The Bubble That Breaks Concrete

In June 1983, with snowmelt filling the reservoir behind Glen Canyon Dam on the Colorado River, engineers opened the left spillway tunnel to pass the flood. Within days they heard it. A loud rumbling came from inside the tunnel, then a heavy thumping on the crest above. When the flow was throttled and the tunnel inspected, the cause was almost absurd in its smallness. A deposit of calcite on the tunnel invert, a quarter of an inch high, about six millimetres, had been enough to start it.

Six millimetres of irregularity, under water moving fast enough, had triggered cavitation. By the time the spillway was shut for the season, the high-energy flow had excavated a hole 35 feet deep, nearly eleven metres, along with 134 feet long and 50 feet wide, cutting clean through the concrete liner and down into the sandstone beneath. Three-quarters of the tunnel's lining was gone. The US Bureau of Reclamation's own field report recorded the moment the damage announced itself.

Within 50 minutes the left spillway bucket suddenly stopped flipping the jet and the water downstream from the spillway bucket turned an ominous amber color as the jet carried chunks of concrete and sandstone into the river.

Burgi, Moyes and Gamble, US Bureau of Reclamation, "Operation of Glen Canyon Dam Spillways, Summer 1983" (PAP-0714).

Cavitation is not abrasion. Nothing in the water is grinding the concrete, the way silt and gravel wear a stilling basin. The damage is done by the water itself. Where high-velocity flow passes a surface irregularity, the local pressure can fall to the vapour pressure of water, and tiny vapour cavities form. A fraction of a second later, carried into a zone of higher pressure, they collapse, imploding against the surface with pressure spikes high enough to fatigue and pluck out concrete, then steel, then rock. The faster the water, the less of an irregularity it takes.

The thresholds are well established. On a spillway, cavitation damage can begin once clear-water velocities reach roughly twelve to fifteen metres per second. Above twenty to thirty metres per second, the surface tolerances needed to avoid it become so tight that no realistic finishing can hold them. Engineers track the margin with a dimensionless cavitation index, and once it falls low enough, below about 0.2 for ordinary concrete, no surface finish is good enough on its own. Something else has to protect the surface.

That something is air. The finding, made at the Bureau of Reclamation's hydraulics laboratory in the 1950s and proven on the spillway at Yellowtail Dam in 1967, is that a small amount of air entrained in the water near the boundary cushions the collapsing cavities before they reach the concrete. The standard puts a number on it.

Surface irregularities will not cause cavitation damage if the air-water ratio in the layers of water near the solid boundary is about 8 percent by volume.

ACI 210R-93, Erosion of Concrete in Hydraulic Structures, Section 5.3.

The device that delivers the air is the aerator: a ramp or offset across the spillway floor that lifts the flow clear for an instant and draws air beneath it through slots open to the atmosphere. Glen Canyon was retrofitted with air slots in both tunnels before the next runoff season, the design taken directly from Yellowtail. That principle now governs every high-velocity spillway in the world.

Indian practice has codified the same defence. IS 12804 sets out how to estimate the aeration demand of a spillway or outlet and how to size the aerators that meet it. On the Sardar Sarovar Dam on the Narmada, where the service-spillway chute runs faster than thirty metres per second, the aerators were proportioned from hydraulic-model studies at the Central Water and Power Research Station and built into the glacis. The outcome is the one a designer wants and a newsletter cannot dramatise: no cavitation damage has been observed on the spillway.

The subcontinent's cautionary case sits across the border, at Tarbela on the Indus, where in 1974 a gate stuck part-open drove flow at thirty-eight metres per second through an outlet tunnel, and cavitation tore into the lining before the tunnel failed. The mechanism was debated afterwards, with silt-laden abrasion argued as a contributor, so Tarbela reads best as a case where cavitation was a principal factor rather than the only one. The lesson it shares with Glen Canyon does not depend on the argument.

The lesson: On a high-velocity spillway, surface alignment governs survival as much as concrete strength does. A bump the height of a few stacked coins, in flow moving fast enough, can cut a crater through a concrete liner and into bedrock in a single season, and the strongest mix on a misaligned surface will still cavitate. Two defences hold, and they work together. Grind and tolerance the surface so the flow has nothing to catch on, and aerate the flow so that even where it does catch, a cushion of air near the boundary takes the blow. Below a cavitation index of about 0.2, the air is the only defence left.

Read more: Spillway Concrete: Designing for Abrasion and Cavitation Resistance →