Nepal’s power stations were built where rivers fall fastest, which is also where mountain floods run. The August 2026 collapse cost the country roughly a tenth of its generating capacity and trapped hundreds of workers underground.
Hydropower needs two things: a reliable flow of water and a steep drop to turn it into power. Himalayan valleys supply both in abundance, which is why the Trishuli, the Bhote Koshi and their neighbours are lined with dams, headworks, tunnels and powerhouses. It is also why, on the morning of 26 August 2026, a debris flood released by a rock-ice avalanche near the Nepal–Tibet border met so many machines.
What was lost
Nepal generates more than 95 per cent of its electricity from hydropower, and installed capacity had reached roughly 4,200 megawatts by May 2026, according to Columbia University’s Center on Global Energy Policy. Its analysis counts 12 operating plants damaged, with a combined 431 megawatts of capacity, and a further 15 projects under construction with 470 megawatts affected. The Associated Press has put the loss at about a tenth of the national power capacity. Other sources give a different count of damaged projects, and the figures should be treated as provisional until the Nepal Electricity Authority publishes its assessment.
Nepal had also become an exporter. It began selling surplus electricity to India in 2024, mainly in the wet season, and after the flood it asked India for imports earlier than usual in the winter. Generation had grown from 6.1 terawatt-hours in 2020 to 11 in 2024, and the government’s stated ambition is 30,000 megawatts by 2035. The timing of the loss matters, because river flows fall through the dry season and Nepal usually turns to imports from October.
Where the plants stand
The Trishuli valley holds a chain of stations. Reporting on the rescue named Trishuli 3A and 3B as the focus of tunnel operations, and the Upper Trishuli-1 project appears in industry accounts of the damage; other reports list Rasuwagadhi, Chilime and Devighat among the affected stations. A typical plant on these rivers has a diversion weir and intake, settling basins, a long headrace tunnel, a powerhouse, a switchyard and a transmission line. Each has a different weakness. Weirs and intakes are exposed to boulders and to the abrasive sediment load of a debris flood, which can also silt up basins. Tunnels are vulnerable at their portals, and powerhouses sit at the bottom of the hydraulic head, close to river level.
The tunnel problem
The human cost was concentrated underground. Hydropower tunnels carry water from an intake to a powerhouse and are staffed by construction and maintenance crews. By 31 August, 933 workers stationed at hydropower projects were reported missing and nearly 300 had been rescued from tunnels, while Nepal’s army said the volume of debris made opening them a major challenge. ICIMOD drew a comparison with the 2021 Chamoli disaster in India, when workers were trapped in tunnels after a rock and ice avalanche.
A tunnel is a confined space with a limited number of portals, usually in the river valley itself. A debris flood carries mud, boulders and timber that can fill a portal in minutes, and stranded workers have few routes to escape.
Fifteen of the damaged projects were still under construction, a stage at which large work crews are employed on site.
A pattern, not an anomaly
Three recent Himalayan disasters have hit hydropower, each with a different trigger. At Chamoli in February 2021, a rock and ice avalanche was identified as the cause by a study in Science, and the flood swept through hydropower works on the Rishiganga and Dhauliganga. In October 2023 a glacial lake outburst in Sikkim destroyed the Teesta III dam at Chungthang, one of the state’s largest projects. In August 2026 a rock-ice avalanche was again the origin. The hazard is not confined to one type of trigger, which complicates any single design response.
What engineers and analysts say
Reporting by the Associated Press quotes several specialists. Jakob Steiner, a geoscientist at the University of Graz, cautioned that physical defences may be overwhelmed by large floods. Ramraj Narasimhan of the Coalition for Disaster Resilient Infrastructure argued for building redundancy into systems in place of trying to make single structures indestructible. Himanshu Thakkar of the South Asia Network on Dams, Rivers and People argued that hydropower projects can amplify damage downstream. These are differing views, and they point to different remedies: better siting, redundant networks, or fewer structures in the corridor.
The system effect
Losing roughly a tenth of capacity in one valley is a system problem as well as a local one. Plants along a single river share its flood risk, so they tend to fail together, and a grid that draws nearly all its electricity from hydropower has limited spare capacity to absorb a correlated loss. That is the sense in which the corridor matters: the concern is less any one plant’s exposure than the common exposure of many.
What is not yet known
Public information does not yet show whether the damaged projects had assessed rock-ice avalanche or debris-flow hazard, how portals were designed, or what warning arrangements protected workers. Those are questions for the regulator and the developers. The Imperial-led analysis of the disaster concluded that no existing early warning system could have given enough lead time, which shifts attention from detection to design and to who is inside a structure when the alarm comes.
The engineering questions are practical ones: where to put a portal relative to the flood path, how to give tunnels a second exit, how to protect switchyards and transmission lines, and how to keep supply going when a valley’s plants are lost together. Answers may differ from valley to valley, but the August event has made the question hard to defer.
Nishtala V Subramanyam
