Joint water-supply and hydropower management in a capacity-depleted multipurpose reservoir: the two-input AD-2I–WE model for Akhangaran

Water reservoirs are the great storage tanks of civilization, holding back the seasonal rush of mountain rivers to release water slowly when the fields are dry and the cities are thirsty. But these tanks have a hidden enemy: silt. Over decades, the sand and mud carried by the river settle on the bottom, slowly filling the tank from the floor up. This process, known as sedimentation, does more than just reduce the total amount of water a reservoir can hold; it changes the very rules of how that water is managed. A water level that once meant a full tank might now mean a nearly empty one, and the timing of when water is released becomes a delicate balancing act between keeping crops alive and generating electricity. In Central Asia, where snowmelt feeds rivers in the spring but farmers need water most in the hot summer, this balance is critical. If the storage capacity shrinks, the ability to carry water from the wet season to the dry season diminishes, threatening both food security and power generation.

Researchers at TIIAME National Research University in Uzbekistan have tackled this problem at the Akhangaran Reservoir, a vital facility near Tashkent that supplies irrigation water and hydroelectric power. By combining decades of underwater surveys with daily records of water levels and power generation, they developed a new way to manage the reservoir that adapts to its shrinking size. The study reveals that the reservoir has lost a significant portion of its capacity since the 1970s, forcing operators to rethink how they move water through the seasons. The team created a flexible decision-making model that uses two main inputs: the current amount of water in the tank and the annual order for how much water farmers need. This approach allows the system to prioritize safety and water delivery first, and then find ways to generate more electricity without compromising the water supply.

The history of the Akhangaran Reservoir shows a clear and troubling trend. Between 1972 and 2022, the total volume of water the reservoir could hold at its normal level dropped from 197.7 million cubic meters to 169.4 million cubic meters. That is a loss of 28.3 million cubic meters, or about 14 percent of its original capacity. The rate of this loss has accelerated in recent years, with nearly two-thirds of the total loss occurring between 2006 and 2022. This reduction means that the physical buffer available to store spring snowmelt for use during the dry summer months has shrunk significantly. When the researchers looked at how the reservoir was actually being operated between 2020 and 2025, they found that the existing rules, which were based on the reservoir's older, larger dimensions, were no longer working effectively. Under the old approach, the system failed to meet water demands in nearly 30 percent of the time periods measured, with the worst shortages occurring in July, August, and September when irrigation needs are highest.

To understand how to fix this, the researchers tested three different management strategies. The first was simply observing how the reservoir was currently run to see where it failed. The second was a rigid approach that tried to guarantee a specific amount of water remained in the tank at the end of every year, regardless of what happened during the summer. The third was their new adaptive model, which set a strict safety floor to ensure the reservoir never dropped to a dangerously low level during the year, while also adjusting how much water was released based on the current situation. The rigid, end-of-year target failed because it forced the reservoir to hold back too much water early on, causing shortages during the critical summer months. The new adaptive model, however, succeeded in keeping a safety buffer of 15 million cubic meters in the tank at all times. By doing so, it improved the reliability of water delivery by about 6 percent and reduced the total amount of water short by roughly 21 percent compared to the current operations.

A key finding of the study is that managing the timing of water releases can significantly boost electricity production without using any extra water. Hydropower depends not just on the volume of water flowing through the turbines, but also on the height of the water above them; water released from a higher level generates more power. The researchers found that by shifting the release of water to times when the reservoir was fuller and the water pressure was higher, they could increase annual energy generation by nearly 34 percent compared to a standard schedule. This increase came entirely from better timing, not from consuming more water or lowering the safety of the water supply. Even with these improvements, the model still produced less energy than what was actually observed in recent years, suggesting that the real-world operations might be taking advantage of conditions that are difficult to predict or that the physical limits of the turbines are being pushed in ways the model does not fully capture.

The study also highlighted a structural limitation that no amount of better management can solve. In 2022, a year with unusually high water inflow, the reservoir was forced to spill over 63 million cubic meters of water. This spill occurred even when the model had perfect knowledge of the weather and water levels in advance. The cause was a combination of the reservoir's reduced storage capacity and the physical limit of the turbines, which can only pass a certain amount of water per second. When the water came in faster than the turbines could handle and the tank was already full, the excess had to be let go. This finding points to a hard infrastructure constraint: the current equipment cannot handle the full flow of wet years, leading to wasted water that could otherwise be used for power or stored for later.

The researchers also tested how the system would respond if the demand for irrigation water increased. They found that asking for more water did not result in significantly more electricity. As the irrigation order was raised from 280 million cubic meters to 350 million cubic meters, the amount of energy generated remained almost exactly the same, while the amount of unmet water demand grew sharply. This indicates that simply demanding more water does not make the system more efficient; instead, it creates a larger gap between what is needed and what can be delivered. The study concludes that the most effective path forward is to treat the irrigation order as a flexible target that can be adjusted based on the actual water available, rather than a fixed promise that must be kept at all costs.

Ultimately, this work demonstrates that in a reservoir that has lost a significant portion of its capacity, the old rules no longer apply. The solution is not to try to force the system to behave as if it were still full, but to adapt the management strategy to the new reality. By using a model that constantly checks the current water level and the specific needs of the season, operators can protect the water supply, avoid dangerous low levels, and still find opportunities to generate more power. The study provides a clear roadmap for how to manage a shrinking resource, showing that with the right approach, it is possible to maintain security for farmers and power grids even as the physical landscape changes beneath them. The findings suggest that the next step for the Akhangaran Reservoir is to update its operational rules to reflect its current size and to consider upgrading the turbines to handle the high flows that currently cause such large spills.