sociological aspects of our energy future, fire off a comment and I’ll post (or possibly consolidate) links here.

  23. Hi Tom,
    Thanks for another excellent article.

    So BAU cannot continue. That is becoming a given for many. But what will continue?

    Eventually industry will take primacy again because it supplies the jobs. So how much pumped storage is needed to keep the industrial sector operating with a smooth power supply? There are many industrial processes that must have consistant power to be efficient. What do we need to keep those running?

  24. Tom, if you do the math on sensible heat storage you will find that three cubic kilometers of basaltic aggregate can store your 7 day energy requirement using 500 C delta T (includes conversion losses from thermal to electricity). This is relevant only to CSP, not PV or wind, but we do have the solar resource required. Cost of such an amount of storage…about $250 billion. That’s a small price to pay for base-load solar.

    Another technology where the storage already exists is OTEC, but temperatures are so low that conversion equipment costs a lot. Floating OTEC facilities in the tropics could generate the teraWatts required and ship the energy as ammonia to where ever needed. Figure $10/W of capacity. That’s a lot of money for one teraWatt, about 2.5 times what the federal government has borrowed in the last three years.

    • You should study the history of why ammonia was dropped from use as those reasons are still relevant even more so if one thinks about shipping the stuff in quantity around the world.

  25. What do you think of Chris Nelder’s post today on crowdsourcing to scale?


    “Second, the town develops local storage. Again, a whole range of technologies are available here, including residential- and commercial-sized battery arrays, pumped water systems, distributed flywheels, compressed air in underground caverns, molten salts, ammonia synthesis, and many others. . . The final step is to deploy switches that would allow the town to disconnect from the main grid when it goes down, and fall back on their own capacity.”

    • Tom and I have taken two completely different tacks on the question of storage in these pieces. I presented some plausible, real-world technologies that can be deployed in a distributed manner, without attempting to do the math on how it might all be done (which would be an extremely complex effort, best attempted by a team at the DoE, NREL or similar). Tom presented a thought experiment to determine if just one storage technology, pumped hydro, could support the entirety of US energy demand if supplied by renewables. That approach isn’t prescriptive, it’s just a useful way of estimating a given technology’s capacity within an order of magnitude.

      In reality, we would never use just one storage technology, just as we have never used, and never would use, just one form of energy generation.

      Ultimately what we have to figure out is what can be done, and where. The answers will vary from place to place.

      Step one would be to cut the problem down to size. Many studies suggest that efficiency gains (everything from insulation and windows, to appliances, to ground-loop geothermal heat pumps) could cut about 30% of our existing demand. When you transition the majority of 240 million cars and light trucks to rail, you can cut another big chunk.

      Step two is to deploy every kind of renewable power generation where it’s technically and economically feasible, as I outlined in that article.

      Step three is to figure out the storage mechanisms. Very modestly-sized battery banks (able to provide one or two days of backup) installed along with solar PV on 100 million homes and commercial buildings could cut the solution space down substantially. Grid interconnection (e.g., linking the Midwest wind farms to the Southwest solar PV capacity) can reduce the need for storage further. Small (or large) hydro storage in places with elevation supplies another wedge of solutions, but it certainly doesn’t need to do it all. CAES where salt caverns exist (like the ones we use for the SPR) can provide another municipal-sized wedge. Ammonia and hydrogen synthesis can be used almost anywhere. Adding thermal storage capacity to solar thermal plants can supply a substantially bigger slice of capacity than they do now. And so on. After all of that has been done, you have a very modest remaining need for on-demand peaker plant capacity powered by natural gas. And, as Germany has shown, the more renewable capacity you add, the more “baseload” capacity from coal and nuclear you actually need to remove. http://www.unendlich-viel-energie.de/uploads/media/35_Renews_Spezial_Renewable_Energies_and_Baseload_Power_Plants-1.pdf

      I am convinced that with a highly distributed, “all of the above” approach to both generation and storage, we can take some big steps in the right direction.

      • Chris -
        re your statement:
        “Tom presented a thought experiment to determine if just one storage technology, pumped hydro, could support the entirety of US energy demand if supplied by renewables. That approach isn’t prescriptive, it’s just a useful way of estimating a given technology’s capacity within an order of magnitude.
        In reality, we would never use just one storage technology, just as we have never used, and never would use, just one form of energy generation.
        Ultimately what we have to figure out is what can be done, and where. The answers will vary from place to place. ”

        Tom’s analysis explored the feasibility of the most cost-effective storage known (pumped hydro) serving a total reliance not on ‘renewables’ but specifically on the intermittent renewables of Wind and PV.
        It actually demonstrates that, if no cheaper scaleable storage tech can be demonstrated, a reliance on the intermittent renewable is physically and economically untenable.
        As you remark, partial reliance on Solar-thermal-salt would diminish the national storage problem, but it would do so specifically by dimininshing reliance on intermittent energy supply options. Thus I’d suggest that the storage problem is actually an artefact of an irrational assumption favouring the intermittent renewables over the baseload sustainable energies, and of a flawed comparison of their costs that excludes the intermittents’ untenable storage costs.



  26. Minor ommission. You mention losses in stored hydro energy is evaporation. The most significant loss, in most cases is leakage of the stored water – through the dam walls, abutments and between valleys.

  27. [shortened by moderator for brevity: follow links for more]

    There are many other points about pumped hydro your readers may be interested in on this thread: http://bravenewclimate.com/2010/04/05/pumped-hydro-system-cost/ . (Note in particular the reviewer’s comments at the end of the article.)

    Some points of interest:

    1. The rated power of the project needs to be available 95% of the time to get funded. So the rated power is based on the vertical distance between minimum supply level in the upper reservoir and full supply level in the lower reservoir. Therefore, if the upper reservoir is 100m above the lower, each reservoir has the same area, and the useable active storage is 10m depth in each, then the power rated power is calculated on 80m vertical head, not 100m. This becomes more significant where there is less topographic relief and we can not get the large head differences we’d like. For 50 m difference between the reservoirs, and 10 m active storage in each, the rated power is based on 30 m of head.

    2. For pumped hydro energy storage (PHES) to be economically viable:

    a. It needs sell lots of electricity to pay for the huge capital investment. So it needs to be used every day, not just intermittently.

    b. It needs to buy energy for storage at about ¼ the price it will get for selling.

    3. This comment explains in more detail why large scale PHES is not viable (mosatly) as energy storage for unreliable renewables: http://bravenewclimate.com/2010/04/05/pumped-hydro-system-cost/#comment-133008

    4. This provides a simple cost estimate of wind power with PHES compared with nuclear. http://bravenewclimate.com/2010/04/05/pumped-hydro-system-cost/#comment-86108

  28. In the end we will have to use less energy. Either nature will force us or we could change our ways – many don’t know this, but our current money system is the main driver of our quest for infinite growth. In a few words, all money is debt; all debt has interest rates; the money for the interest does not exist -> more debt. You can read “Modern money mechanics” by FED to have a complete picture.

    As a consequence of this, the real economy must expand faster than the interest rate. This is irrespective of the actual need for products and services, so a lot of “demand” must be artificially created.

    Anyway, there are solid grounds to maintain that a change in the system will rapidly reduce energy needs by 50% or more. Not to mention all the greenhouse gases that will not be released into the atmosphere…

  29. [shortened by moderator]

    I haven’t seen methane as a means of storing energy from electricity mentioned here.

    Basically your ingredients are H20, CO2 and electricity and you get CH4, H20 and O2 as output.

    The efficiency electricity -> methane -> electricity is reported to vary from 28% to 45%.

    The reasoning is:
    - the infrastructure to handle and store methane is already established (natural gas is >93% methane)
    - there’s no other viable means of long term electricity storage

    I think this is one of the most important solutions for a future powered by renewable energy because:
    - very little investment needed, only the conversion plants for the process mentioned above
    - cars can realiably be powered by methane. Due to tax reasons, natural gas powered cars are cheaper to operate over here in Germany than cars with gasolin
    - heating can easily be done by methan (about 50% of all homes over here are heated with natural gas)
    - can easily be converted to electric energy (about 14% of german electrical energy is based on natural gas, mostly for peak electrical loads)

  30. Tom,

    A nice post, but it needs a bit of touching up. Also, another model pumped hydro model is that of the Seneca Pumped hydro and Blenheim-Gilboa unit. These use a 100 acre pond that is made on the top of the hill/mountain -very little concrete needed an lots of storage for te acreage/volume stored.

    Anyway, on the US west coast are steep hills/mountians often averaging 500 to 1000 meter drops, which would use ocean water. You can store up one quarter or so of our pumped hydro needs this way -minimal concrete neededacreage needed, adjacent to ocean water.

    Lake Superior offers excellant potential for about 75% of its shoreline -from Sault St Marie north all the way back to near Marquette, Mich. And you could also do more with the Michigan sand dunesalong Lake Michigan.

    Next comes the Appalachians and New England/NY/Pa. This makes a great case for stopping mountain top removal for coal mining -those mountains would be better used as pumped hydro ponds like The Seneca Project. There is enormous potential associated with Vermont/NY and Lake Champlain, and many of the Finger Lakes.

    Some parts of the country are not well suited for PH (Great Plains, for example), but there is lots of wind and not many people there. Places like Florida, the Gulf Coast and much of Texas may have to go with more expensive versions of storage, like compressed air and batteries-but so what, many of them are living large on more productive regions, anyway (esp Florida).

    Also, a great way to put people back to work, too -shades of the CCC and WPA…

    Dave Bradley

    • Dave – your assertions contradicting Tom’s post offer no mathematical refutation, just confident proposals.

      As I’m unable to fault any part of the analysis you reject, I’d be interested to read the mathematical justifications of your numerous assertions.

  31. Is it easier to store heat than it is to store potential energy? My assumption is that it wipes the floor compared to storing electricity, but alas, I have not done the math.

    I noticed a couple of comments mentioned storing heat in gravel, which is odd. I don’t have a link handy, but I’ve seen numbers per cubic centimeter that show water is by far the better choice. Yes, I know, liquid salt is better still, but at what cost?

    Granted, storing heat at relatively low temperature would really only be applicable to space heating and maybe domestic hot water. Not a solution to the problems of liquid fuels for transportation or steam for electricity, which together I believe account for about two thirds of our energy use in the wealthy industrialized nations.

    Then again, if one holds the view that we are already committed to a world without cars and coal fired power plants, wouldn’t it be great if we could still enjoy relatively warm houses and the occasional hot shower?

    • The trouble with storing energy in water is that there is a limit to the temperature range. Storing vast quantities of steam is not practical on the TWh scale nor is pressurising the water to keep it liquid. Although gravel only has one fifth the specific heat of water, if can be heated to 500°C and cooled down to -150°C on a very large scale as is suggested by Isentropic Ltd. At this temperature range it is storing as much heat as water confined to the liquid range on a weight for weight basis and about 2.5 times as much on a volume for volume basis.

  32. I appreciate all the work that has gone into this estimate, but I think this report evaluates pumped storage for a use regime that is not currently economical and will likely never be economic.
    Pumped storage needs to be used very frequently to be economic, and the current 7 GW of pumped storage in Europe is used this way. Current pumped storage in Europe either adjusts static nuclear output to fluctuating demand levels, or adjusts fluctuating wind output to fluctuating demand levels.
    For long (~7 day) downward fluctuations in renewable output, the US will likely continue to use the existing thermal power stations for decades (already paid for and operating, why not) burning either the remnants of fossil fuels or biomass (wood, mostly). Sizing pumped storage to meet tail of the distribution probability does not make economic sense and will not happen.
    On the other hand, using pumped storage to flatten the daily distribution of solar and wind power already makes economic sense and will likely continue to make economic sense and continue to be expanded.

    • The logic of your proposal – that extant coal-plant capacity equal to the entire output of Wind and PV should be maintained as back-up for their intermittency – seems rather perverse. The more Wind and PV are deployed, the more extant fossil plant must be maintained and, to defray costs, operated ?

      There is of course a sustainable baseload biomass energy potential, but firstly, its use via dispersed gasifiers and combined-cycle-gas-turbines offers far better efficiency than its diesel-haulage to very large extant coal plants that are of far lower thermal efficiency.

      Secondly, the priority of maximizing the carbon efficiency of using that biomass resource demands the interrment of about 70% of its carbon as ‘biochar’ (~35% of wood tonnage), whose production provides ~28% of the wood’s energy potential as a crude syngas. This gas can either be combusted for ‘carbon-negative’ power or, more likely in view of the looming liquid fuel scarcity the DOE has forecast, converted to methanol for use in transportation roles.

      As to your question “why not” maintain fossil plants to make good the intermittency of Wind and PV, anything less than our most rapid practicable ending of fossil fuel dependence is in effect complicity in genocide (by serial famines) as intensified climate destabilization disrupts global food production.

      I hope we’d agree that the real objective is to end fossil fuel dependency ASAP, not to maximize the deployment of fossil-energy-dependent Wind and PV.



  33. Interesting look at this. You can of course just ignore the whole section about the concrete… I work at a pump hydro called Blenheim Gilboa in the Catskills region of New York. We have earth dams. NYPA operates another pump hydro at Niagara. Ours has about 900 feet of head. We can generate 1.2 Gw and our reservoir is good for 16 GWh. I’d say that you’d want to consider a massive reduction in fossil generation and use the gas for emergencies… having 10-20% capacity in natural gas that’s used rarely with an installed capacity of around 2 days of storage is much more practical. If you’re looking at a system like this also it’s best to figure in things like load avoidance programs.

    You don’t need concrete, you just need the topology and the free space. Concrete is safer, though.

    • Thanks—Just to clarify and put in the same units/terms in the article, the capacity is 16 million kWh, which is half the Raccoon Mountain example used in the article. The run time is about 13 hours. The head is approximately 300 meters (comparable to Raccoon Mountain, which also uses an earth dam, BTW).

  34. Tom, very interesting post, as usual, but I have several disagreements. The most important disagreement is: you are rebutting the worst possible renewable/storage scheme, which nobody serious was suggesting.

    The question is not whether the worst possible scheme is implausible, but whether the best one is implausible, since it’s the best one which we would choose if we wanted to go renewable.

    Well, what are the best schemes? One of the best schemes is to use a combination of solar thermal with overnight molten salt storage, for the summer, and windmills without any storage for the winter. Molten salt storage has a far higher energy density than pumped storage; a 2 GwHt tank is about 30 feet high. Also, solar and wind power are complementary, insofar as the wind is much stronger (in general) during the winter, and solar energy is stronger during non-winter months. Any brief periods when it’s both overcast in the desert (where the solar thermal plants are located) and simultaneously non-windy during the winter in Wyoming, would be met with natural gas intermediate load-following plants. This would probably be only a few days per year.

    This scheme is entirely plausible (we have more than enough resources and land area), and it would reduce fossil fuel usage for electricity generation by about 95%.

    Engineers and financial decision-makers will use the best and most plausible combinations of renewable energy generation.

    Tom S

  35. We have a *seasonal* electrical energy storage mechanism in which the storage has *already* been deployed across the upper midwest agricultural belt. There is more hydrogen in a tanker-truck of agricultural fertilizer grade ammonia than in a tanker truck full of liquid hydrogen. (The NH3 truck is a mild steel shell, the liquid hydrogen truck has a cryogenic rated inner vessel, superinsulation, and an outer shell, thus cutting way down on the liquid payload rate)

    There is enough ammonia in transit on Iowa roadways right now to supply generation for a day or two. This ammonia is in transit to the field to be applied before the ground freezes. There will be another burst of transport and application in the spring. Now combine that with the -33F atmospheric pressure liquid ammonia tanks.. and you’ve got a *seasonal* wind energy storage mechanism. It works even better when the ammonia being applied comes from the wind turbine on the farm, and then the excess can be stored for and collected for peaker power generation.

    For some numbers (and comparisons to pumped hydro and CAES, take a look at http://grid.coop/storage.pdf , and please give me feedback if you think the numbers are off.)

    • Troy,

      It’s quite possible to manufacture hydrogen, or ammonia, or some other combustible using excess power from renewables. However doing so has a big cost.

      The problem with chemical storage, is that it requires TWO energy conversions: first converting electrical energy to chemical energy (synthesizing ammonia or some other combustible); and then converting the combustible back into electricity when needed, using turbines or something similar. If each of the conversions is 50% efficient (which is optimistic) then the total round-trip efficiency is 25%. In other words, we would lose 75% of the energy stored that way.

      That’s the difficulty with energy storage. We can choose either efficiency or vast amounts. The efficient methods (like batteries, pumped storage, and flywheels) retrieve 90% of the energy, but are very limited in terms of how much energy they can store. The large-scale methods (like chemical storage) can store virtually unlimited amounts of energy but are very inefficient (75% losses). Nothing is simultaneously efficient and large-scale.

      That’s why it’s crucial to reduce the amount of storage needed, by using multiple kinds of renewables (wind and solar) simultaneously, since they peak at different seasons and complement each other. This reduces but does not eliminate the need for storage.

      If we use only 4 days’ worth of storage yearly, then something with 75% losses might be acceptable, especially since renewables are often generating excess power anyway and that power would have been wasted entirely anyway.

      • “The efficient methods (like batteries, pumped storage, and flywheels) retrieve 90% of the energy, but are very limited in terms of how much energy they can store. The large-scale methods (like chemical storage) can store virtually unlimited amounts of energy but are very inefficient (75% losses). …”

        Then choose both! That is, a hybrid storage system can have the best of both. Recognize that the seven ‘days’ of storage proposed here are not be equal in rate of use: the first day is used perhaps every day, the seventh seldom, if ever. The storage system cost and size improves when designed accordingly. For instance, use a high efficiency system (e.g. battery) for that first ‘day’ of storage (48 billion kWh) which might see a charge/discharge cycle every day to handle backing up solar at night. For the remaining six ‘days’ or so, use a lower efficiency chemical fuel storage system where the low efficiency costs little because it is seldom used, nor are concerned with depleting a finite resource if the fuel is produced by solar or other renewable sources. Infrequent use also allows bridge a fuel like natural gas to back up intermittent renewable sources with existing infrastructure while extending current reserves of natural gas. Meanwhile alternative chemical storage sources like energy density king hydrogen could be gradually brought up to scale.

      • I do not know why few people realize what you have just posted as it domes all chemical energy storage systems to a low conversion rate and in this case what it is converted to is highly toxic.

  36. How about rather than trying to build an enormous energy storage infrastructure, can we not just connect opposite sides of the earth with HDVC so that the sunlit side can power the part of the earth that’s in shadow? Isn’t this feasible or at least more feasible than pumping air in caves, etc. etc.?

    • I’d call it energy-storage scaremongering to surreptitiously talk down renewable energy into insignificance: a lot of technical blabber all based on the wrong assumptions. Now take a look at this for a change: http://www.beyondzeroemissions.org/

      • I’m not opposed to renewables (built my own off-grid PV at home and get most of my electricity from it). But I don’t pretend that storage isn’t a huge issue. I’m primarily addressing the damaging attitude that we have abundant viable solutions. If this attitude is correct, then fine—ignore my math/warnings: things will take care of themselves and there is no need for concern. If it’s wrong, then we will ignore the scale of the problem and not prepare adequate mitigation. I hope we go whole-hog on renewables, but we could be dangerously deluding ourselves to assume it is easy and will just happen when the time is right.

  37. Tom,

    What are your thoughts on a transition of fossil fuels/chemical fuels for use as a storage medium only?

    In your previous post about storage you showed how small a gasoline supply is needed to power a home for 7 days compared to other solutions.

    I know that (in theory) power from renewable energy could be converted into chemical storage. If that was done simply for our battery solution, how feasible would it be?

  38. Here’s an interesting reference. The Bath Country Pumped Storage station in Virginia is the largest in the world by power output. As it happens I walked some of the flow tunnels while it was under construction years ago. The BCPS is 2.7GW, dams are mostly earth and rock fill, cost $1.6B (1985 dollars), so this was a storage project in the neighborhood of a dollar a Watt.

    • It’s storage you really care about and pay for, and for Bath County, this is 27 million kWh. So the cost (in 1985 $) is about $60/kWh. Call it $100/kWh in today’s dollars. This is approximately the cost of a lead acid battery today (can get 1.8 kWh Trojan T-1275 for about $150).

      You can get a lead acid battery capable of hundreds of amps (thus thousands of Watts) for $50, making this pennies per Watt. This further illustrates that for storage, it’s the cost per kWh, not the cost per Watt that is most relevant.

      • That logic assumes the upper lake is drained *once* and the plant mothballed. Just as with utility grade battery storage, the cost per kWh attributed to storage is the capital cost divided by the total number of kWh they system will yield over its *total life time* (adjusted for the time value of money), plus the cost of round trip energy losses per cycle. If we guess BCPS runs the equivalent of six thousand 27 million kWh cycles over twenty years the cost is $60 / 6000 cycles, or $0.01 per kWh, again attributed to the storage system.

        • I’m sorry, but you are way off base here. No such assumption is built in. The lead acid battery in my example, at about $100/kWh does not mean you only get to use it once: you can get (some sizable fraction of) the 1.8 kWh many hundreds of times, bringing the cost per lifetime-stored-kWh down into the 10–20 cents/kWh range. What you pay for is how much capacity can be stored per cycle. That’s the primary metric of a battery. A small battery cycled many times may have the same amount of lifetime energy storage as a much larger battery (if used/charged at the same power and duty cycle), but the cost and usefulness of the small battery is nowhere near as great—especially in a pinch.

  39. I’m thinking that the right approach is to use the solar hot water heater I already have on the roof and marry it to a huge swimming pool size hot water tank. The tank can take months to get up to temperature and stay there over the winter. Thus providing heating for the house and a source of electrical energy from the waste heat:

    this turbine has the benefit of being small and (hopefully) cheap, but you can get a normal turbine for about $10,000.

  40. (note: I teach this material in the physics department at a large US university)

    You mix electricity generation and total energy consumption, which leads to some inflated numbers. US electricity demand is 450 GW, varying from 300 GW at night to 600 GW in the day. (reduce your numbers by a factor of 4)

    A more reasonable scenario is to meet all US electricity needs with 450 GW of nuclear, storing the excess at night in pumped storage and using it during the day. I’m ignoring the 10-20% loss but it won’t change the conclusion much. Storing 150 GW for 10 hours means we’re down a factor of 200 from your numbers, so only one facility like you describe is needed for this to work. 50 Ludington/Raccoon Mtn sized facilities could hold the needed energy, but the power is too low and we would need 100 of these.

    No one would build a facility in the way you describe. Dams are built in bottlenecks, as seen in the photos of L/RM. So the construction materials demands are much much lower than you calculate.

    • I appreciate your thoughts, and agree with your numbers. But our pretexts are different. You have 100% nuclear for electricity only. I have renewables for all our current energy needs: thus the “mixing” of electricity with total energy—I’m looking to the larger problem. Replacing current electricity demands with 100% nuclear does not address oil depletion, for instance. Then of course 100% (non-breeder) nuclear runs into resource limits in mere decades. Why would we knowingly commit ourselves to this path? But who knows—we might just do that.

      Experience tells me that these comments will launch a full-on debate on nuclear pros/cons. I’m going to stop that here: I’ll get to nuclear in another post—I promise…

  41. I’m a bit late to the party here, but I’d like to question the assumption that we’d need any significant amounts of storage at all.

    We’re talking about regional scales, not personal or even local. We already have a grid that ships power at least halfway across the continent, and we can reasonably assume that a smart grid will be part of any massive infrastructure upgrade like what we’re discussing.

    With that in mind, we’d build renewables such that the total continent-wide generation capacity is enough to meet demand. And, when you start thinking at that scale, you realize that you really don’t need to worry much about local cloud cover or calm conditions. When the Southwest is cloudy and the PV output drops, there’ll be a storm blowing through the Midwest and the turbines will be at capacity. And at no time will output in either reigon drop to nothing, or even all that far off maximum output.

    If you figure out the actual real-world worst-case regional conditions and design your grid to provide adequate capacity for those situations, you don’t need any storage at all.

    The only possible way that storage makes sense is if it’s cheaper than building excess capacity, and all these storage solutions sound like they’re far more expensive.

    By way of a real-world example, my PV array went live a couple months ago. I sized the PV for 100% offset and added SHW so I’d have enough for an EV I hope to purchase sometime in the coming years. I also had insulation blown in the attic shortly afterward, and the weather has been mild enough that I’ve yet to turn on either heat or A/C.

    And, with all of that adding up, I’ve yet to go a day when I’ve used more electricity than I’ve generated. And we’ve had a few stormy days in that mix — I just didn’t do laundry (etc.) those days.

    If this were to hold throughout the year, I’d only need enough battery backup for 12-18 hours at most to go off the grid and never run out.

    There’s no reason a continental grid couldn’t operate similarly.



    • A transcontinental grid is certainly a likely partner to a large-scale renewable infrastructure, but unless that infrastructure is seriously overbuilt, your personal experience (which is presumably supported by non-renewables at night) will not convey. I think you’re an outlier saying that storage will not be necessary for a renewable future. The grid can’t create the necessary energy on a much calmer-than usual night, no matter how smart.

  42. The Caspian Sea is well below sea level.
    It might be easier to use the Black Sea as upper and the Caspian Sea as lower reservoir instead of building them both.
    You just need to put the difference in height close together horizontally,
    this could be done by an aquaeduct, with a pumb/turbine at it´s eastern end.
    Since the Caspian Sea is large, the absolute change in its level
    wouldn´t be large for amounts of energy typically requiring to be stored, and evaporation would help you with pumping the water back.

  43. The fact that just one of the “small” dams considered here has as much concrete as the Three Gorges and Grand Coulee dams combined is humbling.

    Both of those dams also store much, much more water than the dams you’re assuming in your calculations, though – 3 to 10 (Hoover) to 20 times as much (albeit with head heights 3-5x lower).

    So there’s clear real-world evidence that storing the quantities of water under consideration is very doable.

    Unfortunately, this appears to be part of a pattern, where your assumptions appear to be biased towards pessimism. That isn’t unreasonable in and of itself, but do keep in mind that using consistently pessimistic arguments means that your calculations do not provide a reasonable basis for concluding that a project is infeasible, as an objective observer could rightfully suggest that reality may be less pessimistic.

    If you want to provide a compelling argument that something cannot be done, you must use optimistic assumptions, just as someone arguing a thing can be done must use pessimistic ones. To reverse those is effectively begging the question, and will sharply limit the persuasive power of your argument for people who do not already agree with it.

    • Point taken, and I agree with the sentiment. But comparing a river gorge storage capacity to a pumped storage installation is not fair either. If you compare to the existing pumped storage reservoirs, the scaling holds up reasonably well. The reason is that pumped storage sites tend to use broader, shallower features than river gorge dams, so that the manmade wall must be broad and voluminous. Fewer good “pinch points” in the mountains.

  44. Tom,

    First off, I absolutely love this blog and thank you for writing it. As another data-driven, green nerd, I really appreciate the work you put into it, and eagerly look forward to each new entry.

    That said, a request. One of the most common criticisms of environmentalists that I’ve heard is the fact that they tend to focus on what’s bad, on what we’re doing wrong, on what can’t be done. (I’m not saying that you fall into that category, but stay with me, if you would.) At the same time, while acknowledging that it’s certainly a comfortable myth as currently practiced, I do think that the notion of sustainability that has come to the fore in the last 20 years has really done a lot to engage much more of the mainstream population with the idea of addressing the problem.

    So, that (lengthy) preamble complete: Acknowledging that a massive reduction in consumption (material and energetic) is an inescapable requirement, what do you envision as a realistic lifestyle in a post fossil fuel world? For example, my mother once opined that sustainability was “shivering in a cold, damp cave in the dark”; what does it look like to you? And what innovations, investments, and sacrifices would it entail to get from here to there?

    Again, thanks for your thoughts here, and I look forward to your continued posts.



    • I am violating my rule of having discussions go off topic, but this one is perhaps of general interest to Do the Math readers, so please forgive me. I can still claim to be new at this business, in any case.

      In any case thanks for the note, and I will certainly be formulating something of a vision in future posts—though with the caveat that “visionary” is not my specialty. First, I am working my way through a process mirroring my own trajectory through this subject: initially trying to get a quantitative handle on the scope of the problem and coming to realize that the scale thwarts me at practically every turn. I think only after I have defined the box we’re in will I have the foundation to start talking about ways out of the box, or redefining our box (and no, the box I picture is not a dank cave). The good news for me is that I think there are physically viable paths to a decent future. But those paths will not automatically unfold at our feet, and lifestyles will likely have to change a fair bit. The political challenges are immense, and will only be undertaken if we collectively appreciate the fix we’re in. So I still have a lot of work to do…

  45. Only 3% of the dams in the U.S. have generators. Generators are not being added because this opens licensing proceedings and environmentalists have successfully demonized hydro to the point that dams are being removed. Did you look at the volume impounded by the other 97%?

    Recent hydro studies generally place significant limits on “acceptable” sites for reasons other than technical merit. You say “Our traditional hydro capacity could not be scaled up by even a factor of two—since the premier river sites have been plucked already.” but the National Hydro Association disagrees. You may wish to explore how much conventional hydro capacity is actually technically feasible in the lower 48 of the U.S. without artificial limits. In addition, Canada and Alaska have immense untapped hydro potential.

    • I have not seen this 3% figure, but I strongly suspect that the number would not apply to large dams (large head, large flow). I’ll bet for those projects the fraction with hydro swells to nearly 100%. Once a dam is in, you’ll find little objection to throwing a generator in from environmentalists. It’s the dam and the upstream impacts they want to avoid.

      As for exploitation, many river systems (Tennessee, Columbia, etc.) are dammed so that the river is all just flowing lake: the lake level for one backs up to the base of the next dam upstream. So those rivers are—for all intents and purposes—fully exploited. I have seen the 50% exploited number in a few places now. If you could provide a link to a study that says otherwise…

      • My company owns dams. Trust me, they avoid opening licensing proceedings to add generation, even when the dam has been there for 100 years. There are certainly many reasonable environmentalists, and we need them to provide balance and restrain the drill, baby drill nutjobs. Unfortunately there are many folks who use the licensing proceedings as leverage for unreasonable aims.

  46. 2,500 dams of the smaller 250 metre type would require a lot of concrete, do we even have that much cement-grade limestone?

    Let’s see: 2,500 dams * 19M m^3 of concrete each * 2,400 kg m^-3 (density of concrete) * 0.1 (10% of concrete is portland cement) = 11 trillion kg of cement.

    That’s more than 3,300 times the global cement output in 2010.

  47. Surely tidal power offers a huge potential source of energy from flowing water, without all the expense and problems of building massive dams? Early days yet, but arrays of sub-sea propellor-type windmills sited in high-tidal areas such as the North opf Scotland could potentially harvest reliable and renewable ebergy all the year round.

    • Nice where you can get it, but small beans overall. I’ll cover this in a post at some point.

  48. Tom –
    There are some further considerations on your findings on which I’d be glad of your thoughts.

    First, I wonder if you’d agree that, until we have a global climate treaty in operation capping and declining fossil fuel combustion, the idea of renewable energy actually displacing fossil fuel usage is wishful thinking, since any fuel that is locally displaced is simply bought and burnt elsewhere ? This being so, carbon efficient energy supply appears a secondary concern compared to the priority of achieving the treaty.

    Second, your analysis of PHES seems to indicate a further pivotal deficiency in the intermittent Wind and PV renewables’ viability, but I may well be missing something and so would appreciate your opinion. Given that the goal is a reliable energy supply with very high carbon efficiency, and given that pumped-hydro is easily the cheapest energy storage option currently available, the carbon debt of the immense 250m facility you posit warrants evaluation, particularly given its economies of scale.
    From your numbers: a 250m dam needs 19m cu metres of concrete using 32,000 GWHrs of energy in its production (32 billion KWHrs), and can supply 0.6GW of power (600MW) for seven days. Assuming that well-sited Wind and PV supplying that PHES facility have an all-year night & day capacity factor of 25%, the dam’s embedded energy and pollution payback is then at best:
    32,000GWHrs /0.6 GW /25% capacity /90% efficiency /8,760 hrs-in-yr = 27.06 yrs to payback.
    Setting a more optimistic capacity factor of 33% it would still take 20.29 years to payback its concrete’s pollution, let alone its additional construction outputs.
    For a dam serving the energy storage needs of just 600MW of Wind & PV plant even the latter period, in addition to the plants’ own payback periods, appears wholly untenable given the urgency of cutting fossil fuel emissions.

    Third, if the intermittent Wind and PV renewables cannot on energy or GHG costs justify new pumped-hydro capacity for their requisite energy storage, I’m wondering what level of deployment of these intermittents could be justified by re-dedicating existing Hydro-power to provide load balancing and the necessary 7-day energy reserve capacity, using their lowest 7-days-of-full-output water-levels + inflow rates as the prudent baseline ?
    This would of course require some additional sustainable baseload capacity to offset that cut in Hydro-power availability, but with good national supervision some of the annual downtime in new baseload plants could be offset in winter by hydro up-time to allow extra late-summer power capacity, thus optimising plant usage. But again, in addressing the issues of reserve energy capacity, I lack both data and expertise in the math and physics, and so would value your opinion.



  49. Good work. You are successfully demonstrating how doable some schemes are.

    Going forward, we (you) should always note the confidence-in and variance-of these schemes. We can’t easily compare something with high-confidence and low-variance to something with low-confidence and high-variance, so we need to make this distinction known up front.

    We can kindly compare hydro storage to solar thermal, for instance, but we can’t very well compare them to thorium reactors and theoretical geothermal. Again to the gentleman who said we must prove impossibility with the optimistic outcome, we risk becoming INTELLECTUALLY HOLLOW if we directly compare the best (or worst) case scenario of a theoretical scheme to just digging holes.

  50. Could you write an article about the sustainability of nuclear power? From what I have been reading it seems that only two things are stopping nuclear power from being the obvious solution: 1) Waste, 2) Accidents. Are these the only limitations or will we run out of fuel for those as well?

  51. Tom,

    The scenario here focuses on the supply side, but what if we looked at it as a demand side problem? (as a note, politically it is far easier to build supply side solutions, but lets continue on the “can it be done” model)

    Imagine a scenario where energy use in the US was divided into 3 tiers:

    1) Tier 3: Very optional power use.
    2) Tier 2: Very important, but not critical energy use.
    3) Tier 1: Critical Energy Use

    So to use a home for a model, your washer/dryer would be tier 3, the refrigerator is tier 2, and the heater in winter is tier 1.

    So in this model, we build enough solar capacity to handle all 3 tiers most of the time. If the capacity drops due to low solar/wind activity, we cut tier 3 power use.

    If extremely low activity is occuring, we cut tier 2. And then when activity is so critically low that tier 1 is affected….we go to the batteries.

    This would allows us to drop the amount of needed storage….but its not the amount I’m curious about but the rate. Right now, synthetic chemical storage is being discounted due to low efficiency of conversion (I believe 25% was quoted above as an optimistic estimate).

    I have no idea what percentage of energy use each tier falls in, but lets say 25% tier 1, 50 % tier 2, 25% tier 3.

    Under this model, one day of total battery use could be replaced by 4 days of regular days where tier 3 energy use was cut off (assuming the 25% efficiency number noted above).

    This puts chemical storage back in the game, and since we already have all of the gas storage infrastructure in place, renewable storage is well on its way to being ready to go.

    I put the big caveat to this whole disscussion, that I recognize such a scenario is hardly ideal (think building a dam is hard, tell millions of Americans to go without TV for a few days!). But if other storage options just are simply not good enough, it may be the way to solve the problems.

    From a math standpoint, does anyone have an idea what percentage of power would actually be considered tier 1 and tier 3?

  52. The author has erected a straw man – and the straw man has failed as it was designed to do.

    The author is not interested in realistic solutions – which do exist.
    And a much expanded pumped storage (say x10 existing to x20 existing) is very doable and very valuable in matching daily supply with daily demand.

    • Our current pumped storage capacity is in the neighborhood of 22 GW for 12 hours (about 200 million kWh; less than 1/1000th the goal I set forth). Expanding by 20 would bring us up to the ballpark of 400 GW for 12 hours. Indeed, this would be a giant step toward reliable power, capable of offsetting baseload demand in the electricity sector (approx 40% of our total energy) and only for 12 hours. My sense is that this still falls far short of what we would really need to preserve our current activities in a renewable infrastructure (which has to ultimately pick up the other 60% of energy).

      So truly I am not trying to erect straw arguments. I am genuinely trying to outline what it would take to transition fully to a renewable infrastructure and still live like kings. It’s super hard, and people who paint it as easy do us all a disservice.

Comments are closed.