- The fundamental principle is to
use pistons in vertical cylinders as masses to be raised by fluid pressure
as a medium for storing energy. Letting down the masses to drive the fluid
through power generators would deliver the power on demand.
- The extent to which the
cylinders are to be built above or below ground level is not essential to
the principle of the device; for any scale of unit and choice of materials the economic output is a function
of height plus depth; the longer the path up and down the cylinder and the
greater the mass to follow that path, the better. Accordingly the ideal
cylinder in any realistic situation should be determined by the relative costs
of the piston material, and of the above-ground and subterranean
construction at various depths and heights. Each of these costs is
significant in calculating the trade-offs and the general economics of any
such project. This point of cost justification will be taken largely for
granted in the following discussion; specific figures would be too
speculative at this point.
- The pistons are to be
functionally “dumb”, inert in themselves, with no internal mechanism. They
are raised and lowered by pumps that control fluid pressure through ports in the
cylinder wall. When a piston is neither charging nor discharging its
stored energy, detents that retract into the cylinder wall can hold it passively suspended
at suitable heights without relying on friction or expenditure of power. This aspect of the design differs radically from some other designs that use brakes or active devices.
- The choice of working fluid is
a matter of choice to suit local needs and approaches; I like the idea of
non-drying, non-gumming oil where that may be practical, but obviously water, possibly brine, has
its own advantages, partly depending on the design and circumstances.
- Notionally each piston should
be monolithic in function and as dense as may be practical. I still see
lead as the most desirable material, followed by various forms of iron,
though I have seen alternative suggestions such as concrete, which to me
seem inferior in several respects. As I explain below, the fact that
the pistons are monolithic in function need not imply that they are
monolithic in structure.
- One problem with lead is that
it is softer and more deformable than rival materials such as steel or even cast iron. It accordingly is vulnerable to damage in accidents during
installation and maintenance. Such damage could compromise a piston’s precision and
options for sealing the contact between the piston and the cylinder wall.
- Another problem with the piston
idea is that the mass would need to be enormous. Various designs assume
piston masses of hundreds of tonnes or even hundreds of thousands of tons.
Installation and handling of such large objects and great masses as indivisible
units could be a bad idea if there is any practical alternative.
- Accordingly instead of simple,
solid slugs of lead, each piston should be jacketed with a suitable
material. The most obvious design might be a hollow box to be filled
effectively solidly with suitably packable lead segments. Originally I
assumed a steel box, but I am increasingly interested in the possibility
of polymer jackets, probably with steel-reinforced floors designed to
hold the vulnerable lead segments and to rest on detents safely and without harm.
- Each segment of lead inside a
piston box should be suitably shaped and coded with a unique, machine-readable
identification number so as to be suitable for installation and removal on
site by intelligent gantries. There are so many possible alternative designs
for such segments that I do not discuss the details here, beyond remarking
that any design needs be easy to place into a stable and precise
configuration, probably in an oil medium. The essential effect is that the
jacket then could be designed for assembly and manipulation on site by
gantry, installed and packed with lead, with no need to handle
the total piston mass by any means other than fluid pressure, and only after it is installed.
- The lead segments might be designed so that once
installed, their weight could exert some outward force on the piston walls to enforce proper
sealing against the cylinder
walls, though this is not an essential feature. One way to achieve that
would be by fitting the slabs of lead at an angle of probably less than one
degree from the horizontal on an oil surface, thereby resting a slight
fraction of their mass against the outer wall of the jacket.
- The jacket itself, if of steel
and of really large scale design, must also be modular. In sizes up to
say, two metres diameter and ten metres long, the jacket fairly routinely might
be transportable and installable, but some of the designs I have seen
online seemed to imply sizes of about ten metres diameter and 125 metres
long (and eight of those above each other in a single cylinder...) Though I do not here deny the
feasibility of such units, nor even their possible desirability once installed,
I do not think it would be worth trying either to transport or install
them as finished units, whether empty or with their internal mass
installed. The very notion of assembling the jackets on installation is
challenging and intriguing, bearing in mind the many technologies for
doing so, ranging from welding to bolting and gluing; it is as sobering as
the challenge of the required precision on the required scale. The whole
idea strikes me as a charming example of an engineering project in its own
But not trivial.
- Nowadays we have options other
than steel jacketing. Some modern polymers, probably fibre-reinforced,
might be equally suitable for the jacketing, either as a one-piece
structure, or floored with steel, or with just a steel rim around the
bottom to accommodate detents that might be designed for extension inward
from the cylinder wall to hold the pistons stationary where necessary. One
advantage of such jackets over steel, is that they could be cast or welded
in place from the bottom up, creating an essentially perfect fit with the
cylinder wall. Depending on the nature of the polymer, they could be cured
with the aid of ultraviolet or gamma ray sources as they are fabricated in
place on site, though for my part I rather favour thermoplastics instead
But those are details that one should not force on the polymer engineers in advance.
- Brakes or detents of some sort clearly
are necessary for a number of purposes. Pumping fluid beneath or between
pistons would achieve nearly all required positive movement, either
raising pistons, usually to store energy, or lowering them, usually to
generate power. However, when it is necessary for a piston to remain in
one position indefinitely, such as while the energy reservoir is full and
the power demand is very low, then under constant pressure one must expect
undesirable leakage past the piston. It then would be desirable to apply
positive static control. Or such controls might become necessary in dealing with a damaged piston. Some schemes proposed in other discussions suggest
brakes for holding the piston in place, but I reject that idea partly
because of the constraints it would place on the piston's density and
complexity, and partly because of the tendency to damage the cylinder
wall, not to mention the problem of brakes slipping on the wall lining, very
likely damaging the wall in the process and jeopardising the integrity of
- Instead I prefer the use of
detents. There are many design options, but what I like offhand is the idea
of detents that fit into gaps in regions in which fluid can be pumped into
or out of the spaces between stacked pistons. The detents could be
cantilever bars recessed into the cylinder walls. They could be deployed
by control machinery when no piston either is in the way or needs to pass.
The detent assemblies might perhaps include some sort of
shock-absorbing mechanism; even at the immensely slow speeds in question,
one does not simply say "whoa!" to a 10000-tonne mass. These
details too are for the engineers to decide. Still, valving the fluid flow
should offer very fine control indeed, so it might be possible to rely on
direct control rather than shock absorbers.
Still, shock absorbers might be an important feature in the event of a catastrophic control failure. I like to have a passive fail-safe option; I was horrified to discover that the Fukushima nuclear reactors had needed active controls to prevent disaster, which could have been prevented by passive controls that, expensive or not, would have been a lot cheaper than cleaning up the mess afterwards.
One attractive idea is to use bi-stable detents designed to remain passively engaged or retracted until once again activated, but mono-stable detents that only let pistons pass when actively permitted, might be safer.
- At the same level as each ring
of detents there would be one or more input/output ports in the cylinder
wall, through which fluid could be forced in by pumps or drawn off either
to generate power or to lower a piston.
- The seal between the piston and
the wall would be of self-lubricating polymer on either the surface of the piston or of the wall, or both.
The cylinder's internal wall might be of polished steel or lined with hard silicone
or other appropriate polymer surface of a type that could readily be repaired or serviced when necessary. In a steel cylinder jacket the seal
could be cylinder rings extending entirely round the piston without any
gap, and made of solid, self-lubricating polymer. If the piston's entire jacket
were itself of polymer, it could be fabricated in place to fit the
cylinder, providing its own seal with no piston rings. To exploit the
flexibility of the polymer in such a jacket, lead segments inside the
piston could be designed to exert some small fraction of their weight outwards,
forcing the piston wall snugly against the cylinder wall. In either case the width of
the sealing surfaces should exceed the width of any interruptions in the
wall, such as inlet-outlet openings or detent recesses, so that passing such gaps would
not present any seepage problems whenever the piston passes over.
- The bottom edge of the
otherwise cylindrical piston should be chamfered into a recessed
rectangular rabbet all round the edge, deep enough to accommodate the
detents, and high enough to enable the matching input-output ducts to work
at full capacity even if the head of the piston immediately below is in actual
contact. See figure. In the design of large pistons this might demand that
at least the floor of the piston be of steel. The surface of the head of each piston should not be such as to permit pistons to contact each other too closely; it should always be possible to inject fluid between, and never be possible for them to damage each other or get stuck together.
- The foregoing designs would
place all the fluid- and energy-handling resources and also all the
controls outside the pistons, and in fact outside the cylinders as well.
Ideally, once the cylinder and pistons are constructed and installed, they
should never need maintenance apart from occasional inspection every few decades. All the
rest of the attention could be devoted to the piping, the fluid, the pumps
etc. None the less, the design should permit inspection and maintenance at any time without dismantling, and as far as may be, without interrupting operation. This might affect the choice of fluid, giving preference to transparency etc. Clear water with traces of harmless corrosion inhibitors and preservatives such as zinc compounds, might have advantages.
- Some of the designs described
by other parties online put multiple pistons into a single cylinder. This entails
both advantages and disadvantages. Most importantly it makes it possible
to limit the size of any mass to be handled as a unit at any point in an
operation; it also reduces the pressure that individual pumps must work against
and so on.
- Multi-piston cylinders do
introduce complications, such as the need for fluid to bypass pistons, and
they complicate the design of semi-open units that either have
mushroom-headed pistons, or that elevate large fractions of the working
mass above the top of the cylinder. But for very large systems multiple piston
designs probably are unavoidable for this approach at least.
- In the multi-piston approach considered
in this essay the pressure pipe or pipes run up the outside of at least
one side of each cylinder, with inlets or outlets for the fluid at least at
each level where the chamfered rabbet around the bottom of a cylinder may be
brought to rest. At each such level, there also should be a set of detents.
- Assuming that we use the
multi-piston approach, I propose that the workable pressures within the
system be at least somewhat greater than twice the pressure exerted by any
- In withdrawing power from a multi-piston
cylinder, first insert the detents into the bottom of the gap through which
the lowest piston selected to deliver power must pass. Then insert enough
fluid beneath the pistons selected for immediate power delivery, to raise them enough to remove
the load from their detents. When the force on the detents stops,
retract the appropriate detents so that the piston can exert its downward working pressure. Thereafter begin to remove the fluid supporting
the selected pistons, permitting it to drive the selected power turbines, then re-inserting their
fluid into the space above the highest of the pistons providing power at this time. As each piston delivering power comes
close to the end of its intended course, the detents at the bottom of that
gap already having been inserted, the next piston above gets released as
required, and the procedure continues.
- Cylinders could in principle be daisy-chained in series
to maximise output pressure, or
combined in parallel to maximise power output at a given lesser pressure.
- In any
closed or semi-closed system in which the fluid is driven back up above
the moving piston or pistons as they sink, the mass of the fluid itself
does not contribute to the total output energy, because it has to be raised
during power output.
- To raise any combination of multiple pistons for accumulating energy, first raise the top piston to just above its highest intended detents, probably inserting intermediate detents as the piston passes, as a precaution against emergencies. Stop just above the top detents and engage them. Then proceed downwards, raising the next pistons in turn.