The Dam That Leaked on Purpose (And the One That Leaked by Mistake)

In February 1983, the US Army Corps of Engineers completed the world's first all-RCC dam at Willow Creek, Oregon. It was 49 metres high, about 550 metres long, and contained approximately 330,000 cubic metres of roller compacted concrete placed in roughly five months. It was built to control the flood risk that had killed 247 people in Heppner, Oregon in 1903.

When the reservoir filled that spring, seepage appeared in the drainage gallery and on the downstream face. Not a trickle. Not a damp patch. Flowing water.

The design philosophy had been elegant in theory. RCC was a new material. The engineers believed that with sufficient paste content, lift joints would bond without bedding mortar. The RCC mass itself would be the water barrier. The upstream face used precast concrete panels, but the joints between them were left unsealed. There were no transverse contraction joints. No secondary impermeable layer behind the panels. The RCC body was expected to do the waterproofing work on its own.

The theory broke when water touched the dam.

Over the next twelve months (April 1983 to April 1984), a roughly $2 million grouting programme was executed through holes drilled into the dam from the crest. It reduced seepage by approximately 75%. The dam did not fail. But the world's first all-RCC dam had to be repaired during its first year of operation because a core assumption about RCC permeability was wrong.

Water Power Magazine, "Lessons from the Past"

That was the learning moment for the entire industry. Modern RCC dam practice calls for bedding mortar on lift joints, or an upstream impermeable facing (conventional concrete, geomembrane, or grout-enriched RCC), or both. ACI 207.5R documents the practice. ICOLD Bulletin 126 covers it in detail. It is now standard textbook material.

And yet the deeper lesson still gets misunderstood on sites today.

The seepage at Willow Creek was not the failure. The design assumption that produced it was the failure. RCC dams will always have higher joint permeability than conventional concrete dams. A 60-metre RCC dam with 300 mm lifts has approximately 200 lift joints stacked through its section, each one a potential seepage plane. The engineering question has never been whether RCC will seep. It has been how to design for the seepage that is certain to occur.

This is the framework we use on site when reading seepage on any RCC dam today.

Baseline. Distributed damp patches across the downstream face. Flow rate that tracks reservoir level. Stable numbers at the same reservoir level over time. Clear water. The dam is working as designed. Record it. Do not react.

Investigate. Flow rate increasing at constant reservoir level. New seepage locations appearing. White calcium deposits at exit points (the concrete is leaching). Something is changing. Raise monitoring frequency. Check internal drainage for blockage. Begin planning intervention.

Mobilise. Flow rate accelerating. Water turbid or carrying particles. Concentrated flow at a single point. These indicate internal erosion, the most dangerous failure mode a dam can develop. Notify the dam safety authority under the Dam Safety Act 2021. Hourly monitoring. A qualified engineer on site within 24 hours.

The difference between the three is not the presence of water. It is whether the water is behaving within design assumptions or outside them.

The lesson: The real learning from Willow Creek was not that RCC leaks. It was that the design must assume the leakage, build the impermeability into a facing system separate from the RCC body, and give engineers a framework to read the seepage trend rather than the absolute volume. Forty-three years later, every new RCC dam in India's pumped storage pipeline will live or die by that same principle.

Read more: RCC Dam Seepage: Causes, Prevention, and Remediation →