The Face That Holds the Water

The Ludington plant on the shore of Lake Michigan stores its water in a reservoir that does not sit on rock. The bedrock is more than eight hundred feet down, so when it was built in the early 1970s the reservoir had to be made watertight by lining it: compacted clay several feet thick, faced with asphaltic concrete on the inner slope. Then the lining leaked, and the way it leaked is the whole lesson of this issue.

Water that gets through a reservoir lining does not simply leave. It collects underneath, and on a pumped-storage reservoir, which is drawn down fast and often, the water trapped below the lining can end up at higher pressure than the water above it. At Ludington that pressure pushed up from beneath.

Water leaking from the reservoir caused hydrostatic overpressures to develop beneath the asphalt liner.

Schnabel Engineering, on the Ludington Pumped Storage Plant.

The result was uplift, bulging and deformation of the lining. The first remedy, a wellpoint system meant to draw the water down from below, was damaged by ice and could not keep up. The reservoir was eventually re-drained with a purpose-built trench drain beneath the lining and redundant pumps, with vibrating-wire piezometers watching the pressure. The face had become the thing the whole reservoir turned on.

That is the defining feature of a pumped-storage upper reservoir, and it is why its concrete is a different problem from a conventional dam's. An upper reservoir is a pond built on high ground with little or no river feeding it, a bathtub on a ridge. Because almost nothing flows in, almost nothing can be allowed to leak out: there is no catchment to make up the loss. And it is filled and emptied every day, pumped full when power is cheap and surplus solar is spilling, and run back down when the grid needs the energy returned. A storage dam holds a nearly constant head for months. A pumped-storage reservoir can rise and fall by tens of metres in a day. Studies of pumped-storage operation record drawdown rates of fifteen to more than thirty metres a day, against something like half a metre a day at a conventional reservoir. The face is not holding a static load. It is being flexed, wetted and dried, and heated and cooled, thousands of times over its life.

So the whole reservoir becomes a face, and two systems dominate. One is the concrete-faced rockfill dam, a rockfill embankment sealed on its upstream slope by a thin reinforced concrete slab. The other is asphaltic concrete, a bituminous facing prized for pumped storage because it flexes with settlement and cycling and can be patched quickly. At Tianhuangping in China, an 1,800 megawatt scheme with its upper reservoir on a hilltop, the entire basin was lined this way.

Asphaltic concrete facing was used for the whole 285000m2 upper reservoir basin.

Water Power Magazine, on the Tianhuangping upper reservoir.

The facing there is built in three layers, a thin closing coat over a hundred-millimetre impervious layer over a bonding course, and beneath it a gravel drainage blanket runs to a gallery at the bottom of the reservoir. That last detail is the point. The design does not pretend the face is perfect. It assumes a little water will get through, and it gives that water a measured path out, so it drains away instead of building pressure and lifting the lining as it did at Ludington.

Where the face is reinforced concrete instead, the slab is thin by dam standards, sized by a rule that has held for decades: roughly three-tenths of a metre plus two-thousandths of the head, so even under a hundred metres of water the slab is only about half a metre thick. The slab itself rarely fails. The joints do. The perimeter joint, where the face meets its foundation, is the critical leakage path, sealed with waterstops, and Indian practice has a code for exactly this, IS 12200, covering PVC waterstops at the transverse contraction joints of concrete and masonry dams. On the highest concrete-faced dams the face has failed in another way, cracking in compression at the vertical joints as the rockfill deforms beneath it; at Barra Grande in Brazil the leakage through a ruptured face passed a thousand litres a second. A pumped-storage face is lower and its loads are different, but the daily cycling is the price it pays instead.

India is about to build a great many of these. At the end of 2025 the country had about seven gigawatts of pumped storage in operation. The Central Electricity Authority's 2026 roadmap sets a course to cross a hundred gigawatts by 2035-36, with about twelve gigawatts already under construction, because pumped storage is the mature way to hold the surplus of a grid filling up with solar. Most of the new schemes are off-stream and closed-loop, which means two purpose-built reservoirs each, on high ground, lined. That is a very large amount of new face going into the ground on a compressed timetable.

The lesson: On a pumped-storage upper reservoir the face is not a detail of the dam. It is the dam. It has no catchment to hide a leak behind and it cycles every day, so it fatigues in ways a storage dam's concrete never does. Design the face for the cycling, not only the static head. Treat the perimeter joint and its waterstops as the thing most likely to decide whether the reservoir holds, and specify them to IS 12200 rather than by habit. And build in an under-drain, as Tianhuangping did and Ludington had to learn, so that the water which inevitably finds its way through is measured and led away, not trapped to lift the lining. The hundred-gigawatt ambition is a face-slab-and-waterstop problem as much as it is a turbine one.

Read more: Pumped Storage Hydropower: Why Concrete Technology Will Define India's 100 GW Ambition →