NOTE: This post does not reflect my latest thinking on water storage. Please see the new summary.
I’ve stated that the reason it’s possible to locate bags at somewhat different elevations within a storage field is the transient valves—remote-controlled shutoff valves in the feed tubes for every bag. How does that work?
I’ll look at some simplified examples in which I assume that the transient valves and distribution pipes are rated for 100 psi (6.8 atm). This corresponds to a pressure head, or water-column height, of 230 ft (70 m). That means we can lay out the bags on the mountain with a 230-foot difference of elevation between the lowest and the highest bag. I’ll also assume that the bags themselves are rated for 10 m of head (the height of the bags in the Capitan design, times a safety factor of two). 10 m is 33 feet, with a pressure of 14 psi (1 atm). There’s nothing magic about those ratings; they’re only for discussion.
These drawings are meant to illustrate the concepts and are not to scale. (The lower storage field won’t be shown, and won’t be discussed here; all the same considerations apply to the lower field as the upper, but the issues are expected to be much more tractable at the bottom, due to simpler geography and fewer constraints at most potential sites.)
First, suppose we have an upper storage area that’s totally flat and level:
Assume these bags are initially at their “empty” level (they’re never completely emptied, to prevent flow obstruction and to keep the bags anchored), and that their transient valves are closed. There would be some benefits to a completely closed, air-free system, with the distribution pipes and the penstocks always kept full of water, and that will work for this layout.
Now, suppose some electricity starts coming in from the grid to be stored. The pump, at the bottom of the hill, activates, ramping up the pressure in the penstock. At the same time, the valve to the first bag opens. Water under a few psi of pressure flows into the bag. (The water coming out of the pump is around 1,400 psi, but the pressure gradually decreases as you go up the penstock.)
When the bag is filled to its 3,600 cubic meter capacity, as indicated by a pressure gauge in its feed tube, the valve closes (over at least several seconds, to avoid pressure spikes), and simultaneously, the valve for the second bag opens. Now water from the penstock fills the second bag. This bag is at the same elevation, so it can be filled at the same pressure.
When the second bag is full, its valve closes and the valve to the third bag opens. This process continues until all the bags are full. Then the pump shuts off, because this mini-storage field can’t accept any more water.
When it’s time to get the stored energy back, the process reverses. A valve in the powerhouse connects the penstock to the turbine instead of the pump, at the same moment that the transient valve for the rightmost bag is opened. This allows water to flow out of the bag, through the distribution piping and penstock, and through the turbine, producing electricity. When that bag has reached its targeted minimum amount of water, its valve closes, and the next one to its left opens. The process repeats until all the bags have been drained. At that point, the powerhouse valve to the turbine is closed, and the system waits in its fully discharged state until there is more energy to store. (A valve in the powerhouse will be closed to keep water from flowing back through either the turbine, or the pump—centrifugal pumps are not positive-displacement, so water can run through them when they’re idle.)
This system is very easy to manage. We could operate the valves exactly as just described—”first in, first out”, that is, fill the nearest bag to the penstock first, then the second, and so on. Or, we could open all the valves at once, and let all the bags fill together. Actually, we can fill the bags in any order, and any number at the same time, because they’re all at the same elevation. I’ll return to this point later.
Next, here is a set of upper-storage bags on terrain that slopes toward the penstock end of the storage field:
If we try to operate this as a closed system, as we did with the previous one, there’s a serious problem. We can only expose the bags to 14 psi of pressure above ambient, so to fill a bag, the pressure on the distribution-pipe side of its valve has to be 14 psi or less, or it isn’t safe to open the valve (the bag will see more than its rated pressure, and may burst). Now suppose we are ready to start filling the leftmost, lowest bag. Suppose the valve to the rightmost, highest bag is 230 feet (70 m) higher; if the pressure at the leftmost bag is 14 psi, the pressure at the rightmost bag will be 14 – (230 * 0.43) = negative 85 psi! This is physically impossible because a vacuum (-15 psi relative to ambient) is the lowest pressure that can exist. What this means is that inside of the distribution pipe, at the top, is a vacuum; the water further down the pipe will be boiling at room temperature, and the pipe will be doing its best to collapse.
So we can’t have a closed, air-free system. We need to have the distribution pipe vented to atmosphere at its top, so the water surface inside it can start low when the first bag is filling, and rise as the series of bags are filled. The water surface in the pipe, being in equilibrium with outside air, will always be at 1 atmosphere, so there will be no partial vacuum, no boiling point depression, and no force trying to collapse the pipe.
When we’re filling the topmost bag, the pressure in the distribution pipe where it joins the penstock (230 feet lower) will be a positive 150 psi, which it’s designed to handle. We protect the bags from that pressure by keeping their transient valves closed.
If we insisted on an air-free, closed system, there are some other things we could try. We would have to keep the top of the system at one atmosphere, and somehow deal with the 150 psi pressure at the lowest bag—maybe we could design valves that are able to open very slightly so water flows into the bag more gradually, and doesn’t burst it. Besides adding complexity, such pressure reducers tend to waste a significant amount of energy as friction, which is the last thing we want to do. So a vent seems best.
To fill the topmost bag, it needs some positive pressure to push the water in (we specified that it can handle up to 14 psi; the actual pressure will be determined by the flow rate we want, i.e. how fast we want the bag to fill). That means the water level in the distribution pipe will rise higher than the top of that bag, so the distribution pipe should turn upward and go up high enough that water will never reach its top (and high enough above the ground to stay clear of snow). There should be a cap and screen over the pipe to ensure that nothing but clean air gets in. Something simple should do:
Next, consider a system where the slope is the other way, away from the penstock:
This design, like the previous one, needs the distribution pipe to be vented, or we can never fill the lower-elevation bags. It also has another problem: if we want to fill the lowest bag at a pressure of 14 psi or less, we can’t, because when the system is at that pressure, the water level in the penstock and distribution tube will be far below the high spot, and there will be no way for water to get over to the bag. And when it’s time to drain the bags, the water from the lowest bag won’t be able to get out to drive the turbine, for the same reason.
The obvious solution is to re-route the main distribution pipe, either around the high spots (if the terrain permits) or through them in a tunnel, so that it never rises higher than the lowest bag, something like this:
Keep in mind that the distribution pipes are all in the low-pressure regime (100 psi maximum in this exercise), so the cost of the extra pipe isn’t that daunting. It will cost something, though, to trench or tunnel a place for the pipes, or route them around. The light blue line shows the lowest water that should ever occur in the feed system; this is low enough to drain the lowest bag, so there’s no reason to let it go lower.
If we need to, we can branch the distribution pipe so it supplies water to different areas of low-elevation bags that are some distance apart, as in this case where there are low bags at either end of the storage field, and higher ones in the middle:
These are all conceptual drawings, and things will look very different in our Capitan site, where the upper storage has 1,000 bags. If a site is particularly difficult, it will be tempting to just get something like a dragline excavator up there and flatten out the mountaintop, to make the water distribution problem simpler. That can certainly be done, and in some cases, it might be the best answer. In other cases, that solution won’t be acceptable or won’t be cost-effective.
Filling or draining one bag at a time is clearly not something that is going to happen in a system with 1,000 bags—many bags will be open at once. This will still proceed in order of elevation, filling the lowest bags first and draining them last. Having a number of bags at each elevation gives the system operator (or strictly speaking, the software) more flexibility.
The storage fields are not there for their own sake. The purpose of the system is to store or release energy exactly as the ever-fluctuating grid requires. What’s required of the upper storage field (and the lower field too) is that it must be able to accept or return whatever amount of hydraulic energy (pressure times flow rate) that’s asked of it, moment to moment. When the upper storage is a large lake, it’s easier to just accept that this is true without thinking about it too hard. It’s not as obvious that our storage field of 1,000 individual bags, each with its own valve, can produce the same result. What is obvious is that we have very fine-grained control—we can go from 20 bags supplying water to the turbine, to 30, within a few seconds. This suggests that we can feed the turbine as well as a large, passive reservoir can, or maybe better. But we will have to do some software simulations to prove that.
It should be quite feasible to write software that will analyze the elevation data for a prospective site and the number of bags to be placed there, and compute the bag layout that will minimize the amount of earth-moving to be done and the amount of distribution piping required, while maximizing the ability of the storage field to respond to energy demand.